Acta Orthopaedica, Vol. 92. Issue 2, 2021 by Acta

2/21 ACTA ORTHOPAEDICA

Medical

IMPROVE THE CHANCES

REDUCE RISK FOR INFECTION Reduction of infection risk* using dual antibiotic-loaded bone cement in high risk patients

in aseptic revision TKA * as reported in study results

10453

www.heraeus-medical.com

in fractured neck of femur

Vol. 92, No. 2, 2021 (pp. 127–248)

34 % 69 % 57 %

in primary hip & knee arthroplasty

Volume 92, Number 2, April 2021

2102_20479_AZ_COPAL_RFI_Acta_Orthopaedica_Europe_215x280_EN.indd 1 TF-IORT210050.indd 1

23.02.21 08:49

29-04-2021 19:41:39

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. 92, No. 2, 2021

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 © 2021 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. 92, No. 2, April 2021 Guest editorial Is there a reason to challenge our current practice in children’s forearm fractures?

127

L S Lohmander and I A Harris

COVID-19 Reducing orthopaedic theatre exposure during the COVID-19 lockdown: does a shift towards virtual reality-based training offer a solution?

129

A Arshad, A Zaveri, and H Atkinson

Lower arm Reliability of recommendations to reduce a fracture of the distal radius

131

E Z Boersma, J T P Kortlever, M W G Nijhuis–van der Sanden, M J R Edwards, D Ring, and T Teunis

137

L L Hermansen, B Viberg, and S Overgaard

143

Y Tyson, C Hillman, N Majenburg, O Sköldenberg, O Rolfson, J Kärrholm, M Mohaddes, and N P Hailer

151

G Rochcongar, M Remazeilles, E Bourroux, J Dunet, V Chapus, M Feron, C Praz, G Buia, and C Hulet U Betz, L Langanki, F Heid, J Spielberger, L Schollenberger, K Kronfeld, M Büttner, B Büchler, M Goldhofer, L Eckhard, and P Drees; the PROMISE Group B Viberg, T Frøslev, S Overgaard, and A B Pedersen

Hip Development of a diagnostic algorithm identifying cases of dislocation after primary total hip arthroplasty—based on 31,762 patients from the Danish Hip Arthroplasty Register Uncemented or cemented stems in first-time revision total hip replacement? An observational study of 867 patients including assessment of femoral bone defect size Reduced wear in vitamin E-infused highly cross-linked polyethylene cups: 5-year results of a randomized controlled trial The PROMISE study protocol: a multicenter prospective study of process optimization with interdisciplinary and cross-sectoral care for German patients receiving hip and knee endoprostheses Mortality and revision risk after femoral neck fracture: comparison of internal fixation for undisplaced fracture with arthroplasty for displaced fracture: a population-based study from Danish National Registries Influence of day of surgery and prediction of LOS > 2 days after fast-track hip and knee replacement C C Jørgensen, K Gromov, P B Petersen, and H Kehlet, on behalf of the Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Collaborative Group Increased rate of complications in myasthenia gravis patients following hip and knee arthroplasty: a nationwide database study in the PearlDiver Database on 257,707 patients Knee Variation and trends in reasons for knee replacement revision: a multi-registry study of revision burden Compensation claims after knee arthroplasty surgery in Norway 2008–2018 Innervation of the distal part of the vastus medialis muscle is endangered by splitting its muscle fibers during total knee replacement: an anatomical study using modified Sihler’s technique Spine National trends in lumbar spine decompression and fusion surgery in Finland, 1997–2018 Infection Are conventional microbiological diagnostics sufficiently expedient in the era of rapid diagnostics? Evaluation of conventional microbiological diagnostics of orthopedic implant-associated infections (OIAI) Increasing but levelling out risk of revision due to infection after total hip arthroplasty: a study on 108,854 primary THAs in the Norwegian Arthroplasty Register from 2005 to 2019 Hospital variation in the risk of infection after hip fracture surgery: a population-based cohort study including 29,598 patients from 2012–2017

156 163

170

C C Jørgensen, K Gromov, P B Petersen, and H Kehlet, on behalf of the Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Collaborative Group

176

W F Sherman, V J Wu, S A Ofa, B J Ross, I D Savage-Elliott, and F L Sanchez

182

P L Lewis, O Robertsson, S E Graves, E W Paxton, H A Prentice, and A W-Dahl P-H Randsborg, T F Aae, I R K Bukholm, A M Fenstad, O Furnes, and R B Jakobsen B Pretterklieber, A Ungersböck, and M L Pretterklieber

189 194

199

V T Ponkilainen, T T Huttunen, M H Neva, L Pekkanen, J P Repo, and V M Mattila

204

H V Aamot, J C Noone, I Skråmm, and T M Leegaard

208

H Dale, P Høvding, S M Tveit, J B Graff, O Lutro, J C Schrama, T S Wik, I Skråmm, M Westberg, A M Fenstad, G Hallan, L B Engesæter, and O Furnes J D Vesterager, P K Kristensen, I Petersen, and A B Pedersen

215

Children Knee and foot contracture occur earliest in children with cerebral palsy: a longitudinal analysis of 2,693 children Risk factors for implant-related fractures after proximal femoral osteotomy in children with developmental dysplasia of the hip: a case-control study Most surgeons still prefer to reduce overriding distal radius fractures in children

222 228

E Cloodt, P Wagner, H Lauge-Pedersen, and E Rodby-Bousquet J Ding, Z-Z Dai, Z Liu , Z-K Wu , Z-M Zhang, and H Li

235

T Laaksonen, J Puhakka, J Kosola, A Stenroos, M Ahonen, and Y Nietosvaara

Emotion Emotional tones in scientific writing: comparison of commercially funded studies and non-commercially funded orthopedic studies

240

A N V Steffens, D W G Langerhuizen, J N Doornberg, D Ring, and S J Janssen

Correspondence Antibiotics should not be used for back/leg pain (Acta Orthop 2021; 92 (1): 1-3. doi: 10.1080/17453674.2020.1855561)

244

C Manniche versus P Fritzell, O Hägg, T Bergström, B Jönsson, S G E Andersson, M Skorpil, P M Udby, and M Andersen B C van der Zwaard, W-Y Liu, J Sprengers, N Verschoor, and R P van Hove versus S Donell

Preparation for the next COVID-19 wave: The European Hip Society and European Knee Associates recommendations (Knee Surgery, Sport Traumatol Arthrosc 2020; 28(9): 2747-55. doi: 10.1007/s00167-020-06213-z) Information to authors (see http://www.actaorthop.org/)

247

Acta Orthopaedica 2021; 92 (2): 127–128

127

Guest editorial

Is there a reason to challenge our current practice in children’s forearm fractures?

Sir,—The paper by Laaksonen et al. (2020) in this issue of Acta Orthopaedica shows practice variation and possible overtreatment for a common orthopedic presentation: completely displaced (including overriding) distal radius fractures in children. While there is a considerable “grey zone” in treating these fractures, based on factors such as the age and sex of the child, and the proximity of the fracture to the wrist, there should be better agreement than has been demonstrated in this paper, and what many of us see in clinical practice. We believe that the results presented by Laaksonen and colleagues raise several important questions about orthopedic practice, variation therein, and the management of these fractures in particular. Questions such as: “What is the cause of the variation? ,” “Is better quality evidence for benefit and harms needed for the different treatments?,” and “How can the practice be brought in line with what is best for the patient?” Practice variation occurs where clear guidance is not provided by high-quality evidence. Sometimes practice variation occurs even when high-quality evidence exists (Grove et al. 2016, Lohmander et al. 2016). For the fractures discussed here, there are no published low-risk-of-bias comparative trials that have compared cast-only treatment with closed reduction alone, or closed reduction and percutaneous fixation (Handoll et al. 2018, Zeng et al. 2018). Evidence from such trials would be helpful, and would need to cover the “grey zone” where most practice variation is likely to occur. For example, there may be little practice variation for patients aged less than 3 years, or those close to skeletal maturity, so inclusion of the ages in between and an analysis based on sex and fracture anatomy would be needed. Such a trial would ideally be randomized at an individual level but could also be done using cluster randomization of institutions with crossover, where all eligible patients at each participating institution would be included as part of routine care, first with one treatment, then with the comparator, over two time periods. This would allow fairer comparisons and better generalizability, and would aid with recruitment. If the randomized trial design is deemed not feasible or acceptable, the next best alternative could be a prospective observational study, comparing 2 or more treatment alterna-

tives. This would need to be as carefully designed and reported as a randomized trial, to provide a fair comparison between study groups, but without the randomization. With this design, however, risk of bias due to insufficient matching of study groups and to unknown factors remains high. Related to the question of “what is best for the patient” is what the primary outcome of such a trial should ideally be. Active range of motion? Patient-reported pain and function? Patient-reported overall satisfaction or quality of life? Imaging-based outcome? For interventions regulated by the FDA and EMA, primary outcomes are to be clinically relevant, i.e., how a patient feels, functions, or survives. Blinding is a challenge in surgical trials, but many, perhaps most, endpoint assessments can be observer blinded. We might argue that given that we do not have strong evidence in favor of more invasive management, the current default treatment should be cast immobilization alone, the assumption being similar benefit, less harm, and lower cost. Observational evidence to support treating these fractures without reduction and with excellent outcome was published several years ago (Do et al. 2003, Crawford et al. 2012). However, as the Laaksonen survey shows, surgeons are currently not inclined to treat these fractures without reduction. Several reasons may contribute to this attitude. The bias towards more invasive management may be due to “defensive medicine” where doctors, because of unmotivated fear of poor outcome with nonoperative treatment, try to avoid possible future complaints. Another reason could be bias introduced due to the mentor–trainee guidance in surgical training. The silo environment of a highly specialized community that has not fully adopted the principles of evidence-based surgery may also contribute: “If your only tool is a hammer, you tend to treat everything as a nail.” Surgeons often tend to intervene when evidence is lacking—of operating until the evidence says otherwise—while the opposite should be the rule. A contributing factor that biases treatment towards more invasive options is that doctors, and patients, are susceptible to overestimating the benefits and underestimating the harms of their interventions (Hoffmann and Del Mar 2017). The particular case discussed here is likely no exception. Until further treatment studies have been published it seems prudent to recommend

© 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 10.1080/17453674.2020.1854505

