Acta Orthopaedica, Vol. 91, Issue 6, 2020 by Acta

6/20

* Sprowson AP et al. Bone Joint J 2016; 98-B: 1534–1541

www.heraeus-medical.com

Vol. 91, No. 6, 2020 (pp. 621–803)

Bone cement with gentamicin ttamicin i i and clindamycin

ACTA ORTHOPAEDICA

69 %

reduction of deep infections in hip hemiarthroplasty after fractured neck of femur *

Volume 91, Number 6, December 2020

TF-IORT200163.indd 1

20-11-2020 16:13:12

Acta Orthopaedica is owned by the Nordic Orthopaedic Federation and is the official publication of the Nordic Orthopaedic Federation

E DITORIAL O F FICE

Acta Orthopaedica Department of Orthopedics Lund University Hospital SE–221 85 Lund, Sweden E-mail: acta.ort@med.lu.se Homepage: http://www.actaorthop.org

EDITOR

THE FOUNDATION BOARD OF

Anders Rydholm Lund, Sweden

THE NORDIC O RTHOPAEDIC F EDERATION AND A CTA O RTHOPAEDICA

DEPUTY EDITOR

Peter A Frandsen Odense, Denmark CO-EDITORS

Li Felländer-Tsai Stockholm, Sweden Nils Hailer Uppsala, Sweden Ivan Hvid Oslo, Norway Urban Rydholm Lund, Sweden Bart A Swierstra Wageningen, The Netherlands Eivind Witsø Trondheim, Norway Rolf Önnerfält Lund, Sweden

Peter Frandsen Denmark Ragnar Jonsson Iceland Heikki Kröger Finland Anders Rydholm Sweden Kees Verheyen the Netherlands

WEB EDITOR

Magnus Tägil Lund, Sweden S TATISTICAL EDITOR

Jonas Ranstam Lund, Sweden P RODUCTION MANAGER

Kaj Knutson Lund, Sweden

Vol. 91, No. 6, 2020

SUBSCRIPTION INFORMATION Acta Orthopaedica [print 1745-3674, online 1745-3682] is a peerreviewed journal, published six times a year plus supplements by Taylor & Francis on behalf of Nordic Orthopaedic Federation.

Airfreight and mailing in the USA by agent named WN Shipping USA, 156-15, 146th Avenue, 2nd Floor, Jamaica, NY 11434, USA. Periodicals postage paid at Jamaica NY 11431.

Annual Institutional Subscription, Volume 91, 2020

US Postmaster: Send address changes to Acta Orthopaedica, WN Shipping USA, 156-15, 146th Avenue, 2nd Floor, Jamaica, NY 11434, USA.

$1,291

£798

€1,035

The subscription fee purchases an online subscription. The price includes access to current content and back issues to January 1997 (if available). Printed copies of the journal are provided on request as a free supplementary service accompanying an online subscription. Supplements to the journal are also included in the subscription price. For more information, visit the journal’s website: http://www.tandfonline.com/IORT Manuscripts should be uploaded at http://www.manuscriptmanager.com/ao/ for further handling at: Acta Orthopaedica Editorial Office, Department of Orthopaedics, Lund University Hospital, SE-221 85 Lund, Sweden Correspondance concerning copyright and permissions should be sent to: Maria Montzka, Portfolio Manager – Medicine P.O. Box 3255, SE-103 65 Stockholm, Sweden, Tel: +46 (0)760 14 24 68. Fax: +46 (0)8 440 80 50. E-mail: maria.montzka@informa.com Ordering information: Please contact your local Customer Service Department to take out a subscription to the Journal: USA, Canada: Taylor & Francis, Inc., 530 Walnut Street, Suite 850, Philadelphia, PA 19106, USA. Tel: +1 800 354 1420; Fax: +1 215 207 0050. UK/ Europe/Rest of World: T&F Customer Services, Informa UK Ltd, Sheepen Place, Colchester, Essex, CO3 3LP, United Kingdom. Tel: +44 (0) 20 7017 5544; Fax: +44 (0) 20 7017 5198; Email: subscriptions@tandf.co.uk Dollar rates apply to all subscribers outside of Europe. Euro rates apply to all subscribers in Europe except the UK and Republic of Ireland. If you are unsure which applies, contact Customer Services. All subscriptions are payable in advance and all rates include postage. Journals are sent by air to the USA, Canada, Mexico, India, Japan and Australasia. Subscriptions are entered on an annual basis, i.e., January to December. Payment may be made by sterling check, US dollar check, euro check, international money order, National Giro, or credit card (Amex, Visa and Mastercard). Back issues: Taylor & Francis retains a two-year back issue stock of journals. Older volumes are held by our official stockists to whom all orders and enquiries should be addressed: Periodicals Service Company, 351 Fairview Ave., Suite 300, Hudson, New York 12534, USA. Tel: +1 518 537 4700; fax: +1 518 537 5899; e-mail: psc@ periodicals.com.

Subscription records are maintained at Taylor & Francis Group, 4 Park Square, Milton Park, Abingdon, OX14 4RN, United Kingdom.

Copyright © 2020 The Author(s). Published by Taylor & Francis on behalf of the Nordic Orthopedic Federation. This is an Open Access article distributed under the terms of the Creative Commons Attribution-Non-Commercial License (https://creativecommons.org/licenses/by-nc/3.0 . Informa UK Limited, trading as Taylor & Francis Group makes every effort to ensure the accuracy of all the information (the “Content”) contained in its publications. However, Informa UK Limited, trading as Taylor & Francis Group, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Informa UK Limited, trading as Taylor & Francis Group. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Informa UK Limited, trading as Taylor & Francis Group shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. Terms & Conditions of access and use can be found at http://www.tandfonline. com/page/terms-and-conditions Indexed/abstracted in: Allied and Complementary Medicine Library (Amed); ASCA (Automatic Subject Citation Alert); Biological Abstracts; Chemical Abstracts; Cumulative Index to Nursing and Allied Health Literature(CINAHL); Current Advances in Ecological and Environmental Sciences; Current Contents/Clinical Medicine; Current Contents/Life Sciences; Developmental Medicine and Child Neurology; Energy Research Abstracts; EMBASE/ Excerpta Medica; Faxon Finder; Focus On: Sports Science & Medicine; Health Planning and Administration; Index Medicus/MEDLINE; Index to Dental Literature; Index Veterinarius; INIS Atomindex; Medical Documentation Service; Nuclear Science Abstracts (Ceased); Periodicals Scanned and Abstracted. Life Sciences Collection; Research Alert; Science Citation Index; SciSearch; SportSearch; Uncover Veterinary Bulletin. Printed in England by Henry Ling

Acta Orthopaedica

ISSN 1745-3674

Vol. 91, No. 6, December 2020 Guest editorial “Great balls on fire:” known algorithm with a new instrument? Annotation To mix or not to mix? Medicolegal implications of mixed components in total hip arthroplasty COVID-19 The impact of COVID-19 on the future of orthopaedic training in the UK Impact of the COVID-19 pandemic on paediatric orthopaedic trauma workload in central London: a multi-centre longitudinal observational study over the “golden weeks:” The COVid Emergency Related Trauma and orthopaedics (COVERT) Collaborative Delayed surgery versus nonoperative treatment for hip fractures in post-COVID-19 arena: a retrospective study of 145 patients Preparing an orthopedic department for COVID-19: Lessons learned from reorganization and educational activities Redeployment of the trainee orthopaedic surgeon during COVID19: a fish out of water? Hip The anatomical SP-CL stem demonstrates a non-progressing migration pattern in the first year: a low dose CT-based migration study in 20 patients Long-term migration of a cementless stem with different bioactive coatings. Data from a “prime” RSA study: lessons learned HipSim — hip fracture surgery simulation utilizing the Learning Curve–Cumulative Summation test (LC-CUSUM) Electromagnetic navigation system for acetabular component placement in total hip arthroplasty is more precise and accurate than the freehand technique: a randomized, controlled trial with 84 patients Does cup position differ between trabecular metal and titanium cups? A radiographic propensity score matched study of 300 hips Do hip precautions after posterior-approach total hip arthroplasty affect dislocation rates? A systematic review of 7 studies with 6,900 patients Migration of the uncemented Echo Bi-Metric and Bi-Metric THA stems: a randomized controlled RSA study involving 62 patients with 24-month follow-up Automated classification of hip fractures using deep convolutional neural networks with orthopedic surgeon-level accuracy: ensemble decision-making with antero-posterior and lateral radiographs A vitamin E blended highly cross-linked polyethylene acetabular cup results in less wear: 6-year results of a randomized controlled trial in 199 patients No association between blood count levels and whole-blood cobalt and chromium levels in 1,900 patients with metal-on-metal hip arthroplasty Non-surgical treatment before hip and knee arthroplasty remains underutilized with low satisfaction regarding performance of work, sports, and leisure activities Femoral shaft fracture Outcome at 1 year in patients with femoral shaft fractures treated with intramedullary nailing or skeletal traction in a low-income country: a prospective observational study of 187 patients in Malawi Knee Predicting the mechanical hip–knee–ankle angle accurately from standard knee radiographs: a cross-validation experiment in 100 patients

621

S M Röhrl

624

R M Peters, J T Hiemstra, Wp Zijlstra, S K Bulstra, and M Stevens

627

R Dattani, C Morgan, L Li, K Bennett-Brown, and R M H Wharton K Sugand, C Park, C Morgan, R Dyke, A Aframian, A Hulme, S Evans, K M Sarraf, and the COVERT Collaborative

633

639 644 650

B Mi, L Chen, D Tong, A C Panayi, F Ji, J Guo, Z Hou, Y Zhang, Y Xiong, and G Liu R D Jensen, M Bie, A P Gundsø, J M Schmid, J Juelsgaard, M L Gamborg, H Mainz, and J D Rölfing G Faria, B J Tadros, N Holmes, S Virani, G K Reddy, B S Dhinsa, and J Relwani

654

O Sandberg, S Tholén, S Carlsson, and P Wretenberg

660 669

P van der Voort, M L D Klein Nulent, E R Valstar, B L Kaptein, M Fiocco, and R G H H Nelissen J D Rölfing, R D Jensen, and C Paltved

675

R Mihalič, J Zdovc, J Mohar, and R Trebše

682

I Laaksonen, N Hjelmberg, K Gromov, A E Eskelinen, O Rolfson, H Malchau, A Troelsen, K T Mäkelä, and M Mohaddes Crompton, L Osagie-Clouard, and A Patel

687 693

K Dyreborg, M R Andersen, N Winther, S Solgaard, G Flivik, and M M Petersen

699

705

Y Yamada, S Maki, S Kishida, H Nagai, J Arima, N Yamakawa, Y Iijima, Y Shiko, Y Kawasaki, T Kotani, Y Shiga, K Inage, S Orita, Y Eguchi, H Takahashi, T Yamashita, S Minami, and S Ohtori J R A Massier, J H J van Erp, T E Snijders, and A de Gast

711

N Honkasaari, O Lainiala, O Laine, A Reito, and A Eskelinen

717

Y van Zaanen, A Hoorntje, K L M Koenraadt, L van Bodegom–Vos, G M M J Kerkhoffs, S Waterval–Witjes, T A E J Boymans, R C I van Geenen, and P P F M Kuijer

724

L Chokotho, H-H Wu, D Shearer, B C Lau, N Mkandawire, J-E Gjertsen, G Hallan, and S Young

732

W P Gielis, H Rayegan, V Arbabi, S Y Ahmadi Brooghani, C Lindner, T F Cootes, P A de Jong, H Weinans, and R J H Custers

The implications of an aging population and increased obesity for knee arthroplasty rates in Sweden: a register-based study Intra-articular injection with Autologous Conditioned Plasma does not lead to a clinically relevant improvement of knee osteoarthritis: a prospective case series of 140 patients with 1-year follow-up Does intraoperative contamination during primary knee arthroplasty affect patient-reported outcomes for patients who are uninfected 1 year after surgery? A prospective cohort study of 714 patients Increase in early wound leakage in total knee arthroplasty with local infiltrative analgesia (LIA) that includes epinephrine: a retrospective cohort study Tibial lengthening Tibial lengthening using a retrograde magnetically driven intramedullary lengthening device in 10 patients with preexisting ankle and hindfoot fusion Ankle Randomized trial comparing suture button with single 3.5 mm syndesmotic screw for ankle syndesmosis injury: similar results at 2 years Shoulder Low arthroplasty survival after treatment for proximal humerus fracture sequelae: 3,245 shoulder replacements from the Nordic Arthroplasty Register Association Rotator cuff repair with biological graft augmentation causes adverse tissue outcomes No need to use both Disabilities of the Arm, Shoulder and Hand and Constant–Murley score in studies of midshaft clavicular fractures Infection Similar risk of complete revision for infection with single-dose versus multiple-dose antibiotic prophylaxis in primary arthroplasty of the hip and knee: results of an observational cohort study in the Dutch Arthroplasty Register in 242,179 patients Correspondence Trauma and orthopaedics training amid COVID-19: A medical student’s perspective Delayed surgery versus nonoperative treatment for hip fractures in post-COVID-19 situation Information to authors (see http://www.actaorthop.org/)

738

A Overgaard, P Frederiksen, L E Kristensen, O Robertsson, and A W-Dahl

743

J V Korpershoek, L A Vonk, T S De Windt, J Admiraal, E C Kester, N Van Egmond, D B F Saris, and R J H Custers T Justesen, J B Olsen, A B Hesselvig, A Mørup–Petersen, and A Odgaard

750 756

B C van der Zwaard, R L Roerdink, and R P van Hove

761

B Vogt, R Roedl, G Gosheger, G Toporowski, A Laufer, C Theil, J N Broeking, and A Frommer

770

B W Ræder, I K Stake, J E Madsen, F Frihagen, S B Jacobsen, M R Andersen, and W Figved

776

D Unbehaun, S Rasmussen, R Hole, A M Fenstad, B Salomonsson, Y Demir, S L Jensen, S Brorson, V Äärimaa, I Mechlenburg, and J V Rasmussen M S Rashid, R D J Smith, N Nagra, K Wheway, B Watkins, S Snelling, S G Dakin, and A J Carr A H Qvist, M T Væsel, C Moss, T Jakobsen, and S L Jensen

782 789

794

E S Veltman, E Lenguerrand, D J F Moojen, M R Whitehouse, R G H H Nelissen, A W Blom, and R W Poolman

801

C M Bigogno and K S Rallis versus C Morgan and R Dattani

803

V Wiwanitkit versus G Liu

Acta Orthopaedica 2020; 91 (6): 621–623

621

Guest editorial

“Great balls on fire:” known algorithm with a new instrument? I am sure Jerry Lee Lewis would excuse my witticism with the title of his famous rock ’n roll song. However, it depicts metaphorically that radiostereometric analysis (RSA) has become even more of a hot topic lately. You might either love or be frustrated about RSA. Journal editors and study investigators love this high-precision measuring method for its high resolution and objectivity, but researchers sometimes despair over the meticulous working steps and processing of data necessary in order not to lose any precious study patients. Implant companies are often positive toward RSA studies but reluctant to mark a batch of implants with beads directly in the production process. Nevertheless, no other imaging entity can so far beat the precision and accuracy of RSA in spite of the great advances of MRI and CT. Since its introduction by Selvik (1989) RSA has over the past 40 years evolved to be the most trusted tool to monitor implant migration just like an orthopedic GPS separating stable and unstable implants or measuring material wear to the tenth of a millimeter. Meanwhile RSA has gained an almost religious status through its ability to predict the survival of implants by extrapolating migration data after a short time in situ (Kärrholm 2012). As far back as the 1990s, Kärrholm showed the predictive value for late failure of early migration in cemented stems for the hip (Kärr holm et al. 1994). By now the evidence for similar patterns is also increasing for acetabular cups (Pijls et al. 2012) and for selected uncemented hip stems (de Vries et al. 2014) as well as for tibial components in total knee arthroplasty (Pijls et al. 2018). However, in general the interpretation of early migration of uncemented implants seems to be more complicated. We do not know for how long after implantation migration of uncemented hip and knee components can last without compromising long-lasting stability. With regard to other types of artificial joints (e.g., in the upper extremity) the threshold for acceptable early migration is still poorly documented. Nevertheless, it is obvious that continuous migration of an implant will eventually end in painful clinical loosening. RSA studies can disclose the success or failure of an implant and thereby save thousands of unaware patients from poorly performing implants. This is strongly needed as the revelation of “implant files” (Lenzer et al. 2018) has shown the insufficiency of preclinical implant testing. If the implants are faulty it also puts the surgeon in the line of fire despite his or her treatment with good intentions. Several authors have therefore

outlined the importance of RSA in the quest for stepwise and safe introduction of new implants (Malchau 2000, Nelissen et al. 2011). The International Radiostereometry Society and other pioneers are still working hard to standardize the execution, analysis, and presentation of RSA results to render the method useful for orthopedic surgeons all over the world (Kärrholm et al. 1997, Ryd et al. 2000, Valstar et al. 2005). Just recently Pijls (2020) commented on the positive effect of standardization of RSA, which even resulted in an ISO norm 16087:2013(E). Over 700 studies are registered in PubMed with Acta Orthopaedica as the primary platform with over 400 articles alone. Would it not therefore be handy to perform a reliable migration study for any implant or even follow up each patient with RSA? An appealing thought, but although digital RSA has become a lot more user friendly compared with the times when the data was stored on punch cards or markers were marked manually, it is still an invasive method and has not yet found its way into the clinical workday as a diagnostic tool except in some RSA centers (Horsager et al. 2017). Sophisticated edgedetection software has mostly obliviated the need for markers in implants (Kaptein et al. 2003, Lindgren et al. 2020) but the human bone still has to be marked with tantalum markers as reference during the surgery. This issue of Acta includes an article from van der Voort et al. (2020) with 25 years’ RSA follow-up, which is the longest of its kind ever. An uncemented stem with 3 different coatings is stable irrespective of implant surface used. The article illustrates the pitfalls and lessons learned with clinical RSA studies as mentioned above. A second article by Sandberg et al. (2020), also in this issue, reports the migration pattern of a new uncemented femoral stem analyzed with a new method based on CT. Likewise, as earlier RSA studies of other stems have shown, after initial minimal migration the stem settles and is stable. Can we rely on this data? Since the turn of the millennium a group of researchers in Sweden have explored the possibilities of providing similar migration data with CT as with RSA. Olivecrona et al. (2002) showed in the hip that it is in principal possible to perform migration and wear measurements with CT. Initially the researchers still used markers to identify the bone as a reference (Olivecrona et al. 2004, Otten et al. 2017) but bony landmarks and implants can be identified by image segmentation and fusion techniques alone without any markers (Noz et al.

© 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group, on behalf of the Nordic Orthopedic Federation. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. DOI 10.1080/17453674.2020.1840029

622

2001). Additionally, earlier downsides of CT have changed. First, CT technology has taken a huge step forward in reducing the irradiation dosage for the patients (Sandgren et al. 2016) and second the amount of metal artefacts caused by implants could be reduced considerably by new measures (Lell et al. 2013, Wellenberg et al. 2016). These developments might have become a game changer for the group around Olivecrona and Weidenhjelm, who have teamed up with a strong imaging company to develop a software interface suitable for radiologists and orthopedic surgeons with a special interest (Olivecrona et al. 2004, Jedenmalm et al. 2011, Maguire et al. 2014, Svedmark et al. 2015, Otten et al. 2017, Eriksson et al. 2019, Broden et al. 2020b). Low-dose CT provides direct 3-dimensional data, which renders obsolete a cage and double examinations as used in classic RSA. Image segmentation and fusion algorithms even make the use of markers for bone and implant unnecessary. The technology comes in 2 forms: computer tomography motion analysis (CTMA) as migration analysis over time (Broden et al. 2020b) or image motion analysis (IMA) as stability testing on the same day between 2 examinations separated by a provocation (Svedmark et al. 2015). The latter is comparable to 2 RSA examinations on the same day in an unloaded and a loaded position (Bragonzoni et al. 2005, Dunbar et al. 2012, Kibsgard et al. 2017, Lam Tin Cheung et al. 2018). The algorithm and software behind the CT analysis, however, are basically the same. The potential of this new technology seems vast. CTMA technology might make in vivo testing of new implants available for anyone, which might finally make a stepwise introduction of all new implants feasible. IMA might open up to become an ubiquitous diagnostic tool for implant loosening or any other type of joint instability, which might give the orthopedic surgeon a more precise indication for surgery. CTMA and IMA have the potential to revolutionize the way we quantify instability. And this not only for the hip but for any other joint—with or without an implant (Olivecrona et al. 2016, Broden et al. 2020a)! We now have to question whether the time is right for a paradigm shift for in vivo migration measurements. Not quite I would suggest. Before widespread use of this new technology it has to live up to the standards of proven RSA. We have to test and validate this method thoroughly against the existing gold standard RSA because there might be some methodological pitfalls we are not aware of this far. Metal artifact reduction will not be equally effective for all implants. Type of metal, size, shape, and thickness will probably affect how precisely an implant can be identified (Radzi et al. 2014). Bone is living matter that changes its shape over time. Bony landmarks might therefore change their form and position. Both might lead to loss of precision and accuracy, rendering data useless. RSA has been an orthopedic domain. Developed by orthopedic surgeons and engineers, it has been performed and

Acta Orthopaedica 2020; 91 (6): 621–623

driven by orthopedic surgeons with the help of radiographers. CT technology and its data analysis are mainly in the hands of radiologists. This might in the future change the working distribution. Orthopedic surgeons provide the clinical questions and the radiologist provides the measuring method and maybe even the analysis. Also, the industry should have an interest but it might not come away financially as easily as before with ambitious orthopedic surgeons willing to devote years or even their lives to meticulous RSA analysis. All parties will have to find a way to collaborate for the sake of our patients. Nevertheless, the new CT technique is here and promising. Offering a new tool, it might open up possibilities not yet even imagined. It is now up to the RSA researchers around the world to play with and test this new method down to the bones in every aspect and detail. Maybe we are on the verge of gaining a new useful tool for all orthopedic surgeons? Referring back to music, no matter what rhythm we play in the future, the algorithm has to be correct. RSA players around the world: let’s test this new instrument! Stephan M Röhrl Division of Orthopaedic Surgery, Oslo University Hospital, Oslo, Norway Email: s.m.rohrl@medisin.uio.no Bragonzoni L, Russo A, Loreti I, Montagna L, Visani A, Marcacci M. The stress-inducible displacement detected through RSA in non-migrating UKR. Knee 2005; 12(4): 301-6. doi: 10.1016/j.knee.2004.09.006. Broden C, Giles J W, Popat R, Fetherston S, Olivecrona H, Sandberg O, Maguire G Q, Jr, Noz M E, Skoldenberg O, Emery R. Accuracy and precision of a CT method for assessing migration in shoulder arthroplasty: an experimental study. Acta Radiol 2020a; 61(6): 776-82. doi: 10.1177/0284185119882659. Broden C, Sandberg O, Skoldenberg O, Stigbrand H, Hanni M, Giles J W, Emery R, Lazarinis S, Nystrom A, Olivecrona H. Low-dose CT-based implant motion analysis is a precise tool for early migration measurements of hip cups: a clinical study of 24 patients. Acta Orthop 2020b; 91(3): 260-5. doi: 10.1080/17453674.2020.1725345. de Vries L M, van der Weegen W, Pilot P, Stolarczyk P A, Sijbesma T, Hoffman E L. The predictive value of radiostereometric analysis for stem survival in total hip arthroplasty: a systematic review. Hip Int 2014; 24(3): 215-22. doi: 10.5301/hipint.5000102. Dunbar M J, Fong J W, Wilson D A, Hennigar A W, Francis P A, Glazebrook M A. Longitudinal migration and inducible displacement of the Mobility Total Ankle System. Acta Orthop 2012; 83(4): 394-400. doi: 10.3109/17453674.2012.712890. Eriksson T, Maguire G Q Jr, Noz M E, Zeleznik M P, Olivecrona H, Shalabi A, Hanni M. Are low-dose CT scans a satisfactory substitute for stereoradiographs for migration studies? A preclinical test of low-dose CT scanning protocols and their application in a pilot patient. Acta Radiol 2019; 60(12): 1643-52. doi: 10.1177/0284185119844166. Horsager K, Kaptein B L, Romer L, Jorgensen P B, Stilling M. Dynamic RSA for the evaluation of inducible micromotion of Oxford UKA during step-up and step-down motion. Acta Orthop 2017; 88(3): 275-81. doi: 10.1080/17453674.2016.1274592. Jedenmalm A, Nilsson F, Noz M E, Green D D, Gedde U W, Clarke I C, Stark A, Maguire G Q Jr, Zeleznik M P, Olivecrona H. Validation of a 3D CT method for measurement of linear wear of acetabular cups. Acta Orthop 2011; 82(1): 35-41. doi: 10.3109/17453674.2011.552777.

Acta Orthopaedica 2020; 91 (6): 621–623

Kaptein B L, Valstar E R, Stoel B C, Rozing P M, Reiber J H. A new modelbased RSA method validated using CAD models and models from reversed engineering. J Biomech 2003; 36(6): 873-82. doi: 10.1016/s00219290(03)00002-2. Kärrholm J. Radiostereometric analysis of early implant migration: a valuable tool to ensure proper introduction of new implants. Acta Orthop 2012; 83(6): 551-2. doi: 10.3109/17453674.2012.745352. Kärrholm J, Borssen B, Lowenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 4–7-year stereoradiographic follow-up of 84 cemented prostheses. J Bone Joint Surg Br 1994; 76(6): 912-17. Kärrholm J, Herberts P, Hultmark P, Malchau H, Nivbrant B, Thanner J. Radiostereometry of hip prostheses: review of methodology and clinical results. Clin Orthop Relat Res 1997 (344): 94-110. Kibsgard T J, Röhrl S M, Røise O, Sturesson B, Stuge B. Movement of the sacroiliac joint during the Active Straight Leg Raise test in patients with long-lasting severe sacroiliac joint pain. Clin Biomech (Bristol, Avon) 2017; 47: 40-5. doi: 10.1016/j.clinbiomech.2017.05.014. Lam Tin Cheung K, Lanting B A, McCalden R W, Yuan X, MacDonald S J, Naudie D D, Teeter M G. Inducible displacement of cemented tibial components ten years after total knee arthroplasty. Bone Joint J 2018; 100B(2): 170-5. doi: 10.1302/0301-620X.100B2.BJJ-2017-0428.R2. Lell M M, Meyer E, Schmid M, Raupach R, May M S, Uder M, Kachelriess M. Frequency split metal artefact reduction in pelvic computed tomography. Eur Radiol 2013; 23(8): 2137-45. doi: 10.1007/s00330-013-2809-y. Lenzer J, ICIJ reporters I. Medical device industry: international investigation exposes lax regulation. BMJ 2018; 363: k4997. doi: 10.1136/bmj.k4997. Lindgren L, Jorgensen P B, Morup R M S, Jensen M, Romer L, Kaptein B, Stilling M. Similar patient positioning: a key factor in follow-up studies when using model-based radiostereometric analysis of the hip. Radiography (Lond) 2020; 26(2): e45-e51. doi: 10.1016/j.radi.2019.10.009. Maguire G Q Jr, Noz M E, Olivecrona H, Zeleznik M P, Weidenhielm L. A new automated way to measure polyethylene wear in THA using a high resolution CT scanner: method and analysis. Sci World J 2014; 2014:528407. doi: 10.1155/2014/528407. Malchau H. Introducing new technology: a stepwise algorithm. Spine (Phila Pa 1976) 2000; 25(3): 285. doi: 10.1097/00007632-200002010-00004. Nelissen R G, Pijls B G, Kärrholm J, Malchau H, Nieuwenhuijse M J, Valstar E R. RSA and registries: the quest for phased introduction of new implants. J Bone Joint Surg Am 2011; 93(Suppl. 3): 62-5. doi: 10.2106/ JBJS.K.00907. Noz M E, Maguire GQ Jr, Zeleznik M P, Kramer E L, Mahmoud F, Crafoord J. A versatile functional-anatomic image fusion method for volume data sets. J Med Syst 2001; 25(5): 297-307. doi: 10.1023/a:1010633123512. Olivecrona L, Crafoord J, Olivecrona H, Noz M E, Maguire G Q, Zeleznik M P, Svensson L, Weidenhielm L. Acetabular component migration in total hip arthroplasty using CT and a semiautomated program for volume merging. Acta Radiol 2002; 43(5): 517-27. Olivecrona H, Weidenhielm L, Olivecrona L, Beckman M O, Stark A, Noz M E, Maguire G Q Jr, Zeleznik M P, Svensson L, Jonson T. A new CT method for measuring cup orientation after total hip arthroplasty: a study of 10 patients. Acta Orthop Scand 2004; 75(3): 252-60. doi: 10.1080/00016470410001169.

623

Olivecrona H, Maguire G Q Jr, Noz M E, Zeleznik M P, Kesteris U, Weidenhielm L. A CT method for following patients with both prosthetic replacement and implanted tantalum beads: preliminary analysis with a pelvic model and in seven patients. J Orthop Surg Res 2016; 11: 27. doi: 10.1186/ s13018-016-0360-7. Otten V, Maguire G Q Jr, Noz M E, Zeleznik M P, Nilsson K G, Olivecrona H. Are CT scans a satisfactory substitute for the follow-up of RSA migration studies of uncemented cups? A comparison of RSA double examinations and CT datasets of 46 total hip arthroplasties. Biomed Res Int 2017; 2017: 3681458. doi: 10.1155/2017/3681458. Pijls B G. Reflections on the RSA guidelines. Acta Orthop 2020; 91(3): 232-3. doi: 10.1080/17453674.2020.1763568. Pijls B G, Nieuwenhuijse M J, Fiocco M, Plevier J W, Middeldorp S, Nelissen R G, Valstar E R. Early proximal migration of cups is associated with late revision in THA: a systematic review and meta-analysis of 26 RSA studies and 49 survival studies. Acta Orthop 2012; 83(6): 583-91. doi: 10.3109/17453674.2012.745353. Pijls B G, Plevier J W M, Nelissen R. RSA migration of total knee replacements. Acta Orthop 2018; 89(3): 320-8. doi: 10.1080/17453674.2018.1443635. Radzi S, Cowin G, Robinson M, Pratap J, Volp A, Schuetz M A, Schmutz B. Metal artifacts from titanium and steel screws in CT, 1.5T and 3T MR images of the tibial Pilon: a quantitative assessment in 3D. Quant Imaging Med Surg 2014; 4(3): 163-72. doi: 10.3978/j.issn.2223-4292.2014.03.06. Ryd L, Yuan X, Lofgren H. Methods for determining the accuracy of radiostereometric analysis (RSA). Acta Orthop Scand 2000; 71(4): 403-8. doi: 10.1080/000164700317393420. Sandberg O, Tholén S, Carlsson S, Wretenberg P. The anatomical SP-CL stem demonstrates a non-progressing migration pattern the first year: a low dose CT-based migration study in 20 patients. Acta Orthop 2020; 91(6): 654-9. Sandgren B, Skorpil M, Nowik P, Olivecrona H, Crafoord J, Weidenhielm L, Persson A. Assessment of wear and periacetabular osteolysis using dual energy computed tomography on a pig cadaver to identify the lowest acceptable radiation dose. Bone Joint Res 2016; 5(7): 307-13. doi: 10.1302/2046-3758.57.2000566. Selvik G. Roentgen stereophotogrammetry: a method for the study of the kinematics of the skeletal system. Acta Orthop Scand 1989; 232(Suppl.): 1-51. Svedmark P, Berg S, Noz M E, Maguire G Q Jr, Zeleznik M P, Weidenhielm L, Nemeth G, Olivecrona H. A new CT method for assessing 3D movements in lumbar facet joints and vertebrae in patients before and after TDR. Biomed Res Int 2015; 2015:260703. doi: 10.1155/2015/260703. Valstar E R, Gill R, Ryd L, Flivik G, Borlin N, Kärrholm J. Guidelines for standardization of radiostereometry (RSA) of implants. Acta Orthop 2005; 76(4): 563-72. doi: 10.1080/17453670510041574. van der Voort P, Klein Nulent M L D, Valstar E R, Kaptein B L. Long-term migration of a cementless stem with different bioactive coatings. Data from a “prime” RSA study; lessons learned. Acta Orthop 2020; 91(6): 660-8. Wellenberg R H, Boomsma M F, van Osch J A, Vlassenbroek A, Milles J, Edens M A, Streekstra G J, Slump C H, Maas M. Computed tomography imaging of a hip prosthesis using iterative model-based reconstruction and orthopaedic metal artefact reduction: a quantitative analysis. J Comput Assist Tomogr 2016; 40(6): 971-8. doi: 10.1097/RCT.0000000000000449.

624

Acta Orthopaedica 2020; 91 (6): 624–626

Annotation

To mix or not to mix? Medicolegal implications of mixed components in total hip arthroplasty Rinne M PETERS 1,2 a, Jantina T HIEMSTRA 3,4 a, Wierd P ZIJLSTRA 1, Sjoerd K BULSTRA 2, and Martin STEVENS 2 1 Department of Orthopedic Surgery, Medical Center Leeuwarden; 2 Department of Orthopedic Surgery, University of Groningen, University Medical Center Groningen; 3 Hausfeld Advocaten, Amsterdam; 4 Department of Private law, University of Groningen, The Netherlands a Shared first authorship Correspondence: rinnepeters@gmail.com Submitted 2020-05-19. Accepted 2020-08-25

About 300 different hip prostheses promoted by a multitude of distributors are available on the European market. Most total hip arthroplasties (THAs) are assembled from components produced by the same manufacturer (non-mixed THAs), yet certain situations require a combination of components from different manufacturers within a single hip prosthesis (mixed THAs). Despite it being against manufacturers’ guidelines (Smith & Nephew 2013, Link 2018), orthopedic surgeons who do this are encouraged by clinical results that are comparable to and sometimes even superior to those obtained without mixed components (Tucker et al. 2015, Peters et al. 2016, Taylor et al. 2018). This mixing and matching is common clinical practice. The question does remain as to whether it is allowable by law. In this annotation paper we assess the legality of mixed THAs based on European law. Mix and match: clinical perspective Mixed prostheses are deﬁned as THAs (stem, head, and cup) comprising components made by different manufactures. With a reported prevalence of 11%, 24%, and 15% in the Netherlands, New Zealand, and England and Wales, respectively, mixing and matching is common clinical practice (Tucker et al. 2015, Peters et al. 2016, Taylor et al. 2018). Based on these national joint registry studies, it was demonstrated that mixed THAs yield at least comparable and for certain combinations even better outcomes than THAs with components from the same manufacturer (Tucker et al. 2015, Peters et al. 2016, Taylor et al. 2018). The concept of mixed THA refers to both fixed (trunnion/ taper) and mobile (head/cup) combinations as well as hard-onsoft and hard-on-hard bearings. A distinction should additionally be made between primary and revision procedures. An

argument for the use of mixed components in primary THA could be the need for a dual mobility cup in case of high risk of instability. Other arguments could be altered anatomy (e.g., developmental dysplasia of the hip), patient characteristics (e.g., frail elderly patients requiring cemented stems), and high-risk patients (e.g., prior lumbar spine fusion with irradiated pelvis). In revision arthroplasty, combining components from different manufacturers could be considered in order to prevent additional patient morbidity (e.g., leaving a well-fixed stem from another company in situ during a cup revision) (Mueller et al. 2018), or to optimize component placement performed by surgeons with extensive clinical experience. This is all in the best interest of the patient. Hard-on-soft mixing and matching across the femoral head and acetabular component (mobile bearings) have demonstrated excellent results for several combinations. For example, data from the National Joint Registry of England and Wales (NJR) showed that cemented stems with mixed polyethylene cups were associated with a lower risk for revision compared with their manufacturer-matched equivalents (Tucker et al. 2015). For fixed combinations, different taper sizes used by the various manufacturers have made it difficult for surgeons to combine the stem and head junction properly, as the stems and head can vary in shape, metallurgy, roughness, inclination, and angle (Werner et al. 2015). Mixed components over the trunnion–taper junction in THAs with large head and hard-on-hard bearings may result in wear of the femoral head– neck interface (trunnionosis), which has been reported as an increasingly prevalent cause of failure (Mistry et al. 2016). In THAs with ceramic heads, a mismatch can result in a fractured femoral head component.

Acta Orthopaedica 2020; 91 (6): 624–626

625

Legal implications The use of mixed components gives rise to legal implications from public and private law. One aspect of public law is that orthopedic implants have to be approved and marked Conformité Européene by an appropriate body before being allowed on the European market. This approval is given if the product meets the requirements of the Medical Devices Directive or its successor, the Medical Device Regulation, e.g., that the implant does not entail a safety risk (Directive 93/42/EEC 1993, Regulation (EU) 2017/745 2017). If a product is altered or a new product is designed by using several components that are not tested together, this approval might no longer be valid. Implications may derive from private law too. The unauthorized mixing of components can give rise to a risk of liability toward patients, as liability could be imposed for (1) producing a defective product or (2) medical negligence.

onstrated in 2015 that the use of heads and stems from different manufacturers in mixed THAs leads to increased revision rates (Tucker et al. 2015). Under certain circumstances, the producer can avoid the liability described above. Article 7 from the Directive sums up several defenses. For example, the producer will not be liable if it was impossible to know the risk that led to the defect because of a lack of knowledge. This refers to the objective scientific and technical knowledge available and accessible at that time, “including the most advanced level of such knowledge” (Commission of the European Communities v. United Kingdom of Great Britain and Northern Ireland 1997). As regards THAs the defense will most likely be unsuccessful if, at the time the THA was used by the surgeon, there was published scientific research available pointing out the risks of mixing components and materials.

Product liability Orthopedic surgeons who combine components from different manufacturers that are not designed, tested, or meant to be combined in compliance with the producers of the components bear a liability risk toward the patient. This risk derives from the European Product Liability Directive (85/374/EEC), which states that the producer of a product is liable for damages suffered by a patient if this product appears to be defective. This Directive is transposed into national law of all member states of the European Union and the European Free Trade Association. A healthcare provider who mixes components, such as a femoral head and stem from manufacturer A with a cup from manufacturer B, into a THA could qualify as a “manufacturer of a finished product” to whom the liability regime of the Directive applies (Gabrielczyk 2017).

Medical negligence Mixing of components could also result in liability of the healthcare provider if it qualifies as negligence. Liability for medical negligence is not regulated at the EU level, so regimes will vary per country. Generally, liability will require negligent behavior from the healthcare provider, meaning that he or she must have breached a standard of care (Cass 1936, HR 1990, BGH 1994, Bolam v. Friern HMC 2015). As opposed to the previous regime of strict liability of the producer, medical negligence generally requires that the healthcare provider commits a fault. Mixing of components might be considered negligent when it is unauthorized and discouraged by the manufacturer, untested by the orthopedic surgeon and unapproved according to public law—all the more in a primary situation, when reasonable alternatives are available and when medical publications have shown clinical risks. Whether or not the healthcare provider has acted negligently will be influenced by the communication of these risks to the patient and the receipt of the patient’s consent. And yet this might not be decisive due to the differences in knowledge and expertise between healthcare providers and patients. So, when mixing is a reasonable option, it is important to inform the patient about the use of mixed components, the benefits and potential risks, and reasonable alternatives, in order to gain the patient’s consent. A search of case law in the United Kingdom (UK), Germany, and the Netherlands revealed that until now no orthopedic surgeon has ever been held responsible as the manufacturer of a finished product of mixed components. In the past one trial in the UK has been started but this has not resulted in a ruling in which the orthopedic surgeon was held responsible as the manufacturer of a finished product of mixed components.

Defective product In order for a producer (manufacturer or orthopedic surgeon) to be liable, the product has to be defective. This means that the product does not provide the safety that an individual is entitled to expect (article 6 of the Directive). Relevant in this respect is a recent English ruling that metal-on-metal (MoM) prostheses were not defective in terms of the entitled expectation of safety of such prostheses in 2002 (Colin Gee and others v. Depuy International Limited 2018). To determine whether a product provides the safety a person is entitled to expect, relevant circumstances are: the presentation of the product, the use to which the product could reasonably be expected to be put, and the time when the product was put into circulation. With regard to the latter: the defectiveness will be determined based on the state of knowledge and safety standards at the time it was put into circulation. The fact that a better product was subsequently put into circulation will not lead to the conclusion that the product in question must be considered defective. For orthopedic surgeons, this means that the state of knowledge at the time of insertion of the prothesis is important. In this respect it is relevant, for instance, that it was dem-

Conclusion Mixing and matching in total hip arthroplasty is common practice worldwide. It is generally done in the interest of the patient, aiming to optimize the outcome of the treatment. We

626

assessed the rules for mixed THAs based on European law, to create awareness of the legality. Despite evident medical benefits and similar or even superior revision rates compared with non-mixed THAs (Tucker et al. 2015, Peters et al. 2016, Taylor et al. 2018), from a legal perspective it is advisable to avoid mixing when reasonable alternatives are available, especially in primary arthroplasty. The unauthorized mixing of components can create a liability risk based on European and national law. An orthopedic surgeon who mixes components from different manufactures could qualify as a “manufacturer of a finished product” and may be held liable without fault if the product appears to be defective. However, to date, no orthopedic surgeon has been held legally responsible or ended up in a lawsuit for the use of mixed components, based on case law review in the United Kingdom, Germany, and the Netherlands. Although no search was done of case laws in other European countries we presume that the situation in these countries can be considered representative of the situation in Europe as a whole. If a situation does require the use of mixed components, surgeons are best advised to (1) avoid mixing across the fixed articulation (i.e., use a head from the same manufacturer as the stem), (2) appropriately match sizes across the mobile articulation in hard-on-soft THAs (Tucker et al. 2015, Taylor et al. 2018), and (3) avoid mixing in hard-on-hard bearings. Surgeons are likewise advised to gain knowledge on the results of specific component combinations (e.g., based on arthroplasty registry results) and to explain the choices to the patient in order to receive his/her consent. Conflict of interest No beneﬁts in any form have been received or will be received in direct or indirect relation to the subject of this article. None of the authors have any financial or personal relationships with other individuals or organizations that could potentially and/or inappropriately influence (bias) this work or its conclusions.

Authors contributed to (1) study design, (2) gathered data, (3) initial draft, and (4) final draft. JTH and RMP contributed to (1), (2), (3), (4); WPZ, SKB, and MS contributed to (1), (4). The authors thank A H Hosman and R W Poolman for their contributions to the conception of this article. Acta thanks Pelle Gustafson and Maziar Mohaddes for help with peer review of this annotation.

Acta Orthopaedica 2020; 91 (6): 624–626

Gabrielczyk S. Produkthaftungsrechtliche Verantwortung bei Mix and Match. Medizinprodukterecht; 2017. p. 121. Mistry J B, Chughtai M, Elmallah R K, Diedrich A, Le S, Thomas M, Mont M A. Trunnionosis in total hip arthroplasty: a review. J Orthop Traumatol 2016; 17(1): 1-6. Mueller U, Panzram B, Braun S, Sonntag R, Kretzer J P. Mixing of head–stem components in total hip arthroplasty. J Arthroplasty 2018; 33(3): 945-51. BGH, November 29 1994, NJW 1995, 776. https://www.jurion.de/urteile/ bgh/1994-11-29/vi-zr-189_93/ (accessed January 25, 2019). Bolam v. Friern HMC [1957] 1 W.L.R. 582; Montgomery v. Lanarkshire Health Board (General Medical Council Intervening) [2015] UKSC11. Cass. Civ 20 mei 1936, DP 1936.1.88. Commission of the European Communities v. United Kingdom of Great Britain and Northern Ireland. CJEU, May 29, 1997, C-300/95. http://curia. europa.eu/juris/showPdf.jsf?docid=100739&doclang=EN (accessed January 25, 2019). Colin Gee and others v. Depuy International Limited [2018] EWHC 1208 (QB). Directive 93/42/EEC. Medical Devices Directive. https://eur-lex.europa.eu/ LexUriServ/LexUriServ.do?uri=CONSLEG:1993L0042:20071011:en: PDF (accessed July 7, 2020). HR, November 9, 1990. ECLI:NL:PHR:1990:AC1103. Link. Product guideline: Lubinus SPII. Anatomically Adapted Hip Prosthesis System; 2018. p. 33. http://www.linknederland.nl/_cache/link/media/ oifdr65606/Link_Nederland_SPII_operatietechniek_2018_04_004. pdf?hash=5e049bab4a3b5a8f (accessed July 7, 2020). Regulation (EU) 2017/745. Medical Devices Regulation. https://eur-lex. europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32017R0745&fro m=EN (accessed July 7, 2020). Smith & Nephew. Product guideline: Anthology primary hip system. Surgical technique; 2013. p. 26. https://www.smith-nephew.com/global/surgicaltechniques/recon/00811%20v1%20anthology%20st%2010.13.pdf (accessed July 7, 2020). Peters R M, Steenbergen van L N, Bulstra S K, Zeegers A V C M, Stewart R E, Poolman R W, Hosman A H. Nationwide review of mixed and nonmixed components from different manufacturers in total hip arthroplasty: a Dutch Arthroplasty Register study. Acta Orthop 2016; 87(4): 356-62. Taylor J W, Frampton C, Rothwell A G. Long-term survival of total hip arthroplasty using implants from different manufacturers. J Arthroplasty 2018; 33(2): 491-5. Tucker K, Pickford M, Newell C, Howard P, Hunt L P, Blom A W. Mixing of components from different manufacturers in total hip arthroplasty: prevalence and comparative outcomes. Acta Orthop 2015; 86: 671-7. Werner P H, Ettema H B, Witt F, Morlock M M, Verheyen C C. Basic principles and uniform terminology for the head–neck junction in hip replacement. Hip Int 2015; 25(2): 115-19.

Acta Orthopaedica 2020; 91 (6): 627–632

627

Perspective

The impact of COVID-19 on the future of orthopaedic training in the UK Rupen DATTANI, Catrin MORGAN, Lily LI, Katharine BENNETT-BROWN, and Rupert M H WHARTON

Department of Trauma and Orthopaedics, Chelsea and Westminster NHS Foundation Trust, London, UK Correspondence: rupen.dattani@nhs.net Submitted 2020-06-11. Accepted 2020-07-07.

ABSTRACT — The COVID-19 pandemic has had a major impact on global healthcare systems, has drastically affected patient care, and has had widespread effects upon medical education. As plans are being devised to reinstate elective surgical services, it is important to consider the impact that the pandemic has had and will continue to have on surgical training. We describe the effect COVID-19 has had at all levels of training in the UK within trauma and orthopaedics and evaluate how training might change in the future. We found that the COVID-19 pandemic has significantly impacted trainees within trauma and orthopaedics at all levels of training. It had led to reduced operative exposure, cancellations of examinations and courses, and modifications to speciality recruitment and annual appraisals. This cohort of trainees is witnessing novel methods of delivering orthopaedic services, which will continue to develop and become part of routine practice even once the pandemic has resolved. It will be important to observe the extent to which the rapid changes currently being introduced will impact the personal health, safety, and career progression of current trainees.

UK, it is important to consider the impact that the pandemic will have upon surgical training. This article describes the effect COVID-19 has had at all levels of training in the UK within T&O and evaluates how training might change in the future. The UK training pathway Medical students spend between 4 and 6 years of study at medical school before entering a foundation training scheme. This is a 2-year, work-based training programme after which doctors can apply to enter a specialty or general practice training programme. Orthopaedic specialty training is divided into 2 stages, core and higher specialty programmes (Figure). Core surgical training is usually 2 years and its purpose is to allow the trainee to develop basic and fundamental surgical skills common to all surgical specialities, together with some speciality-specific surgical skills. On completion of core training, trainees enter a 6-year specialty training scheme, which is akin to the residency programme in North America. Foundation year (FY) doctor experience There are estimated to be just under 14,000 FY doctors in the UK (The UK Foundation Programme, 2019). There are 6 rotations during foundation training and only 1 of these is in a surgical specialty. For many foundation doctors (FY), this may be the only exposure to surgery prior to choosing a specialty

The COVID-19 pandemic has had a major impact on global healthcare systems, has drastically affected patient care, and has had widespread effects upon medical education. On the 23rd of March 2020, the UK government imposed a lockdown and Speciality Certificate of introduced stringent social distancrecruitment Completion of MRCS training (CCT) ing measures in response to the rising completed FRCS completed number of COVID-19 infections. In the field of trauma and orthopaedics (T&O), COVID-19 led to an immediate restructuring of services, redeployment of doctors, and cancellation of elective operating. As plans are being devised to reinstate elective surgical services in the Training pathway for trauma and orthopaedic surgery in the UK.

628

training programme. Those with a rotation in trauma and orthopaedics often already have an interest in the specialty. The rotation gives an invaluable insight into the field of T&O and provides the trainees an opportunity to build their portfolio for the application to core surgical training. During the early stages of the pandemic, FY doctors were redeployed away from surgical specialties, including T&O, to medical specialties to support the emergency response. As a result, many FY doctors missed out on the opportunity to work within T&O and this may ultimately influence their future career choices (Peel et al. 2018). Core surgical trainee (CST) experience There are currently 1,284 CSTs in the UK (Health Education England 2019). The initial impact of COVID-19 on CSTs was the immediate suspension of all surgical rotations, which resulted in trainees remaining in their current hospital and surgical speciality. This meant that some CSTs did not rotate into T&O and will have missed out on an important training opportunity. During the height of the pandemic, CSTs who remained within T&O were also redeployed to other specialities. This resulted in a significant reduction in surgical exposure and some CSTs may not meet the current minimum requirements to complete core training. At our institution, over a 3-month period (March 23–June 23, 2020), CSTs reported a 76% reduction in logbook numbers compared with the same timeframe in 2019. Speciality registrars (StR) experience There are currently 1,158 StRs enrolled in the UK T&O training programme (Joint Committee on Surgical Training 2019). In many institutions, an emergency COVID rota was introduced and StRs became “first on call” for emergency referrals and looking after inpatients. In some regions, StRs were also redeployed to intensive care units. With the introduction of the national lockdown, there was a dramatic reduction in the number of trauma cases (Hampton et al. 2020). This, coupled with advice encouraging non-surgical management of trauma, has resulted in a significant reduction in trauma operating (British Orthopaedic Association 2020a). Furthermore, where cases have required surgery, there has been an emphasis towards consultant-led operating in order to minimise operative time. All these factors have resulted in a substantial reduction in surgical exposure to orthopaedic trauma. At our institution, over a 3-month period (March 23– June 23, 2020), StRs have reported a 90% reduction in numbers of trauma case operations compared with the same timeframe in 2019. Similar reduction in trainee operating volume has also been reported in other parts of the world during the current pandemic (Jones et al. 2020). As in most parts of the world, elective orthopaedic surgery in the UK almost completely stopped. Consequently, trainees have not had any elective surgical exposure since the middle of March 2020. A significant component of the orthopaedic curriculum is focused on achieving competencies through

Acta Orthopaedica 2020; 91 (6): 627–632

the use of the online Intercollegiate Surgical Curriculum Programme (ISCP) and electronic surgical logbook. There are a number of indicative procedures required for the Certificate of Completion of Surgical Training (CCST): a minimum requirement of 1,800 cases with 70% of these as the primary surgeon. As a result of reduced surgical exposure as outlined above, many trainees may fail to achieve the minimum requirements to progress to the next stage of their training. Fellowship experience In an era of increasing subspecialisation, fellowship has become an essential component of orthopaedic education and a prerequisite to application for consultant posts. It is estimated that approximately 90–95% of UK StRs will undertake a specialist fellowship after award of CCST (Ruddell et al. 2018). Many current UK fellows have been redeployed to their home deaneries to support the local COVID-19 response. For those who still remain in their posts, both in the UK and abroad, surgical exposure has reduced dramatically due to suspension of elective services. Consequently, many fellows may feel they have not gained enough experience to transition to consultant grade and may decide to do an additional fellowship to gain more exposure and further develop their subspecialty skills. Many trainees who were due to commence fellowships have had to delay the start of their training until elective orthopaedic services resume. For others, even though units were keen to receive them, trainees were apprehensive about starting in hospitals given the constant fear of sickness or redeployment whilst living far from home. For those hoping to start fellowships abroad, the closure of international borders has created additional uncertainty. Finally, COVID has struck at the time of year when most fellows are either finalising their future employment or have already committed to consultant jobs. The uncertainty surrounding future services as a result of the pandemic has placed additional anxiety and apprehension on this cohort of trainees. The future of orthopaedic training post-COVID The impact of COVID-19 on orthopaedic training in the UK will be felt for months and possibly years to come. This will necessitate adaptability and restructuring of training programmes to allow trainees access to the learning opportunities that they require to progress through their surgical career. Some of these changes may need to be temporary to ensure current trainees progress through their training schemes, whilst other changes will need to be long term and in preparedness for a possible second wave of COVID-19 infections. From recruitment to completion of training and beyond, educational modalities are likely to change with positive and negative effects associated with all aspects. Training structure At the time of writing (July 7, 2020), the first peak of virus infections was diminishing and it remains to be seen whether

Acta Orthopaedica 2020; 91 (6): 627–632

further peaks will necessitate alternative plans. Currently, the British Orthopaedic Association (BOA), following guidance from NHS England, is devising plans to allow elective services to resume (British Orthopaedic Association. 2020b). In order to ensure patient and staff safety, it will be essential to develop COVID-free “green” pathways to deliver elective orthopaedic services (British Orthopaedic Association 2020b). This may lead to the separation of trauma training from elective orthopaedic training to minimise cross-contamination. It is difficult to predict whether this change is short term, requiring readjustment of the normal process, or is long term, necessitating the need for restructuring of the training programme. This will have implications for the management of the on-call system as those involved in planned care will not be able to be part of the on-call rota. Trainees will have to ensure that their trauma skills do not decline during elective periods and further reassessment prior to CCST may be required to ensure day one consultants remain able to deliver a safe orthopaedic trauma service. Speciality recruitment and Annual Review of Competency Progression In the UK, the traditional national recruitment process for orthopaedic surgeons takes place every April; this year it was cancelled due to the COVID-19 pandemic. A contingency plan was put in place that did not include a formal interview process. Instead, national training numbers were allocated using a unvalidated self-assessment score, which ranked the trainees. This process was poorly received by applicants. With social distancing measures likely to stay in place for the foreseeable future, it is possible that virtual interviews may become the norm. A streamlined virtual interview process using a centralized virtual interview platform or the use of a virtual interview as a screening tool for face-to-face interviews would be possible methods by which recruitment could proceed in the future (Jones and Abdelfattah 2020). Furthermore, an assessment of how implementation of videoconferencing technology in the orthopaedic matching scheme may affect the selection process will need to be undertaken from both a programmatic and candidate standpoint (Jones and Abdelfattah 2020). It may still be possible to keep the current format of the recruitment process and assess prospective candidates via a virtual interview process using online methods, using this as an opportunity to review and improve the interview process for future trainees. Such platforms have been shown to be successful for surgical training programmes in the USA during the current pandemic (Day et al. 2020). CSTs may lack experience and skill with virtual interviews given that this is not a widely used method of assessing prospective trainees; teaching of such skills may therefore need to be incorporated into future training programmes. Some institutions have offered online interviews for recruitment into other specialities and have shared their experience on how trainees can best show their interpersonal skills during this process (Jones and Abdelfattah 2020).

629

The Annual Review of Competency Progression (ARCP) process is the means by which training CSTs and StRs are reviewed each year to ensure that they are offering safe, quality patient care, and to assess their progression against standards set down in the curriculum for their training programme. The ARCP process is conducted through face-to-face meetings with a panel consisting of the training programme director and at least 2 other surgeons. Although these could recommence with appropriate social distancing measures in place, it is likely that these will become virtual assessments in the future. The quality of the ARCP process, whether performed as a face-to-face meeting or virtual assessment, is unlikely to be affected as trainees will still have an opportunity to raise any issues and receive feedback from the panel. Clinics With the introduction of social distancing measures there has been a massive expansion in the use of telemedicine and video-assisted consultations during the COVID-19 crisis. Numerous medical schools in the USA have incorporated telemedicine training into the preclinical undergraduate curriculum (Waseh and Dicker 2019). The continued use of telemedicine in the UK is likely to become part of normal clinical practice in the future, especially for patients who cannot easily attend face-to-face consultations or patients deemed to be at high risk from COVID-19. Orthopaedic trainees will need to receive formal training to use telemedicine in a professional, safe, and evidence-based manner. Telemedicine competencies may need to be built into existing components of the orthopaedic curriculum. Physical examination of patients in clinics forms a crucial part of training. With the increased use of telemedicine, modified interactive orthopaedic examination techniques have been developed and trainees will need to become familiar with such virtual examinations (Tanaka et al. 2020). Despite this a thorough clinical examination will not be possible. Faceto-face consultation is fundamental in every doctor–patient relationship and it remains to be seen whether telemedicine will equally allow such trusting relationships to be built. In the future, virtual clinics may be useful as an initial screening consultation during which a pertinent clinical history can be taken, thus minimising the length of face-to-face consultations. In addition, remote clinics may also have a place in the assessment of postoperative patients where they have been shown to be a safe and a cost-effective means of delivering patient care (Parisien et al. 2020, Buvik et al. 2019, Goldstein et al. 2019, Tanaka et al. 2020). Surgical skills In the short term, it is likely that there will be fewer cases being done on each operating list and hence there will be a reduction in operative training opportunities. In the longer term, with an increased backlog of cases, it is plausible there may be more operating lists and therefore training opportunities

630

could improve. In order to optimise surgical skills, simulation may play an important role in preparing a trainee for performing an operation for the first time, as well as being deployed alongside operating to practise newly acquired skills. The use of simulation training has been shown to reduce intraoperative complications in other surgical specialities (Staropoli et al. 2018). In T&O, it has been demonstrated to be effective in skill acquisition for shoulder and knee arthroscopy (Aim et al. 2016, Bartlett et al. 2018) as well as wrist fracture reduction (Jackson et al. 2020). Training on a virtual simulator can be realistic and the ability to perform surgical steps countless times without additional cost per attempt makes this an attractive training tool. Another advantage of simulation training is that skills can be measured and evaluated in a standardized manner. Virtual reality training has been shown to improve technical skills in orthopaedic surgery and may need to be integrated into the orthopaedic curriculum in the UK, following in the steps of other surgical specialities (Aim et al. 2016, Staropoli et al. 2018). The COVID-19 pandemic may act as a catalyst for the use of augmented reality (AR) and immersive simulation within orthopaedic surgery. Although AR is a field in its infancy, it has been shown to have a wide variety of applications in both elective and trauma surgery. AR enables hands-free real-time access to operating room resources, helping promote telemedicine and education (Laverdiere et al. 2019). It has been used as a platform that allows a surgeon to deliver virtual assistance remotely to another surgeon by layering a live video of their hands reaching into the local surgeon’s operative field in real time, to provide complex visual instructions within shoulder arthroscopy (Ponce et al. 2014). Its use could be particularly valuable for surgical education during times of limited operating combined with the benefit of reducing the number of surgeons in the operating theatre and hence reducing virus transmission risk. Immersive simulation has also been shown to improve translational technical and non-technical skill acquisition over traditional learning in orthopaedic residents in cadaveric shoulder surgery (Lohre et al. 2020). Examinations The Fellowship of the Royal College of Surgeons (FRCS) examination undertaken by trainees in the UK and Ireland has served as a benchmark for the standard required of a Day 1 Consultant Orthopaedic Surgeon in a District General Hospital. Currently, all examinations have been suspended pending a review of capacity for delivery of these examinations effectively and safely during the pandemic. The FRCS (Tr &Orth) examination encompasses 2 sections: section 1 is the written exam and section 2 the clinical exam. Section 1 of the examination is already conducted remotely using computer-based tests held at local driving test centres throughout the UK and Ireland. This part of the exam could easily be reintroduced as long as appropriate social distancing measures are implemented at testing centres. Other Royal Col-

Acta Orthopaedica 2020; 91 (6): 627–632

leges are moving towards a new method of online proctored exam delivery where the written component of the examination can take place in the candidate’s own home or workplace (Royal College of Opthalmologists 2020). Section 2 of the examination consists of the clinical short and long cases and structured oral interviews. Reintroduction of the clinical section will prove difficult to deliver as candidates, examiners, and patients will need to wear personal protective equipment (PPE) during the process. Clinical examinations are usually held in hospitals and it would be unethical to ask patients to attend institutions where they may be at a higher risk of exposure to COVID-19. In the short term, with appropriate planning, the venue could be changed to a hotel or conference centre with a smaller number of candidates sitting the examination to minimise risk to the patient. The use of virtual patients has previously been explored as a method for creating high-fidelity simulated patient interactions that can overcome many of the challenges associated with using live standardised patients (Triola et al. 2006). The virtual patient has the advantage of being easily modified to demonstrate a variety of clinical scenarios and has successfully been used to assess clinical competencies (Botezatu et al. 2010). The oral component of the examination consists of four 30-minute patient-based scenarios. With appropriate planning and validation, it may be possible to conduct this part of the assessment remotely. Similar types of oral examinations have previously been assessed using telemedicine technology while giving formative feedback in a way that is financially feasible for the organisers and well received by candidates (Hannon et al. 2020). Trauma meetings Most orthopaedic departments across the UK hold a morning trauma meeting, reviewing acute trauma admissions to discuss their management. With the introduction of social distancing measures, many orthopaedic departments have implemented a remote “tele-conference” system with a screenshot broadcast of presented radiographs (Pearce et al. 2020). This avoids the potential risk of transmission within the orthopaedic workforce as well as facilitating access to the trauma meeting for those in self-isolation or those who have been advised to shield. With social distancing measures likely to stay in place for the foreseeable future, the use of virtual trauma meetings is likely to continue and may even become part of normal orthopaedic practice in the post-COVID era. Teaching, courses, and conferences Most orthopaedic rotations hold weekly regional teaching designed to cover the syllabus for all trauma and orthopaedic topics to FRCS level. Currently, teaching in most rotations has stopped and will most likely resume in the form of virtual teaching with the use of tele-conferencing. The use of online platforms to deliver speciality teaching has been shown to be well received by trainees, and can result in high attendance

Acta Orthopaedica 2020; 91 (6): 627–632

rates and improved success in postgraduate examinations (MacDonald et al. 2013, Smith et al. 2013). Orthopaedic journal clubs form a crucial part of orthopaedic training and are important in educating trainees in the skills of critically appraising scientific papers. As with regional teaching, journal clubs are likely to move to an online platform in the future. Virtual journal clubs can have many advantages including improved trainee participation as well as maintenance of an electronic record of the discussion threads, the cases discussed, and the papers flagged for future review, which can easily be accessed by all trainees at a later date (Palan et al. 2012). With trainees being able to attend virtually, this means less time away from family and a step in the right direction in improving a surgeon’s work/life balance with no detrimental effects to educational benefits. Attendance at courses is an important component of orthopaedic training and is one of the mandatory elements assessed at the final ARCP (Joint Committee on Surgical Training. 2017). At present, all courses have been cancelled and there has been a rapid expansion of web-based learning resources with increased availability of online lectures and webinars (Plancher et al. 2020). Even before the COVID crisis, webbased teaching and video-based surgical learning had gained popularity among trainees and had been shown to be an effective education tool (Rogers et al. 2019). Such means of learning are likely to become an essential method of delivering orthopaedic education even after the current pandemic has been resolved. Attending regional and national orthopaedic conferences also forms an important part of orthopaedic training. Even after the post-COVID era, organisers of these meetings may feel pressure to change to virtual platforms. Web-based meetings are becoming increasingly popular as in-person meetings can be costly, time-intensive, and involve time away from home (Moran et al. 2018). Some emerging platforms have interactive components such as chat and messaging whereby attendants can interact actively with presenters in a similar manner to the traditional question-and-answer periods. Until we find better ways of networking virtually, trainees will not have the benefit of this type of mentorship without in-person meetings. Conclusion The COVD-19 pandemic is an evolving crisis that has had a profound impact on global healthcare systems. Along with many in society, the impact on orthopaedic training will have far-reaching implications, hence the term the “COVID Generation”. As an immediate response to the pandemic, this cohort of trainees have had to adapt quickly to using novel methods of delivering orthopaedic services. It will be important to observe the extent to which the rapid changes will impact the personal health, safety, and career progression of current trainees. Looking to the future, this pandemic has provided a unique opportunity for educators to change the way surgical

631

training has traditionally been delivered and to help shape the future of medical education. There are numerous examples (e.g. HIV, wars, terrorist attacks) whereby learning from challenging situations has helped transform science and healthcare. Isaac Newton revolutionised the scientific world while in isolation during the plague. Benjamin Franklin once said, “Out of adversity comes opportunity”. All medical trainees now have the opportunity to help make seminal changes in healthcare, and for orthopaedic trainees this is their time to contribute to the advancement of their specialty. Funding and conflicts of interest No funding was received in connection with this article. The authors have no conflicts of interest to declare.

RD, CM, and RW designed the research. RD, CM, LL, KBB, and RW wrote the first draft of the manuscript. RD and CM further edited and prepared the final manuscript. All authors contributed substantially to the contents of the article. Acta thanks Magnus K Karlsson, Laurie Anne Hiemstra, and Caroline Hing for help with peer review of this study.

Aim F, Lonjon G, Hannouche D, Nizard R. Effectiveness of virtual reality training in orthopaedic surgery. Arthroscopy 2016; 32 (1): 224-32. Bartlett J D, Lawrence J E, Stewart M E, Nakano N, Khanduja V. Does virtual reality simulation have a role in training trauma and orthopaedic surgeons? Bone Joint J 2018; 100-B (5): 559-65. Botezatu M, Hult H, Tessma M K, Fors U G. Virtual patient simulation for learning and assessment: superior results in comparison with regular course exams. Med Teach 2010; 32 (10): 845-50. British Orthopaedic Association. Management of patients with urgent orthopaedic conditions and trauma during the coronavirus pandemic. London: BOA; 2020a. British Orthopaedic Association. Re-starting non-urgent trauma and orthopaedic care: Summary guidance. London: BOA; 2020b. Buvik A, Bugge E, Knutsen G, Smabrekke A, Wilsgaard T. Patient reported outcomes with remote orthopaedic consultations by telemedicine: a randomised controlled trial. J Telemed Telecare 2019; 25 (8): 451-9. Day R W, Taylor B M, Bednarski B, Tzeng C D, Gershenwald J E, Lee J E, Grubbs E G. Virtual interviews for surgical training program applicants during COVID-19: lessons learned and recommendations. Ann Surg 2020; May 20: 10.1097/SLA.0000000000004064. Goldstein Y, Schermann H, Dolkart O, Kazum E, Rabin A, Maman E, Chechik O. Video examination via the smartphone: a reliable tool for shoulder function assessment using the constant score. J Orthop Sci 2019; 24 (5): 812-6. Hampton M, Clark M, Baxter I, Stevens R, Flatt E, Murray J, Wembridge K. The effects of a UK lockdown on orthopaedic trauma admissions and surgical cases. Bone & Joint Open 2020; 1 (5): 137-43. Hannon P, Lappe K, Griffin C, Roussel D. Objective structured clinical examination: from exam room to zoom breakout room. Med Educ 2020; May 16. doi: 10.1111/medu.14241. Online ahead of print. Health Education England. Specialty recruitment competition ratios 2019. Leeds: HEA; 2020. p. 4. Jackson T J, Shah A S, Buczek M J, Lawrence J T R. Simulation training of orthopaedic residents for distal radius fracture reductions improves radiographic outcomes. J Pediatr Orthop 2020; 40 (1): e6-e13.

632

Joint Committee on Surgical Training. Certification guidelines for trauma and orthopaedic surgery. JCST; 2017. Joint Committee on Surgical Training. Trauma and orthopaedics. JCST; 2019. Jones R E, Abdelfattah K R. Virtual interviews in the era of COVID-19: a primer for applicants. J Surg Educ 2020; 77(4): 733-4. Jones S D, Thon S, Frank R M. Shoulder and elbow education during COVID19: perspectives from the resident, fellow, and attending level. J Shoulder Elbow Surg 2020; 29 (7): 1297-9. Laverdiere C, Corban J, Khoury J, Ge S M, Schupbach J, Harvey E J, Reindl R, Martineau P A. Augmented reality in orthopaedics: a systematic review and a window on future possibilities. Bone Joint J 2019; 101-B (12): 1479-88. Lohre R, Bois A J, Athwal G S, Goel D P, Canadian Shoulder and Elbow Society (CSES). Improved complex skill acquisition by immersive virtual reality training: a randomized controlled trial. J Bone Joint Surg Am 2020; 102 (6): e26. MacDonald E, Nicol L, Ahmed I. Providing a specialty trainee teaching programme over long distances. Clin Teach 2013; 10 (6): 374-8. Moran J, Briscoe G, Peglow S. Current technology in advancing medical education: perspectives for learning and providing care. Acad Psychiatry 2018; 42 (6): 796-9. Palan J, Roberts V, Bloch B, Kulkarni A, Bhowal B, Dias J. The use of a virtual learning environment in promoting virtual journal clubs and casebased discussions in trauma and orthopaedic postgraduate medical education: the Leicester experience. J Bone Joint Surg Br 2012; 94 (9): 1170-5. Parisien R L, Shin M, Constant M, Saltzman B M, Li X, Levine W N, Trofa D P. Telehealth utilization in response to the novel coronavirus (COVID-19) pandemic in orthopaedic surgery. J Am Acad Orthop Surg 2020; 28 (11): e487-92. Pearce O, Yeomans D, Kelly M. The impact of a global pandemic on orthopaedic services: The collaborative approach in a UK major trauma centre. Transient Journal of Trauma, Orthopaedics and the Coronavirus 2020; April 6. Peel J K, Schlachta C M, Alkhamesi N A. A systematic review of the factors affecting choice of surgery as a career. Can J Surg 2018; 61 (1): 58-67.

Acta Orthopaedica 2020; 91 (6): 627â&#x20AC;&#x201C;632

Plancher K D, Shanmugam J P, Petterson S C. The changing face of orthopedic education: searching for the new reality after COVID-19. Arthrosc Sports Med Rehabil 2020; April 27 [online]. Ponce B A, Jennings J K, Clay T B, May M B, Huisingh C, Sheppard E D. Telementoring: use of augmented reality in orthopaedic education: AAOS exhibit selection. J Bone Joint Surg Am 2014; 96 (10): e84. Rogers M J, Zeidan M, Flinders Z S, Presson A P, Burks R. Educational resource utilization by current orthopaedic surgical residents: a nationwide survey. J Am Acad Orthop Surg Glob Res Rev 2019; 3 (4): e041. Royal College of Opthalmologists. Move to online proctored exams for FRCOphth part 1 and part 2 written examinations. London: Royal College of Ophthalmologists; 2020. Ruddell J H, Eltorai A E M, DePasse J M, Kuris E O, Gil J A, Cho D K, Paxton E S, Green A, Daniels A H. Trends in the orthopaedic surgery subspecialty fellowship match: assessment of 2010 to 2017 applicant and program data. J Bone Joint Surg Am 2018; 100 (21): e139. Smith P J, Wigmore S J, Paisley A, Lamb P, Richards J M, Robson A J, Revie E, McKeown D, Dewhurst D, Garden O J. Distance learning improves attainment of professional milestones in the early years of surgical training. Ann Surg 2013; 258(5): 838-42; discussion 842-3. Staropoli P C, Gregori N Z, Junk A K, Galor A, Goldhardt R, Goldhagen B E, Shi W, Feuer W. Surgical simulation training reduces intraoperative cataract surgery complications among residents. Simul Healthc 2018; 13 (1): 11-15. Tanaka M J, Oh L S, Martin S D, Berkson E M. Telemedicine in the era of COVID-19: the virtual orthopaedic examination. J Bone Joint Surg Am 2020; Jun 17;102(12):e57. doi: 10.2106/JBJS.20.00609. The UK Foundation Programme. 2019 recruitment stats and facts report. 2019; 2020: 36 Triola M, Feldman H, Kalet A L, Zabar S, Kachur E K, Gillespie C, Anderson M, Griesser C, Lipkin M. A randomized trial of teaching clinical skills using virtual and live standardized patients. J Gen Intern Med 2006; 21 (5): 424-9. Waseh S, Dicker A P. Telemedicine training in undergraduate medical education: mixed-methods review. JMIR Med Educ 2019; 5 (1): e12515.

Acta Orthopaedica 2020; 91 (6): 633–638

633

Impact of the COVID-19 pandemic on paediatric orthopaedic trauma workload in central London: a multi-centre longitudinal observational study over the “golden weeks” The COVid Emergency Related Trauma and orthopaedics (COVERT) Collaborative Kapil SUGAND 1a, Chang PARK 1a, Catrin MORGAN 2, Rory DYKE 1, Arash AFRAMIAN 2, Alison HULME 2, Stuart EVANS 2, Khaled M SARRAF 1, and the COVERT Collaborative 1 1 Imperial College Healthcare a shared first authorship

NHS Trust, London; 2 Chelsea and Westminster Hospital, London, UK

COVERT Collaborative collaborators: Camilla Baker 2, Katharine Bennett-Brown 2, Henry Simon 2, Edward Bray 2, Lily Li 2, Noel Lee 2, Nadia Pakroo 2, Kashed Rahman 2, and Andrew Harrison 1 Correspondence: ks704@ic.ac.uk Submitted 2020-06-28. Accepted 2020-07-25.

Background and purpose — The COVID-19 pandemic has been recognised as an unprecedented global health crisis. This study assesses the impact on a large acute paediatric hospital service in London, evaluating the trends in the acute paediatric orthopaedic trauma referral caseload and operative casemix before (2019) and during (2020) COVID-19 lockdown. Patients and methods — A longitudinal retrospective observational prevalence study of both acute paediatric orthopaedic trauma referrals and operative caseload was performed for the first 6 “golden weeks” of lockdown. These data were compared with the same period in 2019. Statistical analyses included median (± median absolute deviation), risk and odds ratios as well as Fisher’s exact test to calculate the statistical significance, set at p ≤ 0.05. Results — Acute paediatric trauma referrals in 2020 were reduced by two-thirds compared with 2019 (n = 302 vs. 97) with a halving risk (RR 0.55) and odds ratios (OR 0.43) of sporting-related mechanism of injuries (p = 0.002). There was a greater use of outpatient telemedicine in the COVID-19 period with more Virtual Fracture Clinic use (OR 97, RR 84, p < 0.001), and fewer patients being seen for consultation and followed up face to face (OR 0.55, RR 0.05, p < 0.001). Interpretation — The impact of the COVID-19 pandemic has led to a decline in the number of acute paediatric trauma referrals, admissions, and operations during the COVID period. There has also been a significant change in the patient pathway with more being reviewed via the means of telemedicine to reduce the risk of COVID-19 transmission and exposure. More work is required to observe for similar trends nationwide and globally as the pandemic has permanently affected the entire healthcare infrastructure.

The novel coronavirus SARS-COV-2 (COVID-19) will be documented as one of the most unprecedented pandemics within modern history. It was first reported in December 2019 with the first patient hospitalised in the city of Wuhan, China (Wu et al. 2020). Declared a pandemic and a global public health emergency by the World Health Organization (2020), as of July 25, 2020 there are over 15 million reported cases and over half a million mortalities. British response to the pandemic The English government responded by implementing stringent social distancing and lockdown measures by mid-March. As of March 23, 2020, all members of the public were required to stay at home unless for essential purposes (UK Government 2020a, b). In response to the National Health Service (NHS) emergency declaration (NHS England 2020a, b), a collaborative effort between NHS England, NHS Improvement, the Royal Colleges, the British Orthopaedic Association (BOA), and the British Society for Children’s Orthopaedic Surgery (BSCOS) resulted in a compilation of guidelines to standardise and protocolise care on a national level during the COVID-19 outbreak (British Orthopaedic Association 2020). Burden of paediatric trauma Trauma is a major cause of mortality and morbidity within the paediatric population in the face of a concurrent decline in paediatric trauma services that has been recognised globally (Danseco et al. 2000, Spady et al. 2004, Peden et al 2008, Tuason et al. 2009). Recent studies of the UK population have found fracture rates in children to be between 76 and 137 per 10,000 person years (Orton et al. 2014, Moon et al. 2016). Although there is complete suspension of elective adult

634

care, semi-elective and urgent elective paediatric care are still deemed to be essential during the pandemic. The unique consequence of the COVID-19 pandemic has meant that schools and nurseries have had to stay closed as a result of the lockdown from March and children, like adults, have had to adjust to life in the constraints of their own homes. Objective This study aimed to evaluate the impact of the COVID-19 pandemic in one of the largest central London multi-centre paediatric hospital services, evaluating the trends in acute orthopaedic trauma referral caseload and operative casemix before (2019) and during the COVID-19 lockdown (2020) from mid-March to the end of April (i.e., over a period of 6 weeks). Alternative hypothesis The alternative hypothesis was that there will be a difference in the prevalence of acute orthopaedic referrals, orthopaedic trauma casemix, and follow-up due to social distancing/lockdown. As an unintended consequence, not being able to leave the household due to social distancing and the lockdown, it was hypothesised that children may be less at risk of accidents and injuries during the COVID-19 pandemic.

Patients and methods Patient sampling All acute referrals, operative notes, inpatient medical records, and discharge summaries were accessed using the electronic medical system within the hospitals as well as from the eTrauma online system (Open Medical Ltd, London, UK), allowing for consistency in data collection. This real-time online database is accessed by those in the trauma and orthopaedic department and records all acute referrals and operations, both retrospectively and prospectively. Hospital sites St Mary’s Hospital is a major trauma centre and equivalent to a level 1 trauma centre, managing the entire spectrum of orthopaedic and non-orthopaedic polytrauma. Chelsea and Westminster Hospital is a large district general hospital equivalent to a level 2 trauma unit, with a dedicated orthopaedic and plastics unit. Both sites have dedicated paediatric orthopaedic departments led by consultant/attending surgeons. Together, they both represent specialist tertiary centres for paediatric orthopaedic trauma and elective surgery. Study period The study period was from the start of the government-imposed social distancing on the morning of March 17, 2020 (which includes the more stringent “lockdown” from the morning of March 24) to the April 28, 2020. This was compared with the

Acta Orthopaedica 2020; 91 (6): 633–638

same 6-week interval in 2019 when there were no such restrictions on social interaction. Inclusion criteria (for both years) All acute orthopaedic trauma cases presenting to the Emergency Department that were subsequently referred to the trauma orthopaedic department either to fracture clinic services or to the acute referral team were included, as were all orthopaedic trauma cases that required an operation within the defined study period 1 year apart. Those patients listed for an operation prior to the period of data collection were included in the final analysis. The study adhered to the STROBE guidelines. Exclusion criteria (for both years) We excluded any patient aged over 18 years old. For operative trauma cases, those undergoing spinal procedures were excluded as the service is delivered jointly by the neurosurgery team. Any non-urgent semi-elective procedures were excluded from further analysis as this would inaccurately assess the impact of any social distancing measures on trauma workloads. Routine elective orthopaedic cases were excluded because this practice was suspended. Data points Demographics including age, sex, and ASA grades were recorded for all patients. Injury characteristics were recorded, including the anatomical location and whether the injury was open or closed. The mechanism of injury was categorised and whether or not the patient was referred as a trauma call. The nature of the operative procedure and the anaesthetic technique were recorded. Patients undergoing multiple procedures were recorded for every episode when they were taken to theatre. Statistics All data were anonymised as well as being verified by 2 authors for accuracy. A Shapiro–Wilk test was conducted for normality, which indicated that the data ought to be treated non-parametrically. The median (± median absolute deviation) was calculated for both age and ASA grade. A Mann– Whitney U-test calculated for significance for continuous data. Both risk (or prevalence) and odds ratios were calculated for discreet datasets as well as a Fisher’s exact test for statistical significance, defined as p ≤ 0.05. Ethics, funding, and potential conflicts of interest There were no ethical objections. Both centres gave permission for the use of the data. This study was assessed using the UKRI / MRC / NHS Health Research Authority Ethics Decision Tool and was considered an ‘audit / not research’; and therefore it was not subject to further ethical review by the NHS Research Ethics Committee (NHS REC). All patient details remained anonymised. All information was stored on NHS encrypted servers. This study received no funding. The authors have no conflict of interests to declare.

Acta Orthopaedica 2020; 91 (6): 633–638

635

Table 1. Results from the acute referrals and requirement for surgery between both time intervals pre-COVID 2019 and COVID 2020. Values are n (%) unless otherwise specified Pre-COVID COVID N = 302 N = 97 Demographic Male 183 (61) Female 119 (39) Median age (MAD) 10 (3) Median ASA (MAD) 1 (0) Injury Upper limb 201 (67) Lower limb 89 (29) Pelvis 3 (1) Infection 6 (2) Other 3 (1) Mechanism of injury Sporting 114 (38) Fall 141 (47) Fall from height >1.5m 8 (3) Road traffic accidents 4 (1) Pathological 0 (0) Other 21 (7) Laceration 0 (0) Infection 6 (2) Trampoline 5 (2) N/A 0 (0) Open injury 4 (1) Trauma call 3 (1) Admission 33 (11) Operative management 48 (16) Safeguarding 25 (8) Comorbidity ADHD 0 (0) Asthma 7 (2) Osteogenesis imperfecta 1 (0) None 276 (91) Autism 2 (1) Eczema 1 (0) Leukaemia 1 (0) Epilepsy 2 (1) G6PD deficiency 1 (0) Sickle cell 1 (0) Type 1 diabetes mellitus 3 (1) Tuberculosis 1 (0) WPW syndrome 1 (0) Febrile convulsions 2 (1) Migraine 1 (0) Thalassaemia 1 (0) COVID status Negative Positive Not tested Clinic outcome Discharge 164 (54) Referral 2 (1) N/A 103 (34) Follow-up 23 (8) DNA 10 (3) Clinic pathway Hybrid – face to face + virtual clinics 0 (0) Discharged from face to face clinic 2 (1) Discharged via virtual clinic 0 (0) Face to face clinic 290 (96) N/A 9 (3) Referral to virtual clinic 0 (0) Referral to hand therapy 1 (0) Number of follow-ups to date 2 Days to discharge 0 Days to first orthopaedic review 7

53 (55) 44 (45) 7 (4) 1 (0)

Table 2. Risk, prevalence and odds ratios and (95% CI) between post- versus pre-COVID era Factor

p-value 0.3 0.02 0.8

a Fisher’s

p-value a

exact test

0.002

Results A comparison between the two cohorts has been tabulated (Table 1). Prevalence, risk, and odds ratios Table 2 outlines the prevalence and odds ratios alongside their 95% confidence intervals and statistical significance. The risk ratio is synonymous with the prevalence ratio. Only the statistically significant ratios were included. There was a significant reduction in the odds of sporting-related mechanism of injuries (by 57%) in 2020 compared with 2019. Although the prevalence and odds ratios had significantly increased for follow-up during the COVID era, the odds of face-to-face consultations had a significant drop of 95%. The follow-up appointments could now be conducted safely using telecommunications such as a virtual fracture clinic, which increased prevalence 59-fold. Consequently, there was also a strong trend to discharge patients safely from the virtual fracture clinic with an odds ratio of 224.

1 (1) 3 (3) 1 (1) 90 (93) 1 (1) 1 (1) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1 (1) 2 (2) 94 (97) 59 (61) 2 (2) 9 (9) 22 (23) 5 (5)

OR (CI)

Sporting related injury 0.55 (0.36–0.83) 0.43 (0.25–0.74) 0.002 Follow-up 3.1 (1.8–5.3) 3.6 (1.9–6.7) < 0.001 Clinic pathway < 0.001 Face to face consultation 0.55 (0.41–0.62) 0.05 (0.02–0.08) Virtual fracture clinic 59 (3.5–1,000) 65 (3.7–1,127) Discharge from virtual fracture clinic 164 (10–2,664) 224 (14–3,723)

67 (69) 23 (24) 1 (1) 6 (6) 0 (0) 20 (21) 50 (52) 6 (6) 2 (2) 2 (2) 8 (8) 1 (1) 4 (4) 3 (3) 1 (1) 4 (4) 1 (1) 18 (19) 16 (16) 11 (11)

RR or PR (CI)

0.001

Comment on alternative hypothesis The alternative hypothesis was not rejected. During the COVID period there was a reduced number of acute orthopaedic referrals consisting of younger patients (p = 0.02) presenting with half odds of sporting-related injuries (p = 0.002). Patients were at greater odds of being followed-up (p < 0.001), while face-to-face consultations halved in favour of the virtual fracture clinic with the capability of safe discharge (p < 0.001).

4 (4) 2 (2) 26 (27) 47 (48) 10 (10) 9 (9) 0 (0) 1 0 0

< 0.001 < 0.001 < 0.001 < 0.001 0.7 < 0.001

Discussion Shift in referrals The data supported the alternative hypothesis, which demonstrated that in the “golden weeks” following the governmental order to socially distance and isolate, there has been a notable reduction in paediatric injuries compared with last year. There

636

were only 97 referrals following the introduction of social distancing measures in 2020 compared with 302 in 2019. This represents a 68% reduction in paediatric injuries. This reduction is likely a direct consequence of the social distancing measures implemented on a national scale. As the aetiology of paediatric fractures is linked to physical activity, often outdoors, the lockdown has significantly reduced the incidence of paediatric injuries (Schalamon et al. 2011 , Joeris et al. 2014). There may also be a shift in patient behaviour, with parents being more anxious about attending hospital due to the risk to themselves and their child of contracting COVID-19. Those who would have previously promptly sought acute services may have delayed and attempted to treat injuries at home. It is possible a number of these injuries or symptoms may have self-resolved without the need for urgent medical attention. This may represent an explanation for the reduction in the number of acute orthopaedic presentations to the Emergency Department seen in 2020. Demographic and mechanism The general demographic of those presenting with injuries changed between the 2 periods, with a significantly younger median age (p = 0.02) in 2020 and more girls (p = 0.3). The younger cohort during the pandemic may reflect a less riskaverse population sample leading to trauma, or more concerned parents with a lower threshold of presenting to ED for reassurance, especially with a lower ability of the child to communicate his/her symptoms. There was no statistically significant difference in the comorbidities in those being referred, with the same median ASA grade of 1. The pattern of injury also remained generally unchanged with upper limb injuries being the most common at 67% and 69% respectively in 2019 and 2020. There was a significant decline in sportingrelated injuries by 17% in 2020 (p = 0.002, RR = 0.55, OR = 0.43). This is unsurprising as following government guidelines all sports were banned during the lockdown and, even with the relative easing in mid-May, team sports remained forbidden. Falls remained the most common mechanism of injury in 2020, and these injuries can have occurred at home as well as during the single episode of exercise that the government allowed during lockdown. Non-accidental injuries (NAI) NAI should always be considered in the paediatric population and the principles have not changed during the pandemic; healthcare workers must remain vigilant during times of increased stress and social isolation (NHS England 2020c). Indeed, with families under lockdown there have been reports of increased rates of domestic violence in households during the pandemic and this raises the concern for more NAI during the COVID-19 period (Alradhawi et al. 2020). In 2020 there were 3% more safeguarding referrals. Whilst there was no statically significant difference between the two periods, with school systems closed the usual pathways for the detection

Acta Orthopaedica 2020; 91 (6): 633â&#x20AC;&#x201C;638

of the unusual and NAI are no longer in place. Nevertheless, there has not been a large increase in delayed presentation of substantial injuries that would prompt such a concern due to the effects of the pandemic. We are aware that delayed presentation is always a possibility that may have not been captured in the set time period for this study. Operative intervention In 2020 there was an approximate 50% reduction in inpatient admissions and two-thirds reduction in those undergoing operative intervention. However, due to the overall reduction in referrals, the percentage of those admitted or undergoing operative intervention remained relatively unchanged. Against the backdrop of such a significant reduction in new acute trauma referrals, there is still a small percentage of substantial injuries that require admission and operative intervention in spite of social distancing and lockdown measures. The BOA guidance published at the beginning of the pandemic recommended increased emphasis on the non-operative management of paediatric injuries during the COVID19 pandemic (British Orthopaedic Association 2020). With the paediatric remodelling potential and remedial options in the form of later corrective surgery (e.g., for malunion), the guidelines recommended changes that deviated from standard practice for managing paediatric injuries. Although the new risks of COVID-19 were considered for the management of all patients, ultimately the decision to operate was based on the best interests of the child. In our practice no patient requiring operative intervention was denied this on the basis of COVID19 alone for an acute issue. This was also enabled due to the overall reduction in referral numbers and operative cases, suggesting that a scenario did not occur whereby limited operative capacity was overwhelmed such that surgical treatment was deferred. Infection Children represent approximately 2% of all COVID-19 cases but to reduce the risk of increased spread schools and nurseries closed during this period (Docherty et al. 2020). Subsequently, there is an expectation of a reduction in acute referrals for infection-related pathology in the paediatric population. The current literature suggests that children appear to be less susceptible to the effects of COVID-19, often displaying milder symptoms than their adult counterparts (Dong et al. 2020, Henry et al. 2020, Lu and Shi 2020, Ludvigsson 2020). Nevertheless, there have been a number of documented cases of a more serious associated illness similar to that of Kawasaki vasculitis (Harahsheh et al. 2020, Viner and Whittaker. 2020). The concern in the paediatric population is that COVID-19 may cause an increased incidence in the presentation of transient synovitis, or indeed COVID-related septic arthritis and osteomyelitis. In our study there has been an increase in overall referrals for infection, from 2% in 2019 to 6% in 2020. As the pandemic evolves and more information surfaces, further

Acta Orthopaedica 2020; 91 (6): 633–638

investigation is required to observe the influence of COVID19 on paediatric musculoskeletal infections. Although we have found an increase in those being referred for infection, this was statistically insignificant. However, a larger and broader population sample may suggest otherwise. There were 2 patients who tested positive for COVID-19 via nasopharyngeal PCR swabs. Both patients were admitted and treated surgically for septic arthritis requiring operative washout of the hip and knee. The joint aspirate in the washout of the knee subsequently grew Neisseria meningitis and Staphylococcus aureus from the hip but no COVID-19 PCR was identified from the operative tissue sample. To our knowledge there has not been any literature investigating the link between COVID-19 and septic arthritis in children, but the fact that 2 COVID-positive paediatric patients in 2020 had washout of a major joint for septic arthritis certainly warrants further investigation. Patient pathway The COVID-19 pandemic has meant that the pathway for managing paediatric patients has changed to minimise risk to both patient and family (British Orthopaedic Association 2020). The threshold for paediatric follow-up tends to be lower due to children being a vulnerable cohort without self-advocacy and with limited communication. Monitoring paediatric patients is an acceptable reason for continuing follow-up. The COVID19 pandemic challenged this policy as most children infected with COVID-19 seem to develop a sub-clinical condition or mount a milder immune response and illness, with the exception of a few. Hence, they are more likely to act as vectors to contribute to viral transmission (Dong et al. 2020, Henry et al. 2020, Lu and Shi 2020, Ludvigsson 2020). Against the backdrop of the pandemic, there was a significant increase in the prevalence and odds of follow-up despite these concerns. However, the justification for this increase was the introduction of virtual fracture clinic (VFC). The VFC can be conducted by surgeons remotely, whereby they review the clinical referral from the Emergency Department while assessing the electronic imaging. The decision tree consists of either follow-up or discharge. Patients can be seen in a face-to-face clinic to either review their progress, arrange for further imaging, or commence treatment. The other option is to discharge the patient either to physiotherapy/hand therapy/ orthotics with patient information leaflets that can be posted or emailed to them. It is due to its availability, accessibility, and affordability that the utilisation of telemedicine has been a key adaptation in the care of patients during the pandemic (Loeb et al. 2020, Tanaka et al. 2020). VFC can also help to avoid the “did not attend” (DNA) situation that was seen in 2019. Consequently, there was a substantial rise in the prevalence and odds of referring and discharging paediatric patients safely from VFC follow-up. Conversely, face-to-face consultation, as expected, was also almost halved with regard to prevalence but the odds

637

of seeing the paediatric patient in the follow-up clinic were reduced by 95%. This was not only statistically significant but also clinically significant as reducing face-to-face consultations was vital in curbing viral transmission, especially in a vulnerable cohort. This was also combined with the number of follow-ups halving (p < 0.001). Furthermore, more urgent reviews were also implemented with a swifter first orthopaedic review of 1-week to same-day review (p < 0.001). This is in keeping with the recommendations made by the BOA (2020). Due to the restructuring of services, same-day physiotherapy and hand therapy were offered to patients either face to face or via video consultations. There is a concern regarding increased NAI during the lockdown, which may go unnoticed, especially with increasing utilisation of virtual consultations to reduce face-to-face follow-up. However, the normal pathway of safeguarding at the point of contact in the ED still applies, but clinicians should remain vigilant with a low threshold of appropriate suspicion as non-verbal and physical clues may be overlooked during virtual consultations. One option is to move from telephonic consultation with the parents to a video-format to examine children, their injuries, and their behaviour visually. Limitations and future work This study, albeit multi-centre in a hotspot of the pandemic, is still not representative of the national picture. Further work is required to look at the national trend. Furthermore, some schools have just started to reopen with the government planning on all children returning to school in September. There needs to be vigilant monitoring to look for signs of a second wave and resurfacing of the pandemic. A future study as we enter a period of gradual relaxation of lockdown measures will provide more data on the continued impact of the pandemic as well as the wider adoption of virtual consultations and active exclusion of delayed presentations of non-accidental injuries. Conclusion This study represents the early experience of the pandemic at one of the largest central London multi-centre paediatric hospital services. The prevalence of paediatric referrals reduced by nearly two-thirds during the COVID-19 period, with a significant reduction in those being referred for sporting injuries. The COVID-19 pandemic has also changed the pathway for paediatric referrals with greater utilisation of telemedicine in the form of virtual fracture clinics, allowing for safer ongoing follow-up without risking contamination, by reducing face-toface consultation. These changes may have been triggered by the pandemic due to the necessity of reducing transmission rates, but these may represent novel methods of delivering care to this unique population in the future. The number of patients requiring operative intervention reduced by two-thirds during the pandemic but an identical 16% of those referred required operative intervention in both 2019 and 2020. Although guidelines at the onset of the pan-

638

demic suggested a shift to non-operative treatment of injuries that may previously have required operative intervention, children must be treated based on their clinical needs and the resources available. Treatment must also continue to support and address any concerns the patient or the family may have regarding COVID-19. Neither hospital compromised on providing paediatric patients with evidence-based management for injuries, including surgery within an acceptable timeframe. All clinical needs were met within an acceptable timeframe and there was no need to consider later surgery for possible malunions. Acta thanks Hanne Hedin for help with peer review of this study.

Alradhawi M, Shubber N, Sheppard J, Ali Y. Effects of the COVID-19 pandemic on mental well-being amongst individuals in society: a letter to the editor on “The socio-economic implications of the coronavirus and COVID-19 pandemic: A review”. Int J Surg 2020; 78: 147-8. British Orthopaedic Association (BOA). British Orthopaedic Association Standards for Trauma and Orthopaedics (BOAST). Management of patients with urgent orthopaedic conditions and trauma during the coronavirus pandemic 21 Apr 2020. Available from https://www.boa.ac.uk/ resources/covid-19-boasts-combined.html Danseco E R, Miller T R, Spicer R S. Incidence and costs of 1987–1994 childhood injuries: demographic breakdowns. Pediatrics 2000; 105: E27. Docherty A B, Harrison E M, Green C A. Features of 16,749 hospitalised UK patients with COVID-19 using the ISARIC WHO Clinical Characterisation Protocol. medRxiv, April 28, 2020. Dong Y, Mo X, Hu Y. Epidemiological characteristics of 2143 pediatric patients with 2019 coronavirus disease in China. Pediatrics 2020. doi: 10.1542/peds.2020-0702. Harahsheh A S, Dahdah N, Newburger J W, Portman MA, Piram M, Tulloh R, McCrindle B W, de Ferranti S D, Cimaz R, Truong D T, Burns J C. Missed or delayed diagnosis of Kawasaki disease during the 2019 novel coronavirus disease (COVID-19) pandemic. J Pediatr 2020; 222: 261-2. Henry B M, Lippi G, Plebani M. Laboratory abnormalities in children with novel coronavirus disease 2019. Clin Chem Lab Med 2020; 58(7): 1135-8. Joeris A, Lutz N, Wicki B, Slongo T, Audigé L. An epidemiological evaluation of pediatric long bone fractures: a retrospective cohort study of 2716 patients from two Swiss tertiary pediatric hospitals. BMC Pediatr 2014; 14: 314. NHS England. Next steps on NHS response to COVID-19: Letter from Sir Simon Stevens and Amanda Pritchard. March 17, 2020a. Available from www.england.nhs.uk/coronavirus/publication/next-steps-on-nhs-responseto-covid-19-letter-from-simon-stevens-and-amanda-pritchard NHS England. Redeploying your secondary care medical workforce safely. March 26, 2020b. Available from www.england.nhs.uk/coronavirus/publication/redeploying-your-secondary-care-medical-workforce-safely/

Acta Orthopaedica 2020; 91 (6): 633–638

NHS England. ICON. Specialty guides for patient management during the coronavirus pandemic. April 2, 2020c. Version 1. Available from https:// www.england.nhs.uk/coronavirus/wp-content/uploads/sites/52/2020/04/ C0097-Specialty-guides-and-coronavirus_ICON-cribsheet-for-midwivesduring-COVID_v1-02-April-2020.pdf Loeb A E, Rao S S, Ficke J R, Morris C D, Riley H 3rd, Levin A S. Departmental experience and lessons learned with accelerated introduction of telemedicine during the COVID-19 crisis. J Am Acad Orthop Surg 2020; 28(11): e469-e476. Lu Q, Shi Y. Coronavirus disease (COVID-19) and neonate: what neonatologists need to know. J Med Virol 2020; 10.1002/jmv.25740. Ludvigsson J F. Systematic review of COVID-19 in children shows milder cases and a better prognosis than adults. Acta Paediatr 2020; 109(6): 1088-95. Moon R J, Harvey N C, Curtis E M, de Vries F, van Staa T, Cooper C. Ethnic and geographic variations in the epidemiology of childhood fractures in the United Kingdom. Bone 2016; 85: 9-14. Orton E, Kendrick D, West J, Tata L J. Persistence of health inequalities in childhood injury in the UK: a population-based cohort study of children under 5. PLoS One 2014; 9(10): e111631. Peden M, Oyegbite K, Ozanne-Smith J, Hyder A, Branche C, Rahman A K M F, Rivara F, Bartolomeos K, editors. World report on child injury prevention. Geneva: World Health Organization; 2008. Schalamon J, Dampf S, Singer G, Ainoedhofer H, Petnehazy T, Hoellwarth M E, Saxena A K. Evaluation of fractures in children and adolescents in a Level I trauma center in Austria. J Trauma 2011; 71(2): E19-E25. Spady D W, Saunders D L, Schopflocher D P, Svenson L W. Patterns of injury in children: a population-based approach. Pediatrics 2004; 113(3 Pt 1): 522-9. Tanaka M J, Oh L S, Martin S D, Berkson E M. Telemedicine in the Era of COVID-19: the virtual orthopaedic examination. J Bone Joint Surg Am 2020; ;102(12): e57. Tuason D, Hohl J B, Levicoff E, Ward W T. Urban pediatric orthopaedic surgical practice audit: implications for the future of this subspecialty. J Bone Joint Surg Am 2009; 91(12): 2992-8. UK Government. Oral statement to Parliament. Controlling the spread of COVID-19: Health Secretary’s statement to Parliament. March 16, 2020a. Available from www.gov.uk/government/speeches/controlling-the-spreadof-covid-19-health-secretarys-statement-to-parliament UK Government. Guidance: Staying at home and away from others (social distancing). March 23, 2020b. Available from https://www.gov.uk/government/publications/full-guidance-on-staying-at-home-and-away-fromothers Viner R M, Whittaker E. Kawasaki-like disease: emerging complication during the COVID-19 pandemic. Lancet 2020; 395(10239): 1741-3. World Health Organization. Coronavirus disease 2019 (COVID-19): situation report, 186. July 24, 2020. Available from https://www.who.int/ docs/default-source/coronaviruse/situation-reports/20200724-covid-19-sitrep-186.pdf?sfvrsn=4da7b586_2 Wu F, Zhao S, Yu B, Chen Y M, Wang W, Song Z G, et al. A new coronavirus associated with human respiratory disease in China. Nature 2020; 579: 265-9.

Acta Orthopaedica 2020; 91 (6): 639–643

639

Delayed surgery versus nonoperative treatment for hip fractures in post-COVID-19 arena: a retrospective study of 145 patients Bobin MI 1,a, Lang CHEN 1,a, Dake TONG 3,a, Adriana C PANAYI 2, Fang JI 4, Junfei GUO 5, Zhiyong HOU 5, Yingze ZHANG 5, Yuan XIONG 1, and Guohui LIU 1 1 Department

of Orthopedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China; 2 Department of Plastic Surgery, Brigham and Women’s Hospital, Harvard Medical College, Boston, USA; 3 Department of Orthopedics, Shanghai Ninth People’s Hospital, Shanghai Jiaotong University School of Medicine, Shanghai, China; 4 Department of Orthopedics, Changhai Hospital, Shanghai, China; 5 Department of Orthopaedic Surgery, Third Hospital of Hebei Medical University, Shijiazhuang, China a Shared first authorship Correspondence: liuguohui@hust.edu.cn Submitted 2020-05-14. Accepted 2020-08-17.

Background and purpose — Following the outbreak of COVID-19 in December 2019, in China, many hip fracture patients were unable to gain timely admission and surgery. We assessed whether delayed surgery improves hip joint function and reduces major complications better than nonoperative therapy. Patients and methods — In this retrospective observational study, we collected data from 24 different hospitals from January 1, 2020, to July 20, 2020. 145 patients with hip fractures aged 65 years or older were eligible. Clinical data was extracted from electronic medical records. The primary outcomes were visual analogue scale (VAS) score and Harris Hip Score. Major complications, including deep venous thrombosis (DVT) and pneumonia within 1 month and 3 months, were collected for further analysis. Results — Of the 145 hip fracture patients 108 (median age 72; 70 females) received delayed surgery and 37 (median age 74; 20 females) received nonoperative therapy. The median time from hip fracture injury to surgery was 33 days (IQR 24–48) in the delayed surgery group. Hypertension, in about half of the patients in both groups, and cerebral infarction, in around a quarter of patients in both groups, were the most common comorbidities. Both VAS score and Harris Hip Score were superior in the delayed surgery group. At the 3-month follow-up, the median VAS score was 1 in the delayed surgery group and 2.5 in the nonoperative group (p < 0.001). Also, the percentage of complications was higher in the nonoperative group (p = 0.004 for DVT, p < 0.001 for pulmonary infection). Interpretation — In hip fracture patients, delayed surgery compared with nonoperative therapy significantly improved hip function and reduced various major complications.

The incidence of hip fractures is on the rise, owing in part to the aging population, with the number of annual hip fractures worldwide expected to exceed 6 million by 2050 (Papadimitriou et al. 2017). Nonoperative treatment of hip fractures requires long-term recovery, which increases the risk of complications such as pulmonary infection, pressure ulcers, urinary tract infection, and lower limb venous thrombosis, consequently leading to high mortality (Sinvani et al. 2020). Surgical treatment, hence, tends to be the preferable treatment for hip fracture patients. In terms of surgery, operation promptness is crucial for better clinical outcomes in the elderly (Ekeloef et al. 2019). Delayed surgery may occur for various reasons, including delayed admission, or surgery in high-risk patients on certain medications or with comorbidities. Folowing the outbreak of COVID-19 in December 2019, in China, and particularly in the city of Wuhan, many hip fracture patients were unable to gain timely admission and surgery. Many factors contributed to this delay, including shortage of medical resources preventing hospitals from treating non-COVID-19 patients. In addition, fear of the high infection and mortality of SARSCoV-2 deterred many patients from attending hospitals. As China is now in the post-outbreak period, many hip fracture patients are beginning to crowd into hospitals seeking surgical treatment. In this retrospective study, we compared the outcomes of delayed surgery and nonoperative therapy in hip fracture patients. The clinical characteristics and surgical outcomes, including complications, were recorded.

640

Acta Orthopaedica 2020; 91 (6): 639–643

Hip fracture patients January 1 – July 20, 2020 n = 175

Patients and methods Study design and participants This was a multicenter retrospective clinical study of delayed surgery versus nonoperative therapy in hip fracture patients during the COVID-19 pandemic. The study included 24 cohorts of adult inpatients (> 65 years old) from Wuhan Union Hospital, the Third Hospital of Hebei Medical University, Zhengzhou Orthopedic Hospital, Zhengzhou Central Hospital, Renhe Hospital of China Three Gorges University, Shanghai Changhai Hospital, People’s Hospital of Xiangzhou District, Xiangyang Central Hospital, People’s Hospital of Xiantao City, Qianjiang Hospital of Traditional Chinese Medicine, Taihe Hospital of Shiyan City, Nanchang Hospital of Traditional Chinese Medicine, Jiangxia People’s Hospital, Tianmen People’s Hospital, Shanghai Tenth People’s Hospital, People’s Hospital of Wuhan University, Wuhan Tongji Hospital, People’s Hospital of Shiyan City, Wuhan General Hospital of Guangzhou Military Command, Guangshui Hospital of Traditional Chinese Medicine, Guangshui Second People’s Hospital, People’s Hospital of Chibi City, Dongxi Lake People’s Hospital, and Badong People’s Hospital. The analyzed period spanned January 1, 2020 to July 20, 2020. All patients were diagnosed with hip fracture by orthopedic doctors on the basis of radiographies or CT scans. Surgeries were postponed due to the COVID-19 pandemic. Patients were included if they met the following criteria: at least 65 years old, admitted between January 1, 2020 and July 20, 2020, had a hip fracture that had not undergone surgery for more than 21 days. Patients with prior hip operation history were excluded. COVID-19 was diagnosed in accordance with the COVID-19 was diagnosed in accordance with the New Coronavirus Pneumonia Prevention and Control Program (6th edition) published by the National Health Commission of China (NHCC). Data collection Using a customized data collection form, we tabulated clinical characteristics, comorbidities, prior history, treatment, laboratory test results, and function scores, as provided by the 24 hospitals. All inpatients and family members received pulmonary CT, SARS-CoV-2 nucleic acid testing, and antibody testing prior to admission. Laboratory test results are the final results obtained prior to discharge. VAS score and Harris hip function score were graded by an orthopedic expert panel after 1 month and 3 months’ follow-up. Statistics Primary data were analyzed with SPSS software (version 23.0; IBM Corp, Armonk, NY, USA). Categorical variables were presented as frequency and percentages, while median and interquartile rage (IQR) values were used to describe continuous variables. A chi-square test was used to compare categorical variables between delayed surgery and nonoperative

Excluded (n = 30): – hip fracture within 21 days, 15 – incomplete information, 12 – less than 65 years old, 2 – had hip operation history, 1 Eligible hip fracture patients n = 145 Surgery group n = 108

Nonoperative therapy group n = 37

Figure 1. Flow chart of study design.

therapy; Fisher’s exact test was used when data were limited. A Kolmogorov–Smirnov test was performed to test whether continuous variables complied with normal distribution. An independent group t-test was applied to normally distributed continuous variables, otherwise the Mann–Whitney U-test was used. We defined excellent, good, not bad, and poor as Harris score 86–100, 71–85, 56–70, and 0–55 respectively, and the specific number and percentage in each group were calculated. P-values < 0.05 were considered statistically significant. Endpoints Due to the urgency and importance of the COVID-19 pandemic, we collected and analyzed the follow-up results of 145 elderly hip fracture patients as of July 20, 2020. Among them, the range of day from admission to the endpoint was 99 to 187 days. Kaplan–Meier curves were constructed to compare surgery and nonoperative therapy. 1-month and 3-month followup results were based on the outpatient clinical system and direct contact with patients or their family member. Ethics, funding, and potential conflicts of interest The study was approved by the Institutional Review Board of Union Hospital, Tongji Medical College, Huazhong University of Science and Technology and the requirement for consent was waived because of the retrospective analysis (IRB no. 2020-01-27). This study was supported by the National Key Research & Development Program of China (Grant Nos. 2018YFB2001502 and 2018YFB1105705), National Science Foundation of China (Grant No. 81772345), the National Health Commission of the People’s Republic of China (Grant Nos. ZX-01-018 and ZX-01-C2016153), and the Health Commission of Hubei Province (Grant No. WJ2019Z009). The authors report no conflicts of interest.