128

nonoperative treatment of these fractures, the burden of proof belonging to those who recommend operative treatment.. The practice of evidence-based surgery in orthopedics faces numerous obstacles (Grove et al. 2016, Robinson et al. 2019, Emara et al. 2020). The randomized controlled trial comparing treatment with a surgical procedure to a treatment without surgery, or even with placebo surgery, while increasing, remains a distinct minority in our field (Lim et al. 2014, Wartolowska et al. 2014, Beard et al. 2020, Harris et al. 2020, Skou et al. 2020). We know from recent examples of orthopedic practice, such as arthroscopic surgery of the knee and shoulder for pain, and vertebroplasty for osteoporotic vertebral fractures, that even if enough scientific evidence were generated to guide practice in this area, the eventual practice change would be a gradual, behavioral one. There is often a lag of many years between evidence being published and a resulting change in practice. Surgeons, like most professions, are insular and look to each other for clues regarding acceptable behavior. While influential surgeons can change practice amongst peers, change still needs to come down to individual surgeons questioning their current beliefs and looking objectively at their current practices within the context of the practice of those around them. The findings of Laaksonen and colleagues should provide a stimulus for low-risk-of-bias comparative studies in this area so that practice can be narrowed to a range that targets the best patient outcomes with the least harm and cost. One such trial was planned and a protocol published (Adrian et al. 2015), but no results appear to have been published. We are encouraged to note that the authors of the survey discussed here have initiated a randomized trial of “Casting Versus Percutaneous Pinning Treatment of Pediatric Overriding Distal Forearm Fractures” (https://clinicaltrials.gov/ct2/show/NCT04323410). We call on surgeons to question their current practice, to participate in planned trials, and to be open to new evidence as it is presented. Evidence-based surgery differs from nonevidence-based surgery in that the former necessitates that judgments are consistent with underlying evidence, while the latter does not. L Stefan LOHMANDER 1 and Ian A HARRIS 2 of Clinical Sciences Lund, Orthopaedics, Lund University, Sweden. 2 South Western Sydney Clinical School, University of New South Wales, and Department of Orthopaedic Surgery, Liverpool Hospital, Liverpool, NSW, Australia. Email: stefan.lohmander@med.lu.se Email: iaharris1@gmail.com 1 Department

Acta Orthopaedica 2021; 92 (2): 127–128

Adrian M, Wachtlin D, Kronfeld K, Sommerfeldt D, Wessel L M. A comparison of intervention and conservative treatment for angulated fractures of the distal forearm in children (AFIC): study protocol for a randomized controlled trial. Trials 2015; 16: 437. DOI 10.1186/s13063-015-0912-x. Beard D J, Campbell M K, Blazeby J M, Carr A C, Weijer C, Cuthbertson B, Buchbinder R, Pinkney T, Bishop F, Pugh J, Cousins S, Harris I, Lohmander L S, Blencowe N, Gillies K, Probst P, Brennan C, Cook A, FarrahHockley D, Savulescu J, Huxtable R, Rangan A, Tracey I, Brocklehurst P, Lee N, Nicholl J, Reeves B, Hamdy F, Rowley S, Cook J S. Considerations and methods for placebo controls in surgical trials: state of the art review and ASPIRE guidance. Lancet 2020; 395: 828-38. Crawford S N, Lorrin S K L, Izuka B H. Closed treatment of overriding distal radial fractures without reduction in children. J Bone Joint Surg Am 2012; 94: 246-52. doi.org/10.2106/JBJS.K.00163. Do T T, Strub W M, Foad S L, Mehlman C T, Crawford A H. Reduction versus remodeling in pediatric distal forearm fractures: a preliminary cost analysis. J Pediatr Orthop B 2003; 12: 109-15. Emara A K, Klika A K, Piuzzi N S. Evidence-based orthopaedic surgery— from synthesis to practice. JAMA Surgery 2020; August 5. doi: 10.1001/ jamasurg.2020.1521. Online ahead of print. Grove A, Johnson R, Clarke A, Currie G. Evidence and the drivers of variation in orthopaedic surgical work: a mixed methods systematic review. Health Syst Policy Res 2016; 3: 6. https://www.hsprj.com/health-maintanance/ evidence-and-the-drivers-ofvariation-in-orthopaedic-surgical-work-amixedmethod-systematic-review.php?aid=8815. Handoll H H G, Elliott J, Iheozor-Ejiofor Z, Hunter J, Karantana A. Interventions for treating wrist fractures in children. Cochrane Database Syst Rev 2018; 12(12): CD012470. DOI: 10.1002/14651858.CD012470.pub2. Harris I A, Sidhu V, Mittal R, Adie S. Surgery for chronic musculoskeletal pain: the question of evidence. Pain 2020; 161 (9, Suppl. 1): S95-S103. Hoffmann T C, Del Mar C. Clinicians’ expectations of the benefits and harms of treatments, screening, and tests: a systematic review. JAMA Intern Med 2017; 177(3): 407-19. doi: 10.1001/jamainternmed.2016.8254. https:// clinicaltrials.gov/ct2/show/NCT04323410 (accessed September 10, 2020). Laaksonen T, Puhakka J, Kosola J, Stenroos A, Ahonen M, Nietosvaara Y. Most surgeons still prefer to reduce overriding distal radius fractures in children. Acta Orthop 2020; (x): 1-5. [Epub ahead of print] doi: 10.1080/17453674.2020.1854502 Lim H C, Adie S, Naylor J M, Harris I A. Randomised trial support for orthopaedic surgical procedures. PLoS One 2014; 9: e96745. Lohmander L S, Thorlund J B, Roos E M. Routine knee arthroscopic surgery for the painful knee of the middle-aged and old: time to abandon ship. Acta Orthop 2016; 87: 2-4. Robinson A H N, Johnson-Lynn S E, Humphrey J A, Haddad F S. The challenges of translating the results of randomized controlled trials in orthopaedic surgery into clinical practice. Bone Joint J 2019; 101-B(2): 121-3. doi:10.1302/0301-620X.101B2.BJJ-2018-1352.R1. Skou S T, Juhl C B, Hare K B, Lohmander L S, Roos E M. Surgical or nonsurgical treatment of traumatic skeletal fractures in adults: systematic review and meta-analysis of benefits and harms Syst Rev 2020; 9(1): 179. doi.org/10.1186/s13643-020-01424-4. Wartolowska K, Judge A, Hopewell S, Collins G S, Dean B J F, Rombach I, Brindley D, Savulescu J, Beard D J, Carr A J. Use of placebo controls in the evaluation of surgery: systematic review. BMJ 2014; 348: g3253. Zeng Z-K, Liang W-D, Sun Y-Q, Jiang P-P, Li D, Shen Z, Yuan L-M, Huang F. Is percutaneous pinning needed for the treatment of displaced distal radius metaphyseal fractures in children? A systematic review. Medicine 2018; 97: 36(e12142).

Acta Orthopaedica 2021; 92 (2): 129–130

129

Perspective

Reducing orthopaedic theatre exposure during the COVID-19 lockdown: does a shift towards virtual reality-based training offer a solution? Adam ARSHAD 1, Amit ZAVERI 2, and Henry ATKINSON 2 1 Department of Emergency Medicine, University College London Hospital; 2 North Middlesex University Hospital, London, UK Correspondence: adam.arshad95@outlook.com Submitted 2020-10-08. Accepted 2020-10-22

Orthopaedic training in the United Kingdom has changed little from the Halstedian apprenticeship model of graduated responsibility, with the mantra “see one, do one, teach one”. Whilst still relevant in surgical teaching, the current and ongoing disruption to surgical training secondary to the coronavirus disease 2019 (COVID-19) outbreak highlights the need for alternative methods of experiential surgical learning, which allow for the development of the knowledge, skills, and attitudes of orthopaedic surgeons, to be sought. Virtual reality-based training (VRBT) involves the trainee independently interacting with a computer-generated simulation of the operative room (OR). Its benefit comes from the deliberate practice of key procedural steps, which the trainee can repeat in the virtual environment before transferring these skills to real patients (Ericsson and Harwell 2019). There is widely published evidence to suggest that repetitive practice is fundamental for orthopaedic surgical training, specifically for arthroscopic surgery and key steps in trauma procedures, e.g., femoral neck guidewire placement for dynamic hip screw fixation (Mabrey et al. 2010, Sadideen et al. 2013, Stirling et al. 2014, Thomas et al. 2014, Gustafsson et al. 2019, Rölfing et al. 2020). In line with such observations, emerging literature evidences improvements in technical aptitude from repeated VRBT (Aim et al. 2016, Bartlett et al. 2018), the eagerness of trainees to practise on the technology (Karam et al. 2013), its potential cost-effectiveness (Bridges and Diamond 1999), and its endorsement from the Royal College of Surgeons (RCS) (RCS 2019). However, in the UK access to this technology is sporadic and not equitable across training regions. Integration appears to have been stifled by a lack of urgency and inflexibility to alter the accepted apprenticeship methods of learning. Prima facie evidence suggests a considerable reduction in the surgical training opportunities during the COVID-19 out-

break. Orthopaedic subspecialty rotations were stopped in March 2020, and trainee involvement in theatres was reduced to make way for consultant-delivered procedures while also making sure that fewer individuals were exposed to these aerosol-generating procedures. Finally, there is an ongoing postponement of elective cases. All this is occurring against a background of many emerging technically challenging techniques (e.g., arthroscopy and navigated surgery) and a reduction in theatre exposure secondary to the European Working Time Directive (Fitzgerald and Caesar 2012). Regardless of these disruptions, the Joint Committee on Surgical Training (JCST) surgical curriculum in the UK has maintained the minimum number of certain index procedures that are deemed crucial for all orthopaedic surgeons to formally complete their training. Adaptive solutions are required to maintain the holistic development of orthopaedic registrars (Kelc et al. 2020). Compared with alternative methods of experiential, non-OR learning, which include cadaver-based dissection and physical mannikins, high-fidelity VRBT provides realistic 3D anatomy and haptic feedback to truly mimic the OR environment. Its use is flexible to the time of the trainee and cases are changeable to the learning requirements of the learner (Vaughan et al. 2016). This has been extensively referenced within the literature (Aim et al. 2016, Bartlett et al. 2018). However, transitioning experiential learning into a virtual platform is not without its disadvantages. The instructivist pedagogy of VR, with an absence of crucial personal interactions, runs against experiential learning theories, which highlight the importance of peer and multidisciplinary learning. The loss of this sensitive interaction renders VRBT incapable of developing skills that centre on communication and analysis, key requirements of the surgical curriculum (ISCP: Intercollegiate surgical curriculum programme 2020). We must

130

therefore consider that VRBT cannot truly fully replace the “hands on” learning of the OR but could become integrated as an adjunct to theatre learning, for example with femoral neck guidewire placement. This would be supplemented with self-reflection and reflective practice, integral aspects for the professional development of orthopaedic trainees (Cruess 2006). Altogether, given these aforementioned deficiencies, the future development of this technology could see the incorporation of “group” VRBT, whereby individuals could communicate and interact together on a case to truly mimic the multidisciplinary nature of the theatre environment. Altogether, the future of orthopaedic teaching is evolving, and technology must be at the centre of this change. With the current relative lack of theatre time for orthopaedic trainees as a result of the COVID-19 pandemic, and the possibility of futures waves, the role of VBRT will become increasingly obvious as a safe, reproducible, adjunctive, and cost-effective way of developing and maintaining surgical training. As we emerge from this pandemic, let us not reach back to the normal, but instead reach out for the better, adapt our practices, and bring orthopaedic learning into the 21st century. The authors declare no conflicts of interest Aim F, Lonjon G, Hannouche D, Nizard R. Effectiveness of virtual reality training in orthopaedic surgery. Arthroscopy 2016; 32: 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: 559-65. Bridges M, Diamond D L. The financial impact of teaching surgical residents in the operating room. Am J Surg 1999; 177: 28-32.

Acta Orthopaedica 2021; 92 (2): 129–130

Cruess R L. Teaching professionalism: theory principles and practices. Clin Orthop Relat Res 2006; 449: 177-85. Ericsson K A, Harwell K W. Deliberate practice and proposed limits on the effects of practice on the acquisition of expert performance: why the original definition matters and recommendations for future research. Front Psychol 2019; 10: 2396. Fitzgerald J E, Caesar B C. The European Working Time Directive: a practical review for surgical trainees. Int J Surg 2012; 10: 399-403. Gustafsson A, Pedersen P, Romer T B, Viberg B, Palm H, Konge L. Hipfracture osteosynthesis training: exploring learning curves and setting proficiency standards. Acta Orthop 2019; 90: 348-53. ISCP. The syllabus, intercollegiate surgical curriculum programme: ICSP; 2020. Karam M D, Pedowitz R A, Natividad H, Murray J, Marsh J L. Current and future use of surgical skills training laboratories in orthopaedic resident education: a national survey. J Bone Joint Surg Am 2013; 95: e4. Kelc R, Vogrin M, Kelc J. Cognitive training for the prevention of skill decay in temporarily non-performing orthopedic surgeons. Acta Orthop 2020; 91 (5): 523-6. Mabrey J D, Reinig K D, Cannon W D. Virtual reality in orthopaedics: is it a reality? Clin Orthop Relat Res 2010; 468: 2586-91. Royal College of Surgeons. Commission on the future of surgery. London: RCS; 2019. Rölfing J D, Jensen R D, Paltved C. Hipsim: hip fracture surgery simulation utilizing the learning curve-cumulative summation test (LC-CUSUM). Acta Orthop 2020; 91(6): 669-74. Sadideen H, Alvand A, Saadeddin M, Kneebone R. Surgical experts: born or made? Int J Surg 2013; 11: 773-8. Stirling E R B, Lewis T L, Ferran N A. Surgical skills simulation in trauma and orthopaedic training. J Orthop Surg Res 2014; 9: 126. Thomas G W, Johns B D, Marsh J L, Anderson D D. A review of the role of simulation in developing and assessing orthopaedic surgical skills. Iowa Orthop J 2014; 34: 181-9. Vaughan N, Dubey V N, Wainwright T W, Middleton R G. A review of virtual reality based training simulators for orthopaedic surgery. Med Eng Phys 2016; 38 59-71.

Acta Orthopaedica 2021; 92 (2): 131–136

131

Reliability of recommendations to reduce a fracture of the distal radius Emily Z BOERSMA 1, Joost T P KORTLEVER 2, Maria W G NIJHUIS-VAN DER SANDEN 3, Michael J R EDWARDS 1, David RING 2, and Teun TEUNIS 4 1 Radboud

University Medical Center, Radboud Institute for Health Sciences, Department of Surgery, Nijmegen, the Netherlands; 2 Department of Surgery and Perioperative Care, Dell Medical School, The University of Texas at Austin, USA; 3 Radboud University Medical Center, Radboud Institute for Health Sciences, Department of IQ Healthcare, Nijmegen, the Netherlands; 4 Department of Plastic Surgery, University Medical Center Utrecht, Utrecht, The Netherlands Correspondence: david.ring@austin.utexas.edu Submitted 2020-05-02. Accepted 2020-10-17.

Background and purpose — It is unclear what degree of malalignment of a fracture of the distal radius benefits from reduction. This study addressed the following questions: (1) What is the interobserver reliability of surgeons concerning the recommendation for a reduction for dorsally displaced distal radius fractures? (2) Do expert-based criteria for reduction improve reliability or not? Methods — We sent out 2 surveys to a group of international hand and fracture surgeons. On the first survey, 80 surgeons viewed radiographs of 95 dorsally displaced (0° to 25°) fractures of the distal radius. The second survey randomized 68 participants to either receive or not receive expertbased criteria for when to reduce a fracture and then viewed 20 radiographs of fractures with dorsal angulation between 5° and 15°. All participants needed to indicate whether they would advise a reduction or not. Results — In the 1st study, the interrater reliability of advising a reduction was fair (kappa 0.31). Multivariable linear regression analyses indicated that each additional degree of dorsal angulation increased the chance of recommending a reduction by 3%. In the 2nd study, reading criteria for reduction did not increase interobserver reliability for recommending a reduction. Interpretation — There is notable variation in recommendations for reduction that is not accounted for by surgeon or patient factors and is not diminished by exposure to expert criteria. Surgeons should be aware of their biases and develop strategies to inform patients and share the decision regarding whether to reduce a fracture of the distal radius.

Many aspects of distal radius fracture (DRF) management are debated (Koval et al. 2014, Mauck and Swigler 2018). For example, it is unclear what degree of malalignment of a DRF benefits from reduction (Mackenney et al. 2006, Dario et al. 2014). Some guidelines address criteria for the adequacy of a reduction, but there is less written about recommendations for when to reduce a fracture. The Clinical Practice Guideline from the American Academy of Orthopedic Surgeons and the Dutch guideline, for example, do not address when to reduce a fracture (AAOS 2009, Brink et al. 2010). The international distal radius fracture study group suggested the following criteria for offering a reduction: dorsal tilt of the articular surface on the lateral radiograph more than 10°, intra-articular displacement more than 2 mm, ulnar positive variance of more than 5 mm, and ulnar-ward inclination of the articular surface on the posteroanterior radiograph less than 15° (Nelson 2006). However, these recommendations are based on little data. In a survey study in the Netherlands, there was limited intersurgeon agreement on recommendations for treatment after reduction of a fracture of the distal radius after introducing a national guideline (Pijls et al. 2016). Understanding the sources of practice variation is the first step towards reducing unhelpful and unwarranted variation. Unwarranted practice variation indicates the need to optimize the balance of benefits and harms, while limiting unhelpful use of resources (Birkmeyer et al. 2013, Saving et al. 2018). In order to learn from variation, we first need to know what drives it. This study addresses the interobserver reliability of surgeons recommending a reduction of a DRF and whether reading expert-based criteria before advising a reduction has an effect on the reliability of advising a reduction. The following questions were addressed: (1) what is the interobserver reli-

132

ability of the recommendation to reduce a DRF? (2) Is there a difference in interobserver reliability based on surgeon characteristics, fracture types, and patient characteristics? (3) What radiographic factors and patient characteristics are independently associated with recommending a reduction? And finally (4), do expert-based criteria increase interobserver reliability?

Methods Study design The 1st survey addressed recommendation for reducing a DRF. The 2nd tested the influence of expert-based criteria for recommending a reduction for a DRF. Surveys were created and distributed through SurveyMonkey (Palo Alto, CA, USA). Participants Members (surgeons) of the Science of Variation Group (SOVG) were invited to participate in the 1st study and 2 months later they were then invited to participate in the 2nd study. The SOVG is an international collaboration of orthopedic surgeons, plastic surgeons, and fracture surgeons that studies variation in the definition, interpretation, classification, and treatment of human illness (https://sites.google.com/site/ scienceofvariationgroup/home). Only a subset participates regularly in the surveys, and even regular participants respond to surveys only in their region of expertise, so it is not possible to measure a meaningful response rate. For these studies we invited surgeons who are specialized in upper extremity and hand surgery. Recommendations for reduction For the survey regarding recommendations for reduction, we selected 95 consecutive radiographs from patients with a DRF treated in the Radboud UMC, Nijmegen and Massachusetts General Hospital, in the first 6 months of 2018. Inclusion criteria for the radiographs were: patients aged between 18 and 90 years old; fracture classification AO Types A and C fractures; fractures with a dorsal angulation of the articular surface on the lateral radiograph close to threshold for acceptable alignment (between 0° and 25°), and good-quality radiographs (as measured by EB according to standardized methods) (Medoff 2005). Radiographs were classified by EB and checked by DR. Because we wanted to study the full spectrum of dorsal angulation, we included 23 radiographs (25%) with dorsal angulation between 0° and 5°, 48 radiographs (50%) between 6° and 15° dorsal angulation, and 24 radiographs (25%) between 16° and 25° of dorsal angulation. For each fracture a posteroanterior and lateral radiograph was presented. Radiographs at both institutions were taken according to similar positioning guidelines. When studying interobserver variability, the study’s power is determined by the number of observers and the number of images. After a certain number of raters, power no longer