Results 175 patients were initially identified (Figure 1). Finally, 145 patients were included, with 108 patients undergoing surgery

Acta Orthopaedica 2020; 91 (6): 639–643

641

Table 1. Demographics, clinical characteristics, and treatment of elderly hip fracture patients during the COVID-19 pandemic Factor

All Nonoperative patients Surgery therapy n = 145 n = 108 n = 37 p-value

Demographics Female sex Age, mean range Chronic comorbidity Hypertension Diabetes Cerebral infarction Coronary heart disease Osteoporosis Prior medical history Smoking Alcohol Other clinical characteristics Simple fall Traffic accident Fall from height Sprain Femoral fracture type Neck Intertrochanteric Subtrochanteric Brain injury Treatment Traction Analgesics Antibiotics

90 72 66–81

70 72 65–79

20 74 66–85

0.2 0.2

81 29 40 23 10

62 18 29 15 9

19 11 11 8 1

0.5 0.09 0.7 0.3 0.4

20 16

15 12

5 4

1.0 1.0

118 15 6 6

88 11 4 5

30 4 2 1

1.0 1.0 1.0 1.0

105 36 4 5

88 17 3 4

17 19 1 1

< 0.001 < 0.001 1.0 1.0

41 116 68

32 86 56

9 30 12

0.5 0.8 0.04

and the remaining 37 patients receiving nonoperative treatment (Table 1). No patients were diagnosed with COVID19 infection. Most patients were female. Hypertension and cerebral infarction were the most common comorbidities. We noted no statistically significant difference between the 2 groups in terms of comorbidities as well as prior medical history (smoking). The most common cause of fracture was simple fall (118, 81%) and most common type of fracture was femoral neck fracture (104, 72%). Only 5 patients suffered

brain injury. Femoral neck fractures were more common in the surgery group (p < 0.001) and intertrochanteric fractures were more common in the nonoperative therapy group (p < 0.001). There was no statistically significant difference in traction treatment (p = 0.5) and analgesic treatment (p = 0.9) between the 2 groups. Antibiotic treatment was more common in the surgery group (p = 0.04) (Table 1). Surgery, which included plate internal fixation, dynamic hip screw, proximal femoral nail antirotation, hip replacement, and hollow screw internal fixation, was selected based on type of fracture. Days from injury to surgery, surgery time, and intraoperative blood loss were noted (Table 2). Prior to discharge, the laboratory test results showed that the nonoperative therapy group displayed lower lymphocyte counts, lower hemoglobin, higher level of CRP, and higher level of D-dimer (Table 3, see Supplementary data). At the 1-month follow-up, the surgery group showed a lower incidence of deep venous thrombosis and pulmonary infection. In terms of functional outcome, the surgery group had a lower VAS score (p < 0.001) and a higher Harris Hip Score (p = 0.04), showing a better early outcome than the nonoperative therapy group (Table 4). At the 3-month follow-up, 9 patients from the surgery group and 3 patients from the nonoperative therapy group did not come to hospital for review or were unable to be contacted. Among the 133 remaining follow-up patients, 13/99 patients in the surgery group suffered from deep venous thrombosis, and 12/34 patients in the nonoperative therapy group (p = 0.004, Table 5). Pulmonary infection occurred in 15/99 patients in the surgery group and 16/34 patients in the nonoperative therapy group (p < 0.001). Patients in the surgery group showed a lower VAS score and better recovery of hip function than patients in the nonoperative group (p < 0.001 for VAS score; p = 0.04 for Harris Hip Score). Only 1 (1%) patient died in the surgery group, while 4 patients (12%) died in the nonoperative therapy group after 3-month follow-up (p = 0.02) (Table 6, Figure 2).

Table 2. Intraoperative data of 108 surgical patients Proximal Plate Dynamic antirotation Hip All fixation hip screw nail replacement n = 108 n = 12 n = 6 n = 15 n = 68 Femoral fracture typ, n neck intertrochanteric subtrochanteric Days from injury to surgery IQR Surgery time, minutes IQR Intraoperative blood loss, mL IQR

Discussion Hollow screw fixation n=7

88 7 5 2 67 7 17 5 0 11 1 0 3 0 1 2 0 0 33 32 45 29 37 23 24–48 25–58 26–71 22–44 25–49 21–38 90 133 95 90 80 90 60–120 60–173 80–110 60–135 60–120 40–120 200 150 225 200 200 50 100–300 100–400 118–425 100–400 120–300 30–100

This study is the first to compare the outcomes of delayed surgery and nonoperative therapy for patients with hip fractures in the post-COVID-19 arena. This study includes 108 patients with delayed surgery and 37 patients with nonoperative therapy. Prior research has reported that the longer the delay in surgery, the higher the risk of poor outcomes and complications (Neuman et al. 2014). Most hip fracture patients

642

Acta Orthopaedica 2020; 91 (6): 639–643

Table 4. 1-month follow-up results of 145 elderly hip fracture patients during the pandemic of COVID-19 Items

All Nonoperative patients Surgery therapy n = 145 n = 108 n = 37 p-value

Deep venous thrombosis, n Pulmonary infection, n VAS score (0–10) IQR Harris Hip Score (0–100) IQR

20 24 3.0 2.0–4.0 69 57–82

11 9 0.03 14 10 0.03 3.0 5.0 < 0.001 1.3–4.0 3.0–6.5 71 66 0.04 60–84 55–82

Table 5. 3-month follow-up results of 133 elderly hip fracture patients during the COVID-19 pandemic Items

All Nonoperative patients Surgery therapy n = 133 n = 99 n = 34 p-value a

Deep venous thrombosis, n Pulmonary infection, n VAS score (0–10) IQR Harris Hip Score (0–100) IQR Dead

25 31 1.0 0.0–3.0 71 63–85 5

13 12 0.004 15 16 < 0.001 1.0 2.5 < 0.001 0.0–2.0 1.0–5.0 73 70 0.04 66–86 58–84 1 4 0.02

a Mann–Whitney

U-test or chi-square test was selected to compare differences between surgery and conservative therapy where appropriate.

are elderly with pulmonary dysfunction, meaning that pulmonary infection is more likely to occur when lying in bed (Sasabuchi et al. 2018). It is reported that as high as 6% of patients experience pulmonary infection during nonoperative management or delayed surgery (Goh et al. 2020). Therefore, prompt surgical repair of hip fractures is recommended. However, it is difficult to ensure all patients receive timely surgery due to various factors. Normal medical activity was disturbed by the outbreak of COVID-19. Although surgeons aimed to treat patients promptly, the limited medical resources resulted in surgical postponements. Such patients with hip fractures eventually crowded into hospitals post-COVID-19, while

Figure 2. Kaplan–Meier curves of survival. Gehan–Breslow–Wilcoxon test p-value = 0.003; hazard ratio, surgery versus nonoperative therapy group = 0.06 (95% confidence interval 0.01–0.38).

others chose nonoperative therapy. In our study, 13% of pulmonary infections were noted in the delayed surgery group, and 27% were noted in the nonoperative therapy group after 1-month follow-up. However, the percentage of pulmonary infection increased to 15% and 47% in the surgery group and nonoperative therapy group respectively after 3-month follow-up. Thus, although delayed surgery is not the optimal choice, it showed a decreased occurrence of pulmonary infection, possibly due to early mobilization and standing postsurgery. Deep venous thrombosis is a common complication in patients with hip fractures. Although all patients are recommended prophylactic treatment including massage, muscle equal-length contraction, and passive joint movement, the odds of deep venous thrombosis (DVT) remain high. Previous studies have reported that the prevalence of preoperative DVT is 8% in hip fracture patients (Shin et al. 2016). In our study, 11 cases (10%) of DVT were seen in the surgery group and 9 cases (24%) of DVT were seen in the nonoperative therapy group after 1-month follow-up; the percentages were increased to 13% in the surgery group and 35% in the nonoperative therapy group after 3-month follow-up. Various reasons contribute to higher DVT rates, including aging, smoking, and surgery. Our results indicate that elevated D-dimer

Table 6. Demographics, clinical characteristics, and causes of death of 6 dead elderly hip fracture patients during the COVID-19 pandemic Hyper- Dia- Cerebral Patient Therapy Sex Age tension betes infarction Other chronic comorbidity Cause of death 1 surgical female 84 + + + 2 nonoperative male 80 – – + 3 nonoperative male 87 – – – 4 nonoperative female 71 + – + 5 nonoperative male 73 + + + 6 nonoperative male 85 + + –

Chronic obstructive pulmonary disease Pleural effusion, arrhythmia, atrial fibrillation Pulmonary infection, pulmonary tuberculosis, pulmonary embolism Coronary heart disease, arrhythmia Coronary heart disease Renal failure

Pulmonary infection Heart failure Pulmonary embolism Heart failure Heart failure Renal failure

Days from admission to death 78 18 27 69 88 169

Acta Orthopaedica 2020; 91 (6): 639–643

is 1 of the risk factors for venous thrombosis, consistent with previous studies (Xing et al. 2018). Thus, antithrombotic drugs were recommended for the patients in both the delayed surgery group and the nonoperative therapy group, especially for patients with elevated D-dimer levels (Forster and Stewart 2016). Treatments should be considered to avoid pulmonary embolus once deep venous thrombosis is noted (Spandorfer and Galanis 2015). This study is limited by its small size and our conclusions may suffer from sample bias. With the extensive development of medical technology, delays in surgical treatment of hip fractures are rare. The COVID-19 outbreak has created a unique worldwide medical situation that has necessitated the delay of elective procedures. Our data strongly suggests that delayed surgery is superior to nonoperative therapy for hip fracture patients. With COVID-19 now under control in China, surgeons are now refocusing on delayed elective patients. As we did not collect data on physicians, possible variations in physician skill level may affect outcomes, especially in terms of complications. Although the 1-month mortality rate has been quoted to range from 3% to 24% post-surgery treatment and 31% post-conservative treatment (Prommik et al. 2019), no differences in death rates were observed between the two groups in our study after 1-month follow-up. However, 1 patient (1%) in the surgery group and 4 patients (12%) in the nonoperative therapy group were dead after 3-month followup. We report only on the short-term prognosis of hip fracture. The outcomes as regards malunion, non-union, infection, and mortality rate will be recorded in a future study. No data was collected on the cost for hip fracture patients. Requirement for nursing and rehabilitation care when staying in a nursing home, and various post-surgical complications including DVT, may increase the cost of treatment. In summary, when elective procedures must be postponed, patients with hip fractures are more likely to benefit from delayed surgery rather than nonoperative therapy. Compared with nonoperative therapy, delayed surgery showed a lower likelihood of major complications. Furthermore, it decreased pain and resulted in increased mobilization and better function. Supplementary data Table 3 is available as supplementary data in the online version of this article, http://dx.doi.org/ 10.1080/17453674. 2020.1816617

643

G L and Y Z designed the study; B M, D T, F J, J G and Z H collected clinical data; B M and L C carried out data analyses; B M, Y X and L C wrote the manuscript; A P contributed to revision of the manuscript. Acta thanks Yan Li, Aare Märtson and Tao Yu for help with peer review of this study.

Ekeloef S, Homilius M, Stilling M, Ekeloef P, Koyuncu S, Münster A B, Meyhoff C S, Gundel O, Holst-Knudsen J, Mathiesen O, Gögenur I. The effect of remote ischaemic preconditioning on myocardial injury in emergency hip fracture surgery (PIXIE trial): phase II randomised clinical trial. BMJ 2019; 367: 16395. Forster R, Stewart M. Anticoagulants (extended duration) for prevention of venous thromboembolism following total hip or knee replacement or hip fracture repair. Cochrane Database Syst Rev 2016; 3: CD004179. Goh E L, Lerner R G, Achten J, Parsons N, Griffin X L, Costa P M L. Complications following hip fracture: results from the World Hip Trauma Evaluation cohort study. Injury 2020, 6: 1331-6. Neuman M D, Silber J H, Magaziner J S, Passarella M A, Mehta S, Werner R M. Survival and functional outcomes after hip fracture among nursing home residents. JAMA Intern Med 2014; 174: 1273-80. NHCC. National Health Commission of China. [The State Council’s joint prevention and control mechanism for pneumonia epidemic in response to new coronavirus infection (6th edition).] Chinese. Accessed 2020 Mar 7. http://www. nhc.gov.cn/jkj/s3577/202003/4856d5b0458141fa9f376853224d41d7.shtml Papadimitriou N, Tsilidis K K, Orfanos P, Benetou V, Ntzani E E, Soerjomataram I, Künn-Nelen A, Pettersson-Kymmer U, Eriksson S, Brenner H, Schöttker B, Saum K U, Holleczek B, Grodstein F D, Feskanich D, Orsini N, Wolk A, Bellavia A, Wilsgaard T, Jørgensen L, Boffetta P, Trichopoulos D, Trichopoulou A. Burden of hip fracture using disability-adjusted lifeyears: a pooled analysis of prospective cohorts in the CHANCES consortium. Lancet Public Health 2017; 2: e239-e46. Prommik P, Kolk H, Sarap P, Puuorg E, Harak E, Kukner A, Pääsuke M, Märtson A. Estonian hip fracture data from 2009 to 2017: high rates of non-operative management and high 1-year mortality. Acta Orthop 2019; 90: 159-64. Sasabuchi Y, Matsui H, Lefor A K, Fushimi K, Yasunaga H. Timing of surgery for hip fractures in the elderly: a retrospective cohort study. Injury 2018; 49: 1848-54. Shin W C, Woo S H, Lee S J, Lee J S, Kim C, Suh K T. Preoperative prevalence of and risk factors for venous thromboembolism in patients with a hip fracture: an indirect multidetector CT venography study. J Bone Joint Surg Am 2016; 98: 2089-95. Sinvani L, Goldin M, Roofeh R, Idriss N, Goldman A, Klein Z, Mendelson D A, Carney M T. Implementation of hip fracture co-management program (AGS CoCare: Ortho®) in a large health system. J Am Geriatr Soc 2020; 389: 1519-27. Spandorfer J, Galanis T. Deep venous thrombosis. Ann Intern Med 2015; 162: ITC1. Xing F, Li L, Long Y, Xiang Z. Admission prevalence of deep vein thrombosis in elderly Chinese patients with hip fracture and a new predictor based on risk factors for thrombosis screening. BMC Musculoskelet Disord 2018; 19: 444.

644

Acta Orthopaedica 2020; 91 (6): 644–649

Preparing an orthopedic department for COVID-19 Lessons learned from reorganization and educational activities Rune Dall JENSEN 1,2, Magnus BIE 1, Anne Plønd GUNDSØ 3, Johannes Martin SCHMID 4, Joachim JUELSGAARD 4, Maria Louise GAMBORG 1,5, Hanne MAINZ 3, and Jan Duedal RÖLFING 1,2,3 1 Corporate

HR, MidtSim, Central Denmark Region, Aarhus; 2 Department of Clinical Medicine, Aarhus University, Aarhus; 3 Department of Orthopaedics, Aarhus University Hospital, Aarhus; 4 Department of Respiratory Disease and Allergy, Aarhus University Hospital; 5 Centre for Health Sciences Education, Aarhus University, Denmark Correspondence: jan.roelfing@rm.dk Submitted 2020-06-29. Accepted 2020-08-24.

Background and purpose — The COVID-19 pandemic has disrupted healthcare services around the world. We (1) describe the organizational changes at a level 1 trauma center, (2) investigate how orthopedic healthcare professionals perceived the immense amount of information and educational activities, and (3) make recommendations on how an organization can prepare for disruptive situations such as the COVID-19 pandemic in the future. Methods — We conducted a retrospective survey on the organizational restructuring of the orthopedic department and the learning outcomes of a needs-driven educational program. The educational activities were evaluated by a nonvalidated, 7-item questionnaire. Results — The hospital established 5 COVID-19 clusters, which were planned to be activated in sequential order. The orthopedic ward comprised cluster 4, where orthopedic nursing staff were teamed up with internal medicine physicians, while the orthopedic team were redistributed to manage minor and major injuries in the emergency department (ED). The mean learning outcome of the educational activities was high–very high, i.e., 5.4 (SD 0.7; 7-point Likert scale). Consequently, the staff felt more confident to protect themselves and to treat COVID-19 patients. Interpretation — Using core clinical competencies of the staff, i.e., redistribution of the orthopedic team to the ED, while ED physicians could use their competencies treating COVID-19 patients, may be applicable in other centers. Insitu simulation is an efficient tool to enhance non-technical and technical skills and to facilitate organizational learning in regard to complying with unforeseen changes.

The COVID-19 pandemic has caused an unprecedented destabilization of healthcare services on a global scale. In Denmark, the case fatality rate was 4.8% in June 2020 compared with the global case fatality rate of 5.4% (Dong et al. 2020). As the COVID-19 pandemic spread throughout Europe, authorities started to reorganize inpatient care at hospitals to ensure the healthcare system would be able to treat the anticipated vast numbers of patients suffering from COVID-19. While research into medical treatment for COVID-19 is ongoing and the development of a vaccine still seems to be many months away, the pressure on the healthcare professionals continues to intensify in certain parts of the world. In spring 2020, the potentially overwhelming burden of illnesses stressed the Danish health system’s capacity and healthcare professionals. Hence, the widespread use of recommended personal protection equipment (PPE) in the care of all patients with respiratory symptoms had the highest priority (Adams and Walls 2020). In the orthopedic domain, Liang et al. (2020a, b) highlighted 3 main principles for surgeries during the pandemic; (1) determining the urgency of the surgery and the possibility of postponing it, (2) protecting the health of professionals and patients by providing PPE and rapid testing, and (3) being mindful of conserving healthcare resources (Liang et al. 2020a). Here, Hirschmann et al. (2020) pointed to the importance of being mindful of how to handle aerosol-generating procedures, which is especially prominent during orthopedic surgeries. Furthermore, interdisciplinary collaboration is argued to be essential for ensuring a high-quality information flow (Liang et al. 2020a). At Aarhus University Hospital in Denmark, several learning modalities, e.g., in-situ simulation, e-learning, classroom teaching, etc. were used to effectively disseminate information

Acta Orthopaedica 2020; 91 (6): 644–649

List of participants of the educational activities n = 100

645

Additional participants from the department’s email distribution list n = 47 Excluded (n = 46): – did not respond to eligibility screening question, 1 – did not participate in the educational activities, 45 due to: • maternity leave • sick leave • no longer / not yet employed • temporary attachment only • sent home during COVID-19 (pregnancy / comorbidities)

Participants in total n = 101 Response rate: 96% (n = 97 / 101) All 7 items required an answer i.e., no missing data

Figure 1. Questionnaire evaluating needs-driven educational activities: identification, eligibility of participants and response rate.

and training to healthcare professionals to meet the demands of the COVID-19 disease. The present study (1) describes the organizational changes of a level 1 trauma center in Denmark, (2) investigates how healthcare professionals in the Department of Orthopedics perceived the immense amount of information and training, and (3) makes recommendations on how an organization can prepare for disruptive situations such as the COVID-19 pandemic in the future.

Methods Organizational restructuring We conducted a retrospective survey on the organizational restructuring of the department and the learning outcomes of the needs-driven educational program at a level 1 trauma center (550–700 cases annually) and orthopedic department during COVID-19. The orthopedic department comprises 48 beds, 5 day beds, 300 employees, 3,000 elective orthopedic procedures, 3,200 day surgeries, 1,700 acute orthopedic procedures, and 34,000 outpatient visits annually. Data regarding the organizational restructuring of the hospital and the orthopedic department was obtained from the administration and the healthcare professionals involved. Questionnaire A non-validated, 7-item questionnaire (in Danish is available from the authors) regarding the healthcare professional’s perception of the educational activities and the learning outcome was developed. The questionnaire was distributed electronically to the participants (n = 101, Figure 1) via SurveyXact.

dk (Rambøll Management Consulting, Aarhus, Denmark) allowing for tracking of individual participation/lack of participation but ensuring anonymous data analysis. Answer options for individual questions were displayed in random order for each individual participant. Participants had the opportunity to write comments at the end of the questionnaire. Data was collected from June 9, 2020 until June 18, 2020. Email reminders were sent twice. Furthermore, healthcare professionals reminded their colleagues to participate verbally and through social media.

Statistics The before/after items of the questionnaire were analyzed using 1-way repeated ANOVA with subsequent pairwise comparison of the 3 subitems. Descriptive statistics, i.e., mean (standard deviation [SD]) were applied for the other items of the questionnaire. A p-value of ≤ 0.05 was considered statistically significant. Ethics, funding, data sharing, and potential conflict of interests Participation in the survey was voluntary and no payment was offered. Ethical approval of an anonymous data analysis was not required from the Danish Committee on Health Research Ethics (according to the Danish Ethical Committee Law § 14, stk. 2). The study was approved by the head of the orthopedic department and conducted in full accordance with ethical principles, including the World Medical Association Declaration of Helsinki. If a response rate > 90% was achieved, all 3 departments (orthopedic wards 1 + 2 and the outpatient clinic) were rewarded with candy for the entire staff. The original data, e.g., the Danish questionnaire and all participants’ responses, can be requested from the corresponding author. No external funding was received for this study. The authors declare that they have no conflicts of interest.

Results Organizational changes at the hospital Shocked by the devastating images of military lorries driving COVID-19 casualties to the crematorium in Northern Italy, social-distancing measures and lockdown of Danish society were carried out at an early stage. Aarhus University Hospital shut down all elective surgery and outpatient visits and started to reorganize the entire hospital for COVID-19 testing, triaging, resuscitation, and treatment. A stepwise approach was made in order to prepare inpatient departments to care for admitted COVID-19 patients in 5 newly established COVID-19 clusters. The first COVID-19 cluster was situated at the Department

646

of Infectious Diseases, which started admitting COVID-19 patients at once, while the other departments were reorganized to prepare for this task. The next cluster should start admitting COVID-19 patients when 50% of the capacity of the previous cluster was in use. The COVID-19 clusters were headed by an infectious disease or respiratory medicine physician and the head of nursing from the respective departments. Organizational changes at the Department of Orthopedics The Department of Orthopedics comprised cluster 4. Even at the height of the pandemic, only COVID-19 clusters 1 and 2 were activated, while clusters 3 and 4 used the time to prepare for handling the disease. Respiratory medicine physician and co-author JMS headed cluster 4 in co-leadership with the orthopedic head of nursing. COVID-19 cluster 4 was staffed by orthopedic nursing staff working in their usual workplace, but preparing and taking care of medically ill patients. Physicians in cluster 4 were recruited from the emergency department (ED) and the departments of endocrinology, geriatrics, and respiratory medicine. Orthopedic surgeons Orthopedic surgeons (approximately 70, including residents) were redistributed to take care of minor and major injuries and trauma in the emergency department (ED) with a 2-day warning. This corresponded to 10 full-time employees. Meanwhile, ED physicians were deployed to staff COVID-19 testing, triaging, and COVID-19 clusters. This was a conscious choice by the administration, optimizing the use of core clinical competencies of the orthopedic team and ED physicians, who are more used to treating respiratory compromised patients suffering from multiple comorbidities. Consequently, this change caused minimal disruption and discomfort among orthopedic surgeons. Moreover, the usual orthopedic acute and subacute services provided by 2 consultants (24/7 and 8 am to 9 pm daily) and 1–2 residents (24/7) were still running 2 acute theaters daily instead of the usual 65 operating theatres on a weekly basis. The remaining orthopedic specialists were asked to work from home or take vacation. These changes took place from March 20, 2020 until May 4, 2020. Thereafter, the orthopedic surgeons as well as the ED staff returned to their regular jobs and orthopedic services gradually returned to normal conditions within 4–6 weeks. During this period, the physical space of the ED was reused as a COVID-19 triaging area and the orthopedic outpatient clinic was therefore refurbished to house the ED, including a casting room and option to reposition fractures under fluoroscopy. However, standardized radiographs and CT scans were also available at the Department of Radiology. Orthopedic surgeons are usually in charge of trauma resuscitation at our level 1 trauma center receiving 550–700 trauma cases annually. Neither the organization of the trauma team nor the physical place were changed during COVID-19.

Acta Orthopaedica 2020; 91 (6): 644–649

The educational activities “personal protection equipment (PPE)” and “hand hygiene” were also offered to orthopedic surgeons. However, almost none participated. Guidelines and availability of PPE when treating patients in the ED and when operating on patients with suspected or confirmed COVID19 infection were often changing and resulted in uncertainty and discomfort among some orthopedic surgeons. Finally, no initiatives were made to acquire new surgical skills or prevent skill decay (Hedeman and Felländer-Tsai, 2020). Orthopedic nursing staff Clinical competencies of the orthopedic nursing staff had to be refreshed and improved in order to be able to provide care to medical COVID-19 patients with multiple comorbidities instead of the regular orthopedic inpatients. Multiple needsdriven educational activities were initiated to enable the staff to protect themselves and their families from the infection and to provide care to COVID-19 patients. Educational activities preparing for COVID-19 A steering committee consisting of the head of the COVID19 cluster and 4 administrative clinical nurses responsible for development, IT, and research ensured that the comprehensive flow of new relevant information from the Central Denmark Region to the hospital was reviewed and passed on to the staff through guidelines, newsletters, and daily morning meetings at the department. All questions and concerns from the staff were reviewed and addressed at staff meetings and on a question-and-answer poster. Based on the staff’s concerns and needs, several educational activities were planned, including the correct use of PPE, care for medical COVID-19 patients, etc. (Table 1). Educational activities were thus needs-driven, supplementing the competencies of the staff and addressing their concerns and suggestions. To ensure adaptability of the educational program, the participants conducted ad hoc assessment of the learning and relevance of the activities. These assessments were continuously evaluated by the committee, who discontinued inefficient initiatives and added new activities as requested. 2 specialty nurses were in charge of monitoring and ensuring attendance. Lists of participants in all activities were drawn up, and activities continued to be offered until the entire staff had the opportunity to attend. In order to prevent the spread of COVID-19 among participants, educational activities were offered multiple times per day, in small groups of 2–5 participants, in locations allowing for 1–2 m spacing between the participants. Table 1 displays the educational activities, which were decisive initiatives to prepare the healthcare personnel in the orthopedic department for COVID-19. Questionnaire 97 of 101 participants in the educational activities answered the questionnaire (Figure 1).

Acta Orthopaedica 2020; 91 (6): 644–649

647

Table 1. Needs-driven educational activities offered at the orthopedic department related to the COVID-19 pandemic response in order to prepare for COVID-19, ordered by mean (CI) recommendation on a 7-point Likert scale Faculty, format, Educational activity schedule, and objectives

Sequential Recommend order Participants to colleagues (needs-driven) Duration (max. 101) mean (SD)

Full-scale in situ simulation team training

ABCDE approach to the patient with COVID-19, teamwork and communication, correct use of PPE

Organizational structure of the COVID-19 department

COVID-19 related deaths

60 min

6.8 (0.5)

Multiple courses mixed with simulation

6.4 (0.8)

Use of personal protective equipment (PPE)

Donning and doffing procedures, types of PPE, 1 precautions required regarding contact and aerosols, nursing and medical staff teamwork within the isolation room E.g., distribution of doctors, examinations and tests, interdisciplinary cooperation

30 min

6.4 (0.8)

Arterial blood gas analysis

Arterial puncture analysis and using the machinery correctly, transport and timeframe for analysis; max. 2 participants/instruction

20 min

6.2 (1.2)

Focus on the COVID-19 guidelines for deaths and the subsequent work practices

30 min

6.1 (0.9)

Lung physiotherapy Basic oxygen therapy and airway suctioning

Patient positioning, mobilization, other aspects

30 min

6.1 (0.9)

Practical skills training

10 min

5.9 (1.1)

The elderly patient and medical issues

Educator launched discussion on considerations regarding the weakened elderly patient who has COVID-19

45 min

5.7 (1.1)

Fluid resuscitation and nutrition

Sufficient nutrition and correct use of feeding tubes, fluid restrictions related to COVID-19

30 min

5.6 (1.2)

Delirium—symptoms and causes

Interventions and confusion and assessment methods (CAM)

45 min

5.6 (1.3)

Hand hygiene

Additional focus to avoid contact 1 transmission of COVID-19

E-learning and video

101

N/A

Following nursing staff at the Departments of Endocrinology and of Respiratory Medicine

Peer-to-peer work observation and teaching Discontinued due to inefficiency

7.4 hours

N/A

For chronological order please refer to the column sequential order (needs-driven).

The majority of educational activities received excellent feedback from the participants (Table 1). The self-reported mean learning outcome of all attended educational activities was high–very high, i.e., 5.4 (CI 5.0– 6.0) on a 7-point Likert scale with anchors 1 = extremely low, 2 = very low, 3 = low, 4 = neither/nor, 5 = high, 6 = very high, 7 = extremely high. The participants felt more comfortable after participation in the educational activities in this unprecedented situation (Figure 2). Free text comments of the participants The participants perceived that the COVID-19 situation resulted in ambivalent experiences for healthcare professionals. On one hand, participants perceived the organizational changes as positive (i.e., increase in interdisciplinary actions, decrease of bureaucracy, and a stronger sense of community). On the other hand, participants reported uncertainty, due to the severity of the disease combined with an overwhelming amount of information, as a participant wrote in the questionnaire:

Very comfortable Comfortable

Before After

p < 0.001 p < 0.001

p < 0.001

Slightly comfortable Neither/nor Slightly uncomfortable Uncomfortable Very uncomfortable

Using PPE Avoiding Providing care to correctly to contract COVID-19 COVID-19 at work patients

Figure 2. Mean results (CI) of the question: “Before/after participating in the educational activities, how comfortable did you feel about: (1) using PPE correctly?, (2) avoiding contracting COVID-19 at work?, (3) providing care to COVID-19 patients?

648

“There was a lot of information to process. In addition, guidelines changed rapidly, making it very difficult to feel certain and up to date.” Simulation training was perceived as key in preparing oneself and the department for the COVID-19 disease by de-mystifying the disease, and providing hands-on experiences with the patient group, effectively improving a sense of self-efficacy (Bandura 1994). Stress management especially was experienced as helpful in reducing potential stressors and increasing a sense of comfort in handling the COVID-19 patients, as also noted by participants in the survey stating that: “especially insitu simulations were very fruitful”; “the introduction to treatment of COVID-19 patients provided some sense of security”; “After the [educational] activities, I felt so ready to do my part in fighting COVID-19 and treating patients”.

Discussion In this retrospective survey, we describe the organizational changes for the entire orthopedic department and investigate how 101 healthcare professionals perceived different educational activities during the rapid global and national spread of COVID-19. While healthcare workers accept an increased risk of infection as part of their chosen profession, they often exhibit concern regarding family transmission, especially involving family members who are elderly, immunocompromised, or have chronic medical conditions (Walton et al. 2020). It is well known that stress and lack of commitment has a negative effect on healthcare professional quality of work. In surgery, Flin et al. (2006) showed that surgeons and nurses generally demonstrate positive attitudes to acts that are likely to enhance teamwork and safety in the operating theatre and surgical department (Flin et al. 2006). They found that the majority of nurses, surgeons, and trainees were more likely to make errors in tense and hostile situations. In our study, the participants reported that they experienced uncertainty in the rapid spread of the COVID-19 virus. Furthermore, they reported that the educational activities and in-situ simulation in particular made them feel more comfortable and helped to demystify the COVID-19 disease. This was further confirmed by the participants, who described a stronger sense of individual and interdisciplinary professional self-efficacy, less uncertainty, and lower stress levels after engaging in simulated training sessions. These findings fit well with the revised model of stress by Palmer et al. (2003). Their model consists of 7 hazards for stress including employees’ demands and employees’ support and training. Demand is defined as the exposure to physical hazards and workload including insufficient personnel and complexity of work. COVID-19 is likely to add complexity of work in the orthopedic department. Hence, employees’ support and training seem essential in order to avoid stress and

Acta Orthopaedica 2020; 91 (6): 644–649

skill decay in a pandemic such as the COVID-19 disease (Kelc et al. 2020, Hedeman and Felländer-Tsai, 2020). Infrastructure is a key pillar supporting the fundamental aim of promoting improved standards of care and well-being for all patients, together with a good experience of the healthcare system (Luxon 2015). The data from the present study and data from a related study (focus-group interviews evaluating the efficacy of in-situ simulation during COVID-19, unpublished data) showed that learning occurred at an organizational level as well as an individual level and that an adaptable infrastructure at the hospital seems essential in order to treat the potentially increasing number of COVID-19 patients. A variety of different learning activities helped participants to acquire knowledge of different aspects of receiving and caring for COVID19 patients. In-situ simulation was described as an efficient tool for learning how to deal with stressful situations, enhancing effective communication, and acquisition of specific technical skills. Furthermore, the instructor was mentioned as a key factor by being engaged in the scenario, ensuring fidelity, and securing a safety net by debriefing the simulation. Another general insight from the participants was the impact and importance of interdisciplinary simulation training sessions. It highlighted each healthcare professional’s value and role when handling these patients, but also in general. Furthermore, interdisciplinary learning was facilitated in the simulated settings, in regard to specific workflows and knowledgesharing, which again led to an increased sense of comfort in teamwork and trust in the handling of COVID-19. Simulation has a huge potential to help in managing the global COVID-19 crisis in 2020 and in potentially similar future pandemics (Lababidi et al. 2020). Simulation can rapidly facilitate hospital preparation and education of large numbers of healthcare professionals and has proven its value in many settings (Dieckmann et al. 2020). It can be utilized to scale-up workforce capacity through experiential learning. Simulation and simulation facilitators can also contribute to the optimization of work structures and processes (Brydges et al. 2020). Limitations This study has 3 main limitations. First, a non-validated retrospective questionnaire was utilized exploring healthcare professionals’ experience of a variety of educational activities during the COVID-19 pandemic. Second, recall bias is likely to occur in retrospective surveys, but, due to the rapid development of COVID-19 and immediate introduction of the first learning activities (Table 1), prospective data collection was not feasible (Althubaiti 2016). Third, 1 item in the questionnaire assessed the learning outcome of the activities attended. However, subjective assessment of the learning outcome of an educational intervention is not as accurate as an objective assessment, i.e., an examination or practical test of what was learned (Lababidi et al. 2020). Nonetheless, the educational activities built confidence and may thus have decreased stress

Acta Orthopaedica 2020; 91 (6): 644–649

Table 2. General recommendations for orthopedic departments on how to use competencies and establish needs-driven educational program in order to prepare for unprecedented situations Prepare • Establish a steering committee to facilitate and support continuous learning activities based on relevant guidelines and the staff’s need and feedback. • Prepare 2–3 educational activities with relevance to protecting employees, i.e., in-situ simulation focusing on use of personal protection equipment and hygiene. Facilitate • Conduct interdisciplinary in-situ simulation. • Adapt simulation scenarios to the needs of the given department/ staff. • Conduct ad hoc assessments of educational program to promote needs-driven initiatives. Organize • Establish knowledge-sharing across departments/clusters. • Use competencies and resources wisely. Redistributing orthopedic surgeons to manage minor and major orthopedic injuries in the ED is a minor disruption compared with taking care of medically ill patients suffering from COVID-19. This use of core clinical competencies may be applicable in a large variety of hospitals in Scandinavia and around the world.

and other mental health problems, which healthcare professionals are prone to during the current health crisis (Walton et al. 2020). Recommendations and lessons learned from reorganizing a level 1 trauma center and orthopedic department to prepare and handle COVID-19 are summarized in Table 2. In particular, in-situ simulation received excellent evaluations and may prove useful in preparing staff for COVID-19 and other disruptive events challenging the core clinical competencies. Many healthcare professionals experienced uncertainty due to the rapid development of the COVID-19 disease.

JDR, MB, and RDJ contributed to the study conception and design. Material preparation and data collection were performed by MB, JDR, and AG. Data analysis was performed by JDR and MLG. The first draft of the manuscript was written by RDJ, JDR, JMS, JJ, MLG, and HM. All authors revised the drafts, and read and approved the final manuscript. The authors would like to express their gratitude to the head of the orthopedic department: Birgit Eg Andersen, Sten Larsen, Lone Dich, Hanne Sørensen, and Pernille Didriksen; secretary Thea Isaksen for transcription of the data; Berit Pedersen Haa, Gitte Schmidt, and Kirsten Rasmussen for being part of the steering committee organizing the educational activities and critical revision of the manuscript; the teachers from the orthopedic and other departments; Neel Toxvig and Line Roest Niemann for their contributions and Corporate HR, MidtSim for supporting the in-situ simulations.

649

Acta thanks Karin Bernhoff, Mats Ericson and Lisbet Meurling for help with peer review of this study.

Adams J G, Walls R M. Supporting the health care workforce during the COVID-19 global epidemic. JAMA 2020; 323(15): 1439-40. Althubaiti A. Information bias in health research: definition, pitfalls, and adjustment methods. J Multidiscip Healthc 2016; 9: 211-17. Bandura A. Self-efficacy. In: Ramachaudran VS, editor. In: Ramachaudran VS, editor. Encyclopedia of Hum Behavior. 4th ed. New York: Academic Press; 1994. p 71-81. Brydges R, Campbell D M, Beavers L, Khodadoust N, Iantomasi P, Sampson K, Goffi A, Caparica Santos F N, Petrosoniak A. Lessons learned in preparing for and responding to the early stages of the COVID-19 pandemic: one simulation’s program experience adapting to the new normal. Advances in Simulation 2020; 5: 8. doi: 10.1186/s41077-020-00128-y Dieckmann P, Torgeirsen K, Qvindesland S A, Thomas L, Bushell V, Langli Ersdal H. The use of simulation to prepare and improve responses to infectious disease outbreaks like COVID-19: practical tips and resources from Norway, Denmark, and the UK. Advances in Simulation 2020; 5(1): 1-10. Dong E, Du H, Gardner L. An interactive web-based dashboard to track COVID-19 in real time. Lancet Infect Dis 2020; 20(5): 533-4. Flin R, Yule S, McKenzie L, Paterson-Brown S, Maran N. Attitudes to teamwork and safety in the operating theatre. Surgeon 2006; 4(3): 145-51. Hedeman L R, Felländer-Tsai L. Simulation-based skills training in nonperforming orthopedic surgeons: skills acquisition, motivation, and flow during the COVID-19 pandemic. Acta Orthop 2020; 91(5). [Epub ahead of print] doi.org/ 10.1080/17453674.2020.1781413. Hirschmann M T, Hart A, Henckel J, Sadoghi P, Seil R, Mouton C. COVID-19 coronavirus: recommended personal protective equipment for the orthopaedic and trauma surgeon. Knee Surg Sport Traumatol Arthrosc 2020; 28(6): 1690-8. Kelc R, Vogrin M, Kelc J. Cognitive training for the prevention of skill decay in temporarily non-performing orthopedic surgeons. Acta Orthop 2020; 3674: 1-4. Lababidi H M S, Alzoraigi U, Almarshed A A, Alharbi W, Alamar M, Arab A A, Mukahal M A, Alasmari F A, Mzahim B Y, Alharastani H A M, Alammi S S, Alawad Y I. Simulation-based training programme and preparedness testing for COVID-19 using system integration methodology. BMJ Simul Technol Enhanc Learn 2020; 1-8. Liang Z C, Chong M S Y, Liu G K P, Della Valle A G, Wang D, Lyu X, Chang C-H, Cho T-J, Haas S B, Fisher D, Murphy D, Hui J H P. COVID-19 and elective surgery. Ann Surg 2020a May 20; [Online ahead of print]. doi: 10.1097/SLA.0000000000004091. Liang Z C, Chong M S Y, Sim M A, Lim J L, Castañeda P, Green D W, Fisher D, Lian K, Murphy D, Hui J H P. Surgical considerations in patients with COVID-19. J Bone Joint Surg 2020b; 102(11): e50. doi: 10.2106/ JBJS.20.00513. Luxon L. Infrastructure: the key to healthcare improvement. Futur Hosp J 2015; 2(1): 4-7. Palmer S, Cooper C, Thomas K. Revised model of organisational stress for use within stress prevention/management and wellbeing programmes: brief update. Int J Heal Promot Educ 2003; 41(2): 57-8. Walton M, Murray E, Christian M D. Mental health care for medical staff and affiliated healthcare workers during the COVID-19 pandemic. Eur Hear J Acute Cardiovasc Care 2020; 9(3): 241-7.

650

Acta Orthopaedica 2020; 91 (6): 650–653

Redeployment of the trainee orthopaedic surgeon during COVID-19: a fish out of water? Giles FARIA, Baha John TADROS, Natalie HOLMES, Siddharth VIRANI, Gaddam Kumar REDDY, Baljinder Singh DHINSA, and Jaikumar RELWANI

East Kent Hospitals University NHS Foundation Trust, Ashford, Kent, UK Correspondence: gilesfaria@doctors.org.uk Submitted 2020-07-22. Accepted 2020-08-30.

Background and purpose — COVID-19 has had a significant impact on health services and the entire healthcare sector, including trauma and orthopaedics, has been compelled to adapt. At the heart of this was the redeployment of the orthopaedic trainees to support “frontline specialties”. This paper sheds light on the experience of orthopaedic trainees in redeployment. Methods — In this retrospective study, we asked orthopaedic trainees in the KSS (Kent, Surrey, Sussex) and London Deaneries to complete a survey regarding their experience in redeployment during the COVID-19 outbreak. The study took place in the Kent, Surrey, Sussex, and London regions of the United Kingdom over a period of 8 weeks from 15th of March 2020 until 15th of May 2020. The study was based at East Kent Hospitals University NHS Foundation Trust and participants were recruited from a number of secondary and tertiary care centres across the region. 120 orthopaedic trainees were contacted, working in 21 teaching hospitals. Of these, 40 trainees (30%) from 13 hospitals responded and completed the survey. Results — 50% of the surveyed trainees were redeployed to other specialties. Trainees spent varying amounts of time in the redeployed speciality and gave differing views on how comfortable they felt and how useful they felt the experience was. One-third of trainees experienced symptoms and/ or tested positive for COVID-19 and the majority of these were redeployed to other specialties. Interpretation — Orthopaedic training appears to have taken a temporary back seat at this time but trainees have made a significant contribution to reinforcing key front-line specialties in the fight against COVID-19.

COVID-19 has had a significant impact on the health services since late December 2019, when the outbreak was reported in Wuhan, China. The entire healthcare sector, including trauma and orthopaedics, has been compelled to adapt to the exponentially increasing number of cases (Chang Liang et al. 2020). In the United Kingdom, the focus of the healthcare system shifted to treating COVID-19 cases and reducing further spread. Hospitals were seeing their resources shifted to reinforce the emergency, acute medical, and intensive care teams as advised by NHS England and the British Orthopaedic Association (2020). This appeared to be consistent with other countries around the world (Sarpong et al. 2020), whose hospitals and orthopaedic services followed a similar tactic (Chang Liang et al. 2020). At the heart of this was the redeployment of the orthopaedic trainees to support “frontline specialties”. The governing body that manages trainees across all specialties, Health Education England (HEE), issued guidance recommending this redeployment (Health Education England 2020), but this was to be organised locally by trusts. This had huge repercussions on the workings of the orthopaedic department and the trainees themselves. This paper sheds light on the experience of orthopaedics trainees in redeployment during the COVID-19 outbreak.

Methods In this retrospective study, we asked orthopaedic trainees in the KSS (Kent, Surrey, Sussex) and London Deaneries to complete a survey regarding their experience in redeployment during the COVID-19 outbreak. The survey was constructed using Survey Monkey (Palo Alto, CA, USA), and comprised 10 questions (Table 1).

Acta Orthopaedica 2020; 91 (6): 650–653

Table 1. Outlining survey questions

651

Required tasks Administration

Redeployment survey 1. Have you been re-deployed to another specialty during the COVID-19 crisis? 2. What’s your current grade? 3. To which department were you re-deployed? 4. In the last 8 weeks, what percentage of your total clinical time was spent in re-deployment? 5. What did your role involve? 6. How comfortable did you feel performing the role given to you? 7. Did you receive any training/ education during this period in that speciality? 8. How valuable did you find your experience in re-deployment? 9. Have you yourself had symptoms and/or tested positive for COVID-19 in the last 8 weeks? 10. Any further comments regarding your experience?

Proning/deproning Ward round support Reviewing/clerking patients Managing patients Performing intervention Other 0

10 20 30 40 50 60 70 80 90 100

Distribution (%) Figure 2. Tasks required of trainees during redeployment.

Responded to survey n = 40

Trainees remained within orthopaedics n = 20

Trainees redeployed to (n = 20): – Intensive Care Unit, 14 – General Medicine, 5 – Accident and Emergency, 1

Experienced symptoms and/or tested positive for COVID-19 n = 14

Trainees remained within orthopaedics n=5

Trainees redeployed to (n = 9): – Intensive Care Unit, 4 – General Medicine, 4 – Accident and Emergency, 1

Figure 1. Pattern of redeployment.

Figure 3. Number of trainees who experienced symptoms and/or tested positive for COVID-19 during redeployment.

120 trainees were contacted, working in 21 teaching hospitals. Of these, 40 orthopaedic trainees (30%) from 13 hospitals responded and completed the survey. Invitations were sent using email, WhatsApp, and Facebook messenger. A week later a reminder was sent. As the variables were binary or categorical, data was described using frequencies and percentages.

“very uncomfortable”, 4 were “fairly comfortable”, 5 were “indifferent”, and 9 were “completely comfortable”. 24 trainees reported having received an induction or training session on COVID-19, with some receiving it prior to their redeployment. The remainder noted an absence of such. The typical activities asked of trainees in their redeployment included administrative tasks, proning and de-proning of intubated and ventilated patients, reviewing patients, or assisting with clerking of new patients. Some trainees reported managing patients themselves and performing interventions (Figure 2). When asked about how useful they found their redeployment with regard to their future careers, 6 trainees found it “interesting but not applicable”, 5 of trainees found it “not interesting and not applicable”, 7 found it “interesting and applicable”, and 2 were unsure. Working in close proximity to COVID-19 confirmed patients was likely to have been of significant concern to many trainees and one-third of trainees reported experiencing symptoms and/or testing positive for COVID-19 in the last 8 weeks from 15th of March 2020 until 15th of May 2020. Looking into the redeployment status of those who experienced symptoms of, or tested positive for, COVID-19, the majority were redeployed to various other clinical areas such as general medicine or intensive care (Figure 3).

Ethics, funding, and potential conflicts of interest Ethical approval and registration—Not applicable. Funding sources—No funding sources to declare. Conflicts of interest—No conflicts of interest to declare.

Results Of the 120 trainees surveyed, 40 responses were recorded, and of the 40 trainees who responded, 20 were redeployed (Figure 1). The time spent in the redeployment role was highly varied. 7 trainees spent a quarter of their time in redeployment, 4 trainees spent half of their time in redeployment, 4 trainees spent three-quarters of their time in redeployment, and 5 spent all their time in redeployment. On surveying the 20 redeployed trainees’ attitudes on the tasks they were asked to perform, 2 reported that they were

652

Discussion COVID-19 redeployment is commonplace in most hospitals in the United Kingdom. The British Medical Association (BMA) compiled a document outlining how the process should occur (British Medical Association 2020a). In summary, it states that the trainee should consent to the redeployment, understand their role, and not be expected to carry out any activities outside their level of training or that they do not feel comfortable with. Employers are also expected to carry out an induction session specific to the redeployment proposed, and provide adequate personal protective equipment (PPE) in line with Public Health England guidance (Public Health England 2020). This includes a special addendum on what to do if this condition is not met (British Medical Association 2020b). Our data shows that half of the surveyed trainees were redeployed to other specialties, with the other half remaining in orthopaedics. This was part of a coordinated response of many hospitals to boost the staff and resource allocation to “front line” specialties (Sarpong et al. 2020) and was endorsed by Health Education England (HEE). This appears to be in line with a national survey of all UK doctors performed by the BMA, which reports that half of UK doctors were redeployed to other specialties (British Medical Association 2020c). Junior trainees were more commonly redeployed and the reasons for this may have included increased retention of general medical knowledge compared with their senior counterparts (Culp and Frisch 2020). It was felt by some university and hospital leaders that the risk of exposure to COVID-19 and the need to conserve PPE as much as possible outweighed the training benefit (Gallagher and Schleyer 2020) to junior surgeons. This was recognised by HEE and the Joint Colleges of Surgical Training (JCST) and the annual review of trainees’ competencies was altered to reflect this. However, this decision did not appear to be favoured by all trainees, with some reporting frustration with the lack of orthopaedic training opportunities: “Blanket exclusion from theatre has been frustrating considering the opportunities have continued to exist in some capacity and the redeployment duties have been low-demand.” The proportion of the time trainees spent in different specialties was managed locally as per guidance from HEE, which states that trainees may be required to assist in managing acutely unwell patients, at the expense of planned training activites, but with protective expectations about not working beyond their competencies and capabilities (Health Education England 2020). Some institutions redeployed their trainees wholly to front-line specialties such as Accident and Emergency and intensive care, whilst others created a rota with a mix of parent and redeployed specialties. Reasons for the discrepancy may have included varying staffing levels within that hospital and anticipated decline in workload levels within

Acta Orthopaedica 2020; 91 (6): 650–653

orthopaedics. It is interesting to note that only one-quarter of trainees spent the entire period in redeployment, whilst the majority continued to have some orthopaedic exposure. Therefore, a significant proportion of trainees could at least partially continue their training whilst supporting the response to the pandemic. Reasons for this could be the equitable distribution of the redeployment duties, which would have ensured that the morale of the trainee surgeons was maintained: “Short redeployment managed between orthopaedics and ITU. Worth doing, but was glad to return to regular duties.” Some trainees felt that they were surplus to requirement in their redeployed specialty. For example, as the number of admissions to intensive care started to decline after the initial wave (ICNARC 2020), workload may have diminished with the added caveat that intensive care trainees may be better placed to carry out tasks and clinical work in their own specialty than orthopaedic trainees: “Intensive Care (ITU) was very supportive, though over staffed and little work to do, however the redeployment was understandable. Better use of the resource could have been if surplus staff were sent back to parent specialty for support of those services.” 1 of the key themes outlined by the BMA in its document surrounding redeployment was to ensure trainees and junior doctors did not perform tasks outside their level of training and competency and were not pressured to do so (British Medical Association 2020a). Our data showed that the majority of trainees were “indifferent” or “completely comfortable” with the tasks they were asked to do. These tasks were mostly completing administrative and documentation aspects of the ward round, assisting with proning and de-proning patients and accompanying intensive care doctors to clerk in new patients. This also suggests a reasonable competency amongst the trainees in carrying out basic medicine-based duties and reflects the value of such doctors in unprecedented situations, such as the pandemic. However, around a quarter of surveyed trainees report they were uncomfortable with their duties and this included being asked to run an A&E minor injuries with no senior support, and managing general medical patients. This appears to be in line with the national BMA survey, which reported that one-fifth of UK doctors were not comfortable with tasks given to them in redeployment (British Medical Association 2020c). Whilst this level of care may be expected of core trainee level doctors, it is understandable that this could be anxiety-provoking (British Medical Association 2020a). Anxiety amongst doctors was a significant theme in the BMA national survey, which reported that one-third of UK doctors suffered depression, anxiety, stress, burnout, emotional distress, or another mental health condition worse than prior to the pandemic, with 40% having to access NHS well-being services (British Medical Association 2020c). This highlights a key issue surrounding well-being of trainees working during the pandemic.

Acta Orthopaedica 2020; 91 (6): 650–653

Another key requirement highlighted by the BMA was to ensure trainees and junior doctors received adequate induction and training sessions to ensure they were as prepared as possible for their redeployment to a completely new specialty (British Medical Association 2020a), possibly one that they have never worked in before. Our results showed that almost half of trainees received no such training, which may have contributed to one-quarter of trainees feeling uncomfortable with the tasks they were asked to perform. The possible reasons for the lack of induction or training could be the rapidly changing requirements of the COVID-19 response, leading to urgent redeployment before delivery of such training was possible: “We were provided with little or no training. Departmental ‘Zoom’ teaching was set up once weekly.” All trainees surveyed were orthopaedic trainees and therefore it was interesting to ascertain how useful they felt the redeployment would be to their future careers. The majority found the redeployment interesting but half did not feel it was actually applicable to their future careers. Some on the other hand felt supported and gained valuable experience/; “Redeployed under the medical team. Great training experience from all grades who were very welcoming—especially given the fact that I was an Orthopaedic trainee.” Finally, a likely concern for trainees as outlined by the BMA was the close and prolonged proximity to COVID-19 positive patients and the risk of being exposed to COVID19 themselves. This was reflected in the BMA national survey, which reported that only four-tenths of UK doctors felt adequately protected from COVID-19 (British Medical Association 2020c). Whilst our survey did not directly address issues around personal protective equipment (PPE), the BMA national survey reported widespread shortages of common items, such as “shortage” or “no supply at all” in 12% of responses. 29% of respondents felt “sometimes” or “often” pressured to see patients without adequate protection for the specific area of care (British Medical Association 2020c). Our data shows that one-third of trainees surveyed displayed symptoms and/or tested positive for COVID-19 during the survey time and the majority of these trainees were redeployed, with the areas of intensive care and general medicine being the most affected (Figure 3). This appears to be higher than the national survey of UK doctors by the BMA, who reported that roughly one quarter of surveyed respondents displayed symptoms and/or tested positive for COVID-19 (British Medical Association 2020c). However, the numbers in our study are too small to draw a reliable conclusion. We acknowledge that responder bias may be present in terms of increased number of responses from those trainees who were redeployed compared with those who remained in orthopaedics. Anecdotally, we believe that the low response rate could be due to the large number of COVID-19 related emails that trainees received at this time, as well as altered shift patterns and activities within shifts. Risk factors

653

for this may have included trainees’ health conditions or lack of adequate PPE. However, it is noteworthy that a significant number of trainees who remained within orthopaedics displayed symptoms and/or tested positive for COVID-19 during the survey period and it stands to reason that the risk factors for this are the same regardless of the department. Conclusion Our data provides the first insight into orthopaedic trainees’ experience of redeployment to other specialties in the United Kingdom. COVID-19 has changed the way the entire healthcare system functioned in the 8-week study period from 15th of March 2020 until 15th of May 2020. Whilst orthopaedic training appears to have taken a temporary back seat at this time, trainees have made a significant contribution to reinforcing key front-line specialties, such as intensive care, in the fight against COVID-19.

GF and BT designed and distributed the survey, with input from BD, and collated the results. GF designed and wrote the majority of the manuscript with BT, NH, SV, GR, BD, and JR providing critique and suggestions for its contents. Acta thanks Cecilia Escher and Roger Skogman for help with peer review of this study.

British Medical Association. 2020a. Available from https://www.bma.org.uk/ advice-and-support/covid-19/returning-to-the-nhs-or-starting-a-new-role/ covid-19-staff-redeployment British Medical Association. 2020b. Available from https://www.bma.org.uk/ advice-and-support/covid-19/ppe/covid-19-refusing-to-treat-where-ppeis-inadequate British Medical Association. 2020c. Available from https://www.bma.org.uk/ media/3070/bma-covid-tracker-survey-full-results-aug-2020.pdf British Orthopaedic Association. 2020. Available from https://www.boa. ac.uk/uploads/assets/ee39d8a8-9457-4533-9774e973c835246d/4e3170c2d85f-4162-a32500f54b1e3b1f/COVID-19-BOASTs-Combined-FINAL. pdf Chang Liang Z, Wang W, Murphy D, Po Hui J H. Novel coronavirus and orthopaedic surgery: early experiences from Singapore. J Bone Joint Surg Am 2020; e000236. doi: 10.2106/JBJS.20.00236 Culp B M, Frisch NB. COVID-19 impact on young arthroplasty surgeons. J Arthroplasty 2020; 35(7S): S42-S44. doi: 10.1016/j.arth.2020.04.058 Gallagher T H, Schleyer A M. “We signed up for this!”—student and trainee responses to the Covid-19 pandemic. N Engl J Med 2020; 382(25): e96. doi: 10.1056/NEJMp2005234 Health Education England. 2020. Available from https://www.hee.nhs.uk/ coronavirus-information-trainees ICNARC. 2020. Available from https://www.icnarc.org/Our-Audit/Audits/ Cmp/Reports Public Health England. 2020. Available from https://www.gov.uk/government/publications/wuhan-novel-coronavirus-infection-prevention-andcontrol/covid-19-personal-protective-equipment-ppe#summary-of-pperecommendations-for-health-and-social-care-workers Sarpong N O, Forrester L A, Levine W N. What’s important: redeployment of the orthopaedic surgeon during the COVID-19 Pandemic: perspectives from the trenches. J Bone Joint Surg Am 2020; 102(12): 1019-21. doi: 10.2106/JBJS.20.00574

654

Acta Orthopaedica 2020; 91 (6): 654–659

The anatomical SP-CL stem demonstrates a non-progressing migration pattern in the first year: a low dose CT-based migration study in 20 patients Olof SANDBERG 1, Simon THOLÉN 2, Sofia CARLSSON 2, and Per WRETENBERG 3 1 Sectra

AB, Linköping; 2 Department of Radiology, Lindesberg Hospital, Örebro University Hospital; 3 Department of Medical Sciences, Section of Orthopaedics, Örebro University Hospital, Sweden Correspondence: olof.sandberg@sectra.com Submitted 2020-05-05. Accepted 2020-09-17.

Background and purpose — RSA is the gold standard for evaluation of early implant migration. We report the results of a new CT-based method Sectra CT micromotion analysis (CTMA) applied to assess the migration pattern in 20 patients in the 1st year after surgery, both with and without the use of tantalum beads in the bone. The patients had an SP-CL anatomical stem that uses an S-shape, designed to better fit the curvature of the femur. Patients and methods — 20 THA patients (mean age 61 years, 10 female) received SP-CL stems, tantalum markers in the femur, and low-dose CT scans at 1 day, 3 months and 12 months postoperatively. In addition, precision as well as inter- and intra-observer variability of the 12-month migration was measured. Results — The 3-month subsidence was median 0.5 mm (95% CI 0.3–1.0) and the internal rotation 1.8° (CI 0.9–2.6). At 12 months the corresponding values were 0.6 (CI 0.3–1.6) mm and 1.9° (CI 0.8–2.4). Precision was 0.1 to 0.3 mm and 0.1° to 0.4° at 3 and 12 months. Intra- and interobserver variability yielded R-values averaging 0.96 and 0.98. Interpretation — The migration mainly took place during the 1st 3 months, in line with other uncemented stems. The number of patients with subsidence over 2 mm in the first year (5) might be due to the design of the prosthesis with an anatomical shape. Alternatively, our results might indicate a challenge when choosing the correct size for these new anatomical stems. CTMA provided precise and highly repeatable measurements of migration without the need for tantalum markers.

Anatomical stems use various features such as an S-shape, surface roughness, and grooves to attempt to increase stability and minimize stress shielding and aseptic loosening. Anatomical stems are designed to fill the proximal femur to allow for better osteointegration. This concept has been demonstrated to give good long-term results (Kim 2008). Compared with other anatomical stems, the uncemented anatomical SP-CL stem from Link (Waldemar Link, Hamburg, Germany) has a longer distal part to help the surgeon place the stem in the correct position. Computer simulations have indicated that this stem design should provide considerable advantages for strain shielding (Heyland et al. 2019), but to our knowledge the stem performance has not been evaluated in a migration study. The long-term survival of an implant can be assessed in a small cohort by using radiostereometric analysis (RSA) (Kärrholm 2012). Problematic implants will already distinguish themselves after 1–2 years through their migration pattern (Pijls et al. 2012). Underperforming designs can be identified and avoided early on (Nelissen et al. 2011). For the last 3 decades implant migration measurements have primarily been done with RSA. Since 2018 a new CT-based alternative, Sectra CT based Micromotion Analysis (CTMA), has become available. This method relies on standard CT machines instead of the specialized RSA-lab equipment. We describe the 1-year migration pattern of the SP-CL anatomical stem and assess the capabilities of the CTMA tool.

Patients and methods Patients and surgical procedure 20 THA patients (mean age: 61 years (range 45–74), 10 female) underwent between May 2017 and May 2018 total hip replacement with anatomical stems (SP-CL, Waldemar Link,

Acta Orthopaedica 2020; 91 (6): 654–659

655

Figure 2. Standard DICOM frame of reference used in this article. Arrows point in the positive direction.

CT scans The patients received CT scans postoperatively on day 1, at month 3, and at month 12. The settings used for the CT (Siemens Somatom Definition AS 64 slices) were a kVp of 120 kV, a tube current of 30 Eff mAs, 0.6 mm increments, 1.0 pitch, and a rotation time of 0.5 seconds. Slice acquisition was done in 64 x 0.6 mm. Reconstruction was done with a model-based Safire level 3 I41f algorithm in 0.6 mm thick slices, using IMAR at level Hip Implants. The dosage received per scan was 1.7 mSv. This protocol was based on a previous study (Eriksson et al. 2019).

Figure 1. CTMA basic workflow. Step 1: identify the reference body (femur) in both postoperative CT and follow-up CT. The resulting computer-generated blue/red/green color scale indicates how well the computer matched postoperative and follow-up. Step 2 repeats the process but now for the moving body (stem). In the 3rd and final step one or several points of interest are defined. These are the points for which data is to be reported. Definition of a custom coordinate system can also be done in the same step. Data acquisition includes migration quantification as well as moving images.

Hamburg, Germany) and tantalum markers. Inclusion criteria were age 30–85 years and normal anatomy. All patients were ASA class 1 or 2. 2 experienced surgeons performed the operations. Preoperative templating was done in 2D using the Sectra Orthopedic package (Sectra, Linköping, Sweden) to preliminarily plan implant size and positioning with the final decision taken during surgery. All surgeries were done with a posterior approach. Implant The core of the SP-CL stem is made of a titanium alloy. The stem has a polished finish on the distal and proximal parts, with a calcium phosphate coating on the central part, which has a ribbed structure. The implant has an S-shape.

Migration measurement CT scans were loaded into a migration measurement tool (CTMA version 21.1.1, Sectra, Linköping, Sweden). The main procedure followed the steps below (Figures 1 and 2): 1. 2 CTs of the same patient are loaded into the CTMA software (for example postoperatively and 1 year). 2. A reference body (such as the femur) is indicated by the user in both CTs and the software then matches the positions of the body as closely to each other as possible (Figure 1, step 1). A color-coded overlay indicates wellness of fit. The user decides based on visual inspection whether the matching is sufficient, whether it needs correction, or in extreme cases whether it merits exclusion. 3. Step 2 is repeated with a moving body (the stem, Figure 1, step 2). 4. The coordinate system is adjusted (in this study we used standard DICOM, i.e., patient-oriented coordinate system, Figure 1, step 3 and Figure 2). 5. Measurements points to be reported are defined (Figure 1, step 3. This study used tip, neck, and center of the head, see Figure 4, Supplementary data). 6. The software reports the movement and rotational changes for the points of interest relative to the reference body, reported in the chosen frame of reference. The data is imported into a suitable statistics software. Moving images can also be collected (see Supplementary Video).

656

Acta Orthopaedica 2020; 91 (6): 654–659

Reported quantities and repeatability measurements The main reported results in this study are internal rotation and subsidence, as they are closely connected to the most common modes of failure. Additional measurements for all 3 migration and rotation directions were made on head, neck, and tip both with and without tantalum markers for the reference body (Figure 5, Tables 4 and 5, Supplementary data). All measurements were done by 2 observers in parallel according to a protocol decided on in advance. One observer (ST) was new to the tool, and the other (OS) was experienced. OS repeated the measurements again after 1 month. The repeated measurements formed the basis of repeatability estimations. Exclusion criteria for images of poor quality If a CTstack deviated from the recommended reconstruction settings, observers 1 and 2 together decided whether an adjustment of Hounsfield unit settings in CTMA could make the CT readable or whether it should be excluded. Default HU settings were 300 for bone and 2,200 for metal. Exclusion criteria for marker-based measurements with poor marker spread In RSA studies a condition number above 150 is typically used as an exclusion criterion for tantalum markers that are insufficiently spread out. Since the CTMA software does not provide this calculation, an RSA expert with over 20 years of experience with RSA was shown the CT stacks of each patient and asked to categorize which patterns could clearly be estimated as “below 150 in condition number.” This was done based on a visual inspection of the CT stacks in 3D visualization software. All patterns that did not fall into this category were excluded from the bead-based data analysis. Precision estimates 9 patients had 2 CT examinations taken on the same occasion, so-called double examinations. The precision definition used was based on the ISO standard for RSA with the standard deviation of the double examinations multiplied by a constant to provide the 95% confidence interval around zero (Standardization IOf: International Standard ISO 16087:2013(E)). In our study precision is reported with the formula (x ) ∑ √––––––––– n n

precision = t * SD = t(N – 1) *

i=1

where t(N-1) is the value from the t-distribution that corresponds to a 95% double-sided interval with N-1 degrees of freedom where N equals the number of double exams. A t-distribution is considered to give a better estimate than a normal distribution due to N being lower than 30. The chosen expression for SD takes height for the measured quantity of total translation, which is not centered around 0. Mean difference between the double examinations is also reported.

Table 1. Median subsidence and internal rotation of the neck at 3 and 12 months, with 95% CI in parentheses Factor Subsidence, mm Internal rotation, °

3 months

12 months

0.5 (0.3–1.0) 1.8 (0.9–2.6)

0.6 (0.3–1.6) 1.9 (0.8–2.4)

Statistics Normality was tested for by looking at Q-Q plots and through a Shapiro–Wilks test. For data sets that were not normally distributed bootstrapping using 1,000 samples was used to generate 95% confidence intervals (CI) of the median. Both inter- and intra -observer variability was tested with a 2-way random intraclass correlation in absolute agreement, single measures mode (ICC(2,1)) R-value calculation and reported with the lower 95% CI. SPSS Statistics subscription version 1.0 (IBM Corp, Armonk, NY, USA) was used for statistics and Graphpad Prism 8.4.0 (Graphpad Prism Software, San Diego, CA, USA) for graphs. Ethics, funding, and potential conflicts of interest This study was approved by the Swedish Ethical Review Authority in Uppsala under approval ID number 219-00117 and funded by Region Örebro County and by Link Sweden AB (Åkersberga, Sweden), which provided institutional support but with no influence on study design or presentation. PW, ST, and SC report no potential conflicts of interest. OS is a full-time employee at Sectra, a company commercializing a CT-based migration measurement tool.

Results No patients or CT exams were excluded from the study. All patients were followed up. There were no complications during surgery and no patients experienced complications. At the time of writing approximately 2 years have passed since surgery and so far no problems have occurred in any patients. The migration data at 3 and 12 months was mostly found not to be normally distributed and we decided that all migration quantities were treated as such. The median 3- and 12-month subsidence for the neck was 0.5 and 0.6 mm and the internal rotation 1.8 and 1.9° (Table 1 and Figure 3, as well as Video link in Supplementary data). Visual inspection of the data (graph not shown) demonstrated no clear correlation between subsidence and internal rotation. Inter- and intra-observer variability was reported to be 0.88– 0.99 and 0.90–0.99 respectively for the different measured quantities and with bone-based or bead-based measurement (Tables 2 and 3). 5 of the 69 CT scans have been saved with reconstruction settings differing from the study protocol (deviations in slice

Acta Orthopaedica 2020; 91 (6): 654–659

Proximal migration, mm 0

657

Table 2. Inter-observer variability, reporting the lower 95% CI for the R-value

Internal rotation, ° 10

–1

Translation Rotation TT, x, y, z, x, y, z, total anterior adduc- internal Location medial posterior proximal translation tilt tion rotation

–2

6 –3

4 –4

–5 –6

0 3

Month

Figure 3. Individual patient values for subsidence and internal rotation of the neck at 3 and 12 months. The median values are shown in red and the precision limit is shown with a dashed red line.

thickness, metal artifact reduction, and/or different soft tissue algorithms) but for which adjustments of the parameters for HU extraction were successful in rendering them possible to analyze in all cases. 7 out of 20 patients had bead patterns that could not be assessed as being most likely below 150 in condition number. Therefore these 7 patients had their marker-based measurements excluded from any further analysis. The precision for subsidence and internal rotation was 0.07 mm and 0.37°. Using the femoral bone rather than the tantalum markers increased precision, especially for rotations. While the head and neck had similar precision, the precision of the tip measurements was worse, in particular when using tantalum markers (Table 5, Supplementary data).

Head Bead 0.99 0.96 0.99 Bone 0.97 0.99 0.99 Neck Bead 0.99 0.95 0.99 Bone 0.96 0.95 0.99 Tip Bead 0.99 0.94 0.99 Bone 0.88 0.95 0.99

0.98 0.96 0.99 0.98 0.99 0.99 0.98 0.98 0.99 0.96 0.99 0.98 0.99 0.99 0.98 0.98 0.98 0.96 0.99 0.98 0.99 0.99 0.98 0.98

Table 3. Intra-observer variability, reporting the lower 95% CI for the R-value Translation Rotation TT, x, y, z, x, y, z, total anterior adduc- internal Location medial posterior proximal translation tilt tion rotation Head Bead 0.95 0.95 0.99 Bone 0.95 0.99 0.99 Neck Bead 0.95 0.97 0.99 Bone 0.97 0.92 0.99 Tip Bead 0.95 0.90 0.98 Bone 0.96 0.93 0.99

Discussion This study describes the migration behavior of the SP-CL cementless anatomical stem, as well as the applicability of the CT-based micromotion measurement tool CTMA. Our results show a median 1-year subsidence of 0.55 mm and that this stem exhibits non-progressive behavior, i.e., that a major part of the migration occur early, prior to the 3rd postoperative month. Thereafter very little additional migration is seen. Similar behavior has been shown in numerous RSA migration studies on other uncemented stems, with follow-up times of up to 10 years (Callary et al. 2012, Weber et al. 2014, Aro et al. 2018, Sesselmann et al. 2018). Different implant types are thought to have different thresholds as to what is an acceptable level of migration. Another anatomical uncemented stem, ABG II, had a 2-year median subsidence of 0.7 mm (Aro et al. 2018), while other designs of uncemented stems have reported 1-year means of 0.6 mm (Callary et al. 2012), 0.3 mm (Sesselmann et al. 2018), (Weber et al. 2014), and 1.4 mm (Wolf et al. 2010). While less migra-

0.98 0.90 0.97 0.98 0.99 0.98 0.94 0.97 0.99 0.90 0.97 0.98 0.99 0.98 0.94 0.97 0.90 0.90 0.97 0.98 0.99 0.98 0.94 0.97

tion is always preferred, the threshold for maximum migration is unknown for an uncemented stem. Progressive migration past the early phase may present the greater risk, as compared with an abating early migration. The shorter uncemented, anatomical CFP stem, which is also designed for metaphysical anchoring with a porous surface structure, has been demonstrated to have a similar non-progressive subsidence behavior, though with smaller absolute values. Röhrl et al (2006) could see a connection between increased subsidence and too small a stem curvature. It has been shown in cemented stems (which intend to recruit absolute stability from the beginning) that a 2-year subsidence of over 1.2 mm implies a 50% risk of revision (Kärrholm et al. 1994). To our knowledge no corresponding long-term follow-up has yet been conducted for uncemented stems. It is noteworthy that no patients needed to be excluded from the markerless migration analysis. There is no risk for marker occlusion when using CTs, as there is when using RSA images. And while unsatisfactory image quality is a theoretical cause for exclusion, all 69 CT scans and reconstructions performed in this study were of acceptable quality. 1 patient had a deviating migration pattern with continued migration past the third month. At 2 years post-surgery this

658

patient yet had to return with any symptoms or complaints. In a study on uncemented stems Klein et al. (2019) had 2 revisions, both of which exhibited this type of migration. We used a new CT-based measurement tool. It demonstrated a markerless precision (0.1–0.3 mm and 0.1–0.4°) which compares excellently with the current gold standard RSA where examples of precision values for uncemented stems are reported at 0.1–0.2 mm and 0.3–2.0° (Wolf et al. 2010, Nysted et al. 2014, Weber et al. 2014). We also included precision measurement of 9 cemented cups as these were visible in the CT stacks. That precision, 0.1–0.2 mm and 0.1–0.2°, is similar to the precision reported in another recently published article on the precision of CTMA for cups: 0.1–0.3 mm and 0.2–0.4° (Brodén et al. 2020b). Both the stem and the femur have an elongated shape, which possibly might explain the worse precision of the rotation measurement along that axis and hence why the precision values of the stem rotations differ more for the different rotational directions while the precision values of a cup rotations appear to be more similar regardless of the direction. We observed that tantalum markers did not seem to offer any benefit to precision. On the contrary, the marker-based precision was markedly inferior for almost all translations and rotations. This indicates that, where possible, tantalum markers in the bone should be avoided. That the precision for the tip was markedly worse with bead-based measurements might be that the tantalum markers are placed preferably in the proximal portion of the femur and so are further away from the tip. This study for the first time reports inter- and intra-observer reliability of CTMA. The results demonstrated that a relatively inexperienced and an experienced user could produce similar excellent results (R-values). This indicates a robustness in the measurement tool. 3 patients had a subsidence of over 3.0 mm at the 1st follow-up. This may have been caused by under-sizing of the implants. 2 years postoperatively none of the patients in the study, including these 3, have returned with any complications. It might be more challenging to correctly template the stem due to the small incremental steps between sizes. This could have been connected to a learning process but the surgeons in our study had performed 10 operations each with this implant prior to study start. Furthermore, a closer look at the distribution of subsidence across the patients indicated no temporal effect, i.e., patients at the end of the series seemed to have had a similar risk of high subsidence as those at the beginning of the series. It has been suggested that anatomical models are less forgiving in certain steps (Giebaly et al. 2016). CTMA has a number of advantages when compared with RSA: it is not dependent on any markers that have to be added and remain visible and stationary on/in the bone and/ or implant. Furthermore, CTMA requires no 3D models of the implants to be acquired from the manufacturers or created via reverse engineering. Neither does it require access to an exclusive RSA lab and image capture does not involve the use

Acta Orthopaedica 2020; 91 (6): 654–659

of a calibration cage. In theory the CT-based approach could simplify follow-up as the patient does not have to return to the same hospital (though all but one follow-ups in this study were performed at the same hospital). All CTs used in this study were low dose, which reduces the radiation to below the thresholds set by the European commission on radiation protection in medical research such as this (European Commission 1998). Our study has limitations. While we chose 1 year as followup time, corresponding RSA studies usually have 2 years. Furthermore, we used a new migration measurement tool based on CT, which is not as validated as RSA. However, the CTMA software has previously been tested with good results (Bakhshayesh et al. 2019, Brodén et al. 2020a, b, Eriksson et al. 2019, Schriever et al. 2020) and the underlying principles of CT-based migration measurement have been studied in over 30 scientific publications over the last 2 decades (Olivecrona et al. 2003, 2004, 2016, Polfliet et al. 2015, Scheerlinck et al. 2015, Boettner et al. 2016a, b, Brodén et al. 2016, Otten et al. 2017). In conclusion, the SP-CL anatomical stem was shown to have a non-progressive migration pattern in line with other uncemented stems, with very little migration past the 3-month mark. The CT-based method was demonstrated to have excellent repeatability and precision. Tantalum markers in the bone were not needed. Supplementary data A supplementary video is available at http://actaorthop.org/ include/sup/14110-video.mp4. Figures 4–5 and Tables 4–5 are available as supplementary data in the online version of this article, http://dx.doi.org/10. 1080/17453674.2020.1832294

All authors took part in the drafting of this manuscript and approved the final version. In addition, PW conceived of, planned, collected patients, and took part in the data analysis of this study. OS took part in data collection, data analysis, and wrote the first draft. ST took part in data collection and analysis. SC took part in data analysis. The authors are grateful to the late Professor Emeritus Lars Weidenhielm and Dr Henrik Olivecrona for their advice on setting up this study. Acta thanks Volker Otten and Stephan Maximilian Röhrl for help with peer review of this study.

Aro E, Alm J J, Moritz N, Mattila K, Aro H T. Good stability of a cementless, anatomically designed femoral stem in aging women: a 9-year RSA study of 32 patients. Acta Orthop 2018; 89(5): 490-5. Bakhshayesh P, Sandberg O, Kumar V, Ali A, Enocson A. Volume fusion of CT images to measure femoral symmetricity. Surg Radiol Anat 2019b; (0123456789): 1-5. Boettner F, Sculco P, Lipman J, Renner L, Faschingbauer M. A novel method to measure femoral component migration by computed tomography: a cadaver study. Arch Orthop Trauma Surg 2016a; 136(6): 857-63.

Acta Orthopaedica 2020; 91 (6): 654–659

Boettner F, Sculco P K, Lipman J, Saboeiro G, Renner L, Faschingbauer M. The effect of a low radiation CT protocol on accuracy of CT guided implant migration measurement: a cadaver study. J Orthop Res 2016b; 34(4): 7258. Brodén C, Olivecrona H, Maguire GQ, Noz M E, Zeleznik M P, Sköldenberg O. Accuracy and precision of three-dimensional low dose CT compared to standard RSA in acetabular cups: an experimental study. Biomed Res Int 2016; 2016: 1-6. Brodén C, Giles J W, Popat R, Fetherston S, Olivecrona H, Sandberg O, Maguire G Q, Noz M E, Sköldenberg O, Emery R. Accuracy and precision of a CT method for assessing migration in shoulder arthroplasty: an experimental study. Acta Radiol 2020a; 61(6): 776-82. Brodén C, Sandberg O, Sköldenberg O, Stigbrand H, Hänni M, Giles J W, Emery R, Lazarinis S, Nyström A, Olivecrona H. Low-dose CT-based implant motion analysis is a precise tool for early migration measurements of hip cups: a clinical study of 24 patients. Acta Orthop 2020b; 3674:1-6. Callary S A, Campbell D G, Mercer G E, Nilsson K G, Field J R. The 6-year migration characteristics of a hydroxyapatite-coated femoral stem. J Arthroplasty 2012; 27(7): 1344-8.e1. European Commission. Radiation Protection 99. Guid Med exposures Med Biomed Res. Brussels: European Commission; 1998. Eriksson T, Maguire G Q, Noz M E, Zeleznik M P, Olivecrona H, Shalabi A, Hänni M. Are low-dose CT scans a satisfactory substitute for stereoradiographs for migration studies? A preclinical test of low-dose CT scanning protocols and their application in a pilot patient. Acta Radiol 2019; 60(12): 1643-52. Giebaly D E, Twaij H, Ibrahim M, Haddad F S. Cementless hip implants: an expanding choice. Hip Int 2016; 26(5): 413-23. Heyland M, Checa S, Kendoff D, Duda G N. Anatomic grooved stem mitigates strain shielding compared to established total hip arthroplasty stem designs in finite-element models. Sci Rep 2019; 9: 482. ISO SIoIS. ISO 16087:2013 Implants for surgery—Roentgen stereophotogrammetric analysis for the assessment of migration of orthopaedic implants. Kärrholm J. Radiostereometric analysis of early implant migration: a valuable tool to ensure proper introduction of new implants. Acta Orthop 2012; 83(6): 551-2. Kärrholm J, Borssén B, Löwenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 4–7-year stereoradiographic follow-up of 84 cemented prostheses. J Bone Joint Surg Br 1994; 76(6): 912-17. Kim Y H. The results of a proximally-coated cementless femoral component in total hip replacement: a five- to 12-year follow-up. J Bone Joint Surg Ser B 2008; 90(3): 299-305. Klein L J, Puretic G, Mohaddes M, Kärrholm J. Similar clinical results and early subsidence between the Collum Femoris Preserving and the Corail stem: a randomized radiostereometric study of 77 hips with 2 years’ followup. Acta Orthop 2019; 90(3): 202-8. Nelissen R G H H, Pijls B G, Kärrholm J, Malchau H, Nieuwenhuijse M J, Valstar E R. RSA and registries: the quest for phased introduction of new implants. J Bone Joint Surg Am 2011; 93(Suppl. 3): 62-5.

659

Nysted M, Foss O A, Klaksvik J, Benum P, Haugan K, Husby O S, Aamodt A. Small and similar amounts of micromotion in an anatomical stem and a customized cementless femoral stem in regular-shaped femurs. Acta Orthop 2014; 85(2): 152-8. Olivecrona L, Olivecrona H, Weidenhielm L, Noz M E, Maguire G Q, Zeleznik M P. Model studies on acetabular component migration in total hip arthroplasty using CT and a semiautomated program for volume merging. Acta Radiol 2003; 44(4): 419-29. Olivecrona H, Weidenhielm L, Olivecrona L, Beckman M, Stark A, Noz M, Maguire G, Zeleznik M, Svensson L, Jonson T. A new CT method for measuring cup orientation after total hip arthroplasty: a study of 10 patients. Acta Orthop Scand 2004; 75(3): 252-60. Olivecrona H, Maguire G Q, Noz M E, Zeleznik M P, Kesteris U, Weidenhielm L. A CT method for following patients with both prosthetic replacement and implanted tantalum beads: preliminary analysis with a pelvic model and in seven patients. J Orthop Surg Res 2016; 11(1): 27. Otten V, Maguire Jr G Q, Noz M E, Zeleznik M P, Nilsson K G, Olivecrona H. Are CT scans a satisfactory substitute for the follow-up of RSA migration studies of uncemented cups? A omparison of RSA double examinations and CT datasets of 46 total hip arthroplasties. Biomed Res Int 2017; 2017: 1-11. Pijls B G, Nieuwenhuijse M J, Fiocco M, Plevier J W, Middeldorp S, Nelissen R G, Valstar E R. Early proximal migration of cups is associated with late revision in THA: a systematic review and meta-analysis of 26 RSA studies and 49 survival studies. Acta Orthop 2012; 83(6): 583-91. Polfliet M, Vandemeulebroucke J, Van Gompel G, Buls N, Deklerck R, Scheerlinck T. Estimation of hip prosthesis migration: a study of zero migration. IFMBE Proc 2015; 45: 126-9. Röhrl S M, Li M G, Pedersen E, Ullmark G, Nivbrant B. Migration pattern of a short femoral neck preserving stem. Clin Orthop Relat Res 2006; 448: 73-8. Scheerlinck T, Polfliet M, Deklerck R, Van Gompel G, Buls N, Vandemeulebroucke J. Development and validation of an automated and marker-free CT-based spatial analysis method (CTSA) for assessment of femoral hip implant migration: in vitro accuracy and precision comparable to that of radiostereometric analysis (RSA). Acta Orthop 2015; 87(2): 1-7. Schriever T, Olivecrona H, Wilcke M. Triquetral motion is limited in vivo after lunocapitate arthrodesis. J Hand Surg Glob Online 2020; 2(1): 42-5. Sesselmann S, Hong Y, Schlemmer F, Hussnaetter I, Mueller L A, Forst R, Tschunko F. Radiostereometric migration measurement of an uncemented Cerafit® femoral stem: 26 patients followed for 10 years. Biomed Tech 2018; 63(6): 657-63. Weber E, Sundberg M, Flivik G. Design modifications of the uncemented Furlong hip stem result in minor early subsidence but do not affect further stability: a randomized controlled RSA study with 5-year follow-up. Acta Orthop 2014; 85(6): 556-61. Wolf O, Mattsson P, Milbrink J, Larsson S, Mallmin H. Periprosthetic bone mineral density and fixation of the uncemented CLS stem related to different weight bearing regimes: a randomized study using DXA and RSA in 38 patients followed for 5 years. Acta Orthop 2010; 81(3): 286-91.

660

Acta Orthopaedica 2020; 91 (6): 660–668

Long-term migration of a cementless stem with different bioactive coatings. Data from a “prime” RSA study: lessons learned Paul VAN DER VOORT 1, Martijn L D KLEIN NULENT 1, Edward R VALSTAR 1,2, Bart L KAPTEIN 1, Marta FIOCCO 3,4, and Rob G H H NELISSEN 1,2 1 Department of Orthopaedics, Leiden University Medical Center, Leiden; 2 Department of Biomechanical Engineering, Faculty of Mechanical, Maritime and Materials Engineering, University of Technology Delft, Delft; 3 Department of Medical Statistics and Bioinformatics, Leiden University Medical Center, Leiden; 4 Mathematical Institute, Leiden University, Leiden, The Netherlands Correspondence: p.van_der_voort@lumc.nl Submitted 2020-04-23. Accepted 2020-09-09.

Background and purpose — Little is known about the long-term migration pattern of cementless stems in total hip arthroplasty (THA). Furthermore, the role of bioactive coatings in fixation, and thus migration, remains uncertain. Hydroxyapatite (HA) is the most commonly used bioactive coating. However, delamination of the coating might induce loosening. Alternatively, fluorapatite (FA) has proved to be more thermostable than HA, thereby potentially increasing longevity. We assessed the long-term migration of cementless stems with different coatings using radiostereometric analysis (RSA), thereby establishing a reference for acceptable migration. Patients and methods — 61 THAs in 53 patients were randomized to receive either a HA, FA, or uncoated MalloryHead Porous stem during the years 1992 to 1994. Primary outcome was stem migration measured using RSA and secondary outcome was the Harris Hip Score (HHS). Evaluation took place preoperatively and postoperatively on the second day, at 6, 12, 25 and 52 weeks, and annually thereafter. At the 25-year follow-up, 12 patients (17 THAs) had died and 1 patient (1 THA) was lost to follow-up. Due to the high number of missing second-day postoperative RSA radiographs, the 1-year postoperative RSA radiograph was used as baseline for the comparative analyses. Results — Mean follow-up was 17 years (SD 6.6). All stems showed initial rapid migration with median subsidence of 0.2 mm (–0.1 to 0.6) and median retroversion of 0.9° (–3.2 to 2) at 12 months, followed by stable migration reaching a plateau phase. No stem was revised, albeit 1 stem showed continuous subsidence up to 1.5 mm. Comparing the different coatings, we could not find a statistically significant difference in overall 25-year migration (p-values > 0.05). Median subsidence at 15-year follow-up was for HA –0.1 mm (–0.4 to 0.2), for FA 0 mm (–0.1 to 0.2), and for uncoated stems 0.2 mm (–0.1 to 0.5). Median internal

rotation at 15-year follow-up was for HA not available, for FA 1.1° (–0.5 to 2.6), and for uncoated stems 0° (–0.5 to 0.4). HHS were also comparable (p-values > 0.05), with at 15-year follow-up for HA 85 points (41–99), for FA 76 points (61–90), and for uncoated stems 79 points (74–90). Interpretation — The long-term migration pattern of cementless stems using different bioactive coatings has not previously been described. No beneficial effect, or side effect at long-term follow-up of bioactive coatings, was found. The provided migration data can be used in future research to establish thresholds for acceptable migration patterns cementless stem designs.

Long-term migration data on cementless femoral stems in total hip arthroplasty (THA) is scarce, with only a few studies reporting migration measured with radiostereometric analysis (RSA) with follow-up beyond 10 years (Sesselmann et al. 2018, Critchley et al. 2020). In a prior meta-analysis we were unable to establish a threshold for acceptable early subsidence for cementless stems, because of the lack of long-term survival and migration data (Van der Voort et al. 2015). As the number of THAs being performed is still on the rise, as well as the number of relatively young patients receiving mostly cementless THAs, the burden of future failure and subsequent revision is expected to increase (Kurtz et al. 2009). Hence longevity of implants is paramount and should be scrutinized. Although bioactive coatings for cementless stems are widely employed, their beneficial effect remains questionable (Hailer et al. 2015, Inacio et al. 2018). Pooled data from randomized and cohort studies showed no clinical benefit of hydroxyapatite (HA)-coated implants (Gandhi et al. 2009, Goosen et al. 2009, Li et al. 2013, Chen et al. 2015) and large registry studies found no difference in risk of revision surgery

Acta Orthopaedica 2020; 91 (6): 660â&#x20AC;&#x201C;668

(Paulsen et al. 2007, Lazarinis et al. 2011, Hailer et al. 2015). A recent registry study found an overall lower risk of revision of HA-coated stems, but this was not consistent among different implant types, suggesting a significant influence of distinct design features on longevity (Inacio et al. 2018). Bioactive coatings were introduced in the 1980s to enhance fixation by osseointegration, with HA used as the most common coating (Geesink 1989, Furlong and Osborn 1991). However, retrieval studies have shown resorption and delamination of the HA coating from the implant, which raised concerns regarding the induction of osteolysis and, ultimately, failure of the implant (Bloebaum et al. 1994, Bauer 1995). Fluorapatite (FA) was introduced as an alternative to HA with comparable biocompatibility and osteoconductive properties (Dhert et al. 1993), but with better thermostability (Lugscheider et al. 1989). Hence, FA might adhere better to the implant during the application process using a plasma-spraying technique, thereby possibly reducing resorption and delamination of the coating (Klein et al. 1994). HA-coated implants have shown reduced migration in comparison with their uncoated counterparts (SĂ¸balle et al. 1993, KĂ¤rrholm et al. 1994b). To our knowledge, FA has not been investigated in RSA studies, or in clinical trials. In 1991, we initiated a trial to investigate the influence of different coatings on migration of cementless stems, the first RSA study performed at our facility. Despite teething problems, patient follow-up was continued to provide long-term migration data on cementless stems in general, and bioactive coatings specifically.

Patients and methods Study design This study was initially designed in 1991 as a multi-center, single blinded, randomized controlled trial comparing the influence of different coatings on the migration and clinical outcome of cementless THA. During the pilot phase of this study logistical problems were encountered with regard to obtaining RSA radiographs at the different participating hospitals, as this was (at that time) possible at only 1 institute (Leiden University Medical Center). Subsequently, it was decided to continue as a single-center study performed at the Leiden University Medical Center. From May 1992 to May 1994, all consecutive patients scheduled to receive a cementless primary THA for osteoarthritis, either primary or secondary to a systematic inflammatory disease and younger than 65 years of age, were approached for participation in a randomized, clinical RSA study. Included patients were randomized to 2 intervention groups receiving either an HA- or FA-coated implant and the control group received an implant without a bioactive coating. Treatment allocation was randomized with the use of a computergenerated randomization scheme and bilateral cases were

661

Figure 1. Mallory-Head Porous stem, with a porous coating on the proximal third, a gritblasted surface on the middle third, and a smooth satin-textured surface on the distal third.

allowed. The study design was single-blind; surgeons were aware of the coating used; clinical observers were blinded to the type of coating. The study was performed in compliance with the Helsinki Declaration, approval of the institutional medical ethical board was obtained, and all patients gave written informed consent. Surgical technique All THAs were implanted by experienced hip surgeons, or under their direct supervision. Surgeries were performed through a direct lateral approach in the lateral decubital position, except for 2 posterolateral approaches. For RSA measurements, 1-mm tantalum markers were inserted into the proximal femur during surgery. All patients received the same rehabilitation program starting with passive and controlled active movements on the first postoperative day and mobilization with full weight-bearing on the second postoperative day, after the first RSA radiograph was obtained. Implants All patients received a Mallory-Head Porous stem with a dual tapered design with a round cross-sectional geometry (Biomet, Warsaw, IN, USA). The stem is characterized by an anterior and posterior flange and wide lateral fin. It is made of a titanium alloy (Ti-6A1-4V), with a porous coating on the proximal third, a grit-blasted surface on the middle third, and a smooth satin-textured surface on the distal third (Figure 1). The implants with a bioactive coating received either HA or FA plasma-sprayed onto the proximal porous coated surface. All patients received a 28 mm cobalt-chromium head and a cementless Mallory-Head finned Ringloc acetabular cup (Biomet, Warsaw, IN, USA).

662

Acta Orthopaedica 2020; 91 (6): 660–668

Table 1. Precision of RSA measurements (upper limits of 95% zero motion confidence interval) Translation, mm Rotation, °

Transverse Longitudinal Sagittal (x-axis) (y-axis) (z-axis) 0.22 0.80

0.17 1.13

0.54 0.31

Follow-up Patients were evaluated preoperatively and postoperatively at 6 weeks, 3 months, 6 months, 1 year, and annually thereafter. At each evaluation, RSA radiographs were obtained and the Harris Hip Score (HSS) was determined. Conventional anteroposterior and lateral radiographs were acquired preoperatively, at 6 weeks, and at 2, 5, 10, 20, 25 years postoperatively, and on indication (e.g., pain or suspected failure). On the 6-week postoperative radiographs the stem orientation (i.e., varus, neutral, or valgus) was determined. Patients unable to attend follow-up moments were contacted to check implant status and whether implant-related problems had arisen. RSA technique RSA radiographs were obtained using a uniplanar setup with the patient in supine position and the calibration cage underneath the examination table. During follow-up, in 2002, the initial calibration box (Large Reference Box, Leiden, The Netherlands) was replaced by a new box (Carbon Box, Leiden, The Netherlands). Furthermore, in 2004 digital radiography was introduced. Both changes had no effect on the accuracy of the RSA measurements (Pijls et al. 2012). A marker-based analysis was carried out to calculate migration over time (Model-Based RSA software, version 3.34; RSAcore, The Netherlands), using 4 stem markers: 3 markers attached to the stem (performed by the manufacturer) and the center of the head acted as a fourth marker. Migration was expressed as translations along and rotation about 3 axes (longitudinal, transverse, and sagittal) of a right-handed orthogonal coordinate system. Since the failure mechanism of stems consists of subsidence and retroversion (Kärrholm et al. 1994a), the primary effect variables were translation along and rotation about the longitudinal axis. The accuracy of RSA measurements was determined by obtaining double examinations of 29 stems. Assuming zero migration in the brief time interval between these double examinations, the limits of the 95% prediction interval of accuracy of zero migration were determined (Table 1) (Ranstam et al. 2000). For all examinations, the mean error of rigid body fitting of the RSA markers in the femur was below 0.35 mm; the mean condition number of the RSA markers was 37 (SD 22; range, 13–111) in the femur. Bone markers were defined as unstable when they moved more than 0.5 mm with respect to the other bone markers. Unstable markers were excluded from the analyses. These values satisfy the marker

stability and distribution criteria of the RSA guidelines and the ISO guideline (ISO 16087:2013) (Valstar et al. 2005). Statistics Measured values of normally distributed data are reported as the mean (SD); measured values of non-normally distributed data are reported as the median (range). Estimates are reported as the mean and the 95% confidence interval (CI). Reported analyses were performed according to the per-protocol principle to reflect the genuine effect of treatment (i.e., HA vs. FA vs. uncoated). To safeguard for attrition bias, all analyses were repeated according to the intention-to-treat principle and compared with the outcomes of the per-protocol analyses. Migration and increase in HHS throughout the follow-up period were analyzed with use of a linear mixed model (LMM) with subject as a random effect. This model deals effectively with repeated measurements, missing values, and variation in duration of follow-up (DeSouza et al. 2009). Differences between the stems were assessed by estimating the main treatment effect and the stem type × time interaction, both as an overall effect over the entire follow-up period taking the repeating measurements into account. The assessment of the interaction term allows for the investigation of possible time-varying mean differences. At the 5- and 15-year follow-up point, the mean differences were assessed using ANOVA. As a sensitivity analysis, separate adjusted analyses were carried out with age, sex, BMI, and diagnosis (primary or secondary osteoarthritis) as covariates. A p-value of <0.05 was considered to be significant (SPSS version 20.0; IBM Corp, Armonk, NY, USA). Ethics, funding, and potential conflicts of interest Approval from the institutional medical ethics committee was obtained and all the patients gave written informed consent. The Department of Orthopaedics of the LUMC received a single unrestricted grant from Biomet. In addition, funding from the European Information and Communication Technologies Community Seventh Framework Program (FP7/20072013) (grant agreement number 248693) and the Dutch Arthritis Association (project number LLP13; 08-1-300) were received in support of this study. None of these sponsors took part in the design or performance of the study, neither in the collection, management, analysis, nor in the interpretation of the data; or in preparation, review, or approval of the manuscript. None of the authors has any potential financial conflict of interest related to this manuscript.

Results Patients From May 1992 to May 1994, 75 consecutive THAs in 67 patients were assessed for inclusion and 61 THAs in 53 patients were randomized (Figure 2). In 19 THAs (19 patients)

Acta Orthopaedica 2020; 91 (6): 660–668

ENROLLMENT

663

Patients assessed for eligibility n = 67 Excluded Patients who declined participation n = 14 Patients randomized n = 53 (61 THAs) HA coated n = 15 Excluded (n = 3): – no bone markers, 2 – no RSA radiographs, 1

ALLOCATION intention-to-treat

ALLOCATION per-protocol

FA coated n = 18

Allocated to HA coating (n = 12): – received HA coating, 7 – received unknown coating, 4 – received uncoated implant, 1

Uncoated n = 28

Excluded (n = 4): – no bone markers, 2 – no RSA radiographs, 2 Allocated to FA coating (n = 14): – received FA coating, 6 – received unknown coating, 8

Excluded (n = 12): – no bone markers, 6 – no RSA radiographs, 6 Allocated to uncoated (n = 16): – received uncoated, 10 – received unknown coating, 6

HA coating (n = 7)

Unknown coating (n = 4)

FA coating (n = 6)

Unknown coating (n = 8)

Uncoated (n = 11)

Unknown coating (n = 6)

Follow-up: – mean 11.4 (SD 6.3) years – died, 4 (4, 6, 14, 17 year FU) – lost to follow-up, 1 (5 year FU) – revisions, 0

Follow-up: – mean 17.5 (SD 10.7) years – died, 1 (3 year FU) – lost to follow-up, 0 – liner revisions, 2 (11, 13 year FU) – cup revision, 1 (23 year FU)

Follow-up: – mean 18.4 (SD 7.4) years – died, 3 (5, 17, 18, 19 year FU) – lost to follow-up, 0 – liner revisions, 2 (11, 14 year FU)

Follow-up: – mean 18.3 (SD 7.9) years – died, 4 (3, 11, 18, 20 year FU) – lost to follow-up, 0 – liner revisions, 2 (12, 13 year FU)

Follow-up: – mean 17.4 (SD 3.8) years – died, 1 (17 year FU) – lost to follow-up, 0 – liner revisions, 2 (13, 15 year FU) – Cup revision (9 year FU)

RSA reference scene: – postoperative, 2 – 1 year, 7

RSA reference scene: – postoperative, 0 – 1 year, 4

RSA reference scene: – postoperative, 1 – 1 year, 5

RSA reference scene: – postoperative, 2 – 1 year, 7

Follow-up: – mean 18.1 (SD 4.1) years – died, 4 (11, 17, 17, 17 year FU) – lost to follow-up, 0 – liner revisions, 5 (13, 13, 15, 17, 18 year FU) – cup revisions, 2 ’ (1 THA, 17 year FU) RSA reference scene: – postoperative, 3 – 1 year, 10

RSA reference scene: – postoperative, 2 – 1 year, 5

Figure 2. CONSORT flowchart of patient recruitment, allocation and follow-up. THA = total hip arthroplasty; HA = hydroxyapatite; FA = fluorapatite; FU = follow-up; SD = standard deviation.

Table 2. Group characteristics at baseline. Values are count unless otherwise specified Characteristic

Hydroxy- Fluorapatite apatite Uncoated Unknown (n = 7) (n = 6) (n = 11) (n = 18)

Sex Male 3 1 5 7 Female 4 5 6 11 BMI a 23 (5.6) 28 (2.7) 24 (4.5) 25 (3.3) Age at surgery a 58 (15) 57 (13) 50 (14) 53 (9) Diagnosis Osteoarthritis 5 2 4 6 Rheumatoid arthritis 1 2 4 9 Hip dysplasia 0 1 1 1 Ankylosing spondylitis 1 1 1 0 Osteonecrosis 0 0 0 2 Perthes disease 0 0 1 0 Side Left 4 5 5 6 Right 3 1 6 12 Surgeon Consultant 7 6 10 14 Resident 0 0 1 4 Stem orientation Varus 0 1 0 1 Neutral (< 3°) 7 5 10 16 Valgus 0 0 1 1 Preoperative HHS a min 0–max 100 points 33 (18) 34 (4.5) 32 (14) 37 (15) a Values

are mean (SD).

RSA analyses could not be performed due to absence of either bone markers (n = 10) or RSA radiographs (n = 9). These 19 patients were comparable to the analyzed group with respect to sex, BMI, age at surgery, surgeon, stem orientation, and preoperative HHS (post-hoc chi-square test and unpaired Student’s t-test; p-values > 0.05). Thus, 34 patients (42 THAs) were analyzed with a mean follow-up of 17 years (SD 6.6). 7 stems were HA-coated, 6 stems were FA-coated, and 11 stems were uncoated (Table 2). 1 patient was randomized for an HAcoated stem, but instead received an uncoated stem. In 18 stems the coating was unknown due to missing implant stickers in 9 cases, missing coating details on the implant sticker in 4 cases, and missing medical (paper) records in 5 cases. 12 patients (17 THAs) died during follow-up and 1 patient (1 THA) emigrated and was subsequently lost after 5 years of follow-up. Migration Second-day postoperative RSA radiographs were available in only 10 out of 42 stems, whereas RSA radiographs at 1-year follow-up were available in 38 out of 42 stems (Figure 3). As the number of available second-day postoperative RSA radiographs was not sufficient to make meaningful comparative analyses between different coatings, the 10 stems available for direct postoperative migration measurement were analyzed as a single cohort, independent of coating, showing relatively rapid subsidence during the first postoperative year with a

664

Acta Orthopaedica 2020; 91 (6): 660–668

Y-translation (mm)

Count 45

Y-rotation (°)

0.5 Remaining THAs Available RSA radiographs

3.0 Second-day-postoperative baseline 1-year-postoperative baseline

0.4 0.3

2.0

0.2

1.5

0.1

25 20

1.0

–0.1

0.5

–0.2

–0.3

–0.5

–0.4

–1.0

–0.5 5 0

Second-day-postoperative baseline 1-year-postoperative baseline

2.5

–1.5

–0.6 0 1/81/41/2 1 2 3 4 5 6 7 8 10 12 14 16 18 20 22 24

–0.7

Years from index operation

Figure 3. Bar graph showing number of RSA radiographs available for analysis per followup point. Line graph showing number of THAs in follow-up (i.e., total minus deceased and lost to follow-up).

0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24

–2.0

Years from index operation

Figure 4. Median Y-translation (i.e., translation along the longitudinal axis) with interquartile ranges of the complete cohort during the 25 years of follow-up, using both the second-day (n = 10) and the 1-year (n = 38) postoperative RSA radiograph as a baseline.

0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24

Years from index operation

Figure 5. Median Y-rotation (i.e.,internal rotation about the longitudinal axis) with interquartile ranges of the complete cohort during the 25 years of follow-up, using the both the second-day (n = 6) and the 1-year (n = 17) postoperative RSA radiograph as a baseline.

Y-translation (mm)

0.5

0.4 0.3

HA FA Uncoated

0.3

0.2

0.1

–0.1

–0.2

–0.3

–0.4

–0.5

–0.6

–0.7

0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24

Years from index operation

Figure 6. Median Y-translation (i.e., translation along the longitudinal axis) with interquartile ranges during the 25 years of follow-up of the HA, FA, and uncoated stems, using intentionto-treat analysis (i.e., all included stems as per randomization group) and the 1-year postoperative RSA radiograph as a baseline (i.e., unknown initial migration).

HA FA Uncoated

0.4

–0.7

1-year-postoperative baseline Outlier

–0.5

–1.0

–1.5

0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24

Years from index operation

Figure 7. Median Y-translation (i.e., translation along the longitudinal axis) with interquartile ranges during the 25 years of follow-up of the HA, FA, and uncoated stems, using per protocol analysis (i.e., only stems with verified coating) and the 1-year postoperative RSA radiograph as a baseline (i.e., unknown initial migration).

median subsidence of 0.2 mm (–0.1 to 0.6) at 12 months, followed by stable subsidence during the remaining follow-up period (Figure 4). During the period of initial subsidence there were also relatively large rotations of these stems in the horizontal plane, which stabilized after 1 year (Figure 5). The 38 stems available for migration measurement using the 1-year postoperative RSA radiograph as baseline were analyzed both as a single cohort, and in a comparative manner using different coatings. Migration of the overall cohort showed rather stable subsidence and rotation until 14 years’ follow-up, after which the migration patterns began to show large variabil-

–2.0

0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24

Years from index operation

Figure 8. Median Y-translation (i.e., translation along the longitudinal axis) with interquartile ranges of the complete cohort during the 25 years of follow-up with the influential outlier excluded and shown separately.

ity (Figures 4 and 5). Also, from this time period onward the number of patients attending the RSA outpatient clinic began to decline (Figure 3). For the comparative migration analyses (i.e., HA vs. FA vs. uncoated stems), both intention-to-treat (ITT) and per-protocol (PP) analyses were performed. The ITT analyses reflect the best-case scenario, using allocation as per randomization group of 42 THAs (including both 24 THAs with verified and 18 THAs with unknown coating) (Figure 6), whereas per-protocol (PP) analyses reflect the genuine effect of the different coatings, including only the 24 THAs with verified coating and excluding the 18 THAs with unknown coating

Acta Orthopaedica 2020; 91 (6): 660–668

665

Table 3. Stem migration during 25 years of follow-up: per-protocol analysis (i.e., only stems with verified coating), using the 1-year postoperative RSA radiograph as a baseline (i.e., unknown initial migration). Values are count, median (range)

Subsidence in 1 stem did not stabilize (Figure 8). This stem was randomized for no coating, but as the implant sticker was FU year Hydroxyapatite Fluorapatite Uncoated p-value missing this could not be veriLongitudinal translation, mm a fied. At the last available radio 2 7 0.04 (–0.19 to 0.32) 5 0.20 (–0.02 to 0.27) 8 –0.03 (–0.16 to 0.12) 0.1 b graph after 17 years’ follow-up 5 2 –0.06 (–0.07 to –0.04) 5 0.09 (–0.16 to 0.44) 6 0.00 (–0.12 to 0.12) 0.6 c there was evidence of subsid 10 2 0.04 (0.02 to 0.06) 4 0.07 (–0.08 to 0.12) 7 –0.07 (–0.41 to 0.37) ence, and around the tip of the 15 3 0.07 (–0.16 to 0.44) 3 0.04 (–0.15 to 0.13) 5 –0.18 (–0.52 to 0.12) 0.5 c 20 0 – 1 – 1 – stem radiolucencies and pedes 25 0 – 1 – 1 – tal formation were noticeable. Internal rotation, degrees However, the HHS remained 2 3 –0.21 (–1.35 to 0.24) 2 –1.02 (–1.23 to –0.81) 3 –0.50 (–0.76 to 0.25) 0.5 b 5 1 – 2 –0.53 (–0.74 to –0.32) 2 –0.10 (–0.48 to 0.28) – higher than 90 points during 10 1 – 2 –0.14 (–0.22 to –0.05) 2 –0.33 (–0.52 to –0.13) follow-up. 15 0 – 2 1.08 (–0.46 to 2.62) 2 –0.02 (–0.45 to 0.42) – 20 0 – 25 0 –

0 – 0 –

a Negative values correspond to subsidence b Main effect, per protocol c Prespecified time point, per protocol

0 – 0 –

(i.e., distal migration).