Acta Orthopaedica 2021; 92 (2): 131–136

Table 1. Surgeon and clinical characteristics, study A. Values are count unless otherwise specified Surgeon characteristics (n = 80) Men 72 Continent of practice United States 51 Europe 17 Other 12 Years in practice 0–5 28 6–10 17 11–20 25 21–30 10 Supervising trainees 64 Main specialty Hand and wrist 62 Shoulder and elbow 15 Other 3 Clinical characteristics (n = 95) Age a 61 (14) Men 18 AO classification 23-A 39 23-B 0 23-C 56 a Dorsal angulation 11 (6.6) AP distance a 21 (2.5) Radial height a 7.4 (2.5) Ulnar-ward inclination a 12 (4.7) a Ulnar variance –0.1 (2.0) a Mean

(SD)

increases; power can then only increase by rating more images. To make sure every rater did not have to review 95 radiographs we divided our 95 radiographs into 4 sets of 23 or 24 radiographs. Every participant was then randomized to 1 of the 4 survey sets with 23 to 24 radiographs. Members of the SOVG were randomized to 1 of the 4 surveys each with 23 or 24 fractures. All observers were asked to indicate whether they would advise a reduction of a fracture of the distal radius. Every set of radiographs was accompanied by patient age and gender. Information on the criteria for selecting the radiographs was not added. 80 surgeons completed the survey, 72 were men, and the majority resided in Europe (n = 17) and North America (n = 51) (Table 1). 62 Surgeons were hand and wrist surgeons. Most of the surgeons supervised trainees (Table 1). Using a sample size calculation for Fleiss kappa, we calculated that a minimum of 94 images would allow us to find a kappa of 0.60 with a 95% confidence interval of 0.10 (half width), alpha set at 0.05 and 20 raters, assuming a proportion of 0.50 positive ratings for the recommendation for reduction study. Because we divided the available radiographs into 4 sets, we would need 80 observers. Influence of expert-based criteria on recommendations for a reduction For the study addressing the influence of expert-based criteria

Acta Orthopaedica 2021; 92 (2): 131–136

133

Table 2. Surgeon characteristics, study B. Values are count Without With Total guideline guideline n = 68 n = 38 n = 30

p-value

Men 63 35 28 1.0 Continent of practice 0.3 United States 43 21 22 Europe 13 9 4 Other 12 8 4 Years in practice 0.5 0–5 17 10 7 6–10 16 11 5 11–20 25 11 14 21–30 10 6 4 Supervising trainees 56 30 26 0.5 Main specialty 0.6 Hand and wrist 53 28 25 Shoulder and elbow 13 9 4 Other 2 1 1

on recommendation for a reduction, we selected 20 consecutive radiographs between November 2017 and February 2018 treated in the Radboud UMC, Nijmegen. Inclusion criteria for the radiographs were: patients aged between 18 and 90 years old, fracture classification AO types A and C fractures, fractures with a dorsal angulation near the threshold of acceptable alignment (dorsal angulation of 5 to 15 degrees), and goodquality radiographs. Radiographs were measured and classified by (EB [researcher]) and checked by a hand surgeon (DR). We included 5 radiographs with dorsal angulation between 5° and 7.5°, 10 radiographs between 7.6° and 12.5°, and 5 radiographs between 12.6° and 15°. All radiographs included a posteroanterior and lateral view of the fractured distal radius. The criteria for when to reduce a DRF were expert based as there are no validated criteria for indication of a reduction after a DRF. The expert-based criteria were chosen by a panel of three trauma surgeons specialized in the upper extremity, and are based on the AAOS criteria for adequacy of DRF alignment. The criteria are: dorsal tilt of more than 10°, ulnar positive variance of more than 3 mm, radial inclination of less than 15°, intra-articular displacement if the fracture is intra-articular. To investigate the influence of the expert-based criteria, the raters were randomized into 2 groups. Both groups received a survey with 20 sets of radiographs. One group also received the expert-based criteria for reduction. All observers were asked to indicate whether they would advise reduction of a DRF. Every set of radiographs was accompanied by patient age and sex. 68 surgeons completed the survey, 63 were men, and the majority were resident in Europe (n = 13) and North America (n = 43) (Table 2). 53 surgeons were hand and wrist surgeons. Most of the surgeons undertook supervision of trainees (Table 2). Assuming we needed a similar number of observers for the other study (influence of expert-based criteria on recommen-

dation for reduction), 20 sets of radiographs would allow us to determine kappa with a 95% confidence interval of 0.18 (half width). Statistics Continuous variables are described with means and standard deviations and categorical variables with absolute numbers. We used the Fleiss kappa to assess the reliability (i.e., interobserver agreement) of advising a reduction for the DRF. We regarded non-overlapping 95% confidence intervals as a significant difference. The 95% confidence intervals were determined by bootstrapping (number of resamples: 1,000). Kappa values were interpreted using the classification of categorical data by Landis and Koch: a value of 0.01 to 0.20 indicates slight agreement; 0.21 to 0.40 fair agreement; 0.41 to 0.60 moderate agreement; 0.61 to 0.80 substantial agreement; and 0.81 to 0.99 near perfect agreement. To determine factors associated with the likelihood of reduction we divided the proportion of recommended reductions by the total number of recommendations for each radiograph. We created a multivariable linear regression model with the likelihood for reduction as the dependent variable and the radiographic factors and patient characteristics as the independent variables. Ethics, funding, and potential conflicts of interest This study received approval from the Institutional Review Board of the University of Texas at Austin, 2017-11-0081. The authors received no financial support for the research, authorship, and/or publication of this article. All the authors declare no conflict of interest related to this study.

Results Interobserver reliability for recommending a reduction Recommendation for reduction had fair interobserver reliability (kappa 0.3, 95% CI 0.2–0.4). Surgeons characteristics, age and sex of the patient, and fracture characteristics had no influence on the interobserver reliability (Table 3). Factors associated with recommending reduction Multivariable linear regression analyses indicated that each additional degree of dorsal angulation increased the chance of recommending a reduction by 3% (beta 0.03, CI 0.02–0.03, p-value < 0.001) (Table 4, see Supplementary data). Dorsal angulation explained 37% of the variation in the likelihood of recommending a reduction (semi-partial R2 0.4). Influence of expert-based criteria Expert-based criteria for reduction did not increase the interobserver reliability for recommending a reduction (no criteria kappa 0.4, CI 0.3–0.6 vs. criteria 0.5, CI 0.3–0.6) (Table 5, see Supplementary data).

134

Acta Orthopaedica 2021; 92 (2): 131–136

Table 3. Interobserver agreement (95% confidence interval) on recommending reduction Factor

Kappa (95% CI)

Surgeon variables Overall Sex Female Male Continent of practice USA Europe Other Years of practice: 0–5 6–10 11–20 21–30 Supervising trainees Yes No Main specialty Hand and wrist Shoulder and elbow Other Set Set 1 Set 2 Set 3 Set 4 Patient variables Sex Female Male Age 18–60 > 60 AO Classification 23-A 23-C Dorsal angulation (°) 0–5 6–15 16–25 AP distance (mm) 16–21 22–29 Radial height (mm): 1–7.4 7.5–13 Radial inclination (°) 1–11.6 > 11.6–23.6 Ulnar variance (mm) –4.5 to –0.8 > –0.8–5.4 a Statistically

0.31 (0.23–0.39) 0.49 (0.29–0.69) 0.30 (0.22–0.37) 0.28 (0.19–0.37) 0.37 (0.27–0.47) 0.45 (0.29–0.62) 0.36 (0.25–0.48) 0.28 (0.16–0.40) 0.20 (0.12–0.27) 0.36 (0.18–0.54) 0.30 (0.22–0.37) 0.45 (0.32–0.59) 0.29 (0.21–0.37) a 0.42 (0.28–0.56) 0.76 (0.42–1.10) a 0.28 (0.20–0.39) 0.32 (0.12–0.52) 0.31 (0.20–0.42) 0.24 (0.11–0.36) 0.31 (0.22–0.41) 0.27 (0.14–0.40) 0.32 (0.21–0.43) 0.28 (0.18–0.38) 0.40 (0.26–0.53) 0.21 (0.15–0.28) 0.14 (0.04–0.24) 0.22 (0.12–0.32) 0.12 (0.06–0.18) 0.38 (0.27–0.50) 0.21 (0.14–0.29) 0.25 (0.12–0.37) 0.32 (0.21–0.42) 0.25 (0.14–0.37) 0.31 (0.23–0.39) 0.34 (0.23–0.45) 0.26 (0.18–0.34)

significant difference.

Discussion Surgeon biases, habits, and preferences contribute to variations in care. In the face of limited evidence, attitudes and beliefs concerning the indications for treatment are important reasons for surgical variation. Reducing unwarranted practice

variation could lead to a reduction in avoidable morbidity and unhelpful use of resources (Birkmeyer et al. 2013). Our study addressed the interobserver reliability of surgeons recommending a reduction of a DRF and the influence of reading expert-based criteria influenced recommendations. This study was not intended to determine a threshold for when to reduce a distal radius fracture. We acknowledge some limitations for the study. 1st, only 68 surgeons completed the influence of expert-based criteria study, and it might have been underpowered. We were close to our estimate, and power analysis for reliability studies is imperfect. Therefore, we do not think this had much influence on the study. 2nd, members of the SOVG are more likely to work in an academic setting than the average surgeon, which could decrease generalizability. 3rd, the SOVG group does not measure intraobserver reliability because it is always greater than interobserver reliability. 4th, our use of dorsal angulation (rather than ulnar variance or other criteria) as selection criteria, might have influenced the regression analysis. A similar study with different radiographic parameters as selection criteria might have slightly different results. 5th, measurements of the radiographic parameters were performed by one researcher and checked by the senior surgeon. Measurements are somewhat imprecise, but it is unlikely that there was any systematic bias and the random variations probably had little influence on the statistics. 6th, we did not ask the observers whether they would advise surgery or not. There may be a subset for which surgical considerations would alter recommendations for reduction, but that subset is likely to be small. Finally, there are important differences between a survey and actual practice. For example, the radiographs were accompanied only by age and sex. In actual practice more factors are important to determine whether a fracture needs reduction, for example patient occupation and hand dominance. In our opinion the relative simplification would be expected to reduce variability. The fair interobserver reliability on whether to reduce a DRF or not (kappa 0.31, CI 0.23–0.39) is consistent with prior studies demonstrating notable variation and limited reliability in surgeon recommendations. The study by Tosti et al. (2014) found that the interobserver agreement for recommending treatment of little finger metacarpal neck fractures was fair. Until we have a better evidence base on whether to reduce a fracture or not, it may be worthwhile to invest in tools such as decision aids to help patients weigh the advantages and disadvantages and participate in the decision. The observation that each additional degree of dorsal angulation increased the chance of recommending a reduction by 3% is consistent with our observation that surgeons often use dorsal angulation as the most important feature used to decide on whether to recommend reduction of a fracture. This is consistent with studies suggesting that dorsal angulation is one form of deformity after fracture that affects function measured with patient reported outcome measures (PROMs) such as the Disability of the Shoulder and Hand (DASH) score (McQueen

Acta Orthopaedica 2021; 92 (2): 131–136

and Caspers 1988, Gliatis et al. 2000, Wilcke et al. 2007, Ali et al. 2018). In a prior study similar to ours, radiographic parameters accounted for about half of the variation in treatment recommendations (Neuhaus et al. 2015). Other studies also show that patient factors such as male sex and age and fewer comorbidities did not explain any more of the variation in the treatment recommendation than radiographic factors alone (Mackenney 2006, Kodama et al. 2013). The observation that expert-based criteria did not influence the interobserver reliability for recommending a reduction is consistent with prior reliability studies. For instance, exposure of observers to a description of staging of wrist arthritis related to scaphoid nonunion did not improve reliability of staging (ten Berg et al. 2017). This study and the study of Christensen et al. (1981) proposed that the interpretation of radiographs with a particular pathology involves the learned concept for what is normal and not normal, meaning that surgeons see what they already know or believe. Using additional guidance or knowledge could be less effective than expected due to the influence of cognitive biases such as anchoring and familiarity (Christensen et al. 1981, ten Berg et al. 2017). For instance, one study found that personality features influence treatment recommendations; a higher pioneer score (associated with innovation and creativity) was associated with a higher rate of recommendation for surgery (Teunis et al. 2015). This surgeon-characteristic influence on recommendation for treatment could have influenced the potential additional value of the expert-based criteria and may also be an explanation for observer variation. Further research should be conducted to investigate whether the expert-based criteria and especially dorsal angulation could increase the reliability among young residents and eventually limit practice variation. In conclusion, limited interobserver reliability contributes to practice variation. There was notable variation in recommendations for reduction that was not accounted for by surgeon or patient factors and was not diminished by exposure to expert criteria. Dorsal angulation was the main driver for recommending a reduction, reflecting the fact that surgeons may focus on a few, relatively simplistic factors in making recommendations. The ability to learn from practice variation is hindered by notable variability that is, to date, unaccounted for by measured factors. Future studies might address surgeon cognitive bias and heuristics and the best methods for nudging people (patients and surgeons) toward evidence and what matters most to patients. In addition, future studies should assess the importance of the effect of dorsal angulation on deciding when to reduce a DRF and the influence on patient-reported outcome after DRF. Supplementary data Tables 4 and 5 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/1745 3674.2020.1846853

135

EB, JK, MN, ME, DR, and TT conducted and planned the study and designed the survey. EB, JK, DR, and TT helped analyze the results. EB, JK, MN, ME, DR, and TT contributed to the interpretation of the results. EB took the lead in writing the manuscript. All authors provided critical feedback and helped shape the research, analysis and manuscript Acta thanks Hebe Désirée Kvernmo and Mats Wadsten for help with peer review of this study. AAOS. Guideline on The Treatment of Distal Radius Fractures. Rosemont, IL: AAOS; 2009. Ali M, Brogren E, Wagner P, Atroshi I. Association between distal radial fracture malunion and patient-reported activity limitations: a long-term follow-up. J Bone Joint Surg Am 2018; 100(8): 633-9. doi: 10.2106/ JBJS.17.00107. Birkmeyer J D, Reames B N, McCulloch P, Carr A J, Campbell W B, Wennberg J E. Understanding of regional variation in the use of surgery. Lancet 2013; 382(9898): 1121-9. doi: 10.1016/S0140-6736(13)61215-5. Brink P R G, Bransz N, Deijkers R L M, van Eerten P V, Kolkman S, van Loon J, et al. Guidelines distal radius fractures, diagnosis and treatment. 2010. http:// www.heelkunde.nl/uploads/h8/VM/h8VMOlRX83oW2NSR_5rTWw/ Richtlijn_Distale_radius_fracturen_definitieve_versie_0511.pdf.] Christensen E E, Murry R C, Holland K, Reynolds J, Landay M J, Moore J G. The effect of search time on perception. Radiology 1981; 138(2): 361-5. doi: 10.1148/radiology.138.2.7455116. Dario P, Matteo G, Carolina C, Marco G, Cristina D, Daniele F, Andrea F. Is it really necessary to restore radial anatomic parameters after distal radius fractures? Injury 2014; 45(Suppl. 6): S21-6. doi: 10.1016/j. injury.2014.10.018. Gliatis J D, Plessas S J, Davis T R. Outcome of distal radial fractures in young adults. J Hand Surg Br 2000; 25(6): 535-43. doi: 10.1054/jhsb.2000.0373. Kodama N, Imai S, Matsusue Y. A simple method for choosing treatment of distal radius fractures. J Hand Surg Am 2013; 38(10): 1896-905. doi: 10.1016/j.jhsa.2013.07.009. Koval K, Haidukewych G J, Service B, Zirgibel B J. Controversies in the management of distal radius fractures. J Am Acad Orthop Surg 2014; 22(9): 566-75. doi: 10.5435/JAAOS-22-09-566. Mackenney P J, McQueen M M, Elton R. Prediction of instability in distal radial fractures. J Bone Joint Surg Am 2006; 88(9): 1944-51. doi: 10.2106/ JBJS.D.02520. Mauck B M, Swigler C W. Evidence-based review of distal radius fractures. Orthop Clin North Am 2018; 49(2): 211-22. doi: 10.1016/j.ocl.2017.12.001. McQueen M, Caspers J. Colles fracture: does the anatomical result affect the final function? J Bone Joint Surg Br 1988; 70(4): 649-51. Medoff R J. Essential radiographic evaluation for distal radius fractures. Hand Clin 2005; 21(3): 279-88. doi: 10.1016/j.hcl.2005.02.008. Nelson D. Eradius 2006. Neuhaus V, Bot A G, Guitton T G, Ring D C. Influence of surgeon, patient and radiographic factors on distal radius fracture treatment. J Hand Surg Eur 2015; 40(8): 796-804. doi: 10.1177/1753193414555284. Pijls B G, Ritchie E D, van den Bremer J, Brink R, Nelissen R G, Clinical Practice G. Management of fractures of the distal radius 4 years after the introduction of national guidelines. J Hand Surg Eur 2016; 41(7): 777-9. doi: 10.1177/1753193416632142. Saving J, Ponzer S, Enocson A, Mellstrand Navarro C. Distal radius fractures: regional variation in treatment regimens. PLoS One 2018; 13(11): e0207702. doi: 10.1371/journal.pone.0207702. ten Berg P W L D, Drijkoningen T, Guitton T G, Ring D. Does a comparison view improve the reliability of staging wrist osteoarthritis? Hand 2017; 12(5): 439-45. Teunis T, Janssen S J, Guitton T G, Vranceanu A M, Goos B, Ring D. Surgeon personality is associated with recommendation for operative treatment. Hand (NY) 2015; 10(4): 779-84. doi: 10.1007/s11552-015-9755-x.

136

Tosti R, Ilyas A M, Mellema J J, Guitton T G, Ring D, Science of Variation G: interobserver variability in the treatment of little finger metacarpal neck fractures. J Hand Surg Am 2014; 39(9): 1722-7. doi: 10.1016/j. jhsa.2014.05.023.

Acta Orthopaedica 2021; 92 (2): 131–136

Wilcke M K, Abbaszadegan H, Adolphson P Y. Patient-perceived outcome after displaced distal radius fractures: a comparison between radiological parameters, objective physical variables, and the DASH score. J Hand Ther 2007; 20(4): 290-8; quiz 9. doi: 10.1197/j.jht.2007.06.001.

Acta Orthopaedica 2021; 92 (2): 137–142 137

Development of a diagnostic algorithm identifying cases of dislocation after primary total hip arthroplasty—based on 31,762 patients from the Danish Hip Arthroplasty Register Lars L HERMANSEN 1,2,3, Bjarke VIBERG 4,5, and Søren OVERGAARD 2 1 Department of Orthopedics, Hospital of South West Jutland, Esbjerg; 2 The Orthopedic Research Unit, Department of Orthopedic Surgery and Traumatology, Odense University Hospital, Odense, Department of Clinical Research, University of Southern Denmark; 3 OPEN, Odense Patient data Explorative Network, Odense University Hospital, Odense; 4 Department of Orthopedic Surgery and Traumatology, Lillebaelt Hospital, University Hospital of Southern Denmark; 5 Department of Regional Health Research, University of Southern Denmark, Denmark Correspondence: lars.lykke.hermansen@rsyd.dk Submitted 2020-02-23. Accepted 2020-05-06.

Background and purpose — Dislocation of total hip arthroplasties (THA) is often treated with closed reduction and traditionally not registered in orthopedic registers. This study aimed to create an algorithm designed to identify cases of dislocations of THAs with high sensitivity, specificity, and positive predictive value (PPV) based on codes from the Danish National Patient Register (DNPR). Patients and methods — All patients (n = 31,762) with primary osteoarthritis undergoing THA from January 1, 2010 to December 31, 2014 were included from the Danish Hip Arthroplasty Register (DHR). We extracted available data for every hospital contact in the DNPR during a 2-year follow-up period, then conducted a comprehensive nationwide review of 5,096 patient files to register all dislocations and applied codes. Results — We identified 1,890 hip dislocations among 1,094 of the included 31,762 THAs. More than 70 different diagnoses and 55 procedural codes were coupled to the hospital contacts with dislocation. A combination of the correct codes produced a sensitivity of 63% and a PPV of 98%. Adding alternative and often applied codes increased the sensitivity to 91%, while the PPV was maintained at 93%. Additional steps increased sensitivity to 95% but at the expense of an unacceptable decrease in the PPV to 82%. Specificity was, in all steps, greater than 99%. Interpretation — The developed algorithm achieved high and acceptable values for sensitivity, specificity, and predictive values. We found that surgeons in most cases coded correctly. However, the codes were not always transferred to the discharge summary. In perspective, this kind of algorithm may be used in Danish quality registers.