Table 6. Harris Hip Score (min 0 – max 100 points) during 25 years of follow-up: perprotocol analysis (i.e., only stems with verified coating). Values are count, median (range)

Hydroxyapatite

Preoperative Year 2 Year 5 Year 10 Year 15 Year 20 Year 25

5 35 (8–57) 2 81(65–96) 2 100 (99–100) 4 86 (72–94) 4 85 (41–99) 0 – 0 –

a Main effect, per protocol b Prespecified time point, per

Fluorapatite 3 2 3 5 5 2 1

34 (30–39) 90 (89–91) 76 (69–84) 73 (62–90) 76 (61–90) 63 (57–69) –

Uncoated p-value 9 4 5 7 6 2 1

37 (8–54) 0.6 a 95 (85–100) 85 (66–100) 86 (61–98) 79 (74–90) 0.9 b 83 (82–83) –

protocol

(Figure 7). Post-hoc verification revealed adequate randomization in both ITT and PP analyses as the 3 different coatings coating groups were comparable with respect to sex, BMI, age at surgery, surgeon, stem orientation, and preoperative HHS (post-hoc chi-square test and 1-way ANOVA; p-values > 0.05). Overall 25-year migration did not reveal a significant difference among the 3 different coatings in both the ITT and PP analyses (LMM; p-values > 0.05; Table 3; Tables 4 and 5 in Supplementary data). Furthermore, we could not find a significant difference in time to stabilization and subsequent migration, that is, no evidence of interaction (coating type × time interaction; LMM; p-values > 0.05; Tables 4 and 5 in Supplementary data). Migration of the stems was comparable among the 3 coatings at the prespecified time points of 5 years and 15 years postoperatively (1-way ANOVA; p-values > 0.05; Table 3). The results of the adjusted analyses were comparable to the results from the unadjusted analyses and age, sex, diagnosis, and BMI did not significantly influence migration.

Clinical outcome We could not find a significant difference in HHS among the groups during follow-up in both the PP and ITT analyses (LMM; p-values > 0.05; Table 6; Tables 7 and 8 in Supplementary data). Between-group differences did not change significantly over time (coating type × time interaction; LMM; p-values > 0.05; Tables 7 and 8 in Supplementary data). Adjusted analyses for age, sex, diagnosis, BMI, and postoperative HHS gave similar results. Furthermore, comparing the HHS at the 15-year follow-up point did not yield any statistically significant differences (1-way ANOVA; p-values > 0.05; Table 6).

Survival None of the stems were revised during followup. There were 13 liner revisions in 13 THAs due to wear. In 1 THA (uncoated) with a liner revision 13 years after follow-up, the cup was revised due to aseptic loosening 20 years after follow-up and subsequently re-revised a couple weeks later due to malpositioning. In addition, 2 more cups in 2 THAs (coating unknown) were revised due to aseptic loosening. During follow-up, 12 patients (17 THAs) died due to causes unrelated to the THA. One patient (1 THA) was lost to follow-up due to emigration, the fate of this THA could not be determined.

Discussion We found stable migration and thus fixation of the cementless Mallory-Head Porous stem over a period of 25 years. After initial migration, all but 1 of 42 stems stabilized after 1-year follow-up and there were no stem revisions. 4 cups were revised due to aseptic loosening and 13 liners were revised due to wear. Comparative analyses did not yield a dif-

666

ference in migration and clinical scores among HA, FA, and uncoated cementless stems. Furthermore, there was no difference in time to stabilization between coated and uncoated stems, thus excluding a delamination problem occurring later in follow-up. To our knowledge, this is the first long-term RSA study with over 20 years’ follow-up, and the first RSA study comparing migration of HA, FA, and uncoated stems. Only a few RSA studies describe migration beyond 10-year follow-up. Sesselman et al. (2018) reported migration of 26 cementless Cerafit stems with a follow-up of 10 years. They found a median subsidence of 0.01 mm at 2-year and 0.09 mm at 10-year follow-up, with most of the subsidence occurring during the first 6 postoperative weeks. Critchley et al. (2020) reported migration of 30 cementless Corail stems with a follow-up of 14 years. They found a mean subsidence of 0.62 mm at 2-year and 0.7 mm at 14-year follow-up, with initial rapid subsidence over 6 weeks and subsequent stabilization. In our study subsidence was 0.15 mm at 2-year, 0.3 mm at 10-year, and 0.1 mm at 14-year follow-up, with most of the subsidence occurring during the first 2 postoperative years. These studies, using different cementless stem designs, all show initial migration of different magnitude with subsequent stabilization, emphasizing the influence of design features on migration, and thus fixation. Søballe et al. (1993) and Kärrholm et al. (1994b) performed RSA studies comparing the migration of HA-coated stems with stems without bioactive coating. Søballe et al. (1993) found more migration (MTPM) of uncoated titanium stems compared with HA-coated stems, although subsidence was comparable after 1-year follow-up with a mean of 0.09 mm for the HAcoated stems. Kärrholm et al. (1994b) compared migration of HA-coated stems with cemented and cementless stems over a 2-year period. They found a median of 0.05 mm proximal migration of HA-coated stems compared with 0.12 mm subsidence of cemented and 0.1 mm of cementless stems. As these differences were statistically significant both authors concluded that HA seems to enhance early fixation. In our study we could not find a benefit of HA on long-term fixation. There might be a benefit of HA during the first postoperative year. However, due to insufficient data we were unable to find such a difference. Our study shows that once a stem has stabilized there is no difference in migration among the different coatings. Several reviews have been published comparing HAcoated implants with uncoated implants. Gandhi et al. (2009) reported no difference in aseptic loosening, or in HHS. Goosen et al. (2009) reported no difference in HHS, endosteal bone ingrowth, and radiolucent lines. Li et al. (2013) could not find a difference in HHS, or radioactive lines, and Chen et al. (2015) reported no benefit of HA in terms of survivorship, but HA-coated implants showed better postoperative HHS and less femoral osteolysis during follow-up. A recent registry study found an overall lower risk of stem revision for any reason for HA-coated stems compared with a non-HA coated stem; however, the rate for stem revision for

Acta Orthopaedica 2020; 91 (6): 660–668

HA-coated Mallory-Head stems was higher compared with the non-HA-coated counterpart (0.11% vs. 0.02%) (Inacio et al. 2018). This study suggests that longevity of implants might be related more to specific implant design than to type of coating. Except for 1 study performed at our institution, there are no RSA studies evaluating the migration of the Mallory-Head Porous stem. Van der Voort et al. (in submission) showed median subsidence of 0.2 mm (range 0.4–4.8) at 5 years’ follow-up of uncoated stems. In the current study we found the same median subsidence at 5 years’ follow-up but the range in this study was considerably smaller, with the largest subsidence being only 0.4 mm. Insufficient data on initial migration during the first postoperative year in this study might explain this difference. The cementless Mallory-Head Porous stem has an excellent 10-year survival record with 48 stem revisions of 5,932 primary THAs in the Dutch Arthroplasty Register (LROI 2019) and 27 stem revisions of 3,303 primary THAs in the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR 2019). All stems in our study showed stable subsidence, except for 1. This stem, with an untraceable coating, was inserted because of severe osteoarthritis at the age of 61 years. There was no preoperative template available but the postoperative radiograph showed a varus position of the stem with insufficient contact with the lateral cortex at the metaphysis, suggesting undersizing. Albeit that initial subsidence was unknown, the stem showed progressive subsidence from the 1-year follow-up onwards. On the 17-year follow-up radiograph there was obvious subsidence visible, next to radiolucencies and pedestal formation. Remarkably, this patient never scored less than 90 points on the HHS scale and at the 17-year followup moment the patient was asymptomatic, although, aged 78 years, she walked only about 200 meters. Overall, there was rather stable migration up to 14 years reaching a plateau phase, but thereafter migration patterns began to show greater variability, which especially holds for subsidence. From then onwards, patient attendance for regular follow-up also decreased dramatically, resulting in only 5 stems being available for analysis at 20 years’ follow-up. This low number of available, analog RSA radiographs in combination with relative low precision compared with the modern RSA technique are the most probable reasons for the great variability in stem migration patterns (Valstar et al. 2000). Furthermore, as most stems available for analyses beyond 20 years were FA coated, it could be reasoned that in this selected group of patients either overall subsidence increased or the FA coating broke down after more than a decade, leading to increased subsidence. However, the latter cannot be substantiated as no data on the non-coated control group was available. Additionally, the increasing retroversion, together with to increasing subsidence, might be related to a more sedentary lifestyle of these slightly older patients. For that matter, standing up from a chair creates a retroversion force at the neck of the femoral stem.

Acta Orthopaedica 2020; 91 (6): 660–668

This was the first RSA study performed at our institution, therefore it gave some insights into initial study set-up logistics. There was experience neither with inserting tantalum markers in periprosthetic bone, nor with the validity of the instrument used at that time to insert markers, which turned out to skip 1 out of 4 markers. This was noticed only after RSA radiographs were evaluated in too late a postoperative period. For that reason a novel tantalum marker inserter was developed. Additionally, there was a lack of the expertise needed for optimal logistics concerning RSA radiographs. Initially, a single researcher (RN) took care of study logistics and RSA radiograph analyses without secretarial assistance. Hence, patients missing a follow-up moment were noticed only weeks later. These technical and logistical shortcomings resulted in the exclusion of 19 THAs. Furthermore, stem allocation to the randomization groups was not adequately documented and implant stickers of the manufacturer were missing or lacking essential information. The latter might have been related to the distinctive manufacturing process for stems in the current study; the attachment of 3 RSA markers and applying 3 different coatings might have interfered with regular application of implant stickers. This study has several limitations. First, there were too few migration measurements available during the first postoperative year to make meaningful analyses using the second-day postoperative radiograph as baseline, which is the conventional manner to calculate migration over time. However, as stems susceptible to failure will typically show progressive migration, using the 1-year postoperative RSA radiograph will also detect stems prone to failure. Second, the given coating could not be verified in 18 stems due to due to missing or insufficient implant stickers. To overcome this problem of unknown coatings, both per-protocol and intention-to-treat analyses were performed. The latter reflects the base case scenario, assuming all stems received the coating as per randomization. This is a plausible assumption as only 1 of 24 known coatings received a different coating as per randomization. Third, the Mallory-Head Porous stem is nowadays seldom used, limiting the clinical applicability of this study. In conclusion, this study could not establish a beneficial effect at long-term follow-up of bioactive coatings on migration, and thus fixation, in this type of stem. Neither could delamination of the less thermostable HA coating be proven. This study provides value migration data that can be used to establish an acceptable migration pattern of cementless stems with which new stem designs can be compared. Supplementary data Tables 4–5 and 7–8 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2020.1840021

667

RGN designed the study. RGN, PV and MKN collected the clinical data. ERV, BLK, and PV collected the RSA data. PV and BLK performed the RSA analyses. PV and MF analyzed the data. PV and RGN wrote the initial draft manuscript. Critical revision of the manuscript was done by all authors. Edward Valstar passed away in the analysis phase of this study but was significantly involved in data collection. We remember Edward Valstar as an inspiring mentor, dedicated researcher and foremost a warm personality (Nelissen et al. 2017). Acta thanks Leif Ryd and Stephan Maximilian Röhrl for help with peer review of this study. AOANJRR. Australian Orthopaedic Association National Joint Replacement Registry, Annual Report 2019. https://aoanjrr.sahmri.com/. Bauer T W. Hydroxyapatite: coating controversies. Orthopedics 1995; 18(9): 885-8. Bloebaum R D, Beeks D, Dorr L D, Savory C G, DuPont J A, Hofmann A A. Complications with hydroxyapatite particulate separation in total hip arthroplasty. Clin Orthop Relat Res 1994; (298): 19-26. Chen Y-L, Lin T, Liu A, Shi M-M, Hu B, Shi Z-l, Yan S-G. Does hydroxyapatite coating have no advantage over porous coating in primary total hip arthroplasty? A meta-analysis. J Orthop Surg Res 2015; 10(1): 21. Critchley O, Callary S, Mercer G, Campbell D, Wilson C. Long-term migration characteristics of the Corail hydroxyapatite-coated femoral stem: a 14-year radiostereometric analysis follow-up study. Arch Orthop Trauma Surg 2020; 140(1): 121-7. DeSouza C M, Legedza A T, Sankoh A J. An overview of practical approaches for handling missing data in clinical trials. J Biopharm Stat 2009; 19(6): 1055-73. Dhert W, Klein C, Jansen J, Van der Velde E, Vriesde R, Rozing P, De Groot K. A histological and histomorphometrical investigation of fluorapatite, magnesiumwhitlockite, and hydroxylapatite plasma–sprayed coatings in goats. J Biomed Mater Res 1993; 27(1): 127-38. Furlong R, Osborn J. Fixation of hip prostheses by hydroxyapatite ceramic coatings. J Bone Joint Surg Br 1991; 73(5): 741-5. Gandhi R, Davey J R, Mahomed N N. Hydroxyapatite coated femoral stems in primary total hip arthroplasty: a meta-analysis. J Arthroplasty 2009; 24(1): 38-42. Geesink R. Experimental and clinical experience with hydroxyapatite-coated hip implants. Orthopedics 1989; 12(9): 1239-42. Goosen J, Kums A, Kollen B, Verheyen C. Porous-coated femoral components with or without hydroxyapatite in primary uncemented total hip arthroplasty: a systematic review of randomized controlled trials. Arch Orthop Trauma Surg 2009; 129(9): 1165-9. Hailer N P, Lazarinis S, Mäkelä K T, Eskelinen A, Fenstad A M, Hallan G, Havelin L, Overgaard S, Pedersen A B, Mehnert F. Hydroxyapatite coating does not improve uncemented stem survival after total hip arthroplasty! An analysis of 116,069 THAs in the Nordic Arthroplasty Register Association (NARA) database. Acta Orthop 2015; 86(1): 18-25. Inacio M C, Lorimer M, Davidson D C, De Steiger R N, Lewis P L, Graves S E. What is the risk of revision surgery in hydroxyapatite-coated femoral hip stems? findings from a large national registry. Clin Orthop Relat Res 2018; 476(12): 2353. Kärrholm J, Borssen B, Lowenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 4–7-year stereoradiographic follow-up of 84 cemented prostheses. J Bone Joint Surg Br 1994a; 76(6): 912-7. Kärrholm J, Malchau H, Snorrason F, Herberts P. Micromotion of femoral stems in total hip arthroplasty: a randomized study of cemented, hydroxyapatite-coated, and porous-coated stems with roentgen stereophotogrammetric analysis. J Bone Joint Surg Am 1994b; 76(11): 1692705.

668

Klein C, Wolke J, de Blieck–Hogervorst J, De Groot K. Calcium phosphate plasma–sprayed coatings and their stability: an in vivo study. J Biomed Mater Res 1994; 28(8): 909-17. Kurtz S M, Lau E, Ong K, Zhao K, Kelly M, Bozic K J. Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030. Clin Orthop Relat Res 2009; 467(10): 2606-12. Lazarinis S, Kärrholm J, Hailer N P. Effects of hydroxyapatite coating on survival of an uncemented femoral stem: a Swedish Hip Arthroplasty Register study on 4,772 hips. Acta Orthop 2011; 82(4): 399-404. Li S, Huang B, Chen Y, Gao H, Fan Q, Zhao J, Su W. Hydroxyapatite-coated femoral stems in primary total hip arthroplasty: a meta-analysis of randomized controlled trials. Int J Surg 2013; 11(6): 477-82. LROI. Dutch Arthroplasty Register, Annual Report 2019. http://www.lroirapportage.nl/. Lugscheider E, Weber T F. Production of biocompatible coatings of hydroxyapatite and fluorapatite. In: Thermal Spray Technology, Proceedings of the National Thermal Spray Conference; 1989. p. 337-43. Nelissen R, Kaptein B, Veeger D. Edward Valstar (1970–2017). Acta Orthop 2017; 88(6): 701. Paulsen A, B Pedersen A, Johnsen S P, Riis A, Lucht U, Overgaard S. Effect of hydroxyapatite coating on risk of revision after primary total hip arthroplasty in younger patients: findings from the Danish Hip Arthroplasty Registry. Acta Orthop 2007; 78(5): 622-8. Pijls B G, Valstar E R, Kaptein B L, Fiocco M, Nelissen R G. The beneficial effect of hydroxyapatite lasts: a randomized radiostereometric trial comparing hydroxyapatite-coated, uncoated, and cemented tibial components for up to 16 years. Acta Orthop 2012; 83(2): 135-41.

Acta Orthopaedica 2020; 91 (6): 660–668

Ranstam J, Ryd L, Önsten I. Accurate accuracy assessment. Acta Orthop 2000; 71(1): 106-8. Sesselmann S, Hong Y, Schlemmer F, Hussnaetter I, Mueller L A, Forst R, Tschunko F. Radiostereometric migration measurement of an uncemented Cerafit® femoral stem: 26 patients followed for 10 years. Biomed Tech (Berl) 2018; 63(6): 657-63. Søballe K, Toksvig-Larsen S, Gelineck J, Fruensgaard S, Hansen E, Ryd L, Lucht U, Bunger C. Migration of hydroxyapatite coated femoral prostheses: a Roentgen stereophotogrammetric study. J Bone Joint Surg Br 1993; 75(5): 681-7. Valstar E, Vrooman H, Toksvig-Larsen S, Ryd L, Nelissen R. Digital automated RSA compared to manually operated RSA. J Biomech 2000; 33(12): 1593-9. Valstar E R, Gill R, Ryd L, Flivik G, Börlin N, Kärrholm J. Guidelines for standardization of radiostereometry (RSA) of implants. Acta Orthop 2005; 76(4): 563-72. Van der Voort P, Pijls B G, Nieuwenhuijse M J, Jasper J, Fiocco M, Plevier J W, Middeldorp S, Valstar E R, Nelissen R G. Early subsidence of shapeclosed hip arthroplasty stems is associated with late revision: a systematic review and meta-analysis of 24 RSA studies and 56 survival studies. Acta Orthop 2015; 86(5): 575-85.

Acta Orthopaedica 2020; 91 (6): 669–674

669

HipSim — hip fracture surgery simulation utilizing the Learning Curve–Cumulative Summation test (LC-CUSUM) Jan Duedal RÖLFING 1–3, Rune Dall JENSEN 1,2, and Charlotte PALTVED 1,2 1 Corporate

HR, MidtSim, Central Denmark Region; 2 Department of Clinical Medicine, Aarhus University; 3 Department of Orthopaedics, Aarhus University Hospital, Denmark Correspondence: jan.roelfing@clin.au.dk Submitted 2020-03-20. Accepted 2020-05-14.

Background and purpose — Virtual reality simulation of hip fracture surgery is available for orthopedic residents nationwide in Denmark. Summative assessment of learning applying the learning curve cumulative summation test (LC-CUSUM) has not been utilized in orthopedic simulation training. The strength of the LC-CUSUM is that it assumes incompetency and signals competency based on solid statistics. We investigated the LC-CUSUM characteristics of novices stepwise mastering the simulated dynamic hip screw (DHS) procedure. Material and methods — 32 1st-year orthopedic residents participated in HipSim and its 3 subsequent LCCUSUM evaluations: placing a Kirschner wire, placing a Kirschner wire in different patients, and performing the entire DHS procedure in different patients. The career status of the participants, i.e., still working in orthopedics or in another specialty was recorded ≥ 2 years after participation and associated with the simulation performance (passed/ failed). Results — 13/14 participants passing HipSim according to LC-CUSUM were still working in orthopedics, while 9/18 participants failing HipSim had quit orthopedics at ≥ 2 years follow-up. The simulator-generated feedback did not statistically significantly differ between the groups. Interpretation — LC-CUSUM and its summative pass/ fail assessment of each simulation was feasible in this formative simulation program. Clinical educators can be reassured that participants passing HipSim are likely to continue to 2nd–5th year of residency, while failing HipSim should raise concerns and trigger career counselling and clinical supervision. The motivational aspect of LC-CUSUM pass/ fail assessment when designing formative simulation training warrants further research.

The high mortality rate after hip fractures is caused by a multitude of factors including comorbidities, the effects of immobilization due to the fracture, and suboptimal surgical treatment and rehabilitation (Brauer et al. 2009, Röck et al. 2019). Ideally, a hip fracture operation will allow the patient to immediately fully weight-bear. Intraoperative adverse events and insufficient stabilization of the fracture significantly prolong recovery and worsen the prognosis. Thus, high-quality surgical training is often emphasized as key. Virtual reality (VR) simulation training has been introduced to address the issue of training, and previous studies have demonstrated that VR training can ensure basic proficiency (Pedersen et al. 2014, Gustafsson et al. 2019). Consequently, VR training in hip fracture surgery is available to 1st-year orthopedic residents in Denmark at an early stage of their orthopedic career and before applying for and enrolling in the 2nd–5th years of orthopedic residency. Since the Learning Curve–Cumulative Summation test (LCCUSUM) was first described by David Biau et al. (2008), it has gained popularity in monitoring skill acquisition and quality control of surgical performance when learning new procedures. In orthopedics, LC-CUSUM has been applied to signal competency of the surgeon in a multitude of procedures, e.g., spinal laminectomy, and total hip and knee replacement surgery (Lee et al. 2014, Zhang et al. 2014, Park et al. 2019). We present the 1st study applying LC-CUSUM in orthopedic simulation training. The major strength of the LC-CUSUM is that it assumes incompetency and can signal competency based on solid statistical methods. Consequently, LC-CUSUM could prove beneficial to determine thresholds in simulation-based training, e.g., if the null hypothesis of an inadequate performance is rejected, the learner can be allowed to continue the learning process in the clinical setting (summative assessment

670

Acta Orthopaedica 2020; 91 (6): 669–674

fulfilment of the < 10 procedure criterion, thus all residents are expected to participate in HipSim training, which takes place during working-hours. Consequently, selection bias is unlikely. Retrospectively, we divided the participants into 2 cohorts depending on whether or not they were pursuing a career in orthopedics in January 2020 (e.g. enrolment in 2nd–5th year of orthopedic residency or orthopedic PhD program).

Figure 1. The haptic-feedback virtual reality simulator (TraumaVision, Swemac Simulation AB, Sweden).

of learning). If this holds true, simulation-based LC-CUSUM training programs can facilitate the education of young surgeons, who can safely learn to master the initial steps of the learning curve without imposing risk on patients. We describe the LC-CUSUM characteristics of DHS simulation training (HipSim) of 1st-year orthopedic residents and associate the results with the participants’ career status after ≥ 2 years follow-up.

Methods Simulator The construct validity of the virtual reality (VR) simulator (TraumaVision, Swemac Simulation AB, Sweden) has been established, i.e., the simulator performance correlates with surgical experience (Pedersen et al. 2014, Akhtar et al. 2015, Gustafsson et al. 2019). Notably, the simulator can discriminate between novices (< 10 DHS), intermediates, and experts (Akhtar et al. 2015). The simulator provides the opportunity to train the basic skills required for the clinical DHS procedure, e.g., hand–eye coordination and the ability to work in 3 dimensions based on 2-dimensional visual information and haptic feedback (Figure 1). Participants The DHS simulation training was available for all 1st-year orthopedic residents who had performed fewer than 10 hip fracture-related surgeries under supervision (cannulated screws, DHS, intramedullary nails, hemi-/total hip arthroplasties) from Central Denmark Region and Northern Denmark Region from November 2016 until January 2018. All departments have written agreements to send all their residents in

Simulation program and LC-CUSUM design The simulation was designed as a time-dispersed masterylearning program with 2–3 training days of a maximum 4 hours of training per day. The DHS simulation program introduced participants stepwise to the surgical procedure. At competency level 0 (CL0), participants had to become acquainted with the simulator and master the key step of placing a Kirschner wire (K-wire) in the fractured left hip of the same patient repetitively. Subsequently, the compexity was increased at CL1 and CL2. Clinical variability was introduced at CL1 with up to 24 different scenarios (different patients and fracture patterns), and the entire DHS procedure was introduced at CL2. Here, the participants also had to choose the optimal DHS angle for the given patient, insert the K-wire, ream and insert the sliding screw, and fixate the plate to the femoral shaft with a single bicortical screw. Advancement to the next CL was granted when LC-CUSUM signaled competency. Pass/fail criteria In order to apply the LC-CUSUM to any procedure, the procedure has to be classified as “passed” or “failed”. Hence, passing/failing criteria had to be established: a tip apex distance (TAD) of more than 20 mm is a scientifically validated predictor of failure of internal fixation (Baumgaertner et al. 1995, De Bruijn et al. 2012). Other failing criteria were, for instance, more than 3 attempts to place the K-wire appropriately or the breach of cortical bone with either the K-wire or the reamer. Besides the immediate simulator-generated feedback of passing/failing a procedure, secondary feedback data included the passing criteria: TAD ≤ 20 mm, center–center or center– inferior placement, no violation of cortical bone, and no more than 3 attempts to place the K-wire. For a complete list of passing/failing criteria for CL0 (= CL1), and CL2 respectively, please see Figure 2, Supplementary data. Furthermore, information regarding the total time, fluoroscopy time, and number of radiographs were provided, but did not influence whether the procedure was passed or failed. Learning Curve–Cumulative Summation test (LC-CUSUM) The statistical design of the LC-CUSUM test was based on personal correspondence with D. Biau. The adequate performance level was set at 10% failure and the equivalence zone (delta) at 5%, i.e., the acceptable deviation from adequate performance. An in-control limit, h = 0.74, was chosen to

Acta Orthopaedica 2020; 91 (6): 669–674

671

Simulator-generated feedback of passed and failed 1st-year orthopedic residents (n = 32) subdivided into the two cohorts: career in orthopedics / career in other specialty 2 years after HipSim training. Values are mean (95% CI) unless otherwise specified

Passed (n = 14) Failed (n = 18) Career in Career in Career in Career in orthopedics other specialty orthopedics other specialty n = 13 n = 1 n = 9 n=9

Training sessions (days) a 2 (1–3) Total simulation time (min.) 196 (152–241) Mean time of last 3 simulations (s) at CL0 66 (47–85) at CL1 77 (60–94) at CL2 206 (168–244) Mean total simulations (n) 89 (73–106) Simulations at CL0 (n) 23 (16–31) Simulations at CL1 (n) 33 (24–41) Simulations at CL2 (n) 33 (25–41) Mean TAD b of last 3 simulations at CL2 (mm) 13 (12–15) Mean cortical drill outside femoral shaft (mm) 6 (5–7) Mean fluoroscopy time of last 3 simulations (s) at CL0 9 (2–45) at CL1 13 (5–49) at CL2 31 (9–55) Mean radiographs of last 3 simulations (n) at CL0 21 (10–37) at CL1 22 (7–51) at CL2 36 (15–91)

1 157

2 (1–3) 186 (119–252)

2 (1–3) 170 (123–217)

52 85 230 108 46 45 17

71 (58–86) 66 (37–95) 199 (82–315) 109 (66–150) 30 (22–49) 38 (25–50) 19 (3–35)

77 (59–95) 109 (76–142) 208 (135–282) 79 (61–96) 32 (22–41) 29 (22–36) 21 (1–41)

13 (10–17)

15 (11–19)

8 (5–10)

7 (5–19)

7 23 46

5 (2–18) 9 (2–27) 28 (11–63)

10 (2–42) 13 (4–50) 24 (8–41)

17 22 26

24 (4–37) 15 (4–28) 41 (17–74)

21 (17–30) 32 (11–37) 44 (22–55)

a median (range) b TAD: tip–apex distance

give a probability of declaring competency if the participant’s true performance is adequate (true positive, akin to power) of 91% and a probability of declaring competency if the participant’s false performance is inadequate (false positive, akin to type I error) of 12% over 50 procedures. The score gain for a successful procedure was +0.057 and the loss for a failure was –0.405. In order to make LC-CUSUM easier to comprehend it was slightly modified by setting the in-control limit (h) to 13, the score gain for a successful procedure to +1 and the loss for a failure to –7. Consequently, 13 successful procedures in a row will reject the null hypothesis of inadequate performance at the given CL 0, 1, or 2, respectively. Thus, a minimum number of 39 procedures was required to complete all 3 CLs, while the maximum number of procedures was capped at 50 procedures for each level. The LC-CUSUM design dictates that a single failure during the learning process causes the subtraction of 7 points on the LC-CUSUM score. This may seem rather punitive, as a single failure thus requires up to 7 successful procedures afterwards in order to regain the same LC-CUSUM score as before the failure. However, it is statistically required to be able to ensure the high standard of adequate performance among successful participants. Furthermore, it should be highlighted that the responsiveness of the LC-CUSUM to learning is preserved, because

early failures during learning are not punished as hard as failures at later stages during the learning process, e.g., the LC-CUSUM score cannot be negative and is bound by 0–13 for CL0, 13–26 for CL1, and 26–39 for CL2 (Biau et al. 2008). Statistics Data are presented as mean (95% confidence interval, CI) or median (min–max). Both a chi-square test and Fisher’s exact test were used to analyze the 2 x 2 contingency table (passed/failed x orthopedics/other). 1-way ANOVA with uncorrected Fisher’s least significant difference test for multiple group comparison was applied to test for statistically significant different means between the three groups (passed orthopedics, failed orthopedics, and failed other specialty) according to Table. A p-value ≤ 0.05 was considered statistically significant; however, predominantly CI is given instead of p-values according to Acta Orthopaedica author guidelines.

Ethics, funding, and potential conflicts of interest Ethical approval (no. 251/2016) was granted by the Ethical Committee, Central Denmark Region on November 23, 2016. Funding was granted by the Central Denmark Region. No external funding was obtained. The current HipSim version of TraumaVision, Swemac Simulation AB, Sweden was programmed by engineers from Swemac Simulation AB according to suggestions by JDR. No direct or indirect financial contributions were received; thus, all authors declare no conflicts of interest.

Results 32 1st-year orthopedic residents fulfilled the inclusion criteria (< 10 hip fracture surgeries before HipSim participation and subsequently at least 2 years follow-up). 22 participants were enrolled in an orthopedic residency or an orthopedic PhD program at latest follow-up. 10 participants changed to another medical specialty and were no longer working within orthopedics at latest follow-up. There was no statistically significant difference in the male/ female ratio, age, and hand dominance between 1st-year orthopedic residents who continued or quit a career in orthopedics. Furthermore, there were no statistically significant

672

Acta Orthopaedica 2020; 91 (6): 669–674

LC-CUSUM score (passed; n = 14) 39 Competency level 2

CL2n = 14

26 Competency level 1

CL1n = 14

13 Competency level 0

CL0n = 14

Continued career in orthopedics Quit career in orthopedics 50

100

Discussion 150

Simulations

LC-CUSUM score (failed; n = 18, 0 at CL0, 6 at CL1 and 12 at CL2) 39 Competency level 2

CL2n = 0

26 Competency level 1

CL1n = 12

CL0n = 18

Competency level 0

100

predictive value was 50% (CI 29–71). Accordingly, the sensitivity was 59% (CI 39–77) and the specificity was 90% (CI 60–99). For details regarding the number of simulations, simulation time, use of fluoroscopy and radiographs, and tip apex distance, see Table. None of the differences in these outcome measures reached statistical significance. However, a higher number of simulations (mean difference 32 [CI -3–67]) was noted among failed participants still working in orthopedics at follow-up vs. failed participants who had quit orthopedics at 2 years follow-up.

150

Simulations

Figure 3. Learning Curve–Cumulative Sum (LC-CUSUM) chart illustrating passing (+1 LC-CUSUM score) and failing (–7 LC-CUSUM score) of each simulation of the 14 participants passing competency levels 1–3 (CL; upper panel) and 18 participants failing at one of the competency levels. 13/14 passed and 9/18 failed participants still pursued a career in orthopedics at ≥ 2 years’ follow-up.

differences in these characteristics between participants who passed or failed HipSim according to LC-CUSUM. 14/32 participants reached an LC-CUSUM score of 39 and thereby passed the simulation program according to the LCCUSUM criteria (max. 50 simulations at each of the 3 CLs to obtain an LC-CUSUM score of 13 and advance to the next CL; Figure 3). 13/14 of the participants who passed the simulation program are currently enrolled in an orthopedic residency program, while only 9/18 of the participants who failed the LCCUSUM test still pursued a career in orthopedic surgery at 2 years follow-up (p = 0.02). Hence, the positive predictive value of passing HipSim and continue to the 2nd–5th-year orthopedic residency was 93% (CI 69–100) and the negative

Overall less than half of all participants (14/32) passed HipSim according to the LC-CUSUM criteria. The positive predictive value of pursuing a career in orthopedics was 93%. However, the majority of participants failing HipSim, but still pursuing a career in orthopedics more than 2 years after the simulation program, were persistent and continued simulation training until they reached the predefined passing LC-CUSUM score of 39, but with more than the maximum allowed simulations, while 9/10 of participants working in other medical specialties 2 years after training failed according to LC-CUSUM. In contrast to previous studies with the simulator, this study is the first to include clinical variability in the simulation program. Previous studies have considered the validity of the simulator and explored learning curves of repetitively operating on the same side of a single patient (Pedersen et al. 2014, Akhtar et al. 2015, Sugand et al. 2015, Gustafsson et al. 2019). We strongly believe that the introduction of clinical complexity is essential to exploit the full learning potential of the VR simulation program. Implications for orthopedic simulation training of the summative LC-CUSUM (pass/fail) assessment One of the participants failed the LC-CUSUM (within the maximal number of 50 procedures per competency level), but was allowed to continue simulation training and performed no less than 239 procedures before reaching the LC-CUSUM score of 3 × 13 = 39 (Figure 4, see Supplementary data). The learning curve of this participant underlines the aim of HipSim to educate and ensure acquisition of basic technical skills rather than providing a summative (pass/fail) assessment of the participant. This participant enrolled in an orthopedic residency program and continuously receives excellent ratings of his/her clinical performance including osteosynthesis of hip fractures. Thus, HipSim is a formative simulation-based training program, but utilizing the strength of LCCUSUM with a summative evaluation of each simulation. Consequently, the results of HipSim should not and cannot be used as a summative assessment of the participant’s capa-

Acta Orthopaedica 2020; 91 (6): 669–674

bilities to become a competent surgeon (Aucar et al. 2005, Strandbygaard et al. 2017). 9/18 of the participants failing HipSim at an early stage of their 1st year in orthopedics did not continue their career in orthopedics and did not enroll in 2nd–5th year residency or an orthopedic PhD program. These results may prove useful for orthopedic educators in the future. Passing HipSim may be a confounder for the determination of the participant to become an orthopedic surgeon. Hence, passing the LC-CUSUM simulation program may reassure clinical supervisors of the potential of their residents, whereas failing the simulation program may draw the educators’ attention to career counselling in order to help their first-year residents in their decision-making on whether to pursue a career in orthopedics or not. This hypothesis is supported by the observation that some of the junior doctors currently employed in other specialties did not return for the second training day compared with the more determined participants, for instance the orthopedic resident who participated on 3 days and simulated 239 procedures compared with a mean of 79 (CI 61–89) simulations for participants failing HipSim and quitting orthopedics. Furthermore, we speculate as to whether participants with the career plan to work as general practitioners or within emergency medicine may have applied for the 1st-year residency in orthopedics with the purpose of acquiring an insight and clinical evaluation skills without an earnest interest in the surgical procedures or surgical simulation. However, these hypotheses require further scientific investigations. Perspectives regarding unsolved issues in medical education The main purpose of hip fracture simulation programs is to advance novices to become pre-trained novices with an accelerated learning curve when starting to learn to operate in the clinical setting. To our knowledge no transfer studies have investigated this issue regarding hip fracture surgery. However, a randomized controlled trial investigated the effect of VR training of total hip arthroplasty and reports the transferability of the acquired technical skill to a cadaver lab (Hooper et al. 2019). Furthermore, the ability of the participants to use their acquired competencies in different, related hip fracture procedures such as cannulated screws and intramedullary nailing should be tested. Notably, the transferability of technical skill from one surgical procedure to another has been a matter of debate (Van Sickle et al. 2006, Akhtar et al. 2016). From a medical education point of view, future studies are needed to investigate whether a summative LC-CUSUM test promotes or hinders formative acquisition of technical competencies. The latter might be the case, because the LC-CUSUM test punishes failed simulations harshly, causing frustration (Rölfing et al. 2019). Thus, it is relevant to investigate if LCCUSUM should be included or if other more motivating learn-

673

ing strategies should be applied in simulation-based training programs. Need for continued clinical supervision and monitoring In the present study, none of the participants had flawless performance, i.e., reaching an LC-CUSUM score of 39 without any mistakes. Furthermore, the vast majority of participants made mistakes not only at early stages of the learning curve, but also at later stages. This highlights the strength of the LCCUSUM as it visualizes these late failed procedures (Figure 1). In the clinical setting, orthopedic residents in Scandinavia are often permitted to operate under no or limited supervision after 3–5 successful supervised procedures only. In our study, even participants with 12 consecutive, successful procedures made critical mistakes and failed the next procedure, which underlines the need for continued clinical supervision. Furthermore, the number of simulations and total simulation time was comparable between groups. Thus, competency cannot be declared based on the number of procedures or the time in training (Gustafsson et al. 2019). This important take-home message and the ability of the LC-CUSUM to monitor successful procedures before signaling competency with predefined statistical certainty was also appreciated in clinical studies regarding spinal laminectomy and acetabular cup placement in total hip arthroplasty (Lee et al. 2014, Park et al. 2019). Finally, we agree with the notion that simulation training can help to acquire the skills needed to perform the surgical procedure, such as hand–eye coordination, and the ability to work in 3 dimensions based on 2-dimensional visual information and haptic feedback (Akhtar et al. 2016, Marcheix et al. 2017). VR simulation is currently neither suited to teaching the participants the entire procedure including reposition of the fracture and soft-tissue handling nor does it encompass the overwhelming complexity of performing the procedure in the operating theatre. However, the major benefit of HipSim resides in learning the skills needed to perform the DHS surgical procedure, including the sequence of surgical steps. Thus, HipSim-certified pretrained novices are prepared to continue the learning process under close supervision in the operating theatre. The orthopedic community is becoming increasingly aware and appreciates this beneficial aspect of simulation training (Morgan et al. 2017, Kalun et al. 2018, Atesok et al. 2019, Gustafsson et al. 2019). In conclusion, LC-CUSUM and its summative pass/fail assessment of each simulation was feasible in this formative simulation program. Clinical educators can be reassured that participants passing HipSim are likely to continue to the 2nd–5th years of residency, while failing HipSim should raise concerns and career counselling and close clinical supervision seem to be appropriate measures. The motivational aspect of LC-CUSUM pass/fail assessment when designing formative simulation training warrants further research.

674

Supplementary data Figures 2 and 4 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/1745 3674.2020.1777511 JDR, RDJ, and CP: study design, critical review and final approval of the manuscript. JDR: data collection, analysis and first draft of the manuscript. Acta thanks Chinmay Gupte and Eleftherios Tsiridis for help with peer review of this study

Akhtar K, Sugand K, Sperrin M, Cobb J, Standfield N, Gupte C. Training safer orthopedic surgeons. Acta Orthop 2015; 86(5): 616-21. Akhtar K, Sugand K, Wijendra A, Sarvesvaran M, Sperrin M, Standfield N, Cobb J, Gupte C. The transferability of generic minimally invasive surgical skills: is there crossover of core skills between laparoscopy and arthroscopy? J Surg Educ 2016; 73(2): 329-38. Atesok K, Hurwitz S, Anderson D D, Satava R, Thomas GW, Tufescu T, Heffernan M J, Papavassiliou E, Theiss S, Marsh J L. Advancing simulation-based orthopaedic surgical skills training: an analysis of the challenges to implementation. Adv Orthop 2019; Sep 2; 2019: 2586034. doi: 10.1155/2019/2586034. eCollection 2019. Aucar J A, Groch N R, Troxel S A, Eubanks S W. A review of surgical simulation with attention to validation methodology. Surg Laparosc Endosc Percutaneous Tech 2005; 15(2): 82-9. Baumgaertner M R, Curtin S L, Lindskog D M, Keggi J M. The value of the tip–apex distance in predicting failure of fixation of peritrochanteric fractures of the hip. J Bone Joint Surg 1995; 77(7): 1058-64. Biau D J, Williams S M, Schlup M M, Nizard R S, Porcher R. Quantitative and individualized assessment of the learning curve using LC-CUSUM. Br J Surg 2008; 95(7): 925-9. Brauer C A, Coca-Perraillon M, Cutler D M, Rosen A B. Incidence and mortality of hip fractures in the United States. JAMA 2009; 302(14): 1573-9. De Bruijn K, den Hartog D, Tuinebreijer W, Roukema G. Reliability of predictors for screw cutout in intertrochanteric hip fractures. J Bone Joint Surg Am 2012; 94(14): 1266-72. Gustafsson A, Pedersen P, Rømer T B, Viberg B, Palm H, Konge L. Hipfracture osteosynthesis training: exploring learning curves and setting proficiency standards. Acta Orthop 2019; 90(4): 348-53.

Acta Orthopaedica 2020; 91 (6): 669–674

Hooper J, Tsiridis E, Feng J E, Schwarzkopf R, Waren D, Long W J, Poultsides L, Macaulay W, Papagiannakis G, Kenanidis E, Rodriguez E D, Slover J, Egol K A, Phillips D P, Friedlander S, Collins M. Virtual reality simulation facilitates resident training in total hip arthroplasty: a randomized controlled trial. J Arthroplasty 2019; 34(10): 2278-83. Kalun P, Wagner N, Yan J, Nousiainen M T, Sonnadara R R. Advances in medical education and practice. Dovepress Surgical simulation training in orthopedics: current insights. Adv Med Educ Pract 2018; 9: 125-131. Lee Y K, Biau D J, Yoon B H, Kim T Y, Ha Y C, Koo K H. Learning curve of acetabular cup positioning in total hip arthroplasty using a cumulative summation test for learning curve (LC-CUSUM). J Arthroplasty 2014; 29(3): 586-9. Marcheix P S, Vergnenegre G, Dalmay F, Mabit C, Charissoux JL. Learning the skills needed to perform shoulder arthroscopy by simulation. Orthop Traumatol Surg Res 2017; 103(4): 483-8. Morgan M, Aydin A, Salih A, Robati S, Ahmed K. Current status of simulation-based training tools in orthopedic surgery: a systematic review. J Surg Educ 2017; 74(4): 698-716. Park S-M, Kim H-J, Kim G-U, Choi M-H, Chang B-S, Lee C-K, Yeom J S. Learning curve for lumbar decompressive laminectomy in biportal endoscopic spinal surgery using the Cumulative Summation test for learning curve. World Neurosurg 2019; 122: e1007-13. Pedersen P, Palm H, Ringsted C, Konge L. Virtual-reality simulation to assess performance in hip fracture surgery. Acta Orthop 2014; 85(4): 403-7. Röck N D, Riahi L B, Christensen H C. Hoftebrud—Dansk tværfagligt register for hoftenære lårbensbrud—National årsrapport 2018; 2019 (April). Rölfing J D, Nørskov J K, Paltved C, Konge L, Andersen S A W. Failure affects subjective estimates of cognitive load through a negative carry-over effect in virtual reality simulation of hip fracture surgery. Adv Simul 2019; 4(1): 1-8. Strandbygaard J, Scheele F, Sørensen J L. Twelve tips for assessing surgical performance and use of technical assessment scales. Med Teach 2017; 39(1): 32-7. Sugand K, Akhtar K, Khatri C, Cobb J, Gupte C. Training effect of a virtual reality haptics-enabled dynamic hip screw simulator. Acta Orthop 2015; 86(6): 695-701. Van Sickle K R, Ritter E M, Smith C D. The pretrained novice: using simulation-based training to improve learning in the operating room. Surg Innov 2006; 13(3): 198-204. Zhang Q, Zhang Q, Guo W, Liu Z, Cheng L, Yue D, Zhang N. The learning curve for minimally invasive Oxford phase 3 unicompartmental knee arthroplasty: cumulative summation test for learning curve (LC-CUSUM). J Orthop Surg Res 2014; Sep 6; 9: 81. doi: 10.1186/s13018-014-0081-8.

Acta Orthopaedica 2020; 91 (6): 675–681

675

Electromagnetic navigation system for acetabular component placement in total hip arthroplasty is more precise and accurate than the freehand technique: a randomized, controlled trial with 84 patients Rene MIHALIČ 1, Jurij ZDOVC 2, Janez MOHAR 1, and Rihard TREBŠE 1,3 1 Valdoltra

Orthopaedic Hospital, Ankaran; 2 University of Ljubljana, Faculty of Pharmacy, Ljubljana; 3 University of Ljubljana, Faculty of Medicine, Ljubljana, Slovenia Correspondence: rene.mihalic@ob-valdoltra.si Submitted 2020-03-09 Accepted 2020-05-25

Background and purpose — The accuracy of conventional navigation systems depends on precise registration of bony landmarks. We investigated the clinical use of electromagnetic navigation (EMN), with a unique device for precise determination of the anterior pelvic plane. Patients and methods — We randomly allocated patients scheduled for total hip arthroplasty into 2 groups of 42 patients each. In the study group, cups were placed at the predetermined target angles (inclination: 42.5°; anteversion: 15°) with the support of EMN. In the control group, cups were placed freehand aiming at the same target angles. Postoperatively the true position of the cup was determined using computed tomography scan of the pelvis. Precision (root mean squared error, RMSE) bias (mean bias error, ME), accuracy, and duration of surgery were compared between the methods. Results — Cup anteversion was more accurate and precise in the navigated group. The ME in the navigated and freehand group was –1.7° (95% CI –2.4 to 1.1) and –4.5° (CI –6.5 to 2.5), respectively. The RMSE in the navigated and freehand group was 2.8° (CI 2.3–3.2) and 8.0° (CI 6.3– 9.5), respectively. The inclination was also more precise in the navigated group, with the RMSE in the navigated and freehand group at 4.6° (CI 3.4–5.9) and 6.5° (CI 5.4–7.5), respectively. The accuracy of the inclination and the duration of surgeries were similar between the groups. Interpretation — Cup placement with the help of EMN is more precise than the freehand technique and it does not affect the duration of surgery.

Optimal cup placement is crucial to the success of total hip arthroplasty (THA) since it is associated with lower rates of dislocation, prolonged implant survival, and better quality of life of the patient (Learmonth et al. 2007). For cup position in THA the safe zone was defined by Lewinnek et al. (1978), with recommended inclination and anteversion angles of 40° ± 10° and 15° ± 10°, respectively. Several studies have demonstrated that using the freehand technique for cup placement within the safe zone remains a challenge even for high-volume surgeons. More than 75% of cups are still inadvertently placed out of the safe zone (Digioia et al. 2002, Saxler et al. 2004, Bosker et al. 2007). Several studies have reported that the cup placement could be optimized using imageless navigation, which is a more reproducible technique compared with a freehand THA (Digioia et al. 2002, Kalteis et al. 2006a, Hohmann et al. 2011, Lass et al. 2014). Those studies were mostly performed with different producers’ stereo-optical navigation systems with the same basic concept (Renkawitz et al. 2009). To assure the accuracy of such a system, the tracker must be large, and therefore placed outside the surgical incision. This is related to additional morbidity (Dorr et al. 2005, Kamara et al. 2017). The accuracy depends on precise registration of bony landmarks (Dorr et al. 2005, Lass et al. 2014), which are necessary for the determination of the reference plane. The registration of the reference points is mostly affected by the thickness of the overlying soft tissues. This can tilt the virtual reference plane and contribute to the systemic error (Hohmann et al. 2011), especially in obese patients (Parratte and Argenson 2007, Wassilew et al. 2012, Buller et al. 2019). To avoid imprecise reference plane determination, a different imageless navigation concept was introduced. This system consists of an electromagnetic transmitter and sensors, which are placed

676

Acta Orthopaedica 2020; 91 (6): 675–681

ENROLLMENT

Assessed for eligibility n = 137 Excluded (n = 53): – not meeting inclusion criteria, 27 – declined to participate, 19 – other reasons, 7 Randomized n = 84

ALLOCATION

Allocated to the study group – EHIP (n = 42): Received allocated intervention (n = 42)

Allocated to control group – freehand (n = 42): Received allocated interventiion (n = 42)

FOLLOW-UP

Lost to follow-up (n = 0) Discontinued intervention (n = 0)

Lost to follow-up (n = 0) Discontinued intervention (n = 0) ANALYSIS

Analyzed (n = 42)

Figure 1. Position of the Navi-frame on the sawbones model of the pelvis.

Analyzed (n = 42)

Figure 2. CONSORT flow diagram of the study.

inside the incision and on instruments. Additionally, we developed a particular tool (Navi-frame) to overcome the difficulties in the registration of bony landmarks for correct anterior pelvic plane (APP) determination. The basic idea was that at least 3 non-collinear points describe a plane. In THA these 3 points are represented by the two anterior superior iliac spines (ASIS) and the pubic tubercle. The real APP is registered with the placement of the Navi-frame on these 3 points (Figure 1). This presents a major improvement compared with the stereooptical systems. Our 2 hypotheses for this study were: 1. An electromagnetic navigation (EMN) system enables more accurate and precise cup placement in THA than the freehand technique, regardless of the patient’s BMI. 2. The EMN system does not affect the duration of surgery.

Patients and methods Study design and patient selection Before the study design, we performed a pilot study (part of the validation process of the EMN system), including 10 patients in each group (navigated and freehand group), which was a basis for power analysis and also represented a learning period for handling the EMN system. This was a randomized, controlled clinical trial of 2 groups of 42 patients, all scheduled for THA between May 4, 2017 and February 2, 2018. The patient data included: diagnosis, age, sex, BMI, Harris Hip Score (HHS), and side. In the study group (EHIP), patients underwent the EMN-assisted cup placement during THA. In the control group (freehand), patients underwent conventional freehand cup placement. The inclusion criteria were age above 18 years, unilateral surgery, osteoarthritis of the hip, no previous surgery on the

affected hip, implantation of the same acetabular component through the same approach, and signed informed consent. 3 high-volume surgeons performed all procedures. We followed the CONSORT guidelines. Within the cohort of 137 consecutive patients scheduled for primary THA, 84 patients met the inclusion criteria and were randomly allocated in a 1:1 ratio into the EHIP group and the control (freehand) group (Figure 2). Randomization was conducted using computer-generated numbers from the Research Randomizer System. Even numbers represented the EHIP group. Surgical procedure and postoperative evaluations A modified Hardinge approach in the supine position was used and a cementless cup and stem from the same manufacturer were implanted in all cases (Allofit/Alloclassic Zimmer Biomet, Warsaw, IN, USA). In the EHIP group, during patient draping a sterilely covered arm with the electromagnetic transmitter of the EMN system (Guiding Star, E-Hip module, Ekliptik d.o.o., Ljubljana, Slovenia) was mounted on the operating table and connected to the central unit equipped with a monitor. No additional preoperative time was needed to prepare the EMN system. After resection of the femoral head, a specially designed Steinmann pin (diameter 4.5 mm) with a reference sensor on it was mounted above the acetabular edge, without additional skin and soft tissue dissection. APP was then determined with the help of the Navi-frame equipped with the measuring sensor. The Navi-frame was place-pressed on both ASIS and the pubic tubercle. When the position of the frame was correct, the APP was registered (Figure 3). After acetabular preparation, the cup was impacted with the help of a conventional cup holder with a measuring sensor on it aiming to place the cup around predefined target angles (42.5° for inclination and 15° for anteversion). The values of both angles were displayed on the monitor and registered by the

Acta Orthopaedica 2020; 91 (6): 675–681

677

Figure 5. Postoperative measurements of cup position (APP = anterior pelvic plane).

Figure 3. Determination of the anterior pelvic plane with Navi-frame (black arrow) and navigation system consisting of electromagnetic transmitter (white arrow) and monitor (white arrowhead).

Figure 4. A. Implantation of the acetabular component with conventional cup applicator equipped with holder (white arrow) for measuring sensor (black arrow). Reference sensor (black arrowhead). B. Monitor with real-time information on acetabular component position.

EMN system. Those values represented the basis for later calculations of accuracy and precision of the EMN system (Figure 4). All manipulations of the navigation system were performed after the skin incision and represent part of the measured surgical duration (skin incision to last skin suture). In the freehand group, the cup was placed freehand, aiming to place it inside the predefined target angles, with the help of visible anatomical landmarks around the acetabulum and relying on the surgeon’s ability to estimate the patient’s real position on the operating table. Predefined target angles represented the basis for later calculations of precision and accuracy of the freehand technique. Postoperatively (up to 48 hours after surgery) all patients underwent CT scans of hip and pelvis for the determination of the actual acetabular component position, which served as a reference (Kalteis et al. 2006b, Lass et al. 2014). The position of the pelvis was standardized by reformatting the images to the APP. Single measurements of the inclination and anteversion angles based on the CT scans of hips and pelvises were made by the independent technician, with the help of special CAD/CAM (computer-aided design/manufacturing) software (EBS software, Ekliptik d.o.o., Ljubljana, Slovenia), where the technician defined the APP, the sagittal plane, the transverse plane, and the axis of the cup (line perpendicular to the plane defined by the outer circumference of the cup), and the measurements occurred automatically based on the combined algorithm by Murray (1993), and by Hohmann et al. (2011) (Figure 5). All patients underwent the same perioperative antibiotic prophylaxis, pain management, and rehabilitation protocols. Statistics The sample size was calculated with the G*Power software tool (http://www.gpower.hhu.de/) based on the preliminary data obtained from the validation of the navigation system (Mihalic and Trebse 2016). Study power (1–β) was set to 0.8

678

Acta Orthopaedica 2020; 91 (6): 675–681

Table 1. Demographic data. Values are mean (standard deviation) unless otherwise specified Factor Sex: female/male, n Age Body mass index HSS, preoperative

Type of operation Freehand EHIP (n = 42) (n = 42) 22/20 66 (11) 29 (5.7) 62 (17)

21/21 67 (11) 30 (4.3) 56 (17)

Inclination deviation (°)

Anteversion deviation (°)

Inclination (°) 55 E-HIP Freehand 50

–10 0

0 –10

EHIP, electromagnetic navigation surgery. HHS, Harris Hip Score –10

–20

EHIP

Freehand

EHIP

Freehand

Figure 6. Deviations of the inclination angle measurements (left panel) and anteversion angle measurements (right panel) in the EHIP (study group) and the freehand group. Median (red line across boxes), 1st and 3rd quartile (lower and upper hinges), minimum and maximum non outlying (< 1.5 times interquartile range) values (whiskers).

and the significance level (α) to 0.05. Descriptive statistics were given as mean and range for continuous variables and as percentages for categorical variables. Continuous variables for both groups were compared using a 2-tailed Student’s independent t-test, and categorical variables were compared with the chi-square test. For the comparison of method performance in EHIP and the freehand group, the target angles during the THA and angles measured on the postoperative CT scan (true angles) were used to calculate the deviation from the target angle for every patient. Based on those deviations, the method bias, precision, and accuracy were determined. Bias was expressed as the mean bias error (ME), which was calculated as the average difference between the target angles and the true angle. The negative difference between the target and true angle is interpreted as the method’s underprediction of the true angle. Precision was expressed as the root mean squared error (RMSE), which was calculated as the standard deviation of the differences between the target angle and the true angle. ME and RMSE for the inclination and anteversion angles were primary outcomes of our study. Accuracy was represented by a combination of ME and RMSE (Sheiner and Beal 1981). The lower absolute value of ME and the lower RMSE represent less method bias and better precision, respectively, and subsequently better method accuracy. The duration of the surgery was the secondary outcome. R software, version 3.6.0 (R Development Core Team, Vienna, Austria) and package boot (http://www.R-project.org) was used to perform bootstrapping with replacement method (number of virtual samples = 10,000) and to obtain the 95% confidence intervals (CI) for the ME and RMSE. Other statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 25.0 (IBM Corp, Armonk, NY, USA). P < 0.05 was considered significant.

Anteversion (°)

Figure 7. Distribution of cup positions for both groups. Area between 30° and 50° for inclination and between 5° and 25° represents Lewinnek safe zone.

Ethics, registration, funding, and potential conflicts of interest The study design was approved by the National Committee of Medical Ethics (77/05/12) and was performed in agreement with the Helsinki II declaration. The study was registered at ClinicalTrials.gov (NCT04101864). This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Regarding this study the authors have no conflicts of interest.

Results Demographic data The demographic characteristics of both groups were similar (Table 1). Surgical duration The average duration of surgeries was similar between the groups: in the EHIP group 70 minutes (SD 10) and 70 minutes (SD 13) in the freehand group. Radiographic parameters For both angles, the observed range of deviations was lower in the EHIP group (Figure 6). Precision presented by the RMSE for both angles was statistically significantly higher in the EHIP group (Table 2). Accuracy, presented by a combination of RMSE and ME, was statistically significantly higher for the anteversion angle in the EHIP group (Table 2). However, we did not observe any difference in method bias for the inclination angle when comparing the two groups (Table 2). Regarding Lewinnek’s safe zone there were 4 outliers in the EHIP, and 9 outliers in the freehand group (Figure 7).

Acta Orthopaedica 2020; 91 (6): 675–681

679

Table 2. Comparison of accuracy and precision for inclination and anteversion angles. Values are mean (95% confidence interval) Factor Inclination ME (°) RMSE (°) Anteversion ME (°) RMSE (°)

Type of operation Freehand EHIP (n = 42) (n = 42)

p-value a

1.9 (0.0 to 3.8) 6.5 (5.4 to 7.5)

1.7 (0.4 to 3.0) 4.6 (3.4 to 5.9)

0.9 0.02

–4.5 (–6.5 to –2.5) 8.0 (6.4 to 9.6)

–1.7 (–2.4 to –1.1) 2.8 (2.3 to 3.3)

0.01 < 0.001

2-tailed Student’s independent t-test, based on the bootstrap sample distributions. EHIP, electromagnetic navigation surgery. ME, mean bias error. RMSE, root mean squared error.

Table 3. Comparison of accuracy and precision of inclination and anteversion angles for patients with body mass index (BMI) ≤ 30 and ≥ 30 in both groups. Values are mean (95% confidence interval) Factor Freehand Inclination ME (°) RMSE (°) Anteversion ME (°) RMSE (°) EHIP Inclination ME (°) RMSE (°) Anteversion ME (°) RMSE (°)

BMI < 30

BMI ≥ 30

n = 15

n = 27

p-value a

3.0 (0.7 to 5.3) 6.8 (5.7 to 7.9)

–0.2 (–3.1 to 2.7) 5.7 (3.6 to 8.2)

0.1 0.4

–5.0 (–7.3 to –2.6) 8.1 (6.0 to 10) n = 22

–3.6 (–7.2 to –0.1) 7.8 (5.4 to 10) n = 20

0.5 0.9

1.9 (0.1 to 3.7) 4.7 (3.7 to 5.8)

1.5 (–0.5 to 3.3) 4.6 (2.2 to 7.2)

0.8 0.9

–1.8 (–2.8 to –0.9) 2.9 (2.3 to 3.5)

–1.6 (–2.5 to –0.7) 0.8 2.6 (1.9 to 3.4) 0.5

For footnotes, see Table 2.

Additionally, no significant association was observed between the precision and accuracy and BMI. The ME and RMSE were similar comparing non-obese patients with BMI < 30 and obese patients with BMI ≥ 30 for both angles, and for both methods (Table 3).

Discussion Currently, there are still various opinions regarding the benefits of navigation and other computer-aided systems in THA. Many studies demonstrated that navigation is superior to freehand techniques regarding accuracy and precision of the cup placement (Dorr et al. 2005, Parratte and Argenson 2007, Najarian et al. 2009, Hohmann et al. 2011, Lass et al. 2014, Buller et al. 2019). Importantly, despite there being several advantages of the existing navigation systems (accuracy and precision) (Dorr et al. 2005, Parratte and Argenson 2007, Najarian et al. 2009, Hohmann et al. 2011, Lass et al. 2014, Buller et al. 2019), they also have considerable disadvantages including tracker-related harm (Dorr et al. 2005, Kamara et al. 2017), longer duration of surgery (Parratte and Argenson 2007, Najarian et al. 2009, Lass et al. 2014, Liu et al. 2015), and higher costs. Our study confirmed the first hypothesis. The presented EMN system with the use of the Navi-frame provides more precise cup placements in THA compared with the freehand technique. Additionally, the EMN system demonstrated less bias in anteversion angle estimation, indicating better accuracy of cup placement for the anteversion angle. The observed accuracy and precision of the anteversion angle of the cup placement were below 3°. Additionally, we observed that accuracy and precision of navigated cup placement were unaffected by the patient’s BMI or thickness of the soft tissue overlaying the bony landmarks. The accuracy and the precision of navigated cup placement were similar comparing the obese

(BMI ≥ 30) and the non-obese patients (BMI ≤ 30) (Table 3). We hypothesize this is due to the specially developed Naviframe for APP determination, which captures all three bony landmarks at once and creates the most accurate approximation of real APP, regardless of the patient’s BMI. In contrast to our observations, several studies have demonstrated the opposite: that the precision and accuracy of conventional imageless navigation systems is mostly affected by the soft tissue overlying the bony landmarks (Parratte and Argenson 2007, Parratte et al. 2007, Wassilew et al. 2012, Buller et al. 2019) and that the most important factor to avoid systemic error is precise registration of landmarks (Digioia et al. 2002, Parratte et al. 2008, Hohmann et al. 2011, Lass et al. 2014). Paratte and Argenson (2007) concluded that the accuracy of acetabular component placement in obese patients is considerably affected by the soft tissue thickness over the bony landmarks, which probably affects the precision of registration of the APP and is the most obvious limitation factor of navigation systems currently on the market. The in vitro study by Paratte et al. (2008) evaluated the accuracy of percutaneous and ultrasound-based registration of bony landmarks for APP determination. They reported no statistically significant difference in terms of inclination. In contrast, anteversion errors were statistically significantly higher with percutaneous registration. Similar results were published by Wassilew et al. (2012), who compared the accuracy of an ultrasoundbased navigation system and an imageless navigation system with surface registration. They also observed a statistically significant correlation between BMI and anteversion error in the surface registration group. Hohmann et al. (2011) reported that one of the most important factors to avoid high systemic errors is precise acquisition of the bony landmarks. Ybinger et al. (2007) also reported that the thickness of the soft tissue overlying bony landmarks influenced the inclination and the

680

anteversion values. In contrast, Lass et al. (2014) found no statistically significant difference in accuracy of acetabular component position in relation to a patient’s BMI. However, they claimed that the main reason that BMI was not affecting the accuracy of navigation was exact acquisition of the bony landmarks with a sharp metal pointer. Given the above, it seems reasonable to assume that the most important factor in determination of APP is the thickness of the soft tissue overlying bony landmarks affecting their precise registration, but it seems this does not affect our EMN system. We also confirmed the second hypothesis and proved that the duration of the surgery was unaffected by the navigation system, which is in contrast to other published studies where the duration of surgery was considerably longer due to navigation (Parratte and Argenson 2007, Najarian et al. 2009, Lass et al. 2014, Liu et al. 2015). Considering that longer surgeries are associated with the increased risk of infection (Cheng et al. 2017, Kong et al. 2017), this represents an important advantage over other navigation systems. We observed an even narrower range of duration of the surgery in the EHIP group, which is probably due to more reproducible and fluent surgical procedures allowed by the EMN support. Since the EMN system with Navi-frame is applicable to every patient’s position, and every surgical approach in THA, the main limitation of our study was to focus on the cup position only. We did not consider the combined acetabular and femoral component anteversion, which could represent another factor in the prevention of impingement and possible implant dislocation. In our study, femoral components were of rectangular, tapered, cementless design with limited ability to adjust their anteversion, which is determined by the femoral canal. The main influence on the combined anteversion was the acetabular component anteversion as noted also by Goudie et al. (2015). Based on a large metanalysis, cup position seems to be important for hip instability (particularly large deviations from the average) as well as many other variables. The target zone is difficult to set because it is influenced by many other factors including the approach and individual anatomical variations of the spinopelvic region (Seagrave et al. 2017). Additionally, given that several primary outcomes were tested, possible multiplicity issues were not excluded, and further studies are necessary to additionally confirm the clinical significance of the EMN system. Nevertheless, based on our results, we could conclude that EMN in THA appears to increase the accuracy and precision of the cup position and does not affect the surgical duration. Consequently, it might become a valuable tool to target patient-specific cup position determined by many individual anatomical factors and judged important for hip stability and longevity. The best property of the EMN is that it is a passive system, providing the surgeon with real-time information on implant position. This is especially important in difficult anatomical situations, without standard landmarks and in miniinvasive procedures (DiGioia et al. 2003) where small inci-

Acta Orthopaedica 2020; 91 (6): 675–681

sions compromise the visibility of different landmarks, which usually help a surgeon in cup placement with the freehand technique.

The authors would like to thank the staff of the Department of Radiology at their institution for their support when performing the postoperative CT scans. Additionally, the authors would like to thank the staff of the A1 department at their institution, including nurses and physiotherapists, for their help in fulfilling pre- and postoperative protocols. Special thanks go to Anže Mihelič, MD, who was the third surgeon performing surgeries for the study. RM: conceptualization; data curation; investigation; methodology; project administration; writing—original draft. JZ: data curation; formal analysis; methodology; writing—review and editing JM: methodology; writing— review and editing. RT: conceptualization; methodology; supervision; writing—review and editing. Acta thanks Fabian van de Bunt and Keijo T Mäkelä for help with peer review of this study

Bosker B H, Verheyen C C P M, Horstmann W G, Tulp N J A. Poor accuracy of freehand cup positioning during total hip arthroplasty. Arch Orthop Trauma Surg 2007; 127(5): 375-9. Buller L T, McLawhorn A S, Romero J A, Sculco P K, Mayman D J. Accuracy and precision of acetabular component placement with imageless navigation in obese patients. J Arthroplasty 2019; 34(4): 693-9. Cheng H, Chen B P-H, Soleas I M, Ferko N C, Cameron C G, Hinoul P. Prolonged operative duration increases risk of surgical site infections: a systematic review. Surg Infect 2017; 18(6): 722-35. Digioia A M, Jaramaz B, Plakseychuk A Y, Moody J E, Nikou C, Labarca R S, Levison T J, Picard F. Comparison of a mechanical acetabular alignment guide with computer placement of the socket. J Arthroplasty 2002; 17(3): 359-64. DiGioia A M, Plakseychuk A Y, Levison T J, Jaramaz B. Mini-incision technique for total hip arthroplasty with navigation. J Arthroplasty 2003; 18(2): 123-8. Dorr L D, Hishiki Y, Wan Z, Newton D, Yun A. Development of imageless computer navigation for acetabular component position in total hip replacement. Iowa Orthop J 2005; 25: 1-9. Goudie S T, Deakin A H, Deep K. Natural acetabular orientation in arthritic hips. Bone Joint Res 2015; 4(1): 6-10. Hohmann E, Bryant A, Tetsworth K. A comparison between imageless navigated and manual freehand technique acetabular cup placement in total hip arthroplasty. J Arthroplasty 2011; 26(7): 1078-82. Kalteis T, Handel M, Bäthis H, Perlick L, Tingart M, Grifka J. Imageless navigation for insertion of the acetabular component in total hip arthroplasty: is it as accurate as CT-based navigation? J Bone Joint Surg Br 2006a; 88(2): 163-7. Kalteis T, Handel M, Herold T, Perlick L, Paetzel C, Grifka J. Position of the acetabular cup: accuracy of radiographic calculation compared to CT-based measurement. Euro J Radio 2006b; 58(2): 294-300. Kamara E, Berliner Z P, Hepinstall M S, Cooper H J. Pin site complications associated with computer-assisted navigation in hip and knee arthroplasty. J Arthroplasty 2017; 32(9): 2842-6. Kong L, Cao J, Zhang Y, Ding W, Shen Y. Risk factors for periprosthetic joint infection following primary total hip or knee arthroplasty: a meta-analysis. Int Wound J 2017; 14(3): 529-36. Lass R, Kubista B, Olischar B, Frantal S, Windhager R, Giurea A. Total hip arthroplasty using imageless computer-assisted hip navigation: a prospective randomized study. J Arthroplasty 2014; 29(4): 786-91.

Acta Orthopaedica 2020; 91 (6): 675–681

Learmonth I D, Young C, Rorabeck C. The operation of the century: total hip replacement. Lancet 2007; 370(9597): 1508-19. Lewinnek G E, Lewis J L, Tarr R, Compere C L, Zimmerman J R. Dislocations after total hip-replacement arthroplasties. J Bone Joint Surg Am 1978; 60(2): 217-20. Liu Z, Gao Y, Cai L. Imageless navigation versus traditional method in total hip arthroplasty: a meta-analysis. Int J Surg 2015; 21: 122-7. Mihalic R, Trebse R. Imageless navigation of the cup in total hip arthroplasty: does it work? Orthopaedic Proceedings 2016; 98-B(SUPP_9):45-45. Murray D W. The definition and measurement of acetabular orientation. J Bone Joint Surg Br 1993; 75(2): 228-32. Najarian B C, Kilgore J E, Markel D C. Evaluation of component positioning in primary total hip arthroplasty using an imageless navigation device compared with traditional methods. J Arthroplasty 2009; 24(1): 15-21. Parratte S, Argenson J-N, Flecher X, Aubaniac J-M. [Computer-assisted surgery for acetabular cup positioning in total hip arthroplasty: comparative prospective randomized study]. Rev Chir Orthop Reparatrice Appar Mot 2007; 93(3): 238-46. Parratte S, Argenson J-N A. Validation and usefulness of a computer-assisted cup-positioning system in total hip arthroplasty: a prospective, randomized, controlled study. J Bone Joint Surg Am 2007; 89(3): 494-9. Parratte S, Kilian P, Pauly V, Champsaur P, Argenson J-N A. The use of ultrasound in acquisition of the anterior pelvic plane in computer-assisted

681

total hip replacement: a cadaver study. J Bone Joint Surg Br 2008; 90(2): 258-63. Renkawitz T, Tingart M, Grifka J, Sendtner E, Kalteis T. Computer-assisted total hip arthroplasty: coding the next generation of navigation systems for orthopedic surgery. Expert Rev Med Devices 2009; 6(5): 507-14. Saxler G, Marx A, Vandevelde D, Langlotz U, Tannast M, Wiese M, Michaelis U, Kemper G, Grützner P A, Steffen R, von Knoch M, Holland-Letz T, Bernsmann K. The accuracy of free-hand cup positioning: a CT based measurement of cup placement in 105 total hip arthroplasties. Int Orthop 2004; 28(4): 198-201. Seagrave K G, Troelsen A, Malchau H, Husted H, Gromov K. Acetabular cup position and risk of dislocation in primary total hip arthroplasty. Acta Orthop 2017; 88(1): 10-17. Sheiner L B, Beal S L. Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm 1981; 9(4): 503-12. Wassilew G I, Perka C, Janz V, König C, Asbach P, Hasart O. Use of an ultrasound-based navigation system for an accurate acetabular positioning in total hip arthroplasty: a prospective, randomized, controlled study. J Arthroplasty 2012; 27(5): 687-94. Ybinger T, Kumpan W, Hoffart H E, Muschalik B, Bullmann W, Zweymüller K. Accuracy of navigation-assisted acetabular component positioning studied by computed tomography measurements: methods and results. J Arthroplasty 2007; 22(6): 812-17.

682

Acta Orthopaedica 2020; 91 (6): 682–686

Does cup position differ between trabecular metal and titanium cups? A radiographic propensity score matched study of 300 hips Inari LAAKSONEN 1, Natalie HJELMBERG 2, Kirill GROMOV 3,4, Antti E ESKELINEN 5, Ola ROLFSON 2, Henrik MALCHAU 2,6, Anders TROELSEN 3, Keijo T MÄKELÄ 1, and Maziar MOHADDES 2 1 Department

of Orthopedics and Traumatology, Turku University Hospital, Turku, Finland, Finnish Arthroplasty Register, Helsinki, Finland; 2 Swedish Hip Arthroplasty Register, Department of Orthopaedics, Institute of Surgical Sciences, Sahlgrenska University Hospital, Gothenburg, Sweden; 3 Department of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark; 4 Danish Hip Arthroplasty Register, Aarhus, Denmark; 5 Coxa Hospital for Joint Replacement, Tampere, Finland, Finnish Arthroplasty Register, Helsinki, Finland; 6 Harris Orthopedic Laboratory, Massachusetts General Hospital, Boston, USA Correspondence: inari.laaksonen@tyks.fi Submitted 2020-01-06. Accepted 2020-06-16.

Background and purpose — The use of trabecular metal cups in primary total hip arthroplasty (THA) is increasing, despite the survival of Continuum cups being slightly inferior compared with other uncemented cups in registries. This difference is mainly explained by a higher rate of dislocation revisions. Cup malpositioning is a risk factor for dislocation and, being made of a highly porous material, Continuum cups might be more difficult to position. We evaluated whether Continuum cups had worse cup positioning compared with other uncemented cups. Patients and methods — Based on power calculation, 150 Continuum cups from 1 center were propensity score matched with 150 other uncemented cups from 4 centers. All patients had an uncemented stem, femoral head size of 32 mm or 36 mm, and BMI between 19 and 35. All operations were done for primary osteoarthrosis through a posterior approach. Patients were matched using age, sex, and BMI. Cup positioning was measured from anteroposterior pelvic radiograph using the Martell Hip Analysis Suite software. Results — There was no clinically relevant difference in mean inclination angle between the study group and the control group (43° [95% CI 41–44] and 43° [CI 42–45], respectively). The study group had a larger mean anteversion angle compared with the control group, 19° (CI 18–20) and 17° (CI 15–18) respectively. Interpretation — Continuum cups had a greater anteversion compared with the other uncemented cups. However, the median anteversion was acceptable in both groups and this difference does not explain the larger dislocation rate in the Continuum cups observed in earlier studies.

Trabecular metal (TM) has become an increasingly popular implant material in both primary and revision total hip arthroplasty (THA) (Laaksonen et al. 2017, 2018). Its highly porous surface provides good initial stability and improves bone ingrowth (Bobyn et al. 1999, Beckmann et al. 2014). Continuum cups (Zimmer Biomet, Warsaw, IN, USA) with TM surface have showed higher revision rates than other uncemented cups after primary THA in some register studies mainly due to a higher dislocation rate (Laaksonen et al. 2018, Hemmilä et al. 2019). Dislocation is one of the most common postoperative complications leading to revision surgery (AOANJRR 2017, Finnish Arthroplasty Register [FAR] 2017). Risk for recurrent dislocation and periprosthetic joint infection increases after revision surgery and therefore prevention of the first dislocation is vital (Ezquerra et al. 2017). Potential risk factors for dislocation are posterior approach, small femoral head size, fracture as the indication for surgery, female sex, and suboptimal acetabular cup positioning (Hailer et al. 2012, Zijlstra et al. 2017). Optimal cup positioning to avoid dislocation is traditionally defined by Lewinnek safe zones. According to this definition optimal cup inclination angle is 40° ± 10° and optimal anteversion angle is 15° ± 10° (Lewinnek et al. 1978. Slight modifications to optimize the stability have also been presented (Danoff et al. 2016). In particular, lower anteversion has been associated with increased dislocation rate (Seagrave et al. 2017a). We theorized that the higher dislocation rate for Continuum cups compared with other uncemented cups may be caused by suboptimal cup positioning due to difficulties in optimizing the acetabular cup position with this highly porous material.

Acta Orthopaedica 2020; 91 (6): 682–686

683

Table 1. Demographic and surgical data of study population. Values are number (%) unless otherwise specified Factor

Study group n = 139

Control group n = 150

Turku a 50 (36) Coxa b 42 (30) Varberg c 47 (34) Hvidovre d 150 (100) Females 53 (38) 60 (40) Age, median (SD) 65 (8.9) 65 (12) BMI, median (SD) 28 (3.4) 27 (3.5) Femoral head size 32 mm 53 (38) 29 (19) 36 mm 86 (62) 121 (81) Liner Hi-wall 150 Neutral 98 (70) Oblique 30 (22) 10° elevation 11 (8)

Figure 1. A Continuum cup (A) and a control group cup (B).

In this observational multicenter cohort study, we analyzed whether there is a difference in acetabular implant positioning while using Continuum acetabular cups compared with other uncemented acetabular cups in primary total hip arthroplasty.

Patients and methods Power calculation Based on a previous publication describing cup positioning and its influence on dislocation, a power calculation (Biedermann et al. 2005) showed that a minimum cohort of 101 patients was needed in each group to detect a difference of 6 degrees abduction (n = 101) and 5 degrees in anteversion (n = 137), with 95% power and an α-error probability of 0.01. Recruiting 137 patients from each center will enable detection of 4 degrees difference in anteversion or inclination with a power of 95% (α = 0.05). To ensure study power in case of difficulties in the radiographic measurements 150 patients were included in both groups. Patients 150 randomly selected primary total hip arthroplasty cases from the 3 joint replacement centers (Turku University Hospital, Turku, Finland; Coxa Hospital for Joint Replacement, Tampere, Finland; Varberg Hospital, Varberg, Sweden), using a porous Continuum tantalum cup were propensity score matched with 150 cases using a porous-coated titanium cup (the control group) from 1 center (Hvidovre, Copenhagen, Denmark) (Figure 1). The Continuum cup is used in most primary THAs in the study centers.

a Turku University Hospital, Turku, Finland b Coxa Hospital for Joint Replacement, Tampere, Finland c Varberg Hospital, Varberg, Sweden d Hvidovre University Hospital, Copenhagen, Denmark

Inclusion criteria All patients were operated on for primary osteoarthrosis (OA) with a posterior approach between 2014 and 2017. Femoral head size used was 32 mm or 36 mm and all patients received an uncemented stem. All patients had a BMI between 19 and 35. Patients were matched using age, sex, and BMI (Table 1). All patients from the study centers fulfilling the inclusion criteria were collected for the matching process. No bilateral cases were included. Radiographic analysis The Martell Hip Analysis Suite software (version 8.0.1.4.3, UCTech, University of Chicago, IL, USA) was used for radiographic measurements (Martell and Berdia 1997, Elson et al. 2015). The first postoperative anteroposterior radiograph of the pelvis (0–3 months) was used for measuring cup inclination and anteversion angles. All measurements in the study group was measured by 1 examiner (NH) during 2019. In the study group there was difficulties measuring abduction and anteversion angles in 11 hips due to suboptimal radiographs and all were excluded from further analysis. Statistics Data selection and matching were applied using the R software (version 3.6.1; R Foundation for Statistical Computing, Vienna, Austria). A random sampling process (without replacement) was used to select 150 patients with Continuum cups. Propensity score matching, controlling for age, sex, and BMI, was used to select the control group of other uncemented cups. The propensity scores were estimated implementing a 1:1 nearest neighbor matching using logistic regression. The

684

Acta Orthopaedica 2020; 91 (6): 682–686

Table 2. Median cup angles and percentages of cups in the Lewinnek safe zone in the Continuum cup study group and the other uncemented cups control group Factor

Study group

Control group

Cup angles, mean (95% CI) Inclination 43 (41–45) 43 (41–44) Anteversion 19 (18–20) 17 (15–18) Turku a Inclination 42 (39–46) Anteversion 15 (13–17) Coxa b Inclination 45 (42–49) Anteversion 24 (22–26) Varberg c Inclination 43 (40–46) Anteversion 19 (16–21) d Hvidovre Inclination 43 (41–44) Anteversion 17 (15–18) Cups in the safe zone, % (n) Inclination 67 (93) 81 (122) Anteversion 76 (105) 67 (100) Both 52 (72) 55 (82) a–d See

Table 1.

normal distribution of data was checked by generating histogram and qq plots and whenever the normal distribution did fulfil assumptions, a nonparametric Mann–Whitney test was used to compare groups (study vs. control). The sex and the operated side distributions in the 2 groups were checked by chi-square test. Data were presented as median (SD). A p-value < 0.05 was considered as statistically significant. Means and 95% confidence interval levels (CI) are presented. Ethics, registration, funding, and potential conflicts of interests Ethical approval was obtained from the Local Ethical Review Board in Turku University Hospital (T01/003/18, date of issue May 8, 2018). This research received funding from the Southwestern Finland State Research Funding and from Turku University Hospital for radiograph transfer costs. All authors declare no conflict of interest.

Results There was no clinically relevant difference in mean inclination angle between the study group and the control group (43° [95% CI 41–44] and 43° [42–45], respectively) (Table 2). The study group had a larger mean anteversion angle compared with the control group, 19° (18–20) and 17° (15–18) respectively. Center-wise inclination and anteversion angles were 43° (41–44) and 17° (15–18), respectively, for the control group and 45° (42–49) and 24° (22–26) for center 1, 43° (40–46) and 19° (16–21) for center 2, and 42° (39–45) and 15° (12–17) for center 3 in the study group.

Figure 2. The scatter plot depicts Martell radiographic analysis, which compares the distribution of 2 groups in the safe zone.

Only 52% (n = 72) of the cups in the Continuum study group and 55% (n = 82) of the cups in the control group were in the Lewinnek safe zone when both inclination and anteversion angles were addressed (Figure 2, Table 2).

Discussion Recent register studies have presented higher revision rates for trabecular metal acetabular cups compared with other uncemented cups. This difference is mainly explained by higher dislocation rates. Acetabular component malpositioning is a known risk factor for dislocation in total hip arthroplasty. Trabecular metal is a highly porous material, which might make cups with a TM surface more difficult to implant in the desired position and therefore predispose to malpositioning. In this study we aimed to assess cup positioning in Continuum and other uncemented devices to evaluate potential malpositioning in Continuum cups. Trabecular metal cups are a good option in demanding cases with large bone defects because of their good osteointegration qualities (Bobyn et al. 1999). The use of this highly porous material has increased significantly during the last decade in both primary and revision THA encouraged by the good results in short- to mid-term clinical studies (Jafari et al. 2010, Baad-Hansen et al. 2011, Mohaddes et al. 2015, Wegrzyn et al. 2015). Despite the improved osteointegration and stability, TM cups have had slightly higher revision rates in register reports, mainly due to revisions for dislocation (Hemmilä et al. 2019). A potential risk for bias is difference in patient selection, as TM surfaced cups are traditionally used in more demanding cases. However, the difference in revision rate is not likely to be fully explained by patient selection as TM surfaced primary cups are the primary acetabular components in

Acta Orthopaedica 2020; 91 (6): 682–686

primary THA in many centers (Laaksonen et al. 2018). The Continuum cup is the most common device used in primary THA in the study centers and therefore in this study there should not be bias in patient selection. The higher revision rate for Continuum cups during shortterm follow-up appears to be explained by higher dislocation revision rates (Hemmilä et al. 2019). One of the known risk factors for THA dislocation is acetabular component malpositioning (Biedermann et al. 2005). Too small anteversion and potential retroversion has traditionally been associated with higher dislocation risk (Seagrave et al. 2017a). There have been several attempts to generate optimal safe zones in cup positioning to minimize the dislocation risk, varying around 30°–45° for inclination and 5°–25° for anteversion (Callanan et al. 2011, Seagrave et al. 2017b). In our material there were no clinically relevant differences in the inclination angle between the study groups. Continuum cups had a higher anteversion angle than the control group; however, anteversion was acceptable in both groups and higher anteversion protects from dislocation than rather predisposes to it (Seagrave et al. 2017a). Even though our study group had a slightly higher median anteversion angle, in our material the median anteversion fitted within all suggested safe zones in both the Continuum study group and the control group. Nevertheless, there is no consensus on the optimal acetabular cup angles (Cotong et al. 2017). It is possible that due to earlier reports of the higher dislocation rate in Continuum cups surgeons are aiming for slightly greater anteversion in these cups than in other uncemented devices. This could explain the higher median anteversion observed in the study group. On the other hand, the use of elevated rim liners might also lead to a decrease in aimed anteversion as high anteversion combined with posterior elevation might lead to impingement. As the anteversion was at an acceptable level and slightly larger in the study group than in the control group and the inclination comparable between the groups, difficulties in cup positioning are not likely to explain the higher dislocation rate for Continuum cups in earlier studies. 1 possible explanation is smaller coverage in neutral Continuum liners and smaller jumping distance, which predisposes to dislocation (Sariali et al. 2009). Oblique and elevated liners appear to assess this problem and reduce dislocation risk (Hemmilä et al. 2019). However, the long-term data on stabilizing liners’ effect on implant survival is limited and it is possible that stabilizing liners might cause posterior impingement and therefore predispose to anterior dislocation. Further, mean anteversion was higher in 1 of the study centers compared with the other 2 study centers. One potential explanation for this is that the ContinuumTM system had been used for only a short time in that center and at the beginning of the study period surgeons in this center were aiming for slightly greater anteversion as recommended in the system they used previously. However, the anteversion was within Lewinnek’s safe zone in all 3 study centers and greater anteversion in patients operated on with a

685

posterolateral incision should protect from rather than predispose to dislocation. Thus far TM components have demonstrated better survival compared with other uncemented cups in primary THA in only 1 register study. This study included other uncemented cups only from the same manufacturer and not the best performing uncemented acetabular devices (Matharu et al. 2018a). More long-term register data, especially including elevated and oblique liners, are needed to assess whether the overall revision risk for Continuum cups is lower compared with other uncemented devices at a later stage when aseptic loosening is the main reasons for revision. There has also been speculation based on a small clinical study that TM as a material might have some qualities protecting against PJI (Tokarski et al. 2015). Unfortunately, these results have not been reproducible in later studies and Matharu et al. (2019) advise clinicians to be cautious regarding such claims. We acknowledge that our study has limitations. 1st, we were unable to reliably collect data with dislocation as the endpoint as some of the patients might have changed their treating hospital during the study time. Therefore, we could not study whether malpositioning predisposed to dislocation or try to create our own cup positioning safe zones according to dislocations. 2nd, due to the observational nature of this study, we were unable to randomly assign patients to the study or control group. However, to avoid potential bias, we have matched the groups by age, sex, and BMI, and all included patients were operated on for primary OA with a posterior approach, had an uncemented stem with femoral head size 32 mm or 36 mm, and BMI between 19 and 35. 3rd, we did not have data on surgeons’ experience that might affect to cup positioning and possibly cause bias. Further, another limitation is that the Martell system uses only pelvic AP radiographs when measuring the cup position. Hence there was a need to double check the lateral radiographs manually to ensure that acetabular components were not in retroversion. In conclusion, Continuum acetabular components had greater anteversion compared with the other uncemented cups. However, anteversion was at acceptable level in both groups and this difference does not explain the larger dislocation rate in Continuum cups observed in earlier studies as greater anteversion protects from dislocation rather than predisposes to it. The authors would like to acknowledge Emma Naucler and Ali Rafati for their help with the statistical analysis. IL: Study design, data collection, interpretation of the results, main responsibility for writing of the manuscript. NH: Radiographic measurements, interpretation of the results, writing the manuscript. KG, AE, AT, KM: Study design, data collection, interpretation of the results, writing the manuscript. OR, HM: Study design, interpretation of the results, writing the manuscript. MM: Study design, statistical analysis, interpretation of the results, writing the manuscript. Acta thanks George Grammatopoulos and Marc Nijhof for help with peer review of this study.

686

AOANJRR. Annual report 2017. https://aoanjrr.sahmri.com/annual-reports2017 [accessed January 6, 2020]. Baad-Hansen T, Kold S, Nielsen P T, Laursen M B, Christensen P H, Soballe K. Comparison of trabecular metal cups and titanium fiber-mesh cups in primary hip arthroplasty: a randomized RSA and bone mineral densitometry study of 50 hips. Acta Orthop 2011; 82(2): 155-60. Beckmann N A, Weiss S, Klotz M C M, Gondan M, Jaeger S, Bitsch R G. Loosening after acetabular revision: comparison of trabecular metal and reinforcement rings. A systematic review. J Arthroplasty 2014; 29(1): 229-35. Biedermann R, Tonin A, Krismer M, Rachbauer F, Eibl G, Stöckl B. Reducing the risk of dislocation after total hip arthroplasty: the effect of orientation of the acetabular component. Bone Joint J 2005; 87(6): 762-9. Bobyn J D, Stackpool G J, Hacking S A, Tanzer M, Krygier J J. Characteristics of bone ingrowth and interface mechanics of a new porous tantalum biomaterial. J Bone Joint Surg 1999; 81(5): 907-14. Callanan M C, Jarrett B, Bragdon C R, Zurakowski D, Rubash H E, Freiberg A A, Malchau H. The John Charnley Award: Risk factors for cup malpositioning: quality improvement through a joint registry at a tertiary hospital. Clin Orthop Relat Res 2011; 469: 319-29. Cotong D, Troelsen A, Husted H, Gromov K. Danish survey of acetabular component positioning practice during primary total hip arthroplasty. Dan Med J 2017; 64(4): A5354. Danoff J R, Bobman J T, Cunn G, Murtaugh T, Gorroochurn P, Geller J A, Macaulay W. Redefining the acetabular component safe zone for posterior approach total hip arthroplasty. J Arthroplasty 2016; 31(2): 506-11. Elson L, Dounchis J, Illgen R, Marchand R C, E Padgett D E, Bragdon C R, Malchau H. Precision of acetabular cup placement in robotic integrated total hip arthroplasty. Hip Int 2015; 25(6): 531-6. Ezquerra L, Quilez M P, Pérez M Á, Albareda J, Seral B. Range of movement for impingement and dislocation avoidance in total hip replacement predicted by finite element model. J Med Biol Eng 2017; 37(1): 26-34. Finnish Arthroplasty Register (FAR). Finnish arthroplasty register [Internet] 2017. Available from: www.thl.fi/far [accessed January 6, 2020]. Hailer N P, Weiss R J, Stark A, Kärrholm J. The risk of revision due to dislocation after total hip arthroplasty depends on surgical approach, femoral head size, sex, and primary diagnosis. Acta Orthop 2012; 83(5): 442-8. Hemmilä M, Karvonen M, Laaksonen I, Matilainen M, Eskelinen A, Haapakoski J, Puhto A P, Kettunen J, Manninen M, Mäkela K T. Survival of 11,390 Continuum cups in primary total hip arthroplasty based on data from the Finnish Arthroplasty Register. Acta Orthop 2019; 90(4): 312-17. Jafari S M, Bender B, Coyle C, Parvizi J, Sharkey P F, Hozack W J. Do tantalum and titanium cups show similar results in revision hip arthroplasty? Clin Orthop Relat Res 2010; 468(2): 459-65.

Acta Orthopaedica 2020; 91 (6): 682–686

Laaksonen I, Lorimer M, Gromov K, Rolfson O, Mäkelä K T, Graves S E, Malchau H, Mohaddes M. Does the risk of rerevision vary between porous tantalum cups and other cementless designs after revision hip arthroplasty? Clin Orthop Relat Res 2017; 475(12): 3015-22. Laaksonen I, Lorimer M, Gromov K, Eskelinen A, Rolfson O, Graves S E, Malchau H, Mohaddes M. Trabecular metal acetabular components in primary total hip arthroplasty:h risk for revision compared with other uncemented cup designs in a collaborative register study including 93,709 hips. Acta Orthop 2018; 89(3): 259-64. Lewinnek G E, Lewis J L, Tarr R, Compere C L, Zimmerman J R. Dislocations after total hip-replacement arthroplasties. J Bone Joint Surg Am. 1978; 60(2): 217-20. Martell J M , Berdia S. Determination of polyethylene wear in total hip replacements with use of digital radiographs. J Bone Joint Surg Am 1997; 79(11): 1635-41. Matharu G S, Judge A, Murray D W, Pandit H G. Trabecular metal acetabular components reduce the risk of revision following primary total hip arthroplasty: a propensity score matched study from the National Joint Registry for England and Wales. J Arthroplasty 2018a; 33(2): 447-52. Matharu G S, Judge A, Murray D W, Pandit H G. Do trabecular metal acetabular components reduce the risk of rerevision after revision THA performed for periprosthetic joint infection? A study using the NJR data set. Clin Orthop Relat Res 2019; 477(6): 1382-9. Mohaddes M, Rolfson O, Kärrholm J. Short-term survival of the trabecular metal cup is similar to that of standard cups used in acetabular revision surgery. Acta Orthop 2015; 86(1): 26-31. Sariali E, Lazennec J Y, Khiami F, Catonné Y. Mathematical evaluation of jumping distance in total hip arthroplasty: iInfluence of abduction angle, femoral head offset, and head diameter. Acta Orthop 2009; 80(3): 277-82. Seagrave K G, Troelsen A, Madsen B G, Husted H, Kallemose T, Gromov K. Can surgeons reduce the risk for dislocation after primary total hip arthroplasty performed using the posterolateral approach? J Arthroplasty 2017a; 32(10): 3141-6. Seagrave K G, Troelsen A, Malchau H, Husted H, Gromov K. Acetabular cup position and risk of dislocation in primary total hip arthroplasty. Acta Orthop 2017b; 88(1): 10-17. Tokarski A T, Novack T A, Parvizi J. Is tantalum protective against infection in revision total hip arthroplasty? Bone Joint J 2015; 97-B(1): 45-9. Wegrzyn J, Kaufman K R, Hanssen A D, Lewallen D G. Performance of porous tantalum vs. titanium cup in total hip arthroplasty: randomized trial with minimum 10-year follow-up. J Arthroplasty 2015; 30(6): 1008-13. Zijlstra W P, De Hartog B, Van Steenbergen L N, Scheurs B W, Nelissen R G H H. Effect of femoral head size and surgical approach on risk of revision for dislocation after total hip arthroplasty. Acta Orthop 2017; 88(4): 395-401.

Acta Orthopaedica 2020; 91 (6): 687–692

687

Do hip precautions after posterior-approach total hip arthroplasty affect dislocation rates? A systematic review of 7 studies with 6,900 patients Jack CROMPTON 1,3, Liza OSAGIE-CLOUARD 1,2, and Akash PATEL 1,3 ¹ Royal

Free Hospital, London; ² Institute of Orthopaedics and Musculoskeletal Sciences, University College London; 3 Division of Surgery and Interventional Science, University College London, UK Correspondence: l.osagie@ucl.ac.uk Submitted 2020-02-24. Accepted 2020-07-01.

Background and purpose — Hip precautions limiting flexion, adduction, and internal rotation have been prescribed traditionally to minimize dislocation rates following THA. We assessed the prevalence of hip dislocation following posterior approach total hip arthroplasty without postoperative hip precautions. Methods — A systematic review of multiple medical databases was performed using the PRISMA guidelines and checklist. All clinical outcome studies that reported dislocation rates and postoperative instructions following posterior approach, primary surgery, published within the last 6 years, were included. Results — 6,900 patients were included from 7 Level I– IV studies, with 3,517 treated with and 3,383 without precautions. There was no statistically significant difference in the rates of dislocation between groups (2.2% in restricted group vs. 2.0% in unrestricted group). All but 1 study demonstrated no statistically significant differences in patientreported outcome scores between restricted and unrestricted groups, including those pertaining to return to function, confidence, and pain. Interpretation — The review found no impact on dislocation rates following total hip arthroplasty performed through a posterior approach, regardless of the use of hip precautions. We also found no impact of the prescription of hip precautions on patient-reported outcome scores.

Prosthesis dislocation is a rare complication of posterior approach THA but with a significant impact on mortality and morbidity. Data suggests 11–24% of revision procedures are secondary to recurrent dislocations with a reported dislocation rate post-THA as high as 2.5%, which was traditionally thought to be influenced by surgical approach (Dargel et al. 2014, Skoogh et al. 2019, van der Weegen et al. 2019). Defunctioned abductors, insufficient capsular or short external rotator repair were purported to lead to an increased risk of dislocation compared with direct anterior surgery. However, multiple large-scale retrospective studies have demonstrated no difference in dislocation rates regardless of approach used (Goldstein et al. 2001, Masonis and Bourne 2002, Chechik et al. 2013, Faldini et al. 2018). Traditionally, hip precautions were prescribed postoperatively to reduce the risk of dislocation, commonly avoiding hip flexion beyond 90°, adduction beyond the midline, and internal and external rotation greater than 20° (Lucas 2008, Smith and Sackley 2016). With a move towards active recovery plans, earlier unrestricted mobilization is thought to have no impact on dislocation rates. Studies suggest unrestricted mobilization leads to improved functionality, improved clinical outcomes (Khan et al. 2006, Mikkelsen et al. 2014), reduced healthcare costs, and fewer demands on nursing (Coole et al. 2013), despite which hip precautions are still commonly used. Multiple systematic reviews have investigated the effect of removing hip precautions and the impact on hip dislocation rates. van der Weegen et al. (2016) concluded that reducing hip precautions will not lead to increased dislocation rates postTHA but will improve patient satisfaction. However, Smith et al. (2016) concluded that the evidence at time of publishing was insufficient to determine the potential risks of removing such precautions. Our systematic review focuses on THAs per-

688

formed using the posterior surgical approach from studies published within the last 6 years and the effect on dislocation rates. The primary outcome evaluates dislocation rates in patients undergoing posterior-approach THA with postoperative hip precautions reduced or removed. The secondary outcomes are time to dislocation and patient-reported outcome measures, such as Hip Disability and Osteoarthritis Outcome Scores (HOOS), Visual Analogue Scale (VAS) scores, Oxford Hip Scores (OHS), and patient satisfaction rates.

Methods The review was conducted in accordance with the PRISMA guidelines. Inclusion criteria for studies were those that assessed posterior approach, demonstrated data regarding use of postoperative instructions, and used patient-reported outcomes. Questionnaires, case studies, and reviews were excluded. The online databases PubMed, MEDLINE, Web of Science, and the Cochrane Library were searched, with limits between the dates of January 2013 to October 2019. The following search string “hip arthroplasty AND (precautions OR restrictions) AND dislocation” was utilized, including MeSH terms for “arthroplasty, replacement, hip” and “hip dislocation.” From the identified studies, duplicates were removed, and systematic reviews’ bibliographies manually checked for any studies not yet identified in the search. Abstracts were screened for eligibility, and the eligible studies’ full texts were obtained, where possible, and read to further assess eligibility. Following this, papers were assessed for risk of bias and data extracted from the final studies, using a pre-set data form, including: study year, type of study, number of centers, number of patients, male:female ratio, age, cemented/uncemented, femoral head size, follow-up duration, diagnosis, types of restriction protocol, dislocation rates, time to dislocation, hip outcome scores, pain scores, and time back to ADLs/sport. Ethics, funding, and potential conflicts of interest No ethical approval was required for this study. No funding was received for this work. The authors have no conflicts of interest to declare.

Results Through online database searching 202 papers were identified. These were reduced to 112 when the date range was set to January 2013 until October 2019. After removal of duplicates, 66 papers were assessed; of these, 30 were excluded based on their title and abstract. A further 29 papers were excluded based on their full text, leaving the 7 papers used in this study (Mikkelsen et al. 2014, Gromov et al. 2015, Kornuijt et al. 2016, Allen et al. 2018, Dietz et al. 2019, Peters et al. 2019, van der Weegen et al. 2019) (Figure).

Acta Orthopaedica 2020; 91 (6): 687–692

Papers identified from online database search n = 116 Duplicates removed n = 46 Papers screened on title and abstract n = 66 Excluded based on abstract (n = 30): – no precautions – wrong surgical approach – study design Papers full texts’ screened and evaluated n = 36 Excluded based on full text (n = 19): – wrong /unknown surgical approach, 7 – no precautions, 2 – study design, 8 – no full text/translation available, 2 Papers identified for use in study n=7

Study flow showing papers searched, screened, and included in the review, with reasons for exclusion.

Study design and quality Of the 7 studies included, 5 were prospective and 2 were retrospective, with 2 randomized studies and 5 cohort studies. The follow-up duration ranged from 3 weeks to 1 year, with 2 studies following up on more than 1 occasion. Included were 1 multicenter study from the United States, 3 studies from the Netherlands 2 based in Denmark, and 1 based in the UK (Table 1). A risk of bias assessment for our included studies was conducted using “Cochrane’s tool for assessing risk of bias” (Higgins et al. 2011) for the randomized studies, and the “tool to assess risk of bias in cohort studies” by the CLARITY Group (CLARITY Group 2020) to evaluate the cohort studies (Tables 2 and 3, see Supplementary data). The randomized studies were both at low risk of bias with Dietz et al. (2019) being at high risk for blinding. This was because blinding was not used in this study for either the patients or surgeons. In contrast, Peters et al. (2019) blinded the surgeons to the intervention group to which their patient was allocated. The cohort studies were at high risk because they were all consecutive studies, comparing the 2 groups at different time points. All of the studies had a low risk with respect to the assessment of the exposure, as surgical records were used in most cases. We can also trust the follow-up for the studies, as any missing data were balanced in both groups and plausible reasons for missing data were stated. 3 studies (Mikkelsen et al. 2014, Kornuijt et al. 2016, Allen et al. 2018) were at a higher risk of bias for matching the exposed and unexposed groups for

Acta Orthopaedica 2020; 91 (6): 687–692

689

Table 1. Data collected from each study using a pre-set data collection form Authors (study design, number of centers, and follow-up duration) Patients Male:female Indication Prosthesis Group n (%) ratio for surgery a fixation, n (%) b

Femoral head size, n (%)

Mikkelsen et al. 2014 (Non-randomized controlled study, 1 center, follow-up: 3 and 6 weeks) Restricted 146 (40) 53:47 Osteoarthritis C: 125 (86) ≤ 32 mm: 5 (4) U: 5 (3) 36 mm: 85 (70) H: 15 (10) 40 mm: 29 (24) ≥ 44 mm: 3 (2) Unrestricted 219 (60) 52:48 Osteoarthritis C:190 (87) U: 8 (4) H: 21 (10) Allen et al. 2018 (Retrospective cohort study, 1 center, follow-up: 6 weeks) Restricted 866 (79.1) Not PO: 754 Not reported reported NOF: 7 c Other : 105 Unrestricted 334 (20.9) Not PO: 304 Not reported reported NOF: 3 Other c: 27 van der Weegen et al. 2019 (Cohort study, 1 center, follow-up: 90 days) Restricted 1,102 (51.2) 37:63 PO: 1,068 C: 597 (54.2) Other d—34 U: 495 (44.9) H: 10 (0.9) Unrestricted 1,049 (48.8) 38:62 PO: 1,011 C: 436 (41.6) Other d: 38 U: 597 (56.9) H: 16 (1.5)

≤ 32 mm: 9 (4) 36 mm: 125 (57) 40 mm: 66 (30) ≥ 44 mm: 9 (4)

Dislocation rate, n/N (%) 2/146 (2.7)

6/219 (1.4)

Not reported

10/866 (1.15)

Not reported

4/334 (1.20)

≤ 28 mm: 594 (53.9) ≥ 32 mm: 508 (46.1)

28/1,102 (2.5)

≤ 28 mm: 443 (42.2) ≥ 32 mm: 606 (57.8)

17/1,049 (1.6)

Peters at al. 2019 (Prospective, randomized, non-inferiority study, 1 center, follow-up: 8 week) Restricted 203 (49.8) 46:54 Osteoarthritis Not reported Unrestricted 205 (50.2) 40:60 Osteoarthritis Not reported 32 mm in all patients

3/203 (1.48) 3/205 (1.46)

Dietz et al. 2019 (Randomized, controlled study, 3 centers, follow-up: 2–6 weeks, 3–6 months, and 1 year) Restricted 145 (51.1) 56:44 NOF excluded Not reported 35.3 mm d [34.9–35.7] Unrestricted 139 (48.9) 49:51 NOF excluded Not reported 34.7 mm d [34–35]

2/145 (1.4) 1/139 (0.7)

Kornuijt et al. 2016 (Prospective, comparative safety study, 1 center, follow-up: 3 month) Restricted 109 (50.2) 31:69 PO: 101 C: 53 (49) Other d: 8 U: 56 (51) Unrestricted 108 (49.8) 36:64 PO: 102 C: 40 (37) Other d: 6 U: 68 (63)

≤ 28 mm: 51 (47) ≥ 32 mm: 58 (53) ≤ 28 mm: 42 (39) ≥ 32 mm: 66 (61)

Gromov et al. 2015 (Retrospective, non-inferiority study, data from DNPR e, follow-up: 90 days) Restricted 946 (41.6) 45:55 Not reported Not reported 28 mm: 946 (100) 32 mm: 0 (0) 36 mm: 0 (0) Unrestricted 1,329 (58.4) 39:61 Not reported Not reported 28 mm: 33 (3) 32 mm: 403 (30) 36 mm: 890 (67) a PO = primary osteoarthritis; NOF = neck of femur b C = cemented; U = uncemented; H = hybrid. c Secondary arthritis or inflammatory arthritis. d Mean (95% CI) e DNPR = Danish National Patient Registry.

1/109 (0.9) 0/108 (0)

32/946 (3.4) 37/1,329 (2.8)

fracture.

confounding variables. No statistical adjustment was used in these studies, unlike Gromov et al. (2015), where the odds ratio was adjusted for factors such as age, sex, and femoral head size. Patient characteristics The overall number of patients included in our review was 6,900, with 3,517 allocated to a restricted (or standard precautions) group (RG) and 3,383 allocated to an unrestricted (or reduced precautions) group (UG).

The male:female ratio varied between the different studies; however, 1 study did not record these data (Allen et al. 2018). Overall, there were more females included than males, with 3,375 females compared with 2,325 males. The average age in the studies ranged from 63 to 72 years with details shown in Table 4. The primary indication for surgery was predominantly osteoarthritis. 1 study (Gromov et al. 2015) did not publish data on the diagnosis while Dietz et al. (2019) specifically excluded patients with neck of femur (NOF) fractures.

690

Acta Orthopaedica 2020; 91 (6): 687–692

Table 4. Average age of patients involved in studies Study Mikkelsen et al. 2014, mean (SD) Allen et al. 2018, mean [IQR] van der Weegen et al. 2019, mean [IQR] Peters et al. 2019, mean (SD) Dietz et al. 2019 mean (95% CI) Kornuijt et al. 2016, mean (IQR) Gromov et al. 2015, mean (range)

Average age Restricted Unrestricted 69 (10) 72 [14]

68 (10) 71 [14]

69 [14] 64 (10) 63 (61–64) 69 [16] 67 (20–99)

69 [13] 64 (10) 63 (62–65) 67 [13] 69 (15–104)

Surgery 5 of the studies (Mikkelsen et al. 2014, Gromov et al. 2015, Kornuijt et al. 2016, Peters et al. 2019, van der Weegen et al. 2019) used the standard posterior approach for their surgeries, whereas Dietz et al. (2019) used the mini-posterior approach. Allen et al. (2018) used a variety of approaches, but we have included only data from the posterior approach surgeries in this study. The femoral head size used varied between studies, with data on the different diameters used in 4 of the studies (Mikkelsen et al. 2014, Gromov et al. 2015, Kornuijt et al. 2016, van der Weegen et al. 2019). Peters et al. used 32 mm femoral head sized prosthetics on all patients, whilst Dietz et al. (2019) had a mean femoral head diameter of 35.3 mm in the restricted group and 34.7 mm in the unrestricted group. No data were recorded on this in 1 of the studies (Allen et al. 2018). 4 of the studies did not record data on whether the prosthesis was cemented or uncemented (Gromov et al. 2015, Allen et al. 2018, Dietz et al. 2019, Peters et al. 2019), and 1 study reported the use of a constrained liner (Peters et al. 2019). Restriction and precaution protocols 3 of the studies used the standard hip precautions, in the restricted groups, of no flexion past 90°, no adduction beyond neutral position and no internal rotation (Mikkelsen et al. 2014, Gromov et al. 2015, Dietz et al. 2019). Mikkelsen et al. and Gromov et al. also gave their patients aids, such as elevated toilet seats, extended shoehorns, sock aids, ergonomic reachers, and bath benches. 2 studies used the same detailed precaution protocol for their restricted groups, including the standard hip precautions as above, with additional restrictions in sleeping, car driving, use of pillows and additional equipment (Kornuijt et al. 2016, van der Weegen et al. 2019). The restricted groups in the study by Allen et al. (2018) seemed to follow the standard hip precautions also, but this was not discussed in detail. For the unrestricted groups, the majority of studies (Mikkelsen et al. 2014, Gromov et al. 2015, Kornuijt et al. 2016, Allen et al. 2018, Dietz et al. 2019, van der Weegen et al. 2019) enforced few or no precautions or restrictions on patients, with

no combined flexion, adduction, and internal rotation being allowed in 3 studies (Mikkelsen et al. 2014, Kornuijt et al. 2016, van der Weegen et al. 2019). Allen et al. (2018) detailed a new protocol for the unrestricted group instructing patients to use movements that were “comfortable” and did not “test their range of movement.” Moreover, Kornuijt et al. (2016) and van der Weegen et al. (2019) advised their patients in the unrestricted group only to drive a car once walking without crutches and to use a pillow for comfort only. Also, in these 2 studies, cross-legged sitting and bending with the operated leg moved backwards was restricted in both groups. Peters et al. (2019) took a different approach by restricting flexion beyond 90°, adduction and rotation past the midline in both groups and only restricting the sleeping position in the restricted group (Table 5, see Supplementary data). Dislocation rates Of the total number of THAs performed (6,900) there were 146 dislocations recorded in the 7 studies, with 78 (2.2%) in the restricted group and 68 (2.0%) in the unrestricted group. Among all the studies, Gromov et al. (2015) had the highest rates of dislocation in both the restricted group (3.4%) and unrestricted group (2.8%), and the lowest hip dislocation rates were from the comparative study by Kornuijt et al. (2016), with 1 (0.9%) dislocation in the restricted group and 0 in the unrestricted group. Of the 69 dislocations in the study by Gromov et al. (2015), two-thirds of them occurred within 30 days postoperatively, with 63% in the restricted group and 70% in the unrestricted group occurring in the same time period. The 1 dislocation in the Kornuijt et al. (2016) study occurred 2 weeks after the surgery and all of the dislocations reported by Peters et al. (2019) occurred within 3 weeks, with 1 being as soon as 2 days after the operation. The median number of days to dislocation was 4 (IQR 1–15) and 8 (IQR 1–19) in the restricted and unrestricted groups, respectively, reported by van der Weegen et al. (2019). Furthermore, Mikkelsen et al. (2014) reported the number of days after surgery for each dislocation, occurring at 0, 8, 9, 14, 37, and 40 days for the unrestricted group and 13 and 33 days for the restricted group, whilst the time to dislocation was not recorded in 2 studies for the posterior approach (Allen et al. 2018, Dietz et al. 2019). Clinical outcomes 3 studies recorded no data on patient-reported outcome (Gromov et al. 2015, Kornuijt et al. 2016, van der Weegen et al. 2019). A form of HOOS was used in 3 studies (Mikkelsen et al. 2014, Dietz et al. 2019, Peters et al. 2019) with Dietz et al. (2019) using the HOOS Jr survey and scores for hip outcomes before and after the intervention. For this study, HOOS Jr scores were taken preoperatively, at 2 weeks, 6 weeks, 3–6 months, and 1 year, with the only significant results being in favor of the restricted group at 2 weeks with a mean score of 68 compared with 64 in the unrestricted group. However, there

Acta Orthopaedica 2020; 91 (6): 687–692

was no statistically significant difference in scores at any other time point in the study. The authors also used VAS scores to compare groups, with significant improvements at each follow up (p < 0.004), although there were no significant differences between the 2 groups at any stage in the study. There was also no significant difference found in the rate of pain scores given by patients during the study. Mikkelsen et al. (2014) reported a statistically significant difference in HOOS ADL scores, with the restricted group having the fastest increase in scores; the scores increased by an average of 38 in the restricted group compared with 30 in the unrestricted group in the first 3 weeks postoperatively. There were no significant differences in any of the other HOOS results, including HOOS symptoms, HOOS pain, and HOOS QoL. However, a significantly higher proportion of patients in the unrestricted group were able to perform ADL functions independently 3 weeks postoperatively, including: stair climbing (RG 33%, UG 51%), getting dressed (RG 40%, UG 72%), bath/shower (RG 68%, UG 88%), and house cleaning (RG 38%, UG: 60%). Patients in both groups reported a significant improvement in function on the HOOS and EQ-5D scores at 8 weeks, according to Peters et al. (2019). The delta scores, calculated by “mean baseline score” minus “mean 8-week score”, were -40 (RG) and -44 (UG) (p = 0.09), for the HOOS scores, and -0.32 (RG) and -0.34 (UG) (p = 0.4) for the EQ-5D scores. VAS scores were also used in this study, again showing no statistically significant differences between the 2 groups (RG 38, UG 40). Allen et al. (2018) reported skewed results with their satisfaction scores, with all the median scores being 100 (IQR 90–100), likely due to high frequencies of 90 and 100 scores. There was also no significant difference found between the Oxford hip scores (OHS) at baseline or 1 year, with the restricted group’s median score increasing by 21 and the unrestricted group’s median score increasing by 22.

Discussion This review found the removal of hip precautions, or reduction of restrictions following a posterior approach THA, did not increase dislocation rates. Using a more restrictive protocol led to an increase in HOOS ADL scores postoperatively, and a statistically non-significant trend for reduced protocols improving ADL scores. Of the 7 studies included, only 2 were randomized and 5 were single-center studies. A heterogenous group of prosthetics were used in these studies, which could have an impact on the dislocation rates. The differences in precautions between the restricted and unrestricted groups varied between the studies, with most using standard hip precautions in 1 group and few or no precautions in the other group. All but 1 of the studies using vastly different restriction protocols showed lower

691

dislocation rates in the unrestricted group; however, none of these differences reached statistical significance. Our findings of no statistically significant differences in dislocation rates between the protocols was in keeping with reviews of dislocation rates that were not stratified for surgical approach. A systematic review conducted by van der Weegen et al. (2016) looked at the efficacy of lifestyle restrictions following primary THAs, yet, unlike our study, they included studies conducted as far back as 2005 and covered all types of surgical approach. From the 6 studies they analyzed, no clinically significant negative effect was found when removing or relaxing hip precautions and they also found an improvement in secondary outcomes like our study. Despite this, and other studies (Barnsley et al. 2015, Jobory et al. 2019) reporting similar results with other surgical approaches, hip precautions are still prescribed to patients in as many as two-thirds of hospitals (Gromov et al. 2019). The variation in whether postoperative restrictions are used for posterior approach THAs might be due to the lack of highquality evidence, with only 2 randomized controlled studies being found by our searches. Views on how well hip precautions benefit patients also varies, with studies suggesting that some patients may find it difficult to understand or follow the restrictions (Coole et al. 2013), and even that use of precautions may increase anxiety in THR patients (O’Grady et al. 2002). Conversely, some clinicians in the same study by Coole et al. (2013) felt that prescribing hip precautions improved patient confidence on discharge, and aided tissue repair. It would be useful to conduct a multicenter RCT, collecting more detailed data such as patient-recorded outcome measures (PROMs). Hip dislocation is a multifactorial complication, therefore causality cannot be assessed simply. Various factors are known to affect postoperative risk of dislocation, including age, BMI, sex, comorbidities, surgeon experience, and a host of component factors (Rowan et al. 2018). Moreover, the incidence for hip dislocation is rare and contributes to the difficulty in performing statistically adequate studies with large sample sizes required. Intraoperative technique can vary and have significant consequences for dislocation rates. A major factor for dislocation risk is the femoral head size of the prosthesis, with hip dislocation incidence increasing with smaller femoral head diameters (Rowan et al. 2018). Therefore, the effectiveness of reducing postoperative restrictions should be interpreted cautiously as no standardized rates were found for posterior-approach THA in our search. The limitations of our study include the variety of study types included and the lack of some secondary outcomes data, including pain scores, time to return to ADL, time back to work, and functional hip outcome measures. We could only find 2 randomized studies based on our criteria and search. Our included studies had a total of 6,900 patients, suggesting a large sample size; however, a large proportion (64%) of the sample came from only 2 of the studies, one of which used

692

similar restrictions for both groups, only changing sleeping position (Peters et al. 2019). Ultimately, recent evidence published since 2013 would suggest that removing or relaxing hip precautions and restrictions given to patients following posterior-approach THA has no significant effect on the rates of early dislocation. The paucity of randomized control trials suggests more evidence is needed to determine the significance of effect on return to ADL, work and pain scores, when employing relaxed precautions following posterior approach surgery. Supplementary data Tables 2, 3, and 5 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2020.1795598 JC: Literature search, screening, data collection, and manuscript writing, LOC: Literature search, manuscript writing, and study design. AP: study design and manuscript review. Acta thanks Kirill Gromov and Ola Rolfson for help with peer review of this study.

Allen F C, Skinner D L, Harrison J, Stafford G H. The effect of precautions on early dislocations post total hip arthroplasty: a retrospective cohort study. Hip Int 2018; 28(5): 485-90. doi: 10.1177/1120700018762175. Barnsley L, Barnsley L, Page R. Are hip precautions necessary post total hip arthroplasty? A systematic review. Geriatr Orthop Surg Rehabil 2015; 6(3): 230-5. doi: 10.1177/2151458515584640. Chechik O, Khashan M, Lador R, Salai M, Amar E. Surgical approach and prosthesis fixation in hip arthroplasty world wide. Arch Orthop Trauma Surg 2013; 133(11): 1595-600. doi: 10.1007/s00402-013-1828-0. CLARITY Group. Tool to assess risk of bias in cohort studies [online].http:// help.magicapp.org/knowledgebase/articles/327941-tool-to-assess-riskof-bias-in-cohort-studies (Accessed January 25, 2020). Coole C, Edwards C, Brewin C, Drummond A. What do clinicians think about hip precautions following total hip replacement? Br J Occup Ther 2013; 76(7): 300-7. doi: 10.4276/030802213x13729279114898. Dargel J, Oppermann J, Bruggemann G P, Eysel P. Dislocation following total hip replacement. Dtsch Arztebl Int 2014; 111(51-52): 884-90. doi: 10.3238/arztebl.2014.0884. Dietz M J, Klein A E, Lindsey B A, Duncan S T, Eicher J M, Gillig J D, Smith B R, Steele G D. Posterior hip precautions do not impact early recovery in total hip arthroplasty: a multicenter, randomized, controlled study. J Arthroplasty 2019; 34(7S): S221-7. doi: 10.1016/j.arth.2019.02.057. Faldini C, Stefanini N, Fenga D, Neonakis E M, Perna F, Mazzotti A, Pilla F, Triantafyllopoulos I K, Traina F. How to prevent dislocation after revision total hip arthroplasty: a systematic review of the risk factors and a focus on treatment options. J Orthop Traumatol 2018; 19(1): 17. doi: 10.1186/ s10195-018-0510-2. Goldstein W M, Gleason T F, Kopplin M, Branson J J. Prevalence of dislocation after total hip arthroplasty through a posterolateral approach with partial capsulotomy and capsulorrhaphy. J Bone Joint Surg Am 2001; 83-A(Suppl 2[Pt ]): 2-7. doi: 10.2106/00004623-200100021-00002. Gromov K, Troelsen A, Otte K S, Orsnes T, Ladelund S, Husted H. Removal of restrictions following primary THA with posterolateral approach does

Acta Orthopaedica 2020; 91 (6): 687–692

not increase the risk of early dislocation. Acta Orthop 2015; 86(4): 463-8. doi: 10.3109/17453674.2015.1028009. Gromov K, Troelsen A, Modaddes M, Rolfson O, Furnes O, Hallan G, Eskelinen A, Neuvonen P, Husted H. Varying but reduced use of postoperative mobilization restrictions after primary total hip arthroplasty in Nordic countries: a questionnaire-based study. Acta Orthop 2019; 90(2): 143-7. doi: 10.1080/17453674.2019.1572291. Higgins J P, Altman D G, Gotzsche P C, Juni P, Moher D, Oxman A D, Savovic J, Schulz K F, Weeks L, Sterne J A. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011; 343: d5928. doi: 10.1136/bmj.d5928. Jobory A, Rolfson O, Akesson K E, Arvidsson C, Nilsson I, Rogmark C. Hip precautions not meaningful after hemiarthroplasty due to hip fracture: cluster-randomized study of 394 patients operated with direct anterolateral approach. Injury 2019; 50(7): 1318-23. doi: 10.1016/j.injury.2019.05.002. Khan R J, Fick D, Khoo P, Yao F, Nivbrant B, Wood D. Less invasive total hip arthroplasty: description of a new technique. J Arthroplasty 2006; 21(7): 1038-46. doi: 10.1016/j.arth.2006.01.010. Kornuijt A, Das D, Sijbesma T, van der Weegen W. The rate of dislocation is not increased when minimal precautions are used after total hip arthroplasty using the posterolateral approach: a prospective, comparative safety study. Bone Joint J 2016; 98-b(5): 589-94. doi: 10.1302/0301-620x.98b5.36701. Lucas B. Total hip and total knee replacement: postoperative nursing management. Br J Nurs 2008; 17(22): 1410-4. doi: 10.12968/bjon.2008. 17.22.31866. Masonis J L, Bourne R B. Surgical approach, abductor function, and total hip arthroplasty dislocation. Clin Orthop Relat Res [Conference Paper] 2002; (405): 46-53. doi: 10.1097/00003086-200212000-00006. Mikkelsen L R, Petersen M K, Soballe K, Mikkelsen S, Mechlenburg I. Does reduced movement restrictions and use of assistive devices affect rehabilitation outcome after total hip replacement? A non-randomized, controlled study. Eur J Phys Rehabil Med 2014; 50(4): 383-93. O’Grady P, Rafiq T, Sherazi S. Sleep deprivation following total hip arthroplasty. Ir J Med Sci 2002; 171: Art no. 61. doi: https://doi.org/10.1007/ BF03170104. Peters A, ter Weele K, Manning F, Tijink M, Pakvis D, Veld R. Less postoperative restrictions following total hip arthroplasty with use of a posterolateral approach: a prospective, randomized, noninferiority trial. J Arthroplasty 2019; 34(10): 2415-9. doi: 10.1016/j.arth.2019.05.038. Rowan F E, Benjamin B, Pietrak JR, Haddad F S. Prevention of dislocation after total hip arthroplasty. J Arthroplasty 2018; 33(5): 1316-24. doi: 10.1016/j.arth.2018.01.047. Skoogh O, Tsikandylakis G, Mohaddes M, Nemes S, Odin D, Grant P, Rolfson O. Contemporary posterior surgical approach in total hip replacement: still more reoperations due to dislocation compared with direct lateral approach? An observational study of the Swedish Hip Arthroplasty Register including 156,979 hips. Acta Orthop 2019; 90(5): 411-16. doi: 10.1080/17453674.2019.1610269. Smith T O, Sackley C M. UK survey of occupational therapists’ and physiotherapists’ experiences and attitudes towards hip replacement precautions and equipment. BMC Musculoskelet Disord 2016; 17:228. Smith T O, Jepson P, Beswick A, Sands G, Drummond A, Davis E T, Sackley C M. Assistive devices, hip precautions, environmental modifications and training to prevent dislocation and improve function after hip arthroplasty. Cochrane Database Syst Rev 2016; 7: Cd010815. doi: 10.1002/14651858. CD010815.pub2. van der Weegen W, Kornuijt A, Das D. Do lifestyle restrictions and precautions prevent dislocation after total hip arthroplasty? A systematic review and meta-analysis of the literature. Clin Rehabil 2016; 30(4): 329-39. doi: 10.1177/0269215515579421. van der Weegen W, Kornuijt A, Das D, Vos R, Sijbesma T. It is safe to use minimal restrictions following posterior approach total hip arthroplasty: results from a large cohort study. Hip Int 2019; 29(6): 572-7. doi: 10.1177/1120700018823504.

Acta Orthopaedica 2020; 91 (6): 693–698

693

Migration of the uncemented Echo Bi-Metric and Bi-Metric THA stems: a randomized controlled RSA study involving 62 patients with 24-month follow-up Karen DYREBORG 1, 2, Mikkel R ANDERSEN 1, Nikolaj WINTHER 1, Søren SOLGAARD 2, Gunnar FLIVIK 3, and Michael M PETERSEN 1 1 Department

of Orthopaedic Surgery, Rigshospitalet, University of Copenhagen, Denmark; 2 Department of Hip and Knee Surgery, Herlev-Gentofte University Hospital, Hellerup, Denmark; 3 Department of Orthopedics, Lund University and Skåne University Hospital, Lund, Sweden Correspondence: karendyreborg@hotmail.com Submitted 2020-04-02. Accepted 2020-07-10.

Background and purpose — Despite the good results after total hip arthroplasty (THA), new implants are continuously being developed to improve durability. The Echo BiMetric (EBM) THA stem is the successor to the Bi-Metric (BM) THA stem. The EBM stem includes many of the features of the BM stem, but minor changes in the design might improve the clinical performance. We compared the migration behavior with radiostereometric analysis (RSA) of the EBM stem and the BM stem at 24 months and evaluated the clinical outcome. Patients and methods — We randomized 62 patients with osteoarthritis (mean age 64 years, female/male 28/34) scheduled for an uncemented THA to receive either an EBM or a BM THA stem. We performed RSA within 1 week after surgery and at 3, 6, 12, and 24 months. The clinical outcome was evaluated using Harris Hip Score (HHS) and Oxford Hip Score (OHS). Results — At 24 months, we found no statistically significant differences in migration between the two implants. During the first 3 months both the EBM and the BM stems showed visible subsidence (2.5 mm and 2.2 mm respectively), and retroversion (2.5° and 2.2° respectively), but after 3 months this stabilized. The expected increase in HHS and OHS was similar between the groups. Interpretation — The EBM stem showed a migration at 24 months not different from the BM stem, and both stems display satisfying clinical results.

To improve the longevity of total hip arthroplasty (THA) new designs are continuously being developed. The introduction of new implants should optimally be done by phased stepwise introduction (Malchau 1995, Nelissen et al. 2011) including radiostereometric analysis (RSA) of implant migration. Some subsidence of hip stems is generally accepted within the first 3 months, but after that osseointegration and stability should have occurred. Mean subsidence of up to 1 mm of the stem at 24 months has been reported (Nysted et al. 2014, Weber et al. 2014, Hoornenborg et al. 2018, Sesselmann et al. 2018, Kruijntjens et al. 2020). This study investigates by RSA potential differences in migration at 24 months, between 2 different designs of porous-coated uncemented hip prosthesis; the Bi-Metric Full Proximal Profile THA stem (BM) and the Echo Bi-Metric (EBM) stem (Zimmer Biomet, Warsaw, IN, USA) (Figure 1). Both stems are press-fit titanium alloy stems with a proximal plasma spray porous titanium coating and a distal part with a roughened titanium surface. The BM has shown good clinical results and excellent stem survival in register studies since its introduction in 1984 (Jacobsen et al. 2003, Davies et al. 2010, Mäkelä et al. 2010, Lazarinis et al. 2011). The EBM is the successor to the BM and has 3 theoretical design improvements: a slimmer design of the neck to increase range of motion; a polished bullet-shaped distal tip to reduce distal stress; and an extended porous coating to support biological ingrowth proximally. Evaluation of adaptive bone remodeling and stress shielding will be addressed in another publication. We hypothesized that the migration of the EBM was less at 24 months, compared with the BM stem.

694

Acta Orthopaedica 2020; 91 (6): 693–698

Figure 2. (A) The RSA coordinate system with movements; (B) model-based RSA screenshot while using the RSAcore software.

Figure 1. The stems: on the left the BiMetric stem and on the right the Echo Bi-Metric stem.

Patients and methods Design and participants This study is a randomized controlled trial, allocation ratio 1:1. The inclusion criteria were: patients with primary osteoarthritis scheduled to undergo THA at the Herlev-Gentofte University Hospital, Department of Hip and Knee Surgery, age 30–75 years, and informed consent. The exclusion criteria were: infection, diseases affecting the bone metabolism (osteoporosis, osteomalacia, Paget’s disease, hypo- or hyperparathyroidism, vitamin D deficiencies, cancer, avascular necrosis, or rheumatoid arthritis), pregnancy, inability to cooperate, inability to communicate in Danish, and medicine or alcohol abuse. The secondary exclusion criterion was too few markers visible. Randomization and blinding Prior to study start the randomization code was generated by a web-based program and envelopes with the individual allocation were sequentially numbered. The allocation was performed as block randomization with blocks of 10 and the sequence was locked away. The randomization allocation sequence and packaging of non-transparent and closed envelopes was done by a colleague outside the project. Screening and enrollment were done by the primary investigator (KD). When the patient was ready in the operating theater, the numbered envelope was opened. Due to visual differences of the 2 prostheses, the surgeon and health personnel were not blinded. The participants were all blinded.

Surgery All patients received either an uncemented EBM or BM THA stem. A 32 mm CoCr head and an Exceed ABT RingLoc-x acetabular shell (Zimmer Biomet) with a highly cross-linked polyethylene liner were implanted in all patients. Surgery was performed with a posterolateral approach by 1 of 4 experienced hip surgeons. Prior to each procedure and allocation, the surgeons templated the stems on calibrated radiographs to anticipate the size and position of the stem and cup. The stems do not differ in length, diameter, or offset for a given size. Surgery was performed under spinal or general anesthesia. On the day of surgery, the patients were mobilized with full weight-bearing using crutches and physiotherapy began. All patients were given oral anticoagulants (rivaroxaban until discharge), and prophylactic antibiotics (dicloxacillin, 2 g preoperatively and 1 g x 2 postoperatively) during the first 24 hours. RSA During surgery 8 to 10 tantalum markers (0.8 mm) were inserted in a well-scattered manner into trochanter minor and trochanter major, respectively. Instead of markers on the stems we used CAD models from the RSA software manufacturer with an added 3D surface model of the spherical head as described by Prins et al. (2008), giving us the possibility of model based-RSA (MB-RSA). Within 1 week of discharge the patients had their baseline RSA radiographs taken at the Department of Diagnostic Radiology at Rigshospitalet, Copenhagen, Denmark (mean days from surgery: BM = 7 and EBM = 6). The required set up with 2 ceilingmounted X-ray tubes was used with a uniplanar calibration cage (RSA Biomedical cage 41), defining the coordinate system, to take 2 simultaneous digital radiographs at an angle of 42° apart. The patients were placed in a supine position and the operated limb fixed in maximum external rotation to visualize as much as possible of the lesser trochanter (Figure 2). The follow-up examinations were scheduled at 3, 6, 12, and 24 months. Migration of the stems was calculated using CAD models in the model-based RSA software (version 4.1;

Acta Orthopaedica 2020; 91 (6): 693–698

RSAcore, Department of Orthopaedic Surgery, LUMC, the Netherlands) at the Biomechanics and RSA laboratory at Skåne University Hospital, Lund, Sweden. Migrations for left-sided prostheses were recalculated, making all results right-handed. 55 double examinations were performed with total repositioning of the patients, to estimate the precision of the RSA set-up, i.e., the random deviation. Precision error (PE), defined as 2 standard deviations, was calculated for the rotational and translational segment motions. PE for X-, Y-, and Z-translation was 0.15 mm, 0.27 mm, and 0.54 mm, respectively. For the X-, Y-, and Z-rotation PE was 0.82°, 2.32°, and 0.25°, respectively. Mean error of rigid body fitting was limited to 0.35 mm. The condition number limit was set at 150. Clinical outcome We used Harris Hip Score (HHS) and Oxford Hip Score (OHS) (Fitzpatrick et al. 2007, Paulsen et al. 2012) preoperatively and at 6, 12, and 24 months to evaluate the clinical outcome. Outcomes The primary outcome measure was Y-translation at 24 months. Secondary outcome measures were (1) Y-rotation, X- and Z-movements at the time intervals 3, 6, 12, and 24 months and (2) clinical outcome, monitored with HHS and OHS (postoperatively and at 6, 12, and 24 months). The minimally important difference estimate for HHS is 18 (Singh et al. 2016); for OHS it is 5 (Beard et al. 2015). Sample size Our power analysis was based on the standard deviation (SD) for the migration after 2 years of follow-up from 2 previously published studies, with information regarding SD (Ström et al. 2006, Wierer et al. 2013). Hence, our SD was 0.69 mm, our minimal relevant difference (MIREDIF) = 0.6 mm, type I error = 5%, and type II error = 15% (resulting in a standardized difference of 0.87). Aiming for a power of 80–90% required a minimum of 23 hips in each group. We included 31 in each group to accommodate future dropouts. Statistics Data were tested for normality by histogram, QQ plot, and Kolmogorov–Smirnov, and the data were not normal distributed. For evaluation of potential differences between the mean migrations, we used a Mann–Whitney U-test. All data are presented as mean (SD) unless otherwise reported. The data on the secondary outcomes were considered as being exploratory in nature, and were adjusted for multiplicity only if statistically significant differences (p < 0.05) were found. 95% confidence intervals (CI) were calculated for RSA data. The statistical software SPSS version 24 (IBM Corp, Armonk, NY, USA) was used.

695

ENROLLMENT

Assessed for eligibility n = 116 Excluded (n = 54): – declined, 48 – pilots, 4 – disease affecting bone metabolism, 2 Randomized n = 62

ALLOCATION

Allocated to Echo Bi-Metric (n = 31) Received allocated interventiion (n = 31)

Allocated to Bi-Metric (n = 31) Received allocated interventiion (n = 31)

FOLLOW-UP

Lost to follow-up (n = 1): – did not respond, 1

Lost to follow-up (n = 4): – revised before 3 months, 2 – died of unrelated causes, 2 ANALYSIS

Analyzed with 24-months RSA (n = 30)

Analyzed with 24-months RSA (n = 27)

Figure 3. Flowchart

Ethics, registration, funding, and potential conflicts of interest Written informed consent was obtained from all the subjects before enrollment. This RCT was approved by the local Regional Ethics Committee (H-4-2014-079) and by the Danish Data Protection Agency (GEH-2015-079, I-Suite no. 03764) and registered at ClinicalTrials.gov (NCT02656771) prior to enrollment as a combined registration for the present study and a study of adaptive bone remodeling (not yet published). We have not changed the endpoints after trial initiation; however, we have specified our primary outcome after trial initiation, since its formulation was too imprecise (Evans 2007). The study was carried out in accordance with the principles of the Helsinki Declaration, and data are presented following the CONSORT statement and The guidelines for standardisation of radiostereometry of implants (Valstar et al. 2005). The study was partly funded by Zimmer Biomet (grant number C004287X). Zimmer Biomet had no access to data or impact on the data interpretation. MMP reports grants from Zimmer Biomet during the conduct of the study. All other authors have no conflict of interests related to the manuscript.

Results From February 2016 to September 2017 we screened 116 patients, enrolled and randomized 62 patients (mean age = 64 years [49–74], female/male = 28/34) to receive either an EBM (n = 31) or a BM (n = 31) THA stem (Figure 3). The distribution of THAs among the 4 surgeons was 5, 6, 23, and 28 patients, respectively. In the EBM group, 1 patient was lost to follow-up for unknown reasons (did not respond to contact attempts). In the BM group, 4 patients were lost to follow-up; 2 were revised before 3 months (1 periprosthetic fracture and

696

Acta Orthopaedica 2020; 91 (6): 693–698

Table 1. Baseline demographics. Values are mean (range) unless otherwise specified

Bi-Metric

Echo Bi-Metric

Age, years Sex, male/female, n Height, m Weight, kg BMI Operated side, R/L, n Cup size Stem size

66 (49–74) 17/14 1.77 (1.60–1.96) 84 (50–124) 27 (18–38) 18/13 56 (50–62) 12 (9–16)

63 (50–74) 17/14 1.76 (1.60–1.91) 83 (54–122) 27 (20–36) 15/16 56 (50–62) 11 (7–16)

Y-rotation (°)

Y-translation (mm)

3.5

BM EBM

–0.5

3.0

–1.0

2.5

–1.5

2.0

–2.0

1.5

–2.5

1.0

–3.0

0.5

–3.5

Follow-up in months

1 undersized stem) and 2 died of causes unrelated to the surgery (Figure 3, Table 1).

Figure 4. Mean Y-translation at each follow-up for the 2 stems (bars are standard error of mean).

Primary outcome Subsidence (negative Y-translation) at 24 months for the BM was 2.3 mm, for EBM 2.7 mm (p = 0.6). For both the BM and the EBM initial subsidence was seen up to 3 months (Table 2, Figure 4). At 3 months the stems had subsided 2.5 mm for the EMB and 2.2 mm for the BM stem with no statistically significant differences between the groups at 3 months or at any of the measuring points from 3 to 24 months. There were no statistical outliers regarding Y-translation (outliers as defined by SPSS). Secondary outcomes The ante-retroversion (Y-rotation) was similar between the groups: from 0–3 months retroversion of 2.2° for the BM and 2.5° for the EBM stem was observed and then both stem types stabilized (Table 2, Figure 5).

BM EBM 0

Follow-up in months

Figure 5. Mean Y-rotation at each follow-up for the 2 stems (bars are standard error of mean).

From 3 to 24 months we found 3 consequent outliers in Y-rotation (Table 3). The varus–valgus tilt (Z-rotation) was similar for both groups, with a little valgus tilt (at 24 months BM = 2.0°, EBM = 2.0°) (Table 2). The anterior/posterior tilt (X-rotation) as well as X- and Z-translation stabilized after 3 months (Table 2). HHS and OHS increased similarly in both groups (Table 4).

Discussion

In this RCT comparing the EBM with the BM, we found no statistically significant differences in migration or clinical outcome. Some degree of stem subsidence and retroversion of uncemented Table 2. Mean segment motion (95% CI) of the Bi-Metric (BM) and the Echo Bi-Metric (EBM) hip prostheses has previously been stems at follow-up until 24 months. Outliers are included in the analysis reported and was confirmed by our study. The stems subsided approxi 3 months 6 months 12 months 24 months mately 2.5–3 mm at 24 months. BM/EBM, n 29/31 29/31 28/31 27/30 Osseointegration can be assumed if an RSA study shows stable migraStem translation, mm Y–translation: proximal (+), distal (–) tion values after 3 months. Never BM –2.2 (–2.3 to –1.7) –2.0 (–2.5 to –1.6) –2.1 (–2.6 to –1.6) –2.3 (–3.0 to –1.7) theless, the subsidence was higher EBM –2.5 (–3.2 to –1.9) –2.4 (–3.0 to –1.7) –2.4 (–3.0 to –1.7) –2.7 (–3.5 to –2.0) than reported in other recent stud X–translation BM –1.1 (–0.5 to –0.1) –0.4 (–0.7 to –0.1) 0.1 (–0.3 to 0.0) –0.4 (–0.6 to –0.1) ies, where Y-translation is between EBM –0.5 (–0.7 to –0.3) –0.6 (–0.9 to –0.4) –0.3 (–0.4 to –0.1) –0.5 (–0.8 to –0.3) –0.03 mm and –0.99 mm at 24 Z–translation months (Nysted et al. 2014, Weber BM –0.8 (–1.0 to –0.5) –0.5 (–0.7 to –0.3) –0.9 (–1.2 to –0.6) –0.7 (–1.1 to –0.4) EBM –0.8 (–1.0 to –0.6) –0.4 (–0.6 to –0.2) –0.9 (–1.2 to –0.7) –0.9 (–1.1 to –0.6) et al. 2014, Nebergall et al. 2016, Stem rotation, degrees Mahmoud et al. 2017, Hoornen Y–rotation: anteversion (–), retroversion (+) borg et al. 2018, Sesselmann et al. BM 2.2 (1.5 to 5.0) 2.4 (1.7 to 3.2) 2.5 (1.7 to 3.2) 2.5 (1.6 to 3.3) EBM 2.5 (1.8 to 3.5) 2.6 (1.9 to 3.5) 2.7 (1.9 to 3.7) 2.7 (1.9 to 3.8) 2018, Klein et al. 2019, Kruijntjens X–rotation et al. 2020). Nebergall et al. (2016) BM –0.6 (–0.9 to –0.3) –0.2 (–0.5 to 0.1) –0.6 (–1.0 to –0.2) 1.2 (0.7 to 1.6) (the Taperloc stem), reported a EBM –0.4 (–0.8 to –0.03) 0.0 (–0.2 to 0.3) –0.8 (–1.2 to –0.5) 1.1 (0.7 to 1.6) Z–rotation: varus (–), valgus (+) single outlier of –0.9 mm with the BM 1.7 (1.0 to 2.5) 1.7 (0.9 to 2.5) 1.8 (1.0 to 2.7) 2.0 (1.1 to 2.8) rest of the cohort in a range of –0.1 EBM 1.7 (0.9 to 2.4) 1.8 (1.0 to 2.5) 1.8 (1.1 to 2.6) 2.0 (1.2 to 2.8) mm to –0.2 mm. Hence, they found

Acta Orthopaedica 2020; 91 (6): 693–698

697

Table 3. Y-rotation (°) outliers and and their 24-month clinical score ID (group) 3 months 6 months 12 months 24 months OHS HHS 6 (EBM) 8 (BM) 23 (BM)

12.8 6.9 –3.3

13.4 7.0 –4.2

14.0 7.1 –4.8

14.3 8.0 –5.2

48 43 48

100 100 100

Table 4. Clinical outcome. Values are mean (range) Score HSS Preoperative 6 months 12 months 24 months OHS Preoperative 6 months 12 months 24 months

EBM p-value

61 (38–82) 92 (61–100) 97 (81–100) 99 (91–11)

67 (31–85) 91 (70–100) 95 (73–100) 98 (49–100)

0.1 0.2 1.0 0.6

24 (10–35) 43 (29–48) 46 (31–48) 47 (43–48)

23 (9–40) 44 (27–48) 44 (26–48) 46 (21–48)

0.7 0.5 0.3 0.9

more than 10 times less migration compared with our findings. However, their median time from operation to first follow-up was 14 days (0–43), which makes their “postoperative value” closer to the “value at stabilization.” In the study by Søballe et al. (1993) of the Bi-Metric with or without hydroxyapatite (HA) coating, a subsidence of 0.07– 0.09 mm was found at 12 months; again, close to 10 times less than in our study. However, their study included only 15 patients, making it somewhat underpowered. Hoornenborg et al. (2018) reported the migration and clinical outcome of the Zweymüller with or without HA coating and found a Y-translation of –0.46 mm (CI –0.73 to –0.19) with HA coating and –0.73 mm (CI –1.18 to –0.29) without HA coating. It has been estimated by Kärrholm et al. (1997) that subsidence of cemented stems should be less than 1.5 mm at 24 months; however, in recent years it has been proposed that for uncemented stems it is not the exact value of the subsidence, but instead the migration pattern, that anticipates the fate of the arthroplasty (Weber et al. 2014, Critchley et al. 2020). Accordingly, if a pattern of early subsidence is followed by stabilization at 3 months, one should not worry about stem failure. This seems to be true for our cohort. A possible explanation for the level of subsidence in our study might be the fear of creating a fissure or fracture by excessively forcing the stem into the femoral canal, combined with the collarless design of both stems. To investigate whether the surgeons could have tended to undersize the femoral components, a post hoc evaluation of each templated radiograph compared with the actual stem size together with a classification of Dorr type was performed (Ashraf 2018). Only in 1 case was a stem selected with a smaller size than templated (templated to EBM 13, actual EBM 12). No femurs were Dorr type C.

We had 3 outliers with retroversion (Y-rotation), up to approximately 14° in participant number 6 (Table 3). All 3 scored maximum points (or close to) in the clinical outcome measurements at 24 months. Number 6 is an otherwise healthy man with a BMI of 25 and he received a standard EBM stem size 12. When reviewing and comparing the regular radiographs postoperatively (before mobilization) with the RSA about 1 week later we could see subsidence with the naked eye. It seems the stem subsided and rotated at mobilization and the stem chosen might have been too small. This had no clinical consequences; he had “no complaints at all” during a telephone call just before submission of this manuscript (approximately 3 years postoperatively). In patient number 8 there is also a visible difference between the regular radiographs before mobilization and the first RSA, but this is not the case with number 23. Concerning the varus–valgus tilt (Z-rotation), normally stems migrate into a slight varus position. However, surprisingly we found the stems going into valgus. A possible explanation could be too extensive preparation of the trochanter major before insertion of the stem. This could also explain some of the subsidence. However, we have no obvious explanation for this phenomenon. Clinical outcomes were satisfying with scores very close to the maximum for both groups at 24 months. The study by Klein et al. (2019), in which the Corail stem was evaluated prospectively, reports the median (range) HHS at 24 months as 100 (44–100) and the median (range) for OHS at 24 months as 45 (15–48), which is compatible with our results. Our precision error, measured by double examinations, was comparable to the Nebergall et al. (2016) study. Nevertheless, we find there is room for improvement and this could be explained by our manual set-up with no automatic setting of X-ray tubes or fixed position of the calibration cage. To reduce bias, only the primary investigator handled the RSA set-up. A limitation of our study might be the use of MB-RSA (Nazari-Farsani et al. 2016) and a phantom trial to determine the accuracy of our MB-RSA compared with marker-based RSA would have been ideal. Still, the advantage of not having to produce new adjusted stems with attached markers, and in some way changing the design, is desirable. Another limitation of this study is the rather small sample size. Although we included 11 more individuals than estimated by the power calculation it is not a large cohort and a non-inferiority/equality study would be interesting. However, this would require an even larger cohort and is a slow process for a single hospital. A certain level of selection bias might have reduced the quality of data, as the patients who accept to participate in this type of study are probably the most resourceful of all eligible, in the sense that they have accepted extra follow-ups with the hassle this entails. The generalizability of our findings is rather narrow because the results apply only to the population of otherwise healthy people who need a THA.

698

In conclusion, our hypothesis that the EBM stem migrates less than the BM stem was rejected. Thus, the added design features do not seem to have any consequences concerning stem migration or clinical outcome, suggesting negligible differences.

The authors thank the hip surgeons, J. Retpen, A. G. Kjersgaard and M. Skettrup, who, besides co-author S. Solgaard, agreed to perform the THA surgery according to the protocol. They thank the surgical staff at Gentofte Hospital and chief radiographer Torben Kragelund and radiographer Mohammad Khonbat Lauritsen at the department of diagnostic radiology, Rigshospitalet. Finally, they thank Håkan Leijon of the RSA laboratory in Lund, Skåne University Hospital. Protocol and planning were performed by KD, MMP, SS, NW, GF, and MRA. Recruitment of patients was done by KD and SS. The RSA examinations and all analysis of RSA were performed by KD. The paper was written by KD and reviewed by MMP, SS, NW, GF, and MRA. Acta thanks Lennard Koster and Marc Nijhof for help with peer review of this study.

Ashraf M. Classifications used in total hip arthroplasty. In: Total hip replacement: an overview. Ed. Bagaria V. IntecOpen; 2018. p 19-33. Beard D J, Harris K, Dawson J, Doll H, Murray D W, Carr A J, Price A J. Meaningful changes for the Oxford hip and knee scores after joint replacement surgery. J Clin Epidemiol 2015; 68(1): 73-9. Critchley O, Callary S, Mercer G, Campbell D, Wilson C. Long-term migration characteristics of the Corail hydroxyapatite-coated femoral stem: a 14-year radiostereometric analysis follow-up study. Arch Orthop Trauma Surg 2020; 140(1): 121-7. Davies H, Ollivere B, Motha J, Porteous M, August A. Successful performance of the Bi-Metric uncemented femoral stem at a minimum follow-up of 13 years in young patients. J Arthroplasty 2010; 25(2): 186-90. Evans S. When and how can endpoints be changed after initiation of a randomized clinical trial? PLoS Clin Trials 2007; 2(4): 1-3. Fitzpatrick R, Carr A, Murray D. The Oxford hip score: A guide to the scoring system. Oxford: Oxford University Innovation; 2007. Hoornenborg D, Sierevelt I N, Spuijbroek JA, Cheung J, van der Vis H M, Beimers L, Haverkamp D. Does hydroxyapatite coating enhance ingrowth and improve longevity of a Zweymuller type stem? A double-blinded randomised RSA trial. Hip Int 2018; 28(2): 115-21. Jacobsen S, Jensen F K, Poulsen K, Stürup J, Retpen J B. Good performance of a titanium femoral component in cementless hip arthroplasty in younger patients: 97 arthroplasties followed for 5–11 years. Acta Orthop Scand 2003; 74(3): 248-52. Kärrholm J, Herberts P, Hultmark P, Malchau H, Nivbrant B, Thanner J. Radiostereometry of hip prostheses: review of methodology and clinical results. Clin Orthop Relat Res 1997; (344): 94-110. Klein L J, Puretic G, Mohaddes M, Kärrholm J. Similar clinical results and early subsidence between the Collum Femoris Preserving and the Corail stem: a randomized radiostereometric study of 77 hips with 2 years’ followup. Acta Orthop 2019; 90(3): 202-8. Kruijntjens D S M G, Koster L, Kaptein B L, Jutten L M C, Arts J J, Ten Broeke R H M. Early stabilization of the uncemented Symax hip stem in a 2-year RSA study. Acta Orthop 2020; May 12 [Epub ahead of print].

Acta Orthopaedica 2020; 91 (6): 693–698

Lazarinis S, Kärrholm J, Hailer N P. Effects of hydroxyapatite coating on survival of an uncemented femoral stem: a Swedish Hip Arthroplasty Register study on 4,772 hips. Acta Orthop 2011; 82(4): 399-404. Mahmoud A N, Kesteris U, Flivik G. Stable migration pattern of an ultrashort anatomical uncemented hip stem: a prospective study with 2 years radiostereometric analysis follow-up. Hip Int 2017; 27(3): 259-66. Mäkelä K T, Eskelinen A, Paavolainen P, Pulkkinen P, Remes V. Cementless total hip arthroplasty for primary osteoarthritis in patients aged 55 years and older: results of the 8 most common cementless designs compared to cemented reference implants in the Finnish Arthroplasty Register. Acta Orthop 2010; 81(1): 42-52. Malchau H. On the importance of stepwise introduction of new hip implant technology: assessment of total hip replacement using clinical evaluation, radiostereometry, digitised radiography and a national hip registry [thesis]. Göteborg University; 1995. Nazari-Farsani S, Finnilä S, Moritz N, Mattila K, Alm J J, Aro H T. Is modelbased radiostereometric analysis suitable for clinical trials of a cementless tapered wedge femoral stem? Clin Orthop Relat Res 2016; 474(10): 2246-53. Nebergall A K, Rolfson O, Rubash H E, Malchau H, Troelsen A, Greene M E. Stable fixation of a cementless, proximally coated, double wedged, double tapered femoral stem in total hip arthroplasty: a 5-year radiostereometric analysis. J Arthroplasty 2016; 31(6): 1267-74. Nelissen R G H H, Pijls B G, Kärrholm J, Malchau H, Nieuwenhuijse M J, Valstar E R. RSA and registries: the quest for phased introduction of new implants. J Bone Joint Surg Am 2011; 93: 62-5. Nysted M, Foss O A, Klaksvik J, Benum P, Haugan K, Husby O S, Aamodt A. Small and similar amounts of micromotion in an anatomical stem and a customized cementless femoral stem in regular-shaped femurs. Acta Orthop 2014; 85(2): 152-8. Paulsen A, Odgaard A, Overgaard S. Translation, cross-cultural adaptation and validation of the Danish version of the Oxford hip score: assessed against generic and disease-specific questionnaires. Bone Joint Res 2012; 1(9): 225-33. Prins A H, Kaptein B L, Stoel B C, Nelissen R G H H, Reiber J H C, Valstar E R. Handling modular hip implants in model-based RSA: combined stem– head models. J Biomech 2008; 41(14): 2912-17. Sesselmann S, Hong Y, Schlemmer F, Hussnaetter I, Mueller L A, Forst R, Tschunko F. Radiostereometric migration measurement of an uncemented Cerafit® femoral stem: 26 patients followed for 10 years. Biomed Tech 2018; 63(6): 657-63. Singh J A, Schleck C, Harmsen S, Lewallen D. Clinically important improvement thresholds for Harris Hip Score and its ability to predict revision risk after primary total hip arthroplasty. BMC Musculoskelet Disord 2016; 17(1): 1-8. Søballe K, Toksvig-Larsen S, Gelineck J, Fruensgaard S, Hansen E S, Ryd L, Lucht U, Bünger C. Migration of hydroxyapatite coated femoral prostheses: a Roentgen stereophotogrammetric study. J Bone Joint Surg Br 1993; 75(5): 681-7. Ström H, Kolstad K, Mallmin H, Sahlstedt B, Milbrink J. Comparison of the uncemented Cone and the cemented Bimetric hip prosthesis in young patients with osteoarthritis: an RSA, clinical and radiographic study. Acta Orthop 2006; 77(1): 71-8. Valstar E R, Gill R, Ryd L, Flivik G, Börlin N, Valstar E R, Gill R, Ryd L, Flivik G, Börlin N, Kärrholm J. Guidelines for standardization of radiostereometry (RSA) of implants. Acta Orthop 2005; 76(4): 563-72. Weber E, Sundberg M, Flivik G. Design modifications of the uncemented Furlong hip stem result in minor early subsidence but do not affect further stability: a randomized controlled RSA study with 5-year follow-up. Acta Orthop 2014; 85(6): 556-61. Wierer T, Forst R, Mueller L A, Sesselmann S. Radiostereometric migration analysis of the Lubinus SP II hip stem: 59 hips followed for 2 years. Biomed Tech 2013; 58(4): 333-41.

Acta Orthopaedica 2020; 91 (6): 699–704

699

Automated classification of hip fractures using deep convolutional neural networks with orthopedic surgeon-level accuracy: ensemble decision-making with antero-posterior and lateral radiographs Yutoku YAMADA 1,2, Satoshi MAKI 1, Shunji KISHIDA 2, Haruki NAGAI 2, Junnosuke ARIMA 1,3, Nanako YAMAKAWA 2, Yasushi IIJIMA 2, Yuki SHIKO 4, Yohei KAWASAKI 4, Toshiaki KOTANI 2, Yasuhiro SHIGA 1, Kazuhide INAGE 1, Sumihisa ORITA 1,5, Yawara EGUCHI 1, Hiroshi TAKAHASHI 6, Takeshi YAMASHITA 3, Shohei MINAMI 2, and Seiji OHTORI 1 1 Department of Orthopaedic Surgery, Graduate School of Medicine, Chiba University, Japan; 2 Department of Orthopaedic Surgery, Seirei Sakura Citizen Hospital; 3 Department of Orthopaedic Surgery, Oyumino Central Hospital; 4 Biostatistics Section, Clinical Research Center, Chiba University Hospital; 5 Center for Frontier Medical Engineering, Chiba University; 6 Department of Orthopaedic Surgery, Toho University Sakura Medical Center, Japan

Correspondence: satoshimaki@gmail.com Submitted 2020-04-17. Accepted 2020-07-15.

Background and purpose — Deep-learning approaches based on convolutional neural networks (CNNs) are gaining interest in the medical imaging field. We evaluated the diagnostic performance of a CNN to discriminate femoral neck fractures, trochanteric fractures, and non-fracture using antero-posterior (AP) and lateral hip radiographs. Patients and methods — 1,703 plain hip AP radiographs and 1,220 plain hip lateral radiographs were included in the total dataset. 150 images each of the AP and lateral views were separated out and the remainder of the dataset was used for training. The CNN made the diagnosis based on: (1) AP radiographs alone, (2) lateral radiographs alone, or (3) both AP and lateral radiographs combined. The diagnostic performance of the CNN was measured by the accuracy, recall, precision, and F1 score. We further compared the CNN’s performance with that of orthopedic surgeons. Results — The average accuracy, recall, precision, and F1 score of the CNN based on both anteroposterior and lateral radiographs were 0.98, 0.98, 0.98, and 0.98, respectively. The accuracy of the CNN was comparable to, or statistically significantly better than, that of the orthopedic surgeons regardless of radiographic view used. In the CNN model, the accuracy of the diagnosis based on both views was significantly better than the lateral view alone and tended to be better than the AP view alone. Interpretation — The CNN exhibited comparable or superior performance to that of orthopedic surgeons to discriminate femoral neck fractures, trochanteric fractures, and non-fracture using both AP and lateral hip radiographs.

Although conventional radiography is the mainstay for diagnosing fractures, the sensitivity of radiographs to detect hip fracture is not ideal. Previous reports have indicated that the rate of initial misdiagnosis varies between 7% and 14% (Chellam 2016), and delayed diagnosis and treatment may lead to malunion, osteonecrosis, and arthritis, resulting in a poor prognosis (Parker 1992). Deep learning is a branch of machine learning that has recently yielded breakthroughs in computer vision tasks. A deep-learning approach based on convolutional neural networks (CNNs) is gaining interest across a variety of domains including medical imaging. CNNs are designed to automatically and adaptively learn features from data through backpropagation by using multiple building blocks, such as convolution layers, pooling layers, and fully connected layers (Greenspan et al. 2016). Owing to large datasets and increased computing power, CNNs have rapidly become a cutting-edge method for enhancing performance in medical image analysis. Recently, an increasing number of clinical applications have been reported in radiology for detection, classification, and segmentation tasks. However, studies using CNNs in the field of orthopedic surgery and traumatology are limited and the field is immature. So far, there are radiographic studies using CNNs for hip fractures (Adams et al. 2019, Badgeley et al. 2019, Cheng et al. 2019, Urakawa et al. 2019), distal radius fractures (Kim and MacKinnon 2018, Gan et al. 2019, Yahalomi et al. 2019, Blüthgen et al. 2020), proximal humeral fractures (Chung et al. 2018), ankle fractures (Kitamura et al. 2019) and hand, wrist, and ankle fractures (Olczak et al. 2017).

700

Acta Orthopaedica 2020; 91 (6): 699â&#x20AC;&#x201C;704

This study evaluates the diagnostic performance of a CNN for detecting and classifying hip fractures using plain anteroposterior (AP) and lateral hip radiographs. We compared the diagnostic performance with that of orthopedic surgeons.

Patients and methods Patients We retrospectively reviewed the medical records of all consecutive patients with hip fractures who were admitted to the Seirei Sakura Citizen Hospital between April 2015 and January 2020 and the Oyumino Central Hospital between March 2014 and January 2019. Diagnosis of the fracture type was made mainly using radiographs and computed tomography by at least 2 board-certified orthopedic surgeons. Particularly for the cases where the fracture pattern was not clear, we also used MRI to make the diagnosis. Basal neck fractures were classified as trochanteric fractures because they are recommended to be treated as an extra-capsular fracture using a sliding hip screw (Mallick and Parker 2004). Other fractures, including femoral head and subtrochanteric fractures, were not included in the study. There were 569 patients with femoral neck fracture and 466 patients with trochanteric fracture. The radiographs of non-fractured hips in the AP view were obtained from the AP radiographs of the hip contralateral to the fractured hip. The radiographs of non-fractured hips in the lateral view were obtained from patients with suspected hip fractures that were diagnosed as sprains or bruises of the hip joint. Radiographic dataset Poor-quality images, such as those with poor image contrast (16 AP images and 24 lateral images), anatomical side markers in the region (34 AP images and 5 lateral images), foreign body interference (8 AP images and 2 lateral images), and metal implants (105 AP images) were excluded. Moreover, 196 AP radiographs of non-fractured hips were chosen randomly and excluded to avoid the problem of imbalanced classes. The dataset used in this study included 1,703 plain hip AP radiographs (556 femoral neck fracture cases, 441 trochanteric fracture cases, and 706 normal hips) from 1,047 patients (801 women, 246 men; 567 from Seirei Sakura Citizen Hospital, 480 from Oyumino Central Hospital) and 1,220 plain hip lateral radiographs (555 femoral neck fracture cases, 431 trochanteric fracture cases, and 234 normal hips) from 1,220 patients (911 women, 309 men; 818 from Seirei Sakura Citizen Hospital, 402 from Oyumino Central Hospital). We used only 1 image per patient to decrease the overfitting of the CNN except for the cases of non-fracture AP radiographs. We reserved 50 of the AP and lateral radiographs from each set of femoral neck fractures, trochanteric fractures, and nonfractures for the validation dataset (i.e., 150 images each for AP and lateral views) and used the remainder of the dataset for training. The validation dataset was taken from the latest

Figure 1. Image preprocessing for the convolutional neural network model training and validation. We cropped images to a minimum region containing the femoral head and the greater and lesser trochanters in both the AP (A) and lateral (B) hip radiographs. On the AP radiographs, the fractured hip (left white box) was cropped and the side contralateral from the fractured hip (right white box) was cropped as the nonfractured hip. AP = anteroposterior.

patient admitted to the Seirei Sakura Citizen Hospital during the study period. Image preprocessing for deep learning Plain antero-posterior (AP) and lateral hip radiographs from digital imaging and communications in medicine (DICOM) files were exported in jpeg format from the picture archiving and communication systems (PACS) in our hospital. An orthopedic surgeon (YY, 3 years of experience) performed the image preprocessing using Paint 3D (Microsoft Corp, Redmond, WA, USA) by cropping the minimum region containing the femoral head and greater and lesser trochanters on both exported AP and lateral hip radiographs to generate an image for the CNN training (Figure 1). Model construction and training of the CNN Python programming language, version 3.6.7 (https://www. python.org) and Keras, version 2.2.4 with Tensorflow, version 1.14.0 (https://www.tensorflow.org) at the backend were used to construct the CNN architecture. We used the Xception architectural model, which had been previously trained using

Acta Orthopaedica 2020; 91 (6): 699–704

images with ImageNet. The Xception architecture has 36 convolutional layers forming the feature extraction base of the network. The 36 convolutional layers are structured into 14 modules, all of which have linear residual connections around them, except for the first and last modules. In short, the Xception architecture is a linear stack of depth-wise separable convolution layers with residual connections, which makes the architecture easy to define and modify (Chollet 2016). The input images were scaled down to 299 × 299 pixels. We then fine-tuned the model with the dataset of radiographs of femoral neck and trochanteric fractures, as well as non-fractures. Weights in the first 108 layers were frozen and weights in the other layers were retrained with our data. The network was trained for 100 epochs with a learning rate of 0.1, which was reduced if no improvement was seen. Convergence of the model training was monitored using cross-entropy loss. All images were randomly augmented using ImageDataGenerator (https://keras.io/preprocessing/image/) by a rotation angle range of 20°, width shift range of 0.2, height shift range of 0.2, brightness range of 0.3–1.0, and a horizontal flip of 50%. In addition, the models were separately constructed for AP and lateral radiographic views. The CNN was trained and validated using a computer with a GeForce RTX 2060 graphics processing unit (NVIDIA, Santa Clara, CA, USA), a Core i7-9750 central processing unit (Intel, Santa Clara, CA, USA), and 16 GB of random-access memory. Performance evaluation The performance of the CNN model for differentiating femoral neck fractures from trochanteric fractures and nonfracture was evaluated using the validation dataset, which was not included in the training dataset. The performance of the CNN was evaluated in 3 ways and a diagnosis was made for each: (1) AP hip radiographs alone; (2) lateral hip radiographs alone; (3) both AP and lateral hip radiographs. The probabilities for femoral neck fractures, trochanteric fractures, and fractures were determined for each view. The final decision was made based on the highest probability score between the 3 diagnoses. When diagnosing the fracture based on both views, the probability of the diagnosis on the AP and lateral view was averaged and final diagnosis was made based on the highest probability score. This enabled a comprehensive decision based on both the AP and lateral views and not based on a single view, which is similar to the way clinicians make a diagnosis from radiographs. Image assessment by orthopedic surgeons 2 board-certified orthopedic surgeons (SK and SM with 22 and 14 years of experience, respectively) and 2 resident orthopedic surgeons (NY and JA with 4 and 3 years of experience, respectively) reviewed the AP and lateral hip radiographs in jpeg format, which had an identical area to those used during the training of the CNN but with the same resolution as the original DICOM image. This was intended to provide fair

701

competition between the CNN and the clinicians, although this situation differs from the clinical setting. They reviewed the hip radiographs in the same 3 ways as the CNN and made a diagnosis for each method: (1) AP hip radiographs alone; (2) lateral hip radiographs; alone (3) both the AP and lateral hip radiographs together. The readers were blinded to clinical information such as the age of the patient and the mechanism of injury. Statistical and data analyses All statistical analyses were conducted using SAS (version 9.4 for Windows) and JMP (version 12.0.1; SAS Institute Inc., Cary, NC, USA). Continuous variables were evaluated using an analysis of variance (ANOVA), and categorical variables were evaluated using a chi-square test. A threshold of p < 0.05 was considered significant in two-sided tests of statistical inference. Inter-rater reliability for fractures was calculated using Cohen’s kappa between the orthopedic surgeons. We calculated the true positive (TP), true negative (TN), false positive (TP), and false negative (FN) rates based on the predictions of the CNN and orthopedic surgeons. To evaluate the performance, we calculated the average values of accuracy, recall, precision, and F1 score. The accuracy, recall, precision, and F1 score were calculated by the following numerical formula: Accuracy = TP + TN/TP + FP + FN + TN; Recall = TP/TP + FN; Precision = TP/TP + FP; F1score = 2 × Recall × Precision/Recall + Precision. The accuracy of the diagnostic performance of the CNN and the orthopedic surgeons was compared using a McNemar test. The accuracy of diagnostic performance difference between the radiographic views (i.e., AP, lateral, and 2 views) was also compared using a McNemar test. Ethics, funding, and potential conflicts of interest The study was approved by the Institutional Review Board of the 3 institutions involved and the requirement for consent was waived because of the retrospective analysis (IRB no. 3329). This work was supported by a research grant funded by the Japanese Orthopaedic Association and Inohana-Shougakukai Grants-in-Aid. The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this article.

Results Patient characteristics The non-fracture group consisted of patients who had normal lateral hip radiographs. The number of non-fracture images was relatively small because obtaining lateral radiographs of a non-fractured hip is difficult (Table 1). The non-fracture group was younger than the patients with femoral neck fractures and trochanteric fractures. There were 6 cases of femoral neck fracture and 5 of trochanteric fracture that could not

702

Acta Orthopaedica 2020; 91 (6): 699–704

Table 1. Baseline patient characteristics

Accuracy 1

Factor Age, mean (SD) Sex (M/F), n

Femoral Trochanteric neck fracture fracture n = 569 n = 466 81.3 (11.4) 136/433

85.2 (10.0) 105/361

Non-fracture n = 234 68.8 (16.2) 81/153

AP view Lateral view 2 views

0.9 0.8 0.7 0.6 0.5

Table 2. Accuracy, p-value of the accuracy compared with the CNN, average recall, precision, and F1 score of the diagnostic performance of the CNN and the 4 orthopedic surgeons based on both the anteroposterior and the lateral radiographs

0.4 0.3 0.2

CNN/ Average Average Average surgeon Accuracy (CI) p-value a recall precision F1 score CNN 0.98 (0.96–1.00) Board certified 1 0.92 (0.88–0.96) 2 0.95 (0.91–0.98) Resident 1 0.87 (0.82–0.93) 2 0.78 (0.71–0.85)

–

0.98

0.01 0.1

0.92 0.95

0.0006 < 0.0001

0.87 0.78

0.89 0.82

0.88 0.80

compared with CNN CI = 95% confidence interval; CNN = convolutional neural network.

be diagnosed by radiographs and CT and were subsequently diagnosed by MRI. Performance of the CNN compared with the orthopedic surgeons The accuracy of the CNN was comparable to or statistically significantly better than that of the orthopedic surgeons regardless of radiographic view used (Table 2 and Tables 3 and 4, see Supplementarys data). Improved recall and precision for the diagnosis of femoral neck fractures, trochanteric fractures, and non-fracture were found with the CNN model using 2 views

p = 0.004

p = 0.0002

p < 0.001

0.1 0

p = 0.001

CNN

p < 0.0001 p = 0.3

p = 0.4 p < 0.0001 p < 0.0001

Board Board certified 1 certified 2

p < 0.0001 p = 0.002 p = 0.03

Resident 1

Resident 2

Figure 2. Comparison of the accuracy between the AP, lateral, and both views of the CNN and the 4 orthopedic surgeons. In the CNN model, the accuracy of the diagnosis based on both views was statistically better than the AP view alone and the lateral view alone. The accuracy of diagnosis based on the AP view alone was statistically better than the lateral view alone. The same trend was also seen with the board-certified orthopedic surgeons. AP = anteroposterior; CNN = convolutional neural network.

compared with the AP or the lateral view alone (Table 5 and Tables 6 and 7, see Supplementary data). A comparison of the accuracy between the AP, lateral, and both views of the CNN and that of the orthopedic surgeons is shown in Figure 2. In the CNN model, the diagnostic accuracy based on both views was statistically better than that from the lateral view and the AP view. The interrater reliability of the orthopedic surgeons showed substantial to almost perfect agreement (Table 8). For reference, representative hip radiographs that were correctly diagnosed by the CNN but misdiagnosed by orthopedic surgeons and vice versa are presented in Figure 3.

Trochanteric fracture Recall Precision F1 score

Non-fracture Recall Precision F1 score

CNN 1.00 0.98 0.99 0.94 1.00 0.97 1.00 0.96 0.98 Board certified 1 0.96 0.89 0.92 0.94 0.92 0.93 0.86 0.96 0.91 2 0.96 0.96 0.96 0.96 0.96 0.96 0.92 0.92 0.92 Resident 1 0.96 0.81 0.88 1.00 0.86 0.93 0.66 1.00 0.80 2 0.90 0.71 0.80 0.96 0.77 0.86 0.48 0.96 0.64 CNN = convolutional neural network.

p < 0.0001 p = 0.001

Orthopedic surgeons

Table 5. Diagnostic performance of the CNN and the 4 orthopedic surgeons based on both the anteroposterior and the lateral radiographs CNN/ Femoral neck fracture surgeon Recall Precision F1 score

p = 0.3

Acta Orthopaedica 2020; 91 (6): 699–704

703

Table 8. Interrater reliability presented with Cohen’s kappa of the orthopedic surgeons Board certified Resident orthopedic surgeon orthopedic surgeon Surgeon 1 2 1 2 Board certified 1 2 Resident 1 2

– 0.85

0.85 –

0.78 0.76

0.73 0.78

0.78 0.73

0.76 0.78

– 0.66

0.66 –

Discussion We demonstrated that the ability of the CNN to differentiate femoral neck fractures from trochanteric fractures and nonfracture using AP and lateral hip radiographs was comparable or superior to that of orthopedic surgeons. The frontal radiograph was better for the extraction of the features for the CNN; however, adding the lateral radiographs to the decision process improved the recall and precision of the diagnosis. The promising performance of our CNN to discriminate femoral neck fractures, trochanteric fractures, and nonfracture with an accuracy of 98% was demonstrated. The CNN model was trained on 1,553 AP hip radiographs and 1,070 lateral hip radiographs and validated on 150 AP and lateral hip radiographs. There are several previous studies using a CNN to diagnose hip fractures. Urakawa et al. (2019) presented a CNN model that predicted trochanteric fractures with an accuracy of 95.6% and an AUC of 0.984 using 2,678 images for training and 334 images for validation. Adams et al. (2019) reported a CNN model to diagnose femoral neck fractures with an accuracy of 94.4% and an AUC of 0.98 using 640 images for training and 160 images for validation. Chen et al. (2019) described a CNN model that predicted both femoral neck and trochanteric fractures while achieving an accuracy of 0.959 and an AUC of 0.98 based on 3,605 training images and 100 images for validation. Badgeley et al. (2019) used not only images, but also patient data and hospital process features as input for their CNN model, which achieved an accuracy of 85% and an AUC of 0.91. Although their CNN model’s accuracy and AUC were limited to 74% and 0.78, respectively, it was based solely on images, and was trained on 17,587 radiographs and 5,970 images for validation. The model presented in this previous study discriminated only femoral neck fractures versus non-fractures, trochanteric fractures versus non-fractures, or hip fractures versus normal hips, and did not distinguish femoral neck fractures from trochanteric fractures. It is clinically important, however, to diagnose hip fractures, including both femoral neck fractures and trochanteric fractures, and to correctly differentiate between the 2 conditions for appropriate surgical management.

Figure 3. Representative radiographs of hip fractures. The AP (A) and lateral (B) radiographs of a trochanteric fracture, which the CNN misdiagnosed as a non-fracture, but all the orthopedic surgeon diagnosed correctly. The AP (C) and lateral (D) radiographs of a neck fracture, which 3 of the 4 orthopedic surgeons misdiagnosed as a non-fracture, but the CNN diagnosed correctly. The AP (E) and lateral (F) radiographs of a trochanteric fracture, which 3 of the 4 orthopedic surgeons misdiagnosed as a non-fracture or a neck fracture, but the CNN diagnosed correctly. AP = anteroposterior; CNN = convolutional neural network.

The study highlights the importance of using both AP and lateral radiographs even with a CNN model to diagnose hip fractures. Acquiring 2 radiographs in 2 perpendicular directions facilitates the assessment of the relative positions for the 2 pieces of fractured bone (Plaats 1969). Occasionally, fractures are visible in only a single view. If that view is not obtained, then the examination will be interpreted as falsely negative. Radiology departments in most hospitals follow protocols that call for orthogonal views in frontal and lateral projections for the suspected fracture of long bones. Ensemble decision-making using the CNN model trained on both the AP and lateral radiographs improved the recall and precision of diagnosing femoral neck fractures and the trochanteric fractures. Using both the AP and lateral radiographs for training was also thought to contribute to the reduction in the number of images required to achieve the same or better accuracy than

704

the previously reported model. However, there are few studies of deep learning in fracture detection using 2 or more radiographic views (Kitamura et al. 2019). It was easier to extract the fracture features from the AP radiograph for both the CNN and the orthopedic surgeons. This is why most of the previous studies achieved good accuracy using only AP radiographs for the training. The other reason to use only AP radiographs for training is because obtaining lateral radiographs of a nonfractured hip is difficult. In our study, we obtained lateral hip radiographs of non-fractured hips from patients who were suspected of having a hip fracture, but with a diagnosis of a sprain or bruise of the hip joint. There are several limitations to our study. First, the number of images included was relatively small. To overcome this problem, we used both AP and lateral radiographs for training and also applied transfer learning and data-augmentation methods. Also, the validation dataset was relatively small. Further investigation in a larger cohort is warranted to better train and validate the CNN for improved diagnostic accuracy and reproducibility of hip fractures and type. Second, the CNN discriminated only among radiographs of femoral neck fractures, trochanteric fractures, and non-fracture. It could not diagnose pubic rami fractures, which also occur when elderly patients fall, and can cause pain near the hip joint. Third, the radiographs need to be cropped before inputting to the model. However, it is not difficult for even non-orthopedic surgeons to crop an image around the hip joint. Constructing an object detection model will solve the second and third limitations, but object detection is more difficult than image classification, as it must identify the accurate localization of the object of interest (Feng et al. 2019). In conclusion, we have successfully demonstrated that the ability of the CNN to discriminate femoral neck fractures, trochanteric fractures, and non-fracture using both AP and lateral hip radiographs was comparable or superior to that of orthopedic surgeons. For the CNN, it was easier to extract the features of the fracture using the frontal radiograph; however, adding the lateral radiographs improved the recall and precision for diagnosing femoral neck fractures versus trochanteric fractures. Supplementary data Tables 3, 4, 6, and 7 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/ 17453674.2020.1803664 Conception and design of this study: YY and SM. Acquisition of the datasets: YY, SK, HN, YI, TK, and TY. Training and validation of models: YY and SM. Analysis and interpretation of the results and data: YY, SM, SK, NY, JA, YS, and YK. Drafting the manuscript: YY and SM. Critically reviewing the manuscript: SK, YS, KI, SO, YE, and HT. Supervision: SM and SO.

Acta Orthopaedica 2020; 91 (6): 699–704

The authors thank Ms. Yuri Ichikawa, Mr. Keisuke Masuda, and Ms. Megumi Abe for assisting with the acquisition of the datasets. Acta thanks Max Gordon and Örjan Smedby for help with peer review of this study.

Adams M, Chen W, Holcdorf D, McCusker M W, Howe P D L, Gaillard F. Computer vs human: deep learning versus perceptual training for the detection of neck of femur fractures. J Med Imaging Radiat Oncol 2019; 63(1): 27-32. Badgeley M A, Zech J R, Oakden-Rayner L, Glicksberg B S, Liu M, Gale W, McConnell M V, Percha B, Snyder T M, Dudley J T. Deep learning predicts hip fracture using confounding patient and healthcare variables. NPJ Digit Med 2019; 2: 31. Blüthgen C, Becker A S, Vittoria de Martini I, Meier A, Martini K, Frauenfelder T. Detection and localization of distal radius fractures: deep learning system versus radiologists. Eur J Radiol 2020; 126(February): 108925. Chellam W B. Missed subtle fractures on the trauma-meeting digital projector. Injury 2016; 47(3): 674-6. Cheng C, Ho T, Lee T, Chang C, Chou C, Chen C, Chung I-F, Liao C-H. Application of a deep learning algorithm for detection and visualization of hip fractures on plain pelvic radiographs. Eur Radiol 2019; 29(10): 5469-77. Chollet F. Xception: Deep learning with depthwise separable convolutions. Cornell University [pdf]; 2016. Chung S W, Han S S, Lee J W, Oh K, Kim N R, Yoon J P, Kim J Y, Moon S H, Kwon J, Lee H, Noh Y, Kim Y. Automated detection and classification of the proximal humerus fracture by using deep learning algorithm. Acta Orthop 2018; 89(4): 468-73. Feng X, Jiang Y, Yang X, Du M, Li X. Computer vision algorithms and hardware implementations: a survey. Integration 2019; 69(August): 309-20. Gan K, Xu D, Lin Y, Shen Y, Zhang T, Hu K, Zhou K, Bi M, Pan L, Wu W, Liu Y. Artificial intelligence detection of distal radius fractures: a comparison between the convolutional neural network and professional assessments. Acta Orthop 2019; 90(4): 394-400. Greenspan H, van Ginneken B, Summers R M. Guest Editorial. Deep learning in medical imaging: overview and future promise of an exciting new technique. IEEE Trans Med Imaging 2016; 35(5): 1153-9. Kim D H, MacKinnon T. Artificial intelligence in fracture detection: transfer learning from deep convolutional neural networks. Clin Radiol 2018; 73(5): 439-45. Kitamura G, Chung C Y, Moore B E. Ankle fracture detection utilizing a convolutional neural network ensemble implemented with a small sample, de novo training, and multiview incorporation. J Digit Imaging 2019; 32(4): 672-7. Mallick A, Parker M J. Basal fractures of the femoral neck: intra- or extracapsular. Injury 2004; 35(10): 989-93. Olczak J, Fahlberg N, Maki A, Razavian A S, Jilert A, Stark A, Sköldenberg O, Gordon M. Artificial intelligence for analyzing orthopedic trauma radiographs: deep learning algorithms—are they on par with humans for diagnosing fractures? Acta Orthop 2017; 88(6): 581-6. Parker M J. Missed hip fractures. Arch Emerg Med 1992; 9(1): 23-7. Plaats G J van der. Medical X-ray technique. Macmillan International Higher Education; June 18, 1969. Urakawa T, Tanaka Y, Goto S, Matsuzawa H, Watanabe K, Endo N. Detecting intertrochanteric hip fractures with orthopedist-level accuracy using a deep convolutional neural network. Skeletal Radiol 2019; 48(2): 239-44. Yahalomi E, Chernofsky M, Werman M. Detection of distal radius fractures trained by a small set of X-Ray images and Faster R-CNN. Adv Intell Syst Comput 2019; 997:971-81.

Acta Orthopaedica 2020; 91 (6): 705–710

705

A vitamin E blended highly cross-linked polyethylene acetabular cup results in less wear: 6-year results of a randomized controlled trial in 199 patients Julie R A MASSIER 1,2, Joost H J VAN ERP 1,2, Thom E SNIJDERS 1, and Arthur DE GAST 1,2 1 Clinical Orthopedic Research Center – mN, Zeist; 2 Department of Orthopedic Surgery, Diakonessenhuis, Utrecht, the Netherlands Correspondence: jverp@diakhuis.nl Submitted 2020-04-17. Accepted 2020-07-20.

Background and purpose — Survivorship of total hip arthroplasty (THA) with the ultra-high molecular weight polyethylene (UHMWPE) monoblock cup has been limited due to periprosthetic osteolysis and aseptic loosening, secondary to wear of the UHMWPE. In response, a vitamin E blended highly cross-linked polyethylene (HXLPE) cup was developed. This study set out to compare the wear and clinical 6-year outcomes of vitamin E blended HXLPE with UHMWPE in an isoelastic monoblock cup in patients with hip osteoarthritis who underwent uncemented THA. The 2-year results have been reported previously. Patients and methods — For this randomized controlled trial 199 patients were included. 102 patients received the vitamin E blended HXLPE uncemented acetabular cup and 97 patients the uncemented UHMWPE monoblock cup. Clinical and radiographic parameters were obtained preoperatively, directly postoperatively, and at 3, 12, 24, and 72 months. Wear rates were compared using the femoral head penetration (FHP) rate. Results — 173 patients (87%) completed the 6-year follow-up. The mean NRS scores for rest pain, load pain, and patient satisfaction were 0.3 (SD 1), 0.6 (SD 1), and 8.6 (SD 1) respectively. The mean Harris Hip Score was 93 (SD 12). The FHP rate was lower in the vitamin E blended HXLPE cup (0.028 mm/year) compared with the UHMWPE cup (0.035 mm/year) (p = 0.002). No adverse reactions associated with the clinical application of vitamin E blended HXLPE were observed. 15 complications occurred, equally distributed between the two cups. The 6-year survival to revision rate was 98% for both cups. There was no aseptic loosening. Interpretation — This study shows the superior performance of the HXLPE blended with vitamin E acetabular cup with clinical and radiographic results similar to the UHMWPE acetabular cup.

The use of HXLPE in monoblock cups used for total hip arthroplasty (THA) has been shown to reduce wear significantly (Muratoglu et al. 2001, Martell et al. 2003, Devane et al. 2017). However, the cross-linking of polyethylene forms free radicals, which leads to a decrease of long-term oxidative stability, causing embrittlement of the HXLPE (Sutula et al. 1995). In order to prevent this embrittlement, a new generation of HXLPE has been developed in which the polyethylene is stabilized with vitamin E. Vitamin E reacts with free radicals, protecting against oxidative degradation, avoiding long-term oxidation, embrittlement, and mechanical failure (Oral et al. 2004). Several studies have shown good in vitro and vivo results of the vitamin E stabilized HXLPE in terms of wear rates and mechanical properties (Oral et al. 2006, Halma et al. 2015, Nebergall et al. 2017, Scemama et al. 2017, Lambert et al. 2018, Snijders et al. 2019, van Erp et al. 2020). However, mid-term results of large randomized controlled trials (RCT) corroborating these findings have not been published. The mechanical characteristics of the monoblock acetabular cup potentially reduce polyethylene wear and acetabular osteolysis. Several studies have shown promising survival rates and reduced wear and osteolysis frequencies (Young et al. 2002, Gwynne-Jones et al. 2009, Pakvis et al. 2011). Survivorship of total hip arthroplasty (THA) with an ultrahigh molecular weight polyethylene (UHMWPE) monoblock cup has been limited due to periprosthetic osteolysis and aseptic loosening, secondary to wear of the UHMWPE (Harris 2001). The number of patients in need of a THA will increase in the future and these patients will have higher expectations regarding performance and durability. Therefore, highly cross-linked polyethylene (HXLPE) was introduced in the late 1990s (Kurtz et al. 2011). This RCT compares the wear rates expressed as femoral head penetration (FHP) between the uncemented vita-

706

Acta Orthopaedica 2020; 91 (6): 705–710

min E blended HXLPE acetabular cup and the conventional UHMWPE acetabular cup after 6 years. Secondary outcomes are the effect of an increased head size on the wear, clinical performance, and complication rates between the two cups.

Pre- and postoperative care were protocolled in order to ensure similar perioperative regimens. All patients received cefazolin prophylaxis during 24 hours perioperatively and thromboprophylaxis with low molecular weight heparin for 6 weeks postoperatively. Patients followed the same rehabilitation regimen, starting on the first day after surgery.

Patients and methods

Radiological assessment Radiological assessment was performed by 1 of the authors (JM) according to a standardized form; see van Erp et al. (2020).

Study design This RCT was carried out at the Diakonessenhuis Hospital Utrecht/Zeist, a medium-sized general hospital in the Netherlands. Between 2011 and 2014, 199 patients were included, and randomly allocated in 2 groups (van Erp et al. 2020). Description of patient inclusion and exclusion criteria, baseline characteristics, surgical procedures, and implants are described in the 2-year follow-up of this RCT by van Erp et al. (2020). After enrollment, baseline characteristics, Harris Hip Score (HHS), and Numeric Rating Scale (NRS) score for patient satisfaction, rest, and load pain were documented. After randomization, 102 patients received an uncemented vitamin E blended HXLPE cup (RM uncemented monoblock Pressfit Vitamys cup, Mathys Ltd, Bettlach, Switzerland) and 97 patients received a conventional UHMWPE cup (RM uncemented monoblock Pressfit, Mathys Ltd, Bettlach, Switzerland). Patients were scheduled for clinical and radiological followup on the first day postoperatively, and at 3, 12, 24, and 72 months. At each follow-up HHS and NRS scores, as well as complications, were documented and anteroposterior radiographs (AP) of the pelvis in supine position were taken. The primary outcome of this study was the linear FHP rate. Secondary outcomes were the clinical outcomes and the effect of increased head size. Follow-up was performed by 1 independent investigator (JE) who was blinded to the intervention. In order to reduce bias, the patient was blinded until completion of 6-year follow-up and the surgeon did not perform the follow-up measurements. Subjects were able to leave the study at any time or for any reason, without any consequences. The investigators were able to withdraw a subject from the study for urgent medical reasons. There was no replacement of subjects after withdrawal. For the sample size calculation see van Erp et al. (2020). Procedure Critical aspects of the surgical procedure were standardized for both groups. Surgeons required knowledge of the allocated treatment. Therefore, they were not blinded to the intervention, and were informed about the cup type during surgery. 7 orthopedic surgeons with vast experience in uncemented THA performed the procedures. Alumina ceramic femoral prosthetic heads (BIONIT2, Mathys Ltd, Bettlach, Switzerland) of 28, 32, or 36 mm were used and an uncemented hydroxyapatite coated stem (Twinsys, Mathys Ltd, Bettlach, Switzerland) was implanted.

Statistics Statistical analysis was performed using SPSS Statistics, version 23.0 (IBM Corp, Armonk, NY, USA). The distribution of the data was checked using the Kolmogorov–Smirnov test. A Pearson chi-square test was used to test for differences between groups in surgical approach, head size, and radiographic specifics (Brooker classification, radiographic lucency around the cup and stem, and osteolysis around the stem). A paired t-test was used to test for significant differences in both groups between the FHP rate in the first year and in the last 5 years to determine the steady state. The Mann–Whitney U-test was used to test for significant differences between groups for the FHP rate at 6 years. Ethics, registration, funding, data sharing, and potential conflicts of interest The procedures performed in this study, involving human participants, were in accordance with the ethical standards of the institutional and/or national research committee, with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standard and the CONSORT statement. All patients gave informed written consent. The protocol was approved by the local institutional review board and registered at the Central Commission Human-Related research (CCMO) Registry as HipVit trial (NL 32832.100.10, R-10.17D/HIPVIT 1). This study was funded by the Clinical Orthopedic Research Center – mN. Data are available from the corresponding author on reasonable request. The authors have no competing interests.

Results 177 patients (89%) completed 6-year follow-up. The mean time to final follow-up was 70 months (SD 9). Patient characteristics and surgery specifics are displayed for both groups (Tables 1 and 2). 22 patients (11%) were lost to follow-up (Figure 1). 6 patients died due to other diseases, 12 patients were unreachable, and 4 patients left the study because of severe comorbidity. 4 patients (2%) were excluded because they underwent revision surgery. In the vitamin E blended HXLPE group 2

Acta Orthopaedica 2020; 91 (6): 705–710

707

Table 1. Demographics. Values are n (%) unless otherwise specified

Table 2. Surgery specifics. Values are n (%)

Vitamin E Total blended HXLPE UHMWPE n = 199 n = 102 n = 97

Total n = 199

Age, mean (SD) 65 (5) 66 (5) Weight (kg), mean (SD) 79 (13) 78 (13) Female 141 (71) 77 (75) Diagnosis Primary osteoarthritis 191 (96) 98 (96) Secondary osteoarthritis 2 (1) 1 (1) Inflammatory arthritis 1 (1) Femoral head avascular necrosis 1 (1) Hip dysplasia 4 (2) 3 (3)

Approach Direct lateral 98 (49) 45 (44) 53 (55) Posterolateral 85 (43) 46 (45) 39 (40) Anterolateral 16 (8) 11 (11) 5 (5) Head size 28 mm 22 (11) 2 (2) 20 (21) 32 mm 112 (56) 35 (34) 77 (79) 36 mm 65 (33) 65 (64) NA Inclination < 35° 16 (8) 5 (5) 11 (11) 35–40° 47 (24) 20 (19) 27 (28) 41–45° 54 (27) 26 (26) 28 (29) 46–50° 51 (26) 31 (30) 20 (21) > 50° 31 (16) 20 (20) 11 (11)

65 (5) 79 (14) 64 (66) 93 (96) 1 (1) 1 (1) 1 (1) 1 (1)

Vitamin E blended HXLPE: vitamin E diffused highly cross-linked polyethylene. UHMWPE: ultra-high molecular weight polyethylene.

Vitamin E blended HXLPE n = 102

UHMWPE n = 97

For abbreviations, see Table 1. NA: Not avaiable

Table 3. Femoral head penetration rates in mm. Values are mean (SD) and [95% confidence intervals]

Randomized n = 199 Received UHMWPE cup n = 97

Excluded (n = 12): – revised, 2 – lost to FU, 10 72-month FU: – radiographs, 85 – clinical scores, 80

Figure 1. Summary of follow-up.

Received vitamin E blended HXLPE cup n = 102 Excluded (n = 14): – revised, 2 – lost to FU, 12 72-month FU: – radiographs, 88 – clinical scores, 85

Follow-up Vitamin E interval (year) blended HXLPE

UHMWPE

p-value

Penetration (mm) 0–1 0.24 (0.14) [0.21–0.27] 0.25 (0.13) [0.22–0.27] 0.9 0–2 0.27 (0.14) [0.25–0.30] 0.28 (0.12) [0.26–0.31] 0.8 0–6 0.38 (0.15) [0.35–0.41] 0.44 (0.16) [0.40–0.47] 0.01 1–6 0.14 (0.07) [0.13–0.16] 0.18 (0.08) [0.16–0.19] 0.002 2–6 0.07 (0.05) [0.06–0.08] 0.09 (0.06) [0.08–0.10] 0.05 Penetration rates (mm/year) at 2 years 0.046 (0.030) [0.040–0.052] 0.056 (0.038 [0.048–0.065] 0.05 6 years 0.028 (0.013) [0.026–0.031] 0.035 (0.016 [0.032–0.039] 0.002 For abbreviations, see Table 1.

patients underwent revision surgery: 1 because of infection and 1 for recurrent instability. In the UHMWPE group, 2 patients underwent revision surgery: 1 for cup malpositioning and 1 for recurrent instability. Furthermore, clinical data were missing for 5 patients in the UHMWPE and 3 in the vitamin E blended HXLPE groups. The head sizes used in the vitamin E blended HXLPE group were larger than the head sizes in the control group (p < 0.001). This is mainly because the 36-mm head size is not available for the UHMWPE cup. Bivariate analysis showed similar rates of heterotopic ossifications, radiographic lucencies around the cup and lucencies and osteolysis around the stem. Femoral head penetration The FHP rates were normally distributed, except for the FHP rate from 1 to 6 years. The total FHP after 6 years was 0.38 mm in the vitamin E blended HXLPE cup and 0.44 mm in the UHMWPE cup (p = 0.01) (Table 3). The mean FHP rate from

1 to 6 years, thus excluding the bedding-in time of the first year, was 0.028 and 0.035 mm/year for the vitamin E blended HXLPE cup and UHMWPE cup respectively (Figure 2). The mean FHP rate of the vitamin E blended HXLPE cup after 6 years was lower compared with the FHP rate of the UHMWPE cup (p = 0.002) (Table 3). FHP rates (in mm/year) for each head size in both groups are shown in Table 4 (see Supplementary data). No significant differences in wear between head sizes of the same cup type were found. A sub-analysis of the 28- and 32-mm head sizes was performed. No statistically significant difference in FHP rates between the UHMWPE (0.035 mm/year) and vitamin E blended HXLPE cup (0.031 mm/year) were found at 6-year follow-up (p = 0.2). There was a statistically significant difference between the FHP rate in the second year and in the last 5 years in both groups (vitamin E blended HXLPE cup; p < 0.001, UHMWPE cup; p < 0.001).

708

Acta Orthopaedica 2020; 91 (6): 705–710

Femoral head penetration (FHP; mm)

Score

0.7

10 Total FHP (mm) vitamin E blended HXLPE cup Total FHP (mm) UHMWPE FHP rate (mm/year) vitamin E blended HXLPE cup FHP rate (mm/year) UHMWPE

0.6

0.5

9 8

p = 0.01

7 HHS/10 (vitamin E blended HXLPE) NRS satisfaction (vitamin E blended HXLPE) NRS load pain (vitamin E blended HXLPE) NRS rest pain (vitamin E blended HXLPE) HHS/10 (UHMWPE) NRS satisfaction (UHMWPE) NRS load pain (UHMWPE) NRS rest pain (UHMWPE)

0.4

5 0.3

4 3

0.2

2 0.1

p = 0.002

0 3

Months follow up

Figure 2. Total femoral head penetration (FHP) in mm and mean FHP rate in mm/year.

Months follow up

Figure 3. Numeric Rating Scale and Harris Hip Score.

Table 6. Postoperative complications after 6-year follow-up Total Complication n = 199 Malposition cup Pulmonary embolism Deep venous thrombosis Neurological damage Superficial wound infection Deep infection Leg length discrepancy Dislocation Total

Vitamin E blended HXLPE UHMWPE n = 102 n = 97

1 1 – 1 1 – 1 1 – 2 – 2 2 1 1 1 1 – 1 1 – 6 2 4 15 (8%)

8 (8%)

7 (7%)

Clinical outcomes NRS and HHS after 6 years were similar for both groups (Figure 3, Table 5, see Supplementary data). The all-cause 6-year survival to revision was 98% for both groups. The 6-year survival rate for aseptic loosening was 100% in both groups. Complications No adverse reactions associated with the clinical application of vitamin E-blended HXLPE were observed during this study. In 2 patients with a vitamin E blended HXLPE cup, a perioperative fracture of the femur occurred. In 1 patient with a UHMWPE cup, a postoperative fracture of the acetabulum occurred. All patients were treated nonoperatively and recovered completely without further complications. Within 6-year follow-up 15 complications occurred, among which were 6 patients with dislocations (3%) (Table 6). In the vitamin E blended HXLPE group 2 dislocations occurred and in the UHMWPE 4 (p = 0.4). All patients recovered completely, except for 2 patients who still had mild paresthesia caused by neuropraxia of the sciatic nerve.

Discussion This large, mid-term RCT compares 2 similar acetabular monoblock cups: a vitamin E blended HXLPE and a UHMWPE cup. The vitamin E blended HXLPE cup showed a statistically significant different 2-dimensional linear FHP rate of 0.028 mm/year compared with the UHMWPE cup (0.035 mm/year) (p = 0.002). Good mid-term clinical and radiological outcomes of both cups are evident. No adverse reactions or abnormal mechanical behavior concerning the vitamin E additive were observed. No signs of osteolysis or aseptic loosening were observed in either group. Many studies have shown a substantial amount of the FHP occurring within the first year after surgery, due to bedding-in or creep of the polyethylene (Sychterz et al. 1999, McCalden et al. 2005). Therefore, the first year of linear FHP was attributed to creep (Rochcongar et al. 2018, Snijders et al. 2019). The total FHP in this study was 0.38 mm in the vitamin E blended HXLPE cup and 0.44 mm in the UHMWPE cup (p = 0.01). This is well within the range of results reported in current literature (Dorr et al. 2005, Lambert et al. 2018, Snijders et al. 2019). After bedding-in, steady-state penetration rate is reached. This study showed a statistically significant difference between the FHP rate in the first and second year of follow-up in both groups (van Erp et al. 2020). However, there was no significant difference between the FHP rate in the second year and last 4 years in either group, suggesting steady-state penetration with a constant wear rate has been reached (Figure 2). Several studies have shown superior performance of vitamin E addition to HXLPE for protection from oxidation, preserving mechanical properties, reduced wear debris, and bacterial adhesion compared withthe UHMWPE cups. This suggests that adding vitamin E may prevent osteolysis, implant loos-

Acta Orthopaedica 2020; 91 (6): 705–710

ening, and eventually revision surgery (Scemama et al. 2017, Yamamoto et al. 2017, Galea et al. 2018, Rochcongar et al. 2018). Shareghi et al. (2017) found in 37 vitamin E-infused HXLPE cups an FHP rate of 0.04 mm/year and in 26 HXLPE cups an FHP rate of 0.07 mm/year between 2 and 5 years, using radiostereometric analysis (RSA). These slightly higher wear rates could be explained by the shorter period of 5 years’ follow-up and a different measuring technique. A small RCT comparing the same cups in 62 patients found a statistically significant difference (p < 0.001) between the steady-state FHP rate in a vitamin E blended HXLPE cup (0.020 mm/ year) and a UHMWPE cup (0.058 mm/year) after 3 years of follow-up, using radiostereometric analysis (RSA) (Rochcongar et al. 2018). These results are comparable to our findings, only with a smaller patient cohort and shorter follow-up time. Furthermore, Dumbleton et al. (2002) showed osteolysis and increased risk of revision surgery due to loosening are rare below a linear wear rate of 0.1 mm/year. Our study demonstrated no osteolysis or revisions for aseptic loosening in either group after 6 years of follow-up. This is the first blinded RCT, using a reliable measuring technique, with substantially more patients and a similar control group to compare the effect of vitamin E on wear rates of 2 similar cups. A high follow-up rate (89%) has been reached. Our results are with a similar order of magnitude in FHP rates compared with current literature (Halma et al. 2015, Snijders et al. 2019). This study has several limitations. First, patients in the vitamin E blended HXLPE group received significantly larger head sizes; however, multivariate analysis showed no statistically significant effect of head size on the wear rates in either cup. Besides, in the literature, the effect of a greater head size on the polyethylene wear rates is disputed (Lachiewicz et al. 2009, 2016, van Erp et al. 2020). Furthermore, due to the availability of the vitamin E blended HXLPE cup on the Dutch market, surgeons only have a choice of larger head sizes in most similar cup sizes. In its manufacturing process thinner polyethylene liners are applied because less wear is expected. The possibility to employ larger heads in the same patient due to a thinner polyethylene liner is an advantage of the RM Pressfit vitamys cup (vitamin E blended HXLPE cup) over the UHMWPE cup, since it results in lower risk of dislocation (Knahr 2012). In our study we saw fewer dislocations in the vitamin E blended HXLPE group compared with the UHMWPE group (2 versus 4). Second, despite our follow-up rate of 89%, the loss of 11% is a limitation as the unreachable patients could have been admitted to other hospitals. This should be taken into account while reviewing the results. Third, only a linear, 2-dimensional penetration measurement was performed, using the View Pro-X software instead of the 3D RSA measuring method (Callary et al. 2015). RSA requires the use of tantalum beads placed into the implants and bone. This limits the use of RSA (Lawrie et al. 2003). The View Pro-X software was found to be a reliable method in

709

clinical practice for the assessment of linear FHP to determine the polyethylene wear, with good inter- and intra-class reliability (Martell et al. 2003, Geerdink et al. 2008). Using this software allows a non-invasive, precise measuring method for large patient cohorts. In conclusion, this large RCT shows the excellent and superior mid-term performance of the vitamin E blended HXLPE cup, with significantly less wear in terms of FHP rates compared with the UHMWPE cup. There was no difference between the two groups in terms of clinical outcomes, revision rates, and complications. Since differences in wear rates are few, it is hard to say whether clinical differences will be found in future follow-up. Further long-term follow-up is needed to assess whether these lower wear rates in vitamin E blended HXLPE cups result in less osteolysis and aseptic loosening, compared with conventional UHMWPE. Supplementary data Tables 4 and 5 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674. 2020.1807220 Study design: AG, JE, TS. Patient inclusion: AG, TS. Follow-up of patients: JE. Measurements: JM. Analysis: JM, JE. Drawing of the manuscript: JM, JE. Critical review of the manuscript: JM, JE, TS, AG. Acta thanks R. Itayem and Michael Wyat for help with peer review of this study.

Callary S A, Solomon L B, Holubowycz O T, Campbell D G, Munn Z, Howie D W. Wear of highly crosslinked polyethylene acetabular components. Acta Orthop 2015; 86(2): 159-68. Devane P A, Horne J G, Ashmore A, Mutimer J, Kim W, Stanley J. Highly cross-linked polyethylene reduces wear and revision rates in total hip arthroplasty. J Bone Joint Surg 2017; 99(20): 1703-14. Dorr L D, Wan Z, Shahrdar C, Sirianni L, Boutary M, Yun A. Clinical performance of a Durasul highly cross-linked polyethylene acetabular liner for total hip arthroplasty at five years. J Bone Joint Surg 2005; 87(8): 1816. Dumbleton J H, Manley M T, Edidin A A. A literature review of the association between wear rate and osteolysis in total hip arthroplasty. J Arthroplasty 2002; 17(5): 649-61. Galea V P, Connelly J W, Shareghi B, Kärrholm J, Sköldenberg O, Salemyr M, Laursen M B, Muratoglu O, Bragdon C, Malchau H. Evaluation of in vivo wear of vitamin E-diffused highly crosslinked polyethylene at five years. Bone Joint J 2018; 100-B(12): 1592-9. Geerdink C H, Grimm B, Vencken W, Heyligers I C, Tonino A. The determination of linear and angular penetration of the femoral head into the acetabular component as an assessment of wear in total hip replacement: a comparison of four computer-assisted methods. J Bone Joint Surg Br 2008; 90(7): 839-46. Gwynne-Jones D P, Garneti N, Wainwright C, Matheson J A, King R. The Morscher press fit acetabular component: a nine- to 13-year review. J Bone Joint Surg Br 2009; 91(7): 859-64. Halma J J, Eshuis R, Vogely H C, van Gaalen S M, de Gast A. An uncemented iso-elastic monoblock acetabular component: preliminary results. J Arthroplasty 2015; 30(4): 615-21. Harris W H. Wear and periprosthetic osteolysis: the problem. Clin Orthop Relat Res 2001; (393): 66-70.

710

Knahr K, editor. Total hip arthroplasty: wear behaviour of different articulations. Heidelberg/Vienna: Springer; 2012. Kurtz S M, Gawel H A, Patel J D. History and systematic review of wear and osteolysis outcomes for first-generation highly crosslinked polyethylene. Clin Orthop Relat Res 2011; 469(8): 2262-77. Lachiewicz P F, Heckman D S, Soileau E S, Mangla J, Martell J M. Femoral head size and wear of highly cross-linked polyethylene at 5 to 8 years. Clin Orthop Relat Res 2009; 467(12): 3290-6. Lachiewicz P F, Soileau E S, Martell J M. Wear and osteolysis of highly crosslinked polyethylene at 10 to 14 years: the effect of femoral head size. Clin Orthop Relat Res 2016; 474(2): 365-71. Lambert B, Neut D, van der Veen H C, Bulstra S K. Effects of vitamin E incorporation in polyethylene on oxidative degradation, wear rates, immune response, and infections in total joint arthroplasty: a review of the current literature. Int Orthop 2018; 43(7): 1549-57. Lawrie D, Downing M, Patrick Ashcroft G, Gibson P. Insertion of tantalum beads in RSA of the hip: variations in incidence of extra-osseous beads with insertion site. Acta Orthop 2003; 74(4): 404-7. Martell J M, Berkson E, Berger R, Jacobs J. Comparison of two and threedimensional computerized polyethylene wear analysis after total hip arthroplasty. J Bone Joint Surgery Am 2003; 85(6): 1111-17. McCalden R W, Naudie D D, Yuan X, Bourne R B. Radiographic methods for the assessment of polyethylene wear after total hip arthroplasty. J Bone Joint Surg Am 2005; 87(10): 2323-34. Muratoglu O K, Bragdon C R, O’Connor D, Perinchief R S, Estok D M, Jasty M, Harris W H. Larger diameter femoral heads used in conjunction with a highly cross-linked ultra-high molecular weight polyethylene: a new concept. J Arthroplasty 2001; 16(8 Suppl. 1): 24-30. Nebergall A, Greene M E, Laursen M B, Nielsen P T, Malchau H, Troelsen A. Vitamin E diffused highly cross-linked polyethylene in total hip arthroplasty at five years: a randomised controlled trial using radiostereometric analysis. Bone Joint J 2017; 99-B(5): 577-84. Oral E, Wannomae K K, Hawkins N, Harris W H W H, Muratoglu O K O K. Alpha-tocopherol-doped irradiated UHMWPE for high fatigue resistance and low wear. Biomaterials 2004; 25(24): 5515-22. Oral E, Christensen S D, Malhi A S, Wannomae K K, Muratoglu O K. Wear resistance and mechanical properties of highly cross-linked, ultrahigh-

Acta Orthopaedica 2020; 91 (6): 705–710

molecular weight polyethylene doped with vitamin E. J Arthroplasty 2006; 21(4): 580-91. Pakvis D, Biemond L, Van Hellemondt G, Spruit M. A cementless elastic monoblock socket in young patients: a ten to 18-year clinical and radiological follow-up. Int Orthop 2011; 35(10): 1445-51. Rochcongar G, Buia G, Bourroux E, Dunet J, Chapus V, Hulet C. Creep and wear in vitamin e-infused highly cross-linked polyethylene cups for total hip arthroplasty. J Bone Joint Surg Am 2018; 100(2): 107-14. Scemama C, Anract P, Dumaine V, Babinet A, Courpied J P, Hamadouche M. Does vitamin E-blended polyethylene reduce wear in primary total hip arthroplasty: a blinded randomised clinical trial. Int Orthop 2017; 41(6): 1113-18. Shareghi B, Johanson P-E, Kärrholm J. Wear of vitamin E-infused highly cross-linked polyethylene at five years. J Bone Joint Surg Am 2017; 99(17): 1447-52. Snijders T E, Halma J J, Massier J R A, van Gaalen S M, de Gast A. The survivorship of the uncemented iso-elastic monoblock acetabular component at a mean of 6-year follow-up. HSS J 2019; 16(1): 15-22. Sutula L C, Collier J P, Saum K A, Currier B H, Currier J H, Sanford W M, Mayor M B, Wooding R E, Sperling D K, Williams I R. The Otto Aufranc Award: Impact of gamma sterilization on clinical performance of polyethylene in the hip. Clin Orthop Relat Res 1995; (319): 28-40. Sychterz C J, Engh C A J, Yang A, Engh C A. Analysis of temporal wear patterns of porous-coated acetabular components: distinguishing between true wear and so-called bedding-in. J Bone Joint Surg Am 1999; 81(6): 821-30. van Erp J H J, Massier J R A, Halma J J, Snijders T E, de Gast A. 2-year results of an RCT of 2 uncemented isoelastic monoblock acetabular components: lower wear rate with vitamin E blended highly cross-linked polyethylene compared to ultra-high molecular weight polyethylene. Acta Orthop 2020; 91(3): 254-9. Yamamoto K, Tateiwa T, Takahashi Y. vitamin E-stabilized highly crosslinked polyethylenes: the role and effectiveness in total hip arthroplasty. J Orthop Sci 2017; 22(3): 384-90. Young A M, Sychterz C J, Hopper R H, Engh C A. Effect of acetabular modularity on polyethylene wear and osteolysis in total hip arthroplasty. J Bone Joint Surg Am 2002; 84(1): 58-63.

Acta Orthopaedica 2020; 91 (6): 711–716

711

No association between blood count levels and whole-blood cobalt and chromium levels in 1,900 patients with metal-on-metal hip arthroplasty Noora HONKASAARI 1,2, Olli LAINIALA 1,2, Outi LAINE 2,3, Aleksi REITO 1,2, and Antti ESKELINEN 1,2 1 Coxa

Hospital for Joint Replacement, Tampere; 2 University of Tampere, Faculty of Medicine and Life Sciences, Tampere; 3 Tampere University Hospital, Department of Internal Medicine, Tampere, Finland Correspondence: olli.lainiala@tuni.fi Submitted 2019-12-01. Accepted 2020-08-03.

Background and purpose — The accelerated wear of poorly functioning metal-on-metal (MoM) hip implants may cause elevated whole-blood cobalt (Co) and chromium (Cr) levels. Hematological and endocrinological changes have been described as the most sensitive adverse effects due to Co exposure. We studied whether there is an association between whole-blood Co/Cr levels and leukocyte, hemoglobin, or platelet levels. Patients and methods — We analyzed whole-blood Co and Cr values and complete blood counts (including leukocytes, hemoglobin, platelets) from 1,900 patients with MoM hips. The mean age at the time of whole-blood metal ion measurements was 67 years (SD 10). The mean time from primary surgery to whole-blood metal ion measurement was 8.2 years (SD 3.0). The mean interval between postoperative blood counts and metal ion measurements was 0.2 months (SD 2.7). Results — The median Co value was 1.9 µg/L (0.2–225), Cr 1.6 µg/L (0.2–125), mean leukocyte count 6.7 × 109/L (SD 1.9), hemoglobin value 143 g/L (SD 13), and platelet count 277 × 109/L (SD 70). We did not observe clinically significant correlations between whole-blood Co/Cr and leukocyte, hemoglobin, or platelet counts. Interpretation — Elevated whole-blood Co and Cr values are unlikely to explain abnormal blood counts in patients with MoM hips and the reason for possible abnormal blood counts should be sought elsewhere.

The abnormal wear of poorly functioning MoM implants may cause elevated whole-blood cobalt (Co) and chromium (Cr) levels (Brodner et al. 2003, Cheung et al. 2016). Soft-tissue reactions termed “pseudotumors” related to poorly functioning MoM hip replacements have been widely described (Boardman et al. 2006, Gruber et al. 2007, Pandit et al. 2008). The use of MoM implants has dramatically decreased but due to their previous popularity there are still a large number of patients with MoM hip replacements under follow-up (Silverman et al. 2016). Even though local reactions have been the most discussed, systemic reactions in patients with high-wearing hip implants have been described. Cardiomyopathy, polyneuropathy, hypothyreosis, and polycythemia have been described in some patients with MoM hip implants and in patients with fractured ceramic-on-ceramic implant revised to metal-on-polyethylene, resulting in abrasive wear of the CoCr head by ceramic fragments (Cheung et al. 2016). Systemic adverse events have been linked mostly to Co, and hematological and thyroid effects have been described as the most sensitive responses to Co in humans (Tvermoes et al. 2014). A case report described polycythemia with hemoglobin 190 g/L due to massive abrasive CoCr head wear when a ceramic-on-ceramic implant had been revised to metal-on-polyethylene after fracture of the ceramic liner (Gilbert et al. 2013). Several studies have suggested that blood lymphocyte counts may be affected by implant metals from MoM hip replacements (Hart et al. 2009, Hailer et al. 2011, Penny et al. 2013, Chen et al. 2014, Briggs et al. 2015, Markel et al. 2018). Although thrombocytopenia has not been linked to implant metals, it has been reported that platelets adhere to and are activated by CoCr (Ollivier et al. 2017).

712

Complete blood count including leukocyte count, hemoglobin, and platelets is among the most used blood tests in the world (Horton et al. 2019) and 10–20% of the measurements include abnormal values (Tefferi et al. 2005). Due to wide media attention to MoM hip replacements, patients with MoM hips are sometimes worried whether their abnormal laboratory findings are related to their hip replacement. We sought to find out whether whole-blood metal ion levels are associated with blood count. Our hypothesis was that if Co or Cr affected leukocytes, hemoglobin, or platelets at concentrations noted in our study group, we would observe an upward or downward trend (depending on variable) when blood Co or Cr concentrations are approaching the highest values.

Patients and methods This is a retrospective study. In September 2010 the United Kingdom Medicines and Healthcare Products Regulatory Agency released a medical device alert concerning MoM hip implants (MHRA 2010) and DePuy Orthopaedics recalled their ASR MoM hip replacement systems (DePuy Orthopaedics 2010). After the recall our institution started the systematic screening of all patients with Articular Surface Replacement (ASR; DePuy Orthopaedics, Warsaw, IN, USA) resurfacing or total hip arthroplasty (THA) hips. Screening included whole blood Co and Cr measurements, imaging, and clinical examination. In 2012 the screening program was expanded to include all our patients with any MoM hip implant brand, but implants other than ASRs were not systematically imaged. For this study, we included all our patients with Co and Cr measurements and blood count drawn within an arbitrary 6-month interval. Between November 1999 and February 2012, 3,013 MoM hips were implanted in 2,520 patients at our institution. Of these patients, whole blood Co and Cr measurements were available for 2,260 patients. Blood Co and Cr samples The blood samples for the Co and Cr measurements were drawn from the antecubital vein. The first 10 mL was discarded to avoid contamination by the metal ions released from the needle. The samples were sent to an independent laboratory for analysis. The measurements were performed with dynamic reaction cell inductively coupled plasma mass spectrometry (Agilent 7500 cx or 8800, Agilent Technologies, Santa Clara, CA, USA) between October 2009 and March 2018. Blood counts For the complete blood counts, 3 mL of blood was drawn into an EDTA tube. Leukocyte count, hemoglobin, hematocrit, erythrocyte count, mean corpuscular volume, and mean hemoglobin concentration as well as platelet count were then analyzed. Reference values for blood count values provided

Acta Orthopaedica 2020; 91 (6): 711–716

by our hospital laboratory were used. The complete blood counts were measured between October 2009 and June 2018. We used the complete blood count closest to the latest Co and Cr values. As the data was not prospectively collected and the whole blood Co and Cr measurements were not performed on the same day as the blood count, we chose 6 months as an arbitrary cut-off to exclude patients with long interval between the measurements. Of the 2,260 patients with whole-blood metal ion levels available, 1,900 had their blood count measured less than 6 months before or after the whole-blood metal ion measurement and were therefore included in the study. 572 of the patients went through a revision and for them last pre-revision measurements were used. For 314 patients there was no complete blood count available within 6 months of the latest whole-blood Co and Cr values, and thus an older metal ion measurement fulfilling the condition of blood count measured within a 6-month interval was used. For 258 patients leukocyte differentials were available and they were analyzed as a subgroup. Statistics As the distributions of whole-blood Co and Cr values are skewed, median values (ranges) are reported. Blood hemoglobin value, leukocyte, and platelet counts are normally distributed, so mean values (SD) are used. Spearman’s correlation coefficients were calculated to test whether there is correlation between whole-blood metal ion levels and either hemoglobin, leukocyte, or platelet counts. For Spearman’s correlation coefficient, we calculated 95% confidence intervals (CI). We chose a correlation coefficient of 0.225 (or –0.225) as the limit for clinically significant, equaling roughly 5% of variance for a dependent variable explained by an independent variable. If the observed CIs for correlation coefficients did not include 0.225 the correlation coefficient was considered as clinically not significant. We also calculated the change in blood count components for those 1,873 (99%) patients with preoperative blood counts available. The analyses were performed for all patients regardless of the implant brand and manufacturer or whether they had unilateral or bilateral MoM hip implants even though the blood concentrations are different among them, because, when considering the systemic effects of Co and Cr, it is irrelevant whether the Co and Cr in the blood are from 1 or 2 hips and which implant brand is used. Ethics, funding, and potential conflicts of interest Permission for the study was obtained from the institutional review board (permission number R17524S). This work was supported by the competitive research funds of Pirkanmaa Hospital District, Tampere, Finland, representing governmental funding. No conflicts of interest declared except for AE, who is paid lecturer for Zimmer Biomet, and receives institutional research support (not related to current study) from DePuy Synthes and Zimmer Biomet.