Dislocation is a feared complication after total hip replacement. To prevent this and other complications, we often have to take advantage of and rely on the enormous amount of data from the many orthopedic registries in use rather than conducting large and costly clinical studies (SHAR 2017, AOANJRR 2018, DHR 2019, Varnum et al. 2019a, b). However, data within these registries are not always representative of the actual occurrence of complications. Gundtoft et al. (2015) demonstrated that infections are underreported by 40%. Likewise, a recent study based on a Danish cohort and administrative registers found that the sensitivity was only 63% when patients with dislocations were identified through a combination of the correct diagnosis and procedure code, ultimately missing more than one-third of the patients (Hermansen et al. 2020). The treatment of choice after reduction of the first hip dislocation is nonoperative, unless there is obvious malpositioning of the inserted components causing instability. Revisions are often performed only after several dislocations (Patel et al. 2007, Devane et al. 2012, Saiz et al. 2019). Therefore, a large group of patients treated with closed reduction are never registered in arthroplasty registers and the true burden of this complication remains uncertain. Ideally, a healthcare system should be able to capture all important complications that have an impact on the patient and the treatment quality. This study aimed to create an algorithm to identify dislocations of THAs with high sensitivity, specificity, and predictive values based on codes from a national health care system.

© 2021 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 10.1080/17453674.2020.1868708

138

Acta Orthopaedica 2021; 92 (2): 137–142

Patients and methods Study design We used data that was collected during a recent retrospective cohort study which used prospectively collected data from the Danish Hip Arthroplasty Register (DHR) designed to find the true frequency of hip dislocation after primary THA (Hermansen et al. 2020). We refer the reader to this paper for study details and will only report the main aspects in this article. The RECORD guidelines were followed.

THAs performed 2010–2014 for osteoarthritis from the Danish Hip Arthroplasty Register n = 36,693 Excluded (n = 4,931): – contralateral primary THA, 3,500 – secondary arthritis, 446 – constrained liner, 392 – missing/incorrect laterality, 270 – revisions, 181 – age < 40 years, 110 – incorrect date of surgery, 32 Eligible THAs n = 31,762

Healthcare contacts within 2 years of surgery registered in the Danish National Patient Register n = 16,437 Excluded (n = 10,450): – contact regarding contralateral THA, 4,191 – planned outpatient contacts, 2,961 – contact after revision date, 1,162 – duplet contacts, 994 – wrong diagnosis, 756 – contact at private clinics, 386 Relevant healthcare contacts n = 5,987 Genuine dislocations n = 1,166

Possible dislocations n = 4,821

Missing laterality or > 1 procedure per contact n = 275 Patient files reviewed n = 5,096 Verified by patient files, 99.3% Verified by radiographs, 0.7%

Participants (Figure 1) We identified all patients with primary osteoarthritis (OA) who underwent a THA from January 1, 2010 to December 31, 2014 and followed each patient for 2 years after index surgery. Follow-up was ended after 2 years or before if revision surgery, emigration, or death occurred, whichever came first. We excluded THAs inserted for indications other than primary OA. For the same reason, patients younger than 40 years of age were excluded (Duffy et al. 2001, Ellison et al. 2006). Any contralateral THA procedures during the inclusion period was also omitted to avoid dependency among observations (Ranstam and Robertson 2010). Data sources and data cleaning Dislocations, together with any other type of patient contact with the Danish healthcare system, are registered in the Danish National Patient Register (DNPR) (Lynge et al. 2011). By means of the DNPR, we were able to extract information for every hospital contact with orthopedic and non-orthopedic departments as well as outpatient emergency room contacts for each patient during the individual 2-year follow-up period. We extracted the admission and discharge date, the date of any surgical procedure, and hospital and department names for all hospital contacts that had been assigned any primary or secondary hip or dislocation related diagnostic or procedural code (see Appendix for the complete list). The DNPR completeness is over 99%, and we did not encounter any missing data regarding diagnoses and procedure codes in our population (Schmidt et al. 2015). The diagnostic codes were extracted from the International Classification of Diseases, 10th revision (ICD-10) and procedural codes were derived from the Danish version of the Nordic Medico-Statistical Committee’s (NOMESCO) Classification of Surgical Procedures (NCSP). We then established the following classification of contacts: 1. Genuine dislocations: Contacts assigned a combination of the correct diagnostic (DT84.0(A)) and surgical procedure (KNFH20) codes. 2. Possible dislocations: Any contact not included as a genuine dislocation. A comprehensive review of all patient files meeting criterion 2 was performed to identify every miscoded dislocation. We also reviewed 20% of the genuine dislocation cases to validate the combination of correct codes. 5,096 patient files were manually reviewed, and all dislocations and the applied codes were registered.

Excluded (n = 4,114): – contacts without hip dislocation, 4,009 – contacts with dislocation of contralateral THA, 31 – double contact (department transfer), 74 Contacts with known laterality and 1 procedure per contact n = 891

Contacts with hip dislocation n = 982 (999 dislocations)

1,890 hip dislocations in 1,094 THAs

Statistics We designed the algorithm using a stepwise approach and calculated the sensitivity, specificity,

Figure 1. Flowchart overview of DHR (upper part) and DNPR (lower part) data cleaning and the following patient file review to identify the true frequency of dislocations. The dotted line indicates that 16,437 hospital contacts were found in the DNPR for the 31,762 included THAs.

Acta Orthopaedica 2021; 92 (2): 137–142 139

Table 1. Description of the 5 groups of diagnostic combinations used in the algorithm Group Codes

Description

Group 1 DT840(A) + Combination of correct diagnosis and KNFH20 procedure code (with identified laterality in DNPR) Group 2 KNFH20 Correct procedure code alone combined with any random diagnosis (with identified laterality in DNPR) Group 3 DS730 Alternative and often used diagnosis and KNFH00 procedure codes (with identified laterality KNFH02 in DNPR) KNFH21 KNFH22 Group 4 Group 1–3 All group 1–3 cases, where laterality is uncertain in DNPR Group 5 DT840(A) Correct diagnoses alone combined with any random procedure code, AND limited to acute readmissions or emergency room contacts (with identified laterality in DNPR) See Appendix for detailed definitions of diagnostic and procedure codes.

Hansens Foundation, the Danish Rheumatism Association, the A.P. Møller Foundation for the Advancement of Medical Science, the Orthopaedic Fund of West Jutland, and Doctor of Bramming, Grethe Marie Justesens Fund. The University of Southern Denmark and Region of Southern Denmark each assigned a 1-year PhD scholarship. There are no conflicts of interest and none of the funding had any influence on the data material or reporting of the results.

Results We identified 1,890 hip dislocations in 1,094 of the included 31,762 THAs (Figure 1), which yielded a 2-year cumulative incidence of 3.4% (95% CI = 3.3–3.6). More than 70 different main diagnoses and 55 different procedural codes were coupled to the hospital contacts with dislocation. The most common mistake was the application of the correct procedure code in combination with the wrong diagnosis code. Thereafter, more than 10% of all dislocations were found to have codes that are intended to describe dislocation and reduction of traumatic hip dislocation of native hip joints rather than THA. The most frequently used codes and combinations were grouped and contributed to our algorithm (Table 1). Step 1 was a combination of the correct codes (DT840+KNFH20) with known laterality resulting in a sensitivity of 63% and a PPV of 98% (Table 2). When we added the contacts with the correct procedure code alone (KFH20) and alternative and often applied codes in 2 additional steps (DS730, KNFH (21;22;00;02)), all with known laterality, we increased the sensitivity to 85%, while the PPV was 96%. Step 4 added the contacts of the above-mentioned codes from steps 1 through 3 with unknown laterality in the DNPR, which increased the sensitivity to 91% but lowered the PPV to 93%.

and the positive and negative predictive values for various combinations of the most frequently used codes. The steps were not pre-specified but, instead, were chosen based on the codes that had been applied nationwide from 2010 to 2016 for verified dislocations. The plan was to add codes in steps and continuously increase the sensitivity (i.e., the proportion of true positives of all dislocations), while at the same time keeping the specificity (i.e., the proportion of true negatives of all not having a dislocation) and the positive predictive value (PPV) (i.e., probability that patients based on the algorithm truly have the dislocation) high. The algorithm will identify patients with at least 1 episode of dislocation for a given period of time (i.e., the risk of dislocation) but it will not necessarily identify all dislocations for each patient. It is also important to note that there is a clear distinction between hospital Table 2. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for each step of the algorithm identifying dislocations. Values are % with contacts with or without denoted laterality (95% confidence interval) in the DNPR. This is an important aspect in order to distinguish between contralatStep Sensitivity Specificity PPV NPV eral THAs. Statistics was performed with STATA software version 15.0 (StataCorp, Step 1 62.7 (59.8–65.6) 99.9 (99.9–99.9) 97.9 (96.5–98.8) 98.7 (98.6–98.8) Step 2 77.0 (74.4–79.4) 99.9 (99.9–99.9) 96.5 (95.0–97.6) 99.2 (99.1–99.3) College Station, TX, USA). Ethics, funding, and potential conflicts of interest The Danish Patient Safety Authority (3-30132128/1), the Danish Data Protection Agency (2008-58-0035), and the head of local departments approved the review of patient medical records. Funding was obtained from the Clara Hansens Memorial Fund, Appropriation Merchant Sven Hansen & Wife Ina

Step 3 Step 4 Step 4A Step 5 Step 5A

85.1 (82.9–87.2) 91.3 (89.5–92.9) 91.3 (89.5–92.9) 95.4 (93.9–96.5) 95.4 (93.9–96.5)

99.9 (99.8–99.9) 99.8 (99.7–99.8) 99.9 (99.8–99.9) 99.2 (99.1–99.3) 99.8 (99.8–99.9)

96.3 (94.9–97.4) 93.3 (91.6–94.7) 96.5 (95.2–97.6) 81.8 (79.4–83.7) 96.6 (95.4–97.7) a

99.5 (99.4–99.5) 99.7 (99.6–99.7) 99.7 (99.6–99.7) 99.8 (99.8–99.9) 99.8 (99.8–99.9)

Step 1 = Group 1 alone; Step 2 = Group 1+2 etc. For each step an additional group is added to the previous step, thereby including more codes and increasing sensitivity at a cost of decreased specificity and the positive predictive value. The two steps marked (A) indicate how Steps 4 and 5 can achieve an increase in the positive predictive value if the patient files for the hospital contact of the specific group are reviewed and the falsepositives are discarded. See appendix for examples of the review burden. a Assumes patient file review of Step 4.