Acta Orthopaedica 2020; 91 (6): 711–716

713

Table 1. Metal-on-metal hip implant brands of the implant included in this study Hip resurfacing implants unilateral bilateral n = 536 n = 156 ASR 198 BHR 194 Durom 92 Other 52

Total hip implants unilateral THA bilateral THA n = 998 n = 189

Total hip and resurfacing implants n = 21

BHR/BHR 52 ASR 278 ASR/ASR 52 BHR/BHR 6 ASR/ASR 52 Pinnacle 259 Pinnacle/Pinnacle 50 ReCap/BHR 3 Durom/Durom 9 ReCap 157 BHR/BHR 11 ASR/ASR 2 Other 43 Other 304 Other 76 Other 10

ASR = Articular Surface Replacement (Depuy Orthopaedics, Warsaw, IN, USA); BHR = Birmingham Hip Resurfacings (Smith & Nephew, Memphis, TN, USA); Durom (Zimmer, Warsaw, IN, USA); Pinnacle (Depuy), ReCap (Biomet, Warsaw, IN, USA) .

Table 2. Spearman’s correlation coefficients (ρ) and 95% confidence intervals among whole-blood cobalt, chromium levels, hemoglobin concentration, and leukocyte and platelet counts Factor Leukocytes Hemoglobin Platelets ΔLeukocytes ΔHemoglobin ΔPlatelets

Cobalt Chromium

–0.05 (–0.09 to 0.000) –0.09 (–0.13 to –0.05) 0.06 (0.02 to 0.11) –0.05 (–0.10 to –0.01) 0.02 (–0.02 to 0.07) 0.02 (–0.03 to 0.07)

–0.08 (–0.12 to –0.03) –0.12 (–0.16 to –0.07) 0.13 (0.09 to 0.18) –0.05 (–0.09 to –0.002) 0.05 (0.007 to 0.10) 0.05 (0.003 to 0.09)

Δ = change from preoperative to last postoperative value.

Results Of the 1,900 patients included in the study, 81% had unilateral and 19% had bilateral MoM replacement (Table 1). 58% of the patients were male. The mean age at the time of whole-blood metal ion measurement was 67 years (SD 10). The mean time between the primary surgery and metal ion measurements was 8.2 years (SD 3.0). The mean time interval between blood metal ion measurements and postoperative blood counts was 0.2 months (SD 2.7). The median value for whole-blood Co concentration was 1.9 µg/L (0.2–225) and for Cr 1.6 µg/L (0.2–125). Mean postoperative leukocyte count was 6.7 × 109/L (SD 1.9), hemoglobin 143 g/L (SD 13), and platelet count 277 × 109/L (SD 70). Mean preoperative leukocyte count was 6.8 × 109/L (SD 2.5), hemoglobin 140 g/L (SD 14), and platelet count 245 × 109/L (SD 71). The mean time between preoperative and last postoperative blood count was 8.3 years (SD 3.0). The mean change in leukocyte (Δleukocytes) count was 0.1 × 109/L (SD 2.4), in hemoglobin (Δhemoglobin) –3.0 g/L (SD 13), and in platelet count (Δplatelets) –32 × 109/L (SD 61). Spearman’s correlation coefficients (ρ) between wholeblood Co/Cr and complete blood count parameters are presented in Table 2 (for scatter plots in Figure 1, see Supplementary data). None of the 95% CIs for ρ included predefined 0.225

Table 3. Spearman’s correlation coefficients (ρ) and 95% confidence intervals among whole-blood cobalt, chromium levels, and leukocyte differentials Factor Lymphocytes Neutrophils Eosinophils Basophils Monocytes

Cobalt Chromium

–0.09 (–0.21 to 0.03) –0.05 (–0.17 to 0.07) 0.11 (–0.01 to 0.23) a –0.02 (–0.14 to 0.11) –0.06 (–0.18 to 0.06)

–0.09 (–0.21 to 0.03) –0.07 (–0.19 to 0.06) 0.05 (–0.07 to 0.17) 0.003 (–0.12 to 0.13) –0.08 (–0.20 to 0.42) a

a 95%

confidence interval of the Spearman correlation coefficient includes value of 0.225, indicating that 5% variance for blood eosinphils explained by whole blood cobalt concentration, and 5% variance for blood monocytes explained by whole blood chromium concentrations cannot be ruled out by this analysis.

or –0.225 (equaling roughly 5% of variance for the dependent variable explained by the independent variable), indicating that clinically significant associations can be excluded. Leukocyte differentials including lymphocytes, neutrophils, eosinophils, basophils, and monocytes were available for 258 patients. This subgroup included 53% men. The mean age at the time of blood metal ion measurement was 67 years (SD 10), mean time from primary surgery to metal ion measurement was 7.8 years (SD 2.7), and interval between metal ion and leukocyte differentials 0.9 months (SD 3.0). The Spearman’s correlation coefficients (ρ) between whole-blood Co and blood eosinophil count (0.11, CI –0.01 to 0.23) and between Cr and monocytes count (–0.08, CI –0.20 to 0.42) were the only ρ between Co/Cr and leukocyte differential counts in which 0.225 was not excluded from 95% CI (Table 3, Figure 2, see Supplementary data). For all other analyses, clinically significant associations were excluded.

Discussion As many factors affect the blood leukocyte, hemoglobin, and platelet values, for this study we chose a rather low ρ of 0.225/–0.225 to represent clinically significant correlation. This ρ equals roughly 5% of variance for a dependent vari-

714

able explained by an independent variable. Despite choosing a low ρ-value to represent clinical significance, none of the CIs for correlation coefficients among whole-blood Co/Cr and leukocytes, hemoglobin, or platelet values included that value, excluding a clinically significant association. Therefore, the exposure to elevated whole-blood Co and Cr levels seems unlikely to explain clinically noticeable or relevant changes in complete blood counts of patients with MoM hip and blood Co/Cr levels comparable to those in this study. Consequently, the cause for abnormal complete blood counts should be primarily sought elsewhere. Of the leukocyte differential counts, only the CIs for ρ between whole-blood Co and blood eosinophils and between blood Cr and monocytes included 0.225. For other analyses, clinically significant association was excluded. There are some limitations in our study. 1st, we examined only the association between whole-blood metal ion levels and patients’ blood count, the study setting limits us from evaluating causality. 2nd, we defined clinically significant correlation as correlation coefficient of 0.225 or higher. To our knowledge, there are no predefined limits for clinical significance of correlation between blood metal ions and blood counts reported in the literature. ρ of 0.225 equals roughly 5% of variance for a dependent variable explained by an independent variable. We consider 5% a rather low value, but as blood counts are affected by countless other factors we did not want to choose too high a limit. 3rd, we had no information on the patients’ comorbidities or treatments, e.g., hematologic malignancies, HIV, chemotherapy, or radiation therapy that could have affected their blood counts. Further, as this was a retrospective study the whole-blood metal ion levels and blood counts were not measured on the same date. Hematological abnormalities caused by Co are likely to be reversible (Leyssens et al. 2017) and may even normalize if exposure is removed. However, the changes in whole-blood Co and Cr levels have been mostly less than 5 µg/L at 1-year follow-up with only a few exceptions, even in our cohort of high-risk ASR implants (Reito et al. 2014, 2016). All measurements included were drawn before revision surgery (if performed), as revision may result in a radical decrease of Co and Cr levels. Furthermore, we measured whole-blood levels, which have been described as more stable compared with serum and therefore offer a better indicator of long-term cobalt exposure (Paustenbach et al. 2014). Even though measurements within a 6-month time interval were included, an SD of 2.7 months indicates that more than two-thirds of measurements were within 3 months. We believe that the interval between measurements does not prevent us from drawing conclusions. A variety of symptoms possibly relating to Co and Cr have been described at case report level in patients with hip arthroplasties (Leyssens et al. 2017). The most devastating wear and systemic adverse events leading to death have been described resulting from ceramic fragments of a fractured ceramic implant causing abrasive third-body wear to a metal

Acta Orthopaedica 2020; 91 (6): 711–716

head implanted in revision (Gilbert et al. 2013, Fox et al. 2016, Peters et al. 2017). Systemic adverse events in patients with hip implants have been mainly linked to Co, which was the main ion of interest in our study (Finley et al. 2012). Exposure to Cr(VI) (hexavalent chromium) may cause hemoglobin, leukocyte, and platelet abnormalities (Ray 2016). However, Cr in blood of patients with hip implants is thought to be not Cr(VI) but less toxic Cr(III) (trivalent chromium), which is not likely to cause health risk to patients (Finley et al. 2017). Cr at higher valences (most likely Cr[VI]) has been reported in postmortem samples of 2 diabetics, suggesting that re-oxidation of Cr(III) to more toxic Cr (VI) might be possible under certain circumstances (Swiatkowska et al. 2018). We are not aware of further studies confirming or annulling this finding. There is a case report concerning polycythemia (Hb 190 g/L) in a patient with an intensively worn CoCr head due to ceramic fragments from the previous fractured implant embedded in a new polyethylene liner resulting in extremely high (1,085 µg/L) serum Co levels (Gilbert et al. 2013). We are not aware of other studies regarding the association between Co and hemoglobin in patients with hip replacements. From the 1950s to 1980s Co was used in the treatment of anemia due to its potential to stimulate erythropoietin production, and polycythemia was described resulting from ingestion of Co. Among the published literature, Co doses resulting in blood Co concentrations of approximately 300 µg/L or less have not been associated with hematological responses, while blood Co concentrations of approximately 600 µg/L and higher have consistently been associated with polycythemia and increased hemoglobin content (Finley et al. 2012). Our findings from a hip arthroplasty population support the findings from dietary Co supplementation studies (Finley et al. 2012), as we found no tendency for polycythemia in patients with blood Co up to 225 µg/L. Several studies have reported associations between implant metals and leukocytes, mainly concerning lymphocytes. A randomized trial including 41 patients with MoM (mean Co 0.9 µg/L and Cr 1.1 µg/L) and 44 with metal-on-polyethylene (MoP) arthroplasty reported a higher percentage of HLA DR+ CD8+ T-cells and lower percentage of B-cells in the MoM group (Hailer et al. 2011). A cross-sectional study including 106 patients with MoM hip (median Co 1.7 µg/L and 2.5 µg/L for unilaterals and bilaterals, and Cr 2.3 µg/L and 2.4 µg/L, respectively) and 58 with non-MoM hip observed a reduction in the number of T-lymphocytes and B-lymphocytes in MoM patients (Hart et al. 2009). The finding was supported by a smaller randomized trial including 19 ASR hip resurfacing patients (median Co 1.4 µg/L and Cr 1.3 µg/L) and 19 patients with MoP implant (Penny et al. 2013). A study with 32 patients with MoM implant (mean serum Co 0.5µg/L and Cr 0.2 µg/L), 32 with MoP implant, and 32 controls pointed out that, in particular, CD3+, CD4+, and CD8+ T-lymphocytes were reduced in patients with MoM. In vitro studies have further characterized the reduction of lymphocyte viability and

Acta Orthopaedica 2020; 91 (6): 711–716

the increase in apoptosis due to implant metals (Akbar et al. 2011, Posada et al. 2015). Most of the studies cited have described changes in only certain subpopulations of lymphocytes. Some of the findings may be random variation interpreted as actual associations, but it is possible that Co and/or Cr may have a depressive effect on certain lymphocyte subpopulations even at lower metal ion levels. Our study did not include lymphocyte subpopulations as they are not routinely measured in clinical work. This study provides information only on total leukocytes and leukocyte differentials, and possible changes in a few lymphocyte subpopulations are unlikely to affect those. We did not observe an association between whole-blood Co/Cr and blood leukocyte, hemoglobin, or platelet counts. Of the leukocyte differential counts, the only correlations that included the predefined limit of 0.225 were those between whole-blood Co and blood eosinophils, and whole-blood Cr and blood monocytes. The CIs for these correlations were rather wide. The upper limits for CI of these correlations were 0.229 (equaling roughly 5.2% of variance explained) and 0.42 (equaling 17%). The reference values for blood eosinophils at our laboratory are 0.01–0.45 × 109/L and for monocytes 0.2–0.8 × 109/L. Of the upper limits of reference values, the 5.2% is 0.02 × 109/L and 17% is 0.1 × 109/L, both of which are negligible figures as absolute concentrations. Therefore, despite the CI including the predefined limit of 0.225, these findings are unlikely to be clinically relevant. Further, as a leukocyte differential was not drawn from all patients, there is a risk for selection bias, and therefore the interpretation of this subgroup analysis has to be made with even more caution. To our knowledge there are no case reports or studies concerning abnormal platelet values that would have been linked to implant metals. An in vitro study concerning CoCr stent materials stated that platelets adhere to and are activated by CoCr (Ollivier et al. 2017). 1 case report presented a patient with thrombocytopenia and anemia after orally ingesting a large amount of chromium picolinate (including trivalent Cr, resulting in peak Cr of 6.5 µg/L) used to enhance weight loss and to improve glycemic control (Cerulli et al. 1998). Thrombocytopenia has also been described after ingestion of chromium acid (including hexavalent Cr) mixed in a drink with homicidal intentions (Quaiser et al. 2014). We did not observe an association between implant metals and platelet count. The problem of determining whether the new abnormalities in patients’ blood counts are related to their MoM hip implants is not an everyday problem at an orthopedic outpatient clinic. Because of patients’ worries due to massive media attention (Cohen 2011) or legal aspects of the problems with MoM hip replacements (Dyer 2010, 2018), orthopedic surgeons may occasionally come across patients asking whether some of their symptoms or laboratory findings could be related to their MoM hip replacement. Our study adds, to the existing literature, a large volume of single-center clinical data concerning the relationship between implant metals and complete

715

blood count values with whole-blood Co concentrations up to 225 µg/L and Cr up to 125 µg/L. As our study included a large number of patients some of whom had abnormally high whole-blood Co and Cr ion concentrations, we believe this study had a good chance to reveal possible adverse hematological effects if those were to develop at concentrations included in this study. Our study is derived from a primary hospital treating patients from the municipality, regardless of their age or socioeconomic status, which makes our results generalizable. Conclusion In this, one of the largest single-center MoM cohorts in the world, we found no evidence of clinically significant correlations between whole-blood Co and Cr levels and leukocytes, hemoglobin, and platelets. Based on previous literature, hematological adverse effects are unlikely at Co concentrations of 300 µg/L and below, and Cr in blood of patients with hip implants is in less toxic trivalent Cr(III) form, which does not pose a health risk to patients. Our results supplement the previous results with clinical data from a hip arthroplasty population. The reasons for possible abnormal blood counts in patients with MoM hip and their Co and Cr below several hundred µg/L should be primarily sought elsewhere. Supplementary data Figures 1 and 2 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674. 2020.1827191

The authors would like to thank Ella Lehto RN, and Heli Kupari RN, Coxa Hospital for Joint Replacement, Tampere, Finland, for maintaining the study database. They thank Reija Autio Dr. Tech., University of Tampere, Faculty of Medicine and Life Sciences, Tampere, Finland, for statistical assistance. NH: literature search, data collection and analysis, interpretation of data, statistics, writing and revision of the manuscript, and final approval. O. Lainiala: study design, literature search, data collection and analysis, interpretation of data, statistics, writing and revision of the manuscript, and final approval. O. Laine, AR, AE: study design, interpretation of data and statistics, writing and revision of the manuscript, and final approval. Acta thanks Alister Hart, Raed Itayem and José M H Smolders for help with peer review of this study.

Akbar M, Brewer J M, Grant M H. Effect of chromium and cobalt ions on primary human lymphocytes in vitro. J Immunotoxicol 2011; 8(2): 140-9. Boardman D, Middleton F, Kavanagh T. A benign psoas mass following metal-on-metal resurfacing of the hip. J Bone Joint Surg Br 2006; 88(3): 402-4. Briggs T, Hanna S A, Kayani B, Tai S, Pollock R C, Cannon S R, Blunn G W, Carrington R. Metal-on-polyethylene versus metal-on-metal bearing surfaces in total hip arthroplasty: a prospective randomised study investigating metal ion levels and chromosomal aberrations in peripheral lymphocytes. Bone Joint J 2015; 97(9): 1183-91.

716

Brodner W, Bitzan P, Meisinger V, Kaider A, Gottsauner-Wolf F, Kotz R. Serum cobalt levels after metal-on-metal total hip arthroplasty. J Bone Joint Surg 2003; 85(11): 2168-73. Cerulli J, Grabe D W, Gauthier I, Malone M, McGoldrick M D. Chromium picolinate toxicity. Ann Pharmacother 1998; 32(4): 428-31. Chen Z, Wang Z, Wang Q, Cui W, Liu F, Fan W. Changes in early serum metal ion levels and impact on liver, kidney, and immune markers following metal-on-metal total hip arthroplasty. J Arthroplasty 2014; 29(3): 612-6. Cheung A C, Banerjee S, Cherian J J, Wong F, Butany J, Gilbert C, Overgaard C, Syed K, Zywiel M G, Jacobs J J, Mont M A. Systemic cobalt toxicity from total hip arthroplasties: review of a rare condition, Part 1: history, mechanism, measurements, and pathophysiology. Bone Joint J 2016; 98-B(1): 6-13. Cohen D. Out of joint: the story of the ASR. BMJ 2011; 342: d2905. DePuy Orthopaedics. ASR Recall; 2010. Dyer C. UK law firm starts action on “metal on metal” hip implants. BMJ 2010; 340: c2394. Dyer C. UK patients lose legal claim that DePuy metal-on-metal hip implant was “defective”. BMJ 2018; 361: k2253. Finley B L, Monnot A D, Gaffney S H, Paustenbach D J. Dose–response relationships for blood cobalt concentrations and health effects: a review of the literature and application of a biokinetic model. J Toxicol Environ Health B, Crit Rev 2012; 15(8): 493-523. Finley B, Scott P K, Glynn M E, Paustenbach D, Donovan E, Thuett K A. Chromium speciation in the blood of metal-on-metal hip implant patients. Toxicol Environ Chemistry 2017; 99(1): 48-64. Fox K A, Phillips T M, Yanta J H, Abesamis M G. Fatal cobalt toxicity after total hip arthroplasty revision for fractured ceramic components. Clin Toxicol 2016; 54(9): 874-7. Gilbert C, Cheung A, Butany J, Zywiel M, Syed K, McDonald M, Wong F, Overgaard C. Hip pain and heart failure: the missing link. Can J Cardiol 2013; 29(5): 639.e1-2. Gruber F W, Bock A, Trattnig S, Lintner F, Ritschl P. Cystic lesion of the groin due to metallosis: a rare long-term complication of metal-on-metal total hip arthroplasty. J Arthroplasty 2007; 22(6): 923-7. Hailer N P, Blaheta R A, Dahlstrand H, Stark A. Elevation of circulating HLA DR CD8 T-cells and correlation with chromium and cobalt concentrations 6 years after metal-on-metal hip arthroplasty: a randomized trial. Acta Orthop 2011; 82(1): 6-12. Hart A J, Skinner J A, Winship P, Faria N, Kulinskaya E, Webster D, Muirhead-Allwood S, Aldam C H, Anwar H, Powell J J. Circulating levels of cobalt and chromium from metal-on-metal hip replacement are associated with CD8 T-cell lymphopenia. J Bone Joint Surg Br 2009; 91(6): 835-42. Horton S, Fleming K A, Kuti M, Looi L, Pai S A, Sayed S, Wilson M L. The top 25 laboratory tests by volume and revenue in five different countries. Am J Clin Pathol 2019; 151(5): 446-51. Leyssens L, Vinck B, Van Der Straeten C, Wuyts F, Maes L. Cobalt toxicity in humans: a review of the potential sources and systemic health effects. Toxicology 2017; 387: 43-56.

Acta Orthopaedica 2020; 91 (6): 711–716

Markel D, Bergum C, Flynn J, Jackson N, Bou-Akl T, Ren W. Relationship of blood metal ion levels and leukocyte profiles in patients with articular surface replacement metal-on-metal hip replacement. Orthopedics 2018; 41(3): e424-31. MHRA. Medical device alert. DePuy ASR™ hip replacement implants (MDA/2010/069). 2010; 2019. Ollivier V, Roques C, Receveur N, Gratz M, Feldman L, Letourneur D, Gachet C, Mangin P H, Jandrot-Perrus M. Bioreactivity of stent material: activation of platelets, coagulation, leukocytes and endothelial cell dysfunction in vitro. Platelets 2017; 2(6): 529-39. Pandit H, Glyn-Jones S, McLardy-Smith P, Gundle R, Whitwell D, Gibbons C L, Ostlere S, Athanasou N, Gill H S, Murray D W. Pseudotumours associated with metal-on-metal hip resurfacings. J Bone Joint Surg Br 2008; 90(7): 847-51. doi: 10.1302/0301-620X.90B7.20213. Paustenbach D J, Galbraith D A, Finley B L. Interpreting cobalt blood concentrations in hip implant patients. Clin Toxicol 2014; 52(2): 98-112. Penny J Ø, Varmarken J, Ovesen O, Nielsen C, Overgaard S. Metal ion levels and lymphocyte counts: ASR hip resurfacing prosthesis vs. standard THA: 2-year results from a randomized study. Acta Orthop 2013; 84(2): 130-7. Peters R M, Willemse P, Rijk P C, Hoogendoorn M, Zijlstra W P. Fatal cobalt toxicity after a non-metal-on-metal total hip arthroplasty. Case Rep Orthopedics 2017; 2017: 9123684. doi: 10.1155/2017/9123684. Posada O M, Tate R J, Grant M H. Toxicity of cobalt–chromium nanoparticles released from a resurfacing hip implant and cobalt ions on primary human lymphocytes in vitro. J Appl Toxicol 2015; 35(6): 614-22. Quaiser S, Sharma A, Khan R. Survival after acute ingestion of chromic acid complicated by acute kidney injury: a case report. Egypt J Intern Med 2014; 26(4): 176. Ray R R. Adverse hematological effects of hexavalent chromium: an overview. Interdiscip Toxicol 2016; 9(2): 55-65. Reito A, Moilanen T, Puolakka T, Pajamäki J, Eskelinen A. Repeated metal ion measurements in patients with high risk metal-on-metal hip replacement. Int Orthop 2014; 38(7): 1353-61. Reito A, Lainiala O, Nieminen J, Eskelinen A. Repeated metal ion measurement in patients with bilateral metal on metal (ASR™) hip replacements. Orthop Traumatol Surg Res 2016; 102(2): 167-73. Silverman E J, Ashley B, Sheth N P. Metal-on-metal total hip arthroplasty: is there still a role in 2016? Curr Rev MusculoskeletaMed 2016; 9(1): 93-6. Swiatkowska I, Mosselmans J F W, Geraki T, Wyles C C, Maleszewski J J, Henckel J, Sampson B, Potter D, Osman I, Trousdale R, Hart A J. Synchrotron analysis of human organ tissue exposed to implant material. J Trace Elements Med Biol 2018; 46: 128-37. Tefferi A, Hanson C A, Inwards D J. How to interpret and pursue an abnormal complete blood cell count in adults. Mayo Clin Proc 2005; 80: 923-36. Tvermoes B E, Unice K M, Paustenbach D J, Finley B L, Otani J M, Galbraith D A. Effects and blood concentrations of cobalt after ingestion of 1 mg/d by human volunteers for 90 d. Am J Clin Nutr 2014; 99 (3): 632-46.

Acta Orthopaedica 2020; 91 (6): 717–723

717

Non-surgical treatment before hip and knee arthroplasty remains underutilized with low satisfaction regarding performance of work, sports, and leisure activities Yvonne VAN ZAANEN 1,a, Alexander HOORNTJE 2,3,a, Koen L M KOENRAADT 2, Leti VAN BODEGOM-VOS 4, Gino M M J KERKHOFFS 3, Suzanne WATERVAL-WITJES 1,2,5, Tim A E J BOYMANS 6, Rutger C I VAN GEENEN 2, and P Paul F M KUIJER 1 1 Department of Public and Occupational Health, Amsterdam UMC, University of Amsterdam, Amsterdam Public Health Research Institute, Amsterdam Movement Sciences, Amsterdam; 2 Department of Orthopaedic Surgery, Amphia Hospital, Foundation FORCE (Foundation for Orthopaedic Research Care and Education), Breda; 3 Department of Orthopaedic Surgery, Amsterdam UMC, University of Amsterdam, Academic Center for Evidence based Sports medicine (ACES), Amsterdam Movement Sciences, Amsterdam; 4 Department of Biomedical Data Sciences, Leiden University Medical Center, Leiden; 5 Personalized Knee Care, Maastricht; 6 Department of Orthopaedic surgery, Maastricht University Medical Center, Maastricht, The Netherlands a Shared first authorship. Correspondence: y.vanzaanen@amsterdamumc.nl Submitted 2020-02-26. Accepted 2020-08-10.

Background and purpose — Guidelines for managing hip and knee osteoarthritis (OA) advise extensive non-surgical treatment prior to surgery. We evaluated what percentage of hip and knee OA patients received non-surgical treatment prior to arthroplasty, and assessed patient satisfaction regarding alleviation of symptoms and performance of activities. Patients and methods — A multi-center cross-sectional study was performed in 2018 among 186 patients who were listed for hip or knee arthroplasty or had undergone surgery within the previous 6 months in the Netherlands. Questions concerned non-surgical treatments received according to the Stepped Care Strategy and were compared with utilization in 2013. Additionally, satisfaction with treatment effects for pain, swelling, stiffness, and activities of daily life, work, and sports/leisure was questioned. Results — The questionnaire was completed by 175 patients, age 66 years (range 38–84), 57% female, BMI 29 (IQR 25–33). Step 1 treatments, such as acetaminophen and lifestyle advice, were received by 79% and 60% of patients. Step 2 treatments, like exercise-based therapy and diet therapy, were received by 66% and 19%. Step 3—intra-articular injection—was received by 47%. Non-surgical treatment utilization was lower than in 2013. Nearly all treatments showed more satisfied patients regarding pain relief and fewer regarding activities of work/sports/leisure. Hip and knee OA patients were mostly satisfied with NSAIDs for all outcomes, while exercise-based therapy was rated second best. Interpretation — Despite international guideline recommendations, non-surgical treatment for hip and knee OA remains underutilized in the Netherlands. Of the patients referred for arthroplasty, more were satisfied with the effect of non-surgical treatment on pain than on work/sports/leisure participation.

Worldwide guidelines for managing osteoarthritis (OA) of the hip and knee advise extensive non-surgical treatment prior to surgery (Zhang et al. 2010, Smink et al. 2011, McAlindon et al. 2014). Non-surgical treatment is cost-effective and may lower the rapidly increasing OA-related healthcare expenditure by delaying or even replacing surgery (Berwick and Hackbarth 2012). The global Choosing Wisely initiative aims to optimize healthcare usage and costs by advocating the use of proven but underused healthcare modalities, including preventive care (Berwick and Hackbarth 2012, Bernstein 2015). Regarding hip and knee OA, studies have found underuse of nonsurgical treatments (Snijders et al. 2011, Hofstede et al. 2015). For example, 1 study showed that 81% of hip and knee OA patients did not receive all recommended non-surgical treatments (Snijders et al. 2011). In the Netherlands, a Stepped Care Strategy (SCS) was developed to stimulate the use of non-surgical treatment before hip and knee replacement (Smink et al. 2011). Moreover, providing adequate non-surgical treatment before hip and knee replacement was recommended by the Dutch Orthopedic Association for their Choosing Wisely Campaign (NOV 2015). Yet, the actual utilization of non-surgical treatment in hip and knee OA patients prior to arthroplasty in the Netherlands is described only by a cohort study from 2013 (Hofstede et al. 2015). Furthermore, no previous study has simultaneously assessed patient satisfaction with non-surgical treatments regarding their effect on symptoms like pain and swelling, and participation as in daily life and work. This is of importance given the increasing number of hip and knee OA patients who want to eliminate their pain and also wish to remain active in daily life, work, and sport/ leisure (Kurtz et al. 2009, Otten et al. 2010, Culliford et al. 2015, Witjes et al. 2017). Given the impact of surgery on work

718

participation, the effect of non-surgical treatment on work participation is also of interest (Kuijer et al. 2016, Stigmar et al. 2017). Therefore, the main aim was to assess preoperative nonsurgical treatment by hip and knee OA patients referred for arthroplasty in 2018, as well as compared with 2013, and their satisfaction regarding alleviation of symptoms and performing activities of daily living (ADL), work, and sports/leisure activities.

Patients and methods Multi-center study A multi-center cross-sectional online questionnaire study was performed by convenience sampling, similar to the Dutch study of 2013 (Hofstede et al. 2015), between October 2017 and April 2018 in 5 Dutch hospitals. Eligible patients were either listed for total hip arthroplasty (THA) or total knee arthroplasty (TKA) or had undergone THA or TKA less than 6 months previously in order to minimize recall bias. The participating hospitals were located in the northern, central, and southern parts of the Netherlands, including city and rural areas, and serving general THA and TKA populations. All patients received written information concerning the study and an online invitation to participate. If they agreed to participate, they received the invitation to fill out the online questionnaire. Furthermore, all participating patients received a gift card to the value of 10 euros after completing the questionnaire. Questionnaire An online questionnaire was developed using an electronic data management system (Castor EDC, www.castoredc.com). Eligible patients received an invitation by email, followed by a maximum of 2 email reminders. The questionnaire started with questions regarding baseline characteristics, including age, sex, bodyweight and height, educational level, comorbidity, work situation, and onset of OA complaints. Non-surgical treatment modalities Received non-surgical treatment modalities were asked about based on the multidisciplinary guideline, which consists of 3 sequenced steps, called the Stepped Care Strategy (SCS) (Smink et al. 2011). These 3 steps are: (1) education, lifestyle advice, acetaminophen, glucosamine sulfate (optional); (2) exercise-based therapy, diet therapy, NSAIDs, tramadol; and (3) intra-articular injection. Multidisciplinary care, a treatment option of step 3, was left out because this consists of treatment modalities similar to the monodisciplinary care of step 1 and 2 (Smink et al. 2011), which patients might find hard to distinguish. To assess whether the Dutch SCS and Choosing Wisely campaign (2015) resulted in higher utilization of non-surgical therapy, our results were compared with a similar Dutch study

Acta Orthopaedica 2020; 91 (6): 717–723

performed in 2013 (Hofstede et al. 2015). Higher utilization was defined as a 10% higher utilization rate (Grimshaw et al. 2004). Patient satisfaction Patients were asked to rate their satisfaction regarding the treatment effect of the above-mentioned non-surgical treatment modalities on a Likert scale from 1 (very unsatisfied) to 10 (very satisfied). For every received treatment 6 satisfaction rates were asked about for the effect on symptoms: pain, swelling, and stiffness; and on participation: daily life, work, and sports/leisure. These outcomes are in line with the OMERACT-OARSI core domain set (Singh et al. 2017, Smith et al. 2019). A cut-off point of 6 or higher was used to distinguish between “satisfied” and “not satisfied.” Statistics Patient characteristics were described for the baseline characteristics such as age, sex, BMI, and onset of OA complaints. Subgroup analyses were performed for hip versus knee OA patients. Statistical differences in baseline characteristics and received treatments were tested between these subgroups of OA patients using a Student’s t-test, Mann–Whitney U-test, or Fisher–Freeman–Halton Exact test if corresponding assumptions were met. 2 sensitivity analyses were performed: (1) among patients who completed the questionnaire preoperatively versus postoperatively to explore whether this biased the results and (2) among patients in paid employment to assess their satisfaction with treatment effects for work participation to decide whether the results were biased by employment status at the time of surgery. All statistical analyses were performed using SPSS for Windows (Version 24.0; IBM Corp, Armonk, NY, USA). A significance level of p ≤ 0.05 was used. Ethics, funding, and potential conflicts of interest The Medical Ethics Review Committee of the Amsterdam UMC, location Academic Medical Center, confirmed that the Medical Research Involving Human Subjects Act (WMO) did not apply to this study and official approval is not required (reference number W17_325 #17.378). Informed consent was obtained from all participants included in the study. This project received funding from the Netherlands Organisation for Health Research and Development (ZonMw) (reference number 516000503). The funder had no role in the conducting of the study or the decision to publish. No competing interests were declared.

Results Of 371 invited patients, 186 were willing to participate (response rate 50%) and 175 patients completed the questionnaire (completion rate 94%; Figure 1). Mean age was 66 (SD 8) years, 57% were female, 3 out of 4 were overweight

Acta Orthopaedica 2020; 91 (6): 717–723

Patients eligible for inclusion October 2017 – April 2019 n = 371

719

Acetominophen (1) Education in OA (1) Education in treatments of OA (1) Lifestyle advice (1) Glucosamine sulfate (1)

Excluded (n = 185): – did not respond, 160 – unwilling to participate, 25

Exercised based therapy (2) (Topical) NSAID (2) Dietary therapy if BMI ≥30 (2)

Participated in the study n = 186 Exclusion: Incomplete questionnaire n = 11 Patients included in analysis n = 175

Figure 1. Flowchart of participating patients.

(median BMI 29 [IQR 25–33]) and 67% had no paid employment at the time of surgery (Table). Finally, in half of the patients, OA complaints lasted longer than 5 years. Differences between hip and knee OA patients The median BMI of knee OA patients was statistically significantly higher (31 [IQR 27–36]) than that of the hip OA patients (28 [IQR 24–31]), and complaints lasted longer in knee OA patients (complaints > 5 years in 62%) than in hip OA patients (complaints > 5 years in 37%; Table). Received non-surgical treatments Of the SCS treatments of step 1, acetaminophen was received most often (hip 73%, knee 79%; p = 0.4) and lifestyle advice least often (hip 60%, knee 62%; p = 0.9) (Figure 2). Glucosamine sulfate was received by 18% of hip OA patients and 21% of knee OA patients (p = 0.7). Of the SCS treatments of step 2, exercise-based therapy was received most often by patients (hip 66%, knee 59%; p = 0.4), while diet therapy among overweight patients (hip 23%, knee 19%; p = 0.3) and tramadol (hip 11%, knee 20%; p = 0.1) were least often received. Intra-articular injections, the SCS step 3 treatment, were more often received by knee OA patients (47%) than by hip OA patients (10%; p < 0.01). Comparison of non-surgical treatment utilization over time In comparison with the data from 2013 patients in 2018 were older (mean age 66 versus 64; p = 0.03), less often female (57% versus 72%; p = 0.01), more overweight (median 29 [IQR 25–33] versus 28 [25–31]), and their OA complaints lasted longer (> 5 years 50% versus 43%; p <0.01). In 2018, the hip and knee OA patients reported having received less exercise-based therapy (–12%, prevalence ratio (PR) = 0.85 (95% confidence interval [CI] 0.73–0.99), less

Dietary therapy if BMI ≥25 and <30 (2) Tramadol (2) a

Intra-articular injection (3) 0

Knee (n = 92) Hip (n = 82)

10 20 30 40 50 60 70 80 90 100

Received non-surgical treatment (%)

Figure 2. Non-surgical treatment received by hip and knee osteoarthritis (OA) patients according to the Stepped Care Strategy (Step 1, 2, or 3). a Significant difference between hip and knee OA patients (Fisher–Freeman–Halton Exact test p ≤ 0.05).

Patient characteristics of the total group of hip and knee OA patients and each group separately. Values are number (%) unless otherwise specified Variable

Total OA group n = 175

Hip OA a n = 82 (47%)

Knee OA a n = 92 (53%)

Mean age (SD) b 66 (8) 66 (9) 65 (7) Female sex, n (%) c 100 (57) 42 (51) 57 (62) BMI, median [IQR] d, e 29 [25–33] 28 [24–31] 31 [27–36] Educational level d Primary 48 (27) 19 (23) 29 (32) Secondary 72 (41) 35 (43) 36 (39) College/university 55 (31) 28 (34) 27 (29) d, f Comorbidity Diabetes mellitus 19 (11) 6 (7) 13 (14) Cerebrovascular accident 10 (6) 3 (4) 7 (8) Cancer 12 (7) 5 (6) 7 (8) Cardiac diseases 20 (11) 5 (6) 15 (16) Migraine/severe headache 9 (5) 5 (6) 4 (4) High blood pressure 54 (31) 23 (28) 31 (34) Lung diseases 20 (11) 6 (7) 14 (15) Rheumatic diseases 20 (11) 8 (10) 12 (13) Other 22 (13) 9 (11) 13 (14) Work circumstances d Paid employment 47 (27) 22 (27) 25 (27) Self-employed 10 (6) 5 (6) 5 (5) No paid employment g 118 (67) 55 (67) 62 (67) Onset of OA complaints d, e < 1 year 9 (5) 4 (5) 5 (5) 1–5 years 78 (45) 48 (59) 30 (33) > 5 years 28 (16) 12 (15) 15 (16) > 10 years 37 (21) 12 (15) 25 (27) > 20 years 23 (13) 6 (7) 17 (19) d Region of the country North 37 (21) 21 (26) 16 (17) Middle 25 (14) 12 (15) 12 (13) South 113 (65) 49 (60) 64 (70) a 1 hip or knee is missing; IQR = interquartile range; SD = standard deviation; b Student’s t-test; c Fisher–Freeman–Halton exact test; d Mann–Whitney U-test; e Significant difference between groups p ≤ 0.05; f counting > 100% while more than one answer possible. g Unemployed, retirement, etc.

720

Acta Orthopaedica 2020; 91 (6): 717–723

HIPS

Acetominophen (1)

KNEES

Glucosamine sulfate (1) (Topical) NSAID (2) Tramadol (2) Exercised based therapy (2) Pain Stiffness Swelling ADL Work Sport/leisure

Dietary therapy (2) Intra-articular injection (3) 0

10 20 30 40 50 60 70 80 90 100 0

Pain Stiffness Swelling ADL Work Sport/leisure

10 20 30 40 50 60 70 80 90 100

Percentage with a satisfaction rate ≥6 Percentage with a satisfaction rate ≥6 Figure 3. Percentage of hip and knee OA patients with a satisfaction rate ≥ 6 for effect of non-surgical treatment according to the Stepped Care Strategy (Step 1, 2, or 3) on pain, stiffness, swelling, activities of daily life (ADL), work, and sports/leisure.

glucosamine sulfate (–16%, PR = 0.56 [CI 0.39–0.87]), fewer NSAIDs (–15%, PR = 0.76 [CI 0.63–0.92]), and less tramadol (–10%, PR = 0.62 [CI 0.40–0.95]) than in 2013. For the other non-surgical treatments, non-significant differences in utilization of less than 10% were found. Satisfaction with non-surgical treatment Regarding acetaminophen, 54% of the hip OA patients were satisfied with the effect on pain, 35% on stiffness, 30% on swelling, 41% on ADL, 31% on work, and 22% on sport and leisure participation (Figure 3). Figure 3 also shows these results of the other non-surgical treatments for hip OA patients and the same results for knee OA patients. Though there were differences in percentages between satisfied hip and knee OA patients of up to 32% for instance for the effect of tramadol on pain, no significant differences were found for the distribution of satisfaction rates between hip and knee OA patients. Sensitivity analyses Of the 175 patients, 112 (64%) completed the questionnaire preoperatively. No statistically significant differences were found in baseline characteristics except for country region: 54% of preoperative patients came from the south versus 84% of postoperative. No statistically significant differences were found in received treatments except for education in treatments of OA, which was received by 74% of the preoperative patients and 57% of postoperative (p = 0.03). Patients on the waiting list for surgery were less often satisfied than patients already operated on regarding physical therapy for stiffness (44% versus 61%, p = 0.03), NSAIDs for swelling (40% versus 65%, p = 0.04), NSAIDs for sport (29% versus 69%, p = 0.01), and tramadol for pain (38% versus 67%, p = 0.05). Of the 175 patients, 57 (33%) had paid employment at the time of surgery. Patients with paid employment were younger

(59 years, SD = 7) than patients without paid employment (69 years, SD = 7; p < 0.001) and they were more highly educated than patients without paid employment (p < 0.01). In addition, diabetes mellitus was less common among patients with paid employment (p = 0.04) and rheumatic diseases were more common (p = 0.01). Lifestyle advice and exercise-based therapy were received by more hip and knee OA patients with paid employment (both 77%) compared with patients without paid employment (52% and 54%, respectively; p < 0.01). Hip and knee OA patients with paid employment were more satisfied with the effect of NSAIDs for work participation (56% versus 33%; p = 0.02), intra-articular injection (54% versus 20%; p < 0.01) and exercise-based therapy (42% versus 40%; p = 0.6) than patients without paid employment at the time of surgery.

Discussion The most important finding of the present study was an underuse of non-surgical treatments reported by hip and knee OA patients. In terms of the SCS, step 1 treatments were received by up to 79%, step 2 by up to 66%, and step 3 by up to 47%. Regarding satisfaction with treatments, more than half of the patients were satisfied with the effect on pain of acetaminophen (hip), NSAIDs (hip and knee), tramadol (knee), exercise-based therapy (hip and knee), and intra-articular injection (knee). However, satisfaction rates with non-surgical treatments for work and sports/leisure participation were generally low. The comparison of non-surgical treatment utilization over time revealed that no treatment was received more often than in 2013 in the Netherlands. Non-surgical treatment remains underutilized and this is especially true for treatments of steps 2 and 3. These results are in line with recent findings from Denmark showing underuse of non-surgical knee OA treat-

Acta Orthopaedica 2020; 91 (6): 717–723

ment before referral to an orthopedic surgeon (Ingelsrud et al. 2020). Underuse of NSAIDs and intra-articular injection might be related to safety implications (Charlesworth et al. 2019). However, it seems unlikely that this explains why these treatments are not used or received by more than 50% of patients given the recommendation in the SCS strategy. This assumption is also supported by the positive results of hyaluronic acid on knee-related function and knee complaints according to a recent RCT among working-age knee OA patients (Hermans et al. 2019). Also, our sensitivity analyses showed that patients with paid employment were especially satisfied with the effects of these intra-articular injections on work participation. The same was true for NSAIDs. Given the possible positive effect of these treatments on work participation for half of the patients, these treatments might be underutilized for working-age patients. Unfortunately, no work-related outcomes are presented by Hermans et al. (2019). Several reasons for the underuse of the non-surgical treatments can be mentioned. For instance, Hofstede et al. (2016) showed that an important barrier might be that orthopedic surgeons have little faith in the effectiveness of these treatments. The same might be true for general practitioners’ attitudes towards non-surgical treatment (Smink et al. 2014). For patients, positive experiences with surgery of people in their own environment may lead to the belief that non-surgical therapy is inferior to surgery (Hofstede et al. 2016), even though about 10% of hip OA patients and 20% of knee OA patients reported persistent pain after joint replacement, even after generic or indicated pain education (Beswick et al. 2012, Louw et al. 2019, Birch et al. 2020). Consequently, better education on the effectiveness of non-surgical treatments for both patients and healthcare professionals might support nonsurgical treatment utilization. Indeed, active regional implementation of the SCS, including a patient education booklet, educational outreach visits, and providing reminder material to general practitioners, resulted in increased adherence to the SCS (Smink et al. 2014). These findings justify further dissemination of this implementation strategy to healthcare providers of hip and knee OA patients. Our comparison of non-surgical treatment utilization over time also showed that initiatives based on the Choosing Wisely campaign will not produce the desired results when no additional implementation measures, as stated above, are taken. That patients were older and complaints lasted longer in the present study compared with the study from 2013 (Hofstede et al. 2015) can be interpreted as a trend towards a “wait and see” policy before arthroplasty. If this is the case, this waiting time can be used more effectively by non-surgical treatment. For instance, being overweight is more common in the present study and diet therapy is used in only a third or less by the responding obese or overweight patients. The fact that at least 30% of patients are satisfied with the effect of diet therapy on pain, and in knee OA patients also on stiffness, ADL, and sport/leisure, may support better use of this treatment.

721

In this study, the core outcome domain set for total joint replacement (TJR) (Singh et al. 2017) recommending the use of pain, satisfaction, and participation is used to align outcomes of received non-surgical treatments with the outcomes of TJR. These outcomes are essential for providing better patient-tailored care, particularly for chronic diseases like OA (Karsdal et al. 2014, Witjes et al. 2017). For example, work-disabled knee OA patients may be advised that hyaluronic acid in combination with exercise-based therapy could provide the optimal non-surgical treatment for pain and work participation (Hermans et al. 2019). Additionally, patients might be encouraged to discuss which non-surgical treatments might be most beneficial for their specific needs regarding symptoms and participation. Lastly, data on patient satisfaction regarding the effects on symptoms and participation of non-surgical treatment might stimulate healthcare professionals and researchers to prescribe and develop more effective treatment combinations. While interest in and social demand for these non-surgical treatments is growing (Karsdal et al. 2014, Skou et al. 2018), we hope that an increasing number of hip and knee OA patients receive nonsurgical treatment, thereby enabling surgery to be postponed. Strengths and limitations Our study has some limitations. Most importantly, we performed a retrospective study relying on self-reported data, thus making our findings prone to recall bias. Therefore, in line with previous recommendations, we limited inclusion to patients who were scheduled for total hip or total knee arthroplasty or who underwent total hip or total knee arthroplasty no longer than 6 months previously (Hofstede et al. 2015). Another limitation might be that the patients’ satisfaction rates with the treatment effects, given the indication for surgery, are probably lower than the rates in patients in an earlier stage of OA. Thus, regarding external validity of our findings, our satisfaction rates may only be applicable to preoperative hip and knee OA patients. A strength of our study is that for the first time, as far as we are aware, we evaluated patient satisfaction regarding the effects of non-surgical treatments in hip and knee OA not only on pain but also on other symptoms like stiffness, and on participation as in work. Moreover, we were able to make a comparison regarding utilization of non-surgical treatment between the present study and data collected in a similar manner in 2013 (Hofstede et al. 2015). Another strength is our evaluation after the implementation of an evidence-based non-surgical SCS and the Choosing Wisely campaign. Therefore, our findings on utilization of non-surgical therapy may be valid for countries with comparable initiatives and healthcare provision, although financial compensation for and social acceptance of treatments may differ according to country. Conclusion Non-surgical treatments for hip and knee OA patients appear underutilized in the Netherlands. Of the patients referred for

722

arthroplasty, generally more were satisfied with the effect of non-surgical treatment on pain than with the effect on participation in work and sports/leisure. Better insight into patients’ satisfaction regarding non-surgical treatment effects on symptoms and participation might stimulate patient-centered care and thereby increase better adherence.

This study was set up by a joint effort of AH, KK, LB, GK, SW, TB, RG, and PK. SW, AH, RG, and PK conceived and designed the study. AH designed and distributed the surveys and completed the data collection. YZ and PK were responsible for the processing and analyses of data. YZ drafted the first version of the article. AH and PK contributed to the further drafting of the article. KK, LB, GK, SW, TB, and RG all agreed on conception and design and contributed equally to interpretation of data and its critical revision. YZ and AH edited the manuscript and YZ wrote the final draft. PK is guarantor of the article. Final approval of the version for publication was agreed upon by all authors. The authors would like to thank all patients who participated in this study. Acta thanks Lina Holm Ingelsrud and Maria C Lindström for help with peer review of this study. Bernstein J. Not the last word: choosing wisely. Clin Orthop Relat Res 2015; 473(10): 3091-7. doi: 10.1007/s11999-015-4490-8. Berwick D M, Hackbarth A D. Eliminating waste in US health care. JAMA 2012; 307(14): 1513-16. doi: 10.1001/jama.2012.362. Beswick A D, Wylde V, Gooberman-Hill R, Blom A, Dieppe P. What proportion of patients report long-term pain after total hip or knee replacement for osteoarthritis? A systematic review of prospective studies in unselected patients. BMJ Open 2012; 2(1): e000435. doi: 10.1136/bmjopen-2011-000435. Birch S, Stilling M, Mechlenburg I, Hansen T B. No effect of cognitive behavioral patient education for patients with pain catastrophizing before total knee arthroplasty: a randomized controlled trial. Acta Orthop 2020; 91(1): 98-103 doi: 10.1080/17453674.2019.1694312. Charlesworth J, Fitzpatrick J, Perera N K P, Orchard J. Osteoarthritis—a systematic review of long-term safety implications for osteoarthritis of the knee. BMC Musculoskelet Disord 2019; 20(1): 151. doi: 10.1186/s12891-019-2525-0. Culliford D, Maskell J, Judge A, Cooper C, Prieto-Alhambra D, Arden N K. Future projections of total hip and knee arthroplasty in the UK: results from the UK Clinical Practice Research Datalink. Osteoarthritis Cartilage 2015; 23(4): 594-600. doi: 10.1016/j.joca.2014.12.022. Grimshaw J, Eccles M, Tetroe J. Implementing clinical guidelines: current evidence and future implications. J Contin Educ Health Prof 2004; 24(Suppl. 1): S31-7. doi: 10.1002/chp.1340240506. Hermans J, Bierma-Zeinstra S M A, Bos P K, Niesten D D, Verhaar J A N, Reijman M. The effectiveness of high molecular weight hyaluronic acid for knee osteoarthritis in patients in the working age: a randomised controlled trial. BMC Musculoskelet Disord 2019; 20(1): 196. doi: 10.1186/s12891019-2546-8. Hofstede S N, Vliet Vlieland T P, van den Ende C H, Nelissen R G, Marangvan de Mheen P J, van Bodegom-Vos L. Variation in use of non-surgical treatments among osteoarthritis patients in orthopaedic practice in the Netherlands. BMJ Open 2015; 5(9): e009117. doi: 10.1136/bmjopen-2015-009117. Hofstede S N, Marang-van de Mheen P J, Vliet Vlieland T P, van den Ende C H, Nelissen R G, van Bodegom-Vos L. Barriers and facilitators associated with non-surgical treatment use for osteoarthritis patients in orthopaedic practice. PLoS One 2016; 11(1): e0147406. doi: 10.1371/journal. pone.0147406.

Acta Orthopaedica 2020; 91 (6): 717–723

Ingelsrud L H, Roos E M, Gromov K, Jensen S S, Troelsen A. Patients report inferior quality of care for knee osteoarthritis prior to assessment for knee replacement surgery: a cross-sectional study of 517 patients in Denmark. Acta Orthop 2020; 91(1): 82-7. doi: 10.1080/17453674.2019.1680180. Karsdal M A, Christiansen C, Ladel C, Henriksen K, Kraus V B, Bay-Jensen A C. Osteoarthritis: a case for personalized health care? Osteoarthritis Cartilage 2014; 22(1): 7-16. doi: 10.1016/j.joca.2013.10.018. Kuijer P P, Kievit A J, Pahlplatz T M, Hooiveld T, Hoozemans M J, Blankevoort L, Schafroth M U, van Geenen R C, Frings-Dresen M H. Which patients do not return to work after total knee arthroplasty? Rheumatol Int 2016; 36(9): 1249-54. doi: 10.1007/s00296-016-3512-5. Kurtz S M, Lau E, Ong K, Zhao K, Kelly M, Bozic K J. Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030. Clin Orthop Relat Res 2009; 467(10): 2606-12. doi: 10.1007/s11999-009-0834-6. Louw A, Puentedura E J, Reed J, Zimney K, Grimm D, Landers M R. A controlled clinical trial of preoperative pain neuroscience education for patients about to undergo total knee arthroplasty. Clin Rehabil 2019; 33(11): 1722-31. doi: 10.1177/0269215519857782. McAlindon T E, Bannuru R R, Sullivan M C, Arden N K, Berenbaum F, Bierma-Zeinstra S M, Hawker G A, Henrotin Y, Hunter D J, Kawaguchi H, Kwoh K, Lohmander S, Rannou F, Roos E M, Underwood M. OARSI guidelines for the non-surgical management of knee osteoarthritis. Osteoarthritis Cartilage 2014; 22(3): 363-88. doi: 10.1016/j. joca.2014.01.003. NOV. Choosing wisely in orthopedics; Verstandige keuzes binnen de orthopedie; 1. geen heup- en knievervanging zonder adequaat conservatieve behandeling. In: Verstandig kiezen. ZonMw (The Netherlands Organisation for Health Research and Development) NOV NPCF; 2015. Otten R, van Roermund P M, Picavet H S. [Trends in the number of knee and hip arthroplasties: considerably more knee and hip prostheses due to osteoarthritis in 2030]. Ned Tijdschr Geneeskd 2010; 154: A1534. Singh J A, Dowsey M M, Dohm M, Goodman S M, Leong A L, Scholte Voshaar M, Choong P F. Achieving consensus on total joint replacement trial outcome reporting using the OMERACT filter: endorsement of the final core domain set for total hip and total knee replacement trials for endstage arthritis. J Rheumatol 2017; 44(11): 1723-6. doi: 10.3899/ jrheum.161113. Skou S T, Roos E M, Laursen M B, Rathleff M S, Arendt-Nielsen L, Rasmussen S, Simonsen O. Total knee replacement and non-surgical treatment of knee osteoarthritis: 2-year outcome from two parallel randomized controlled trials. Osteoarthritis Cartilage 2018; 26(9): 1170-80. doi: 10.1016/j. joca.2018.04.014. Smink A J, van den Ende C H, Vliet Vlieland T P, Swierstra B A, Kortland J H, Bijlsma J W, Voorn T B, Schers H J, Bierma-Zeinstra S M, Dekker J. “Beating osteoarthritis”: development of a stepped care strategy to optimize utilization and timing of non-surgical treatment modalities for patients with hip or knee osteoarthritis. Clin Rheumatol 2011; 30(12): 1623-9. doi: 10.1007/ s10067-011-1835-x. Smink A J, Bierma-Zeinstra S M, Schers H J, Swierstra B A, Kortland J H, Bijlsma J W, Teerenstra S, Voorn T B, Dekker J, Vliet Vlieland T P, van den Ende C H. Non-surgical care in patients with hip or knee osteoarthritis is modestly consistent with a stepped care strategy after its implementation. Int J Qual Health Care 2014; 26(4): 490-8. doi: 10.1093/intqhc/mzu058. Smith T O, Hawker G A, Hunter D J, March L M, Boers M, Shea B J, Christensen R, Guillemin F, Terwee C B, Williamson P R, Dodd S, Roos E M, Loeser R F, Schnitzer T J, Kloppenburg M, Neogi T, Ladel C H, Kalsi G, Kaiser U, Buttel T W, Ashford A E, Mobasheri A, Arden N K, Tennant A, Hochberg M C, de Wit M, Tugwell P, Conaghan P G. The OMERACTOARSI core domain set for measurement in clinical trials of hip and/ or knee osteoarthritis. J Rheumatol 2019; 46(8): 981-9. doi: 10.3899/ jrheum.181194. Snijders G F, den Broeder A A, van Riel P L, Straten V H, de Man F H, van den Hoogen F H, van den Ende C H, Group N S. Evidence-based tailored conservative treatment of knee and hip osteoarthritis: between knowing and doing. Scand J Rheumatol 2011; 40(3): 225-31. doi: 10.3109/03009742.2010.530611.

Acta Orthopaedica 2020; 91 (6): 717â&#x20AC;&#x201C;723

Stigmar K, Dahlberg L E, Zhou C, Jacobson Lidgren H, Petersson I F, Englund M. Sick leave in Sweden before and after total joint replacement in hip and knee osteoarthritis patients. Acta Orthop 2017; 88(2): 152-7. doi: 10.1080/17453674.2016.1269051. Witjes S, van Geenen R C, Koenraadt K L, van der Hart C P, Blankevoort L, Kerkhoffs G M, Kuijer P P. Expectations of younger patients concerning activities after knee arthroplasty: are we asking the right questions? Qual Life Res 2017; 26(2): 403-17. doi: 10.1007/s11136-016-1380-9.

723

Zhang W, Nuki G, Moskowitz R W, Abramson S, Altman R D, Arden N K, Bierma-Zeinstra S, Brandt K D, Croft P, Doherty M, Dougados M, Hochberg M, Hunter D J, Kwoh K, Lohmander L S, Tugwell P. OARSI recommendations for the management of hip and knee osteoarthritis, part III: Changes in evidence following systematic cumulative update of research published through January 2009. Osteoarthritis Cartilage 2010; 18(4): 47699. doi: 10.1016/j.joca.2010.01.013.

724

Acta Orthopaedica 2020; 91 (6): 724–731

Outcome at 1 year in patients with femoral shaft fractures treated with intramedullary nailing or skeletal traction in a low-income country: a prospective observational study of 187 patients in Malawi Linda CHOKOTHO 1,2, Hao-Hua WU 3, David SHEARER 3, Brian C LAU 4, Nyengo MKANDAWIRE 1,6, Jan-Erik GJERTSEN 2,5, Geir HALLAN 2,5, and Sven YOUNG 5,7 1 Department

of Surgery, College of Medicine, University of Malawi; 2 Department of Clinical Medicine, University of Bergen, Bergen, Norway; 3 Institute for Global Orthopedics and Traumatology, Orthopedic Trauma Institute, University of California San Francisco, San Francisco, CA, USA; 4 Department of Orthopedic Surgery, Duke University Medical Centre, Durham, NC, USA; 5 Department of Orthopedic Surgery, Haukeland University Hospital, Bergen, Norway; 6 School of Medicine, Flinders University, Adelaide, Australia; 7 Department of Surgery, Kamuzu Central Hospital, Lilongwe, Malawi Correspondence: lindachokotho@gmail.com Submitted 2020-03-26. Accepted 2020-06-30.

Background and purpose — Intramedullary nailing (IMN) is underutilized in low-income countries (LICs) where skeletal traction (ST) remains the standard of care for femoral shaft fractures. This prospective study compared patient-reported quality of life and functional status after femoral shaft fractures treated with IMN or ST in Malawi. Patients and methods — Adult patients with femoral shaft fractures managed by IMN or ST were enrolled prospectively from 6 hospitals. Quality of life and functional status were assessed using EQ-5D-3L, and the Short Musculoskeletal Function Assessment (SMFA) respectively. Patients were followed up at 6 weeks, 3, 6, and 12 months post-injury. Results — Of 248 patients enrolled (85 IMN, 163 ST), 187 (75%) completed 1-year follow-up (55 IMN, 132 ST). 1 of 55 IMN cases had nonunion compared with 40 of 132 ST cases that failed treatment and converted to IMN (p < 0.001). Quality of life and SMFA Functional Index Scores were better for IMN than ST at 6 weeks, 3 and 6 months, but not at 1 year. At 6 months, 24 of 51 patients in the ST group had returned to work, compared with 26 of 37 in the IMN group (p = 0.02). Interpretation — Treatment with IMN improved early quality of life and function and allowed patients to return to work earlier compared with treatment with ST. Approximately one-third of patients treated with ST failed treatment and were converted to IMN.

The gold standard treatment for femoral shaft fractures is intramedullary nailing (IMN), with low complication rates ranging from 1.2% to 5% for postoperative infection (Brumback et al. 2006, Young et al. 2013a, Salawu et al. 2017) and high union rates ranging from 72% to 100% (Ricci et al. 2001, El Moumni et al. 2009, Young et al. 2013b). However, nonoperative treatment using skeletal traction (ST) for at least 6 weeks remains the mainstay treatment for these fractures in low-resource settings (Hollis et al. 2015, Kramer et al. 2016). Nonoperative treatment is associated with increased risk of both medical and surgical complications, reported as high as 55% in some studies (Bucholz and Jones 1991, Doorgakant and Mkandawire 2012, Kramer et al. 2016, Parkes et al. 2017). In Malawi, femoral shaft fractures are most commonly treated by ST. IMN, when performed, is done using the SIGN IM nail, which is donated by SIGN Fracture Care International (Richland, WA, USA) (Shah et al. 2004). Most studies comparing IMN with ST in LICs used conventional measures such as fracture union, complications, and range of motion (Swai 2005, Kamau et al. 2014, Parkes et al. 2017). No prior study has measured quality of life or function using a validated patient-reported outcome instrument to compare ST and IMN in any context. This study compared the quality of life and functional status of patients with femoral shaft fractures treated with either ST or IMN in Malawi.

Acta Orthopaedica 2020; 91 (6): 724–731

Eligible femoral fractures n = 426

Excluded (n = 178): – less than 18 years old, 4 – timing exclusion criteri, 14 – proximal fracture, 97 – other lower extremity injury, 4 – pathological fracture, 2 – distal fracture, 15 – additional injury requiring admission, 7 – open fracture, 5 – clinical infection at surgical site, 1 – prior surgery involving affected femur, 1 – other, 10 – non-union, 18 Patients enrolled n = 248 Excluded Missed 1-year follow-up n = 61 Patients with 1-year follow-up n = 187 SIGN nail n = 55

Skeletal traction n = 132 Converted to SIGN nail n = 40

Skeletal traction only n = 92

Figure 1. Flow chart showing eligibility, exclusion, enrolment and loss to follow-up of patients.

Patients and methods Study setting and patient enrolment This is a prospective multicenter observational study where adult patients aged 18 years and older, with isolated unilateral femoral shaft fractures (AO/OTA class 32) in 6 hospitals in Malawi, were enrolled from March 2016 to July 2018. Patients with associated major injuries, pathological or open fractures, infection at the surgical site, or prior surgery involving the affected femur were excluded (Figure 1). The type of treatment (ST or IMN) was determined by the treating orthopedic clinical officer (OCOs) or surgeon. OCOs are non-physician clinicians trained to provide nonoperative care for orthopedic conditions and emergency orthopedic surgery for selected cases, such as acute infections and open fractures (Mkandawire et al. 2008). The patients were recruited from Queen Elizabeth Central Hospital (QECH), Kamuzu Central Hospital (KCH), Beit Cure International Hospital (BCIH), and 3 district hospitals: Chiradzulu, Thyolo, and Chikwawa. In both QECH and KCH, patients with femoral shaft fractures were treated with ST or IMN based on the treating clinician’s assess-

725

ment, which was based largely on surgical capacity of the hospital at that time. In the district hospitals all patients were treated by ST. IMN patients who met the inclusion criteria were recruited into the study if they had surgery within 6 weeks from the time of injury. ST patients either continued with skeletal traction until clinical and radiological signs of fracture union were present or were offered IMN if, in the opinion of the treating clinician, union was unlikely without further intervention. The diagnosis of delayed union was made by the treating clinician, if at 6 weeks or more postinjury there was still tenderness and mobility at the fracture site, and no radiological evidence of callus formation. Nonunion was defined as no evidence of fracture healing both clinically and radiologically after at least 3 months on ST or 6 months after IMN. Consequently, the ST group had 2 subgroups: those who started with skeletal traction but later converted to IMN because of either delayed union or nonunion and those who had skeletal traction as definitive treatment until union. A sample size of 110 patients in each group was initially calculated using OpenEpi software (www.openepi. com) (Sullivan et al. 2009) at 95% confidence interval and 80% power using the minimal clinically important difference (MCID) (Jaeschke et al. 1989) of 0.1 between the 2 groups for the EQ-5D, with a standard deviation of 0.12 (Luo et al. 2010, Ibrahim et al. 2018) and a more conservative standard deviation of 0.2 was used for the ST group. The calculation was adjusted to account for 20% loss to follow-up. However, at the 1-year interim analysis there were 65 patients in the IMN group and 120 patients in the ST group. A new sample size was calculated with an allocation ratio of 2:1, resulting in a required sample size of 80 cases in the IMN group and 160 patients in the ST group. Treatment The SIGN nail was used in all IMN patients. This is a solid locking IM nail that can be inserted without need for a fracture table or intraoperative fluoroscopy. At KCH and QECH, the SIGN nail was inserted using open reduction on a standard operating table. At BCIH, fluoroscopy guidance was used. All ST patients had straight leg extension skeletal traction with a Steinmann pin inserted into the proximal tibia under local anesthesia, using an aseptic technique. A stirrup, rope and weights assembly was hung over a bar, pulley, or directly over the end of the bed. Counter-traction and anti-rotating mechanisms were used at the treating clinician’s discretion. Pin site care was performed daily by the patients’ guardians. All patients received physiotherapy by either the hospitals’ physiotherapists or rehabilitation technician. Outcomes The primary outcomes were quality of life determined by European Quality of Life 5-Dimensions Survey (EQ-5D-3L) index score (Brooks and Group 1996) and the Short Musculoskeletal Functional Assessment (SMFA) Function and Both-

726

ersome index scores (Swiontkowski et al. 2005). Both tools have been translated to Chichewa and validated in Malawi (Chokotho et al. 2017, 2019). Both tools were administered verbally by the research assistants who recorded the responses on Microsoft surface computers. Index utility scores for the EQ-5D-3L were generated using the value set for the Zimbabwean population (Jelsma et al. 2003). At each follow-up, patients were asked if they had returned to their pre-injury work, whether employed or otherwise. No specification was made as to whether the patients did not return to work because of the injury or because they were laid off due to injury-related absenteeism. Follow-up Follow-up assessments were performed 6 weeks, 3 months, 6 months, and 1 year after injury. If patients missed scheduled appointments, a telephone interview to answer the EQ-5D-3L and SMFA questionnaires was undertaken. Patients who failed to come for an appointment and were not reached by phone were assessed by research assistants in their homes. Patients who could not be contacted by telephone and could not be found in person were regarded as lost to follow-up. Statistics Data were collected using RedCap electronic data capture tools hosted at the University of California San Francisco (UCSF) (Harris et al. 2009). Data were analyzed using Stata version 10.0 (StataCorp, College Station, TX, USA). Unadjusted analysis was done between IMN and ST groups using Satterthwaite’s t-test for means with unequal variances. Subgroup analysis was also done between the IMN group and successful ST patients. Potential confounders associated (not necessarily causally related) with the outcome were first identified in a univariate regression analysis. Marital status, mechanism of injury, and education level were identified as significantly associated with both the EQ-5D and SMFA scores. The potential confounders and other independent variables were then added in a generalized linear regression model using the forward stepwise regression approach to come up with a final model. Comparison of categorical data was done using a chi-square test, or Fisher’s exact test when any expected cell frequency was less than 5. Listwise deletion of missing data was used in unadjusted and adjusted regression analysis. Findings were considered statistically significant if the p-value was less than 0.05, thus “significant” results refers to statistical significance. Clinical significance is presented using MCID. Estimates were presented with their 95% confidence intervals (CI). Ethics, funding, and potential conflicts of interest The study was approved by the College of Medicine Research Ethics Committee, in Malawi, and the University of Bergen

Acta Orthopaedica 2020; 91 (6): 724–731

and University of California San Francisco Institutional Review Boards. Written informed consent was obtained from all patients in the study. The study was funded by the Institute of Global Orthopedics and Traumatology (IGOT), University of California San Francisco, James O. Johnston Research Grant, and a PhD grant through the Norhed Project, financed by Norad. Author DS is a non-paid member of the Board of Directors for SIGN Fracture Care International. The rest of the authors declare no conflict of interest.

Results There were 426 eligible cases, of which 248 were enrolled in the study. 1-year follow up was achieved in 187 cases (75%) (Figure 1). 55 and 132 cases were treated with IMN and ST respectively. Baseline demographic and injury details The mean age of patients was 38 (SD 13) years in the IMN group and 40 (SD 16) years in the ST group (Table 1). In both groups the majority of patients were male. The most common cause of injury was road traffic injury followed by falls. More people in the ST group had primary school as their highest level of education, whereas there were more people with postsecondary education in the IMN group (p < 0.001). Most fractures were AO/OTA type 32A, but there were more type 32B in the IMN group than in the ST group, p = 0.02 (Table 1). Treatment The mean waiting time from injury to definitive treatment was 13 (SD 12) days for the IMN group and 4.4 (SD 5) days for the ST group, p < 0.001 (Table 1). 1 patient in the IMN group had a nonunion and was treated with an exchange nail, whereas 40 patients (30%) in the ST group had either nonunion or delayed union and subsequently converted to IMN during the course of the study (p < 0.001). Details on duration from time of injury to conversion were available for 20 patients out of 40, with a median of 63 days and a range of 50 to 252 days. Quality of life IMN versus all ST patients The unadjusted mean EQ-5D index scores were higher in the IMN group than ST group at 6 weeks (p = 0.03) and 3 months (p = 0.03) after injury (Figure 2) but not at 6 months and 1 year. The mean EQ-5D index scores were lower at 1-year post injury compared with baseline, (p < 0.001). Patients in the IMN group reported significantly better quality of life than those in the ST group at 6 weeks, 3 months, and 6 months after the injury, with an adjusted mean difference of –0.14 (CI –0.27 to –0.02); –0.07 (CI –0.14 to –0.0001); –0.08 (CI –0.15 to –0.01) respectively. The mean difference was greater than MCID at 6 weeks and equal to MCID at 3 months and 6 months (Table 3).

Acta Orthopaedica 2020; 91 (6): 724–731

727

Table 1. Baseline details

Unadjusted mean (CI) EQ-5D scores 1.2

Successful IM All skeletal traction nailing traction Convert only Variable n = 55 n = 132 p n = 40 n = 92 Age, mean (SD) 38 (13) 40 (16) 0.3 median 37 37 IQR 28–45 26–48 Sex, n (%) 0.7 Female 12 22 (17) Male 42 107 (81) Missing 1 3 (2) Marital status 0.8 Single 16 39 (30) Married 36 79 (60) Divorced/separated 1 5 (3.8) Widow/widower 2 7 (5) Missing 0 2 (2) Education < 0.001 Primary 13 76 (58) Secondary 18 40 (30) Post-secondary 22 12 (9) Missing 0 4 (3) Mechanism of injury 0.4 Fall 13 45 (34) RTI 37 68 (52) Other 4 16 (12) Missing 1 3 (2) Smoking 0.3 No 52 112 (85) Yes 2 13 (10) Missing 1 7 (5) OTA classification 0.02 A (simple) 37 97 (74) B (wedge) 13 15 (11) C (complex) 4 5 (4) Missing 1 15 (11) OTA 32A subclass 0.06 Oblique 10 11 (8) Spiral 6 16 (12) Transverse 18 67 (51) Missing 21 38 (29) Location 0.4 Distal zone 3 16 (12) Middle zone 35 82 (62) Subtrochanteric 9 14 (11) Missing 8 20 (15) Side of injury 0.7 Right 29 70 (53) Left 23 53 (40) Missing 3 9 (7) Duration before treatment < 0.001 mean (SD) 13 (12) 4.4 (5) median 10 3 IQR 3–18 1–6

p = 0.03 p = 0.03

1.0

0.8

37 (14) 41 (17) 0.6

0.4

6 33 1

16 74 2

10 26 2 1 1

29 53 3 6 1

16 18 5 1

60 22 7 3

12 24 2 2

33 44 14 1

33 4 3

79 9 4

31 5 1 3

66 10 4 12

4 7 20 9

7 9 47 29

6 27 3 4

10 55 11 16

22 14 4

48 39 5

6 (6) 3 1–8

5 (12) 3 1–5

0.2

Skeletal traction Intramedullary nail

Baseline 6 weeks 3 months 6 months 1 year

Figure 2. Unadjusted mean EQ-5D scores for IM nailing vs. skeletal traction. Unadjusted mean (CI) SMFA FI scores 60

p = 0.005 p = 0.004 p = 0.02

Skeletal traction Intramedullary nail

Baseline 6 weeks 3 months 6 months 1 year

Figure 3. Unadjusted mean SMFA Functional Index scores for IM nailing vs. skeletal traction. Unadjusted mean (CI) SMFA BI scores 60

p = 0.01

Skeletal traction Intramedullary nail

Baseline 6 weeks 3 months 6 months 1 year

Successful skeletal traction versus IMN There were no significant differences in the unadjusted and adjusted mean EQ-5D index scores between patients who were treated successfully with ST (without converting to IMN) and those patients who were treated primarily with IMN (Tables 2 and 3). However, the adjusted mean difference in index scores was similar to MCID at the 6 weeks (–0.09, CI –0.2 to 0.06) and 3 months intervals (–0.07, CI –0.2 to 0.03) (Table 3).

Figure 4. Unadjusted mean SMFA Bothersome Index for IM nailing vs. skeletal traction.

Functional status IMN versus all ST patients Both unadjusted and adjusted analyses showed significantly lower mean SMFA functional index scores at 6 weeks, and 3 and 6 months post-injury in the IMN group, indicating better function compared with the ST group (Figure 3 and Table 3).

728

Acta Orthopaedica 2020; 91 (6): 724–731

Table 2. Unadjusted results for sub-group analysis. Values are mean (CI) Variable

Pre-injury/baseline

6 weeks

3 months

EQ-5D Successful ST 0.99 (0.98–1) 0.40 (0.31–0.49) 0.64 (0.50–0.73) IMN 0.95 (0.92–0.99) 0.50 (0.42–0.59) 0.72 (0.68–0.77) SMFA FI Successful ST 1.5 (1.0–2.0) 52 (48–57) a 36 (29–42) a IMN 2.5 (0.8–4.1) 43 (38–47) 27 (23–31) SMFA BI Successful ST 0 48 (43–54) a 30 (24–37) IMN 1 (–0.4 to 2) 39 (34–44) 24 (19–29) a statistically significant (p < 0.05) SMFA FI, SMFA Function Index. SMFA BI, SMFA Bothersome Index

6 months

1 year

0.80 (0.74–0.86) 0.85 (0.78–0.91)

0.91 (0.88–0.93) 0.91 (0.87–0.95)

23 (18–28) a 16 (11–20)

6.7 (4.9– 8.5) 9.3 (5.7–13)

18 (13–23) 13 (7.9–18)

6.3 (4.1–8.4) 7.6 (3.6–12)

Table 3. Adjusted results Variable

Pre-injury/baseline coefficient (CI)

ST vs. IMN EQ5D score 0.03 (–0.004 to 0.1) p-value 0.1 SMFA FI –1.0 (–2.5 to 0.6) p-value 0.2 SMFA BI –0.5 (–1.9 to 0.9) p-value 0.5 IMN vs. successful ST EQ5D score 0.03 (–0.002 to 0.1) p-value 0.1 SMFA FI –1.1 (–2.6 to 0.5) p-value 0.2 SMFA BI –0.9 (–2.2 to 0.4) p-value 0.2

6 weeks coefficient (CI)

3 months coefficient (CI)

6 months coefficient (CI)

–0.14 (–0.27 to –0.02) 0.03 8.7 (2.6 to 15) 0.01 9.2 (2.4 to 16) 0.01

–0.07 (–0.14 to –0.0001) –0.08 (–0.15 to –0.01) 0.05 0.04 8.4 (2.6 to 14) 7.9 (1.7 to 14) 0.01 0.01 7.7 (1.2 to 14) 6.7 (–0.3 to 14) 0.02 0.1

–0.09 (–0.2 to 0.06) 0.2 8.5 (1.8 to 15) 0.01 8.8 (0.9 to 17) 0.03

–0.07(–0.2 to 0.03) 0.2 7.6 (0.4 to 15) 0.04 5.5 (–2.5 to 14) 0.2

Further, the unadjusted mean SMFA Bothersome index was significantly lower in the IMN group compared with the ST group at 6 weeks and 3 months post-injury, indicating that patients in the IMN group were less bothered by their condition (Figure 4). Adjusted analysis showed a similar trend with mean difference in the SMFA Bothersome index of 9.2 (CI 2.4–16) at 6 weeks and 7.7 (CI 1.2–14) at 3 months (Table 3). Successful skeletal traction vs. IMN The mean SMFA functional index scores were significantly lower in the IMN group compared to the successful ST group at 6 weeks, and 3 and 6 months post-injury for both unadjusted (Table 2) and adjusted analysis (8.5, CI 1.8–15; 7.6, CI 0.4–15; 7.2, CI 0.4–14), (Table 3). The unadjusted and adjusted mean SMFA Bothersome index scores were significantly lower in the IMN group compared with the successful ST group at 6 weeks (Tables 2 and 3). Return to work 88 of 103 cases followed up at 6 months responded to the question of whether they had returned to work. No reasons

1 year coefficient (CI) 0.001 (–0.05 to 0.05) 1 –2. (–5.8 to 1.7) 0.3 –1.2 (–5.4 to 2.9) 0.6

–0.05 (–0.14 to 0.03) –0.0001 (–0.05 to 0.05) 0.2 1 7.2 (0.4 to 14) –2.4 (–6.3 to 1.5) 0.04 0.2 4.1(–3.6 to 12) –1.2 (–5.6 to 3.1) 0.3 0.6

were specified for non-response to this question in the remaining 15 cases (9 in the IM group and 6 in the ST group). 24 of 51cases in the ST group had returned to work compared with 26 of 37 in the IMN group (p = 0.02). There were no significant differences in proportions of patients who had returned to work at the other follow-up time points.

Discussion This study found improved quality of life and function up to 6 months post-injury for IMN compared with ST in patients treated for femoral shaft fractures in Malawi. Almost one-third of patients treated with ST failed treatment and were ultimately converted to IMN due to delayed union or nonunion, typically between 6 and 12 weeks after initiating traction. Nonetheless, patients achieving union with skeletal traction had equivalent outcomes to those treated with early IMN at 1 year. As far as we know, this is the first study comparing quality of life and functional status in femoral shaft fracture patients treated with ST or IMN. Haug et al. (2017) looked at quality

Acta Orthopaedica 2020; 91 (6): 724–731

of life in femoral shaft fracture patients treated with skeletal traction and found that patients had both physical and psychological pain as well as emotional distress due to prolonged hospitalization and the associated negative economic impact on their families. Tay et al. (2014) found that patients with long bone diaphyseal fractures treated surgically still had residual physical impairment and pain in the first year postinjury, which was worse among those with delayed union and nonunion even after treatment. Ibrahim et al. (2018) also found that EQ-5D scores did not return to the pre-injury level after operative treatment of femoral shaft fractures, a finding that was also replicated in our study. These studies support the concept that long bone fractures affect long-term quality of life and functional status even after operative treatment. Patients treated with skeletal traction are normally admitted to hospital for at least 6 weeks, which is likely to have substantial financial implications for the patients, their guardians, and the health service providers. In our study, less than half of the ST patients had returned to work at 6 months after the injury compared with approximately three-quarters in the IMN group. The direct and indirect costs associated with skeletal traction may be more than the cost of intramedullary nailing. A cost-effectiveness study of the 2 treatment modalities is needed to give a complete picture of the impact of the treatment modalities and the findings could assist in better priority setting and resource allocation. One-third of the ST patients were converted to IMN due to either delayed union or nonunion. These findings highlight the unmet need for operative fracture treatment in Malawi, where patients are offered operative treatment mostly after failure of primary nonoperative treatment, despite clear evidence in the literature that operative treatment is superior (Brumback et al. 2006, Kamau et al. 2014, Chagomerana et al. 2017). Femoral shaft nonunion is incapacitating and its impact on healthrelated quality of life is comparable to severe hip osteoarthritis and worse than medical conditions such as myocardial infarction and congestive cardiac failure (Brinker et al. 2017). In addition, nonunion surgery is more complex than acute fracture surgery and has an increased risk of infection and other complications (Mahomed 2008, Young et al. 2013b), and also has the potential to use more resources. Efforts should therefore be made to improve surgical services and avert the problem of converting to IMN after failed skeletal traction. Conducting clinical research in low-resource settings presents many challenges, and our study has several limitations. First, the IMN group was not homogeneous. The delay from time of injury to treatment ranged from 1 day to 6 weeks, signifying the challenges faced by orthopedic surgeons in Malawi to provide operative fracture care in a setting where theatre time is limited, and the few available specialists are overwhelmed by the large burden of fractures needing surgery. This baseline discrepancy may have contributed to suboptimal quality of life and function in the IMN group, as early operative stabilization of these fractures is associated with

729

fewer complications and better outcomes in the short term (Mahomed 2008, El-Menyar et al. 2018). Lack of homogeneity also limits its external validity. Another limitation is that there was a high rate of conversion from ST to IMN due to either delayed union or nonunion. This occurred after at least 6 weeks on skeletal traction, and as a result there was no bias at 6 weeks. However, the remaining time points were likely biased towards the null hypothesis of no difference between groups because those patients who failed traction would have experienced a poor outcome had they continued with ST for the entire follow-up period. Details on post-treatment physiotherapy, which plays a crucial role in improving function after injury (Paterno and Archdeacon 2009), were not collected. However, patients in both groups were provided with standard rehabilitation by either the hospitals’ physiotherapists or rehabilitation technicians. Thus it is unlikely that post-treatment rehabilitation affected the functional outcome in 1 group more than the other. We also did not collect detailed information on comorbidities. However, as the mean age in both groups was less than 40 years it is unlikely that there were patients with substantial comorbidities. Loss to follow-up at the different time intervals may have reduced the power of the study to detect a statistically significant difference. Nonetheless, the differences were significant at early time points, and the mean difference found at 1 year was far below the MCID for the EQ-5D. Loss to follow-up also causes uncertainty with regard to the true effect of the treatment modalities, due to unknown outcomes of those who missed follow-up. However, Young et al. (2013b) found that the majority of the femoral shaft fracture patients in Malawi who did not return to hospital for follow-up were doing well. Another limitation was that there was no standard definition of delayed union and nonunion in the study’s facilities. As most patients routinely have only one radiographic view, either anteroposterior (AP) or lateral, it was not possible to use standard scoring systems such as the RUST Score (Whelan et al. 2010) or the criteria used by Tsang et al. (2016). Finally, because patients’ assignment to the 2 study groups was not randomized, there is a potential for confounding due to unmeasured baseline characteristics. Further, regression models may not adequately control for confounding (Shrier and Platt 2008). However, since only confounders measured at baseline were included, we argue that none of these can be colliders in the analysis. Nevertheless, this prospective observational study is the first to compare the quality of life and functional status of femoral shaft fractures treated with either an intramedullary nail or skeletal traction in a LIC. In conclusion, this study found that treatment with IMN improved early (≤ 6 months) postoperative quality of life and function and allowed patients to return to work earlier compared with those treated with ST. Treatment of femoral shaft fractures with ST in a resource-limited setting may achieve similar outcomes to IMN in quality of life and function at 1-year post-injury if fracture union is achieved. However, approximately 1 in every 3 patients treated with straight-leg

730

ST failed treatment, requiring conversion to surgical treatment. There is a need for a cost-effectiveness study comparing these 2 treatment modalities to gain a broader picture of the impact of treatment for femoral shaft fractures in low-resource settings.

LC designed the concept of study, supervised and monitored data collection, analyzed the data, and drafted the manuscript. BL provided input on the concept of the study, supervised and monitored data collection, and helped with database design and critical revision of the manuscript. HHW set up the database, and supervised and monitored data collection, and critical revision of the manuscript. DS provided input on the concept of the study, monitored data collection, and undertook critical revision of the manuscript. NM, JEG, GH, and SY provided input on the concept of the study, and undertook critical revision of the manuscript. The authors would like to thank all the orthopedic surgeons and orthopedic clinical officers for their support during the study. They would also like to thank the study’s physiotherapists who did the clinical assessments. Special gratitude is due to Mr Foster Mbomuwa, the project’s coordinator, for all his efforts during the study. Acta thanks Simon Graham for help with peer review of this study.

Brinker M R, Trivedi A, O’Connor D P. Debilitating effects of femoral nonunion on health-related quality of life. J Orthop Trauma 2017; 31(2): e37e42. Brooks R, Group E. EuroQol: the current state of play. Health Policy 1996; 37(1): 53-72. Brumback R J, Uwagie-Ero S, Lakatos R P, Poka A, Bathon G H, Burgess A R. Intramedullary nailing of femoral shaft fractures, Part II: Fracturehealing with static interlocking fixation. Orthop Trauma Dir 2006; 4(01): 29-33. Bucholz R W, Jones A. Fractures of the shaft of the femur. J Bone Joint Surg 1991; 73(10): 1561-66. Chagomerana M B, Tomlinson J, Young S, Hosseinipour M C, Banza L, Lee C N. High morbidity and mortality after lower extremity injuries in Malawi: a prospective cohort study of 905 patients. Int J Surg 2017; 39: 23-9. Chokotho L, Mkandawire N, Conway D, Wu H-H, Shearer D D, Hallan G, Gjertsen J E, Young S, Lau B C. Validation and reliability of the Chichewa translation of the EQ-5D quality of life questionnaire in adults with orthopedic injuries in Malawi. Malawi Med J 2017; 29(2): 84-8. Chokotho L, Lau B C, Conway D, Wu H, Shearer D, Hallan G, Gjertsen J E, Mkandawire N, Young S. Validation of Chichewa Short Musculoskeletal Function Assessment (SMFA) questionnaire: a cross-sectional study. Malawi Med J 2019; 31(1): 65-70. Doorgakant A, Mkandawire N C. The management of isolated closed femoral shaft fractures in a district hospital in Malawi. Trop Doct 2012; 42(1): 8-12. El-Menyar A, Muneer M, Samson D, Al-Thani H, Alobaidi A, Mussleman P, Latifi R. Early versus late intramedullary nailing for traumatic femur fracture management: meta-analysis. J Orthop Surg Res 2018; 13(1): 160. El Moumni M, Leenhouts P A, Ten Duis H J, Wendt K W. The incidence of non-union following unreamed intramedullary nailing of femoral shaft fractures. Injury 2009; 40(2): 205-8. Harris P A, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap): a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009; 42(2): 377-81.

Acta Orthopaedica 2020; 91 (6): 724–731

Haug L, Wazakili M, Young S, Van den Bergh G. Longstanding pain and social strain: patients’ and health care providers’ experiences with fracture management by skeletal traction; a qualitative study from Malawi. Disabil Rehabil 2017; 39(17): 1714-21. Hollis A C, Ebbs S R, Mandari F N. The epidemiology and treatment of femur fractures at a northern Tanzanian referral centre. Pan Afr Med 2015; 22(1): 338. Ibrahim J M, Conway D, Haonga B T, Eliezer E N, Morshed S, Shearer D W. Predictors of lower health-related quality of life after operative repair of diaphyseal femur fractures in a low-resource setting. Injury 2018; 49(7): 1330-5. Jaeschke R, Singer J, Guyatt G H. Measurement of health status: ascertaining the minimal clinically important difference. Control Clin Trials 1989; 10(4): 407-15. Jelsma J, Hansen K, De Weerdt W, De Cock P, Kind P. How do Zimbabweans value health states? Popul Health Metr 2003; 1(1): 11. Kamau D M, Gakuu L N, Gakuya E M, Sang E K. Comparison of closed femur fracture: skeletal traction and intramedullary nailing cost-effectiveness. East Afr Orthop J 2014; 8(1): 4-9. Kramer E J, Shearer D, Morshed S. The use of traction for treating femoral shaft fractures in low- and middle-income countries: a systematic review. Int Orthop 2016; 40(5): 875-83. Luo N, Johnson J A, Coons S J. Using instrument-defined health state transitions to estimate minimally important differences for four preference-based health-related quality of life instruments. Med Care 2010; April 1: 365-71. Mahomed H. Late nailing of femoral shaft fractures. Orthopedic Proceedings, the British Editorial Society of Bone & Joint Surgery 2008; 90(SUPP_III): 464. Mkandawire N, Ngulube C, Lavy C. Orthopedic clinical officer program in Malawi: a model for providing orthopedic care. Clin Orthop Relat Res 2008; 466(10): 2385-91. Parkes R J, Parkes G, James K. A systematic review of cost-effectiveness, comparing traction to intramedullary nailing of femoral shaft fractures, in the less economically developed context. BMJ Glob Health 2017; 2(3): e000313. Paterno M V, Archdeacon M T. Is there a standard rehabilitation protocol after femoral intramedullary nailing? J Orthop Trauma 2009; 23: S39-S46. Ricci W M, Bellabarba C, Evanoff B, Herscovici D, DiPasquale T, Sanders R. Retrograde versus antegrade nailing of femoral shaft fractures. J Orthop Trauma 2001; 15(3): 161-9. Salawu O N, Ibraheem G H, Babalola O M, Kadir D M, Ahmed B A, Agaja S B, Olawepo A, Nasir A A. Clinical outcomes after open locked intramedullary nailing of closed femoral shaft fractures for adult patients in a Nigerian Hospital. Niger J Clin Pract 2017; 20(11): 1316-21. Shah R K, Moehring H D, Singh R P, Dhakal A. Surgical Implant Generation Network (SIGN) intramedullary nailing of open fractures of the tibia. Int Orthop 2004; 28(3): 163-6. Shrier I, Platt R W. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008; 8(1): 70. Sullivan K, Dean A, Soe M. OpenEpi: a web-based epidemiologic and statistical calculator for public health. Public Health Rep 2009; 124(3): 471-4. Swai S P M. Conservative management of closed femoral shaft fractures: is skeletal traction still justified or cost effective in a developing country like Tanzania? (Doctoral diss., Muhimbili University of Health and Allied Sciences; 2005). Swiontkowski M F, Engelberg R U, Martin D P, Agel J. Short musculoskeletal function assessment questionnaire: validity, reliability, and responsiveness. Orthop Trauma Dir 2005; 3(02): 29-34. Tay W H, de Steiger R, Richardson M, Gruen R, Balogh Z J. Health outcomes of delayed union and nonunion of femoral and tibial shaft fractures. Injury 2014; 45(10): 1653-8. Tsang S T, Mills L A, Frantzias J, Baren J P, Keating J F, Simpson A H. Exchange nailing for nonunion of diaphyseal fractures of the tibia: our results and an analysis of the risk factors for failure. Bone Joint J 2016; 98(4): 534-41.

Acta Orthopaedica 2020; 91 (6): 724–731

Whelan D B, Bhandari M, Stephen D, Kreder H, McKee M D, Zdero R, Schemitsch EH. Development of the radiographic union score for tibial fractures for the assessment of tibial fracture healing after intramedullary fixation. J Trauma Acute Care Surg 2010; 68(3): 629-32. Young S, Lie S A, Hallan G, Zirkle L G, Engesæter L B, Havelin L I. Risk factors for infection after 46,113 intramedullary nail operations in low-and middle-income countries. World J Surg 2013a; 37(2): 349-55.

731

Young S, Banza L N, Hallan G, Beniyasi F, Manda K G, Munthali B S, Dybvik E, Engesæter L B, Havelin L I. Complications after intramedullary nailing of femoral fractures in a low-income country: a prospective study of follow-up, HIV infection, and microbial infection rates after IM nailing of 141 femoral fractures at a central hospital in Malawi. Acta Orthop Scand 2013b; 84(5): 460-7

732

Acta Orthopaedica 2020; 91 (6): 732–737

Predicting the mechanical hip–knee–ankle angle accurately from standard knee radiographs: a cross-validation experiment in 100 patients Willem Paul GIELIS 1, Hassan RAYEGAN 2, Vahid ARBABI 1–3, Seyed Y AHMADI BROOGHANI 2, Claudia LINDNER 4, Tim F COOTES 4, Pim A de JONG 5, H WEINANS 1, and Roel J H CUSTERS 1 1 Department

of Orthopedic Surgery, UMC Utrecht, Utrecht, The Netherlands; 2 Department of Mechanical Engineering, Faculty of Engineering, University of Birjand, Birjand, Iran; 3 Department of Biomechanical Engineering, Faculty of Mechanical, Maritime, and Materials Engineering, Delft University of Technology (TU Delft), Delft, The Netherlands; 4 Division of Informatics, Imaging & Data Sciences, University of Manchester, Manchester, UK; 5 Department of Radiology, UMC Utrecht and Utrecht University, Utrecht, The Netherlands Correspondence: w.p.gielis@umcutrecht.nl Submitted 2020-04-02. Accepted 2020-05-06.

Background and purpose — Being able to predict the hip–knee–ankle angle (HKAA) from standard knee radiographs allows studies on malalignment in cohorts lacking full-limb radiography. We aimed to develop an automated image analysis pipeline to measure the femoro-tibial angle (FTA) from standard knee radiographs and test various FTA definitions to predict the HKAA. Patients and methods — We included 110 pairs of standard knee and full-limb radiographs. Automatic search algorithms found anatomic landmarks on standard knee radiographs. Based on these landmarks, the FTA was automatically calculated according to 9 different definitions (6 described in the literature and 3 newly developed). Pearson and intra-class correlation coefficient [ICC]) were determined between the FTA and HKAA as measured on full-limb radiographs. Subsequently, the top 4 FTA definitions were used to predict the HKAA in a 5-fold cross-validation setting. Results — Across all pairs of images, the Pearson correlations between FTA and HKAA ranged between 0.83 and 0.90. The ICC values from 0.83 to 0.90. In the crossvalidation experiments to predict the HKAA, these values decreased only minimally. The mean absolute error for the best method to predict the HKAA from standard knee radiographs was 1.8° (SD 1.3). Interpretation — We showed that the HKAA can be automatically predicted from standard knee radiographs with fair accuracy and high correlation compared with the true HKAA. Therefore, this method enables research of the relationship between malalignment and knee pathology in large (epidemiological) studies lacking full-limb radiography.

The mechanical axis of the lower limb, which determines knee (mal)alignment, is historically measured using the hip–knee– ankle-angle (HKAA), an angle between the mechanical axes of the femur and tibia. The femoral axis runs through the centers of the femoral head and knee joint. The tibial axis runs through the centers of the knee and ankle joints (Figure 1). A standard knee radiograph is one of the primary tools in the diagnostic process of knee complaints and it is undertaken for a majority of patients. Correspondingly, many epidemiological studies focusing on the knee include standard knee radiographs. However, to verify and measure involvement of malalignment in the pathophysiology, the HKAA should be measured. This requires a full-limb radiograph (Figure 1). Compared with a standard knee radiograph, full-limb radiography involves higher costs, the need for specialized equipment, and a larger effective radiation dose for the patient. These are important reasons for knee OA cohort studies not to include full-limb radiographs. As standard knee radiographs are available for the majority of patients with knee complaints and participants of epidemiological knee-focused studies, it is desirable to have a method for defining knee (mal)alignment from a standard knee radiograph. The femoro-tibial angle (FTA), an angle between the anatomic axes of the femur and tibia (Figure 2), can be used to predict the mechanical axis from a standard knee radiograph. The FTA is an important measurement that can predict the development of knee OA (Brouwer et al. 2007, Moyer et al. 2016). Multiple definitions for the FTA have been proposed (Table 1) (Kraus et al. 2005, Hinman et al. 2006, Brouwer et al. 2007, Issa et al. 2007, Colebatch et al. 2009, Felson et al.

Acta Orthopaedica 2020; 91 (6): 732–737

733

Fem1

Fem2 Tib1

Figure 1. Measurement of the Hip–knee–ankle angle on full limb radiograph. The hip–knee–ankle angle (HKAA, in green) is measured between 2 axes (in red). One axis runs from the middle of the femoral head to the middle of the femoral notch, and a second axis from the middle of the tibial notch to the middle of the talar head.

Fem3 Tib2

Fem4 Tib3

Tib4

Figure 2. Measurement of the femoro-tibial angle (FTA) on standard knee radiographs. Definitions of the femoral axis from left to right (top): Fem1—mid-shaft at approximately 10 cm proximal of the femoral notch + mid-shaft in the area where the meta- and epiphysis meet (van Raaij et al. 2009, Iranpour-Boroujeni et al. 2014, Zampogna et al. 2015). Fem2—mid-shaft at approximately 10 cm proximal of the femoral notch + center of the femoral notch (Felson et al. 2009, van Raaij et al. 2009, McDaniel et al. 2010, Sheehy et al. 2011). Fem3—mid-shaft at approximately 10 cm proximal of the femoral notch + base of the tibial spines (Kraus et al. 2005, Hinman et al. 2006, Issa et al. 2007, McDaniel et al. 2010, Navali et al. 2012, Iranpour-Boroujeni et al. 2014, Zampogna et al. 2015). Fem4—mid-shaft at approximately 10 cm proximal of the femoral notch + middle of tibial plateau (McDaniel et al. 2010). Definitions of the tibial axis from left to right (bottom): Tib1—mid-shaft at approximately 10 cm distal of the base of the tibial spines + mid-shaft in the area where the meta- and epiphysis meet (van Raaij et al. 2009, Iranpour-Boroujeni et al. 2014, Zampogna et al. 2015). Tib2 Mid-shaft at approximately 10 cm distal of the base of the tibial spines + center of the femoral notch (McDaniel et al. 2010). Tib3—mid-shaft at approximately 10 cm distal of the base of the tibial spines + base of the tibial spines (Kraus et al. 2005, Hinman et al. 2006, Issa et al. 2007, Colebatch et al. 2009, van Raaij et al. 2009, McDaniel et al. 2010, Sheehy et al. 2011, Navali et al. 2012, Iranpour-Boroujeni et al. 2014, Zampogna et al. 2015). Tib4—mid-shaft at approximately 10 cm distal of the base of the tibial spines + middle of tibial plateau (McDaniel et al. 2010). The 2 pictures on the left show the measurement of the FTA using method 2 for the femoral axis and method 1 for the tibial axis on a standard AP knee radiograph from the present data set.

2009, van Raaij et al. 2009, McDaniel et al. 2010, Sheehy et al. 2011, Navali et al. 2012, Zampogna et al. 2015). However, a direct comparison between all FTA definitions on the same data is lacking. Additionally, no studies used cross- or external validation to confirm results. As such, there is no consensus on which FTA definition should be used to predict the HKAA. In addition to morphological measurements, statistical shape modelling is used in OA research to quantify variation in joint shape. Multiple studies showed the shape of a joint is a major factor in the incidence and progression of OA (Haverkamp et al. 2011, Waarsing et al. 2011, Agricola et al. 2015). A key step in the statistical shape modelling process is to outline the structures of interest in the medical images (e.g., radiographs) using anatomical landmarks. Manually placed

landmarks on a set of radiographs can be used to train automated search models to place the respective points on new unseen images automatically, paving the road to analyzing large datasets (Lindner et al. 2013a, 2013b). Furthermore, the landmark positions obtained by the search models can easily be used to calculate morphologic measurement such as joint space width or the FTA. The ability to predict the HKAA using automated FTA measurements from standard knee radiographs would make studies on malalignment feasible in large cohorts that lack full-limb radiography. This study aimed (i) to develop an automated image analysis pipeline to measure the FTA from a standard knee radiograph; and (ii) to analyze the performance of various FTA definitions in predicting the HKAA as measured on a full-limb radiograph.

734

Patients and methods Patients We included 100 full-limb (50 males) radiographs, acquired for clinical care at the department of Orthopaedic Surgery of the UMC Utrecht, the Netherlands, in a consecutive series between March and November 2017. All patients were 40 years or older. For inclusion at least 1 standard knee radiograph made on the same day had to be available. We excluded patients with femoral and tibial deformities due to fractures, surgeries (including joint replacement of knee, hip, or ankle, and osteotomies) and developmental disorders. When radiographs of both legs were available for 1 subject, both were included in the study. Radiographic acquisition Weight-bearing extended anteroposterior full-limb radiographs, with the patella facing straight towards the X-ray tube, were undertaken. On the same day, weight-bearing extended anteroposterior knee radiographs with the patella facing forward were taken. Standard knee radiographs were assessed by WPG for Kellgren–Lawrence (KL) grades. Before the assessment, the rater completed the tutorial for KL grading by Hayes et al. (2016). The tutorial includes 19 cases and an answer sheet to test the effect of the tutorial. The square weighted kappa for inter-rater reliability between WPG and the answer sheet was 0.969. Alignment measurements The mechanical HKAA was used as gold standard and was measured on the full-limb radiographs (Moreland et al. 1987). An axis was drawn from the middle of the femoral head to the center of the femoral notch. A second axis was drawn from the base of the tibial spines to the center of the ankle joint. The HKAA was defined as the angle between these axes (Figure 1) (Sharma et al. 2001). The FTA was measured on standard knee radiographs as the angle between the axis of the femur and tibia. We used a bespoke search model in BoneFinder (www.bone-finder.com, Centre for Imaging Sciences, University of Manchester, UK) to automatically outline the distal femur, patella, and proximal tibia using 111 landmarks (Lindner et al. 2013b). All automatically obtained landmarks were checked and manually corrected if necessary. The identified landmarks were used to automatically calculate the FTA. For measuring the femoral and tibial axes, 4 definitions each were considered based on previous literature (Figure 2) (Kraus et al. 2005, Hinman et al. 2006, Brouwer et al. 2007, Issa et al. 2007, Colebatch et al. 2009, Felson et al. 2009, van Raaij et al. 2009, McDaniel et al. 2010, Sheehy et al. 2011, Navali et al. 2012, Zampogna et al. 2015). 9 combinations between the femoral and tibial axes measurements were used to calculate the FTA (Fem1 + Tib1, Fem1 + Tib3, Fem1 + Tib4, Fem2 + Tib1, Fem2 + Tib2, Fem2

Acta Orthopaedica 2020; 91 (6): 732–737

Table 1. Correlations between femoro-tibial angle and hip–knee– ankle angle as reported in the literature Pearson Landmarks Radiograph correlation Reference Fem3 + Tib3 AP extended 0.26 Zampogna et al. 2015 Fem1 + Tib1 AP extended 0.71 Fem3 + Tib3 AP extended 0.81 Colebatch et al. 2009 Fem2 + Tib2 PA semi-flexed a 0.50 McDaniel et al. 2010 a Fem3 + Tib3 PA semi-flexed 0.65 Fem4 + Tib4 PA semi-flexed a 0.55 Fem3 + Tib3 b PA semi-flexed a 0.64 Fem2 + Tib3 PA semi-flexed a 0.59 Fem3 + Tib3 PA semi-flexed a 0.86 Issa et al. 2007 Fem2 + Tib3 PA semi-flexed a 0.66 Felson et al. 2009 Fem1 + Tib1 c PA semi-flexed a 0.76 Iranpour-Boroujeni et al. 2014 Fem3 + Tib3 PA semi-flexed a 0.68 Fem3 + Tib3 PA semi-flexed a 0.75 Kraus et al. 2005 Fem3 + Tib3 Full-limb 0.65 Fem3 + Tib3 Full-limb 0.88 Hinman et al. 2006 Fem2 + Tib3 Full-limb 0.34 van Raaij et al. 2009 Fem1 + Tib1 Full-limb 0.65 Fem2 + Tib3 Full-limb 0.88 Sheehy et al. 2012 Fem3 + Tib3 Full-limb 0.93 Navali et al. 2012 a Positioning aided with b Slight variation where

Synaflexer frame. the tips of the tibial spines are used instead

of the base. variation where the bottom point at the femur is determined using the middle femoral condyles instead of the shaft.

c Slight

+ Tib3, Fem2 + Tib4, Fem3 + Tib3, Fem4 + Tib4). A varus angle is displayed as a negative number, a valgus angle as a positive number. As the standard knee radiographs were not calibrated for absolute distance, we used the width of the femoral condyles and tibial plateau to place circles needed for FTA measurements at approximately 10 cm from the joint line (Figure 2). Based on data from previous work we used a ratio of 1.52 for the femur and 1.42 for the tibia (Wesseling et al. 2016). We used the center of a circle touching the medial and lateral cortex to determine the mid-shaft points. Statistics We used Pearson correlation coefficients to study which FTA method has the strongest correlation with the HKAA. Across all image pairs, we predicted the HKAA from the FTA using linear regression models. A simple model using only 1 FTA predictor, a model including a quadratic term, and a model including sex were considered. Using these predictions, we calculated 2-way mixed single-measures intra-class correlation coefficients for absolute agreement (ICC) between predicted HKAA and observed HKAA. For the 4 FTA definitions with the strongest correlation to the HKAA, we performed 5-fold cross-validation experiments on the same data set. We randomly distributed the dataset in 5 parts, each accounting for 20% of the cases. In each fold, we calculated a linear regression formula to predict the HKAA based on the FTA using 80% of the data and

Acta Orthopaedica 2020; 91 (6): 732–737

735

Table 2. Pearson correlation coefficients and intra-class correlations (ICC) between FTA and HKAA measurements (across all pairs of images)

Error (°) Male Female

5.0

Method Fem1 + Tib1 Fem1 + Tib3 Fem1 + Tib4 Fem2 + Tib1 Fem2 + Tib2 Fem2 + Tib3 Fem2 + Tib4 Fem3 + Tib3 Fem4 + Tib4

Pearson correlation 0.88 0.86 0.86 0.90 0.87 0.89 0.89 0.84 0.83

ICC (95% CI) 0.87 (0.82–0.91) 0.86 (0.80–0.90) 0.86 (0.80–0.90) 0.90 (0.85–0.93) 0.86 (0.80–0.90) 0.89 (0.84–0.92) 0.89 (0.84–0.92) 0.83 (0.76–0.88) 0.82 (0.74–0.87)

2.5

–2.5

–5.0

–15

Table 3. Pearson correlation coefficients and intra-class correlations (ICC) between FTA and HKAA measurements (cross-validation experiments) Method Pearson correlation ICC (95% CI) Fem1 + Tib1 0.88 0.87 (0.82–0.91) Fem2 + Tib1 0.90 0.90 (0.85–0.93) Fem2 + Tib3 0.89 0.89 (0.84–0.92) Fem2 + Tib4 0.89 0.87 (0.80–0.91)

used it to predict the HKAA in the remaining 20% of cases. We repeated this process 5 times, so each case will have a predicted HKAA. For this dataset, we present the Pearson correlation and ICC between the predicted HKAA and gold standard. In addition, we present a Bland–Altman plot displaying absolute measurement errors of predicted HKAA vs. the gold standard in our cross-validation experiments. As no similar experiment was published previously, a valid samplesize calculation was not possible and we applied the minimum of 100 cases as suggested by Vergouwe et al. (2005). We chose to include 50 males and 50 females, to account for sex-specific differences. Ethics, funding, and potential conflicts of interest All radiographs were anonymized, and a waiver of consent was obtained from the local medical ethical committee (no. 17-760/C). This work was supported by Reuma Nederland (LLP-22) and the APPROACH project. APPROACH has received support from the Innovative Medicines Initiative Joint Undertaking under Grant Agreement no. 115770, resources of which are composed of a financial contribution from the European Union’s Seventh Framework Programme (FP7/20072013) and an EFPIA companies’ in-kind contribution. See www.imi.europa.eu. C. Lindner and T.F. Cootes were funded by the Engineering and Physical Sciences Research Council, UK (EP/M012611/1) and by the Medical Research Council, UK (MR/S00405X/1). Drs Cootes and Lindner have a patent US 9928443, EP 2893491 issued.

–10

–5

Mean between observed HKA and predicted HKA

Figure 3. Bland–Altman plot depicting the error between the observed HKAA (gold standard) and the predicted HKAA in the cross-validation setting. Negative numbers represent the degree of varus alignment and positive numbers represent the degree of valgus alignment. The solid line depicts the mean error and the dotted lines the 95% confidence interval.

Results Of 100 full-limb radiographs, 11 had radiographs of both knees available and 89 had only 1 knee radiograph available. 1 knee radiograph was of insufficient quality to perform FTA measurements and was excluded, resulting in 110 full-limb/standard knee radiograph pairs. The mean age was 54 (SD 7.4) and 53 knees were male. Of all knees, 9 were KL 0, 23 were KL 1, 30 were KL 2, 36 were KL3, and 12 were KL 4. Correlation between FTA and HKAA measurements across all pairs of images Across all pairs of images, the Pearson correlations between FTA and HKAA ranged between 0.83 and 0.90 (Table 2). The ICC values ranged from 0.83 to 0.90. The best correlations between HKAA and FTA measurements were found using the FTA defined as a femoral axis between the mid-shaft of the femur (approximately 10 cm above the joint line) and the femoral notch (Fem2), and a tibial axis running through 2 points in the mid-shaft of the tibia (approximately 4 cm and 10 cm beneath the tibial plateau (Tib1). Linear regression to predict the HKAA using the optimal FTA method (Fem2 + Tib1) produced the formula: HKAA = –2.182 + FTA*0.995. The mean absolute error between the predicted HKAA and the observed HKAA was 1.7° (SD 1.2°, range 0.1–5.4). Correlation between FTA and HKAA predictions in cross-validation experiments The correlation statistics found in the cross-validation setting were comparable to those found across all pairs of images,

736

albeit minimally weakened (Table 3). Again, the combination of femoral axis 2 and tibial axis 1 showed the best correlation (Pearson correlation 0.90, ICC 0.90). The mean absolute error between the predicted HKAA and the observed HKAA was 1.8° (SD 1.3°, range 0.1–5.3) in the cross-validation setting. The Bland–Altman plot depicts the error between the observed HKAA (gold standard) and the predicted HKAA in the cross-validation setting (Figure 3). No systematic errors or outliers were found in this plot. Although females were more likely to have a valgus alignment compared with males, the error between observed HKAA and predicted HKAA was similar between sexes (p = 0.9). Linear regression models containing an interaction between FTA and sex, or a quadratic term, performed slightly better across all pairs of images, but performed slightly worse in the cross-validation experiments (data not shown).

Discussion This study showed that the mechanical HKAA can be predicted from a standard knee radiograph using our automated pipeline. We used several FTA definitions and compared their performance in predicting the HKAA. The best-performing FTA definition used a femoral axis between the mid-shaft of the femur (approximately 10 cm above the joint line) and the femoral notch, and a tibial axis running through 2 points in the mid-shaft of the tibia (approximately 4 cm and 10 cm beneath the tibial plateau). This combination to measure FTA had not been reported in the literature. Compared with previous work, the Pearson correlation coefficient between FTA and HKAA measurements (0.83 to 0.90) was high across all pairs of images (Tables 1 and 2) (Kraus et al. 2005, Hinman et al. 2006, Brouwer et al. 2007, Issa et al. 2007, Colebatch et al. 2009, Felson et al. 2009, van Raaij et al. 2009, Sheehy et al. 2011, Navali et al. 2012, Zampogna et al. 2015). The results of the cross-validation can be used to estimate the performance of the HKAA predictions in new cases. In the cross-validation the Pearson correlation between the automatically calculated FTA and the predicted HKAA was 0.90. The mean absolute error was 1.8° (SD 1.3°). To the best of our knowledge the performance of the predicted HKAA based on FTA has not been reported in any cross- or external validation studies. The automatically calculated FTA provides an easy tool to study the influence of varus/valgus malalignment in OA cohorts or trials for which standard knee radiographs are available. We tested only the FTA produced by our automatic analysis pipeline to predict mechanical HKAA. However, the pipeline may be used to collect other measurements automatically and enables the rapid analysis of a collection of measurements for a large number of radiographs. The search algorithms we used were trained on only a small database (293 knees). Small corrections to the landmarks had to be made,

Acta Orthopaedica 2020; 91 (6): 732–737

costing approximately 1 minute per radiograph. In the future we expect the search model to have sufficient accuracy to run fully automatically without the need for manual correction. A database containing around 1,000 knees should be sufficient to achieve this (Gielis et al. 2020). While the standard AP radiograph is most often used in the clinics, numerous OA studies use a semi-flexed PA radiograph. This technique aims to compensate for the tibial slope and give a more accurate reading on the joint space width. The generalizability of our methods to PA radiographs should be checked, and the formula to calculate the predicted HKAA may need adaptation. Additionally, some of the FTA definition may be applied in knees with a prosthesis, notably the definition using only landmarks in the femoral and tibial shafts. However, due to prosthesis placement, translation between the joint center and the femur and/or tibia or changes in the joint angle might occur. This has not been validated as we studied only native knees. Clinically, a validated method to measure leg malalignment from standard knee radiographs would be very useful, as this would make a large proportion of full-limb radiographs unnecessary. A full-limb radiograph has several disadvantages, such as higher costs, more radiation, and the need for specialized equipment. However, it is important to question whether the mean observed error of 1.8° is of sufficient accuracy for clinical applications. Odenbring et al. (1993) suggested that a 3-degree accuracy in measuring the mechanical HKAA is sufficient, as this resembles the precision of a correction osteotomy. To our knowledge the scan–rescan error for determining the HKAA using full-limb radiographs is described in only one study including 8 cases. The authors reported a mean error of 1.3°, but their measurements are rounded to the full degree. Sanfridsson et al. (1996) found a correlation of 0.91 when comparing standard HKAA radiography with the novel QUESTOR method (using a specific positioning platform and software to perform and analyze the full-limb radiography). A statistically significant mean difference of 0.7° in HKAA between double and single leg weight-bearing full-limb radiographs was reported by Yazdanpanah et al. (2017) but they did not report the mean absolute error. More research is needed to investigate the scan–rescan reliability of the HKAA from full-limb radiographs. Our study has a number of strengths. We used standardized clinical radiographs, with a protocol feasible in clinical care. We included an equal number of males and females. We tested a large number of FTA definitions using the same set of radiographs to directly compare their performance. Finally, we used a cross-fold validation experiment to test our predictions in unseen radiographs. The main limitation of our study is that the reliability of the gold standard (the HKAA) has been poorly studied. However, the HKAA is the most commonly used measurement to determine the mechanical angle of the lower extremity in both research studies and clinical care.

Acta Orthopaedica 2020; 91 (6): 732–737

Conclusion We have developed an automated image analysis pipeline to calculate the FTA from standard knee radiographs. We directly compared multiple FTA definitions and tested their performance in predicting the HKAA, as measured from full-limb radiographs. The best-performing FTA definition correlated strongly with the HKAA and predicted it with high accuracy. The proposed image analysis pipeline can be used for epidemiological research on lower-limb alignment in cohorts with standard knee radiographs.

All authors provided significant contribution to drafting and/or revising the manuscript and to research design. Additionally, all authors were involved in either acquisition, analysis, or interpretation of the data. Acta thanks Kaj Knutson for help with peer review of this study.

Agricola R, Leyland K M, Bierma-Zeinstra S M A, Thomas G E, Emans P J, Spector T D, Weinans H, Waarsing J H, Arden N K. Validation of statistical shape modelling to predict hip osteoarthritis in females: data from two prospective cohort studies (Cohort Hip and Cohort Knee and Chingford). Rheumatology 2015; 54(11): 2033-41. Brouwer G M, Tol A W Van, Bergink A P, Belo J N, Bernsen R M D, Reijman M, Pols H A P, Bierma-Zeinstra S M A. Association between valgus and varus alignment and the development and progression of radiographic osteoarthritis of the knee. Arthritis Rheum 2007; 56(4): 1204-11. Colebatch A N, Hart D J, Zhai G, Williams F M, Spector T D, Arden N K. Effective measurement of knee alignment using AP knee radiographs. Knee 2009; 16(1): 42-5. Felson D T, Cooke T D V, Niu J, Goggins J, Choi J, Yu J, Nevitt M C. Can anatomic alignment measured from a knee radiograph substitute for mechanical alignment from full limb films? Osteoarthr Cartil 2009; 17(11): 1448-52. Gielis W P, Weinans H, Welsing P M J, van Spil W E, Agricola R, Cootes T F, de Jong P A, Lindner C. An automated workflow based on hip shape improves personalized risk prediction for hip osteoarthritis in the CHECK study. Osteoarthr Cartil 2020; 28(1): 62-70. Haverkamp D J, Schiphof D, Bierma-Zeinstra S M, Weinans H, Waarsing J H. Variation in joint shape of osteoarthritic knees. Arthritis Rheum 2011; 63(11): 3401-7. Hayes B, Kittelson A, Loyd B, Wellsandt E, Flug J, Stevens-Lapsley J. Assessing radiographic knee osteoarthritis: an online training tutorial for the Kellgren–Lawrence grading scale. MedEd PPORTAL 2016; Published online 2016 Nov 18. doi: 10.15766/mep_2374-8265.10503 Hinman R S, May R L, Crossley K M. Is there an alternative to the full-leg radiograph for determining knee joint alignment in osteoarthritis? Arthritis Rheum 2006; 55(2): 306-13. Iranpour-Boroujeni T, Li J, Lynch J A, Nevitt M, Duryea J. A new method to measure anatomic knee alignment for large studies of OA: data from the Osteoarthritis Initiative. Osteoarthr Cartil 2014; 22(10): 1668-74. Issa S N, Dunlop D, Chang A, Song J, Prasad P V, Guermazi A, Peterfy C, Cahue S, Marshall M, Kapoor D, Hayes K, Sharma L. Full-limb and knee radiography assessments of varus–valgus alignment and their relationship

737

to osteoarthritis disease features by magnetic resonance imaging. Arthritis Care Res 2007; 57(3): 398-406. Kraus V B, Vail T P, Worrell T, McDaniel G. A comparative assessment of alignment angle of the knee by radiographic and physical examination methods. Arthritis Rheum 2005; 52(6): 1730-5. Lindner C, Thiagarajah S, Wilkinson J, The arcOGEN Consortium, Wallis G, Cootes T. Fully Automatic segmentation of the proximal femur using random forest regression voting. IEEE Trans Med Imaging 2013a; 32(8): 1462-72. Lindner C, Thiagarajah S, Wilkinson J M, The acrOGEN Consortium, Wallis G A, Cootes T F. Development of a fully automatic shape model matching (FASMM) system to derive statistical shape models from radiographs: application to the accurate capture and global representation of proximal femur shape. Osteoarthr Cartil 2013b; 21(10): 1537-44. McDaniel G, Mitchell K L, Charles C, Kraus V B. A comparison of five approaches to measurement of anatomic knee alignment from radiographs. Osteoarthr Cartil NIH Public Access 2010; 18(2): 273-7. Moreland J R, Bassett L W, Hanker G J. Radiographic analysis of the axial alignment of the lower extremity. J Bone Jt Surg Ser A 1987; 69(5): 745-9. Moyer R, Wirth W, Duryea J, Eckstein F. Anatomical alignment, but not goniometry, predicts femorotibial cartilage loss as well as mechanical alignment: data from the Osteoarthritis Initiative. Osteoarthr Cartil 2016; 24(2): 254-61. Navali A M, Bahari L A S, Nazari B. A comparative assessment of alternatives to the full-leg radiograph for determining knee joint alignment. Sports Med Arthrosc Rehabil Ther Technol 2012; 4(1): 40. Odenbring S, Berggren A M, Peil L. Roentgenographic assessment of the hip–knee–ankle axis in medial gonarthrosis: a study of reproducibility. Clin Orthop Relat Res 1993; (289):195-6. Sanfridsson J, Ryd L, Eklund K, Kouvaras Y, Jonsson K. Angular configuration of the knee: comparison of conventional measurements and the QUESTOR precision radiography system. Acta Radiol 1996; 37(5): 633-8. Sharma L, Song J, Felson D T, Cahue S, Shamiyeh E, Dunlop D D. The role of knee alignment in disease progression and functional decline in knee osteoarthritis. JAMA 2001; 286(2): 188. Sheehy L, Felson D, Zhang Y, Niu J, Lam Y-M, Segal N, Lynch J, Cooke T D V. Does measurement of the anatomic axis consistently predict hip–knee– ankle angle (HKA) for knee alignment studies in osteoarthritis? Analysis of long limb radiographs from the multicenter osteoarthritis (MOST) study. Osteoarthr Cartil NIH Public Access 2011; 19(1): 58-64. van Raaij T M, Brouwer R W, Reijman M, Bierma-Zeinstra S M A, Verhaar J A N. Conventional knee films hamper accurate knee alignment determination in patients with varus osteoarthritis of the knee. Knee 2009; 16(2): 109-11. Vergouwe Y, Steyerberg E W, Eijkemans M J C, Habbema J D F. Substantial effective sample sizes were required for external validation studies of predictive logistic regression models. J Clin Epidemiol 2005; 58(5): 475-83. Waarsing J H, Kloppenburg M, Slagboom P E, Kroon H M, Houwing-Duistermaat J J, Weinans H, Meulenbelt I. Osteoarthritis susceptibility genes influence the association between hip morphology and osteoarthritis. Arthritis Rheum 2011; 63(5): 1349-54. Wesseling J, Boers M, Viergever M A, Hilberdink W K H A, Lafeber F P J G, Dekker J, Bijlsma J W J. Cohort profile: Cohort Hip and Cohort Knee (CHECK) study. Int J Epidemiol 2016; 45(1): 36-44. Yazdanpanah O, Mobarakeh M K, Nakhaei M, Baneshi M R. Comparison of double and single leg weight-bearing radiography in determining knee alignment. Arch Bone Jt Surg 2017; 5(3): 174-80. Zampogna B, Vasta S, Amendola A, Uribe-Echevarria Marbach B, Gao Y, Papalia R, Denaro V. Assessing lower limb alignment: comparison of standard knee X-ray vs long leg view. Iowa Orthop J 2015; 35: 49-54.

738

Acta Orthopaedica 2020; 91 (6): 738–742

The implications of an aging population and increased obesity for knee arthroplasty rates in Sweden: a register-based study Anders OVERGAARD 1, Peder FREDERIKSEN 1, Lars Erik KRISTENSEN 1, Otto ROBERTSSON 2,3, and Annette W-DAHL 2,3 1 The Parker Institute, Copenhagen University Hospital Frederiksberg, Copenhagen, Denmark; 2 The 3 Faculty of Medicine, Department of Clinical Sciences, Orthopedics, Lund University, Sweden

Swedish Knee Arthroplasty Register, Lund;

Correspondence: anders.overgaard5@gmail.com Submitted 2019-12-03. Accepted 2020-06-22.

Background and purpose — Total knee arthroplasty (TKA) has increased substantially in Sweden. We quantified the relative risk for TKA in the Swedish population for different BMI categories and age groups to investigate whether the continued increase in TKA is attributable to increased prevalence of obesity and elderly people in the population, and to put forward model predictions for coming needs for TKA. Patients and methods — We used the Swedish Nationwide Health Survey (SNHS) and the Swedish Knee Arthroplasty Register (SKAR) 2009–2015 to calculate the relative risk (RR) of TKA by age (middle-aged 45–64 years and elderly 65–84 years) and BMI (BMI 18.5–24.9 normal weight; BMI 25.0–29.9 overweight; BMI > 30 obese). The RR for TKA was applied to the demographic forecasts for the Swedish population as a forecasting model. Results — Population size increased 5.2% from 2009 to 2015 to 40,000 middle-aged and 250,000 elderly, and the prevalence of obesity increased from 16% to 18% in these 2 age categories. Compared with those of normal weight, the RR for TKA was 2.7 (95% CI 2.5–3.0) higher for the overweight and 7.3 (6.7–8.0) higher for the obese, aged 45–64. The corresponding figures for individuals aged 65–84 were 2.1 (2.0–2.2) and 4.0 (3.8–4.3) higher, respectively. The changes in the prevalence of obesity and an increase in the elderly population accounted for an estimated increase of 1,700 TKAs over the 7 years. Interpretation — The increase in obesity frequency and the rise in the population of middle-aged and elderly may, to some extent, explain the rise in TKA utilization in Sweden.

A steady increase in knee arthroplasty has been observed (Swedish Knee Arthroplasty Register [SKAR] 2019). The broader acceptance of knee arthroplasty as treatment for knee osteoarthritis (OA) may have led to the rapid increases in utilization and is predicted by some to continue (Kurtz et al. 2007, Kim et al. 2012). Multinational and regional register comparisons show similar tendencies (Robertsson et al. 2010, Kurtz et al. 2011). Different modeling forecasts have attempted to predict future operation needs (Nemes et al. 2014, Inacio et al. 2017) to prepare the healthcare providers. Obesity is 1 of the major risk factors for development and progression of knee OA that may lead to knee arthroplasty (Felson et al. 1997, Lohmander et al. 2009, Wang et al. 2009). Therefore, as obesity among the population increases, the number of operations is expected to follow a similar pattern. Studies have questioned whether the dramatic rise in operation frequency is attributable to the rise in the prevalence of obesity and the shift in age in the population (Kurtz et al. 2007, Losina et al. 2012, Wills et al. 2012). Following trends and predictions from the United States (Kurtz et al. 2007) where the prevalence of obese individuals in the population has grown at a faster rate than in Sweden, one may speculate that the frequency will continue to increase in Sweden as well. As the population of the world increases in age and weight the strain on healthcare systems is evident (WHO aging and health[REF?]). Also, the demands for specialized, qualified personnel treating knee OA yield a requirement for studies on the association between demographic changes and the need for arthroplasty surgery to understand and handle coming demands.

Acta Orthopaedica 2020; 91 (6): 738–742

739

Table 1. Swedish population size and changes in population size by age group, 2009–2015 (SCB) Factor

2009 2010 2011 2012 2013 2014 2015 Change

Overall 9,340,682 9,415,570 9,482,855 9,555,893 9,644,864 9,747,355 9,851,017 +510,335 (5.5%) Distribution by age 45–64 2,407,091 2,412,844 2,418,771 2,425,951 2,429,634 2,434,058 2,445,729 +38,638 (1.6%) 65–84 1,442,601 1,486,029 1,531,341 1,575,315 1,617,178 1,656,400 1,688,756 +246,155 (17.1%)

We investigated whether the increase in TKA is associated with changes in the prevalence of obesity and the growing elderly population by quantifying the relative risk for TKA in the Swedish population for different BMI categories and age groups.

TKAs, we composed a forecast based on current incidence of TKAs (fixed model) and a linear regression model for development in overweight and obesity with population predictions conducted by the SCB. To calculate the increase in operations for the predicted increase in population the data are extracted from the SCB predicted populations estimate from 2015.

Patients and methods

Statistics Under the assumption of constant age and BMI surgery rates, we calculated the expected number of operations as the weighted sum of the age and BMI-specific incidence and the corresponding population distribution to illustrate the result of population-composition changes and growth in operation frequency.

We used the Swedish Nationwide Health Survey (SNHS) and the Swedish Knee Arthroplasty Register (SKAR) to calculate relative risk (RR) of TKA by age and BMI category. The SNHS comprises a randomly selected national sample of 9,000–10,000 individuals combined with a randomly selected supplementary sample from 4 county councils (Halland, Jönköping, Östergötland, and Kronoberg) and 3 municipalities (Gotland, Göteborg, and Jönköping) consisting of 67,236 individuals (46% men and 54% women) aged 16–84 years. The Sweden Statistics (SCB) provides data on the number of inhabitants divided into different age groups. The SKAR has registered knee arthroplasties since 1975 and has high completeness and correctness of data (SKAR 2019). The response rate for weight and height is > 98% in the SKAR. Weight and height has been validated and was found to have correctness with differences of < 1.5 kg in weight and 1 cm in height between hospital records and what is reported to the SKAR (SKAR 2019). The increase in TKA utilization for different age groups (16–44, 45–64, and 65–84 years) between 2009 and 2015 was estimated by comparing the number of TKAs performed in the period and the operations performed in each BMI category. Increase in population size and prevalence of obesity were calculated similarly. We compared the magnitude of changes in TKA utilization with the changes over the same period regarding population and BMI status. Analyses were performed on the overall population and stratified by age group. We focused on the 2 major age groups who receive TKA (middle-aged 45–64 and elderly 65–84 years), as knee arthroplasty predominates in patients over 45 and under 85 years. To evaluate whether, assuming all other factors remain constant, changes in BMI and population alone could account for the increase in TKA, we combined 2009 with 2015 TKA data with the percentage change in population size, stratified by age group and BMI category. These projections are shown in comparison with the number of TKAs registered in Sweden. Lastly, to assess the future scope of demands on

n = ∑Rate age,BMI × Population age,BMI Lastly, RR is applied to the demographics of the population ranging from 2005 to 2030 and estimates for the future development of the population from the annual report on development in population done by the SCB. A fixed model was used as static incidents are expected. We used the fixed threshold that was calculated as the average age and BMI-specific incidence rate during the study period. To evaluate our methodology our model predictions are compared with the number of primary TKAs registered in the SKAR. Ethics, funding, and conflicts of interest The data gathering of the Swedish Knee Arthroplasty Register was approved by the Ethics Board of Lund University (LU2002). The project has been performed in accordance with the Declaration of Helsinki. The study was supported by a core grant from the Oke Foundation to the Parker Institute. No competing interests were declared.

Results Population growth The Swedish population increased by 510,335 (5.5%) individuals between 2009 and 2015. In the 7-year period, an increase of 38,638 (1.6%) was seen in the middle-aged (45–64 year) population and an increase of 246,155 (17.1%) in the elderly (65–84 year) population (Table 1).

740

Acta Orthopaedica 2020; 91 (6): 738–742

Yearly number of primary TKAs by age group

Distribution (%) of BMI categories

Incidence of TKA per 100,000/year by BMI

Incidence of TKA per 100,000/year by age

14,000

100

400

12,000

85– 65–84 45–64 –44

45–64 65–84

≥ 30 25–29.9 18.5–24.9 < 18.5

300

10,000

8,000

200

6,000

300

≥ 30 25.0–29.9 18.5–24.9 < 18.5

200

4,000

100

20 2,000 0 2004

2006

2008

2010

2012

2014

2016

SKAR

SWEDEN

Figure 1. TKAs performed 2004–2016 Figure 2. Distribution of BMI by age group (SKAR). categories in the TKA population (SKAR) and the Swedish population (SWEDEN) in 2009–2015.

2009 2010 2011 2012

2013 2014 2015

Figure 3. Incidence per 100,000 TKAs 2009–2015 in the different BMI categories.

Table 2. Calculated relative risk (RR) by age and BMI group compared with normal-weight BMI

20,000

Factor Swedes a TKAs b Risk/105/year RR (95% CI)

15,000

Age 45–64 BMI 18.5–24.9 954,746 3,946 59 Reference BMI 25.0–29.9 1,027,390 11,463 159 2.7 (2.4–2.9) BMI ≥ 30.0 425,485 12,860 431 7.3 (6.6–8.0) Age 65–84 BMI 18.5–24.9 602,479 10,736 251 Reference BMI 25.0–29.9 662,284 24,725 526 2.1 (2.0–2.2) BMI ≥ 30.0 265,800 19,136 1,015 4.0 (3.8–4.3) a

Using the proportion of BMI to calculate the number of Swedes in each group. b Total number of total knee arthroplasties in each group 2009–2015.

Increase in obesity The prevalence of obesity was relatively static in the period 2009 to 2015 with an increase of less than 3%. When stratifying for age the most relative change in the prevalence of obesity was an increase of 13% (16% to 18%) among both groups of interest (45 to 84 years of age), with a subsequent reduction in the normal-weight population of 9.5% and 7.3% (42% to 38% and 41% to 38%) for the 2 age groups. The change in the number of overweight and obese individuals by an increase in population size between 2009 and 2015 resulted in a relative increase of 14% for overweight and 32% for obese, while the proportion of non-obese individuals decreased by 8% for the middle-aged and increased by 9% for the elderly. Obesity and age changes and their association with rates of operations The annual number of TKAs during 2009–2015 (varying between 12,044 and 12,763), as well as the proportion of patients in the different age groups, was relatively stable (Figure 1). The recipients of TKA compared with the general Swedish population show a higher percentage of overweight

2009 2010 2011 2012

2013 2014 2015

Figure 4. Incidence per 100,000 TKAs 2009–2015 in the different age groups.

Yearly number of primary TKAs Predicted Actual

10,000

5,000

0 2004

2008

2012

2016

2020

2024

2028

Figure 5. Number of registered and predicted primary TKAs per year 2004–2030.

(BMI 25–29.9: 44% and 35% respectively) and obese (BMI >30: 38% and 14% respectively) individuals than the Swedish population as a whole (Figure 2). 82% of the TKA population are overweight or obese compared with 49% of the entire Swedish population (Figure 3). The incidence of TKA per 100,000/year in the population of obese elderly shows a slight reduction during the end of the study period (Figures 3 and 4). The RR for TKA increased with both age and BMI. For the middle-aged patients (aged 45–64), there was a 2.7-fold (95% CI 2.5–3.0) increase with overweight and a 7.3-fold (95% CI 6.7–8.0) higher risk with obesity compared with those of normal weight. Corresponding figures for elderly patients (aged 65–84) were RR 2.1 (95% CI 2.0–2.2) and RR 4.0 (95% CI 3.8–4.3) (Table 2). The growth of the population resulted in an estimated increased need for 1,259 TKAs. Including the increase in obesity in the model resulted in a calculated increase of 1,744 TKAs between 2009 and 2015. Applying forecast calculations to the predicted Swedish population increase by the SCB we estimate an increase of primary TKAs for OA to approximately 14,200 operations in 2020 and 16,600 in 2030 (Figure 5).

Acta Orthopaedica 2020; 91 (6): 738–742

Discussion We found that patients undergoing TKA were more obese than the general Swedish population, and that the magnitude of the Swedish elderly population was increasing at a faster rate than other age groups. Moreover, the increasing BMI in itself was statistically significantly associated with the risk of undergoing knee replacement. The increase in the number of individuals suffering from obesity in the population together with the increase in population size, especially among the elderly, may to some extent explain the continuing increase in TKAs in Sweden. These findings exemplify the growing healthcare costs attributable to obesity and a growing elderly population through increases in operations for OA. The historical increase in the use of TKA is influenced by a multitude of factors, including a growing prevalence of sportsrelated knee injuries, a change in physical workloads and a sedentary lifestyle (Felson et al. 1997). Major contributors may also involve the broader acceptance that TKA as a treatment for knee OA has earned among the general population and healthcare providers at large over time. We found a striking association between obesity and risk of surgery. The level of obesity was higher for TKA patients than for the rest of the Swedish population. Further, that increasing obesity as indicated by BMI category was associated with an increased RR for TKA, and that increasing obesity had the most dramatic association with younger patients. For the obese (BMI > 30), the RR for a TKA increased 7-fold compared with the normal weight individual aged 45–64. Using these data we produced a prediction model that may give an explanation for the continued increase in numbers of TKAs performed in Sweden. If the indications, economic, and TKA use trends remain unchanged, as appears to have been the case over the 7 years, demographic changes in general could explain the increase in TKAs. This seconds the notion that “overweight driven OA” is the major contributor to the rise in operation rates in combination with an increasing elderly population, and that normal or healthy weight may protect against TKAs (Lohmander et al. 2004). This notion is confounded by the use of TKA as a surrogate endpoint for late-stage OA in the population. Though there is a shared pathogenetic phenotype of OA and obesity the frequency of OA in the population is hard to quantify. In fact, only for a lesser subgroup of patients suffering from OA who need an arthroplasty might obesity be the deciding factor driving the OA patients to surgery, as obesity alone affects locomotion and self-rated health (Marks 2007, Silverwood et al. 2015). Kurtz et al. (2007 revised 2014) calculated an increase of over 600% TKA operations in the United States from 2005 to 2030 based on trajectories from the rapid increases seen between 1990 and 2003, when the increases in numbers of arthroplasty performed increased rapidly. This model is not applicable to the Swedish population, where trends of incidences have slowed. Cross-sectional studies from this period

741

(Kurtz et al. 2007, Losina et al. 2012) show that obesity and demographic changes cannot explain the full extent of the explosive TKA increase in the United States and Germany. Increases in use in younger patients due to sports-related injuries and direct-to-consumer advertisements may influence the public awareness of treatment options as described by Dieppe et al. (1999) for publications from 1990–1998 with considerable variance in the use and demand across nations and healthcare models. A more conservative estimate may be warranted in the Scandinavian countries where the incidence historically is lower. More recent calculations expecting continued explosive growth is rejected and logarithmic development to Poisson’s models describes the differences in mathematical approaches (Sasieni and Adams 1999, Inacio et al. 2017). The main criticism of such data-driven models as presented by Culliford et al. (2015) is that they do not involve any estimate for planning and supply-side constraints (capacity of surgeons and hospital units) and restrictions, as is the case in this model. Regardless of modeling attempts, aging and increasing BMI is part of what is driving the TKA patient increase. And mathematical modeling per se is an attempt to “fit” the trends expressed in the registries. We do not expect current limitations in specialized personnel in Sweden to be an issue. However, for the Nordic countries use of TKA seems to have reached a plateau, therefore modeling expecting continued explosive growth as in the United States is rejected. The fact that TKA use in Sweden follows the needs and demands from the public and lawmakers is evident in upholding the waiting-list guarantees of no longer than 90 days wait for surgery (before COVID19). Though there are strengths in the data sources, our analyses are limited by the fact that population, obesity, and TKA data were each obtained from different sources, limiting the ability to compare in-group changes. Additionally, we did not have the possibility to analyze supply-chain limitations and therefore our estimate may only be indicative for the immediate future TKA frequencies. The SKAR has collected data on BMI since 2009, thus limiting the opportunity to compare earlier more marked changes in operation rates. Prior to 2009 more dramatic changes in the number of operations were seen and might to an extent be influenced by the factors mentioned above. This deems the data unfit for a fixed calculation model as BMI-defined incidence would follow trends of usage and therefore would not explain their contribution. Nationwide register-based studies like the present one have the apparent strength of being population-based, thus reducing the risk of selection bias. However, some degree of residual confounding, bias, and misclassification cannot be ruled out. For the Swedish population and Swedish healthcare providers, assuming unchanged indications and utilization patterns, we can evaluate the consequence of demographic changes and changes in overweight and obesity in the population. The reliability of the forecast model diminishes as new prosthesis types and centralization of treatment and political intervention as “treatment guarantees” change the way patients are treated

742

over time; though stable at the moment, although the increase in obesity in Sweden in a longer perspective than 2009–2015 is relatively “large,” for the studied period increases were meager. These observations on RR highlight the need for information on BMI if one is to compare and combine data from registers as this is shown to have large implications for the incidence of operations. The increase in obesity is likely to be a contributor to the increases seen in national registers globally. The lack of medical treatment options for OA leaves patients seeking surgery, and preventing or postponing surgery by treatment of modifiable risk factors like obesity should be a key objective for halting increasing numbers of TKAs being performed. The attributable fraction among exposed knee OA patients with concomitant overweight or obesity is confounded by the inherent limitations that patients with knee OA and overweight/obesity suffer from the consequent additional problem constituted by difficulties in exercising. In conclusion, we have shown a striking association of obesity and an increasingly elderly population with the rise in TKA rates. Most likely, the continuous increase in, and aging of, the Swedish population will lead to a higher demand for intervention for OA and an increased burden on healthcare resources and hospital budgets.

The study was conceived by AO, LEK, OR, and AWD. AO, PF, and OR performed the analyses. AO wrote the initial draft. All authors contributed to the interpretation of the data and to the revision of the manuscript. The authors would like to thank all the contact surgeons and associated staff at the hospitals in Sweden for their dedicated registration work over the years. Acta thanks Hannu Miettinen, Søren Overgaard, and Reinhard Windhager for help with peer review of this study.

Culliford D, Maskell J, Judge A, Cooper C, Prieto-Alhambra D, Arden N K. Future projections of total hip and knee arthroplasty in the UK: results from the UK Clinical Practice Research Datalink. Osteoarthritis Cartilage 2015; 23(4): 594-600. doi: 10.1016/j.joca.2014.12.022. Dieppe P, Basler H D, Chard J, Croft P, Dixon J, Hurley M, Lohmander S, Raspe H. Knee replacement surgery for osteoarthritis: effectiveness, practice variations, indications and possible determinants of utilization. Rheumatology (Oxford) 1999; 38(1): 73-83. doi: 10.1093/rheumatology/38.1.73. Felson D T, Zhang Y, Hannan M T, Naimark A, Weissman B, Aliabadi P, Levy D. Risk factors for incident radiographic knee osteoarthritis in the elderly: the Framingham Study. Arthritis Rheum 1997; 40(4): 728-33. doi: 10.1002/1529-0131(199704)40:4<728::aid-art19>3.0.co;2-d. Inacio M C S, Paxton E W, Graves S E, Namba R S, Nemes S. Projected increase in total knee arthroplasty in the United States: an alternative

Acta Orthopaedica 2020; 91 (6): 738–742

projection model. Osteoarthritis Cartilage 2017; 25(11): 1797-803. doi: 10.1016/j.joca.2017.07.022. Kim S H, Gaiser S, Meehan J P. Epidemiology of primary hip and knee arthroplasties in Germany: 2004 to 2008. J Arthroplasty 2012; 27(10): 1777-82. doi: 10.1016/j.arth.2012.06.017. Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am 2007; 89(4): 780-5. doi: 10.2106/JBJS.F.00222. Kurtz S M, Ong K L, Lau E, Widmer M, Maravic M, Gomez-Barrena E, de Pina Mde F, Manno V, Torre M, Walter W L, de Steiger R, Geesink R G, Peltola M, Roder C. International survey of primary and revision total knee replacement. Int Orthop 2011; 35(12): 1783-9. doi: 10.1007/s00264-0111235-5. Kurtz S M, Ong K L, Lau E, Bozic K J. Impact of the economic downturn on total joint replacement demand in the United States: updated projectionsto 2021. J Bone Joint Surg Am 2014; 96(8): 624-30. doi: 10.2106/ JBJS.M.00285. Lohmander L S, Felson D. Can we identify a ‘high risk’ patient profile to determine who will experience rapid progression of osteoarthritis? Supported by the Swedish Research Council (Medicine) and NIH AR47785. Osteoarthritis Cartilage 2004; 12: 49-52. doi: 10.1016/j.joca.2003.09.004. Lohmander L S, Gerhardsson de Verdier M, Rollof J, Nilsson P M, Engstrom G. Incidence of severe knee and hip osteoarthritis in relation to different measures of body mass: a population-based prospective cohort study. Ann Rheum Dis 2009; 68(4): 490-6. doi: 10.1136/ard.2008.089748. Losina E, Thornhill T S, Rome B N, Wright J, Katz J N. The dramatic increase in total knee replacement utilization rates in the United States cannot be fully explained by growth in population size and the obesity epidemic. J Bone Joint Surg Am 2012; 94(3): 201-7. doi: 10.2106/JBJS.J.01958. Marks R. Obesity profiles with knee osteoarthritis: correlation with pain, disability, disease progression. Obesity 2007; 15(7): 1867-74. Nemes S, Gordon M, Rogmark C, Rolfson O. Projections of total hip replacement in Sweden from 2013 to 2030. Acta Orthop 2014; 85(3): 238-43. doi: 10.3109/17453674.2014.913224. Public Health Agency of Sweden. Data from the Public Health Survey. Available at https://www.folkhalsomyndigheten.se/[AQ5] Robertsson O, Bizjajeva S, Fenstad A M, Furnes O, Lidgren L, Mehnert F, Odgaard A, Pedersen A B, Havelin L I. Knee arthroplasty in Denmark, Norway and Sweden: a pilot study from the Nordic Arthroplasty Register Association. Acta Orthop 2010; 81(1): 82-9. doi: 10.3109/17453671003685442. Sasieni P D, Adams J. Standardized lifetime risk. Am J Epidemiol 1999; 149(9): 869-75. Silverwood V, Blagojevic-Bucknall M, Jinks C, Jordan J, Protheroe J, Jordan K. Current evidence on risk factors for knee osteoarthritis in older adults: a systematic review and meta-analysis. Osteoarthritis Cartilage 2015; 23(4): 507-15. Statistics Sweden SCB. The future population of Sweden 2015−2060. Demographic reports 2015:2. Available at www.scb.se/en/ Swedish Knee Arthroplasty Register. Annual Report, 2019; 2019. Available at http://www.myknee.se Wang Y, Simpson J A, Wluka A E, Teichtahl A J, English D R, Giles G G, Graves S, Cicuttini F M. Relationship between body adiposity measures and risk of primary knee and hip replacement for osteoarthritis: a prospective cohort study. Arthritis Res Ther 2009; 11(2): R31. doi: 10.1186/ar2636. Wills A K, Black S, Cooper R, Coppack R J, Hardy R, Martin K R, Cooper C, Kuh D. Life course body mass index and risk of knee osteoarthritis at the age of 53 years: evidence from the 1946 British birth cohort study. Ann Rheum Dis 2012; 71(5): 655-60. doi: 10.1136/ard.2011.154021.

Acta Orthopaedica 2020; 91 (6): 743–749

743

Intra-articular injection with Autologous Conditioned Plasma does not lead to a clinically relevant improvement of knee osteoarthritis: a prospective case series of 140 patients with 1-year follow-up Jasmijn V KORPERSHOEK 1, Lucienne A VONK 1, Tommy S DE WINDT 1, Jon ADMIRAAL 1, Esmee C KESTER 1, Nienke VAN EGMOND 1, Daniël B F SARIS 1,2, and Roel J H CUSTERS 1 1 University Medical Center Utrecht, Utrecht, the Netherlands; 2 Mayo Clinic, Rochester, USA Correspondence: r.j.h.custers@umcutrecht.nl Submitted 2020-06-02. Accepted 2020-06-28.

Background and purpose — Platelet-rich plasma (PRP) is broadly used in the treatment of knee osteoarthritis, but clinical outcomes are highly variable. We evaluated the effectiveness of intra-articular injections with Autologous Conditioned Plasma (ACP), a commercially available form of platelet-rich plasma, in a tertiary referral center. Second, we aimed to identify which patient factors are associated with clinical outcome. Patients and methods — 140 patients (158 knees) with knee osteoarthritis (Kellgren and Lawrence grade 0–4) were treated with 3 intra-articular injections of ACP. The Knee Injury and Osteoarthritis Outcome Score (KOOS), pain (Numeric Rating Scale; NRS), and general health (EuroQol 5 Dimensions; EQ5D) were assessed at baseline and 3, 6, and 12 months’ follow-up. The effect of sex, age, BMI, Kellgren and Lawrence grade, history of knee trauma, and baseline KOOS on clinical outcome at 6 and 12 months was determined using linear regression. Results — Mean KOOS increased from 37 at baseline to 44 at 3 months, 45 at 6 months, and 43 at 12 months’ followup. Mean NRS-pain decreased from 6.2 at baseline to 5.3 at 3 months, 5.2 at 6 months, and 5.3 at 12 months. EQ5D did not change significantly. There were no predictors of clinical outcome. Interpretation — ACP does not lead to a clinically relevant improvement (exceeding the minimal clinically important difference) in patients suffering from knee osteoarthritis. None of the investigated factors predicts clinical outcome.

Platelet-rich plasma (PRP) has emerged as a potential treatment for osteoarthritis (OA). High levels of growth factors and cytokines present in platelets stimulate production of cartilage extracellular matrix, proliferation of chondrocytes, and migration of chondrocytes in vitro (Fortier et al. 2011, Fice et al. 2019). The potential beneficial effect of PRP in OA, together with the lack of regulatory restrictions in the use of these minimally manipulated autologous products, has rushed the field forward. The efficacy of PRP for the treatment of OA in clinical trials varies between no clinically relevant effect and a strong analgesic effect (Sánchez et al. 2012, Patel et al. 2013, Filardo et al. 2015, Gobbi et al. 2015, Forogh et al. 2016, Cole et al. 2017, Lin et al. 2019). The efficacy of a commercially available PRP, Autologous Conditioned Plasma (ACP, Arthrex GmbH, Munich, Germany) has been proven in the setting of RCTs (Cerza et al. 2012, Smith 2015, Cole et al. 2017), but effectiveness has not been investigated in daily clinical practice. Moreover, the effect of different patient factors on the clinical outcome after ACP treatment is unknown. This prospective case series aims to assess the effectiveness of ACP in clinical practice and to investigate the effect of sex, age, BMI, radiographic OA grade (Kellgren and Lawrence), history of knee trauma, and baseline Knee Injury and Osteoarthritis Outcome Score (KOOS) on clinical outcome. Since there is no consensus on whether PRP is more effective in mild or advanced OA (Lana et al. 2016, Jubert et al. 2017, Burchard et al. 2019), we included patients with symptomatic OA of all grades. We hypothesize that treatment with ACP leads to clinically relevant improvement in KOOS5 and that clinical outcome can be predicted with any of the investigated patient factors.

744

Acta Orthopaedica 2020; 91 (6): 743–749

Patients receiving Autologous Conditioned Plasma treatment during study period n = 165

Table 1. Baseline characteristics of 140 included patients (158 knees) Characteristic

Excluded patients (n = 6): – did not receive a third injection, 4 – did not have an e-mail address, 1 – did not understand Dutch language, 1 Eligible patients n = 159 Did not give broad consent n = 19 Patients included in analysis (n = 140) – patients with one-sided treatment, 123 – patients with bilateral treatment, 18 Knees included in analysis (n = 158) Excluded knees (n = 44): – received alternative treatment within 1 year, 6 - knee joint distraction, 3 - ligament surgery, 1 - total knee arthroplasty, 2 – wished to leave study, 13 – did not fill in 12 months survey for unknown reason, 25 Knees included in the 12–month follow-up n = 114

Figure 1. Patient recruitment.

Patients and methods Study design and setting This prospective case series includes patients treated with ACP in an academic hospital (University Medical Center Utrecht, the Netherlands) between March 2017 and October 2018. A minimal follow-up of 1 year was chosen, because the effect of ACP reaches its maximum between 6 and 12 months (Cerza et al. 2012, Filardo et al. 2013, Cole et al. 2017). Inclusion criteria were: first series of ACP, symptomatic OA (Kellgren and Lawrence grade 0 to 4), sufficient understanding of the Dutch language to fill in the questionnaires and written informed consent. Exclusion criteria were: less than 3 ACP injections and earlier treatment with ACP. Patients 140 patients (158 knees) could be included (Figure 1). 43 patients received 1 of the 3 injections with a 2-week interval (due to public holidays and other scheduling issues), all others received 3 consecutive injections with a 1-week interval. Sex, age, and BMI were collected from the patient records. History of knee trauma was defined as having a previous diagnosis of traumatic meniscus tear, cartilage defect or cruciate ligament tear. Baseline data were complete for all

Age, mean (SD) Female sex, n (%) BMI, mean (SD) History of traumatic injury, meniscus, anterior cruciate ligament, cartilage defect, n(%) Baseline KOOS5, mean (SD) Baseline NRS-pain, mean (SD) Baseline EQ5D, mean (SD) Bilateral treatment, n Kellgren and Lawrence grade, n (%) 0 1 2 3 4

N = 158 49 (10) 80 (51) 28 (4.1) 79 (50) 37 (14) 62 (1.9) 63 (19) 18 8 (5.1) 40 (25) 55 (35) 43 (27) 12 (7.6)

Abbreviations: EQ5d, EuroQol 5 dimensions; KOOS5, average of the 5 subscales of the Knee Injury and Osteo arthritis Outcome Score (KOOS); NRS pain, numeric rating scale; SD, standard deviation.

patient factors except BMI (35% missing) (Table 1). We did not monitor or correct for the use of other medications during the study period. Radiographic assessment Patients underwent anteroposterior and lateral view radiographies prior to treatment. Kellgren and Lawrence grade was assessed by 3 blinded observers. In any case where 1 observer rated the radiograph with 1 grade lower or higher than the others, the grade of the 2 observers was accepted. If the grades of 2 observers were 2 or more apart, agreement was reached in a consensus meeting. Interobserver reliability was assessed using a 2-way random intraclass correlation coefficient. The internal consistency of the Kellgren and Lawrence grade was good with a Cronbach’s alpha of 0.89. ACP preparation The Arthrex ACP Double-Syringe System (Arthrex GmbH, Munich, Germany) was used for preparation of ACP. 15 mL of peripheral blood was drawn and centrifuged at 360G for 5 minutes to separate the blood components. Approximately 3–6 mL ACP was drawn into the inner syringe and injected into the knee joint using a superolateral approach with the patient in supine position. ACP composition Using the CELL-DYN Emerald hematology analyzer (Abbott B.V., Abbott Park, IL, USA), 28 random samples of leftover material from ACP syringes were analyzed anonymously in order to characterize the administered PRP. Platelet, erythrocyte, and leucocyte concentration were measured in duplicate. The volume of injected material was documented.

Acta Orthopaedica 2020; 91 (6): 743–749

Patient reported outcome measures Patients completed all questionnaires using an online survey tool (OnlinePROMS, InterActive Studios, Rosmalen, the Netherlands) at baseline and at 3, 6, and 12 months’ followup. Possible scores ranged from 0 to 100 (worst–best) for KOOS and EuroQol 5 Dimensions (EQ5D), and 0–10 (best– worst) for Numeric Rating Scale for pain (NRS pain). Dutch translations of KOOS (de Groot et al. 2008), EQ5D (EuroQol Research Foundation 2009), and NRS pain (LROI 2018) were used. In cases of bilateral treatment, patients filled in 2 separate surveys. Patients received a reminder after 5 and 10 days, and were contacted by telephone after 2 weeks in order to increase compliance. 89% of the patients filled out the survey at baseline, 87% at 3 months, 76% at 6 months and 75% at 12 months’ follow-up. Of patients who were lost to follow-up, data collected up to that point were included in the analyses. Data processing and statistics Data were analyzed using IBM Statistical Package for the Social Sciences (SPSS) (version 15.0.0.2, IBM Corp, Armonk, NY, USA). Baseline patient factors are reported by means and standard deviation (SD) or number of patients and percentages. Outcomes are shown as average and 95% confidence intervals (CI). Missing data were not imputed; patients with missing outcome variables were not included in the analysis of those specific variables. P-values < 0.05 were considered significant. The primary outcome, the effectiveness of ACP at 1 year, was evaluated using the change from baseline to 1-year follow-up in the average score on the 5 subscales of the KOOS (pain, symptoms, activities of daily living, sport and recreation, and knee-related quality of life). Change from baseline (ΔKOOS5) was estimated as an average population change using generalized estimating equations (GEE). ΔKOOS5 was compared with the minimal clinically important difference (MCID) recommended for KOOS (Roos 2020) using the CI. Since an MCID for non-operative OA treatment has not been defined and the MCID is highly variable based on calculation method and subscale of KOOS (Mills et al. 2016), we compare our data with the MCID of 8–10 recommended by the developers of the KOOS (Roos 2020). In order to address selective loss to follow-up, using a subgroup analysis, patients lost to follow-up at 12 months were compared with the group that completed the follow-up. In another subgroup analysis, patients who returned for a second series of ACP injections after more than 1 year were compared with patients who did not undergo second ACP treatment. Baseline factors were compared between subgroups using t-tests for continuous variables and Pearson’s chi-square for quantitative variables. Correlation between the ΔKOOS5 and sex, age, BMI, Kellgren and Lawrence grade, history of knee trauma, and baseline KOOS5 was assessed using GEE. As a rule of thumb, minimal sample size for a linear model is 10 patients per

745

factor included in the model, therefore a minimum of 120 patients was included. Collinearity was assessed using correlation matrices, linearity using a scatterplot. Variables reaching a p-value lower than 0.2 in the univariate regression were entered in a multivariate regression model. Variables were removed from the multivariate model in order of p-value (highest first). Variables reaching a p < 0.05 in the multivariate model were retained. Ethics, funding, data-sharing, and potential conflicts of interest This study was submitted to the institutional ethical review board of the University Medical Center Utrecht (METC 19-242, 03-04-2019; METC 17-005, 10-01-2017) and was conducted according to the World Medical Association Declaration of Helsinki. Written informed consent was obtained from all individual participants included in the study. This research was supported by the Dutch Arthritis Foundation (LLP-12). The study dataset is available from the corresponding author upon reasonable request. The authors declare that they have no competing interests.

Results Patients (Figure 1) Of all patients, 89% filled out the survey at baseline, 87% at 3 months, 76% at 6 months, and 75% at 12 months’ follow-up. ACP composition Platelet concentration of 28 random anonymous samples of 18 patients was 513 (184) × 109/L, leucocyte concentration was 6.0 (10) × 109/L, and erythrocyte concentration was 0.07 (0.08) × 109/L. The average volume of the injected ACP from which these 28 samples were derived was 4.4 (0.8) mL. Patient-reported outcomes at 3, 6, and 12 months’ follow-up Compared with baseline, KOOS5 increased at 3, 6, and 12 months after treatment (all p < 0.05; Figure 2). There were no statistically significant improvements between the followup-assessments. ΔKOOS5 partially overlapped with the MCID of 8–10 at 3 months (CI 4.9–9.5), 6 months (CI 4.7–11), and 12 months (CI 2.8–9.0) after treatment. At 6 months, 28% of patients reached the MCID of 8 or higher, 23% reached the MCID at 12 months. The change from baseline was comparable and statistically significant in all KOOS subscales (Figure 3). Pain (NRS) decreased from baseline to 3, 6, and 12 months after treatment, but did not improve statistically significant between follow-up assessments. EQ5D was similar in all of the assessments (Table 2). Loss to follow-up At baseline, age, BMI, history of knee trauma, Kellgren

746

Acta Orthopaedica 2020; 91 (6): 743–749

KOOS5 (95% CI)

100

KOOS (mean) 100

0 Baseline

3 months 6 months 12 months

Figure 2. Mean (95% confidence interval) KOOS5 at baseline and after Autologous Conditioned Plasma treatment. KOOS5 is the average of the 5 subscales of the Knee Injury and Osteoarthritis Outcome Score (KOOS).

Symptoms Pain Sport and recreation Knee-related quality-of-life Activities of daily living

Table 2. Patient-reported outcomes after treatment with Autologous Conditioned Plasma. Values are mean (confidence interval). Scale

Baseline

KOOS5 37 (35–39) EQ5D 63 (60–66) NRS 6.2 (5.8–6.5)

0 Baseline

3 months 6 months 12 months

Figure 3. Mean Knee Injury and Osteoarthritis Outcome Score (KOOS) in the subscales pain, symptoms, function in activities of daily living, function in sport and recreation, and kneerelated quality of life at baseline and after treatment with Autologous Conditioned Plasma.

and Lawrence grade, and KOOS5 of patients who were lost to follow-up at 12 months did not differ from patients who completed the follow-up. The group that was lost to followup consisted of more men (64%). At 3 months, patients who were lost to follow-up at 12 months had a ΔKOOS5 of 6.8 (CI 1.9–12), and patients who completed the follow-up had a ΔKOOS5 of 7.2 (CI 4.6–9.8). The missing values in KOOS5 at 12 months were imputed using the values of KOOS5 at 3 months in order to assess the effect of this loss to follow-up. The KOOS5 of the complete dataset, including the imputed data, is 43 (CI 40–46). Second series of ACP injections After more than a year, a second series of ACP injections was given to 31 patients (34 knees). At baseline, these 31 patients did not differ from the others in sex, age, BMI, Kellgren and Lawrence grade, history of knee trauma, and KOOS5. At 6 months, the patients who later returned for a second series had a ΔKOOS5 of 15 (CI 9.4–21), whereas the patients who did not return for a second series of ACP injections had a ΔKOOS5 of 5.4 (CI 2.2–8.6). At 12 months, the patients who returned for a second series of injections had a ΔKOOS5 of 9.5 (CI 4.2–15), and the others had a ΔKOOS5 of 4.7 (CI 0.1–8.2). Linear regression Sex, age, BMI, Kellgren and Lawrence grade, history of knee trauma, and baseline KOOS5 were not associated with clinical outcome (KOOS5) (Table 3). The variables sex, history of knee trauma, baseline KOOS5, and BMI were entered in a multivariate model, but not retained due to a p-value higher than 0.05.

3 months

6 months

12 months

44 (41–47) 45 (41–48) 43 (39–47) 64 (61–68) 67 (63–70) 66 (62–70) 5.3 (4.9–5.7) 5.2 (4.8–5.6) 5.3 (4.8–5.7)

Table 3. Univariate linear regression with coefficients of several factors in the prediction KOOS5 Factor

Generalized estimating equations b (CI) p-value

Age –0.1 (–0.3 to 0.1) Sex (male) –4.0 (–8.6 to 0.7) BMI 0.6 (–0.2 to 1.1) History of traumatic injury a –0.5 (–5.1 to 4.1) –0.1 (–0.3 to 0) KOOS5 at baseline Kellgren and Lawrence grade 0 Reference category 1 –8.2 (–19 to 2.1) 2 –1.5 (–12 to 8.7) 3 –4.5 (–15 to 5.9) 4 0.4 (–13 to 13.1)

0.4 0.1 0.1 0.8 0.1 0.1 0.1 0.8 0.4 0.9

a Meniscus

injury, anterior cruciate ligament rupture, cartilage defect Abbreviations: CI, 95% confidence interval; KOOS5, average of the 5 subscales of the Knee Injury and Osteoarthritis Outcome Score (KOOS).

Discussion In this prospective case series, treatment with intra-articular ACP for knee OA led to a statistically significant, but not clinically relevant, improvement of the KOOS5 after 3, 6, and 12 months’ follow-up. None of the investigated patient factors predicted clinical outcome, in contrast to our hypothesis. The highest change from baseline (ΔKOOS5) was observed at 6 months and did not exceed the MCID for KOOS (Roos 2020). In patients who returned for a second series of ACP injections after 1 year, the ΔKOOS5 exceeded the MCID at 6 months, but decreased at 12 months. 79% of patients did not return for a second series, due to a longer-lasting improvement, or, based on the low ΔKOOS5 in these patients at 6 months, more likely due to insufficient improvement. Poor clinical results were described previously in an RCT using a different PRP composition (Di Martino et al. 2019). After treatment with PRP, no superior clinical improvement was found compared with hyaluronic acid and the improvement in IKDC score (International Knee Documentation Committee) did not reach the MCID (Irrgang et al. 2006). However, reported results of PRP treatment are predominantly good (Shen et al. 2017, Belk et al. 2020) and we expected a higher ΔKOOS5 after treatment.

Acta Orthopaedica 2020; 91 (6): 743–749

An important source of variation and possible explanation for our findings is the different settings in which studies are executed. In an RCT, the efficacy of PRP is investigated under controlled circumstances. The participants are selected in order to minimize comorbidity and the protocol is designed to reach maximal patient and caregiver compliance. In this prospective case series, the effectiveness of PRP was investigated in the setting of daily clinical practice (Haynes 1999, Revicki and Frank 1999) and our real-world data show that ΔKOOS5 does not exceed the MCID. Moreover, the observed improvement might be largely attributable to a placebo effect, as a recent meta-analysis showed that placebo injections can lead to a clinical improvement above the MCID in RCTs (Previtali et al. 2020). The placebo effect in clinical practice might be even larger (Dieppe et al. 2016). Additionally, regression to the mean might contribute to the observed effect in our study, especially since the population is highly selected by inclusion from a tertiary referral center (Morton and Torgerson 2003). Furthermore, difficulty of publication of negative results, especially of non-randomized studies, might lead to publication bias, which is not considered in recently published metaanalyses (Shen et al. 2017, Belk et al. 2020). Lastly, differences in rehabilitation protocols, number of injections, varying composition between different preparations (Fitzpatrick et al. 2017), and administration intervals might influence clinical outcome. This remains a black box for PRP and hampers comparability of studies. The poor results cannot be attributed to the composition of ACP, as the current composition is similar to that reported by Cole et al. (2017) and the manufacturer (Arthrex 2018), with approximately twice the platelet concentration of peripheral blood (Biino et al. 2013) and a leucocyte concentration classified as minimal (Delong et al. 2012). However, we found a notable variability in platelet and leucocyte concentration. In addition, we did not measure concentrations of cytokines and growth factors, which could provide useful information on the bioactivity of ACP. Notable differences in patient populations do exist between our study and other ACP studies. We included patients in a tertiary referral center for joint preservation, with severe complaints and almost 10 points lower baseline KOOS compared with another ACP study (Filardo et al. 2013). This could mean that ACP is not effective in patients with severe complaints, even though our regression analysis indicated that baseline KOOS does not predict clinical outcome. Second, our patient age (mean 49 years) was lower than that in other ACP studies (mean 55–59 years) (Cerza et al. 2012, Filardo et al. 2013, Cole et al. 2017), but we found no effect of age on clinical outcome, similar to the results of Cole et al. (2017). In 2 studies (Filardo et al. 2011, 2013), younger age was even associated with a better outcome. Third, patients with posttraumatic OA were included in our series, while other studies have excluded patients with a history of knee surgery (Cerza et al. 2012) or treatment for a cartilage defect (Smith 2015),

747

but in our case series history of knee trauma did not predict clinical outcome. Lastly, we included 18 patients with bilateral complaints, whereas these patients were excluded in other studies (Smith 2015, Cole et al. 2017). Patients with bilateral complaints have lower physical function and lower probability of improvement than patients with unilateral OA, and PROMs are influenced by contralateral knee pain (White et al. 2010, Riddle and Stratford 2013). To summarize, notable differences exist in patient population, but based on the results of our regression analysis and the small number of patients with bilateral complaints, these differences cannot fully explain our poor clinical outcome. Limitations First, this is a prospective case series, thus lacking a control group. Since previous RCTs showed efficacy of ACP under ideal circumstances, we explicitly chose to investigate effectiveness in clinical practice. As a result, 43 patients received 1 of the intra-articular injections with a 2-week interval, while the others received all injections with a 1-week interval. This might result in variation in effectiveness, which is also a drawback for implementation of PRP in daily practice and could explain the differences between outcomes in RCTs and our clinical data. Second, within this heterogeneous patient population, various patient factors could influence clinical outcome, but limiting our exclusion criteria allowed us to study a population representative of the (heterogeneous) population in our clinical practice and to evaluate the influence of patient factors on treatment outcome. At the same time, the small number of included patients with Kellgren and Lawrence grade 0 and 4 limits generalizations in these groups. Effectiveness will need to be investigated in a larger cohort of patients with early (non-radiographic) or end-stage (grade 4) OA. Third, a relatively large patient group was lost to followup. However, the average KOOS5 did not change substantially when missing data at 12 months were imputed using data at 3 months. We therefore estimate the effect of this loss to followup to be small. Lastly, the MCID recommended for KOOS is 8–10 (Roos 2020), but the MCID in OA patients can actually range between 1.5 and 21 depending on calculation method and KOOS subscales (Mills et al. 2016), and does not account for the invasiveness of the treatment or its placebo effect. An MCID for non-invasive OA therapy should be established in order to determine whether the demonstrated effectiveness reaches a meaningful level for patients. Implications There was no clinically relevant improvement in the majority of patients, nor did most patients return for additional ACP treatment. No predictors of improved clinical outcome were identified. In the limited number of patients who reached the MCID, the effect of ACP decreased between 6 and 12 months, necessitating a second series of treatment after 1 year. In our view, ACP should not be used in daily clinical practice in the

748

current form and population. Future research should try to improve the clinical outcome of this treatment by optimization of the composition of PRP and/or patient selection, before implementation in daily practice. This study demonstrates the gap between efficacy in RCTs and effectiveness in clinical practice, which underlines the importance of evaluating effectiveness after market approval.

The authors would like to thank Emmie Giessen for her help with the surveys. All authors contributed to the study conception and design. Data collection was performed by EK, RC, and NE; data analysis was performed by JK, JA, and TW. The first draft of the manuscript was written by JK and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Acta thanks Stefan Lohmander for help with peer review of this study. Arthrex IR and D. The Arthrex Autologous Conditioned Plasma Double Syringe System . [Internet] 2018 [cited 2020 May 19]. Available from: https://www. arthrex.com/resources/white-paper/sjjfgPkEEeCRTQBQVoRHOw/thearthrex-autologous-conditioned-plasma-double-syringe-system Belk J W, Kraeutler M J, Houck D A, Goodrich J A, Dragoo J L, McCarty E C. Platelet-rich plasma versus hyaluronic acid for knee osteoarthritis: a systematic review and meta-analysis of randomized controlled trials. Am J Sports Med 2020; 1-12. [Online ahead of print] Biino G, Santimone I, Minelli C, Sorice R, Frongia B, Traglia M, Ulivi S, Di Castelnuovo A, Gögele M, Nutile T, Francavilla M, Sala C, Pirastu N, Cerletti C, Iacoviello L, Gasparini P, Toniolo D, Ciullo M, Pramstaller P, Pirastu M, de Gaetano G, Balduini C L. Age- and sex-related variations in platelet count in Italy: a proposal of reference ranges based on 40987 subjects’ data. PLoS One 2013; 8(1): 1-7. Burchard R, Huflage H, Soost C, Richter O, Bouillon B, Graw J A. Efficiency of platelet-rich plasma therapy in knee osteoarthritis does not depend on level of cartilage damage. J Orthop Surg Res 2019; 14(1): 1-6. Cerza F, Carnì S, Carcangiu A, Di Vavo I, Schiavilla V, Pecora A, De Biasi G, Ciuffreda M. Comparison between hyaluronic acid and platelet-rich plasma, intra-articular infiltration in the treatment of gonarthrosis. Am J Sports Med 2012; 40(12): 2822-7. Cole B J, Karas V, Hussey K, Merkow D B, Pilz K, Fortier L A. Hyaluronic acid versus platelet-rich plasma: a prospective, double-blind randomized controlled trial comparing clinical outcomes and effects on intra-articular biology for the treatment of knee osteoarthritis. Am J Sports Med 2017; 45(2): 339-46. de Groot I B, Favejee M M, Reijman M, Verhaar J A N, Terwee C B. The Dutch version of the knee injury and osteoarthritis outcome score: a validation study. Health Qual Life Outcomes 2008; 6: 1-11. Delong J M, Russell R P, Mazzocca A D. Platelet-rich plasma: the PAW classification system. J Arthrosc Relat Surg 2012; 28(7): 998-1009. Dieppe P, Goldingay S, Greville-Harris M. The power and value of placebo and nocebo in painful osteoarthritis. Osteoarthritis Cartilage 2016; 24(11): 1850-7. Di Martino A, Di Matteo B, Papio T, Tentoni F, Selleri F, Cenacchi A, Kon E, Filardo G. Platelet-rich plasma versus hyaluronic acid injections for the treatment of knee osteoarthritis: results at 5 years of a double-blind, randomized controlled trial. Am J Sports Med 2019; 47(2): 347-54. EuroQol Research Foundation. EQ-5D-5L [Internet] 2009 [cited 2017 Dec 18]. Available from: https://euroqol.org/euroqol/

Acta Orthopaedica 2020; 91 (6): 743–749

Fice M P, Miller J C, Christian R, Hannon C P, Smyth N, Murawski C D, Cole B J, Kennedy J G. The role of platelet-rich plasma in cartilage pathology: an updated systematic review of the basic science evidence. J Arthrosc Relat Surg 2019; 35(3): 961-76.e3. Filardo G, Kon E, Buda R, Timoncini A, Di Martino A, Cenacchi A, Fornasari P M, Giannini S, Marcacci M. Platelet-rich plasma intra-articular knee injections for the treatment of degenerative cartilage lesions and osteoarthritis. Knee Surg Sports Traumatol Arthrosc 2011; 19(4): 528-35. Filardo G, Kon E, Di Matteo B, Di Martino A, Sessa A, Merli M L, Marcacci M. Leukocyte-poor PRP application for the treatment of knee osteoarthritis. Joints 2013; 1(3): 112-20. Filardo G, Di Matteo B, Di Martino A, Merli M L, Cenacchi A, Fornasari P, Marcacci M, Kon E. Platelet-rich plasma intra-articular knee injections show no superiority versus viscosupplementation: a randomized controlled trial. Am J Sports Med 2015; 43(7): 1575-82. Fitzpatrick J, Bulsara M K, McCrory P R, Richardson M D, Zheng M H. Analysis of platelet-rich plasma extraction: variations in platelet and blood components between 4 common commercial kits. Orthop J Sport Med 2017; 5(1): 1-8. Forogh B, Mianehsaz E, Shoaee S, Ahadi T, Raissi G R, Sajadi S. Effect of single injection of platelet-rich plasma in comparison with corticosteroid on knee osteoarthritis: a double-blind randomized clinical trial. J Sports Med Phys Fitness 2016; 56(7-8): 901-8. Fortier L A, Barker J U, Strauss E J, McCarrel T M, Cole B J. The role of growth factors in cartilage repair. Clin Orthop Relat Res 2011; 469(10): 2706-15. Gawaz M, Vogel S. Platelets in tissue repair: control of apoptosis and interactions with regenerative cells. Blood 2013; 122(15): 2550-4. Gobbi A, Lad D, Karnatzikos G. The effects of repeated intra-articular PRP injections on clinical outcomes of early osteoarthritis of the knee. Knee Surg Sports Traumatol Arthrosc 2015; 23(8): 2170-7. Haynes B. Can it work? Does it work? Is it worth it? Br Med J 1999; 319(7211): 652-3. Irrgang J J, Anderson A F, Boland A L, Harner C D, Neyret P, Richmond J C, Shelbourne K D, Bergfeld J, Feagin J, Fulkerson J, Kocher M, Howell S, Staubli H, Hefti F, Hoher J, Jacob R, Mueller W, Chan K M, Kurosaka M. Responsiveness of the International Knee Documentation Committee Subjective Knee Form. Am J Sports Med 2006; 34(10): 1567-73. Jubert N J, Rodríguez L, Reverté-Vinaixa M M, Navarro A. Platelet-rich plasma injections for advanced knee osteoarthritis: a prospective, randomized, double-blinded clinical trial. Orthop J Sport Med 2017; 5(2): 1-11. Lana J F S D, Weglein A, Sampson S E, Vicente E F, Huber S C, Souza C V, Ambach M A, Vincent H, Urban-Paffaro A, Onodera C M K, AnnichinoBizzacchi J M, Santana M H A, Belangero W D. Randomized controlled trial comparing hyaluronic acid, platelet-rich plasma and the combination of both in the treatment of mild and moderate osteoarthritis of the knee. J Stem Cells Regen Med 2016; 12(2): 69-78. Lin K Y, Yang C C, Hsu C J, Yeh M L, Renn J H. Intra-articular injection of platelet-rich plasma is superior to hyaluronic acid or saline solution in the treatment of mild to moderate knee osteoarthritis: a randomized, doubleblind, triple-parallel, placebo-controlled clinical trial. J Arthrosc Relat Surg 2019; 35(1): 106-17. LROI. NRS pain 2018. [Internet] [cited 2020 May 27]. Available from: https:// ww w.lroi.nl/base/downloads/proms-vragenlijst-knie.pdf Mills K A G, Naylor J M, Eyles J P, Roos E M, Hunter D J. Examining the minimal important difference of patient-reported outcome measures for individuals with knee osteoarthritis: a model using the knee injury and osteoarthritis outcome score. J Rheumatol 2016; 43(2): 395-404. Morton V, Torgerson D J. Effect of regression to the mean on decision making in health care. Br Med J 2003; 326(7398): 1083-4. Nurden A T. Platelets, inflammation and tissue regeneration. Thromb Haemost 2011; 105(Suppl. 1): 13-33. Patel S, Dhillon M S, Aggarwal S, Marwaha N, Jain A. Treatment with plateletrich plasma is more effective than placebo for knee osteoarthritis: a prospective, double-blind, randomized trial. Am J Sports Med 2013; 41(2): 356-64.

Acta Orthopaedica 2020; 91 (6): 743–749

Previtali D, Merli G, Di Laura Frattura G, Candrian C, Zaffagnini S, Filardo G. the long-lasting effects of “placebo injections” in knee osteoarthritis: a meta-analysis. Cartilage 2020; Mar 18. Online ahead of print. Revicki D A, Frank L. Pharmacoeconomic evaluation in the real world: effectiveness versus efficacy studies. Pharmacoeconomics 1999; 15(5): 423-34. Riddle D L, Stratford P W. Unilateral vs bilateral symptomatic knee osteoarthritis: associations between pain intensity and function. Rheumatol (UK) 2013; 52(12): 2229-37. Roos E. KOOS FAQ. [Internet] 2020 [cited 2020 Mar 31]. Available from: http://www.koos.nu/koosfaq.html Sánchez M, Fiz N, Azofra J, Usabiaga J, Aduriz Recalde E, Garcia Gutierrez A, Albillos J, Gárate R, Aguirre J J, Padilla S, Orive G, Anitua E. A randomized clinical trial evaluating plasma rich in growth factors (PRGF-

749

Endoret) versus hyaluronic acid in the short-term treatment of symptomatic knee osteoarthritis. J Arthrosc Relat Surg 2012; 28(8): 1070-8. Shen L, Yuan T, Chen S, Xie X, Zhang C. The temporal effect of platelet-rich plasma on pain and physical function in the treatment of knee osteoarthritis: systematic review and meta-analysis of randomized controlled trials. J Orthop Surg Res 2017; 12(1): 1-12. Smith P A. Intra-articular autologous conditioned plasma injections provide safe and efficacious treatment for knee osteoarthritis. Am J Sports Med 2015; 44(4): 884-91. White D, Zhang Y, Felson D, Niu J, Keysor J, Nevitt M, Lewis C, Torner J, Neogi T. The independent effect of pain in one versus two knees on the presence of low physical function: the MOST Study. Arthritis Care Res 2010; 62(7): 938–43.

750

Acta Orthopaedica 2020; 91 (6): 750–755

Does intraoperative contamination during primary knee arthroplasty affect patient-reported outcomes for patients who are uninfected 1 year after surgery? A prospective cohort study of 714 patients Tobias JUSTESEN a, Jakob B OLSEN a, Anne B HESSELVIG, Anne MØRUP-PETERSEN, and Anders ODGAARD

Department of Orthopedic Surgery, Copenhagen University Hospital Herlev-Gentofte, Copenhagen, Denmark a Shared first authorship   Correspondence: tobiasjust@msn.com Submitted 2020-03-20. Accepted 2020-08-04.

Background and purpose — It is well recognized that some knee arthroplasty (KA) patients present with prolonged postoperative inflammation and some develop persistent pain. It can reasonably be speculated that some of these problems develop because of low-grade infections with low virulence bacteria caused by intraoperative contamination. This prospective study was performed to investigate whether intraoperative contamination results in lower patient-reported outcomes (PRO) for patients who were clinically uninfected in the first year after surgery. Patients and methods — We combined data from 2 major prospective studies on patients undergoing primary KA at 2 Danish hospitals between September 2016 and January 2018. Pre- and postoperative (1.5, 3, 6, and 12 months) PROs and intraoperative microbiological cultures were obtained on a total of 714 patients who were included in the study. Based on the microbiological cultures, the patients were divided into 2 groups, contaminated and non-contaminated, and differences in PROs between the 2 groups were analyzed. Results — 84 of 714 (12%) patients were intraoperatively contaminated; none of the 714 patients developed clinical infection. The preoperative Oxford Knee Score was 24 and 23 for contaminated and non-contaminated patients, respectively, improving to 40 and 39 at 1 year (p = 0.8). 1-year AUC for Oxford Knee Score and absolute improvement at each postoperative time point for Forgotten Joint Score and EQ-5D-5L also were similar between contaminated and noncontaminated patients. Interpretation — Patient-reported outcomes from 714 patients do not indicate that intraoperative contamination affects the knee-specific or general health-related quality of life in primary KA patients who are clinically uninfected 1 year after surgery.

Intraoperative bacterial contamination occurs in up to onefourth of joint replacement procedures (Byrne et al. 2007, Font-Vizcarra et al. 2011, Frank et al. 2011, Lindeque et al. 2014, Hesselvig et al. 2020). Periprosthetic joint infections (PJIs) causing clinical signs occur in approximately 1% of patients and it is suspected that most contaminations are of no clinical importance. Some microorganisms remain viable and dormant in biofilms with a potential of eliciting an inflammatory response leading to tissue destruction and pain (Zimmerli et al. 2004, Arnold et al. 2013, Antony and Farran 2016). It has been shown that bacteria are indeed present on a high percentage of implants (Jakobsen et al. 2018) and it has been speculated that they can be responsible for cases of aseptic loosening (Ribera et al. 2014, Rothenberg et al. 2017). The immediate postoperative inflammatory response associated with recovering from the surgical injury (Bilgen et al. 2001) may be potentiated by the inflammatory response caused by bacterial contamination. It is well recognized that some patients present with prolonged postoperative inflammation, some develop persistent pain, and others develop a swollen and stiff joint and it can reasonably be speculated that some of these problems develop because of low-grade infections due to bacterial contamination. Our hypothesis is that intraoperative contamination that does not result in an acute or delayed infection will result in a prolonged inflammatory response that causes increased discomfort and prolonged rehabilitation, which will be reflected in patient-reported outcomes (PROs). To our knowledge only 1 study (Ibrahim et al. 2011) has looked into the relationship between intraoperative contamination and PROs. However, that study investigated patients undergoing hip arthroplasty and did not have any data for the first 8 postoperative years. We investigated whether intraoper-

Acta Orthopaedica 2020; 91 (6): 750–755

ative contamination in knee arthroplasties without subsequent clinical infection results in lower Oxford Knee Score (OKS), lower Forgotten Joint Score (FJS), or lower EQ-5D-5L score in the first year after surgery.

Patients and methods The study design was according to STROBE guidelines. This prospective cohort study combines data from 2 prospective multicenter studies, ICON (Hesselvig et al. 2020) and SPARK (Mørup-Petersen et al., personal communication). These 2 studies enrolled patients from multiple hospitals over a 2-year period (2016–2018) with a partially shared patient cohort. The patients included in this study were enrolled in both the ICON and SPARK studies at 2 high-volume knee arthroplasty centers (Copenhagen University Hospital Gentofte and Aarhus University Hospital) between September 1, 2016 and January 1, 2018. The ICON study included 1,187 patients who underwent primary knee arthroplasties. Patients were instructed to shower preoperatively on the day of the procedure using a normal body wash and no moisturizer afterwards. Local guidelines did not include any pre-admission skin or nasal decontamination. Before surgery, all patients were disinfected twice using a 0.5% chlorhexidine gluconate solution with 80% alcohol. Afterwards, around half of the patients (603 of 1,187) were draped with an antimicrobial incision drape (Ioban2, 3M Health Care, St. Paul, MN, USA). All surgeons routinely use 2 pairs of gloves (inner and outer gloves). According to local guidelines the surgeons changed the outer gloves after preparation of the surgical field, prior to handling the prosthesis, and when using bone cement. All cases of cemented knee arthroplasty were done with cement containing antibiotics (i.e., gentamycin); no vancomycin was placed within the knee/wound. All operations were performed in laminar airflow operation rooms and all patients were given prophylactic antibiotics consisting of either dicloxacillin 2 g or cefuroxime 1.5 g, depending on the hospital routine and allergy status of the patient. This information was not recorded by individual patient. Information on age, operation date, sex, location (left or right), and duration of surgery was collected. Exposure data, i.e., intraoperative contamination, was obtained by 2 dry wound swabs (Copan Diagnostics Inc. Murrieta, CA, USA) from each patient and a wash from the surgeon’s glove during surgery. Both swabs were taken of the lateral wound edge, the first just after incision and the second swab just prior to closure of the skin. Approximately 30 minutes into the operation, prior to handling the prosthesis and possibly bone cement, the surgeon changed the outer gloves. The glove from the dominant hand was turned inside out and washed with 10 mL of isotonic saline. The samples were cultured according to Danish guidelines and susceptibility tested using Eucast breakpoints (eucast.org 2019). Identification

751

was done using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Maldi-Tof, Bruker Daltonics, Hamburg, Germany). Contamination was defined as any amount of bacterial growth from 1 or more of the swabs from either the wound edge or the surgeon’s glove, no matter the type of bacteria. The SPARK study was an observational cohort study of 1,452 patients who underwent primary knee arthroplasty surgery. The patients completed a set of PRO questionnaires preoperatively and at 1.5, 3, 6, and 12 months postoperatively, sent by either email or letter (Procordo Software, Copenhagen, Denmark). The PRO set included OKS, FJS, and EQ5D-5L. The inclusion and exclusion criteria are described in Table 1 (see Supplementary data). The following outcomes were investigated: 1-year AUC for OKS changes from baseline, absolute differences in OKS and EQ-5D-5L between baseline and 1.5, 3, 6, and 12 months after surgery, and differences in absolute postoperative scores for FJS at 3, 6, and 12 months. The modified version of OKS was applied (0–48, 48 best). The FJS comprises 12 items (total score 0–100) with higher scores reflecting better outcomes. The EQ-5D-5L consists of the EQ visual analogue scale (VAS) and the EQ-5D-5L descriptive system. The EQ VAS records the patients’ self-rated health on a scale ranging from 0 to 100, 100 being the best. The EQ-5D-5L comprises 5 dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. The results from the 5 dimensions were converted into index values (–0.22 to 1, 0 corresponding to death, negative numbers health states worse than death, and 1 being perfect health) based on data from the Danish population (Janssen et al. 2013). To determine whether any absolute difference in changes from baseline in OKS between the contaminated and non-contaminated patients was likely to be perceived as relevant by the patients, the minimally important difference (MID) was used. The MID between the responses “a little better” and “about the same” was found to be 5 OKS points in a study by Beard et al. (2015). Our study is considered blinded because the patients were not informed whether they were contaminated or not. Statistics Patients with missing postoperative data at 12 months or missing data at more than 2 time points were excluded from the analysis of AUC. AUC was calculated for PRO changes from baseline using the trapezium rule (Matthews et al. 1990) as an overall time-adjusted measure for changes in OKS. The x-axis (Figure 1) indicates time in months and the y-axis is a normalized PRO measure (0–1, no dimension), giving AUC the dimension “time (months)” (Odgaard et al. 2018). Since an MID for AUC has not been suggested, AUC data is presented as an equivalent measure of “gained months with optimal OKS” (value 48) for each group. When calculating the AUC for OKS we used linearly interpolated values for missing data at the time points 1.5, 3, and 6 months. We analyzed a

752

Acta Orthopaedica 2020; 91 (6): 750–755

OKS

Normalized PRO measure

1.00

Mean improvement

Eligible knee arthroplasties n = 1,499 Excluded (n = 766): – Patients not included in the SPARK study, 361 – Patients not included in the ICON study, 405

0.75

AUC 24

Baseline value

Patients with complete SPARK and ICON data n = 733

0.25

0 Preop. 1.5

0.50

B 3

AUC = area A = area B

Excluded (n = 19): – Patients diagnosed with PJI after inclusion, 12 – Patients who underwent knee revision surgery within the first year due to other reasons than PJI after inclusion, 7

0.00

Months from index operation

Figure 1. Example of area under the curve for a random patient in the study. The AUC is the blue area above the baseline. Full circles indicate values of OKS from 20 at baseline (preoperative value) to 39 at 12 months postoperatively. The AUC is the same size as rectangle A or B. A represents the average improvement in OKS during the first postoperative year by the left y-axis. B represents a translation of A into months with optimal (value 48) OKS.

subgroup of contaminated patients with 2 positive cultures. Analyses were done using a univariable model and multiple linear regression models when adjusting for confounders. The analyses included adjustment of the parameter estimates for differences in the distributions of sex, age, type of prosthesis, and duration of surgery. These variables are known to influence either contamination (Byrne et al. 2007, Hesselvig et al. 2020) or PRO improvements (Weber et al. 2018, Tolk et al. 2019), and based on clinical experience they may reasonably be suspected of being confounders. None of the variables can induce bias when adjusting for these, as neither exposure nor outcome can affect the variables (i.e., they cannot be either a mediator or collider) (Shrier and Platt 2008). P-values < 0.05 were considered statistically significant. Confidence intervals (CI) are defined as 95%. The analyses were performed using the SAS Enterprise Guide (version 7.15 HF3, SAS Institute, Cary, NC, USA). Ethics, registration, funding, and potential conflicts of interest Ethical approval was provided by the Regional Committee of Health Research Ethics (September 2, 2016, Jr. No. H-15012754) and data management was approved by the Danish Data Protection Agency (August 1, 2016, Jr. No. HGH-2016-087, I-Suite no: 04819). Permission to use the EQ-5D-5L questionnaires was given by the EuroQol Research Foundation (January 17, 2019, ID number 28583). All included participants gave informed consent. The SPARK study was funded by the Health Research Fund of the Capital Region of Denmark and the ICON study was funded by 3M Health Care and the University of Copenhagen. However, this particular study did not receive any specific grant from funding agencies in the public, commercial, or notfor-profit sectors. Furthermore, no sponsors were involved in conduct of the research or preparation of the article.

Patients included in the study n = 714

Figure 2. Flow diagram of the inclusion process. PJI = periprosthetic joint infection

AO is paid speaker by Stryker and DePuy, paid consultant by Stryker and DePuy, receives research support by ZimmerBiomet, Stryker, and DePuy, and is Chairman for Danish Knee Arthroplasty Register.

Results At the start of analyzing data for this study May 2019, 1,499 patients were included in either the SPARK study, the ICON study, or both studies (Figure 2). 766 patients were only included in either the ICON or SPARK study due to different enrollment centers and enrollment periods and were thus excluded. 19 patients were excluded due to PJIs or revision surgery. 2 of the 12 patients excluded due to PJIs were intraoperatively contaminated. 1 patient was contaminated with Micrococcus species while joint fluid and biopsy at revision surgery showed Streptococcus dysgalactiae. The other patient was contaminated with Staphylococcus capitis and epidermidis, and joint fluid and biopsy at revision surgery revealed Staphylococcus epidermidis. None of the 7 patients who underwent revision surgery, for reasons other than PJIs, were intraoperatively contaminated. Furthermore, none of the intraoperative biopsies from the revisions, which were done on 4 of the patients on the slightest suspicion of infection, revealed any positive culture. The reasons for revision surgery were: rupture of the posterior cruciate ligament, medial tibial plateau fracture, instability, loosening of the prosthesis, progression of arthrosis, and in 2 cases pain and instability. A sufficient PRO sequence and contamination data were available for 714 patients (389 women and 325 men), who were included in the final analysis. The types of knee arthroplasties included total knee arthroplasty (n = 510), medial unicompartmental knee arthroplasty (n = 170), lateral unicompartmental knee arthroplasty (n = 6), and patellofemoral arthroplasty (n = 28). 12% (84) of the

Acta Orthopaedica 2020; 91 (6): 750–755

753

Table 2. Contaminating organisms and contaminated samples Sample Lateral Lateral wound wound edge after Surgeon’s edge prior Organism incision glove to closure Coagulase-negative staphylococci 23 19 36 Micrococcus 3 1 6 Streptococcus 5 0 2 Gram-positive rods 4 2 3 Gram-negative rods 1 0 1 Staphylococcus aureus 0 0 0

Table 3. Univariable analysis of patients regarding negative vs. positive intraoperative cultures. Values are mean (standard deviation) unless other wise specified Variables

Intraoperative cultures negative positive n = 630 n = 84

Age 68 (9) Women, n (%) 352 (56) BMI 29 (5) Duration of surgery, min 65 (15) Total knee arthroplasty, n (%) 445 (71) Baseline OKS 23 (7) Baseline EQ-5D-5L index value 0.6 (0.1) Baseline EQ VAS 62 (22)

67 (9) 37 (44) 29 (5) 69 (15) 65 (77) 24 (6) 0.6 (0.1) 62 (20)

p-value 0.8 0.04 0.9 0.01 0.2 0.5 0.9 0.7

BMI = body mass index; OKS = Oxford Knee Score (0–48).

Table 4. Univariable analysis: measures of change in Oxford Knee Score for contaminated vs. non-contaminated patients. Values are number of patients and absolute difference unless otherwise specified and (95% confidence interval)

1.5 months

Non-contaminated 586 4.2 (3.6–4.8) Contaminated 77 4.2 (2.7–5.7) p-value 1.0 a b

3 months 580 10.2 (9.6–10.9) 73 10.2 (8.3–12.0) 1.0

6 months 580 13.7 (13.1–14.4) 75 13.0 (11.0–15.1) 0.5

1 year 570 15.6 (15.0–16.3) 78 15.4 (13.6–17.1) 0.8

AUC 1 year a 544 2.9 (2.7–3.0) b 70 2.8 (2.4–3.2) b 0.7

1-year area under the curve for the difference in OKS from baseline; number of gained months with optimal (48) OKS.

patients were intraoperatively contaminated, 1.1% (8) had 2 or more contaminated samples, and 1.3 % (9) were contaminated by more than one organism. Coagulase-negative staphylococci were the most common contaminating organisms (Table 2). The patients had a mean age of 68 years (SD 9, range 28–93) and mean BMI was 29 (SD 5, range 18–52). The mean duration of surgery was 66 minutes (SD 15, range 30–130). Baseline data for age, BMI, OKS, EQ-5D-5L index value, EQ VAS, and type of prosthesis were similar between the contaminated and non-contaminated groups. The groups differed slightly regarding sex and duration of surgery (Table 3), i.e., male sex and longer operation time were associated with increasing contamination. The patient-reported outcomes OKS, FJS, and EQ-5D-5L did not differ statistically significantly between the contaminated and non-contaminated groups when analyzed by AUC, absolute values at any of the postoperative time points, or by absolute differences from baseline (Tables 4, 5, and 6 [for Tables 5 and 6, see Supplementary data]). Mean 1-year OKS for the contaminated and non-contaminated groups was 40 (SD 6) and 39 (SD 8), respectively. The effect sizes of contamination on absolute differences in OKS between baseline and 1.5, 3, 6, and 12 months after surgery ranged from 0.01 to 0.70 (CI –2.0 to 2.7). All effect sizes were statistically insignificant and the range of the effect sizes was not greater than the minimally important difference (MID) of 5.

Furthermore, all outcomes were assessed by multiple linear regression models adjusting for sex, age, type of prosthesis, and duration of surgery. All p-values were still insignificant (range 0.4–1.0) and results were consistent with the unadjusted values. Patients with 2 or more positive samples were analyzed using a univariable model and compared with the non-contaminated patients. The effect size of contamination regarding 1-year AUC (OKS) was 0.3 (CI –0.9 to 1.5), 0.5 to 0.9 (CI –5.3 to 7.1) regarding OKS changes between baseline and 1.5, 3, 6, and 12 months after surgery, and –1.1 to 7.3 (CI –19 to 26) regarding FJS at 3, 6, and 12 months. The same subgroup of patients was analyzed using a multiple linear regression model as well, which revealed results consistent with the univariable analyses. None of the analyses showed any statistically significant difference between the subgroup and the non-contaminated group.

Discussion Some primary knee arthroplasty patients experience prolonged postoperative inflammation, persistent pain, or a swollen and stiff joint. Our hypothesis was that some of these problems develop because of bacterial contamination, but which has not resulted in a clear-cut clinical infection. To our knowledge, this hypothesis has not been tested previously.

754

There is no standardized way of collecting data on intraoperative contamination during knee surgery. Similar studies have used 1 to 3 swab samples collected from a range of locations such as knife blades, suction tips, suture lines, the subcutaneous tissue when closing, fluid residues, and the splash basin (Byrne et al. 2007, Frank et al. 2011, Fuchs et al. 2018). The swab cultures used in this study are less sensitive in detecting intraoperative contamination than for example tissue samples (Aggarwal et al. 2013). This study does not account for any contamination that might occur postoperatively through the non-healed wound or by bacteremia (Zimmerli 2006). Only 2 of 12 patients excluded due to PJIs were intraoperatively contaminated, while the bacteria found during revision surgery in 1 of these cases matched the intraoperative contamination. These results are to be seen in relation to the above-mentioned limitations and the fact that no former studies have been able to prove a correlation between intraoperative contamination and subsequent infection (Davis et al. 1999, Byrne et al. 2007, Jonsson et al. 2014). None of the symptoms in 7 patients, who underwent revision surgery for reasons other than PJI, can be readily explained by bacteria, since none of them were intraoperatively contaminated and in the 4 cases where intraoperative biopsies from the revisions were done, none of the biopsies revealed any positive culture. The contamination rate of 12% is within the range found in similar studies (Byrne et al. 2007, Font-Vizcarra et al. 2011, Frank et al. 2011, Lindeque et al. 2014) and in between the 10% (use of antimicrobial drape) and 15% (no use of antimicrobial drape) found in the study by Hesselvig et al. (2020) (for 363 of the 714 included patients antimicrobial drapes were used). Average age and sex distribution of included patients were comparable to those reported in the Danish Knee Arthroplasty Register (DKR), which has a completion rate of 95%. The analyses of the patient-reported outcomes were based on the AUC analysis of OKS and complementing analyses of all PROs (OKS, FJS, EQ-5D-5L index value, and EQ VAS) at different sequential time points. All PRO results at baseline and postoperatively in our study were of the same magnitude as those found in similar knee arthroplasty studies (Nerhus et al. 2012, Hamilton et al. 2017, Bilbao et al. 2018, Odgaard et al. 2018). We found similar PRO scores in the contaminated and non-contaminated groups. Thus, the enhanced inflammatory response that intraoperative contamination hypothetically could cause was not severe or prolonged enough to significantly potentiate the general postoperative inflammatory response and to be reflected in the PROs. Our results do not support the hypothesis that intraoperative contamination, not resulting in an acute or delayed infection, would lead to increased discomfort in the first postoperative year. The results in our study are in line with a former casecontrol study (Ibrahim et al. 2011), which did not find a correlation between contamination of the femoral head during hip replacement surgery and the patient-reported outcome measures Oxford Hip Score and EQ-5D. Since no other stud-

Acta Orthopaedica 2020; 91 (6): 750â&#x20AC;&#x201C;755

ies have investigated whether intraoperative contamination during knee or hip surgery is associated with lower PROs, our study brings new and needed data on the patient consequences of intraoperative contamination in arthroplasty surgery. The cohort of 714 patients who underwent primary knee arthroplasty surgery will be further followed with the purpose of investigating the possible association of intraoperative contamination with late infections, aseptic loosening of prostheses, and revision surgery for reasons other than infections and loosening. So far, no such associations have been established within the first year after surgery. In summary, patient-reported outcomes from 714 patients do not indicate that intraoperative contamination affects the knee-specific or general health-related quality of life in primary knee arthroplasty patients who are clinically uninfected within the first year after surgery. Supplementary data Tables 1, 5, and 6 as supplementary data in the online version of this article, http://dx.doi.org/ 10.1080/17453674.2020.1811552

TJ and JBO: Wrote the manuscript, performed the formal analysis, and contributed to the investigation. ABH and AM collected data and supervised the data analysis. AO conceptualized the study and contributed to supervision and the formal analysis. All the authors contributed to editing and revision of the manuscript. Acta thanks Ricardo Sousa and Rihard Trebse for help withÂ peer review of this study.

Aggarwal V K, Higuera C, Deirmengian G, Parvizi J, Austin M S. Swab cultures are not as effective as tissue cultures for diagnosis of periprosthetic joint infection. Clin Orthop Relat Res 2013; 471(10): 3196-203. doi: 10.1007/s11999-013-2974-y. Antony S, Farran Y. Prosthetic joint and orthopedic device related infections: the role of biofilm in the pathogenesis and treatment. Infect Disord Drug Targets 2016; 16(1): 22-7. Arnold W V, Shirtliff M E, Stoodley P. Bacterial biofilms and periprosthetic infections. J Bone Joint Surg Am 2013; 95(24): 2223-9. Beard D J, Harris K, Dawson J, Doll H, Murray D W, Carr A J, Price A J. Meaningful changes for the Oxford hip and knee scores after joint replacement surgery. J Clin Epidemiol 2015; 68(1): 73-9. doi: 10.1016/j. jclinepi.2014.08.009. Bilbao A, Garcia-Perez L, Arenaza J C, Garcia I, Ariza-Cardiel G, TrujilloMartin E, Forjaz M J, Martin-Fernandez J. Psychometric properties of the EQ-5D-5L in patients with hip or knee osteoarthritis: reliability, validity and responsiveness. Qual Life Res 2018; 27(11): 2897-908. doi: 10.1007/ s11136-018-1929-x. Bilgen O, Atici T, Durak K, Karaeminogullari, Bilgen M S. C-reactive protein values and erythrocyte sedimentation rates after total hip and total knee arthroplasty. J Int Med Res 2001; 29(1): 7-12. doi: 10.1177/147323000102900102. Byrne A M, Morris S, McCarthy T, Quinlan W, Oâ&#x20AC;&#x2122;Byrne J M. Outcome following deep wound contamination in cemented arthroplasty. Int Orthop 2007; 31(1): 27-31. doi: 10.1007/s00264-006-0121-z.

Acta Orthopaedica 2020; 91 (6): 750â&#x20AC;&#x201C;755

Davis N, Curry A, Gambhir A K, Panigrahi H, Walker C R, Wilkins E G, Worsley M A, Kay P R. Intraoperative bacterial contamination in operations for joint replacement. J Bone Joint Surg Br 1999; 81(5): 886-9. doi: 10.1302/0301-620x.81b5.9545. eucast.org. The European Committee on Antimicrobial Susceptibility Testing. Breakpoint tables for interpretation of MICs and zone diameters. Version 9.0, 2019. http://www.eucast.org; 2019. Font-Vizcarra L, Tornero E, Bori G, Bosch J, Mensa J, Soriano A. Relationship between intraoperative cultures during hip arthroplasty, obesity, and the risk of early prosthetic joint infection: a prospective study of 428 patients. Int J Artif Organs 2011; 34(9): 870-5. doi: 10.5301/ ijao.5000026. Frank C B, Adams M, Kroeber M, Wentzensen A, Heppert V, Schulte-Bockholt D, Guehring T. Intraoperative subcutaneous wound closing culture sample: a predicting factor for periprosthetic infection after hip- and kneereplacement? Arch Orthop Trauma Surg 2011; 131(10): 1389-96. doi: 10.1007/s00402-011-1327-0. Fuchs M, von Roth P, Pfitzner T, Kopf S, Sass F A, Hommel H. Contamination of irrigation fluid during primary total knee arthroplasty. J Am Acad Orthop Surg Glob Res Rev 2018; 2(5): e027. doi: 10.5435/JAAOSGlobalD-17-00027. Hamilton D F, Loth F L, Giesinger J M, Giesinger K, MacDonald D J, Patton J T, Simpson A H, Howie C R. Validation of the English language Forgotten Joint Score-12 as an outcome measure for total hip and knee arthroplasty in a British population. Bone Joint J 2017; 99-B(2): 218-24. doi: 10.1302/0301-620X.99B2.BJJ-2016-0606.R1. Hesselvig A B, Arpi M, Madsen F, Bjarnsholt T, Odgaard A, Group I S. Does an antimicrobial incision drape prevent intraoperative contamination? A randomized controlled trial of 1187 patients. Clin Orthop Relat Res 2020; 478(5): 1007-15. doi: 10.1097/CORR.0000000000001142. Ibrahim T, Aswad M G, Dias J J, Brown A R, Esler C N. Long-term outcome of total hip replacement in patients with or without femoral head contamination. J Orthop Surg (Hong Kong) 2011; 19(2): 174-6. doi: 10.1177/230949901101900208. Jakobsen T H, Eickhardt S R, Gheorghe A G, Stenqvist C, Sonderholm M, Stavnsberg C, Jensen P O, Odgaard A, Whiteley M, Moser C, Hvolris J, Hougen H P, Bjarnsholt T. Implants induce a new niche for microbiomes. APMIS 2018; 126(8): 685-92. doi: 10.1111/apm.12862. Janssen M F, Pickard A S, Golicki D, Gudex C, Niewada M, Scalone L, Swinburn P, Busschbach J. Measurement properties of the EQ-5D-5L compared to the EQ-5D-3L across eight patient groups: a multi-country study. Qual Life Res 2013; 22(7): 1717-27. doi: 10.1007/s11136-012-0322-4.

755

Jonsson E O, Johannesdottir H, Robertsson O, Mogensen B. Bacterial contamination of the wound during primary total hip and knee replacement: median 13 years of follow-up of 90 replacements. Acta Orthop 2014; 85(2): 159-64. doi: 10.3109/17453674.2014.899848. Lindeque B, Hartman Z, Noshchenko A, Cruse M. Infection after primary total hip arthroplasty. Orthopedics 2014; 37(4): 257-65. doi: 10.3928/0147744720140401-08. Matthews J N, Altman D G, Campbell M J, Royston P. Analysis of serial measurements in medical research. BMJ 1990; 300(6719): 230-5. Nerhus T K, Heir S, Svege I, Skramm I, Jervidalo T, Madsen J E, Ekeland A. Time-dependent improvement in functional outcome following Oxford medial unicompartmental knee arthroplasty: a prospective longitudinal multicenter study involving 96 patients. Acta Orthop 2012; 83(1): 46-52. doi: 10.3109/17453674.2011.652890. Odgaard A, Madsen F, Kristensen P W, Kappel A, Fabrin J. The Mark Coventry Award: Patellofemoral arthroplasty results in better range of movement and early patient-reported outcomes than TKA. Clin Orthop Relat Res 2018; 476(1): 87-100. doi: 10.1007/s11999.0000000000000017. Ribera A, Morata L, Moranas J, Agullo J L, Martinez J C, Lopez Y, Garcia D, Cabo J, Garcia-Ramiro S, Soriano A, Murillo O. Clinical and microbiological findings in prosthetic joint replacement due to aseptic loosening. J Infect 2014; 69(3): 235-43. doi: 10.1016/j.jinf.2014.05.003. Rothenberg A C, Wilson A E, Hayes J P, Oâ&#x20AC;&#x2122;Malley M J, Klatt B A. Sonication of arthroplasty implants improves accuracy of periprosthetic joint infection cultures. Clin Orthop Relat Res 2017; 475(7): 1827-36. doi: 10.1007/ s11999-017-5315-8. Shrier I, Platt R W. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008; 8:70. doi: 10.1186/1471-2288-8-70. Tolk J J, Janssen R P A, Haanstra T M, van der Steen M M C, Bierma Zeinstra S M A, Reijman M. Outcome expectations of total knee arthroplasty patients: the influence of demographic factors, pain, personality traits, physical and psychological status. J Knee Surg 2019. doi: 10.1055/s-00391692632. Weber M, Craiovan B, Woerner M L, Schwarz T, Grifka J, Renkawitz T F. Predictors of outcome after primary total joint replacement. J Arthroplasty 2018; 33(2): 431-5. doi: 10.1016/j.arth.2017.08.044. Zimmerli W. Infection and musculoskeletal conditions: prosthetic-joint-associated infections. Best Pract Res Clin Rheumatol 2006; 20(6): 1045-63. doi: 10.1016/j.berh.2006.08.003. Zimmerli W, Trampuz A, Ochsner P E. Prosthetic-joint infections. N Engl J Med 2004; 351(16): 1645-54. doi: 10.1056/NEJMra040181.

756

Acta Orthopaedica 2020; 91 (6): 756–760

Increase in early wound leakage in total knee arthroplasty with local infiltrative analgesia (LIA) that includes epinephrine: a retrospective cohort study Babette C VAN DER ZWAARD, Ramon L ROERDINK, and Ruud P VAN HOVE

Department of Orthopedics, Jeroen Bosch Hospital, ’s Hertogenbosch, The Netherlands Correspondence: b.v.d.zwaard@jbz.nl Submitted 2020-02-03. Accepted 2020-08-10.

Background and purpose — After introducing a new local infiltration anesthesia (LIA) protocol with addition of 30 mL ropivacaine 2% and 1 mg epinephrine, we noted an increase in early wound leakage. As wound leakage is associated with prosthetic joint infection, our department aims to minimize postoperative wound leakage. This study evaluates the incidence of early wound leakage and postoperative pain after knee arthroplasty (KA) following adjustment of the LIA protocol with addition of 30 cc ropivacaine 2% and 1 mg epinephrine. Patients and methods — In this retrospective medical dossier study all patients (n = 502) undergoing a primary total or unicondylar knee arthroplasty between January 1, 2018 and July 1, 2019 were included. Patients received an LIA protocol containing 120 mL 2 mg/mL ropivacaine (ROPI– group; n = 256). After October 30, patients received an LIA protocol containing 150 mL 2 mg/mL ropivacaine with 1 mg epinephrine in the first 100 mL (ROPI+ group; n = 246). The primary outcome measure was early wound leakage (< 72 hours postoperatively), defined as wound fluid leaking past the barrier of the wound dressing. Secondary outcome measure, 10-point numeric rating scale (NRS) pain (< 72 hours postoperatively) was also assessed. Data was evaluated using logistic regression. Results — The incidence of wound leakage was higher in the ROPI+ group: 24% versus 17% in the ROPI– group (p = 0.06). After adjusting for the differences between surgeons the relative risk of this increase was 1.4 (1.0–2.0). The ROPI+ and ROPI– group were similar regarding postoperative pain assessment. Interpretation — Adjustment of the LIA protocol with 30 mL 2% ropivacaine and 1 mg epinephrine led to an increase in early wound leakage in knee arthroplasty but no difference in pain scores.

Local infiltration anesthesia (LIA) during knee arthroplasty (KA) has shown positive results on pain control, early mobilization, hospital discharge, and reduced opiate use (Vendittoli et al. 2006, Toftdahl et al. 2007, Andersen et al. 2008, GómezCardero and Rodríguez-Merchán 2010, Affas et al. 2011). Several years ago, LIA for KA was introduced in our hospital. Amongst changes in preoperative, peroperative, and postoperative medication, 120 mg of 2 mg/mL ropivacaine was infiltrated in the periarticular tissues during surgery. Pain is considered adequately treated by Dutch quality parameters when the numeral rating score for pain (NRS) is below 4 in 90% of patients during the first 72 hours after surgery (Nederlandse Vereniging voor Anesthesiologie [NVA] 2012). An analysis of our postoperative clinical data showed that this quality parameter could not be achieved with this pain management protocol. Therefore, the LIA protocol was updated to the current Dutch guideline for LIA, in which 150 mL 2 mg/ mL ropivacaine with 1 mg of epinephrine was used to infiltrate the periarticular tissues (NVA 2012). KA can be associated with complications such as wound leakage (Saleh et al. 2002, Patel et al. 2007, Kremers et al. 2019). Early wound leakage, during the first 72 hours after surgery, is generally accepted as normal (Kremers et al. 2019). Prolonged wound leakage can result in delayed wound healing, delayed mobilization, patient dissatisfaction, prolonged hospital stay, and increased costs and has been strongly associated with prosthetic joint infections (PJI) (Kurtz et al. 2008, Bozic et al. 2010, Rietbergen et al. 2016). After adjustment of the LIA protocol in near conformation with the national guideline, an increase in the number of patients with postoperative wound leakage was observed at our department. The main aim of this retrospective medical dossier study was to evaluate the effect of both LIA protocols on early wound leakage and if a difference was found to assess

Acta Orthopaedica 2020; 91 (6): 756–760

whether there was a difference in postoperative pain scores between the LIA protocols.

Patients and methods Study design To evaluate the effect of change in LIA protocol on early wound leakage, this retrospective medical dossier study was conducted. On October 30, 2018 the LIA protocol was adjusted for all knee arthroplasty (KA) patients, which includes both total knee arthroplasties and unicondylar knee arthroplasties. Incidence of early wound leakage was compared for KA patients operated on between January 1 and October 29, 2018 with those operated on between October 30, 2018 and July 1, 2019. Participants Data from primary KA patients operated on between January 1, 2018 and July 1, 2019 were included in the study. Patients operated on before October 30, 2018 were included in the ROPI– group (ROPI–) while those operated on after October 29 were included in the ROPI+ group (ROPI+). There were no exclusion criteria. Data was collected from the electronic medical records. Oral pain medication protocol Preoperatively, on the day of surgery, 1,000 mg of acetaminophen, 75 mg of pregabalin, 500 mg naproxen, and 20 mg omeprazol were administered orally. Postoperatively, 1,000 mg of acetaminophen was continued 4 times daily, as well as 75 mg pregabalin twice daily, and 500 mg naproxen twice daily, for 72 hours postoperatively if necessary. Controlledrelease oxycodone 5–10 mg twice daily, and immediate release oxycodone 5 mg, 6 times daily if necessary, were also provided for postoperative pain control. Omeprazole 20 mg once daily, granisetron 10 mg thrice daily if necessary and Movicolon (macrogol) 13.8 g once or twice daily were administered to address side effects of pain medication. Local infiltration analgesia In the ROPI– group, 120 mL 2 mg/mL ropivacaine was infiltrated in the periarticular tissues, after the trial components of the total knee prosthesis were removed and before the tourniquet was inflated. The ropivacaine was equally distributed over the posterior and anterior capsule and the subcutaneous tissue around the wound. In the ROPI+ group, 100 mL 2 mg/ mL ropivacaine with 1 mg epinephrine was infiltrated in the periarticular tissues, as follows: 10 mL for the medial and lateral condyle, 20 mL for medial and lateral posterior capsule, 10 mL on the medial side of the proximal tibia, 10 mL for the patellar ligament, and 10 mL along the distal and proximal quadriceps tendon. 50 mL ropivacaine 2 mg/mL was used to infiltrate the subcutaneous tissue around the wound.

757

Surgery KA was performed either under spinal or general anesthesia by 1 of 5 knee surgeons. Preoperative antibiotics, 2 g cefazoline iv, were administered as well as 8 mg dexamethasone iv and 1,000 mg tranexamic acid iv, prior to the incision. After a medial parapatellar approach, a cemented total knee prosthesis with a fixed bearing (cruciate retained or posterior stabilized, Vanguard, Zimmer Biomet, Warsaw, IN, USA; n = 398 and n = 33 respectively) was placed. In the case of UKA, a cemented unicondylar knee prosthesis with a mobile bearing (Oxford, Zimmer Biomet, Warsaw, IN, USA; n = 71) was placed. A tourniquet was inflated to 250 mm Hg before the components were cemented up to application of the pressure bandage, for approximately 30 minutes. The skin was closed in 4 layers as described elsewhere (Roerdink et al. 2019) After wound closure whilst still in the sterile field the wound was dressed with a 9×30 cm Aquacel surgical dressing (Convatec, Greensboro, NC, USA). A pressure bandage was applied for 24 hours. Anticoagulants During hospital stay after surgery, patients were treated with low molecular weight heparin, subcutaneous nadroparine 0.3 mL. Acetylsalicylic acid and carbasalate calcium were only discontinued on the day of surgery and restarted on the first postoperative day. Other antiplatelet therapy and coumarin therapy were discontinued 7 days before surgery and restarted on the 3rd postoperative day only when the wound was dry. Therapeutic direct oral anticoagulants were discontinued. Measurements and data sources The outcome measure is early wound leakage, defined as wound fluid leaking past the barrier of the wound dressing during the first 72 hours after surgery and is defined dichotomously: early wound leakage yes/no. Excessive wound leakage will pass through the sides of the dressing and this is noted in the electronic medical record (EMR). For the search within the EMR for the outcome measure early wound leakage, we used CTCue (CTcue B.V., Amsterdam, https://ctcue.com/). This software can data mine through EMR text and search for keywords regarding wound leakage: wound leaking, wound leakage, leaking wound, dressing saturated, dressing changed. For each of the 3 postoperative days the EMR of each patient was text mined for these keywords. EMRs not containing 1 of the keywords were searched through manually. Pain was verbally assessed thrice daily by the nursing staff using a numeric rating scale (NRS) ranging from 0 = no pain at all to 10 = most pain imaginable. The results are noted in the EMR of the patient. As NRS pain was a non-normally distributed variable we dichotomized; ≤ 3 and ≥ 4 (van Dijk et al. 2012, Boonstra et al. 2016). Variables such as smoking, diabetes, use of anticoagulants, and surgeon are associated with wound infection. Because early wound leakage might be a symptom of wound infection (Patel et al. 2007, Kremers et al. 2019), these variables were assessed as possible confound-

758

Acta Orthopaedica 2020; 91 (6): 756–760

I Pain

Surgeon´s technique

Use of anticoagulants

Smoking

Diabetes

▲

Exposure

Outcome Ancestor of outcome

Vasoconstriction/ –dilatation

Coagulation ability

Atherosclerosis

Ancestor of exposure and outcome Unobserved (latent) Causal path

▲

Direct acyclic graph describing the hypothesized causal pathways of the selected possible confounders.