140

Acta Orthopaedica 2021; 92 (2): 137–142

RISK OF DISLOCATION Sensitivity (A), specificity (B), positive predictive value (C), and negative predictive value (D) for each step in pure register studies

Increase of the positive predictive value and specificity by review of patient files of particular steps a

Step 1 DT840(A)+KNFH20 a,d + Step 2 KNFH20 a,e + Step 3 DS730 a,e KNFH(00;02;21;22) a,e + Step 4 Steps 1–3 b + Step 5 DT840(A) a,c,e

63%

>99%

98%

99%

77%

>99%

97%

>99%

85%

>99%

96%

>99%

91%

>99%

93%

>99%

95%

>99%

82%

>99%

Clear laterality description in DNPR Uncertain laterality description in DNPR Limited to acute readmissions d Combination of codes e Alone, no combination f Percent of total cohort to be reviewed g Including review of Step 4 b c

0.3% f

1.0% f,g

91%

>99%

97%

>99%

95%

>99%

97%

>99%

Step 4A

Step 5A

Figure 2. Development of the algorithm for identifying dislocation following primary THA. Flowchart presenting sensitivity, specificity, and predictive values in the attempt to identify the risk of dislocation in a pre-defined cohort of THA patients. Researchers can then decide on what level of, e.g., sensitivity and positive predictive value is acceptable for their specific study design. Flowchart description: For each step, additional codes are added to the previous step, thereby including more codes and increasing sernsitivity at the cost of decreased specificity and positive predictive value. Step 4 and 5 can achieve an increase in positive predictive value if the patient files describing the hospital contact for the specific group are reviewed and the false positives discarded.

The only way to increase the sensitivity further was to include contacts with the correct diagnosis code alone (DT840). However, this code is often related to many other aspects of prosthesis complications and is not used solely for dislocations. Therefore, in the last step, the sensitivity increased up to 95% but at the expense of an unacceptable decrease in the PPV to 82%. Specificity was in all steps greater than 99%. Steps 4A and 5A shows an achievable increase in the PPV for these 2 steps if the patient files for the particular step are reviewed. The results from Table 2 were combined into a flowchart (Figure 2), which states the achievable values for pure register purposes and highlights the expected burden of patient file review, which is an option in clinical studies.

Discussion We performed a comprehensive nationwide review and validated the applied codes for THA dislocation. We were able to improve the sensitivity by 28% without sacrificing PPV/ specificity using our approach by adding alternative and validated codes.

Accurate measurements and truthful monitoring of specific complications are important in order to decrease the risk of complications. Dislocation is a feared complication after hip replacement, leading to pain, anxiety, and reduced quality of life as well as increased costs in the healthcare system. Moreover, re-dislocations happen in 40% to 68% of patients, increasing the risk of reoperation (Brennan et al. 2012, Hermansen et al. 2020). In our study, an acceptable sensitivity of 91%, a specificity of more than 99%, and a PPV of 93% are achievable when combining the most frequently used codes into an algorithm. The algorithm provides information regarding the expected sensitivity, specificity, and predictive values in a stepwise approach. Importantly, the sensitivity, specificity, and predictive values presented in the algorithm are derived from a specified cohort of primary OA patients. We have not included patients with secondary OA or femoral neck fracture, which is a limitation. However, there is no reason to believe that the results in other such populations would be different than those in the present study. To our knowledge, codes for hip dislocation have never before been validated in a Danish setting. The accuracy of the DNPR for several other diseases has shown both low to moderate completeness (Nymark et al. 2003, Gundtoft et al. 2015, Jorgensen et al. 2016, Kristensen et al. 2019) and high PPV (Viborg et al. 2017), indicating that great variation is introduced by coding personnel in different specialties. Internationally, there is also significant variation documented between professional hospital coders and orthopedic surgeons for both diagnoses and complications, emphasizing the need for caution when analyzing these data (Mears et al. 2002, Mont et al. 2002). There are unique possibilities for unambiguous linkage and complete follow-up of all patient contacts with the Danish healthcare system. Therefore, our study is based on the application of codes from the broadest range of surgeons possible. Our follow-up was not limited to readmissions to orthopedic departments, as we also reviewed all non-orthopedic readmissions and outpatient emergency room contacts. Upon reviewing numerous patient files from admission to discharge, it is our experience that the surgeons in most cases are coding correctly. However, the codes labeled in the surgery descriptions are not transferred to the discharge summary on all occasions, which forms the basis of a patient’s DNPR registration. Instead, codes used by the staff who meet the patient in the emergency department are chosen and, often, these codes are assigned by untrained and younger doctors when the diagnosis is not yet verified. Remarkably, the sensitivity was only 63% when patients with dislocation were identified with a combination of the correct diagnosis, procedure code, and known laterality, thereby missing one-third of the patients (step 1). The majority of the remaining dislocations can be found using either the correct diagnosis or procedure code alone. In particular, the procedure code alone is trustworthy, while the effect of the diagnosis code DT48.0(A) is far more unpredictable. The diagnosis

Acta Orthopaedica 2021; 92 (2): 137–142 141

is applied to any kind of mechanical complication, making it widely used in planned outpatient contacts. This supports the inclusion of several false-positive cases if the diagnosis alone is uncritically used in the algorithm without review of every patient case. We included only acute admissions or emergency room contacts with DT840 alone to keep the PPV as high as possible. With this algorithm, we focused on the risk of dislocation, thus finding all patients with at least 1 dislocation and not necessarily identifying all dislocations for every patient. This is typically useful in larger register settings monitoring complications. In smaller clinical studies with closer follow-up and involvement of patient-reported outcome measures, it may be of greater importance to report all events of dislocation. Our algorithm possesses lower sensitivity for this scenario (step 4: sensitivity = 88% and PPV = 94%; step 5: sensitivity = 95% and PPV = 84%). In steps 1 to 3 and 5 of the algorithm, laterality is known in both DHR (laterality of the THA) and DNPR (laterality of the hospital contact), which is why a mix in laterality of bilateral THA cases is non-existent. In step 4 we included hospital contacts with unknown or uncertain DNPR laterality to increase sensitivity. A decrease in the PPV is therefore obvious and can be managed by review of a few patient files. The developed algorithm based on the ICD-10 and NOMESCO codes achieved a sensitivity of 91% and a PPV at 93% using register data alone, which we consider acceptable. As the rates of missed patients with dislocation and false positivity were almost equal, the algorithm gives a precise measure for the risk of dislocation in this study. Higher sensitivity is possible but at the expense of drastically lowering the PPV, which is not feasible for register studies. In perspective, this algorithm is meant for incorporation in national registers for the reliable registration of dislocations and will be of major importance for monitoring this severe complication. Also, because the settings of both hip registers and coding algorithms in other Nordic countries are similar to the Danish, it would be an obvious recommendation also to validate the algorithms within the Nordic Arthroplasty Register Association (NARA) collaboration. Supplementary data Appendices A–D ((A) a list of the diagnostic ICD-10 codes, (B) the procedural NCSP codes, which were applied in the National Patient Registry, (C) interpretation of correct coding of a THA dislocation, and (D) description of the review burden related to Table 2) are available in the online version of the article, http://dx.doi.org/10.1080/17453674.2020.1868708

LLH, BV, and SO were responsible for the conception of the study; LLH was responsible for data analysis before and after review of patient files and coordinated the national review; LLH drafted the manuscript; BV and SO revised it critically.

The authors would like to thank the heads of departments from the participating hospitals across Denmark. This study would not have been possible without their permission. They also thank the local heads of research and their secretaries, who came across this study and helped to solve local problems as they occurred. Acta thanks Max Gordon and Maziar Mohaddes for help with peer review of this study.

AOANJRR. Australian National Joint Replacemant Registry—Annual Report. Available from: https://aoanjrrsahmricom/annual-reports-2018; 2018. Brennan S A, Khan F, Kiernan C, Queally J M, McQuillan J, Gormley I C, O’Byrne J M. Dislocation of primary total hip arthroplasty and the risk of redislocation. Hip Int 2012; 22(5): 500-4. doi: 10.5301/hip.2012.9747. Devane P A, Wraighte P J, Ong D C, Horne J G. Do joint registries report true rates of hip dislocation? Clin Orthop Relat Res 2012; 470(11): 3003-6. doi: 10.1007/s11999-012-2323-6. DHR. Danish Hip Arthroplasty Register—Annual Report. Available from: “http://danskhoftealloplastikregisterdk/wp-content/uploads/2019/09/ DHR-årsrapport-2019_til-offentliggørelsepdf; 2019. Duffy G P, Berry D J, Rowland C, Cabanela M E. Primary uncemented total hip arthroplasty in patients < 40 years old: 10- to 14-year results using first-generation proximally porous-coated implants. J Arthroplasty 2001; 16(8 Suppl. 1): 140-4. Ellison B, Berend K R, Lombardi A V Jr, Mallory T H. Tapered titanium porous plasma-sprayed femoral component in patients aged 40 years and younger. J Arthroplasty 2006; 21(6 Suppl. 2): 32-7. doi: 10.1016/j. arth.2006.03.008. 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. doi: 10.3109/17453674.2015.1011983. Hermansen L L, Viberg B, Hansen L, Overgaard S. “True” cumulative incidence of and risk factors for hip dislocation within 2 years after primary total hip arthroplasty due to osteoarthritis: a nationwide population-based study from the Danish Hip Arthroplasty Register. J Bone Joint Surg Am 2020; Publish Ahead of Print. doi: 10.2106/JBJS.19.01352. Jorgensen L K, Dalgaard L S, Ostergaard L J, Andersen N S, Norgaard M, Mogensen T H. Validity of the coding for herpes simplex encephalitis in the Danish National Patient Registry. Clin Epidemiol 2016; 8: 133-40. doi: 10.2147/CLEP.S104379. Kristensen M H, Schmidt S A J, Kibsgaard L, Mogensen M, Sommerlund M, Koppelhus U. Validity of first-time diagnoses of congenital epidermolysis bullosa in the Danish National Patient Registry and the Danish Pathology Registry. Clin Epidemiol 2019; 11: 115-24. doi: 10.2147/CLEP.S184742. Lynge E, Sandegaard J L, Rebolj M. The Danish National Patient Register. Scand J Public Health 2011; 39(7 Suppl.): 30-3. doi: 10.1177/1403494811401482. Mears S C, Bawa M, Pietryak P, Jones L C, Rajadhyaksha A D, Hungerford D S, Mont M A. Coding of diagnoses, comorbidities, and complications of total hip arthroplasty. Clin Orthop Relat Res 2002; (402): 164-70. doi: 10.1097/00003086-200209000-00014. Mont M A, Mears S C, Jones L C, Rajadhyaksha A D, Krackow A M, Bawa M, Hungerford D S. Is coding of diagnoses, comorbidities, and complications in total knee arthroplasty accurate? J Arthroplasty 2002; 17(6): 76772. doi: 10.1054/arth.2002.33549. Nymark T, Thomsen K, Rock N D. [Diagnosis and procedure coding in relation to the DRG system]. Ugeskr Laeger 2003; 165(3): 207-9. Patel P D, Potts A, Froimson M I. The dislocating hip arthroplasty: prevention and treatment. J Arthroplasty 2007; 22(4 Suppl. 1): 86-90. doi: 10.1016/j. arth.2006.12.111.