LIA (± epinephrine)

Wound healing

Table 1. Patient characteristics. Values are number (%) unless otherwise specified

Control group Study group n = 251 n = 232

Mean age (SD) a Female sex b Smoking b Use anticoagulants b Diabetes b Early wound leakage b a

69 (8.8) 154 (61) 34 (14) 80 (32) 47 (19) 42 (17)

69 (8.6) 134 (58) 11 (5) 81 (35) 36 (16) 55 (24)

p-value 0.5 0.4 0.001 0.5 0.4 0.06

Biasing path

Postoperative wound leakage

Table 2. Results of log-binomial models of the ROPI+ protocol versus ROPI– protocol Factor Early wound leakage—crude model Early wound leakage—adjusted for surgeon NRS pain assessment

RR (95% CI) 1.4 (0.99–2.0) 1.4 (1.0–2.0) 0.95 (0.9–1.0)

RR = relative risk; CI = confidence interval; NRS = numerical rating scale.

T-test; b chi-square test.

ers. Smoking, having diabetes, and use of anticoagulants are all assessed during the preoperative screening and recorded in the EMR of each patient. Study size and statistics Rule of thumb is to have a minimum of 10–15 cases per variable. An incidence of 24% early wound leakage was found in preliminary data assessment of the ROPI– group. During the inclusion periods 502 patients received a KA. Therefore, this sample had the required power for testing 5 variables, i.e., 1 outcome measure and 4 possible confounders. To evaluate whether type of LIA protocol had an effect on early wound leakage a regression log-binomial model was used. Pain was assessed thrice daily; six consecutive moments of pain assessment were evaluated using generalized estimating equations— a log-binomial model. The confounder selection was based on the directed acrylic graph (DAG) approach (Shrier and Platt 2008). Possible confounders were selected based on literature on wound infection (Patel et al. 2007, Kremers et al. 2019) and hypothesized relationships. A model of the hypothesized relationships between exposure (type of LIA protocol), primary outcome variable (early wound leakage), and possible confounders (smoking, diabetes, use of coagulants, and surgeon) was created. The final model was chosen after expert-based adjustments and constructed using the online DAGitty (version 3.0) software (Figure) (Textor et al. 2016). Based on the final model, “surgeon” was added to the model as a confounder. 5 orthopedic

surgeons performed KA; this ordinal variable was recoded into dummy variables, all of which were evaluated as confounders. If the primary regression coefficient changed > 10% “surgeon” was kept in the final model as confounder. All analyses were performed with SPSS (IBM SPSS Statistics for Windows, Version 25.0; IBM Corp, Armonk, NY, USA). Ethics, funding, and potential conflicts of interest This study lies outside the scope of the Medical Research involving Human Subject Act (WMO), as declared by the Medical Ethical Board Brabant (NW2019-49). The study was not funded by external parties. The authors declare that there are no conflicts of interest regarding the publication of this study.

Results Participants The cohort consisted of 502 patients, 256 in the ROPI– group and 246 in the ROPI+ group (Table 1). The percentage of smokers was significantly higher in the ROPI– group: 14% versus 5% in the ROPI+ group (p < 0.001). None of the other baseline characteristics were found to statistically significantly differ between groups. Main results The incidence of wound leakage during the first 72 hours after KA surgery was 17% in the ROPI– group compared with

Acta Orthopaedica 2020; 91 (6): 756–760

24% in the ROPI+ group (p = 0.06). When the outcome was adjusted for the differences found between surgeons, the RR of this increase was 1.4 (95% CI 1.0–2.0) (Table 2). Neither smoking nor diabetes was found to confound the difference in wound leakage between the groups. As expected, the use of anticoagulants did significantly increase the chance of early wound leakage after KA; RR 1.9 (CI 1.4–2.5). The use of anticoagulants did not, however, confound the difference in wound leakage between the ROPI– and ROPI+ group. Postoperative NRS ≥ 4 were similar between the ROPI+ and ROPI– group (Table 2).

Discussion On October 30, 2018 the LIA protocol for KA surgeries was adjusted to be in line with the guideline on postoperative pain management from the Dutch Society of Anesthesiology (NVA) because of inadequate pain management (NVA 2012). 30 mL ropivacaine 2 mg/mL and 1 mg epinephrine were added to the existing 120 mL ropivacaine 2 mg/mL. This transition of LIA protocols has led to an increase in the incidence of wound leakage during the first 72 hours after surgery. The protocols did not differ in pain evaluation. The Dutch guideline for LIA after TKA was constructed in 2013 (NVA 2012) and the advice to add 1.5 mg of epinephrine to the 150 mL of ropivacaine was based on expert opinion. When adjusting our LIA protocol, it was hypothesized that the vasoconstrictive effect might inhibit wound healing. Therefore, the new protocol describes the addition of 1 mg epinephrine to the first 100 mL for the periarticular tissue, but not to the 50 mL used for subcutaneous tissue. In the subsequent years after publication of the guideline, several studies with varying LIA protocols have been published (Karlsen et al. 2017). However, due to the heterogeneity of the protocols, it is not possible to determine an optimal dose and drug regimen based on the current evidence. Our study is the first to assess the effect of 2 different LIA protocols on early onset wound leakage. We found a higher risk of developing wound leakage during the first 3 days after surgery in the ROPI+ group after adjusting for the differences between surgeons. A study comparing the use of ropivacaine with and without epinephrine on postoperative pain perception found “major wound leakage” as adverse event in the epinephrine group (Schotanus et al. 2017). In the epinephrine group 3 out of 25 patients had major wound leakage versus none in the group that received LIA with ropivacaine only. However, the study was underpowered for wound leakage as outcome. It is unclear what was considered “major wound leakage,” but it does seem to corroborate our findings. The causal pathway in which the use of epinephrine could lead to an increased incidence of wound leakage is unclear. One characteristic of epinephrine is that it leads to vasoconstriction. A study on the scalp has shown that the duration of vasoconstriction can vary,

759

depending on the dosage (Na et al. 2016). So one explanation could be that wound leakage increases when the constrictive effect ceases. Another study showed that the incidence of patients with a rise of < 20% of their baseline systolic blood pressure was higher at the moment of release of the tourniquet when epinephrine was used in LIA during KA (Yoo et al. 2020). Although they used a different dosage epinephrine (0.6 mg) and a different consistency of the LIA (ropivacaine 180 mg, morphine sulfate 5 mg, ketorolac 30 mg, cefazolin 1 g, and methylprednisolone 40 mg) (Yoo et al. 2020), the increased blood pressure could account for the increased incidence of wound leakage. We evaluated 6 consecutive pain assessments with a longitudinal evaluation between the 2 groups and no differences were found between the 2 protocols. Schotanus at al. (2017) found no differences between an 150 mL ropivacaine protocol with or without epinephrine either, which would endorse our findings. Limitations The main limitation of this study is its retrospective nature, in which it is impossible to distinguish between the 2 differences between protocols: the added epinephrine and 30 mL extra ropivacaine. An assumption could be that the extra 30 mL injected LIA could leak out of the wound in its entirety, leading to the increased incidence. In our opinion this is unlikely. The Aquacel surgical bandage used for our wound care is able to absorb 77.5 g/24 hours of wound fluid (1 mL water = 1 g) (absorption information obtained through personal communication with Convatec, C. Lindsay, November 27, 2019). In this study, a surgical incision has to lose more than approximal 77.5 mL/24 hours to be considered as a leaking wound since we defined it as: “wound fluid leaking past the barrier of the wound dressing during the first 72 hours after surgery.” Therefore, it would be likely that the primary reason for our findings is the added epinephrine and not the 30 mL extra ropivacaine. Another limitation of the study is that we found the surgeon to confound the result. The “surgeon” involves a whole set of small actions and proceedings that might lead to differences between surgeons and it is unclear which of these might lead to differences between surgeons in postoperative wound leakage. This presents an interesting opportunity for future research, but was not feasible to address in the current study. In this study we have done the utmost to evaluate possible confounders. However, due to the retrospective nature of the study and the lack of studies with wound leakage as outcome variables it is feasible that possible confounders have not been assessed or are currently unknown. A clinical relevant outcome measure would be differences in incidence of PJIs between the 2 ROPI protocols. However, the incidence of PJI in this cohort was 0.6% (n = 3), 2 in the ROPI– and 1 in the ROPI+ group. This low incidence does not allow any clinical conclusions nor statistical analyses.

760

Conclusion Our results suggest that an LIA protocol of 150 mL ropivacaine with 1 mg epinephrine increases the incidence of early wound leakage after KA compared with a 120 mL ropivacaine protocol without epinephrine. Our findings combined with other studies suggest that the addition of epinephrine to LIA protocols for KA surgeries should be reevaluated. Concept and design: BvdZ, RvH; data acquisition: BvdZ, RR; data analysis: BvdZ; drafting the manuscript: BvdZ, RR, RvH. Affas F, Nygårds E-B, Stiller C-O, Wretenberg P, Olofsson C. Pain control after total knee arthroplasty: a randomized trial comparing local infiltration anesthesia and continuous femoral block. Acta Orthop 2011; 82(4): 441-7. Andersen L Ø, Husted H, Otte K S, Kristensen B B, Kehlet H. High-volume infiltration analgesia in total knee arthroplasty: a randomized, double-blind, placebo-controlled trial. Acta Anaesthesiol Scand 2008; 52(10): 1331-5. Boonstra A M, Stewart R E, Köke A J A, Oosterwijk R F A, Swaan J L, Schreurs K M G, Schiphorst Preuper H R. Cut-off points for mild, moderate, and severe pain on the numeric rating scale for pain in patients with chronic musculoskeletal pain: variability and influence of sex and catastrophizing. Front Psychol. Frontiers Media SA; 2016; 7: 1466. Bozic K J, Kurtz S M, Lau E, Ong K, Chiu V, Vail T P, Rubash H E, Berry D J. The epidemiology of revision total knee arthroplasty in the United States. Clin Orthop Relat Res 2010; 468(1): 45-51. Gómez-Cardero P, Rodríguez-Merchán E C. Postoperative analgesia in TKA: ropivacaine continuous intraarticular infusion. Clin Orthop Relat Res 2010; 468(5): 1242-7. Karlsen A P H, Wetterslev M, Hansen S E, Hansen M S, Mathiesen O, Dahl J B. Postoperative pain treatment after total knee arthroplasty: a systematic review. PLoS One 2017; 12(3): e0173107. Kremers K, Leijtens B, Camps S, Tostmann A, Koëter S, Voss A. Evaluation of early wound leakage as a risk factor for prosthetic joint infection. J Am Assoc Nurse Pract 2019; 31(6): 337-43. Kurtz S M, Lau E, Schmier J, Ong K L, Zhao K, Parvizi J. Infection burden for hip and knee arthroplasty in the United States. J Arthroplasty 2008; 23(7): 984-91.

Acta Orthopaedica 2020; 91 (6): 756–760

Na Y-C, Park R, Jeong H-S, Park J H. Epinephrine vasoconstriction effect time in the scalp differs according to injection site and concentration. Dermatologic Surg 2016; 42(9): 1054-60. Nederlandse Vereniging voor Anesthesiologie (NVA). Richtlijn Postoperatieve pijn. Utrecht: NVA; 2012. Patel V P, Walsh M, Sehgal B, Preston C, DeWal H, Di Cesare P E. Factors associated with prolonged wound drainage after primary total hip and knee arthroplasty. J Bone Joint Surg Am 2007; 89(1): 33-8. Rietbergen L, Kuiper J W P, Walgrave S, Hak L, Colen S. Quality of life after staged revision for infected total hip arthroplasty: a systematic review. HIP Int 2016; 26(4): 311-18. Roerdink R L, Plat A W, van Hove R P, Leenders A C A P, van der Zwaard B C. Reduced wound leakage in arthroplasty with modified wound closure: a retrospective cohort study. Arch Orthop Trauma Surg 2019; 139(11): 1505-10. Saleh K, Olson M, Resig S, Bershadsky B, Kuskowski M, Gioe T, Robinson H, Schmidt R, McElfresh E. Predictors of wound infection in hip and knee joint replacement: results from a 20 year surveillance program. J Orthop Res 2002; 20(3): 506-15. Schotanus M G M, Bemelmans Y F L, van der Kuy P H M, Jansen J, Kort N P. No advantage of adrenaline in the local infiltration analgesia mixture during total knee arthroplasty. Knee Surgery, Sport Traumatol Arthrosc 2017; 25(9): 2778-83. Shrier I, Platt R W. Reducing bias through directed acyclic graphs. BMC Med Res Methodol; 2008; 8: 70. Textor J, van der Zander B, Gilthorpe M S, Liśkiewicz M, Ellison G T. Robust causal inference using directed acyclic graphs: the R package “dagitty.” Int J Epidemiol 2016; 45(6): 1887-94. Toftdahl K, Nikolajsen L, Haraldsted V, Madsen F, Tønnesen E K, Søballe K. Comparison of peri- and intraarticular analgesia with femoral nerve block after total knee arthroplasty: a randomized clinical trial. Acta Orthop 2007; 78(2): 172-9. van Dijk J F, Kappen T H, van Wijck A J, Kalkman C J, Schuurmans M J. The diagnostic value of the numeric pain rating scale in older postoperative patients. J Clin Nurs 2012; 21(21-22): 3018-24. Vendittoli P-A, Makinen P, Drolet P, Lavigne M, Fallaha M, Guertin M-C, Varin F. A multimodal analgesia protocol for total knee arthroplasty: a randomized, controlled study. J Bone Joint Surg Am 2006; 88(2): 282-9. Yoo S, Chung J-Y, Ro D H, Han H-S, Lee MC, Kim J-T. The hemodynamic effect of epinephrine-containing local infiltration analgesia after tourniquet deflation during total knee arthroplasty: a retrospective observational study. J Arthroplasty 2020; 35(1): 76-81.

Acta Orthopaedica 2020; 91 (6): 761–769

761

Tibial lengthening using a retrograde magnetically driven intramedullary lengthening device in 10 patients with preexisting ankle and hindfoot fusion Bjoern VOGT 1, Robert ROEDL 1, Georg GOSHEGER 2, Gregor TOPOROWSKI 1, Andrea LAUFER 1, Christoph THEIL 2, Jan Niklas BROEKING 1, and Adrien FROMMER 1 1 Children’s

Orthopedics, Deformity Reconstruction and Foot Surgery, University Hospital of Muenster; 2 General Orthopedics and Tumor Orthopedics, University Hospital of Muenster, Germany Correspondence: bjoern.vogt@ukmuenster.de Submitted 2020-04-30. Accepted 2020-07-11.

Background and purpose — Motorized intramedullary lengthening nails (ILNs) have been developed as an alternative to external fixators for long bone lengthening. The antegrade approach represents the standard method for tibial ILN insertion. In patients with preexisting ankle and hindfoot fusion a retrograde approach provides an alternative technique that has not been evaluated so far. We report the outcome of this method in 10 patients. Patients and methods — This retrospective study included 10 patients (mean age 18 years [13–25]) with preexisting ankle and hindfoot fusion who underwent tibial lengthening with a retrograde ILN (PRECICE). The mean leg length discrepancy (LLD) was 58 mm (36–80). The underlying conditions were congenital (n = 9) and post tumor resection (n = 1). The main outcome measures were: ILN reliability, distraction achieved, distraction index (DIX), time to bone healing, consolidation index (CIX), complications, and functional results. Results — All patients achieved the goal of lengthening (mean 48 mm [26–80]). Average DIX was 0.6 mm/day (0.5– 0.7) and mean CIX was 44 days/cm (26–60). Delayed consolidation occurred in 2 patients and healed after ILN dynamization or nail exchange with grafting. Toe contractures in 2 other patients were resolved with physiotherapy or tenotomy. Until last follow-up (mean 18 months [12–30]) no true complications were encountered, knee motion remained unaffected, and full osseous consolidation occurred in all patients. Interpretation — In patients with LLD and preexisting ankle and hindfoot fusion distal tibial lengthening using a retrograde ILN is a reliable alternative to the standard approach with equivalent bone healing potential and low complication rates leaving the knee unaffected.

Fully implantable intramedullary lengthening nails (ILNs) with mechanical (Guichet and Casar 1997, Cole et al. 2001) and motorized (Baumgart et al. 1997, Schiedel et al. 2014) drive systems have been developed as an alternative to external fixators for bone lengthening (Mahboubian et al. 2012, Black et al. 2015, Laubscher et al. 2016). Recently, magnetically driven ILNs in particular have become increasingly popular (Kirane et al. 2014, Wagner et al. 2017) and in contrast to external fixation provide an equally safe and more comfortable option for limb lengthening and deformity correction (Szymczuk et al. 2019, Horn et al. 2019). Antegrade or retrograde femoral and antegrade tibial lengthening with the PRECICE limb lengthening system (NuVasive, San Diego, CA, USA) has been assessed by several studies (Kirane et al. 2014, Schiedel et al. 2014, Shabtai et al. 2014, Tiefenboeck et al. 2016, Wiebking et al. 2016, Fragomen and Rozbruch 2017, Wagner et al. 2017, Iobst et al. 2018, Cosic and Edwards 2020, Nasto et al. 2020). In tibial lengthening the antegrade approach represents the standard method for ILN implantation (Fragomen and Rozbruch 2017). In patients with preexisting ankle and hindfoot fusion a retrograde approach provides an alternative technique for tibial nail insertion. Approach-associated affections of the knee joint like anterior knee pain (Rothberg et al. 2019) and—in immature patients—damage to the proximal tibial growth plate (Wagner et al. 2017, Frommer et al. 2018) can be avoided. Despite these potential advantages, the use of a retrograde tibial nailing approach and distal tibial osteotomy in patients with preexisting ankle and hindfoot fusion has not been evaluated so far.

762

Acta Orthopaedica 2020; 91 (6): 761–769

Table 1. Patient data—preoperative parameters Months No. of Previous Ankle and between Preop. Patient previous Previous deformity hindfoot fusion LLD Persistent no. Sex Side Underlying etiology surgeries a lengthening correction fusion and ILN (mm) deformity 1 F Left Tumor resection > 5 1, femur 1, tibia Prior 2-stage, 16 44 No (osteosarcoma) (ILN) (nail) (nail) 2 F Right Congenital 1 No No Prior 2-stage, 6 36 No clubfoot (screws) 3 F Left Pes calcaneus in 1 No No Prior 2-stage, 12 38 No caudal regression (nail) syndrome 4 F Right Fibular hemimelia > 5 1, femur 2, femur Ankylosis N/a 55 Distal valgus (ILN) (1 GG, 1 ILN) and flexion 1, tibia (exFix) 5 F Right Fibular hemimelia > 5 1, tibia 2, tibia Prior 2-stage, 16 60 No (exFix) (1 exFix, 1 nail) (nail) 6 M Right Fibular hemimelia > 5 2, tibia 3, tibia 1-stage N/a 72 Distal (exFix) (2 exFix, 1 GG) (ILN) flexion 7 F Right Nail-patella 1 No No Prior 2-stage, 62 60 No syndrome (screws) 8 M Right Tibial hemimelia 1 1, tibia 1, tibia Prior 2-stage, 60 55 No (exFix) (exFix) (exFix) 9 M Right Fibular hemimelia > 5 1, femur 3, femur Prior 2-stage, 5 75 Mid-shaft (exFix) (1 exFix, 2 GG) (exFix) valgus 2, tibia 3, tibia (exFix) (2 exFix, 1 GG) 10 M Right Tibial hemimelia > 5 1, tibia 3, tibia Prior 2-stage, 10 80 No (exFix) (1 exFix, 1 nail, (nail) 1 GG) a in

affected leg LLD—leg length discrepancy, ILN—intramedullary lengthening nail, exFix—external fixator, GG—growth guidance, N/a—not applicable.

Patients and methods Patients and indications We performed a retrospective analysis of patients treated with retrogradely implanted magnetically driven ILNs for tibial lengthening between 2015 and 2019. 10 patients (6 females) with preexisting ankle and hindfoot fusion (right = 8) were treated for correction of leg-length discrepancy (LLD). The underlying conditions were congenital (n = 9) and post tumor resection (n = 1) (Table 1). Ankle and hindfoot fusion was established by arthrodesing operations either precedingly in 8 patients or 1-stage with the ILN insertion in 1 individual. A congenital ankylosis was present in 1 patient (Figure 1). The mean preoperative LLD was 58 mm (36–80). 3 patients presented additional persistent angular deformities of the tibia (Nos 4, 6, and 9). The mean period between prior arthrodesis and intramedullary lengthening procedure was 23 months (5–62) (Table 1). The mean age at the date of ILN implantation was 18 years (13–25). The planned distraction distance averaged 49 mm (26–80) and thus was 9.0 mm (0–22) shorter than the LLD (Table 2).

Figure 1. Different etiologies of ankle and hindfoot fusion in patients who were considered for tibial lengthening with retrograde ILN: A— congenital ankylosis (Patient No. 9) , B—post tumor resection, subsequent segment transport, and docking with hindfoot nail (Patient No. 1), C—corrected deformity with screw arthrodesis in congenital clubfoot (Patient No. 2), D –fusion achieved with prior external fixation in tibial hemimelia (Patient No. 8).

Acta Orthopaedica 2020; 91 (6): 761–769

763

Table 2. Patient data—perioperative parameters Age at Patient surgery no. (years)

ILN diameter/ /stroke Concomitant Osteotomy Deformity (mm) operations level (mm) correction

Surgery Fluortime oscopy Intra- Planned cut to suture time operative distraction (min) (min) injuries (mm)

1 21.3 8.5/50 Implant removal 82.9 from prior surgeries (meta-diaphyseal) No 2 16.9 10.7/50 Implant removal from 84.8 prior surgeries (diaphyseal) No 3 18.0 8.5/50 Implant removal 62.8 from prior surgeries (meta-diaphyseal) No 4 17.0 10.7/50 No 107.6 Single osteotomy, (diaphyseal) distal varization and extension 5 16.1 8.5/80 Implant removal 66.9 from prior surgeries, (meta-diaphyseal) No tibiotalar bone grafting, screw osteosynthesis 6 13.3 10.7/50 No 52.3 Single osteotomy, (meta-diaphyseal) distal extension 7 20.9 8.5/50 No 80.6 (meta-diaphyseal) No 8 24.9 8.5/50 No 52.1 (meta-diaphyseal) No 9 17.7 10.7/80 Plate stabilization of 70.4 Second osteotomy, second osteotomy (meta-diaphyseal) mid-shaft varization 10 14.7 10.7/80 Epiphysiodesis of 112.2 proximal fibula (diaphyseal) No

127

3.0

156

0.9

105 126

2.5 3.0

No No

33 40

141

1.9

1.4

2.7

4.7 5.1

No No

50 65

1.1

128 167 91

Implants and surgical technique The second generation PRECICE (P2) limb-lengthening system (NuVasive, San Diego, CA, USA) was used in all individuals (Table 2, Figures 3–4, for Figures 2 and 5, see Supplementary data). The patients were placed in a supine position on a radiolucent surgical table. All nails were implanted retrogradely according to the preoperative planning (Figure 2, see Supplementary data). The corticotomy was performed on the distal tibia with a multiple drill hole technique and subsequent chiseling. In patients with distal malalignment (Nos 4 and 6) a single osteotomy was carried out on the apex deformity for realignment and callus distraction. In the patient with mid-shaft valgus deformity (No. 9) a second osteotomy was executed on the apex of the deformity for angular correction in addition to a distal corticotomy for callus distraction. The mid-shaft osteotomy was bridged by the ILN and additionally fixed using a 4-hole 3.5 mm locking plate (VariAx, Stryker, Kalamazoo, MI, USA) to prevent proximal distraction (Figure 4). If present, the fibula was osteotomized at the border from the proximal to the distal third (n = 7). Adequate function of the implanted ILN was fluoroscopically proven by intraoperative distraction of 1 mm. In 6 patients 7 concomitant operations were performed (Table 2).

The radiographs were taken using the Centricity PACS calibrated digital radiology system (GE Healthcare, Chalfont St Giles, UK). All radiographic measurements were performed with the TraumaCad (Brainlab, Munich, Germany) post-processing software. Anteroposterior (AP) long standing radiographs were obtained preoperatively and at last follow-up in order to assess LLD and coronal alignment of the lower extremity (Paley 2002) (Figures 3 and 4). In immature patients (Nos 6 and 10) final LLD was predicted using the multiplier method (Paley et al. 2000).

Pre- and postoperative clinical and radiographic evaluation The patients underwent clinical and radiographic evaluations preoperatively and periodically after surgery until final follow-up.

Outcome parameters Perioperative parameters Implant type, surgery time including concomitant operations (cut–suture), fluoroscopy time, and amount of intraoperative blood loss were acquired from the surgery protocol (Table 2).

Postoperative lengthening and follow-up protocol Distraction commenced after a latency period of 10 days with an initial distraction speed of 0.66 mm/day in all patients. The follow-up protocol involved clinical and radiographic examinations every 2 weeks during the lengthening period and every 6 weeks during consolidation. The patients were allowed 20 kg partial weight-bearing during distraction and after osseous consolidation full weight-bearing was permitted. All patients were treated with physiotherapy at least once a week during the lengthening period. The average postoperative follow-up was 18 months (12–30).

764

Acta Orthopaedica 2020; 91 (6): 761–769

Figure 3. A—Patient No. 3 with pes calcaneus and leg shortening due to caudal regression syndrome. B—Prior to lengthening the foot deformity was corrected by ankle and subtalar arthrodesis using a retrograde nail. C—After ankle and hindfoot fusion the residual LLD measured 38 mm. D—Tibial distraction osteogenesis of 31 mm using a retrograde magnetically driven ILN. E—Postoperative long standing radiograph after full consolidation. F—Postoperative long standing radiograph after ILN removal 26 months postoperatively showing intended residual LLD of 7 mm.

Figure 4. Patient No. 9 with additional osteotomy for angular correction. A—Preoperative radiographs showing LLD of 75 mm and mid-shaft tibial valgus malalignment. B—A second osteotomy was performed on the apex of the deformity for angular correction in addition to a distal corticotomy for callus distraction. The osteotomy level (OL) is measured as the distance from the tibial osteotomy site (drawn as a line parallel to the nail) to a perpendicular line aligned tangentially with the most plantar aspect of the calcaneus on the lateral radiograph (70.4 mm, meta-diaphyseal).The mid-shaft osteotomy was bridged by the ILN and additionally fixed using a locking plate to prevent proximal distraction. C—Full consolidation of the mid-shaft osteotomy and beginning callus consolidation after distraction of 65 mm. D—Postoperative radiograph showing a residual LLD of 11 mm and correction of the tibial valgus malalignment.

Acta Orthopaedica 2020; 91 (6): 761–769

Functional outcome The presence of postoperative pain was assessed. Postoperative functional results were evaluated using the AOFAS ankle– hindfoot scale. Limb-lengthening parameters The osteotomy level (OL) was measured on lateral radiographs and categorized by the corresponding anatomical bone segment (Figure 4B). The amount of distraction achieved was calculated subtracting the length of the initially exposed male portion of the nail from the length of the extended male portion at the end of distraction. Accuracy, precision, and reliability were calculated as previously described (Schiedel et al. 2014, Wagner et al. 2017). The distraction index (DIX) was determined by dividing the achieved length (mm) by the duration of lengthening (days). The consolidation index (CIX) was calculated by dividing the time from surgery until full weight-bearing was possible (days) by the distraction length (cm). Delayed osseous consolidation (non-union) was defined as absent bone healing 6 months after the end of distraction. Limb alignment parameters Post-surgical bone results were documented according to Chappell et al. (2019). This includes the ankle and hindfoot alignment grading as proposed by Katsenis et al. (2005) and the tibial alignment grading as proposed by Schoenleber and Hutson (2015). Difficulties In accordance with Paley (1990) difficulties during the lengthening procedure were subclassified into problems, obstacles, and true complications. Ethics, funding and potential conflicts of interest The study was approved by the ethical committee of the university of Muenster on July 1, 2019 (registration number: 2019-368-f-S) and was fully financed by the research funds of the University Hospital of Muenster, Germany. BV and RR are paid consultants of NuVasive (San Diego, USA) and have received remuneration during the study period outside the submitted work. The other authors have no conflict of interest.

Results Surgical procedure and perioperative parameters No intraoperative injuries were recorded. The mean surgery and fluoroscopy time including concomitant operations was 121 minutes (83–167) and 2.6 minutes (0.9–5.1), respectively, with an average documented blood loss of 258 mL (80–500). No perioperative blood transfusion was necessary. Corticotomies were conducted 7 times on the metadiaphyseal and 3 times on the diaphyseal region with a mean OL of 77 mm (52–112) (Table 2).

765

Limb-lengthening parameters All 10 patients successfully completed the lengthening procedures. The average achieved distraction measured 48 mm (26–80). The mean difference between the planned and the actual distraction was –1.1 mm (–2 to 0). The accuracy of distraction was 97.9% (SD 2.2) and the precision was 97.8%. The mean distraction period was 82 days (50–128) and the mean consolidation period was 6.6 months (5.1–8.9). The mean DIX was 0.6 mm/day (0.5–0.7) and CIX 1.5 days/cm (0.9– 2.0). All lengthening procedures were completed with the ILN remaining in situ until the end of distraction showing excellent reliability (100%). At the time of manuscript submission (April 30, 2020) 7 ILNs were explanted after a mean of 16 months (7–35). Residual postoperative LLD measured 8.9 mm (0–22). According to Chappell et al. (2019) the parameter “LLD” was excellent in 8 patients and good in 2 cases (Nos 4 and 6) (Table 3). Limb alignment parameters The Hindfoot-Alignment Score according to Katsenis et al. (2005) was excellent in 9 patients and good in 1 case. The Tibia-Alignment Score according to Schoenleber and Hutson (2015) was excellent in all patients. According to Chappell et al. (2019) the subscales “deformity hindfoot” and “deformity tibia” correspond to the aforementioned scores. Difficulties Problems 3 patients complained continuously of moderate pain under distraction alleviated by an oral non-steroidal antiphlogistic. Delayed wound healing was successfully treated nonoperatively in 3 patients. Flexion contractures of the toes due to relative tendon shortening occurred in 2 patients and were resolved either by physiotherapy (No. 3) or tenotomy and arthrodesis during ILN removal (No. 4). In half of the patients an adjustment of the distraction rate was necessary due to either the risk of premature consolidation (temporary acceleration = 2) or insufficient callus formation and toe flexion contractures (temporary deceleration = 3) (Table 3). Obstacles No additional surgeries were necessary under distraction. Non-union led to a prolonged course of treatment in 2 patients. Full osseous consolidation was achieved by proximal dynamization of the nail (No. 1) or nail exchange combined with autologous bone grafting (No. 5) (Figure 5, see Supplementary data). Another nail exchange was performed preventively in a patient (No. 8) who did not abide by the weight-bearing restrictions. Premature callus consolidation was not observed. No implant-associated complications were documented. Minor ILN bending (< 5°) was found in 3 patients. No deep infections were recorded.

766

Acta Orthopaedica 2020; 91 (6): 761–769

Table 3. Patient data—postoperative parameters A B C D E F

K L M N

1 35 –1 9 59 N/A a 0.63 N/A a Delayed wound Non-union No E E 83 (G) healing, acceleration of distraction rate 2 26 0 10 51 5.1 0.51 1.9 Deceleration of No E E 84 (G) distraction rate 3 31 –2 7 50 6.2 0.62 2.0 Toe flexion contracture No E E 86 (G) 4 38 –2 17 57 7.1 0.67 1.9 Toe flexion contracture, No E E 86 (G) deceleration of distraction rate 5 53 –2 7 85 N/A a 0.62 N/A a Delayed wound Non-union No E E 81 (G) healing 6 50 0 22 93 5.1 0.54 1.0 Delayed wound No G E 79 (F) healing 7 49 –1 11 78 6.5 0.65 1.3 Pain Slight ILN No E E 86 (G) bending 8 50 0 5 95 8.9 0.53 1.8 Pain, deceleration of Slight ILN No E E 77 (F) distraction rate bending b 9 64 –1 11 121 6.7 0.53 1.0 Slight ILN No E E 86 (G) bending 10 80 0 0 128 7.0 0.63 0.9 Pain, acceleration of No E E 81 (G) distraction rate a Non-union

A Patient no. B Achieved distraction (mm) C Difference achieved – planed distraction (mm) D Final leg length discrepancy (mm) E Distraction period (days) F Consolidation period (months) G Distraction index (mm/d) H Consolidation index (months/cm) I Problems J Obstacles ILN—intramedullary lengthening nail b preventive nail exchange

True complications No true complications were observed. No intraoperative injuries were recorded. All difficulties during limb lengthening were resolved before the end of treatment. No additional surgery was necessary after consolidation. No lengthening procedure failed to achieve the planned distraction by more than 1 cm. Functional outcome Range of motion of the knee was not affected by the lengthening procedure in any patient. According to the AOFAS ankle– hindfoot score grading categories (maximum score 100) 8 patients reached a good and 2 patients a fair result. The average AOFAS ankle–hindfoot score measured 83 (77–86). Due to the preexisting ankle and hindfoot fusion the maximum score in our cohort is 86 instead of 100 (Rochman et al. 2008). At the time of manuscript submission all patients were pain free and able to ambulate under full weight-bearing load without walking aids. All except 1 patient (No. 4) relied on an orthosis or shoe customizing for ambulation pre- and postoperatively (Table 3).

0 Shoe finishing

Done

30.1

Orthopedic Done shoes Orthopedic Done shoes None Done

15.9 26.3 21.6

Orthopedic Done 19.5 shoes Orthosis Scheduled 14.9 Orthopedic Done shoes Orthosis Done

12.3 12.3

Orthosis

Scheduled 13.0

Orthosis

Scheduled 12.8

K True complications L Katsenis-Hindfoot-Alignment score E—Excellent G—Good M Schoenleber-Tibia-Alignment score E—Excellent N AOFAS-Score G—Good F—Fair O Orthopaedic assist devices P ILN removal Q Follow-up (months) N/A—Not applicable

Discussion This is the first reported series of tibial lengthening through retrogradely implanted ILNs in patients with preexisting ankle and hindfoot fusion. The basic idea regarding the preference for a retrograde approach for tibial ILN implantation in contrast to the standard antegrade insertion technique is to avoid an entry-related effect on the knee joint. Antegrade tibial nail implantation always requires a knee arthrotomy (Fragomen and Rozbruch 2017) and bears the risk of joint effusion or hemarthrosis associated with pain and restricted ROM. Some patients develop persisting anterior knee pain and may become unable to kneel (Rothberg et al. 2019). Due to the retrograde ILN implantation inherently no approach-related effects on the knee joint were observed in our patients. On the other hand, plantar nail entry carries the risk of approach-related problems such as painful soft tissue irritation or delayed wound healing (Mosca et al. 2020). 3 of our patients had delayed plantar wound healing. No cases of deep infections or neurovascular injury occurred.

Acta Orthopaedica 2020; 91 (6): 761–769

In patients undergoing proximal tibial lengthening procedures using antegrade ILNs persistent knee joint contractures have been observed in up to 40% (Shabtai et al. 2014, Tiefenboeck et al. 2016, Wiebking et al. 2016, Nasto et al. 2020), whereas persistent ROM restrictions of the ankle joint were encountered in up to 25% of patients (Kirane et al. 2014, Shabtai et al. 2014, Tiefenboeck et al. 2016, Wiebking et al. 2016, Cosic and Edwards 2020, Nasto et al. 2020). In our series all patients had a fused ankle and hindfoot, and consequential development of equinus contracture or other ankle or hindfoot deformity did not present a problem. No patient developed temporary or permanent ROM restrictions of the knee joint. Toe-associated problems such as flexor tendon contractures were encountered in 2 patients (Nos 3 and 5). However, toe clawing resolved by flexor tendon release has also been described following proximal tibial lengthening using antegrade ILNs (Kirane et al. 2014, Tiefenboeck et al. 2016). Patients with an unstable knee due to congenital deficiencies are exposed to a risk of knee dislocation (Shabtai et al. 2014, Black et al. 2015, Szymczuk et al. 2019). Despite 6 of our 10 patients having a congenital deficiency, no knee dislocation was observed. We believe that distal tibial distraction via retrogradely inserted ILN might especially be beneficial for patients with an unstable knee and preexisting ankle and hindfoot fusion, i.e., due to congenital deficiencies. To prevent physeal damage in immature patients antegrade ILN implantation is considered to be contraindicated and consequently lengthening is usually carried out using external fixators (Wagner et al. 2017, Frommer et al. 2018). In patients with preexisting ankle and hindfoot fusion and concomitant closure of the distal tibial growth plate, the retrograde approach allows the use of a tibial ILN and eliminates the need for external fixators. The distraction goal should be defined on the basis of the predicted LLD at skeletal maturity when conducting leg lengthening in immature patients with a closed distal tibial physis (Frommer et al. 2018). In patients with fused ankle and hindfoot the goal of the treatment should not be exact LLD equalization. An intentional under-correction to a residual LLD of 0.5–1.0 cm is advisable to permit adequate “swing-through” during gait (McCoy et al. 2012). Achieving this goal can become challenging in skeletally immature patients. Tibial ILN insertion through an antegrade approach is technically demanding (Fragomen and Rozbruch 2017) and susceptible to the development of iatrogenic deformities (Lee et al. 2017). Usually blocking screws are necessary to prevent flexion and coronal (especially valgus) deformities (Kirane et al. 2014, Fragomen and Rozbruch 2017, Lee et al. 2017). On the other hand, the same techniques facilitate deformity corrections close to the knee joint (Fragomen and Rozbruch 2017). Using a retrograde ILN with a distal tibial osteotomy, knee-associated deformities can hardly be addressed. In our patients only diaphyseal or distal residual deformities of the tibia were simultaneously realigned during ILN implantation. Although most of the

767

patients had congenital conditions that are usually associated with coronal tibial deformities close to the knee joint (especially valgus), the tibial alignment grading by Schoenleber and Hutson (2015) showed excellent results in all patients as most of the patients had previously undergone deformity reconstructions either by antegrade nail or external fixator. Using a retrograde ILN the corticotomy for distraction osteogenesis is located in the distal tibia. Several studies investigated the results of tibial lengthening with ankle arthrodesis using external fixation. Most authors carried out proximal tibial distractions of 40–55 mm and found an external fixation index of 54–76 days/cm. Non-union of the regenerate was found in 4–25% of the patients (Katsenis et al. 2005, Rochman et al. 2008, Tellisi et al. 2008, Fragomen et al. 2012). Only a few studies evaluated the results of ankle and hindfoot reconstruction and concurrent tibial lengthening through a distal corticotomy. The described distraction index was 0.55 mm/day and the external fixation index was 70–144 days/cm. Non-union occurred in 0–10% of patients. In accordance with our findings the authors demonstrated successful tibial lengthening via a distal corticotomy but did not recommend this approach for distractions of more than 3–4 cm (Sakurakichi et al. 2003, Chappell et al. 2019). Based on our observations, sufficient callus regenerate and bone healing in the distal tibia can also be achieved after longer distraction distances. These findings might indicate that the osseous healing potential of proximal or distal tibial distraction osteogenesis is equivalent. Independent of concomitant ankle and hindfoot pathologies various authors have analyzed proximal tibial lengthening procedures with antegrade magnetically driven ILNs (Kirane et al. 2014, Schiedel et al. 2014, Shabtai et al. 2014, Tiefenboeck et al. 2016, Wiebking et al. 2016, Wagner et al. 2017, Cosic and Edwards 2020, Nasto et al. 2020). The achieved distraction was 26–45 mm (Schiedel et al. 2014, Wiebking et al. 2016, Cosic and Edwards 2020). The DIX was reported as 0.48–0.84 mm/day (Tiefenboeck et al. 2016, Wiebking et al. 2016, Wagner et al. 2017, Nasto et al. 2020) and the CIX amounted to 0.5–3.3 months/cm (Shabtai et al. 2014, Tiefenboeck et al. 2016, Horn et al. 2019, Cosic and Edwards 2020, Nasto et al. 2020). Non-union was encountered in 11–75% (Kirane et al. 2014, Shabtai et al. 2014, Tiefenboeck et al. 2016, Wiebking et al. 2016, Wagner et al. 2017, Cosic and Edwards 2020, Nasto et al. 2020). These findings are comparable to our results for distal tibial distraction osteogenesis using a retrograde ILN with non-union in one-fifth of patients and a CIX of 1.5 months/cm, indicating at least equivalent bone healing potential. Slight bending (< 5°) of the ILN occurred in 3 patients. However, this radiographic finding was not associated with malfunction of the nail, diminished bone healing, or secondary malalignment of the tibia. In all of these cases the distraction distance was close to or greater than 50 mm and the smallest nail diameter of 8.5 mm was chosen in 2 of these patients. Thus, greater

768

lengthening distances with increasing leverage and smaller nail diameters with reduced strength of the device might be risk factors for ILN bending. Whenever an 8.5 mm ILN was implanted, this was due to the limited anatomy of the hindfoot or tibia, taking into account the necessary amount of over-reaming. This study has limitations due to its retrospective, observational character and the small number of patients treated with only 1 type of ILN. In order to identify potential risk factors for treatment failure prospective groupwise and comparative study designs are needed, i.e, comparison with antegrade tibial ILN. The results of the study do not provide any comparison with the potential benefit of tibial lengthening with external fixators or other types of ILNs. There is a selection bias towards congenital etiologies, which is explained by the specification of our department. In summary, distal tibial lengthening via a retrograde ILN is a reliable alternative to proximal tibial distraction and antegrade ILN insertion for correction of LLD in patients with preexisting ankle and hindfoot fusion, showing equivalent bone healing potential and low complication rates, and avoiding approachrelated effects on the knee. Monitoring of callus consolidation and critical assessment of potential soft tissue complications, especially of the toes under distraction, are mandatory. Supplementary data Figures 2 and 5 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/1745 3674.2020.1807222

BV: wrote the manuscript, performed and supervised the measurements, performed statistical analysis, prepared the figures. RR: provided the radiographs, analyzed the data, supervised the work, made substantial changes to the manuscript. GG: provided the radiographs, made substantial changes to the manuscript. GT: performed the measurement for the inter-rater reliability analysis, critically assessed and corrected the manuscript. AL: assessed and corrected the manuscript, arranged the data, and prepared the tables. CT: revised the manuscript. JNB: performed the measurements, critically assessed and corrected the manuscript, AF: designed the study, analyzed the data, supervised the work, performed the measurements, critically assessed and corrected the manuscript. Acta thanks Janet D. Conway and Ulrik Kähler Olesen for help with peer review of this study.

Baumgart R, Betz A, Schweiberer L. A fully implantable motorized intramedullary nail for limb lengthening and bone transport. Clin Orthop Relat Res 1997; (343): 135-43. Black S R, Kwon M S, Cherkashin A M, Samchukov M L, Birch J G, Jo C H. Lengthening in congenital femoral deficiency: a comparison of circular external fixation and a motorized intramedullary nail. J Bone Joint Surg Am 2015; 97(17): 1432-40. doi: 10.2106/JBJS.N.00932. Chappell T M, Ebert C C, McCann K M, Hutchinson B L, Rodriguez-Collazo E. Distal tibial distraction osteogenesis: an alternative approach to addressing limb length discrepancy with concurrent hindfoot and ankle reconstruction. J Orthop Surg Res 2019; 14(1): 244. doi: 10.1186/s13018-019-1264-0.

Acta Orthopaedica 2020; 91 (6): 761–769

Cole J D, Justin D, Kasparis T, DeVlught D, Knobloch C. The intramedullary skeletal kinetic distractor (ISKD): first clinical results of a new intramedullary nail for lengthening of the femur and tibia. Injury 2001; 32(Suppl 4): SD129-39. Cosic F, Edwards E. PRECICE intramedullary nail in the treatment of adult leg length discrepancy. Injury 2020. doi: 10.1016/j.injury.2020.03.004. Fragomen A T, Rozbruch S R. Lengthening and deformity correction about the knee using a magnetic internal lengthening nail. SICOT J 2017; 3: 25. doi: 10.1051/sicotj/2017014. Fragomen A T, Borst E, Schachter L, Lyman S, Rozbruch S R. Complex ankle arthrodesis using the Ilizarov method yields high rate of fusion. Clin Orthop Relat Res 2012; 470(10): 2864-73. doi: 10.1007/s11999-012-2470-9. Frommer A, Rodl R, Gosheger G, Vogt B. [Application of motorized intramedullary lengthening nails in skeletally immature patients: indications and limitations]. Unfallchirurg 2018; 121(11): 860-7. doi: 10.1007/s00113018-0541-4. Guichet J M, Casar R S. Mechanical characterization of a totally intramedullary gradual elongation nail. Clin Orthop Relat Res 1997; (337): 281-90. doi: 10.1097/00003086-199704000-00032. Horn J, Hvid I, Huhnstock S, Breen A B, Steen H. Limb lengthening and deformity correction with externally controlled motorized intramedullary nails: evaluation of 50 consecutive lengthenings. Acta Orthop 2019; 90(1): 81-7. doi: 10.1080/17453674.2018.1534321. Iobst C A, Rozbruch S R, Nelson S, Fragomen A. Simultaneous acute femoral deformity correction and gradual limb lengthening using a retrograde femoral nail: technique and clinical results. J Am Acad Orthop Surg 2018; 26(7): 241-50. doi: 10.5435/JAAOS-D-16-00573. Katsenis D, Bhave A, Paley D, Herzenberg J E. Treatment of malunion and nonunion at the site of an ankle fusion with the Ilizarov apparatus. J Bone Joint Surg Am 2005; 87(2): 302-9. doi: 10.2106/JBJS.C.01421. Kirane Y M, Fragomen A T, Rozbruch S R. Precision of the PRECICE internal bone lengthening nail. Clin Orthop Relat Res 2014; 472(12): 3869-78. doi: 10.1007/s11999-014-3575-0. Laubscher M, Mitchell C, Timms A, Goodier D, Calder P. Outcomes following femoral lengthening: an initial comparison of the Precice intramedullary lengthening nail and the LRS external fixator monorail system. Bone Joint J 2016; 98-B(10): 1382-8. doi: 10.1302/0301-620X.98B10.36643. Lee D H, Kim S, Lee J W, Park H, Kim T Y, Kim H W. A comparison of the device-related complications of intramedullary lengthening nails using a new classification system. Biomed Res Int 2017; 2017: 8032510. doi: 10.1155/2017/8032510. Mahboubian S, Seah M, Fragomen A T, Rozbruch S R. Femoral lengthening with lengthening over a nail has fewer complications than intramedullary skeletal kinetic distraction. Clin Orthop Relat Res 2012; 470(4): 1221-31. doi: 10.1007/s11999-011-2204-4. McCoy T H, Goldman V, Fragomen A T, Rozbruch S R. Circular external fixator-assisted ankle arthrodesis following failed total ankle arthroplasty. Foot Ankle Int 2012; 33(11): 947-55. doi: DOI: 10.3113/ FAI.2012.0947 Mosca M, Caravelli S, Fuiano M, Massimi S, Censoni D, Grassi A, Vocale E, Ceccarelli F, Zaffagnini S. Tibiotalocalcaneal arthrodesis through retrograde nailing for the treatment of juxtaarticular distal tibia aseptic nonunions: a retrospective study at a minimum follow-up of 4 years. Injury 2020; 51(6): 1377-81. doi: 10.1016/j.injury.2020.03.039. Nasto L A, Coppa V, Riganti S, Ruzzini L, Manfrini M, Campanacci L, Palmacci O, Boero S. Clinical results and complication rates of lower limb lengthening in paediatric patients using the PRECICE 2 intramedullary magnetic nail: a multicentre study. J Pediatr Orthop B 2020. [Online ahead of print] doi: 10.1097/BPB.0000000000000651. Paley D. Problems, obstacles, and complications of limb lengthening by the Ilizarov technique. Clin Orthop Relat Res 1990; (250): 81-104. Paley D. Principles of deformity correction. Berlin/New York: Springer; 2002. Paley D, Bhave A, Herzenberg J E, Bowen J R. Multiplier method for predicting limb-length discrepancy. J Bone Joint Surg Am 2000; 82(10): 1432-46.

Acta Orthopaedica 2020; 91 (6): 761â&#x20AC;&#x201C;769

Rochman R, Jackson Hutson J, Alade O. Tibiocalcaneal arthrodesis using the Ilizarov technique in the presence of bone loss and infection of the talus. Foot Ankle Int 2008; 29(10): 1001-8. doi: 10.3113/ FAI.2008.1001. Rothberg D L, Stuart A R, Presson A P, Haller J M, Higgins T F, Kubiak E N. A comparison of the open semi-extended parapatellar versus standard entry tibial nailing techniques and knee pain: a randomized controlled trial. J Orthop Trauma 2019; 33(1): 31-6. doi: 10.1097/ BOT.0000000000001309. Sakurakichi K, Tsuchiya H, Uehara K, Kabata T, Yamashiro T, Tomita K. Ankle arthrodesis combined with tibial lengthening using the Ilizarov apparatus. J Orthop Sci 2003; 8(1): 20-5. doi: 10.1007/s007760300003. Schiedel F M, Vogt B, Tretow H L, Schuhknecht B, Gosheger G, Horter M J, Rodl R. How precise is the PRECICE compared to the ISKD in intramedullary limb lengthening? Reliability and safety in 26 procedures. Acta Orthop 2014; 85(3): 293-8. doi: 10.3109/17453674.2014.913955. Schoenleber S J, Hutson J J, Jr. Treatment of hypertrophic distal tibia nonunion and early malunion with callus distraction. Foot Ankle Int 2015; 36(4): 400-7. doi: 10.1177/1071100714558509. Shabtai L, Specht S C, Standard S C, Herzenberg J E. Internal lengthening device for congenital femoral deficiency and fibular hemimelia. Clin

769

Orthop Relat Res 2014; 472(12): 3860-8. doi: 10.1007/s11999-0143572-3. Szymczuk V L, Hammouda A I, Gesheff M G, Standard S C, Herzenberg J E. Lengthening with monolateral external fixation versus magnetically motorized intramedullary nail in congenital femoral deficiency. J Pediatr Orthop 2019; 39(9): 458-65. doi: 10.1097/BPO.0000000000001047. Tellisi N, Fragomen A T, Ilizarov S, Rozbruch S R. Limb salvage reconstruction of the ankle with fusion and simultaneous tibial lengthening using the Ilizarov/Taylor spatial frame. HSS J 2008; 4(1): 32-42. doi: 10.1007/ s11420-007-9073-0. Tiefenboeck T M, Zak L, Bukaty A, Wozasek G E. Pitfalls in automatic limb lengthening: first results with an intramedullary lengthening device. Orthop Traumatol Surg Res 2016; 102(7): 851-5. doi: 10.1016/j. otsr.2016.07.004. Wagner P, Burghardt R D, Green S A, Specht S C, Standard S C, Herzenberg J E. PRECICE((R)) magnetically-driven, telescopic, intramedullary lengthening nail: pre-clinical testing and first 30 patients. SICOT J 2017; 3: 19. doi: 10.1051/sicotj/2016048. Wiebking U, Liodakis E, Kenawey M, Krettek C. Limb lengthening using the PRECICE(TM) nail system: complications and results. Arch Trauma Res 2016; 5(4): e36273. doi: 10.5812/atr.36273.

770

Acta Orthopaedica 2020; 91 (6): 770–775

Randomized trial comparing suture button with single 3.5 mm syndesmotic screw for ankle syndesmosis injury: similar results at 2 years Benedikte Wendt RÆDER 1, Ingrid Kvello STAKE 2, Jan Erik MADSEN 3,4, Frede FRIHAGEN 3, Silje Berild JACOBSEN 5, Mette Renate ANDERSEN 1,6, and Wender FIGVED 1 1 Department

of Orthopaedic Surgery, Baerum Hospital, Vestre Viken Hospital Trust; 2 Kalnes Hospital, Østfold Hospital Trust; 3 Division of Orthopaedic Surgery, Oslo University Hospital; 4 Institute of Clinical Medicine, Faculty of Medicine, University of Oslo; 5 Department of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo; 6 Aleris Hospital, Tromsø, Norway Correspondence: wendtraeder@gmail.com Submitted 2020-04-28. Accepted 2020-08-14.

Background and purpose — Better outcomes are reported for suture button (SB) compared with syndesmotic screws (SS) in patients treated for an acute ankle syndesmotic injury. One reason could be that screws are more rigid than an SB. A single tricortical 3.5 mm syndesmotic screw (TS) is the most dynamic screw option. Our hypothesis is that 1 SB and 1 TS provide similar results. Therefore, in randomized controlled trial, we compared the results between SB and TS for syndesmotic stabilization in patients with acute syndesmosis injury. Patients and methods — 113 patients with acute syndesmotic injury were randomized to SB (n = 55) or TS (n = 58). The American Orthopedic Foot & Ankle Society (AOFAS) Ankle–Hindfoot Score was the primary outcome measure. Secondary outcome measures included Manchester Oxford Foot Questionnaire (MOXFQ), Olerud–Molander Ankle score (OMA), visual analogue scale (VAS), EuroQol- 5D (EQ-5D), radiologic results, range of motion, complications, and reoperations (no implants were routinely removed). CT scans of both ankles were obtained after surgery, and after 1 and 2 years. Results — The 2-year follow-up rate was 84%. At 2 years, median AOFAS score was 97 in both groups (IQR SB 87–100, IQR TS 90–100, p = 0.7), median MOXFQ index was 5 in the SB group and 3 in the TS group (IQR 0–18 vs. 0–8, p = 0.2), and median OMA score was 90 in the SB group and 100 in the TS group (IQR 75–100 vs. 83–100, p = 0.2). The syndesmotic reduction was similar 2 years after surgery; 19/55 patients in the SB group and 13/58 in the TS group had a difference in anterior syndesmotic width ≥ 2 mm (p = 0.3). 0 patients in the SB group and 5 patients in the TS group had complete tibiofibular synostosis (p = 0.03). At 2 years, 10 TS were broken. Complications and reoperations were similar between the groups. Interpretation — We found no clinically relevant differences regarding outcome scores between the groups. TS is an inexpensive alternative to SB.

Since 2018, several meta-analyses have been published evaluating treatment of acute ankle syndesmotic injury, reporting better outcomes for suture button (SB) fixation compared with syndesmotic screw (SS) (Shimozono et al. 2018, McKenzie et al. 2019). Shimozono concluded that the SB technique resulted in improved outcome and lower rates of joint malreduction. These results are based on heterogenous studies: different fracture types were compared; different numbers of implants were used and different diameters and cortices were engaged for SS fixation (Shimozono et al. 2018). Andersen et al. (2018) reported superior results for SB compared with a quadricortical 4.5 mm SS. A quadricortical SS necessitates routine screw removal, with a 5–9% reported risk of wound infection (Schepers et al. 2011, Andersen et al. 2015) and potential loss of reduction after implant removal (Laflamme et al. 2015). A quadricortical SS is a rigid fixation, inhibiting tibiofibular movement throughout the gait cycle (Riedel et al. 2017, Ramsey et al. 2018). The SB has a higher implant cost compared with SS (Ramsey et al. 2018), may not be sufficient to maintain fibular length in Maisonneuve fractures (Riedel et al. 2017), and has an implant removal rate of 6%, mainly due to irritation from the lateral knot (Andersen et al. 2018). The single tricortical 3.5 mm syndesmotic screw (TS) allows for some tibiofibular movement (Clanton et al. 2017), making the TS an inexpensive alternative, without need for routine implant removal. In this study we compare outcomes between a knotless SB and TS. Our hypothesis was that there is no difference in outcomes in patients treated with SB and a 3.5 mm TS.

Patients and methods Patients and procedures 3 hospitals participated in recruiting and treating patients. Surgery was conducted by 45 different surgeons. Patients were included by the orthopedic resident on call, from January 2016

Acta Orthopaedica 2020; 91 (6): 770–775

ENROLLMENT

771

Assessed for eligibility n = 902 Excluded (n = 789): – not AO 44C fracture, 597 – not meeting inclusion criteria, 172 – declined to participate, 7 – other reasons, 13 Randomized n = 113

ALLOCATION

Allocated to suture button (SB) (n = 58)

Allocated to tricortical screw (TS) (n = 55)

Received allocated intervention (n = 55)

Received allocated intervention (n = 53)

Did not receive allocated intervention (n = 3): – no syndesmosis fixation, 2 – received SS, 1

Did not receive allocated intervention (n = 2): – received SB, 2 – received 2 SS, 1

FOLLOW-UP

Lost to follow-up (n = 0)

Lost to follow-up (n = 1)

Did not attend follow-up at: – 6 months, 3 – 1 year, 4 – 2 years, 10

Did not attend follow-up at: – 6 months, 1 – 1 year, 3 (1 dead) – 2 years, 8 (2 dead)

Incomplete data sets: – 6 months, 0 – 1 year, 1 missing CT scan – 2 years, 2 missing CT scans

Incomplete data sets: – 6 months, 0 – 1 year, 2 missing CT scans – 2 years, 2 missing CT scans ANALYSIS

Analyzed at 2 years (n = 48)

Analyzed at 2 years (n = 47)

Excluded from analysis (n = 0)

Figure 2. CT of injured ankle (upper panel) and uninjured ankle (lower panel) in a 20-year-old woman, 2 years after injury. Tibiofibular distance is measured on axial CT 1 cm proximal to the ankle joint. Distance measured anterior (A); central (C); and posterior (P).

Figure 1. CONSORT flowchart of the trial enrollment and analysis.

to September 2017. Patients aged 18 to 69 who had suffered an acute AO type 44-C ankle fracture assessed by radiographs were asked to participate (Figure 1). Exclusion criteria were polytrauma, open fractures, previous fracture or arthritis of the same ankle, or neurologic impairment of the lower limbs. A web-based randomization system was used, developed and administered by Clinical Research Unit Central Norway, Norwegian University of Science and Technology, Trondheim, Norway. Surgery was performed according to AO principles. The syndesmosis was reduced and fixed in a closed manner, guided by fluoroscopy. Surgeons were recommended to fix the syndesmosis at a level just proximal to the inferior tibiofibular joint (Barbosa et al. 2020), the use of temporary fixators (K-wire or reduction clamp) was decided by the surgeon. Patients allocated to SB were treated with a single knotless SB (Ziptight, Zimmer Biomet, Warsaw, IN, USA). Patients allocated to TS were treated with a fully threaded self-tapping, 3.5 mm tricortical screw (DePuy Synthes, West Chester, PA, USA). The screw length was not specified, but standardized to engage 3 cortices. Surgery was performed by the on-call team, either by an experienced resident, or a less experienced resident accompanied by a consultant or senior resident. Antibiotic prophylaxis was given as a single dose peroperatively. All patients followed the same protocol postoperatively: implants

were not routinely removed; plaster casts and thrombosis prophylaxis were not used routinely. Patients were advised partial weight-bearing (20–30 kg) directly after surgery (Barbosa et al. 2020), then weight-bearing as tolerated after 6 weeks. Outcome measures Patients were assessed by an orthopedic surgeon and a physiotherapist at 6 weeks, 6 months, 1 and 2 years. The physiotherapists who conducted the physical examinations were blinded to the treatment allocation. The main outcome measure was the American Orthopedic Foot & Ankle Society (AOFAS) Ankle–Hindfoot scale. AOFAS incorporates subjective and objective factors into a numerical scale of 0 to 100, 100 being the best. Secondary outcome measures included the Manchester Oxford Foot Questionnaire (Dawson et al. 2007, 2011) (MOXFQ), a 16-item (each item scored 0–4) patient reported outcome measure (PROM). MOXFQ has 3 separate underlying dimensions: pain, activity, and social interaction. The raw score of maximum 64 was converted to a metric index from 0 (best) to 100 (worst) (Morley et al. 2013). MOXFQ is available in Norwegian and is not validated for ankle fractures. The MOXFQ is validated for hallux valgus surgery and has been found to be highly responsive (Dawson et al. 2007). Other secondary measures were the Olerud–Molander Ankle (OMA) score (Olerud and

772

Molander 1984), EuroQol-5D (EQ-5D) index, EQ-5D visual analogue scale (VAS), and VAS scores for pain during rest, during walking, at night, and during daily activities. OMA is a self-reported scale validated for ankle fractures, ranging from 0 (worst) to 100 (best). EQ-5D is a well-validated generic health-related quality-of-life instrument. Ankle range of motion was measured, comparing injured with non-injured ankle. The examination was standardized by a blinded physiotherapist, measuring dorsal and plantar flexion with a goniometer, with the foot placed on a 25 cm high foot stool with the knee in flexion. Radiological measurements Plain radiographs of the injured ankle were obtained after surgery, and at 6 weeks and 6 months. CT scans of both ankles were obtained postoperatively, and after 1 and 2 years. CT scans were standardized with the patient in a supine position, placing the feet in a purpose-made device, keeping the ankles in neutral position with 20° internal rotation of the legs. Radiological measurements were performed by 1 senior musculoskeletal radiologist (SBJ) and one orthopedic surgeon (BWR). The syndesmosis was assessed postoperatively and after 1 and 2 years by measuring the tibiofibular distance on axial CT scans, 1 cm proximal to the midpoint of the tibial plafond (Figure 2). The difference between injured and uninjured side was calculated. A criterion of < 2 mm difference in tibiofibular distance was selected for acceptable syndesmotic reduction (Andersen et al 2019, Patel et al. 2019). Signs of ankle osteoarthritis (OA), synostosis, talar exostoses, broken screws, and osteochondral lesions were reported. When assessing OA on CT scans, we defined mild OA as presence of osteophytes, and advanced OA as narrowing of the joint space and presence of cysts and sclerosis (Ray et al. 2019). Statistics Sample size was calculated according to the equivalence criterion (Piaggio et al. 2012). The minimal clinically important difference (MCID) for ankle fracture patients is not defined for the AOFAS score but has been suggested to be half of the standard deviation (SD) (Norman et al. 2003). Based on data from previous trials with a similar population, the SD was estimated to 12 points (Wikeroy et al. 2010, Andersen et al. 2018), giving an MCID of the AOFAS score of 6 points. A between-group difference of 10 points (AOFAS) was used to ensure a sufficient inclusion of patients. 38 patients had to be included in each group to achieve a power of 0.95 and a significance level of 0.05. To strengthen the data and compensate for loss to follow-up, we planned to include 60 patients in each group. Analyses of endpoint results were performed as both intention-to-treat and per-protocol. Student’s T-test was used to compare means of normally distributed data. The Mann–Whitney U-test was used in cases of skewed data. Fisher’s exact test was used for categorial data. Data is reported as numbers, mean with SD, or median

Acta Orthopaedica 2020; 91 (6): 770–775

Table 1. Patient characteristics at time of enrolment. Values are number of patients unless otherwise specified Characteristic Mean age (SD) Male sex Right side Mean BMI (SD) Medial malleolar fracture Posterior malleolar fracture Medial and posterior malleolar fracture Maisonneuve fracture Osteochondral damage of the talus Intra-articular loose bodies Temporary external fixator an

SB (n = 55)

TS (n = 58)

44 (15) 48 (14) 35 30 32 26 27 (5) 26 (4) 14 19 37 31 10 15 26 20 2 a 4 a 9 10 7 2

= 54

with interquartile range (IQR). We considered a probability of less than 5% as statistically significant and used 95% confidence intervals (CI) to describe uncertainty. Data analysis was conducted in IBM SPSS Statistics for Mac version 26 (IBM Corp, Armonk, NY, USA). Ethics, registration, funding, and potential conflicts of interest Patients gave their written consent prior to randomization. The trial was conducted in accordance with the Declaration of Helsinki, approved by the National Committees for Research Ethics in Norway 2015/1860 and registered at ClinicalTrials. gov (NCT02930486). The study did not receive external funding. The authors have no conflicts of interest to declare.

Results Results are reported according to the CONSORT guidelines. 113 patients were randomized and allocated to SB (= 58) or TS (= 55) (Figure 1). The 2-year follow-up rate was 84%; the radiological follow-up rate was 81%. The baseline demographic patient characteristics and fracture treatment were reported (Table I). Clinical outcomes The groups did not differ statistically regarding clinical outcome: at 2 years, the median AOFAS score was 97 in both groups (IQR SB 87–100 vs. TS 90–100, p = 0.7) (Table 2). The difference in mean AOFAS was < 2, equivalent at all controls (Figure 3). Median MOXFQ was 5 in the SB group and 3 in the TS group (IQR SB 0–18 vs. TS 0–8, p = 0.2) (Table 2), and median OMA score was 90 in the SB group and 100 in the TS group (IQR SB 75–100 vs. TS 83–100, p = 0.2). Similarly, no statistically significant difference was detected in VAS, EQ-5D VAS, or EQ-5D (Table 2). Fracture pattern affected clinical outcome when we stratified the groups according to fracture pattern: after 2 years, patients with

Acta Orthopaedica 2020; 91 (6): 770–775

773

Table 2. Primary and secondary outcome measures Outcome SB a TS a measure n score n score AOFAS 6 weeks 54 67 (10) 6 months 53 87 (82–98) 1 years 53 93 (82–100) 2 years 48 97 (87–100) MOXFQ 6 weeks 52 29 ( (11) 6 months 55 14 (3–31) 1 year 52 5 (0–32) 2 years 48 5 (0–18) OMA 1 year 53 90 (73–100) 2 years 47 90 (75–100) VAS for pain during rest 6 weeks 53 1.0 (0–2) 6 months 54 0.0 (0–1) 1 year 53 0.0 (0–1) 2 years 48 0.0 (0–1) VAS for pain during walking 6 weeks 53 2.0 (1–4) 6 months 54 1.0 (0–3) 1 year 53 1.0 (0–2) 2 years 48 0.0 (0–1) VAS for pain at night 6 weeks 53 1.0 (0–2) 6 months 54 0.0 (0–0) 1 year 53 0.0 (0–0) 2 years 48 0.0 (0–1) VAS for pain during daily activity 6 weeks 53 3.0 (2–6) 6 months 54 1.0 (0–3) 1 year 53 0.0 (0–2) 2 years 48 0.0 (0–29 EQ-5D index 6 weeks 53 0.7 (0.6–0.8) 6 months 54 0.8 (0.7–1.0) 1 year 53 1.0 (0.8–1.0) 2 years 48 1.0 (0.8–1.0) EQ-5D VAS 6 weeks 52 73 (15) 6 months 53 89 (70–95) 1 year 52 85 (71–95) 2 year 48 85 (70–95)

52 54 52 47

66 (13) 88 (77–98) 90 (84–99) 97 (90–100)

48 31 (13) 53 14 (3–36) 51 6 (0–13) 47 3 (0–8)

Tricortical screw better

p-value 0.7 c 1.0 b 0.5 b 0.7 b 0.4 c 0.7 b 0.9 b 0.2 b

52 90 (76–100) 0.4 b 45 100 (83–100) 0.2 b 49 54 52 47

1.0 (0–2) 0.0 (0–2) 0.0 (0–1) 0.0 (0–0)

0.9 b 0.1 b 0.5 b 0.6 b

49 54 52 47

3.0 (2–4) 1.0 (0–2) 1.0 (0–2) 0.0 (0–1)

0.3 b 0.8 b 0.9 b 0.2 b

49 54 52 47

1.0 (0–3) 0.0 (0–1) 0.0 (0–0) 0.0 (0–0)

0.6 b 0.01 b 1.0 b 0.2 b

49 54 52 47

4.0 (2–7) 1.0 (0–2) 1.0 (0–2) 0.0 (0–0)

0.4 b 0.9 b 0.6 b 0.03 b

53 54 52 47

0.7 (0.3–0.7) 0.8 (0.7–1.0) 1.0 (0.8–1.0) 1.0 (0.9–1.0)

0.1 b 0.9 b 1.0 b 0.3 b

51 54 52 45

63 (18) 80 (74–90) 88 (76–90) 90 (77–95)

0.004 c 0.2 b 0.6 b 0.6 b

For not normally distributed data values are given as median (IQR) in parentheses and for normally distributed data as mean (SD). b Nonparametric (Mann–Whitney U) test. c 2-sided t-test for independent samples.

a Maisonneuve fracture pattern had better outcome scores with a median AOFAS at 100 in the Maisonneuve patients group compared with 95 in all other injuries (IQR 95–100 vs. 85–100, p = 0.001), while patients with trimalleolar fractures did worse, with a median AOFAS at 92 compared with 99 in other injuries (IQR 85–97 vs. 90–100, p = 0.03) (Table 3, see Supplementary data). The ability to plantar- and dorsiflex the ankle was similar between the groups. At 2 years, the mean difference between injured and uninjured ankle in plantar and dorsiflexion was ≤ 5° (Table 4, see Supplementary data). Perprotocol analyses supported the intention-to-treat findings.

Suture button better

6 weeks

p = 0.7

6 months

p = 1.0

1 year

p = 0.5

2 years

p = 0.7

–15

–10

–5

Mean AOFAS difference between groups (95% CI)

Figure 3. AOFAS equivalence diagram. Blue area indicates margins of equivalence defined as the between-group difference of 10 points. Results at all time intervals are equivalent since the 95% CI lies wholly inside the margins.

Radiological results At 2 years, 30 patients in the SB group and 27 patients in the TS group had radiological signs of ankle OA (RR 1.1, CI 0.7– 1.7). When analyzing for advanced OA, there was a difference between the groups at 2 years: 8 patients in the SB group and 1 patient in the TS group had advanced OA (RR 8, CI 1–60). The groups displayed similar results when analyzing presence of talar osteophytes at 2 years: 12 in the SB group and 7 in the TS group (p = 0.3). At 2 years, 0 patients in the SB group and 5 patients in the TS group had complete synostosis (p = 0.03) (Figure 4, see Supplementary data). When stratifying the complete cohort at 2 years according to fracture pattern, patients with a Maisonneuve fracture had less OA (15 vs. 42, RR 0.7, CI 0.4–1.0), patients with a trimalleolar fracture had more OA (19 vs. 38, RR 1.6, CI 1.2–2.1). The tibiofibular distance measured on CT scans postoperatively and after 1 and 2 years was similar between the groups. At 2 years, the mean difference in tibiofibular distance was ≤ 1 mm for anterior, central, and posterior measurement in both groups (Table 5). When applying a tibiofibular difference of < 2 mm between injured and uninjured ankle as a criterion for acceptable reduction the groups had similar results at all controls; 19 patients in the SB group and 16 patients in the TS group had an anterior difference > 2 mm postoperatively (RR 1.2, CI 0.7–2.1) (Table 6, see Supplementary data). After 2 years, 35 of 45 patients still had their TS implanted; 10 screws were broken. Complications and reoperations 10 patients in the SB group and 17 patients in the TS group had ≥ 1 reoperation (p = 0.2) (Table 7, see Supplementary data). 5 patients in the SB group and 11 patients in the TS group had their implants removed because of local irritation alone (p = 0.2). 3 patients in the SB group and 3 patients in the SS group required early reoperation (< 3 weeks) after CT postoperatively revealed unacceptable reduction of the fracture or of the syndesmosis (3 syndesmosis malreductions, 1 fibula malreduction, 2 medial malleolus malreduction). 2 patients (male, age 50 and female, age 52 years) suffered a low-energy tibia fracture through the suture button canal (Fig-

774

Acta Orthopaedica 2020; 91 (6): 770–775

inferior to an SB, while Kortekangas et al. (2015) found no difference when comparing an TS with an SB. The first SBs available required a suture SB TS Mean between-group knot on the lateral side, with irritation and a Factor n difference n difference difference (95% CI) p-value a reported removal rate of 6% (Andersen et al. 2018). We used a knotless SB to Difference in anterior distance ≤ 2 weeks 54 0.1 (1.9) 56 0.7 (1.8) –0.5 (–1.2 to 0.2) 0.1 potentially lower this rate. Despite this, our 1 year 54 1.1 (2.0) 50 0.7 (1.8) 0.3 (–0.4 to 1.1) 0.4 removal rate was 9%. Changing to a knot 2 years 46 0.9 (1.9) 45 0.7 (1.6) 0.2 (–0.5 to 1.0) 0.5 less SB did not affect the removal rate. This Difference in central distance ≤ 2 weeks 54 0.1 (1.2) 56 –0.7 (1.1) 0.2 (–0.2 to 0.6) 0.3 could be due to other factors, such as irrita 1 year 54 1.2 (1.9) 50 0.9 (1.4) 0.3 (–0.3 to 1.0) 0.3 tion from the fibula plate, present in almost 2 years 46 1.4 (0.0–2.0) 45 1.0 (0.0–1.0) 0.7 (0.0 to 1.4) 0.2 b half of the SB patients. 6 patients required Difference in posterior distance ≤ 2 weeks 54 –0.4 (2.2) 56 –0.6 (2.1) 0.2 (–0.6 to 1.0) 0.7 early reoperation, based on postoperative 1 year 54 0.1 (1.8) 50 0.4 (1.8) –0.3 (–1.0 to 0.4) 0.5 CT scans. We advocate a low threshold 2 years 46 0.0 (2.3) 45 0.3 (2.0) –0.4 (–1.2 to 0.5) 0.4 for obtaining postoperative CT scans after a Levene’s test was used to assess equality of the variances. Statistical analysis was consyndesmotic reduction (Garner et al. 2015, ducted using the 2-sided t-test for independent samples in normally distributed data; Barbosa et al. 2020). otherwise the Mann–Whitney U-test was used. b The Mann–Whitney U-test was used. Trauma is the most common cause of ankle OA (Saltzman et al. 2005). The rate of radiologic OA after 2 years was high in ures 5, 6, see Supplementary data). The male patient presented this study. The reason for this could be the use of CT, which is 6 months postoperatively with a healed tibia fracture with more sensitive than radiographs when assessing OA. Most of 13° varus deformity. Since this patient had no complaints the the patients (48 of 57) displayed only minor signs of OA. The fracture was not addressed surgically. The female patient pre- rate of advanced OA in 9 patients is in line with previous studies sented initially with a large posterior malleolar fracture. She (Lübbeke et al. 2012, Ray et al. 2019). The observation period presented with pain while walking 4 months after her initial of 2 years is short and the study population is underpowered injury. She had suffered a tibia fracture and was reoperated on to conclude on the differences in advanced OA between the with open reduction and internal fixation. A dual energy X-ray groups. More patients had complete synostosis in the TS group, absorptiometry (DEXA scan) showed osteoporosis. supporting the findings by Hinds et al. (2014) that SS fixation is a risk factor for synostosis development. 2 patients treated with SB suffered a non-traumatic fracture through the suture button canal. This specific complication and its incidence have Discussion not been reported in the literature. We suggest a syndesmotic The main findings in this study are equivalent clinical results in screw as a better alternative in patients with poor bone quality. patients treated with either an SB or an TS 2 years after acute A weakness in the study is our choice of outcome score. The syndesmotic injury. The mean AOFAS difference between the ideal outcome score should be validated for the injury in quesgroups was overlapping and inside the margins of the 95% CI tion, have high reliability, and be available in the language of at all controls. The rate of appropriate syndesmotic reduction, the patients examined. Our primary outcome, the AOFAS, is reoperations, and rate of OA was similar between our groups. not validated; it is criticized for low precision, and for producIn the SB group, 2 patients experienced fractures through the ing skewed data due to ceiling effects (Veltman et al. 2017). suture button canal. 5 patients in the TS group had synostosis Even so, the AOFAS was chosen because of its widespread after 2 years. Fracture pattern affected clinical outcome. use. We decided to add the MOXFQ, since it was available The clinical results are in contrast to earlier studies report- in Norwegian. It is validated for hallux valgus surgery, not ing better results for SB fixation (Shimozono et al. 2018). ankle fractures, hence its properties for ankle fractures are not An explanation for this discrepancy could be the different known. After initiation of our trial, a comparison of 3 differmechanical properties between the screw options for fixation. ent PROMs available in Norwegian were published, recomThe dynamic properties of syndesmotic implants in vivo are mending the Self-Reported Foot and Ankle Score (SEFAS) for unknown, but there are mechanical studies on the subject. evaluating patients with ankle fractures (Garratt et al. 2018). Fixation of the syndesmosis with several 3.5 mm tricorti- Another weakness is the lack of standardization in the syncal SS or a 4.5 mm quadricortical SS locks the fibula in the desmosis fixation and several surgeons treating the patients. incisura, while the TS has in a cadaver study displayed more This could be a source of uncontrolled variability between the dynamic properties (Clanton et al. 2017). This may explain groups. On the other hand, it makes our results transferable to why Andersen et al. (2018) found a quadricortical SS to be the day-to-day practice of fracture surgery. Table 5. Radiological results: difference measured in mm in tibiofibular distance at level of syndesmosis (1 cm proximal to the ankle joint) between injured and uninjured side. Values are mean (SD) or median (IQR) unless otherwise specified

Acta Orthopaedica 2020; 91 (6): 770–775

The primary strengths of this study are the randomized prospective design with blinded scoring of clinical outcome measures, comparable groups at baseline, a high follow-up rate, and CT evaluation 2 years after treatment. In addition, all hospitals participating in the study used both implants as standard treatments before initiation of the trial, minimizing problems with the learning curve associated with new treatments. The procedure was performed by the on-call team, providing generalizability. Our outcome scores after 2 years are in line with scores from similar studies (Wikeroy et al. 2010, Laflamme et al. 2015, Andersen et al. 2018), supporting previous data on outcomes after syndesmotic injury. Interpretation In this RCT comparing a knotless SB and an TS we found no clinically relevant differences regarding outcome scores between the groups. TS is an inexpensive alternative to SB when treating acute syndesmotic injury. Supplementary data Tables 3, 4, 6, 7 and Figures 4–6 are available as supplementary data in the online version of this article, http://dx.doi.org/ 10.1080/17453674.2020.1818175

BWR, JEM, MRA, and WF planned and designed the study. BWR, IKS, JEM, FF, MRA, and WF were active in inclusion, treatment, and follow-up. SBJ and BWR did analysis of the radiologic examinations. BWR and MRA did statistical analysis with feedback from IKS, JEM, FF, and WF. BWR designed the tables and wrote the first draft of the paper; all authors revised the paper and tables. Acta thanks Maria C Cöster and Tatu Mäkinen for help with peer review of this study.

Andersen M R, Frihagen F, Madsen J E, Figved W. High complication rate after syndesmotic screw removal. Injury 2015; 46(11): 2283-7. Andersen M R, Frihagen F, Hellund J C, Madsen J E, Figved W. randomized trial comparing suture button with single syndesmotic screw for syndesmosis injury. J Bone Joint Surg Am 2018; 100(1): 2-12. Andersen M R, Diep L M, Frihagen F, Hellund J C, Madsen J E, Figved W. Importance of syndesmotic reduction on clinical outcome after syndesmosis injuries. J Orthop Trauma 2019; 33(8): 397-403. Barbosa P, Bonnaire F, Kojima K. AO surgery reference: Syndesmosis injury. https://surgeryreference.aofoundation.org/orthopedic-trauma/adulttrauma/malleoli/suprasyndesmotic-simple-fibular- fracture-medial-injuryposterior-fracture/orif?searchurl =%2fSearchResults#fixation; 2020. Clanton T O, Whitlow S R, Williams B T, Liechti D J, Backus J D, Dornan G J, et al. Biomechanical comparison of 3 current ankle syndesmosis repair techniques. Foot Ankle Int 2017; 38(2): 200-7. Dawson J, Doll H, Coffey J, Jenkinson C, Oxford and Birmingham Foot and Ankle Clinical Research Group. Responsiveness and minimally important change for the Manchester–Oxford foot questionnaire (MOXFQ) compared with AOFAS and SF-36 assessments following surgery for hallux valgus. Osteoarthritis Cartilage 2007; 15(8): 918-31. Dawson J, Boller I, Doll H, Lavis G, Sharp R, Cooke P, et al. The MOXFQ patient-reported questionnaire: assessment of data quality, reliability and validity in relation to foot and ankle surgery. Foot 2011; 21(2): 92-102.

775

Garner M R, Fabricant P D, Schottel P C, Berkes M B, Shaffer A D, Ni A, et al. Standard perioperative imaging modalities are unreliable in assessing articular congruity of ankle fractures. J Orthop Trauma 2015; 29(4): e161-5. Garratt A M, Naumann M G, Sigurdsen U, Utvåg S E, Stavem K. Evaluation of three patient reported outcome measures following operative fixation of closed ankle fractures. BMC Musculoskelet Disord 2018; 19: 134. Hinds R M, Lazaro L E, Burket J C, Lorich D G. Risk factors for posttraumatic synostosis and outcomes following operative treatment of ankle fractures. Foot Ankle Int 2014; 35(2): 141-7. Kortekangas T, Savola O, Flinkkila T, Lepojarvi S, Nortunen S, Ohtonen P, et al. A prospective randomized study comparing TightRope and syndesmotic screw fixation for accuracy and maintenance of syndesmotic reduction assessed with bilateral computed tomography. Injury 2015; 46(6): 1119-26. Laflamme M, Belzile E L, Bédard L, van den Bekerom M P J, Glazebrook M, Pelet S. A prospective randomized multicenter trial comparing clinical outcomes of patients treated surgically with a static or dynamic implant for acute ankle syndesmosis rupture. J Orthop Trauma; 2015; 29(5): 216-23. Lübbeke A, Salvo D, Stern R, Hoffmeyer P, Holzer N, Assal M. Risk factors for post-traumatic osteoarthritis of the ankle: an eighteen-year follow-up study. Int Orthop 2012; 36(7): 1403-10. McKenzie A C, Hesselholt K E, Larsen M S, Schmal H. A systematic review and meta-analysis on treatment of ankle fractures with syndesmotic rupture: suture-button fixation versus cortical screw fixation. J Foot Ankle Surg 2019; 58(5): 946-53. Morley D, Jenkinson C, Doll H, Lavis G, Sharp R, Cooke P, et al. The Manchester–Oxford Foot Questionnaire (MOXFQ): development and validation of a summary index score. Bone Joint Res 2013; 2(4): 66-9. Norman G R, Sloan J A, Wyrwich K W. Interpretation of changes in healthrelated quality of life: the remarkable universality of half a standard deviation. Med Care 2003; 41(5): 582-92. Olerud C, Molander H. A scoring scale for symptom evaluation after ankle fracture. Arch Orthop Trauma Surg 1984; 103(3): 190-4. Patel S, Malhotra K, Cullen N P, Singh D, Goldberg A J, Welck M J. Defining reference values for the normal tibiofibular syndesmosis in adults using weight-bearing CT. Bone Joint J 2019; 101-B(3): 348-52. Piaggio G, Elbourne DR, Pocock S J, Evans S J W, Altman D G, CONSORT Group. Reporting of noninferiority and equivalence randomized trials: extension of the CONSORT 2010 statement. JAMA 2012;3 08(24): 2594-604. Ramsey D C, Friess D M. Cost-effectiveness analysis of syndesmotic screw versus suture button fixation in tibiofibular syndesmotic injuries. J Orthop Trauma 2018; 32(6): e198-e203. Ray R, Koohnejad N, Clement N D, Keenan G F. Ankle fractures with syndesmotic stabilization are associated with a high rate of secondary osteoarthritis. Foot Ankle Surg 2019; 25(2): 180-5. Riedel M D, Miller C P, Kwon Y. Augmenting suture-button fixation for Maisonneuve injuries with fibular shortening: technique tip. Foot Ankle Int 2017; 38(10): 1146-51. Saltzman C L, Salamon M L, Blanchard G M, Huff T, Hayes A, Buckwalter J A, et al. Epidemiology of ankle arthritis: report of a consecutive series of 639 patients from a tertiary orthopedic center. Iowa Orthop J 2005; 25: 44-6. Schepers T, Van Lieshout E M, de Vries M R, Van der Elst M. Complications of syndesmotic screw removal. Foot Ankle Int 2012; 32(11): 1040-4. Shimozono Y, Hurley E T, Myerson C L, Murawski C D, Kennedy J G. Suture button versus syndesmotic screw for syndesmosis injuries: a metaanalysis of randomized controlled trials. Am J Sports Med 2018; 100(1): 363546518804804. Veltman E S, Hofstad C J, Witteveen A G H. Are current foot- and ankle outcome measures appropriate for the evaluation of treatment for osteoarthritis of the ankle: evaluation of ceiling effects in foot- and ankle outcome measures. J Foot Ankle Surg 2017; 23(3): 168-72. Wikeroy A K, Hoiness P R, Andreassen G S, Hellund J C, Madsen J E. No difference in functional and radiographic results 8.4 years after quadricortical compared with tricortical syndesmosis fixation in ankle fractures. J Orthop Trauma 2010; 24(1): 17-23.

776

Acta Orthopaedica 2020; 91 (6): 776–781

Low arthroplasty survival after treatment for proximal humerus fracture sequelae: 3,245 shoulder replacements from the Nordic Arthroplasty Register Association Ditte UNBEHAUN 1, Sigrid RASMUSSEN 1, Randi HOLE 2, Anne Marie FENSTAD 2, Björn SALOMONSSON 3, Yilmaz DEMIR 3, Steen Lund JENSEN 4, Stig BRORSON 5, Ville ÄÄRIMAA 6, Inger MECHLENBURG 7, and Jeppe Vejlgaard RASMUSSEN 8 1 Department of Public Health, Aarhus University, Denmark; 2 Norwegian Arthroplasty Register, Department of Orthopaedic Surgery, Haukeland University Hospital, Bergen, Norway; 3 Department of Clinical Sciences, Danderyd Hospital, Division of Orthopedics, Karolinska Institutet, Stockholm, Sweden; 4 Department of Orthopaedic Surgery, Aalborg University Hospital and Clinical Medicine, Aalborg University, Aalborg, Denmark; 5 Department of Orthopaedic Surgery, Zealand University Hospital, Department of Clinical Medicine, University of Copenhagen, Denmark; 6 Departments of Orthopaedics and Traumatology, Turku University and University Hospital, Turku, Finland; 7 Department of Orthopaedic Surgery, Aarhus University Hospital, Department of Clinical Medicine, Aarhus University; 8 Department of Orthopaedic Surgery, Herlev Hospital, Department of Clinical Medicine, University of Copenhagen, Denmark Correspondence: ditte.unbehaun@gmail.com Submitted 2019-03-25. Accepted 2020-06-09.

Background and purpose — Proximal humerus fractures (PHF) may result in sequelae indicating arthroplasty. We report cumulative survival rates and reasons for revision after arthroplasty for proximal humerus fracture sequelae (PHFS). Patients and methods — Data were derived from the Nordic Arthroplasty Register Association. The Kaplan– Meier method was used to illustrate survival rates. A scaled Schoenfeld residual plot was used to report the risk of revision for men relative to women in patients who were treated with reverse shoulder arthroplasty (RSA). Revision was defined as removal or exchange of any component or the addition of a glenoid component. Results — 30,190 primary arthroplasties were reported from 2004 to 2016, of which 3,245 were for PHFS. The estimated 1-, 5-, and 10-year cumulative survival rates (95% CI) were 96% (95–97), 90% (89–92), and 86% (83–88) for stemmed hemiarthroplasty and 94% (92–95), 89% (87–91), and 86% (82–90) for RSA with a median time to revision of 18 months (IQR 9–44) and 3 months (IQR 0–17). The risk of revision for men relative to women in patients who were treated with RSA was 3.2 (1.9–5.1) 0–1 year after surgery and 1.9 (0.9–4.1) 1–8 years after surgery. The estimated 1-, 5-, and 10-year cumulative survival rates (95% CI) were 94% (92–96), 88% (85–90), and 80% (75–86) for men and 95% (94–96), 86% (84–89), and 81% (77–84) for young patients. Interpretation — Shoulder arthroplasty for PHFS was associated with lower survival rates, compared with previously published results of shoulder arthroplasty for acute PHF. The low arthroplasty survival rates for men and young patients especially are worrying.

Both non-surgical and surgical treatments of proximal humerus fractures (PHF) are associated with risk of nonunion, malunion, or avascular necrosis and secondary osteoarthritis of the glenohumeral joint. These complications may later clinically manifest as proximal humerus fractures sequelae (PHFS) with pain, stiffness, and decreased range of motion (Mansat and Bonnevialle 2015, Boileau 2016). There is not yet consensus on the optimal treatment of PHFS (Kilic et al. 2010, Alentorn-Geli et al. 2014, Jacobson et al. 2014, Raiss et al. 2014). Stemmed hemiarthroplasty (SHA) came into common usage in the treatment of acute PHF and PHFS in the 1990s, but during the last decade reverse shoulder arthroplasty (RSA) has become increasingly popular (Wand et al. 2012). By using the RSA design, shoulder stability and function can be improved even with a compromised rotator cuff (Namdari et al. 2013, Nikola et al. 2015). The majority of previous studies of shoulder arthroplasty for PHFS have focused on pain, range of motion, and functional outcome scores (Boileau et al. 2006, Murray et al. 2011, Moineau et al. 2012, Alentorn-Geli et al. 2014, Raiss et al. 2014, Nikola et al. 2015, Hattrup et al. 2016, Raiss et al. 2016, Raiss et al. 2017). Only a few case series have reported revision rates (Boileau et al. 2001, Mansat and Bonnevialle 2015). We report cumulative survival rates and reasons for revision after shoulder arthroplasty for PHFS.

Patients and methods Data were derived from the Nordic Arthroplasty Register Association (NARA), which is a collaboration between the

Acta Orthopaedica 2020; 91 (6): 776–781

777

Arthroplasty survival (%) national shoulder arthroplasty registries Count 100 400 in Sweden, Denmark, Norway, and FinStemmed hemiarthroplasty Anatomical total shoulder arthroplasty land (Rasmussen et al. 2016). The FinnReverse shoulder arthroplasty Other types ish data were not included in the pres300 ent study because of incomplete format. 90 The completeness of the shoulder registries is above 90% in Denmark and Norway and above 80% in Sweden for 200 both primary and revision arthroplasties 80 (Rasmussen et al. 2016). From January 100 2004 to December 2016 NARA contains data from Sweden, Denmark and Norway on 30,190 shoulder arthroplas70 0 ties. 9,137 arthroplasties were used for 0 2 4 6 8 10 12 2004 2006 2008 2010 2012 2014 2016 Years from surgery acute PHF, 3,245 for PHFS, 1,976 for Figure 1. Number of stemmed hemiarthroplas0 1 2 3 4 5 6 7 8 9 10 inflammatory arthritis, 3,324 for rotator ties, anatomical total shoulder arthroplasties, Year At risk 3,245 2,598 2,072 1,577 1,185 881 606 368 205 88 5 No. at risk reverse shoulder arthroplasties and other cuff problems, and 11,647 for osteoar3245 2598 2072 1577 1185 881 606 368 205 88 Figure 2. Cumulative survival for all types5 of types including stemless, resurfacing, andSubjects: thritis. In 861 arthroplasties, the diag- metaphyseal fixed implant arthroplasties. arthroplasties from 2004 to 2016 with 95% CI and numbers at risk. nosis was recorded as “Others” or was missing. Statistics PHFS is defined by NARA as fractures reported as nonunion, malunion, fracture with previous Descriptive statistics were used to report demographic data, osteosynthesis, or a healed fracture reported together with follow-up time, time to revision, and reasons for revision. osteoarthritis or humeral head necrosis. The NARA data set We used the Kaplan–Meier method to illustrate the unadincludes patient-related data (nationality, age, sex, and diag- justed survival rates with 95% confidence interval (CI). nosis), operative data (date, arthroplasty type, and brand), and Due to violation of the proportional hazards assumption, a revision data (date, reason for revision, and new arthroplasty scaled Schoenfeld residual plot was used to report the risk brand). of revision for men relative to women in patients who were Type of shoulder arthroplasty is reported as stemmed hemi- treated with RSA. The estimated risk of revision was calarthroplasty, anatomical total shoulder arthroplasty, reverse culated for 2 separate intervals to fulfill the proportional shoulder arthroplasty, resurfacing hemiarthroplasty, resurfac- hazard assumption. The comparison was adjusted for age ing, total shoulder arthroplasty, stemless hemiarthroplasty, and period of surgery. Although it violates the assumption of or stemless total shoulder arthroplasty. Information on stem independence, patients with bilateral shoulder arthroplasty length and fixation technique is not included in the dataset. procedures were included in the survival analysis as if they In the comparison of arthroplasty types, we included only were independent. The level of statistical significance was stemmed hemiarthroplasty and reverse shoulder arthroplasty. set at p < 0.05 and all tests were 2-tailed. The analyses were The other arthroplasty types were not included because of few performed using SPSS version 22.0 (IBM Corp, Armonk, cases. NY, USA). Revision of an arthroplasty was defined as removal or exchange of any components or the addition of a glenoid com- Ethics, funding, and potential conflicts of interest. ponent. If more than 1 reason for revision had been reported Ethics committee approval was not required. No funding was to the individual registries, the following hierarchy of rea- received for this study. No competing interests are declared. sons for revision was used, so only 1 reason for revision was registered in the common data set: Infection; Periprosthetic fracture; Luxation and instability; Loosening; Rotator cuff problem; and “Other reasons,” which include glenoid wear, Results malposition of the arthroplasty, and pain with no other reasons The annual number of arthroplasties and especially the number reported (Rasmussen et al. 2016). In all the Nordic countries of RSAs increased during the study period (Figure 1). Mean each person is identified by a unique civil registration number age was 69 years (SD 12). Median follow-up time was 45 given at birth. The civil registration number was used in the months (IQR 20–81). There were 306 (9.4%) revisions of all national registries to accurately link the revision procedure to types of arthroplasty with a 12-year cumulative survival rate the primary arthroplasty and to check for death and emigration of 84% (Figure 2). Median time to revision was 15 months in the national population registries. The end of follow-up was (IQR 3 to 33). Overall, the most common reasons for revision were Luxation and instability (3.2%), “Other reasons” the date of revision, date of death or December 31, 2016.

778

Acta Orthopaedica 2020; 91 (6): 776–781

Arthroplasty survival (%)

Reverse shoulder arthroplasty survival (%)

Log RR(t) reverse shoulder arthroplasty – Sex

100

Stemmed hemiarthroplasty Reverse shoulder arthroplasty

Women Men

4 90

80 RR = 1.0

70 0

–2 0

Years from surgery

Year 0 1 2 3 4 5 6 7 8 9 10 RSA 1,152 829 595 393 247 153 92 46 19 6 0 SHA 1,587 1,357 1,147 946 755 591 423 268 158 71 5

Year 0 1 2 3 4 5 6 7 8 9 10 Men 250 181 146 118 92 66 45 32 18 11 6 Women 902 769 623 502 393 315 235 178 136 105 80

Figure 3. Cumulative survival for stemmed hemiarthroplasty and reverse shoulder arthroplasty from 2004 to 2016 with 95% CI and numbers at risk.

Figure 4. Cumulative survival for women and men treated with RSA with 95% CI and numbers at risk.

Reasons for revision for all types of arthroplasties (All), stemless, resurfacing, and metaphyseal fixed implant arthroplasties (Other types), stemmed hemiarthroplasty (SHA), and reverse shoulder arthroplasty (RSA). Values are number (n), percentage of primary arthroplasties in parentheses, and percentage (%) of revisions Reason

All n (%)

Infection 53 (1.6) 17 Periprosthetic fracture 13 (0.4) 4 Luxation and instability 105 (3.2) 34 Loosening 21 (0.6) 7 Rotator cuff problems 35 (1.1) 11 a Other reasons 65 (2.0) 21 Missing 14 (0.4) 5 Total 306 (9.4) 100

Other types n (%) % 6 (1.2) 3 (0.6) 10 (2.0) 8 (1.6) 9 (1.8) 11 (2.2) 3 (0.6) 50 (9.9)

12 6 20 16 18 22 6 100

SHA n (%) %

RSA n (%)

25 (1.6) 16 7 (0.4) 5 39 (2.5) 26 4 (0.3) 3 25 (1.6) 16 41 (2.6) 27 11 (0.7) 7 152 (9.6) 100

22 (1.9) 3 (0.3) 56 (4.9) 9 (0.8) 1 (0.1) 13 (1.1) 0 (0) 104 (9.0)

22 3 54 9 1 13 0 100

a includes glenoid wear, malposition of the arthroplasty, and pain with no other reasons reported.

(including glenoid wear) (2.0%) and Infection (1.6%) (Table). 889 (27%) patients died within the study period. Type of arthroplasty There were 1,587 SHAs and 1,152 RSAs. 502 arthroplasties were categorized as “Others” and for 4 arthroplasties the arthroplasty type was missing. 152 (9.6%) SHAs and 104 (9.0%) RSAs were revised. The median time to revision was 18 months (IQR 9–44) for SHA and 3 months (IQR 0–17) for RSA. The most common reason for revision was Luxation and instability for RSA and “Other reasons” (including glenoid wear) for SHA (Table). The estimated 1-, 5-, and 10-year cumulative survival rates (95% CI) were 96% (95–97), 90% (89–92), and 86% (83–88) for stemmed hemiarthroplasty and 94% (92–95), 89% (87–91) and 86% (82– 90), for reverse shoulder arthroplasty (Figure 3).

Years from surgery

Figure 5. Risk of any revision (solid black line) with 95% CI (dashed lines) for men relative to women treated with RSA adjusted for age and period of surgery. The horizontal red line indicates no difference in risk of revision (RR = 1). The RR estimates are divided into 2 intervals to fulfill the proportional hazard assumption: 0–1 year: RR = 3,2 (1.9–5.1) and 1–8 years: RR = 1.9 (0.9–4.1).

250 men and 902 women were treated with RSA during the study period. The estimated cumulative survival rates (95% CI) for these patients at 1 and 5 years were 87% (83–91), and 78% (71–85) for men and 96% (94–97) and 92% (90–94) for women (Figure 4). The risk of revision for men relative to women in patients who were treated with RSA was 3.2 (1.9–5.1) 0–1 year after surgery and 1.9 (0.9–4.1) 1–8 years after surgery (Figure 5).

Sex 2,422 (75%) of the study population were women. 97 (12%) men and 209 (9%) women were revised. The estimated 1-, 5-, and 10-year cumulative survival rates (95% CI) were 94% (92–96), 88% (85–90), and 80% (75–86) for men and 96% (95–97), 90% (89–92), and 87% (85–89) for women (Figure 6). Age 2,061 (63%) of the study population were older than 65 years at the time of surgery. A total of 152 (13%) patients at the age of 65 years or younger and 154 (7.5%) patients older than 65 years were revised. The estimated 1-, 5-, and 10-year cumulative survival rates (95% CI) were 95% (94–96), 86% (84–89), and 81% (77–84) for young patients and 96% (95–97), 92%

Acta Orthopaedica 2020; 91 (6): 776–781

779

Arthroplasty survival (%)

Stemmed hemiarthroplasty survival (%)

100

100 > 65 years ≤ 65 years

Women Men

> 65 years ≤ 65 years

70 0

Years from surgery

Year 0 1 2 3 4 5 6 7 8 9 10 Men 823 629 501 380 272 198 126 84 45 21 0 Women 2,422 1,969 1,517 1,197 913 683 480 284 160 67 5

Figure 6. Cumulative survival for women and men from 2004 to 2016 with 95% CI and numbers at risk.

Years from surgery

Year 0 1 2 3 4 5 6 7 8 9 10 ≥ 65 1,184 962 777 616 467 368 267 168 99 46 2 > 65 2,061 1,636 1,295 961 718 413 339 200 106 42 3

Year 0 1 2 3 4 5 6 7 8 9 10 ≥ 65 621 531 447 384 308 251 190 118 73 36 2 > 65 966 826 700 562 447 340 233 150 85 35 3

Figure 7. Cumulative survival for patients who were 65 years or younger and older than 65 years from 2004 to 2016 with 95% CI and numbers at risk.

Figure 8. Cumulative survival for patients who were 65 years or younger and older than 65 years treated with SHA from 2004 to 2016 with 95% CI and numbers at risk.

(90–93), and 89% (86–91) for older patients (Figure 7). In particular, young patients treated with SHA had a low arthroplasty survival rate (Figure 8).

Discussion We found that the 12-year cumulative survival rate after PHFS was 85% for SHA and 86% for RSA. This is clearly lower than survival rates found for acute PHF (Brorson et al. 2017). A plausible explanation may be that it is more difficult, with more surgical trauma added, to insert an arthroplasty when the indication is PHFS relative to acute PHF. Furthermore, patients with PHFS often have a long history of pain and poor range of motion prior to the operation, which may induce a subsequent insufficiency and stiffness of both rotator cuff muscles and tendons. This might influence the choice of implant as well as the postoperative outcome to an unknown extent. Finally, some patients are treated for PHFS because of failed osteosynthesis. This may increase the risk of revision because of periprosthetic joint infection relative to acute PHF. As the threshold for revision is expected to be high, the low survival rate for PHFS compared with acute PHF is noteworthy. 5% of RSAs were revised because of luxation and instability, which can be caused by difficulties placing the arthroplasty at the right height with the right tension when the natural anatomy cannot be used as a guidance. The literature is not unanimous in its reports on complication rates and revision rates regarding RSA and SHA for PHFS. A study (Alentorn-Geli et al. 2014) comparing 12

patients treated with SHA and 20 patients treated with RSA for PHFS found no complications that required revision, but SHA demonstrated a higher number of complications compared with RSA. Another study (Kilic et al. 2010) comparing 19 patients treated with RSA and 36 patients treated with an anatomic arthroplasty found a higher revision rate (11%) and complication rate (25%), for RSA compared with anatomic arthroplasty. However, Alentorn-Geli et al. reported only complications that lead to revision, whereas Kilic et al. reported both minor and major complications. Comparison of complications described in the literature is challenging due to different definitions and reporting. Moreover, most studies on PHFS include small retrospective series and thus a high risk of bias. We found lower arthroplasty survival at 1 year and shorter time to revision for RSA compared with SHA. This is in line with other studies, where the complications associated with RSA appeared early after surgery (Namdari et al. 2013,). Our study confirm that a short follow-up time shows a different complication and revision rate for RSA relative to SHA, compared with what is seen with a long follow-up time (Ferrel et al. 2015). A systematic literature review (Mansat and Bonnevialle 2015) reported the risk of revision to be 3.5–35% after treatment with arthroplasty for PHFS. In this review, the revision rate did not differentiate between RSA and SHA and the wide range in risk of revision illustrates the uncertainty of the results in small case series. Several studies suggest that SHA and RSA must be differentiated, which is why confounding by indication may influence the comparison of SHA and RSA. SHA is suitable for less severe PHFS without the need for a greater tuberosity osteotomy, whereas RSA is recommended for severe PHFS when

780

a greater tuberosity osteotomy is needed (Kilic et al. 2010, Raiss et al. 2014, Mansat and Bonnevialle 2015). A higher complication and revision rate for RSA can be expected if the arthroplasty is used for the most severe cases (Kilic et al. 2010, Raiss et al. 2014). However, a revision after failed RSA can be more challenging than after hemiarthroplasty and some surgeons may hesitate to revise an RSA despite a poor functional outcome. This would lead to an underestimation of failures of RSA overall and relative to hemiarthroplasty (Namdari et al. 2013, Brorson et al. 2017). There may be different indications for both primary and revision arthroplasty not only among countries but also among regions, hospitals, surgeons, and maybe also for the same surgeon from time to time. This may also be the reason for the different revision rates reported by single-center studies. Our study has limitations. Information on the fracture, such as morphology, initial fracture treatment, type of sequelae, and migration of the greater tuberosity as well as patient-related factors such as smoking, obesity, and comorbidity were not included in the dataset. Also, the level of surgical experience may influence the choice of arthroplasty, and subsequently the revision rates (Murray et al. 2011). The completeness of the shoulder registries must also be addressed. It is not known whether the number of non-registered patients differs from the patients who were registered. Finally, it is important to be aware that an unknown number of failures are never revised and that some revisions can lead to a good functional outcome. Therefore, they cannot be considered as failures in a later follow-up. Thus, the reported survival rates may not reflect the functional outcome for the patients. Inclusion of patient-reported outcome could have added valuable information, but this was not possible due to the lack of comparable reporting of patient-reported outcomes in the Nordic countries. 27% of the patients died during the study period which, of course, precludes the occurrence of a subsequent revision. This introduces competing risk to the Kaplan–Meier method and the Cox proportional hazard regression model, which, in theory, would overestimate the revision rates. Nevertheless, the Kaplan–Meier method and the Cox proportional hazard regression model is believed to give adequate estimates of the revision risk (Ranstam et al. 2011, Ranstam and Robertsson 2017, Sayers et al. 2018). In summary, shoulder arthroplasty for PHFS was associated with a lower survival rate, especially for men and younger patients, compared with previously published results of shoulder arthroplasty for acute PHF. The low arthroplasty survival rates especially for men treated with RSA and young patients treated with SHA are worrying. These results are pertinent when deciding on the treatment of PHFS. The low survival rate also indicates that it is important to be critical in the choice of treatment when it comes to initial fracture management, to avoid increased risk with joint replacement as treatment of PHFS.

Acta Orthopaedica 2020; 91 (6): 776–781

All authors took part in conception and design of study and in interpretation of the results. AMF, RH, BS, SLJ, and JVR prepared data from the national registries. JVR, AMF, DU, and SR performed the statistical analysis. DU, SR, IM, VA, and JVR participated in the preparation of the manuscript. DU and SR incorporated input from all the other authors and were responsible for writing the manuscript. Acta thanks Esa Jämsen and Patrick Sadoghi for help with peer review of this study.

Alentorn-Geli E, Guirro P, Santana F, Torrens C. Treatment of fracture sequelae of the proximal humerus: comparison of hemiarthroplasty and reversed total shoulder arthroplasty. Arch Orthop Trauma Surg 2014; 134(11): 1545-50. Boileau P. Complications and revision of reverse total shoulder arthroplasty. Orthop Traumatol Surg Res 2016; 102: 33-43. Boileau P, Trojani C, Walch G, Krishnan S G, Romeo A, Sinnerton R. Shoulder arthroplasty for the treatment of the sequelae of fractures of the proximal humerus. J Shoulder Elbow Surg 2001; 10(4): 299-308. Boileau P, Chuinard C, Le Huec J C, Walch G, Trojani C. Proximal humerus fracture sequelae: impact of a new radiographic classification on arthroplasty. Clin Orthop Relat Res 2006; (442): 121-30. Brorson S, Salomonsson B, Jensen S L, Fenstad A M, Demir Y, Rasmussen J V. Revision after shoulder replacement for acute fracture of the proximal humerus. Acta Orthop 2017; 88(4): 446-50. Ferrel J R, Trinth T Q, Fischer R A. Reverse total shoulder arthroplasty versus hemiarthroplasty for proximal humeral fractures: a systematic review. J Orthop Trauma 2015; 29(1): 60-8. Hattrup S J, Waldrop R, Sanchez-Sotelo J. Reverse total shoulder arthroplasty for posttraumatic sequelae. J Orthop Trauma 2016; 30(2): 41-7. Jacobson A, Duquin T R, Sanchez-Sotelo J, Schleck C D, Sperling J W, Cofield R H. Anatomic shoulder arthroplasty for treatment of proximal humerus malunions. J Shoulder Elbow Surg 2014; 23(8): 1232-9. Kilic M, Berth A, Blatter G, Fuhrmann U, Gebhardt K, Rott O, Zenz P, Irlenbusch U. Anatomic and reverse shoulder protheses in fracture sequelae of the humeral head. Acta Orthop Traumatol Turc 2010; 44(6): 417-25. Mansat P, Bonnevialle N. Treatment of fracture sequelae of the proximal humerus: anatomical vs reverse shoulder prosthesis. Int Orthop 2015; 39(2): 349-54. Moineau G, McClelland W B, Trojani C, Rumian A, Walch G, Boileau P. Prognostic factors and limitations of anatomic shoulder arthroplasty for the treatment of posttraumatic cephalic collapse or necrosis (type-1 proximal humeral fracture sequelae). J Bone Joint Surg Am 2012; 94(23): 2186-94. Murray I R, Amin A K, White T O, Robinson C M. Proximal humeral fractures: current concepts in classification, treatment and outcomes. J Bone Joint Surg Br 2011; 93(1): 1-11. Namdari S, Horneff J G, Baldwin K. Comparison of hemiarthroplasty and reverse arthroplasty for treatment of proximal humeral fractures. J Bone Joint Surg 2013; 95(18): 1701-8. Nikola C, Klobucar H, Medancic N. Reverse shoulder arthroplasty in acute fractures provides better results than in revision procedures for fracture sequelae. Int Orthop 2015; 39(2): 343-8. Raiss P, Edwards T B, Silva M R, Bruckner T, Loew M, Walch G. Reverse shoulder arthroplasty for the treatment of nonunions of the surgical neck of the proximal part of the humerus (type-3 fracture sequelae). J Bone Joint Surg 2014; 96(24): 2070-6. Raiss P, Edwards T B, Collin P, Bruckner T, Zeifang F, Loew M, Boileau P, Walch G. Reverse shoulder arthroplasty for malunions of the proximal part of the humerus (type-4 fracture sequelae). J Bone Joint S 2016; 98(11): 893-9. Raiss P, Edwards T B, Bruckner T, Loew M, Zeifang F, Walch G. Reverse arthroplasty for patients with chronic locked dislocation of the shoulder (type 2 fracture sequela). J Shoulder Elbow Surg 2017; 26(2): 279-87.

Acta Orthopaedica 2020; 91 (6): 776–781

Ranstam J, Kärrholm J, Pulkkinen P, Mäkelä K, Espehaug B, Pedersen A B, Mehnert F, Furnes O; NARA study group. Statistical analysis of arthroplasty data, II: Guidelines. Acta Orthop 2011; 82(3): 258-67. Ranstam J, Robertsson O. The Cox model is better than the Fine and Gray model when estimating relative revision risks from arthroplasty register data. Acta Orthop 2017; 3674: 1-3. Rasmussen J V, Brorson S, Hallan G, Dale H, Äärimaa V, Mokka J, Jensen S, Fenstad A M, Salomonsson B. Is it feasible to merge data

781

from national shoulder registries? A new collaboration within the Nordic Arthroplasty Register Association. J Shoulder Elbow Surg 2016; 25(12): 369-377. Sayers A, Evans J T, Whitehouse MR, Blom AW. Are competing risks models appropriate to describe implant failure? Acta Orthop 2018; 89(3): 256-8. Wand R J, Dear K E, Biqsby E, Wand J S. A review of shoulder replacement surgery. J Perioper Pract 2012; 22(11): 354-9.

782

Acta Orthopaedica 2020; 91 (6): 782–788

Rotator cuff repair with biological graft augmentation causes adverse tissue outcomes Mustafa S RASHID, Richard D J SMITH, Navraj NAGRA, Kim WHEWAY, Bridget WATKINS, Sarah SNELLING, Stephanie G DAKIN, and Andrew J CARR

Nuffield Department of Orthopaedics, Rheumatology, and Musculoskeletal Sciences (NDORMS), Botnar Research Centre, Oxford, UK Correspondence: mustafa.rashid@ndorms.ox.ac.uk Submitted 2020-04-11. Accepted 2020-06-30.

Background and purpose — Biological patches can be used to augment rotator cuff tendon repair in an attempt to improve healing and reduce rates of re-rupture. However, little is known about the in vivo tissue response to these patches. We assessed native rotator cuff tissue response after surgical repair and augmentation with 2 commercially available extracellular matrix (ECM) patches. Patients and methods — Patients underwent a rotator cuff repair augmented with either GraftJacket (Wright Medical), Permacol (Zimmer Biomet), or no patch (Control), applied using an onlay technique. A sample of supraspinatus tendon was collected intraoperatively and 4 weeks post-surgery, using ultrasound-guided biopsy. Histology and immunohistochemistry were performed on all samples. Results — The Permacol group (n = 3) and GraftJacket group (n = 4) demonstrated some changes in native tendon ECM compared with the control group (n = 3). Significant disruption of the extracellular matrix of the repaired native supraspinatus, underlying both patches, was observed. The patches did not generally increase cellularity, foreign body giant cell count, or vascularity compared to the control group. 1 patient in the Permacol group had an adverse tissue immune response characterized by extensive infiltration of IRF5+, CD68+, and CD206+ cells, suggesting involvement of macrophages with a pro-inflammatory phenotype. No significant differences in protein expression of CD4, CD45, CD68, CD206, BMP7, IRF5, TGFß, and PDPN were observed among the groups. Interpretation — Histological and immunohistochemical analysis of native tendon tissue after patch augmentation in rotator cuff repair raises some concerns about a lack of benefit and potential for harm from these materials.

Rotator cuff tendon tears occur in 1 in 3 people aged over 60 years (Tempelhof et al. 1999). Around 17,000 rotator cuff repairs are performed in the National Health Service (NHS) in the UK each year (Digital 2016). The incidence of rotator cuff repair is increasing in the UK and the USA (Colvin et al. 2012). Numerous observational studies have attempted to describe the healing rate following cuff repair (Russell et al. 2014, Shen et al. 2014, Yang et al. 2017). Despite the evolution in technique and implants, the overall healing rate is around 60% (Carr et al. 2017). This has led surgeons to develop innovative strategies that aim to augment tendon repair and improve healing rate. One rotator cuff tendon repair augmentation strategy involves the application of a patch overlying the repair. These patches may be biological or synthetic. Biological patches are designed to become incorporated and vascularized by the native tendon, adding essential matrix proteins for healing (Zimmer 2006). Biological graft sources may be from the patient him/herself (e.g., fascia lata autograft), from cadaveric donors (e.g., dermal allograft), or from porcine tissues (e.g., dermal or small intestine submucosa, xenograft). These biological patches, sometimes called extracellular matrix (ECM) patches, are processed to remove donor cells, and sometimes chemically crosslinked, before sterilization for clinical use (Zimmer 2006, Group 2017). 2 popular biological patches, available for clinical use in rotator cuff repair, are GraftJacket (Wright Medical, Memphis, TE, USA) and Permacol (Zimmer Biomet, Warsaw, IN, USA). GraftJacket Regenerative Tissue Matrix (RTM) (manufactured by LifeCell Corporation, Branchburg, NJ, USA) is a cadaveric human dermis graft that is not crosslinked and undergoes decellularization by a proprietary process (Group 2017). Permacol (manufactured by Tissue Science Laboratories PLC, Aldershot, UK) is a porcine dermis graft that is

Acta Orthopaedica 2020; 91 (6): 782–788

chemical crosslinked with 4,4′-Diisocyanato-methylenedicyclohexane (HMDI), and is decellularized by a proprietary process. Both GraftJacket and Permacol patches are marketed with some supporting information from in vitro and animal studies, showing cellular infiltration and neovascularization; however, the mechanisms underpinning these observations are unclear (McQuillan and Harper 2007, Xu et al. 2009, O’Brien et al. 2011, Xu et al. 2012). The in vivo tissue response to xenograft and allograft tissue is important to consider in patch augmentation in humans. Patch augmentation in rotator cuff repair carries some additional risks. These include foreign-body reaction, sterile inflammatory response, transmission of undiagnosed malignancy, and infectious disease transmission (Hinsenkamp et al. 2012). We ascertained the tissue response of the native supraspinatus tendon to 2 biological patches at 4 weeks compared with a control (no patch), using histology and immunohistochemistry.

783

Figure 1. a and b: The needle site used at the 4-week postoperative biopsy; c and d: the core biopsy needle and automated firing device used; e and f: the technique for ultrasound-guided core biopsy. Red line denotes supraspinatus bony footprint. Dotted yellow line denotes core biopsy needle; solid yellow line denotes trajectory of core biopsy needle within supraspinatus tendon when deployed.

Patients and methods This study was conducted on patients undergoing rotator cuff repair. 2 groups of patients underwent rotator cuff repair and augmentation with 2 types of patches. These were GraftJacket (Wright Medical, Memphis, TE, USA) and Permacol (Zimmer Biomet, Warsaw, IN, USA). The control group received conventional rotator cuff repair without augmentation. The primary endpoint was an ultrasound-guided core biopsy sample of all patients in the 3 groups (GraftJacket, Permacol, or control), 4 weeks after surgery. The 4-week time point was chosen to represent early tissue response, and initial inflammatory response. Inclusion criteria included patients with a symptomatic, atraumatic, full-thickness tear involving the supraspinatus ± infraspinatus tendon(s), confirmed with ultrasound or magnetic resonance imaging (MRI). Patients who failed nonoperative treatment including physiotherapy, rest, analgesia, and/or corticosteroid injection(s) were included. Exclusion criteria included partial thickness tears, irreparable tears, acute traumatic tears, subscapularis tears, and patients who had not undergone a trial of conservative management. The primary outcome was native supraspinatus tendon tissue response, assessed by H&E staining, at 4 weeks post-surgery. Secondary outcomes included inflammatory response within the native supraspinatus tendon, as assessed by immunohistochemistry panel, at 4 weeks post-surgery.

Baseline measurements including Oxford Shoulder Score (OSS), Visual Analogue Score (VAS), and EuroQol 5D questionnaire (EQ-5D) were recorded to help define the patient cohort, but not for clinical outcome evaluation. 13 individuals were sequentially allocated into 3 groups, Group 1 (GraftJacket augmentation, 4 patients), Group 2 (Permacol augmentation, 4 patients), and Group 3 (control – standard repair without augmentation, 5 patients). 3 patients were excluded from the tissue analysis. The reasons for exclusion were: partial thickness tear (n = 1), irreparable massive tear (n = 1), and postoperative deep infection requiring arthroscopic washout and debridement (n = 1). The patient with the partial thickness tear was allocated to the Permacol group and the latter 2 patients were in the control group. Thus, Group 1 (GraftJacket augmentation) included 4 patients; Group 2 (Permacol augmentation) included 3 patients; and Group 3 (control group / no patch augmentation) included 3 patients. Demographics, baseline, and 4-week outcomes of patients are listed in Table 2. The patients were treated from March 2016 to March 2017. Surgical procedures were performed under general anesthesia with regional blockade (interscalene block). At the time of surgery, prior to supraspinatus tendon repair, a sample of the torn free edge of the tendon was harvested for histology and immunohistochemistry. Subjects underwent a single-row rotator cuff repair with Versalok (Depuy Mitek, Warsaw, IN, USA) and Healix BR (Depuy Synthes, Warsaw, IN, USA)

784

Acta Orthopaedica 2020; 91 (6): 782–788

Table 1. Primary antibodies used in immunohistochemistry (IHC), including details and rationale for use Primary antibody Source

Host

Clonality

Product code

Conc.

Rationale

CD4 Biorbyt Rabbit Polyclonal (IgG) Orb182470 1:500 Glycoprotein co-receptor on surface of CD4+ T-helper cells CD45 LSBio Mouse Monoclonal (IgG1) LS-C187484 1:250 Leucocyte common antigen. Pan-leucocyte marker CD68 Dako Agilent Mouse Monoclonal (IgG1) IR609 1:1000 Transmembrane glycoprotein in monocyte lineage cells, e.g., monocytic phagocytes CD206 Abcam Rabbit Polyclonal (IgG) Ab64693 1:2000 Mannose receptor. Cell surface marker on macro phages and immature dendritic cells BMP7 Abcam Mouse Monoclonal (IgG1) Ab54904 1:4000 BMP7 is part of TGFß superfamily. Counteracts TGFß1 in fibrosis, anti-fibrotic marker IRF5 Proteintech Rabbit Polyclonal 10547-1-AP 1:300 Interferon Regulatory Factor 5. A transcription factor expressed by pro-inflammatory macrophages TGFß Abcam Rabbit Monoclonal (IgG) Ab170874 1:150 Cytokine with pro-fibrotic effects PDPN Abcam Mouse Monoclonal (IgG1) Ab10288 1:200 Stromal cell activation marker

Table 2. The 13 patients enrolled in the study including demographic data, patient-reported outcome scores (Oxford Shoulder Score, OSS, Euroqol-5D, EQ-5D, and Visual Analogue Scale, VAS) at baseline Baseline Size of Age tear (cm) Patch

OSS (0–48)

52 5 GraftJacket 49 6 GraftJacket 63 5 GraftJacket 55 4 GraftJacket 67 4 Permacol 60 6 Permacol 34 3 Permacol 72 5 Control 68 3 Control 63 5 Control 55 a 4 Control 55 a 8 Irreparable tear 62 a PTT N/A

27 27 30 12 7 28 41 15 27 18 31 39 32

VAS (0–10)

EQ-5D (0–100)

6.9 65 5.2 80 5.7 95 5.0 85 10 45 4.7 96 1.3 70 8.1 75 7.0 80 7.3 90 4.5 75 2.5 90 3.0 15

PTT = partial thickness tear. excluded for reasons stated.

a Patients

suture anchors via a mini-open, deltoid-splitting approach. After the repair was completed, patients in Groups 1 and 2 had either a GraftJacket or Permacol patch applied in an onlay technique over the repair and secured with 3-0 PDS sutures as per the manufacturers’ instructions. Postoperative rehabilitation was guided by a physical therapist and included 4 weeks of sling immobilization, followed by 2 weeks of passive range-of-motion exercises, and 6 weeks of active mobilization. At 4 weeks, an ultrasound (USS) guided biopsy under local anesthesia (2 mL 2% lignocaine) of the repaired supraspinatus tendon was performed with a Bard Magnum core biopsy system and a 16g tissue biopsy needle (Bard Peripheral Vascular, Inc, Tempe, AZ, USA) (Figure 1). This technique has been previously validated (Murphy et al. 2013).

All samples were fixed in 10% formalin, processed with a Leica ASP300S tissue processor (Leica Biosystems Nussloch GmbH, Nußloch, Germany), and embedded in paraffin wax. Using a rotary RM-2135 microtome (Leica Microsystems Ltd, Milton Keynes, UK), sections of 5–6 µm were cut and mounted on glass slides (Leica Microsystems Ltd, Milton Keynes, UK). These were baked at 60° C for 30 minutes, and then at 37° C for 60 minutes. For histology, H&E staining was performed using a Tissue Tek DRS automated stainer (Sakura Finetech Europe, Leiden, Netherlands). For immunohistochemistry (IHC), sections underwent high pH, heat-mediated, epitope retrieval using a PT Links machine (Dako Agilent Pathology Solutions, Santa Clara, CA, USA). Sections were then stained using a Dako Autostainer Link 48 machine (Dako Agilent Pathology Solutions, Santa Clara, CA, USA). Antibody staining was performed with the EnVision FLEX visualization system and an Autostainer Link 48 machine (Dako Agilent Pathology Solutions, Santa Clara, CA, USA) (Table 1). Stained sections were imaged using a Zeiss AX10 inverted microscope with an Axiom HRc camera and Axiovision software (Zeiss, Cambridge, UK) at 40× and 100× magnification for H&E and IHC respectively. For histology sections, total cell count, foreign-body giant cell count (FBGC), and vascularity grading (0–3 scale) were performed on 6 random fields using ImageJ software (National Institutes of Health, Bethesda, MD, USA). This system has been previously validated (Rashid 2018). For IHC sections, all images were imported into CellProfiler software (Broad Institute, Cambridge, MA, USA) and a bespoke pipeline was applied to determine percentage of 3,3′-diaminobenzidine (DAB) staining per nuclei for each sample. For the purposes of analysis, histology and immunohistochemistry results were analyzed in groups as described earlier. Median change from preoperative to 4 weeks postoperative was used to compare across groups. Data were presented using GraphPad PRISM software (GraphPad software, Inc, La Jolla, CA, USA); however, numbers are too small to conduct meaningful statistical analysis.

Acta Orthopaedica 2020; 91 (6): 782–788

785

Change in cellularity / field

FGBC count / field

Vascularity grade (0–3)

400

300

200

100

–100

GraftJacket

Perma- Control col

augmentation) resembled a similar appearance to normal tendon. Permacol sections had more disruption of the ECM than GraftJacket sections on qualitative assessment. Results for total cell count, foreign-body giant cell count, and vascularity grading are presented in Figure 2. There was generally no significant difference between the groups; however, the tissue sections of one patient in Group 2 (Permacol patch) had a distinctly different histological appearance (Figure 2D). These sections showed markedly increased cellularity. Morphological features of cells suggest they are not tendon fibroblasts, as was seen in other patients’ tissue sections, but rather a dense infiltration of immune cells. Clinically, this patient complained of a painful arthrofibrosis 1-week post-surgery, which settled with analgesia from the general practitioner. The serum C-reactive protein level at 4 weeks was 10 mg/dL. Immunohistochemistry staining revealed no differences between the 3 groups for the primary antibodies tested (Figure 3). However, the patient with the abnormal reaction in Group 2 demonstrated significantly increased immunopositive staining for CD68+, CD206+, and IRF5+ cells (marked by * in Figures 3C/D/F).

GraftJacket

Perma- Control col

GraftJacket

Perma- Control col

Figure 2. Representative histology showing tissue response to patch augmentation compared with control (no patch) group. A–D: Typical 4-week biopsy sections stained with hematoxylin and eosin (H&E) for control (A), GraftJacket (B), and Permacol (C) patch augmentation, showing increasing disruption of the tendon extracellular matrix (ECM). D: Abnormal tissue response from patient receiving Permacol showing dense infiltration of immune cells. E–G: Histology results comparing 3 groups (GraftJacket, Permacol, and Control) for change in cellularity (E), foreign-body giant cell (FBGC) count (F), and vascularity grade (G). * symbol denotes the patient receiving Permacol with grossly different tissue response.

Ethics, funding, and potential conflicts of interest This study was granted ethical approval by the National Healthcare Service (NHS) Research and Ethics Committee (REC) Ref: 15/SC/0697. This study was supported by a National Institute for Health Research (NIHR) Biomedical Research Unit (BRU) infrastructure grant. The authors declare no financial disclosures.

Results Histology using H&E staining demonstrated significant disruption, on qualitative review, of the extracellular matrix (ECM) in the patch augmentation groups compared with the control group (Figure 2A–C). Specifically, sections demonstrated reduced crimp pattern, increased friability of the matrix, and lack of parallel oriented collagen fibers. Sections from the control group (conventional repair without patch

Discussion Patch augmentation is occasionally used to improve the healing rate in rotator cuff repair surgery. There are currently no prospective clinical studies that utilize post-implantation biopsies to investigate the human tissue response to these materials. This study is the first to characterize the early in vivo tissue response in humans undergoing rotator cuff repair with, and without, biological patch augmentation. We observed significant extracellular matrix disruption of the native supraspinatus tendon in response to both GraftJacket and Permacol patches compared with the control (no patch augmentation) group. The Permacol xenograft group demonstrated more ECM disruption of the native underlying supraspinatus tendon than the GraftJacket allograft group. At the early (4-week) time point, there was generally no increase in foreign-body giant cells or vascularity in both patch groups compared with the control (no patch) group. 1 patient who received the Permacol patch demonstrated significantly increased cellularity on rotator cuff tissue biopsy. This patient experienced a painful arthrofibrosis postoperatively. The tendon tissue sections from this patient were densely infiltrated with immune cells on histological evaluation. Immunohistochemistry did not reveal any significant differences between the groups; however, the sections from the aforementioned patient in group 2 showed a significant increase in CD68+/CD206+/IRF5+ cells, suggestive of an adverse immune response. Another patient, in the control group, developed a deep infection following uncomplicated rotator cuff repair, further highlighting the potential, albeit rare, risks of surgery.

786

Acta Orthopaedica 2020; 91 (6): 782–788

by 6 weeks on histology and chronic inflammation was noted. 2 0.020 0.0010 0.0010 0.004 * of 10 failed repairs demonstrated 0.015 increased inflammatory infiltrate 0.010 0.0005 0.0005 0.002 that the authors concluded may 0.005 0.0000 0.0000 0.000 represent rejection of the human 0.000 dermal matrix graft (Adams et al. –0.005 –0.0000 –0.002 –0.0005 GraftPerma- Control GraftPerma- Control GraftPerma- Control GraftPerma- Control 2006). There is a paucity of cliniA Jacket col B Jacket col C Jacket col D Jacket col cal studies that have evaluated Change in BMP7 (%/nucleus) Change in IRF5 (%/nucleus) Change in TGFß (%/nucleus) Change in PDPN (%/nucleus) 0.004 0.008 0.003 0.004 in vivo human tissue response to * 0.002 0.006 0.002 biological patches. Histological 0.003 0.001 samples from an individual case 0.004 0.000 0.002 0.000 report revealed extensive infiltra0.002 –0.002 –0.001 0.000 tion of noninflammatory host cells 0.000 –0.004 –0.002 and blood vessels (Snyder et al. –0.002 –0.006 –0.001 –0.003 GraftPerma- Control GraftPerma- Control GraftPerma- Control GraftPerma- Control 2009). A case series of 4 patients E Jacket col F Jacket col G Jacket col H Jacket col undergoing bridging repair of Figure 3. Immunohistochemistry (IHC) results showing change in immunopositive (DAB) staining (difference between preoperative and 4 week postoperative staining) per nucleus in 3 groups (GraftJacket, Per- large rotator cuff tears with Permacol, and Control). A–G: Comparing change immunopositive staining per nuclei in 3 experimental groups macol showed very poor results for anti-CD4 (A), anti-CD45 (B), anti-CD68 (C), anti-CD206 (D), anti-BMP7 (E), anti-IRF5 (F), anti-TGFβ (Soler et al. 2007). All 4 patients (G), and anti-PDPN (H). * symbol denotes patient receiving a Permacol patch, with a grossly different tissue failed to improve. Fluid in the response, showing significantly higher immunopositive staining against CD68, CD206, and IRF5. subdeltoid bursa was seen in all The philosophy behind biological patch augmentation is patients on MRI (1 aspirated to confirm sterile effusion). 2 that the native tendon tissue often lacks a capacity to heal of 4 patients went on to have a reverse total shoulder replaceto the bony footprint and that application of a collagen and ment. At the time of surgery, it was noted that histology demelastin-rich decellularized graft, which can become integrated onstrated chronic inflammation with necrotic fibrous material with native cells and blood vessels, would improve the healing (Soler et al. 2007). rate (Zimmer 2006, Group 2017). Thus, the aim is to improve Despite over 1 million implantations of GraftJacket (in a healing rate by improving the biological environment. wide range of surgical applications) (Group 2017), and over Several groups have investigated the response of tendon 100,000 implantations for Permacol (mainly in genitouricells on GraftJacket and Permacol patches in the labora- nary and hernia repair surgery) (Zimmer 2006), there are no tory (Derwin et al. 2006, Fini et al. 2007, Smith et al. 2016). high-quality, low risk of bias, clinical studies evaluating their Human tendon-derived cells cultured on various synthetic and clinical efficacy. There are several clinical studies investigatbiological patches exhibited different morphology, indicating the application of GraftJacket or Permacol in rotator cuff ing that physical cues may influence cell characteristics and repair augmentation, all with significant bias. Most studies are protein expression (Smith et al. 2016). In particular, synthetic observational, all of which demonstrated some improvement patches demonstrated healthy tenocyte morphology with in patient-reported outcome measures (Burkhead et al. 2007, extended lamellipodia, and increased collagen I:collagen III Bond et al. 2008, Wong et al. 2010, Gupta et al. 2012, Kokratio. Cells cultured on biological patches such as GraftJacket kalis et al. 2014). Only 1 study included a comparator group, and Permacol demonstrated atypical cell morphology (Smith reported as a level II randomized controlled trial (Barber et et al. 2016). al. 2012). This quasi-randomized controlled trial of patients Residual microDNA fragments have been observed in undergoing rotator cuff repair for large posterosuperior rotacommercially available biological patches, raising concerns tor cuff tears included 42 patients, randomized to either conregarding a potential immune response (Derwin et al. 2006). ventional arthroscopic repair (n = 20), or arthroscopic repair Biological patches are derived from harvested cadaveric or plus augmentation of GraftJacket as an onlay (n = 22). They animal tissue and processed under proprietary processes for observed no adverse events related to the patches, although one implantation. Despite this, no 2 patches are the same, with patient developed bursitis postoperatively. This study suffers variation in protein content and residual DNA content being from several limitations and there are serious concerns for risk observed. In our study, we did not biopsy each patch prior to of bias in trial design, conduct, and reporting, as assessed by implantation. Hence, we cannot comment on the cause of the the Cochrane risk of bias tool domains (Higgins et al. 2011). sterile inflammatory reaction seen in the patient who received Our study has limitations that must be considered in light the Permacol patch. of the results. First, the numbers are small and our findings In an infraspinatus repair canine model augmented with a of no difference in the immunohistochemistry are, therefore, human dermal graft, observed infiltration of the grafts occurred not conclusive. Additionally, the small numbers in each group Change in CD4 (%/nucleus)

Change in CD45 (%/nucleus)

Change in CD68 (%/nucleus)

Change in CD206 (%/nucleus)

0.0015

0.006

0.025

Acta Orthopaedica 2020; 91 (6): 782–788

precluded robust statistical analysis. This study does demonstrate that postoperative biopsy is possible and well tolerated by patients. Second, with any research involving core biopsies, despite best efforts to standardize there is some variance with the location of each biopsy. All ultrasound-guided biopsies were performed by an experienced shoulder ultrasonographer (AJC). We were limited only in commenting on the native tendon tissue response because not all biopsies included a sample of the patch. Given that the time point for these biopsies is early (4 weeks), we would not expect to see much cellular infiltration or neovascularization within the patch tissue at this time. Third, the tissue response was assessed at only 1 time point. We chose to focus on the early response in the belief that the early phases of healing are most important, and any acute inflammatory reaction would be observed at this time. We cannot comment on changes that may occur at earlier or later time points. Fourth, there are currently no specific and sensitive cell markers for identifying tendon fibroblasts exclusively. Most commonly used cell markers also stain other immune cell types. Hence, we chose to use total cellularity in the histology sections to identify differences between the groups. Finally, this study was not designed to demonstrate clinical efficacy of one patch over another. For this, a well-designed randomized controlled trial would answer the question of whether patch augmentation is superior to a control group (e.g., no patch augmentation). Both patches are licensed and used widely for rotator cuff repair augmentation, despite initially being developed for other, non-musculoskeletal applications. The human rotator cuff tendon enthesis is complex, and if patch augments are to make significant contributions towards improved healing, a more tailored approach of scaffolds specifically designed for this purpose may be beneficial. Conclusions This is the first study to systematically examine the native tissue response to commercially available biological patches applied in rotator cuff augmentation in humans. Significant disruption of the native supraspinatus tendon ECM was observed in the GraftJacket and Permacol patch augmented groups, compared with the control (no patch) group. 1 patient in the Permacol group had an adverse tissue reaction characterized by extensive infiltration of pro-inflammatory CD68+/ CD206+/IRF5+ cells. These finding raise concerns regarding the use of these patches in rotator cuff augmentation on the basis of early tissue response in vivo.

MSR, RS, SS, SGD, and AJC were involved in study design. KW and BW were involved in ethical approval application, patient recruitment, and coordination of tissue samples. AJC conducted all surgeries and ultrasound-guiding sample biopsies. MSR and NN conducted the histological and immunohistochemical experiments. SS, SGD, and AJC supervised the study. All authors were involved in the preparation and proof-reading of the manuscript.

787

Acta thanks Hanna Cecilia Björnsson Hallgren and Hans Rahme for help with peer review of this study.

Adams J, Zobitz M, Reach J, An K, Steinmann S. Rotator cuff repair using an acellular dermal matrix graft: an in vivo study in a canine model. Arthroscopy 2006; 22(7): 700-9. doi: 10.1016/j.arthro.2006.03.016. Barber F A, Burns J P, Deutsch A, Labbe M R, Litchfield R B. A prospective, randomized evaluation of acellular human dermal matrix augmentation for arthroscopic rotator cuff repair. Arthroscopy 2012; 28(1): 8-15. doi: 10.1016/j.arthro.2011.06.038. Bond J L, Dopirak R M, Higgins J, Burns J, Snyder S J. Arthroscopic replacement of massive, irreparable rotator cuff tears using a GraftJacket allograft: technique and preliminary results. Arthroscopy 2008; 24(4): 403-9 e1. doi: 10.1016/j.arthro.2007.07.033. Burkhead W Z, Schiffern S C, Krishnan S G. Use of Graft Jacket as an augmentation for massive rotator cuff tears. Seminars in Arthroplasty 2007; 18(1): 11-18. doi: 10.1053/j.sart.2006.11.017. Carr A, Cooper C, Campbell M, Rees J, Moser J, Beard D, Fitzpatrick R, Gray A, Dawson J, Murphy J, Bruhn H, Cooper D, Ramsay C. Effectiveness of open and arthroscopic rotator cuff repair (UKUFF): a randomised controlled trial. Bone Joint J 2017; 99-B(1): 107-15. doi: 10.1302/0301620X.99B1. Colvin A, Egorova N, Harrison A, Moskowitz A, Flatow E. National trends in rotator cuff repair. J Bone Joint Surg Am 2012; 94(3): 227-33. doi: 10.2106/JBJS.J.00739. Derwin K A B A, Spragg R K, Leigh D R, Iannotti J P. Commercial extracellular matrix scaffolds for rotator cuff tendon repair. J Bone Joint Surg 2006; 88A(12): 2665-72. Digital N. Hospital admitted patient care activity, 2015–2016 [NS]. In: Hospital Episodes Statistics (HES) Online. http://content.digital.nhs.uk/ article/2021/Website-Search?productid = 23488&q = title%3a%22Hosp ital+Episode+Statistics%2c+Admitted+patient+care+-+England%22+o r+title%3a%22Hospital+Admitted+Patient+Care+Activity%22&sort = Relevance&size = 10&page = 1&area = both#top: NHS Digital; 2016. Fini M, Torricelli P, Giavaresi G, Rotini R, Castagna A, Giardino R. In vitro study comparing two collageneous membranes in view of their clinical application for rotator cuff tendon regeneration. J Orthop Res 2007; 25(1): 98-107. doi: 10.1002/jor.20295. Group. WM. GRAFTJACKET™ Regenerative Tissue Matrix—“How is it processed?”. http://www.wright.com/healthcare-professionals/graftjacket; 2017. Gupta A K, Hug K, Berkoff D J, Boggess B R, Gavigan M, Malley P C, Toth A P. Dermal tissue allograft for the repair of massive irreparable rotator cuff tears. Am J Sports Med 2012; 40(1): 141-7. doi: 10.1177/0363546511422795. Higgins J P, Altman D G, Gotzsche P C, Juni P, Moher D, Oxman A D, Savovic J, Schulz K F, Weeks L, Sterne J A, Cochrane Bias Methods G, Cochrane Statistical Methods G. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011; 343: d5928. doi: 10.1136/bmj.d5928. Hinsenkamp M, Muylle L, Eastlund T, Fehily D, Noel L, Strong D M. Adverse reactions and events related to musculoskeletal allografts: reviewed by the World Health Organisation Project NOTIFY. Int Orthop 2012; 36(3): 63341. doi: 10.1007/s00264-011-1391-7. Kokkalis Z T, Mavrogenis A F, Scarlat M, Christodoulou M, Vottis C, Papagelopoulos P J, Sotereanos D G. Human dermal allograft for massive rotator cuff tears. Orthopedics 2014; 37(12): e1108-16. doi: 10.3928/0147744720141124-59. McQuillan D, Harper J. Extracellular wound matrices: a novel regenerative tissue matrix (RTM) technology for connective tissue reconstruction. Wounds 2007; 19(6): 163-8. doi: Retreived from https://www.woundsresearch.com/. Murphy R, Dean B, Wheway K, Watkins B, Morrey M, Carr A. A novel minimally invasive ultrasound-guided technique to biopsy supraspinatus tendon. Operative Techniques in Orthopaedics 2013; 23(2): 56-62. doi: 10.1053/j.oto.2013.05.003.

788

O’Brien J A, Ignotz R, Montilla R, Broderick G B, Christakis A, Dunn R M. Long-term histologic and mechanical results of a Permacol abdominal wall explant. Hernia 2011; 15(2): 211-15. doi: 10.1007/s10029-010-0628-5. Rashid M. Augmentation strategies in rotator cuff repair. In: Nuffield Department of Orthopaedics, Rheumatology, and Musculoskeletal Sciences (NDORMS). Oxford: University of Oxford; 2018. Russell R D, Knight J R, Mulligan E, Khazzam M S. Structural integrity after rotator cuff repair does not correlate with patient function and pain: a meta-analysis. J Bone Joint Surg Am 2014; 96(4): 265-71. doi: 10.2106/ JBJS.M.00265. Shen C, Tang ZH, Hu JZ, Zou GY, Xiao RC. Incidence of retear with double-row versus single-row rotator cuff repair. Orthopedics 2014; 37(11): e1006-13. doi: 10.3928/01477447-20141023-58. Smith R D J, Carr A, Dakin S G, Snelling S J B, Yapp C, Hakimi O. The response of tenocytes to commercial scaffolds used for rotator cuff repair. European Cells and Materials 2016; 31:107-18. doi: 10.22203/eCM. v031a08. Snyder S J, Arnoczky S P, Bond J L, Dopirak R. Histologic evaluation of a biopsy specimen obtained 3 months after rotator cuff augmentation with GraftJacket Matrix. Arthroscopy 2009; 25(3): 329-33. doi: 10.1016/j. arthro.2008.05.023. Soler J A, Gidwani S, Curtis M J. Early complications from the use of porcine dermal collagen implants (Permacol) as bridging constructs in the repair of massive rotator cuff tears: a report of 4 cases. Acta Orthop Belg 2007;

Acta Orthopaedica 2020; 91 (6): 782–788

73(4): 432-6. doi: retrieved from http://www.actaorthopaedica.be/. Tempelhof S, Rupp S, Seil R. Age-related prevalence of rotator cuff tears in asymptomatic shoulders. J Shoulder Elbow Surg 1999; 8(4): 296-9. doi: https://doi.org/10.1016/S1058-2746(99)90148-9. Wong I, Burns J, Snyder S. Arthroscopic GraftJacket repair of rotator cuff tears. J Shoulder Elbow Surg 2010; 19(2 Suppl.): 104-9. doi: 10.1016/j. jse.2009.12.017. Xu H W H, Zuo W, Sun W, Owens R T, Harper J R, Ayares D L, McQuillan D J. A porcine-derived acellular dermal scaffold that supports soft tissue regeneration: removal of terminal galactose-a-(1,3)-galactose and retention of matrix structure. Tissue Eng Part A 2009; 15(7): 1807-19. doi: 10.1089/ ten.tea.2008.0384. Xu H, Sandor M, Qi S, Lombardi J, Connor J, McQuillan D J, Iannotti J P. Implantation of a porcine acellular dermal graft in a primate model of rotator cuff repair. J Shoulder Elbow Surg 2012; 21(5): 580-8. doi: 10.1016/j. jse.2011.03.014. Yang J Jr, Robbins M, Reilly J, Maerz T, Anderson K. The clinical effect of a rotator cuff retear: a meta-analysis of arthroscopic single-row and double-row repairs. Am J Sports Med 2017; 45(3): 733-41. doi: 10.1177/0363546516652900. Zimmer. Zimmer® Collagen Repair Patch. http://www.zimmer.co.uk/ content/dam/zimmer-web/documents/en-GB/pdf/medical-professionals/shoulder/Zimmer-Collagen-Repair-Patch-Brochure97-4100-002-00-Rev-1-12-2006.pdf; 2006.

Acta Orthopaedica 2020; 91 (6): 789–793

789

No need to use both Disabilities of the Arm, Shoulder and Hand and Constant–Murley score in studies of midshaft clavicular fractures Andreas H QVIST 1, Michael T VÆSEL 2, Carsten MOSS 3, Thomas JAKOBSEN 4, and Steen L JENSEN 4 1 Department

of Orthopedics, Aarhus University Hospital; 2 Department of Orthopedics, Viborg Regional Hospital; 3 Department of Orthopedics, Randers Regional Hospital; 4 Department of Orthopedics, Alborg University Hospital, Denmark Correspondence: Aqvistchristensen@gmail.com Submitted 2020-03-04. Accepted 2020-08-11.

Background and purpose — Most newer randomized studies examining plate fixation and nonoperative treatment of midshaft clavicular fractures utilize both Disabilities of the Arm, Shoulder and Hand (DASH) and Constant–Murley score (CS) in the evaluation of patient outcomes. Compared with DASH, the use of CS requires on-site trained personnel and patient visits to obtain the score. The use of both DASH and CS should provide extra value compared with the use of a single functional outcome score; if this value is not provided, the combined use is not necessary. We evaluated the agreement between DASH and CS in patients with displaced midshaft clavicular fractures. Patients and methods — We used prospectively collected data from 146 patients enrolled in a randomized study comparing operative and nonoperative treatment of midshaft clavicular fractures. We determined correlation between DASH and CS at all follow-up points and calculated mean bias in the Bland–Altman plot. Results — We found moderate to high correlation (from 0.82 at 6 weeks’ follow-up to 0.58 at 1-year follow-up) between DASH and CS score, and a small bias (2.21 [95% CI 0.22–4.20]) in the Bland–Altman plot. Interpretation — In patients with displaced midshaft clavicular fractures DASH and CS measures the same degree of disability. Unless specifically studying strength and range of motion, we recommend the sole use of DASH as it would eliminate potential observer-induced bias along with removing the economic and logistic burden of obtaining CS without compromising the value of the collected data.

Almost all newer randomized studies (Canadian Orthopaedic Trauma Society 2007, Mirzatolooei 2011, Virtanen et al. 2012, Robinson et al. 2013, Ahrens et al. 2017, Woltz et al. 2017b, Qvist et al. 2018) examining plate fixation with nonoperative treatment of midshaft clavicular fractures utilize both Disabilities of the Arm, Shoulder and Hand (DASH) (Hudak et al. 1996) and Constant–Murley score (CS) (Constant and Murley 1987) in the evaluation of patient outcomes. DASH is a self-reported questionnaire, developed in 1996 to describe disability experienced by patients with a musculoskeletal condition of the upper extremity and to monitor change in symptoms and upper limb function over time. Developed in 1987, the CS evaluates shoulder function in general by combining subjective and objective measurements. The use of CS requires on-site trained personnel and ambulatory patient visits to obtain the score and is more time consuming than obtaining DASH (Michener and Leggin 2001). The correlation between and measurement properties of DASH and CS have been examined in patients following nonoperatively treated clavicular fractures (Ban et al. 2016), rotator cuff repair (Skutek et al. 2000), and in patients with a humeral shaft fracture (Mahabier et al. 2017). These studies show a good correlation between DASH and CS, which indicates that either score alone may replace the use of both. In a research setting with finite resources the use of both DASH and CS should provide extra value compared with the use of a single functional outcome score. If this extra value is not provided, the combined use of DASH and CS is not necessary. Should the combined use of these instruments not be necessary DASH has the potential to be used as the sole instrument in register studies with large cohorts. This study evaluates the agreement between DASH and CS in patients receiving both nonoperative and operative treatment of displaced midshaft clavicular fractures.

790

Acta Orthopaedica 2020; 91 (6): 789–793

Table 1. Demographic data 146 patients with midshaft clavicular fractures

Score

Percent of patients with maximum score

100

iDASH

Parameter Demographic Age (range) Age groups, n 18–30 31–45 46–60 Male:female ratio, n Fracture type, n noncomminuted comminuted fracture missing data Shortening, n < 1 cm 1–2 cm > 2 cm missing data

40 (18–60) 38 59 49 119:27 46 94 6

20 iDASH CS 0

38 64 37 7

1.5

Months after fracture

Figure 1. Boxplot of inverted Disabilities of the Arm, Shoulder and Hand (iDASH) and Constant–Murley score (CS) at each follow-up. Red line is the median. Top and bottom box borders indicate the interquartile range (IQR). Whiskers mark minimum and maximum value no further than 1.5 × IQR from the hinge. Outliers beyond that are plotted individually.

Patients and methods We used prospectively collected data from 146 patients enrolled in a randomized study comparing operative and nonoperative treatment of midshaft clavicular fractures (Qvist et al. 2018). DASH and CS were collected at 6 weeks, 3 months, 6 months, and 1-year follow-up. 146 patients were included in the study (Table 1). 71 patients were included in the nonoperative treatment group and 75 were included in the operative treatment group. 22 patients were lost to follow-up. The DASH questionnaire consists of 30 items, each scored 1 to 5 on an ordinal scale (Hudak et al. 1996). The DASH score ranges from 0 to 100 points with 0 points representing normal function and increasing score representing increasing dysfunction. For analysis purposes only an inverted DASH (iDASH equal to 100-DASH) score was used. CS evaluates shoulder function in general by combining subjective and objective measurements (Constant and Murley 1987). An observer measures range of motion (ROM) and power for a total of 35 points. In the original randomized trial this observer was blinded. The subjective measurement of CS is 2 patient-reported items for pain and activities of daily life (ADL) for a total of 65 points. CS ranges from 0 to 100 points with decreasing score representing dysfunction. A CS of 100 points equals a shoulder with full range of motion, no pain, no problems with performing activities of daily life, and an abduction force of 12 kg. Statistics We used R (R Core Team 2018) version 1.1.456 and the blandr package (Datta 2017). We did not impute missing

1.5

Months after fracture

Figure 2. Ceiling effect for inverted Disabilities of the Arm, Shoulder and Hand (iDASH) and Constant–Murley score (CS) at each follow-up.

data. We used QQ plots and histograms to examine data for normality. Patient demographics were reported using descriptive statistics. Ceiling effect was calculated at each follow-up. Spearman’s rank correlation coefficient (r) was calculated to describe correlation between iDASH and CS, as iDASH and Constant scores were not normally distributed at any follow-up point. Correlation was regarded as high if r was over 0.70, moderate if r was between 0.50 and 0.70, and low if r was lower than 0.50 (Mukaka 2012). We expected correlation to decrease with increasing follow-up, as some patients would be expected to reach the celling in iDASH before CS and vice versa. Spearman’s rank correlation coefficient for operative and nonoperative groups was calculated to ensure that combining the 2 groups did not overestimate correlation (Hassler and Thadewald 2003). A Bland–Altman plot was used to describe the mean bias between iDASH and CS. The Bland–Altman plot requires that the difference between the 2 scores is normally distributed, which occurred only at 6 weeks’ follow-up. 95% confidence intervals (CI) were calculated. Ethics, registration, funding, and potential conflicts of interest The original randomized study was approved by the Regional Ethical Committee Board in the North Denmark Region (N-20090054), and registered in the ClinicalTrials.gov database (Identifier: NCT01078480). This study received funding from Swemac Orthopaedics Aps. None of the authors received payments or services, either directly or indirectly in support of any aspect of this work.

Acta Orthopaedica 2020; 91 (6): 789–793

791

Figure 3. Scatterplots of inverted Disabilities of the Arm, Shoulder and Hand (iDASH) and Constant-Murley score (CS) at (A) 6 weeks’ follow-up, (B) 3 months’ follow-up, (C) 6 months’ follow-up, and (D) 1-year follow-up. r = Spearman’s rank correlations coefficient. n = number of subjects.

coefficient from 0.82 at 6 weeks’ follow-up to 0.58 at 1-year follow-up. Correlation between iDASH and CS at 6 weeks, 3 months, and 6 months’ follow-up was high, while correlation at 1-year follow-up was moderate. We found no severe overestimation of correlation coefficients when comparing groups (Table 2). The Bland–Altman plot (Figure 4) showed a mean bias towards iDASH of 2.2 (CI 0.2–4.2) with a upper limit of agreement of 25 (CI 21–28) and a lower limit of –20 (CI –17 to –24).

Figure 4. Bland–Altman plot of the means of inverted Disabilities of the Arm, Shoulder and Hand (iDASH) and Constant–Murley score (CS) versus the differences between iDASH and CS. Top dashed line indicates upper limits of agreement, while lower dashed line indicates lower limits of agreement. Middle dashed line indicates mean bias. Dotted lines show 95% confidence intervals around agreements and mean bias.

Table 2. Number of subjects (n) and correlation (r) between DASH and CS at each follow-up point Treatment

6 weeks n r

Operative 67 0.77 Nonoperative 63 0.78 Combined 130 0.82

3 months n r

6 months n r

1 year n r

65 0.64 65 0.72 130 0.71

63 0.71 63 0.69 126 0.71

63 0.65 60 0.52 123 0.58

Results Both iDASH and CS improved at each follow-up (Figure 1). Ceiling effect increased during follow-up and was present in 44% (CI 35–52) of all iDASH scores and in 49% (CI 41–58) of all CS after 1 year (Figure 2). Figure 3 shows scatterplots and Spearman’s rank correlations coefficients of iDASH and CS at each follow-up point. We saw a decrease in correlation

Discussion We found moderate to high correlation between iDASH and CS score, and a small mean bias in the Bland–Altman plot. The correlation between iDASH and CS decreased with increasing follow-up and was moderate at 1-year follow-up. Scatterplots (Figure 1) of iDASH and CS show the impact of ceiling effect with increasing follow-up, which could explain the decrease in rank correlation, as some patients reach the celling in iDASH before CS and vice versa. The decrease in correlation may also be related to the loss of patients to follow-up, as the power of the correlation test decreases with the lower number of subjects late in the study. The combination of operative and nonoperative treatment groups increases sample heterogeneity and could overestimate the correlation coefficient (Hassler and Thadewald 2003); however, we found no severe overestimation of correlation coefficients when analyzing groups separately. The high correlation between iDASH and CS before the 1-year follow-up point is consistent with findings in previous studies. Ban et al. (2016) found a Pearson correlation coefficient between DASH and CS of –0.92 after 6 weeks of nonoperative treatment in 36 patients with a wide range of clavicular fracture types. Mahabier et al. (2017) found a Spearman rank correlation of -0.78 between DASH and CS at 6 months’ follow-up in a large cohort of patients from a randomized controlled trial comparing operative and nonoperative treatment of humeral shaft fractures. In a large consecutive series of 372 patients

792

with rotator cuff disorders, a Lin’s concordance correlation coefficient of 0.89 was found between DASH and CS at 24 months follow-up (Allom et al. 2009). None of the above studies investigated correlation when ceiling effect was present, and our finding of moderate correlation at 1-year follow-up is different from these previous findings. Correlation coefficient alone does not provide enough information on agreement between methods of measurement (Bland and Altman 1995). To further compare the agreement between iDASH and CS, we constructed a Bland–Altman plot. We found a minimal mean bias between iDASH and CS of 2.2 (CI 0.2–4.2), meaning that on average iDASH would measure 2.2 points more than CS on a group level. We consider the mean bias to be low, as it is lower than the 10-point clinically relevant difference in DASH. 95% of the differences between the iDASH and CS measurements lie between upper (25 points) and lower limits of agreement (-20 points), which is up to 2.5 times more than the clinically relevant difference in DASH. This may seem high, but the point of our Bland–Altman plot was not to investigate a possible direct translation from CS to DASH on an individual level, but to illustrate the mean bias between DASH and CS on a multisubject scale when comparing groups in clinical studies. We do not consider the wide limits of agreement to be an issue in this regard. Supporting our finding of a low mean bias a recent meta-analysis found a similar absolute mean difference of 5.1 (CI 0.1–10.1) vs. 4.4 (CI 0.9–7.9) in DASH and CS score comparing operative and non-operative treatment of midshaft clavicular fractures at 1-year follow-up (Woltz et al. 2017a). The overall high correlation and low mean bias shows similarity between the 2 scores and we believe that DASH and CS measures the same degree of disability. Support for this claim also comes from recent randomized trials (Canadian Orthopaedic Trauma Society 2007, Mirzatolooei 2011, Virtanen et al. 2012, Robinson et al. 2013, Ahrens et al. 2017, Woltz et al. 2017b, Qvist et al. 2018), where in all cases DASH and CS follows the same trends when comparing operative and nonoperative treatment. DASH is a limb-specific instrument developed with the purpose of assessing symptoms and functional status in populations with upper extremity musculoskeletal conditions (Hudak et al. 1996) and CS was developed as a clinical method of shoulder function assessment (Constant and Murley 1987). The comparison of a limb-specific instrument with a shoulderspecific instrument may not be valid if the patient suffers from any other illness of the affected extremity. In our original randomized study these patients were excluded, and in this patient group of otherwise healthy patients we expect that any change in DASH would be related to disability following a midshaft clavicular fracture, making DASH comparable to CS. The obtainment of CS requires on-site trained personnel and ambulatory patient visits, which can pose a logistic and economic challenge. This challenge has previously been recognized as a component in the difficulties of assessing long-

Acta Orthopaedica 2020; 91 (6): 789–793

term outcome of shoulder disability (Dawson et al. 2001). The sole use of DASH does not require on-site trained personnel or ambulatory patient visits, as the DASH questionnaire could be completed at home and mailed in or submitted online via free tools such as REDCap (Harris et al. 2009). The sole use of DASH in the future evaluation of cuff disorders has been proposed in a study comparing DASH, CS, and Oxford score in patients with rotator cuff disorders (Allom et al. 2009). Furthermore, compared with CS, the DASH questionnaire was found to be the most reliable instrument in evaluating outcome in humeral shaft fractures (Mahabier et al. 2017). QuickDASH exists as a shortened version of the DASH questionnaire (Gummesson et al. 2006). QuickDASH contains only 11 questions and could potentially increase response rates compared with DASH. Although some detail is lost with the reduction of questions (Angst et al. 2009), QuickDASH has been shown to be a relevant substitute for DASH (Abramo et al. 2008, Macdermid et al. 2015, Tsang et al. 2017), and could potentially be used in future studies instead of DASH. This study has some limitations. We did not perform a test–retest analysis, as no retest was performed. Lacking this analysis, we were not able to evaluate the test–retest reliability of CS and DASH. However, the scope of this study was not to validate CS and DASH in patients with midshaft clavicular fractures and we do not consider the lack of test–retest analysis to be a severe limitation. Ideally, the Bland–Altman plot should compare DASH or CS against a gold standard, but no such standard exists in the measurement of outcomes following midshaft clavicular fractures. Considering external validity our study population contains only patients with a displaced midshaft clavicular fracture and the results cannot be generalized to all types of clavicular fractures. In conclusion, as DASH and CS in this study measure the same degree of disability in patients with midshaft clavicular fracture, we propose that DASH may be used as the only measurement instrument in future studies comparing outcomes following treatment of midshaft clavicular fractures, unless these studies have a specific aim of studying strength and range of motion, where CS could be used as the only instrument.

CMJ and SLJ designed the study. AHQ, MTV, CMJ, and SLJ acquired the data. AHQ, TJ, and SLJ did the statistical analysis. All authors discussed the results and contributed to the final manuscript. The authors would like to acknowledge Kirsten Duch and Søren LundbyeChristensen for their contribution to the statistical analysis. Acta thanks Laurent Audige and Anders Troelsen for help with peer review of this study.

Abramo A, Kopylov P, Tägil M. Evaluation of a treatment protocol in distal radius fractures. Acta Orthop 2008; 79(3): 376-85.

Acta Orthopaedica 2020; 91 (6): 789–793

Ahrens P M, Garlick N I, Barber J, Tims E M. The Clavicle Trial: a multicenter randomized controlled trial comparing operative with nonoperative treatment of displaced midshaft clavicle fractures. J Bone Joint Surg Am 2017; 99(16): 1345-54. Allom R, Colegate-Stone T, Gee M, Ismail M, Sinha J. Outcome analysis of surgery for disorders of the rotator cuff: a comparison of subjective and objective scoring tools. J Bone Joint Surg Br 2009; 91(3): 367-73. Angst F, Goldhahn J, Drerup S, Flury M, Schwyzer H-K, Simmen B R. How sharp is the short QuickDASH? A refined content and validity analysis of the short form of the disabilities of the shoulder, arm and hand questionnaire in the strata of symptoms and function and specific joint conditions. Qual Life Res 2009; 18(8): 1043-51. Ban I, Troelsen A, Kristensen M T. High inter-rater reliability, agreement, and convergent validity of Constant score in patients with clavicle fractures. J Shoulder Elbow Surg 2016; 25(10): 1577-82. Bland J M, Altman D G. Comparing two methods of clinical measurement: a personal history. Int J Epidemiol 1995; 24(Suppl. 1): S7-14. Canadian Orthopaedic Trauma Society. Nonoperative treatment compared with plate fixation of displaced midshaft clavicular fractures: a multicenter, randomized clinical trial. J Bone Joint Surg Am 2007; 89(1): 1-10. Constant C R, Murley A H. A clinical method of functional assessment of the shoulder. Clin Orthop Relat Res 1987; (214): 160-4. Datta D. blandr: a Bland–Altman method comparison package for R. Zenodo. org; 2017. Dawson J, Hill G, Fitzpatrick R, Carr A. The benefits of using patient-based methods of assessment: medium-term results of an observational study of shoulder surgery. J Bone Joint Surg Br 2001; 83(6): 877-82. Gummesson C, Ward M M, Atroshi I. The shortened disabilities of the arm, shoulder and hand questionnaire (QuickDASH): validity and reliability based on responses within the full-length DASH. BMC Musculoskelet Disord 2006; 7: 44. Harris P A, Taylor R, Thielke R, Payne J, Gonzalez N, Conde J G. Research electronic data capture (REDCap): a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009; 42(2): 377-81. Hassler U, Thadewald T. Nonsensical and biased correlation due to pooling heterogeneous samples. J R Stat Soc Ser D 2003; 52(3): 367-79. Hudak P L, Amadio P C, Bombardier C. Development of an upper extremity outcome measure: the DASH (disabilities of the arm, shoulder and hand) [corrected]. The Upper Extremity Collaborative Group (UECG). Am J Ind Med 1996; 29(6): 602-8. Macdermid J C, Khadilkar L, Birmingham T B, Athwal G S. Validity of the QuickDASH in patients with shoulder-related disorders undergoing surgery. J Orthop Sports Phys Ther 2015; 45(1): 25-36.

793

Mahabier K C, Den Hartog D, Theyskens N, Verhofstad M H J, Van Lieshout E M M. Reliability, validity, responsiveness, and minimal important change of the Disabilities of the Arm, Shoulder and Hand and Constant–Murley scores in patients with a humeral shaft fracture. J Shoulder Elbow Surg 2017; 26(1): e1-12. Michener L A, Leggin B G. A review of self-report scales for the assessment of functional limitation and disability of the shoulder. J Hand Ther 2001; 14(2): 68-76. Mirzatolooei F. Comparison between operative and nonoperative treatment methods in the management of comminuted fractures of the clavicle. Acta Orthop Traumatol Turc 2011; 45(1): 34-40. Mukaka M M. Statistics corner: a guide to appropriate use of correlation coefficient in medical research. Malawi Med J 2012; 24(3): 69-71. Qvist A H, Vaesel M T, Jensen C M, Jensen S L. Plate fixation compared with nonoperative treatment of displaced midshaft clavicular fractures: a randomized clinical trial. Bone Joint J 2018; 100-B(10): 1385-91. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2018. Robinson C M, Goudie E B, Murray I R, Jenkins P J, Ahktar M A, Foster C J, Clark K, Brooksbank A J, Arthur A, Crowther M A, Packham I, Chesser T J, Read E O, Foster C J, Clark K, Brooksbank A J, Arthur A, Crowther M A, Packham I, Chesser T J. Open reduction and plate fixation versus nonoperative treatment for displaced midshaft clavicular fractures: a multicenter, randomized, controlled trial. J Bone Joint Surg Am 2013; 95(17): 1576-84. Skutek M, Fremerey R W, Zeichen J, Bosch U. Outcome analysis following open rotator cuff repair: early effectiveness validated using four different shoulder assessment scales. Arch Orthop Trauma Surg 2000; 120(7-8): 432-6. Tsang P, Walton D, Grewal R, MacDermid J. Validation of the QuickDASH and DASH in patients with distal radius fractures through agreement analysis. Arch Phys Med Rehabil 2017; 98(6): 1217-22.e1. Virtanen K J, Remes V, Pajarinen J, Savolainen V, Björkenheim J-M, Paavola M. Sling compared with plate osteosynthesis for treatment of displaced midshaft clavicular fractures: a randomized clinical trial. J Bone Joint Surg Am 2012; 94(17): 1546-53. Woltz S, Krijnen P, Schipper I B. Plate fixation versus nonoperative treatment for displaced midshaft clavicular fractures: a meta-analysis of randomized controlled trials. J Bone Joint Surg Am 2017a; 99(12): 1051-7. Woltz S, Stegeman S A, Krijnen P, van Dijkman B A, van Thiel T P H, Schep N W L, de Rijcke P A R, Frolke J P M, Schipper I B. Plate fixation compared with nonoperative treatment for displaced midshaft clavicular fractures: a multicenter randomized controlled trial. J Bone Joint Surg Am 2017b; 99(2): 106-12.

794

Acta Orthopaedica 2020; 91 (6): 794–800

Similar risk of complete revision for infection with single-dose versus multiple-dose antibiotic prophylaxis in primary arthroplasty of the hip and knee: results of an observational cohort study in the Dutch Arthroplasty Register in 242,179 patients Ewout S VELTMAN 1,2, Erik LENGUERRAND 3, Dirk Jan F MOOJEN 1, Michael R WHITEHOUSE 3,4, Rob G H H NELISSEN 2, Ashley W BLOM 3,4, and Rudolf W POOLMAN 1,2 1 Department

of Orthopaedic and Trauma Surgery, Joint Research, OLVG, Amsterdam, the Netherlands; 2 Department of Orthopaedics, Leiden University Medical Center, Leiden, the Netherlands; 3 Musculoskeletal Research Unit, Translational Health Sciences, Bristol Medical School, University of Bristol, Bristol, UK; 4 National Institute for Health Research Bristol Biomedical Research Centre, University Hospital Bristol NHS Foundation Trust and University of Bristol, UK Correspondence: e.s.veltman@olvg.nl Submitted 2020-02-09. Accepted 2020-06-22.

Background and purpose — The optimal type and duration of antibiotic prophylaxis for primary arthroplasty of the hip and knee are subject to debate. We compared the risk of complete revision (obtained by a 1- or 2-stage procedure) for periprosthetic joint infection (PJI) after primary total hip or knee arthroplasty between patients receiving a single dose of prophylactic antibiotics and patients receiving multiple doses of antibiotics for prevention of PJI. Patients and methods — A cohort of 130,712 primary total hip and 111,467 knee arthroplasties performed between 2011 and 2015 in the Netherlands was analyzed. We linked data from the Dutch arthroplasty register to a survey collected across all Dutch institutions on hospital-level antibiotic prophylaxis policy. We used restricted cubic spline Poisson models adjusted for hospital clustering to compare the risk of revision for infection according to type and duration of antibiotic prophylaxis received. Results — For total hip arthroplasties, the rates of revision for infection were 31/10,000 person-years (95% CI 28–35), 39 (25–59), and 23 (15–34) in the groups that received multiple doses of cefazolin, multiple doses of cefuroxime, and a single dose of cefazolin, respectively. The rates for knee arthroplasties were 27/10,000 person-years (95% CI 24–31), 40 (24–62), and 24 (16–36). Similar risk of complete revision for infection among antibiotic prophylaxis regimens was found when adjusting for confounders. Interpretation — In a large observational cohort we found no apparent association between the type or duration of antibiotic prophylaxis and the risk of complete revision for infection. This does question whether there is any advantage to the use of prolonged antibiotic prophylaxis beyond a single dose.

Annually around 1 million patients receive a total hip or total knee prosthesis in the United States and over 190,000 hip and knee replacements are performed in England and Wales (Maradit et al. 2015, National Joint Registry for England and Wales 2018). The incidences of prosthetic replacement of the hip and knee are expected to increase (Kurtz et al. 2014). Prosthetic joint infection (PJI) following total hip or knee arthroplasty and the treatment thereof are catastrophic for patients and pose tremendous costs to healthcare systems (Poultsides et al. 2010, Zmistowski et al. 2013, Moore et al. 2015). Perioperative antibiotic prophylaxis remains an effective method of reducing the risk of PJI (Illingworth et al. 2013, Thornley et al. 2015). The type and duration of antibiotic prophylaxis are subject to debate. Both single-dose and multiple-dose antibiotic prophylaxis regimens have been advocated with comparable results (Thornley et al. 2015, Tan et al. 2019). The recommendations provided by the Second International Consensus Meeting of the MusculoSkeletal Infection Society (MSIS) and the European Bone and Joint Infection Society (EBJIS) advise that antibiotic prophylaxis should be administered 30–60 minutes before incision and discontinued within 24 hours after surgery (Hansen et al. 2014, Parvizi and Gehrke 2018). Large variation in prophylaxis regimens has been observed in the United Kingdom (Hickson et al. 2015). The Dutch national orthopedic association advises administration of antibiotic prophylaxis using a first- or second-generation cephalosporin starting 30–60 minutes preoperatively and discontinuing the antibiotic prophylaxis within 24 hours (Swierstra et al. 2009, Nederlandse Orthopaedische Vereniging 2018). The World Health Organization and, in the USA, the Centers for Disease Control and Prevention (CDC) recommend against the

Acta Orthopaedica 2020; 91 (6): 794â&#x20AC;&#x201C;800

use of postoperative continuation of antibiotic prophylaxis and advocate for a single dose of antibiotics delivered preoperatively (Berrios-Torres et al. 2017). This recommendation is vehemently challenged by the American Association of Hip and Knee Surgeons and the International Consensus Meeting, which encourage their members to proceed with the current common practice of multiple-dose antibiotic prophylaxis protocols until more evidence is available (Yates 2018). We compared the risk of complete revision for infection in the 1st year following primary hip and knee arthroplasty according to the perioperatively administered antibiotic prophylaxis regimen by using data from the Dutch Arthroplasty Register (LROI).

Patients and methods This study was structured using the STROBE guideline. In this observational cohort study, we report analyses of data for the Netherlands from the Dutch Arthroplasty Register (LROI) between January 1, 2011 and December 31, 2015. We included in the study all patients who had a primary hip or knee replacement during this period. Patient consent was obtained for data collection and linkage by the LROI. Using data on patient level was not possible due to the legislation of the General Data Protection Regulation. In the absence of individual patient-level data on antibiotic prophylaxis, we performed a national audit of hospital perioperative antibiotic prophylaxis regimens in the Netherlands (Veltman et al. 2018). All 99 Dutch hospitals or clinics performing primary total hip arthroplasty (THA) or total knee arthroplasty (TKA) were contacted and all completed a survey to identify the existence of treatment protocols concerning primary joint replacement, the existence of protocols regarding treatment strategy in case of suspected early postoperative infection, and tendency to register procedures in the LROI database. We asked, in particular, about type and duration of antibiotic prophylaxis. This survey showed a variance in postoperative duration of antibiotic prophylaxis. 10 Dutch hospitals administered a single-shot antibiotic prophylaxis, while the remaining 89 administered a multiple-shot antibiotic prophylaxis. This variance facilitated an observational cohort study using the LROI. The LROI has a completeness of over 95% for primary hip and knee arthroplasties and of 91% and 92% for the hip and knee revision procedures respectively (Dutch Arthroplasty Register [LROI] 2014, 2017, van Steenbergen et al. 2015). The translated survey form can be found in Appendix 1, Supplementary data. Each patient who had a primary THA or TKA was followed up for a minimum of 12 months until the end of the observation period (December 31, 2015) or until the date of 1- or 2-stage revision for infection, revision for another indication, death or end of follow-up (January 1, 2018). Revisions for infection included only complete revision of the total system,

795

obtained by a 1- or 2-stage revision procedures. All partial revisions (e.g., debridement, antibiotics, and implant retention procedures [DAIR]) were excluded because these partial revisions are inconsistently recorded compared with total revisions (Dutch Arthroplasty Register [LROI] 2017, Veltman et al. 2018). We chose to end the follow-up period at 1 year after surgery as with longer follow-up the influence of hematogenous infections on the measured outcome may increase to become larger than the influence of the duration of antibiotic prophylaxis at primary surgery. We defined infection status using the surgical indication reported on the LROI revision arthroplasty form following surgery by the treating orthopedic surgeon. We included patients whom had undergone complete revision captured by the LROI where the reason for revision was defined as infection in the infected group and patients in whom the reason for revision was not reported, or reason for revision other than infection was reported, in the non-infected group. The diagnosis and treatment strategy for complete revision for infection was at the discretion of the surgeon and treating unit and it reflected contemporary practice over the study period, with raised inflammatory markers, joint-specific symptoms, sinuses, and positive microbiological cultures being common diagnostic features over that period (Parvizi et al. 2013). We compared the risk of complete revision surgery for infection in the 1st year following primary arthroplasty by the type and duration of antibiotic prophylaxis regimen administered at primary surgery. We considered the patient characteristics age, sex, BMI, ASA grade, and previous surgery. We considered surgical factors such as indication for surgery, surgical approach, type of fixation, and bearing surface. Data from the LROI database were combined at hospital level with the results of the national survey on antibiotic prophylaxis. Results of the survey show there were 3 types of antibiotic regimens that are used in the Netherlands: multiple doses of cefazolin (MCZ), multiple doses of cefuroxime (MCX), and single dose of cefazolin (SCZ), which are all in concordance with the Dutch guideline for perioperative antibiotics in total hip and knee arthroplasty (Veltman et al. 2018). No other antibiotic regimens were encountered in the survey. Patients were divided into 3 groups (MCZ, MCX, and SCZ) according to the antibiotic prophylaxis protocol of the hospital where they were treated. Statistics We investigated the association between hospital antibiotic prophylaxis regimen policies (MCZ used as the reference) and the risk of complete revision for infection in the first 12 months following the index primary surgery with Poisson regression to account for time at risk and to produce hazard ratios including 95% confidence intervals (CI). The baseline hazard rate was modelled with restricted cubic splines. The optimum numbers of knots (3 degrees of freedom [d.f.] for the hip models, 4 d.f. for the knee models) was identified with AIC

796

and BIC criteria (Appendix Table 1, Supplementary data). Interaction terms between the splines and the main exposure covariates were included to estimate the time-dependent hazard ratio for complete revision for infection of the different antibiotic prophylaxis regimens (Royston and Lambert 2011). Huberâ&#x20AC;&#x201C;White sandwich estimates of variance were computed to adjust for within-hospital correlation. The models were stratified by surgical site and adjusted for age, sex, BMI, and ASA classification. Multiple imputation by chained equations (5 imputations sets) under a missing at random framework was used to account for missing data. The imputation model incorporated the PJI status, time at risk, the main exposure, the aforementioned adjustment factors and indication for surgery, surgical approach, method of fixation, bearing surface, and year of surgery as ancillary variables. All statistical analyses were performed using Stata, version 15.1 (StataCorp, College Station, TX, USA). Ethics, registration, funding, and potential conflicts of interest This study was approved by the scientific committee of the LROI. The database was constructed by the LROI office. All data provided by the LROI were anonymized, no patient identifiable data were available to the researchers. The study protocol was registered on ClinicalTrials.gov (reference NCT03348254). This study was partially supported by the NIHR Biomedical Research Centre at University Hospitals Bristol NHS Foundation Trust and the University of Bristol. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research, or the Department of Health and Social Care. The National Institute for Health Research had no role in study design, data collection analysis, interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. The authors have no conflicts of interest to declare.

Acta Orthopaedica 2020; 91 (6): 794â&#x20AC;&#x201C;800

Rate of revision for infection per 1,000 PYs

40 Cefazolin, multiple doses Cefazolin, single dose Cefuroxime, multiple doses

During 2011 to 2015, 130,712 primary total hip arthroplasties and 111,467 primary total knee arthroplasties were performed across 99 centers.

Months after surgery (S)

Figure 1. Rate of complete revision for infection in the first 12 months following primary hip replacement by type of antibiotics regimen.

Figure 2. Rate of complete revision for infection in the first 12 months following primary knee replacement by type of antibiotics regimen.

Hazard ratio

4.0

4.0 Cefazolin, single dose vs. Cefazolin, multiple doses (ref)

Cefuroxime, multiple doses vs. Cefazolin, multiple doses (ref)

3.0

2.0

1.0

0.5

Months after surgery (S)

Figure 3. Hazard ratio and 95% CI* of complete revision for infection in the first 12 months following primary hip replacement by type of antibiotics regimen (reference: cefazolin multiple dose). * Derived from unadjusted Poisson model with restricted cubic splines (3 degree of freedom) (see Appendix Table 2). Hazard ratio

Hazard ratio

Cefazolin, single dose vs. Cefazolin, multiple doses (ref)

Cefuroxime, multiple doses vs. Cefazolin, multiple doses (ref)

5.0

4.0

3.0

2.0

1.0

Results

Cefazolin, multiple doses Cefazolin, single dose Cefuroxime, multiple doses

Months after surgery (S)

Figure 4. Hazard ratio and 95% CI* of complete revision for infection during the first 12 months following primary knee replacement by type of antibiotics regimen (reference: cefazolin multiple dose). *Derived from unadjusted Poisson model with restricted cubic splines (3 degree of freedom) (see Appendix Table 3).

Acta Orthopaedica 2020; 91 (6): 794–800

797

Table 1. Hazard ratio (HR) of complete revision for infection in the first 12 months following primary hip replacement (reference: cefazolin multiple dose) Months from primary procedure Unadjusted HR 1 2 3 6 9 12 Adjusted HR a 1 2 3 6 9 12 a Adjusted

cefazoline cefuroxime single dose multiple dose HR (95% CI) HR (95% CI) 0.45 (0.17–1.20) 0.50 (0.17–1.42) 0.60 (0.19–1.87) 1.04 (0.43–2.49) 1.59 (0.82–3.09) 2.18 (1.09–4.38)

1.82 (0.92–3.62) 1.92 (0.92–4.01) 1.59 (0.78–3.25) 1.03 (0.61–1.74) 0.76 (0.36–1.61) 0.61 (0.21–1.78)

0.45 (0.17–1.20) 0.49 (0.17–1.38) 0.59 (0.19–1.79) 1.02 (0.43–2.39) 1.59 (0.83–3.02) 2.21 (1.12–4.38)

1.80 (0.92–3.52) 1.88 (0.92–3.86) 1.54 (0.77–3.08) 1.00 (0.60–1.68) 0.75 (0.35–1.61) 0.61 (0.20–1.81)

for age, sex, BMI, and ASA grade.

399 hips and 303 knees were revised within 1 year of the primary arthroplasty for an indication of infection (Tables 2 and 3, see Supplementary data). Multiple-dose cefazolin (MCZ), multiple-dose cefuroxime (MCX), or single-dose cefazolin (SCZ) antibiotic prophylaxes were respectively administrated to 87%, 4%, and 9% of patients. Hereafter, “revision” refers to “1 and 2-stage revisions.” For total hip arthroplasties, the 1-year rates of revision for infection (CI) were respectively 31/10,000 person-years (28– 35), 39 (25–59), and 23 (15–34) in the groups that received MCZ, MCX, and SCZ; the rates for knee arthroplasties were 27 (24–31), 40 (24–62), and 24 (16–36) respectively. The rates of revision for infection over time according to antibiotic prophylaxis regimen are shown in Figures 1 and 2. Revision for infection was performed most frequently in the first 3 months postoperatively for both hip and knee replacements. While the risk of complete revision for infection appeared to differ over time, no or little evidence of differences between antibiotic prophylaxis regimens was found (Figures 3 and 4). In the first 11 months after primary hip arthroplasty, the risk of revision was comparable between SCZ and MCZ (adjusted HR SCZ vs. MCZ at 3 months 0.59 [0.19–1.8], at 6 months 1.02 [0.43–2.4]), but the risk of revision was higher in the SCZ group thereafter (HR 2.2 [1.1–4.4]). No evidence of difference was found between MCZ and MCX following hip arthroplasty (adjusted HR MCX vs. MCZ at 3 months 1.5 [0.77–3.1], at 6 months 1.0 [0.60–1.7], at 12 months 0.61 [0.20–1.8]). For patients receiving a primary total knee arthroplasty revision rates between SCZ and MCZ were comparable (adjusted HR SCZ vs. MCZ at 3 months 1.8 [0.87–3.8], at 6 months 0.89 [0.15–5.3], at 12 months 0.47 [0.09–2.4]). The risk of revision for infection was also comparable between MCZ and MCX

Table 2. Hazard ratio (HR) of complete revision for infection in the first 12 months following primary knee replacement (reference: cefazolin multiple dose) Months from primary procedure Unadjusted HR 1 2 3 6 9 12 Adjusted HR a 1 2 3 6 9 12 a Adjusted

cefazoline cefuroxime single dose multiple dose HR (95% CI) HR (95% CI) 0.78 (0.33–1.84) 1.52 (0.78–2.95) 1.77 (0.86–3.63) 0.89 (0.15–5.26) 0.58 (0.26–1.26) 0.47 (0.09–2.40)

2.24 (0.48–10.5) 2.70 (1.15–6.30) 1.72 (0.54–5.50) 1.13 (0.66–1.91) 1.36 (0.59–3.11) 1.88 (0.58–6.10)

0.78 (0.33–1.83) 1.55 (0.80–3.02) 1.81 (0.87–3.76) 0.89 (0.15–5.31) 0.58 (0.26–1.28) 0.47 (0.09–2.37)

2.34 (0.49–11.2) 2.70 (1.16–6.29) 1.71 (0.54–5.37) 1.15 (0.65–2.02) 1.38 (0.58–3.30) 1.88 (0.56–6.31)

for age, sex, BMI, and ASA grade

(adjusted HR MCX vs. MCZ at 3 months 1.7 [0.54–5.4], at 6 months 1.2 [0.65–2.0], at 12 months 1.9 [0.56–6.1]). The patterns observed were comparable in the unadjusted and adjusted models (Tables 1 and 2).

Discussion In this large observational cohort study of primary total hip and knee replacement, our findings suggest a comparable risk of complete revision for infection between the antibiotic prophylaxis regimens in terms of type of antibiotic and duration of prophylaxis during the first 12 months following surgery. When examining the hazard ratios, it is important to note that the majority of infections occurred within the first 3 months of surgery. Comparing single and multi-dose prophylaxis with cefazolin for hip replacement, the hazard ratio for complete revision for infection following single-dose prophylaxis steadily increased over time from less than half of that with multi-dose to over double the incidence of infection by month 12. This may be due to low-virulence micro-organisms that are more susceptible to multi-dose therapy presenting with infection later. If this is true, the differences between the different regimes should become more apparent with longer follow-up. This was not the case following knee replacement and alternatively may simply reflect either a chance occurrence, differences in patient- and surgery-related factors, or residual confounding. Adjustment for established confounding variables (age, sex, BMI, ASA grade) did not change these results. We observed that the highest risk of complete revision for infection in the year following surgery occurred within the

798

first 3 months after the operation. Rates then appear to rise again towards the end of the follow-up period. These patterns are consistent with contemporary patterns found in other registries (Dale et al. 2012, Lenguerrand et al. 2017a, 2017b). This may be due to the effect of more virulent microorganisms presenting during the first 3 months and less virulent microorganisms presenting later. Since the LROI does not provide data on which microorganism is causing the PJI, this remains speculative. Another reason might be a genuine increase in the incidence of PJI or may reflect more rapid diagnosis and aggressive treatment of PJI in recent years. We have not analyzed procedures where only debridement or partial revision (including debridement and implant retention [DAIR] with modular exchanges) were performed as these procedures are not reliably captured by the LROI registry (Veltman et al. 2018). DAIR has been shown to treat infection effectively in approximately 46â&#x20AC;&#x201C;76% of cases (Wouthuyzen-Bakker et al. 2020). We have no reason to believe that the use of DAIR is related to type or duration of antibiotic prophylaxis, but it is a possible cause of residual confounding. It has been suggested that the most appropriate perioperative prophylactic antibiotic is a first- or second-generation cephalosporin (i.e., cefazolin or cefuroxime) administered intravenously within 30 to 60 minutes prior to incision as a single and weight-adjusted dose (AlBuhairan et al. 2008, Stefansdottir et al. 2009, Steinberg et al. 2009). This policy is part of antibiotic stewardship, performed in countries with a low prevalence of MRSA (Illingworth et al. 2013, American Academy of Orthopaedic Surgeons/American Association of Orthopaedic Surgeons 2014). While consensus exists on type of antibiotic prophylaxis (Parvizi and Gehrke 2018) the postoperative duration of antibiotic prophylaxis remains subject to discussion. A recent systematic review and meta-analysis by Thornley et al. (2015) explored whether or not a single preoperative antibiotic dose is adequate for arthroplasty patients. The review included 4 RCTs including 4,036 patients (Heydemann and Nelson 1986, Ritter et al. 1989, Wymenga et al. 1992, Kanellakopoulou et al. 2009). They concluded that additional postoperative antibiotic doses did not reduce the rates of infections (3.1% versus 2.3% postoperative PJI for multiple-dose and single-dose prophylaxis respectively). However, they reported that the quality of the included studies was very low. 3 of these studies were performed more than 20 years ago, while the other study used teicoplanin, which is no longer recommended for use as antibiotic prophylaxis (Berrios-Torres et al. 2017). Heydemann and Nelson (1986) randomized 211 patients between single-dose and 48-hour multiple-dose prophylaxis, but found no cases of PJI in either group. Ritter et al. (1989) compared a single dose of cefuroxime to 24 hours of postoperative prophylaxis in 196 patients, and found no cases of PJI in either group. Wymenga et al. (1992) randomized 3,013 patients in a multicenter RCT comparing a single preoperative dose of cefuroxime to a group receiving three doses and

Acta Orthopaedica 2020; 91 (6): 794â&#x20AC;&#x201C;800

found no significant differences in PJI rates between groups. Engesaeter et al. (2003) reported the lowest rate of infection for patients who received 4 doses of antibiotic prophylaxis in 24 hours, compared with patients who received 1, 2, or 3 doses in their study of the Norwegian Arthroplasty Register. All authors of these studies recognized their study sample to be underpowered for determining a difference in PJI rates and recommended further studies to provide a definite answer. Based on these studies, the CDC has recently recommended against the use of postoperative continuation of antibiotic prophylaxis (Berrios-Torres et al. 2017). The recent International Consensus meeting advises to continue antibiotics postoperatively for 24 hours until better quality evidence is available (Parvizi and Gehrke 2018). A protocol for an RCT randomizing patients receiving a total knee arthroplasty between singledose versus multiple-dose antibiotic prophylaxis has been registered on clinicaltrials.gov (NCT03283878). The study aims to answer definitively what duration of antibiotic prophylaxis is best. However, the planned follow-up of 90 days seems too short to capture all relevant infections. Also, the sample size is not justified in the trial registration, but with the aim of including 8,000 patients the study seems underpowered. Our study has several strengths. The large numbers studied allows adequate power to detect rare outcomes such as complete revision for infection. Data capture represents over 98% of national activity (Dutch Arthroplasty Register 2017). This rate of coverage provides excellent external validity and generalizability of our findings. The rate of complete revision for infection within 1 year of primary arthroplasty is higher for males, patients with higher BMI, or higher ASA grade in all groups, independent of the type of antibiotic prophylaxis (Dale et al. 2012, Lenguerrand et al. 2018). This is in agreement with the literature and highlights the comparability of this Dutch arthroplasty cohort to other studied cohorts (Dale et al. 2012, Lenguerrand et al. 2018, Kunutsor et al. 2018b). In order to establish the current practice for antibiotic prophylaxis regimes, we conducted a comprehensive national survey to determine current practice. The outcome of interest is a binary endpoint, and whilst this may mean that not all cases of PJI are captured, as many may be treated without complete revision surgery, it does make the end-point easily defined (Blom et al. 2003). In the absence of randomized controlled trials on the type and duration of antibiotic prophylaxis, this natural experiment in a large and generalizable national registry represents the best data currently available to determine whether there is a difference in the risk of complete revision for infection according to the antibiotic prophylaxis regimen. The study does have limitations. The LROI database was established as an arthroplasty register, and whilst one of the outcomes of interest is complete revision for infection, the register was not designed to capture all infection outcomes and thus there is likely to be underreporting of infection as may also be the case in other national arthroplasty registries (Gundtoft et al. 2015, Kunutsor et al. 2018b). The most nota-

Acta Orthopaedica 2020; 91 (6): 794–800

ble effect of this is the lack of capture of further procedures performed after the primary surgery to manage infection, such as DAIR procedures. The Dutch survey showed only 64% of hospitals registered DAIR procedures in the LROI, thus we did not include these in our analysis. As about 50% of PJI may be treated only with DAIR and arthroplasty registries are known to provide an underestimation of the rate of prosthetic revisions due to PJI of 20%, we may be missing as much as 70% of all treated infections (Gundtoft et al. 2015, Kunutsor et al. 2018a). Although prospectively collected our data are observational, and we can only draw conclusions on the nature and magnitude of the associations but cannot establish causative relation due to the possibility of residual confounding and estimation uncertainty. Whilst we conducted a comprehensive survey to establish the current practice in terms of antibiotic prophylaxis regimes, it is likely that for various reasons, including allergy, intolerance, and surgeons’ preference, not all patients received the antibiotics as per hospital protocol. However, a recent large retrospective study in the USA showed that 95% of patients received standard antibiotic prophylaxis (Wyles et al. 2019). The three types of antibiotics all are cephalosporins with the same allergy profile, therefore the percentage of patients with allergies should be comparable in all groups. Changes to the local antibiotic protocols during the study period have not been captured by the survey. The Dutch guideline for antibiotic prophylaxis around primary hip and knee arthroplasty did not change during the time period. However, changes to the antibiotic protocols may have occurred between the groups in all directions. Due to the quasi-randomized allocation of our patients, this should not introduce systematic bias. Thus, this study resembles a natural experiment. Rather than controlling for observed confounders and expecting no unobserved confounders to be present (as in multiple regression, matching, and reweighting), natural experiments identify variation in the exposure, known to be independent of other confounders (Bor 2016). In our study quasi-random variation in the exposure (antibiotic prophylaxis regimen after total hip or knee arthroplasty) arises from naturally occurring random variation due to allocation of patients to the regional hospital near their residence. Natural experiments minimize the risk of confounding due to selective exposure to the intervention or residual confounding, and have internal validity and transparency of assumptions (Bor 2016). To establish true causality, a superiority or non-inferiority randomized controlled trial is still needed. However, as PJI is rare, the numbers needed for such a trial would be very large. Nonetheless, as the impact of PJI is so devastating (Moore et al. 2015) we recommend that such a trial is undertaken and suggest that embedding such a trial in a national arthroplasty registry may reduce costs and improve feasibility. Until such time, the data represented here are the best available evidence and it must be questioned whether there is any advantage to the use of prolonged antibiotic prophylaxis beyond a single dose.

799

Supplementary data The Appendix Tables 1–3 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/ 17453674.2020.1794096

The authors thank the patients and staff of all the hospitals who have contributed data to the LROI database. They are grateful to the Netherlands Orthopaedic Association (NOV) and to the LROI for granting access to this database. We thank Liza van Steenbergen of the LROI office for her help with data extraction and her prompt support with our data management queries. ESV, EL, DJM, and RWP designed the study. The data were extracted from the LROI database by Liza van Steenbergen of the LROI. ESV performed the literature search. EL performed the data analysis. All authors interpreted data, drafted, and reviewed the final manuscript. Acta thanks Håvard Dale and Javad Parvizi for help with peer review of this study.

AlBuhairan B, Hind D, Hutchinson A. Antibiotic prophylaxis for wound infections in total joint arthroplasty: a systematic review. J Bone Joint Surg Br 2008; 90(7): 915-19. American Academy of Orthopaedic Surgeons/American Association of Orthopaedic Surgeons. Information statement: recommendations for the use of intravenous antibiotic prophylaxis in primary total joint arthroplasty; 2014. https://www.aaos.org/globalassets/about/bylaws-library/information-statements/1027-recommendations-for-the-use-of-intravenous-antibiotic-prophylaxis-in-primary-total-joint-arthroplasy.pdf Berrios-Torres S I, Umscheid C A, Bratzler D W, Leas B, Stone E C, Kelz R R, Reinke C E, Morgan S, Solomkin J S, Mazuski J E, Dellinger E P, Itani K M F, Berbari E F, Segreti J, Parvizi J, Blanchard J, Allen G, Kluytmans J A J W, Donlan R, Schecter W P. Centers for Disease Control and Prevention Guideline for the Prevention of Surgical Site Infection 2017. JAMA Surg 2017; 152(8): 784-91. Blom A W, Taylor A H, Pattison G, Whitehouse S, Bannister G C. Infection after total hip arthroplasty: the Avon experience. J Bone Joint Surg Br 2003; 85(7): 956-9. Bor J. Capitalizing on natural experiments to improve our understanding of population health. Am J Public Health 2016; 106(8): 1388-9. Dale H, Fenstad A M, Hallan G, Havelin L I, Furnes O, Overgaard S, Pedersen A B, Karrholm J, Garellick G, Pulkkinen P, Eskelinen A, Makela K, Engesaeter L B. Increasing risk of prosthetic joint infection after total hip arthroplasty. Acta Orthop 2012; 83(5): 449-58. Dutch Arthroplasty Register (LROI). Annual report 2014: Arthroplasty in the picture; 2014. Dutch Arthroplasty Register (LROI). LROI annual report 2017: 10 years of registration, a wealth of information; 2017. Engesaeter L B, Lie S A, Espehaug B, Furnes O, Vollset S E, Havelin L I. Antibiotic prophylaxis in total hip arthroplasty: effects of antibiotic prophylaxis systemically and in bone cement on the revision rate of 22,170 primary hip replacements followed 0-14 years in the Norwegian Arthroplasty Register. Acta Orthop Scand 2003; 74(6): 644-51. Gundtoft P H, Overgaard S, Schonheyder H C, Moller J K, KjaersgaardAndersen P, Pedersen A B. The “true” incidence of surgically treated deep prosthetic joint infection after 32,896 primary total hip arthroplasties: a prospective cohort study. Acta Orthop 2015; 86(3): 326-34. Hansen E, Belden K, Silibovsky R, Vogt M, Arnold W, Bicanic G, Bini S, Catani F, Chen J, Ghazavi M, Godefroy K M, Holham P, Hosseinzadeh H, Kim K I, Kirketerp-Moller K, Lidgren L, Lin J H, Lonner J H, Moore C C, Papagelopoulos P, Poultsides L, Randall R L, Roslund B, Saleh K, Salmon J V, Schwarz E, Stuyck J, Dahl A W, Yamada K. Perioperative antibiotics. J Orthop Res 2014; 32(Suppl. 1): S31-S59.

800

Heydemann J S, Nelson C L. Short-term preventive antibiotics. Clin Orthop Relat Res 1986; (205): 184-7. Hickson C J, Metcalfe D, Elgohari S, Oswald T, Masters J P, Rymaszewska M, Reed M R, Sprowson A P. Prophylactic antibiotics in elective hip and knee arthroplasty: an analysis of organisms reported to cause infections and national survey of clinical practice. Bone Joint Res 2015; 4(11): 181-9. Illingworth K D, Mihalko W M, Parvizi J, Sculco T, McArthur B, el B Y, Saleh K J. How to minimize infection and thereby maximize patient outcomes in total joint arthroplasty: a multicenter approach: AAOS exhibit selection. J Bone Joint Surg Am 2013; 95(8): e50. Kanellakopoulou K, Papadopoulos A, Varvaroussis D, Varvaroussis A, Giamarellos-Bourboulis E J, Pagonas A, Stergiou A, Papadelis P, Nikolaidis V, Giamarellou H. Efficacy of teicoplanin for the prevention of surgical site infections after total hip or knee arthroplasty: a prospective, open-label study. Int J Antimicrob Agents 2009; 33(5): 437-40. Kunutsor S K, Beswick A D, Whitehouse M R, Wylde V, Blom A W. Debridement, antibiotics and implant retention for periprosthetic joint infections: a systematic review and meta-analysis of treatment outcomes. J Infect 2018a; 77(6): 479-88. Kunutsor S K, Whitehouse M R, Blom A W, Board T, Kay P, Wroblewski B M, Zeller V, Chen S Y, Hsieh P H, Masri B A, Herman A, Jenny J Y, Schwarzkopf R, Whittaker J P, Burston B, Huang R, Restrepo C, Parvizi J, Rudelli S, Honda E, Uip D E, Bori G, Munoz-Mahamud E, Darley E, Ribera A, Canas E, Cabo J, Cordero-Ampuero J, Redo M L S, Strange S, Lenguerrand E, Gooberman-Hill R, Webb J, MacGowan A, Dieppe P, Wilson M, Beswick A D. One- and two-stage surgical revision of peri-prosthetic joint infection of the hip: a pooled individual participant data analysis of 44 cohort studies. Eur J Epidemiol 2018b; 33: 933-46. Kurtz S M, Ong K L, Lau E, Bozic K J. Impact of the economic downturn on total joint replacement demand in the United States: updated projections to 2021. J Bone Joint Surg Am 2014; 96(8): 624-30. Lenguerrand E, Whitehouse M R, Beswick A D, Jones S A, Porter M L, Blom A W. Revision for prosthetic joint infection following hip arthroplasty: evidence from the National Joint Registry. Bone Joint Res 2017a; 6(6): 391-8. Lenguerrand E, Whitehouse M R, Beswick A D, Toms A D, Porter M L, Blom A W. Description of the rates, trends and surgical burden associated with revision for prosthetic joint infection following primary and revision knee replacements in England and Wales: an analysis of the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. BMJ Open 2017b; 7(7): e014056. Lenguerrand E, Whitehouse M R, Beswick A D, Kunutsor S K, Burston B, Porter M, Blom A W. Risk factors associated with revision for prosthetic joint infection after hip replacement: a prospective observational cohort study. Lancet Infect Dis 2018; 18(9): 1004-14. Maradit K H, Larson D R, Crowson C S, Kremers W K, Washington R E, Steiner C A, Jiranek W A, Berry D J. Prevalence of total hip and knee replacement in the United States. J Bone Joint Surg Am 2015; 97(17): 1386-97. Moore A J, Blom A W, Whitehouse M R, Gooberman-Hill R. Deep prosthetic joint infection: a qualitative study of the impact on patients and their experiences of revision surgery. BMJ Open 2015; 5(12): e009495. National Joint Registry for England and Wales. 15th Annual Report 2018; 2018. https://www.hqip.org.uk/resource/national-joint-registry-15th-annualreport-2018/ Nederlandse Orthopaedische Vereniging. Richtlijn perioperatieve zorg van totale heupprothese / preventie van wondinfectie bij heupprothese; 2018. https://www.hqip.org.uk/resource/national-joint-registry-15th-annualreport-2018/ Parvizi J, Gehrke T. Proceedings of the Second International Consensus on Musculoskeletal Infection; 2018. https://www.arthroplastyjournal.org/ issue/S0883-5403(18)X0012-6

Acta Orthopaedica 2020; 91 (6): 794–800

Parvizi J, Zmistowski B, Della V C, Bauer T W, Malizos K N, Alavi A, Bedair H, Booth R E, Choong P, Deirmengian C, Ehrlich G D, Gambir A, Huang R, Kissin Y, Kobayashi H, Kobayashi N, Krenn V, Lorenzo D, Marston S B, Meermans G, Perez J, Ploegmakers J J, Rosenberg A, Simpendorfer C, Thomas P, Tohtz S, Villafuerte J A, Wahl P, Wagenaar F C, Witzo E. Diagnosis of periprosthetic joint infection. J Arthroplasty 2013. PMID: 24342275 Poultsides L A, Liaropoulos L L, Malizos K N. The socioeconomic impact of musculoskeletal infections. J Bone Joint Surg Am 2010; 92(11): e13. Ritter M A, Campbell E, Keating E M, Faris P M. Comparison of intraoperative versus 24 hour antibiotic prophylaxis in total joint replacement: a controlled prospective study. Orthop Rev 1989; 18(6): 694-6. Royston P, Lambert P C. Time-dependent effects. In: Flexible parametric survival analysis using Stata: beyond the Cox model. College Station,TX: Stata Press 2011. p. 184-9. Stefansdottir A, Robertsson O, Dahl A, Kiernan S, Gustafson P, Lidgren L. Inadequate timing of prophylactic antibiotics in orthopedic surgery: we can do better. Acta Orthop 2009; 80(6): 633-8. Steinberg J P, Braun B I, Hellinger W C, Kusek L, Bozikis M R, Bush A J, Dellinger E P, Burke J P, Simmons B, Kritchevsky S B. Timing of antimicrobial prophylaxis and the risk of surgical site infections: results from the Trial to Reduce Antimicrobial Prophylaxis Errors. Ann Surg 2009; 250(1): 10-16. Swierstra B A, Bijlsma J W, de Beer J J, Kuijpers T. [Guideline “Diagnostics and treatment of osteoarthrosis of the hip and knee”]. Ned Tijdschr Geneeskd 2009; 153: B39. Tan T L, Shohat N, Rondon A J, Foltz C, Goswami K, Ryan S P, Seyler T M, Parvizi J. Perioperative antibiotic prophylaxis in total joint arthroplasty: a single dose is as effective as multiple doses. J Bone Joint Surg Am 2019; 101(5): 429-37. Thornley P, Evaniew N, Riediger M, Winemaker M, Bhandari M, Ghert M. Postoperative antibiotic prophylaxis in total hip and knee arthroplasty: a systematic review and meta-analysis of randomized controlled trials. CMAJ Open 2015; 3(3): E338-E343. van Steenbergen L N, Denissen G A, Spooren A, van Rooden S M, van Oosterhout F J, Morrenhof J W, Nelissen R G. More than 95% completeness of reported procedures in the population-based Dutch Arthroplasty Register. Acta Orthop 2015; 86(4): 498-505. Veltman E S, Moojen D J F, Nelissen R G, Poolman R W. Antibiotic prophylaxis and DAIR treatment in primary total hip and knee arthroplasty, a national survey in the Netherlands. J Bone Jt Infect 2018; 3(1): 5-9. Wouthuyzen-Bakker M, Sebillotte M, Huotari K, Escudero S R, Benavent E, Parvizi J, Fernandez-Sampedro M, Barbero-Allende J M, Garcia-Canete J, Trebse R, Del T M, Diaz-Brito V, Sanchez M, Scarborough M, Soriano A. Lower success rate of debridement and implant retention in late acute versus early acute periprosthetic joint infection caused by Staphylococcus spp: results from a matched cohort study. Clin Orthop Relat Res 2020; 478(6): 1348-55. Wyles C C, Hevesi M, Osmon D R, Park M A, Habermann E B, Lewallen D G, Berry D J, Sierra R J. 2019 John Charnley Award: Increased risk of prosthetic joint infection following primary total knee and hip arthroplasty with the use of alternative antibiotics to cefazolin: the value of allergy testing for antibiotic prophylaxis. Bone Joint J 2019; 101-B(6_Supple_B): 9-15. Wymenga A, van H J, Theeuwes A, Muytjens H, Slooff T. Cefuroxime for prevention of postoperative coxitis: one versus three doses tested in a randomized multicenter study of 2,651 arthroplasties. Acta Orthop Scand 1992; 63(1): 19-24. Yates A J Jr. Postoperative prophylactic antibiotics in total joint arthroplasty. Arthroplast Today 2018; 4(1): 130-1. Zmistowski B, Karam J A, Durinka J B, Casper D S, Parvizi J. Periprosthetic joint infection increases the risk of one-year mortality. J Bone Joint Surg Am 2013; 95(24): 2177-84.

Acta Orthopaedica 2020; 91 (6): 801–802

801

Correspondence

Trauma and orthopaedics training amid COVID-19: A medical student’s perspective

Sir,—We read with interest the article by Dattani et al. (2020), describing the effects of coronavirus disease 2019 (COVID19) on Trauma and Orthopaedics (T&O) training. As medical students of the “COVID Generation”, we take interest in this perspective. Yet, we notice there is little reflection on the pandemic’s impacts on undergraduate T&O education. Hence, we offer strategies to adapt undergraduate T&O teaching, highlighting telemedicine opportunities and limitations. The COVID-19 pandemic has affected all fields worldwide, particularly healthcare training. Practical anatomy, clinical skills sessions, and hospital placements were suspended, raising concerns over medical students’ competency development. Restricted opportunities to practice physical examinations and observe clinical management (BOA 2014) has reduced students’ T&O specialty exposure, which can influence career selection (Johnson et al. 2012). Cancellation of undergraduate research placements, conferences, and electives has hindered learning, networking, and personal and professional development, while examination disruptions have prevented selfassessment. Although universities offered online resources, these are mainly limited to theoretical education, neglecting practical experience and simulation (Yaghobian et al. 2020). Moreover, although virtual patients are rated higher than recorded lectures (Courteille et al. 2018), these are underutilised and, still, cannot replace real-life clinical training. Dattani et al.’s telemedicine strategies for postgraduate trainees should be implemented more extensively in undergraduate medical education. Telemedicine can enrich student education during COVID19 and will likely continue being used, as the authors mention. Simulation and augmented reality are effective learning aids, especially for human anatomy assimilation (Moro et al. 2017). Simulation can prepare students for real-life T&O emergency situations, improving diagnostic reasoning, while reducing risks to patients (Lateef 2010). Small-group virtual teaching could maintain practical skill development which is essential for surgical T&O education. Encouraging students to observe and, when appropriate, participate in virtual clinics and surgeries would maintain access to healthcare environments, facilitating regular student-patient interaction whilst ensuring COVID-19 safety (Courteille et al. 2018, Stelian and Lacramioara 2018).

However, even though teleconsultations offer regular patient contact, it is harder to build rapport and trusted relationships with patients (Ekeand et al. 2010), and physical examinations are restricted (Romanick-Schmiedl and Raghu 2020). Moreover, with the expanding use of e-health there are concerns about quality of care and confidentiality of patient information, whist telemedicine can be technically challenging for patients (Romanick-Schmiedl and Raghu 2020). Hence, telemedicine may pose barriers for physicians, students, and patients. It is therefore paramount to introduce teachings in medical school to familiarise students with telemedicine. Early training should teach students how to interact with patients through telecommunications, adapting communication skills to the online setting. Moreover, students should be learning how to triage patients that require hospital assessment and realise that telemedicine requires adaptation to the patient. Medicine faces an unprecedented need for short- and longterm flexibility and adaptation. Doctors’ training must be adapted to minimise disruptions to career progression and quality of care. Equally, undergraduate medical education must be refined and reinvented to ensure future physicians’ competence and confidence in practice. Increasing understanding of telemedicine strengths and limitations is essential for physicians, students and patients. Carola Maria Bigogno 1 and Kathrine S Rallis 1,2 and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK 2 Barts Cancer Institute, Queen Mary University of London, London, UK Email: c.m.bigogno@se15.qmul.ac.uk 1 Barts

Sir,—Our article focused on the impact of the COVID-19 pandemic on postgraduate training within trauma and orthopaedics. However, as mentioned in the letter by Bigogno et al, it is also important to consider the impact of the pandemic on undergraduate education, as without medical students there is no next generation of orthopaedic surgeons. Thus, it is para-

802

mount to consider this perspective and to form strategies to help adapt undergraduate education in trauma and orthopaedics during the evolving COVID-19 pandemic. As mentioned in the letter, the COVID-19 pandemic has caused large disruption to the undergraduate medical curriculum affecting medical students across all years, including many students in their clinical years being moved to the front line (Representatives of the STARSurg Collaborative, EuroSurg Collaborative, and TASMAN Collaborative. 2020). For many students, interest in a specialty begins in medical school and the medical school rotation experience has been shown to have an important role in shaping interest and perceptions within trauma and orthopaedics (Baldwin et al. 2011). It is vital that this undergraduate exposure to orthopaedics is not lost. The COVID-19 pandemic should be used as an unexpected opportunity for medical schools to review and improve their current curriculum content by adopting novel teaching methods to ensure the education of medical students’ is not compromised. A lot of the strategies and concepts discussed in our article can be applied to both undergraduate and postgraduate education. The use of surgical simulation, virtual reality and telemedicine within the field of trauma and orthopaedics, is particularly relevant during the COVID-19 pandemic and medical schools should look at ways to integrate this further into the curriculum. The letter refers to the use of augmented reality as an effective learning aid for human anatomy (Moro et al. 2017). Mixed-reality headsets have also been used for ward rounds during the COVID-19 pandemic, helping reduce the number of people on the round and the amount of personal protective equipment used (Imperial College London. 2020). This reduces the time doctors and students spend in high-risk areas as well as providing an educational opportunity. As orthopaedic surgeons, at all levels of training, it is our duty to work together with medical schools to help deliver a safe but effective undergraduate orthopaedic teaching rotation during the COVID-19 pandemic. As the possibility of a second wave approaches, medical student education should remain a priority.

Acta Orthopaedica 2020; 91 (6): 801–802

Catrin Morgan and Rupen Dattani Chelsea and Westminster NHS Foundation Trust, London, UK Email: rupen.dattani@nhs.net Baldwin K, Namdari S, Bowers A, Keenan M A, Levin L S, Ahn J. Factors affecting interest in orthopedics among female medical students: A prospective analysis. Orthopedics 2011; 34 (12): e919-32. British Orthopaedic Association. Trauma and Orthopaedic Undergraduate Sillabus. 2014. Available at: file:///C:/Users/cmbig/Downloads/to-undergraduate-syllabus_website.pdf [accessed 08/09/2020] Courteille O, Fahlstedt M, Ho J, Hedman L, Fors U, von Holst H, FallanderTsai L, Moller H. Learning through a virtual patient vs recorded lecture: a comparison of knowledge retention in a trauma case. Int J Med Educ 2018; 9: 86-92. doi: 10.5116/ijme.5aa3.ccf2 Dattani R, Morgan C, Li L, Bennet-Brown K, Wharton R M H. The impact of COVID-19 on the future of orthopaedic training in the UK. Acta Orthop 2020: 1-6 [Online ahead of print] doi: 10.1080/17453674.2020.1795790 Ekeand AG, Bowes A, Flottorp S. Effectiveness of telemedicine: A systemic review of review. Int J Med Informatics 2010; 79(11): 736-71. doi. org/10.1016/j.ijmedinf.2010.08.006 Imperial College London. Mixed-reality headsets in hospitals help protect doctors and reduce need for PPE. 2020. Available at https://www.imperial. ac.uk/news/197617/mixed-reality-headsets-hospitals-help-protect-doctors/ Johnson A L, Sharma J, Chinchilli V M, Emery S E, McCollister Evarts C, Floyd M W, Keading C C, Lavelle W F, Marsh J L, Pellegrini V D Jr, Van Heest A E, Black K P. Why do medical students choose orthopaedics as a career? J Bone Joint Surg Am 2012; 94(11): e78. doi: 10.2106/ JBJS.K.00826 Lateef F. Simulation-based learning: Just like the real thing. J Emerg Trauma Shock 2010; 3(4): 348-52. doi: 10.4103/0974-2700.70743 Moro C, Stronberga Z, Raikos A, Stjrling A. The effectiveness of virtual and augmented reality in health sciences and medical anatomy. Anat Sci Educ 2017; 10(6): 549-59. doi: 10.1002/ase.1696 Representatives of the STARSurg Collaborative, EuroSurg Collaborative, and TASMAN Collaborative. Medical student involvement in the COVID-19 response. Lancet 2020; 395 (10232): 1254,6736(20)30795-9. Epub 2020 Apr 2. Romanick-Schmiedl S, Raghu G. Telemedicine – maintaining quality during times of transition. Nature Reviews 2020; 6(1): 45. doi.org/10.1038/ s41572-020-0185-x Stelian N, Lacramioara ST. Mixed reality supporting modern medical education. Stud Health Technol Inform 2018; 255: 242-6. Yaghobian S, Ohannessian R, Iampetro T, Riom I, Salles N, Medeiros de Bustos E, Moulin T, Mathieu-Fritx A. Knowledge, attitudes and practices of telemedicine education and training of French medical students and residents. J Telemed Telecare 2020. [Online ahead of print] doi:1357633X20926829.

Acta Orthopaedica 2020; 91 (6): 803

803

Correspondence

Delayed surgery versus nonoperative treatment for hip fractures in post-COVID-19 situation

Sir,—I would like to comment on the publication “Delayed surgery versus nonoperative treatment for hip fractures in post-COVID-19 arena: a retrospective study of 145 patients [Mi et al. 2020].” Mi et al. concluded that “In hip fracture patients, delayed surgery compared with nonoperative therapy significantly improved hip function and reduced various major complications.” The result from this study is to be expected, as surgical management is generally recommended for treatment of hip fractures. The important consideration in this retrospective study is the reason for selection of a therapeutic alternative in each patient. However, the analysis of Mi et al. is in a “post-COVID-19 arena,” hence, there should be no reason for delayed surgery or nonoperative treatment of hip fractures.

Sir,—I thank Viroj Wiwanitkit for his comments. There is a consensus that patients with hip fractures should be operated on without delay under ordinary circumstances. However, the outbreak of COVID-19 in Wuhan, China caused a large number of infected patients, who occupied a large amount of medical resources resulting in some hip fracture patients being unable to gain timely admission and surgery. Furthermore, some patients with hip fractures chose to stay at home, worried about being infected with COVID-19 in hospital. After the post-outbreak period several hip fracture patients began to visit hospitals seeking surgical treatment, but some patients still chose nonoperative treatment. We will continue to follow up these patients and provide them with free individual rehabilitation programs.

Viroj WIWANITKIT Honorary professor, Dr DY Patil University, Pune, India; adjunct professor, Joseph Ayobaalola University, Ikeji-Arakeji, Nigeria Email: wviroj@yahoo.com

Guohui LIU Department of Orthopedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China Email: liuguohui@hust.edu.cn

Mi B, Chen L, Tong D, Panayi A C, Ji F, Guo J, Ou Z, Zhang Y, Xiong Y, Liu G. Delayed surgery versus nonoperative treatment for hip fractures in postCOVID-19 arena: a retrospective study of 145 patients. Acta Orthop 2020; Sep 8: 1-5. doi: 10.1080/17453674.2020.1816617. Online ahead of print.

6/20

* Sprowson AP et al. Bone Joint J 2016; 98-B: 1534–1541

www.heraeus-medical.com

Vol. 91, No. 6, 2020 (pp. 621–803)

Bone cement with gentamicin ttamicin i i and clindamycin

ACTA ORTHOPAEDICA

69 %

reduction of deep infections in hip hemiarthroplasty after fractured neck of femur *

Volume 91, Number 6, December 2020

TF-IORT200163.indd 1

20-11-2020 16:13:12