142

Ranstam J, Robertsson O. Statistical analysis of arthroplasty register data. Acta Orthop 2010; 81(1): 10-14. doi: 10.3109/17453671003587168. Saiz AM, Lum ZC, Pereira GC. Etiology, Evaluation, and Management of Dislocation After Primary Total Hip Arthroplasty. JBJS Rev 2019; 7(7): e7. doi: 10.2106/JBJS.RVW.18.00165. Schmidt M, Schmidt S A, Sandegaard J L, Ehrenstein V, Pedersen L, Sorensen H T. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7: 449-90. doi: 10.2147/ CLEP.S91125. SHAR. Swedish Hip Arthroplasty Register—Annual Report. Available from: https://registercentrumblobcorewindowsnet/shpr/r/Eng_Arsrapport_2017_ Hoftprotes_final-Syx2fJPhMNpdf; 2017.

Acta Orthopaedica 2021; 92 (2): 137–142

Varnum C, Pedersen A B, Gundtoft P H, Overgaard S. The what, when and how of orthopaedic registers: an introduction into register-based research. EFORT Open Rev 2019a; 4(6): 337-43. doi: 10.1302/20585241.4.180097. Varnum C, Pedersen A B, Rolfson O, Rogmark C, Furnes O, Hallan G, Makela K, de Steiger R, Porter M, Overgaard S. Impact of hip arthroplasty registers on orthopaedic practice and perspectives for the future. EFORT Open Rev 2019b; 4(6): 368-76. doi: 10.1302/20585241.4.180091. Viborg S, Sogaard K K, Jepsen P. Positive predictive value of peptic ulcer diagnosis codes in the Danish National Patient Registry. Clin Epidemiol 2017; 9:261-6. doi: 10.2147/CLEP.S132628.

Acta Orthopaedica 2021; 92 (2): 143–150 143

Uncemented or cemented stems in first-time revision total hip replacement? An observational study of 867 patients including assessment of femoral bone defect size Yosef TYSON 1,2, Christer HILLMAN 3,4, Norbert MAJENBURG 1,5, Olof SKÖLDENBERG 3,4, Ola ROLFSON 2,6, Johan KÄRRHOLM 2,6, Maziar MOHADDES 2,6, and Nils P HAILER 1,2 1 Section

of Orthopaedics, Department of Surgical Sciences, Uppsala University Hospital, Uppsala, Sweden; 2 The Swedish Hip Arthroplasty Register, Gothenburg, Sweden; 3 Department of Orthopaedics, Danderyd University Hospital Corp, Stockholm, Sweden; 4 Department of Clinical Sciences, Karolinska Institutet, Danderyd Hospital, Division of Orthopaedics, Stockholm, Sweden; 5 University of Groningen, Groningen, The Netherlands; 6 Department of Orthopaedics, Institute of Clinical Sciences, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden Correspondence: yosef.tyson@surgsci.uu.se Submitted 2020-06-26. Accepted 2020-10-14.

Background and purpose — Uncemented stems are gradually replacing cemented stems in hip revision surgery. We compared the risk of re-revision between uncemented and cemented revision stems and assessed whether the different fixation methods are used in similar femoral bone defects. Patients and methods — 867 patients operated on with uncemented or cemented stems in first-time hip revision surgery due to aseptic loosening performed 2006–2016 were identified in the Swedish Hip Arthroplasty Register. Preoperative femoral bone defect size was assessed on radiographs of all patients. Cox regression models were fitted to estimate the adjusted risk of re-revision during different postoperative time periods. Re-revision of any component for any reason, and stem re-revision, as well as risk of cause-specific re-revision was estimated. Results — Most patients in both fixation groups had Paprosky class IIIA femoral bone defects prior to surgery, but there were more severe bone defects in the cemented group. The adjusted risk of re-revision of any component for any reason was higher in patients with uncemented compared with those with cemented revision stems during the first 3 years after index surgery (hazard ratio [HR] 4, 95% confidence interval [CI] 2–9). From the 4th year onward, the risk of re-revision of any component for any reason was similar (HR 0.5, CI 0.2–1.4). Uncemented revision stems conferred a higher risk of dislocation compared with cemented stems (HR 5, CI 1.2–23) during the first 3 years. Interpretation — Although not predominantly used in more complex femoral defects, uncemented revision stem fixation confers a slightly higher risk of re-revision during the first years, but this risk is attenuated after longer followup.

The increased use of primary total hip replacement (THR) has been followed by a steady rise in the frequency of revision surgery (Kurtz et al. 2007, Rajaee et al. 2018), and the use of uncemented revision stems is increasing in most countries (Swedish Hip Arthroplasty Register [SHAR] 2015). Some surgeons consider uncemented revision stems to be more appropriate in situations with extensive femoral bone loss, but others use long cemented revision stems, sometimes in conjunction with bone impaction grafting. Ultimately, the choice of fixation method in revision surgery is a matter not only of science and evidence, but also of taste and local tradition. Register-based studies indicate that uncemented revision stems may have inferior implant survival when compared with cemented stems, especially in the older population (Weiss et al. 2011, Tyson et al. 2019). However, these register studies lack information on femoral bone defect size, a factor that can affect the choice both of fixation method and of outcome in terms of re-revision rates (Paprosky et al. 1999, Pekkarinen et al. 2000, Della Valle and Paprosky 2004, Ten Have et al. 2012). Some smaller observational studies address bone defect size: in 86 patients with comparable femoral bone defects the choice of fixation has no certain influence on implant survival (Iorio et al. 2008), whereas uncemented revision stems conferred inferior implant survival compared with cemented revision stems in 209 patients with comparable femoral bone defects (Hernigou et al. 2015). However, both studies included different reasons for revision surgery, and the second study included both first-time and multiply revised patients. Taken together, the available evidence on the optimal mode of revision stem fixation is hampered by small cohort sizes and lack of control groups (Berry et al. 1995, Iorio et al. 2008, Ornstein et al. 2009, Lakstein et al. 2010, Hernigou et al. 2015, Stigbrand and Ullmark 2017), there is a lack of

144

information on indications underlying revision surgery (Iorio et al. 2008, Weiss et al. 2011, Hernigou et al. 2015), and, most importantly, in large register studies there is no information on the femoral bone defect sizes present at revision surgery (Weiss et al. 2011, Tyson et al. 2019). Our primary aim was therefore to compare the risk of re-revision of any component for any reason between uncemented and cemented stems in hip revision surgery with adjustment for preoperative femoral bone defect size in a large cohort of patients. Our secondary aims were to investigate whether uncemented and cemented revision stems were used in patients with different bone defect sizes, and to assess if the risk of stem re-revision, as well as risk of re-revision of any component due to aseptic loosening, dislocation, fracture, deep infection, and other reasons differed between the 2 fixation techniques.

Patients and methods This is an observational cohort study on patients registered in the SHAR, which collects data on patients who have undergone primary or revision THR since 1979 (Herberts et al. 1989), and the completeness is estimated at 92% (Swedish Hip Arthroplasty Register [SHAR] 2018). The study cohort is based on a subgroup of patients from a previous study, evaluating stem fixation after revision THR, but without assessment of bone defect size (Tyson et al. 2019). Patients with first-time stem revision due to aseptic loosening performed 2006–2016 were identified in the SHAR. We chose 2006 as the starting point of our observation period in order to increase availability of preoperative radiographs because most hospitals archive their radiographs for a maximum of 15 years. We included revision surgeries performed at 9 orthopedic clinics that represented the national praxis in terms of the distribution of uncemented and cemented revision stem fixation according to the SHAR. During the investigated period 66% of revision stems were uncemented, and 34% cemented. We first selected clinics such that the most commonly used revision stems would be represented in our material. Second, only clinics with a minimal annual volume of 50 revisions were included in order to ascertain that low-volume clinics would not bias the results. A lower limit of 50 primary THRs has been reported as the minimal annual hospital volume to maintain a low revision rate in the Nordic countries, but we are unaware whether similar figures exist for the minimal annual volume of revision surgery (Glassou et al. 2016). Whenever patients were bilaterally revised during the study period, only the first revised side was included. Cement-in-cement revisions (Cnudde et al. 2017), in which the cement–bone interface is intact, were considered a separate technique and were therefore excluded from this analysis. Patients who were revised for reasons other than aseptic loosening at index surgery were also excluded in order to avoid bias introduced by the different outcomes after revisions performed due to infection, periprosthetic fracture, or dislocation.

Acta Orthopaedica 2021; 92 (2): 143–150

The variables method of revision stem fixation, stem type, date of re-revision, reason for re-revision, cup type at index revision, sex, age, comorbidity, date of death, and follow-up time were collected from the SHAR. Follow-up started on the day of surgery, and ended on the date of re-revision, death, emigration, or December 31, 2016, whichever came first. Bone defect size The preoperative radiographs (pelvic view, anterior-posterior hip view, and lateral hip view) were retrieved and femoral bone defect size was independently assessed using the Paprosky classification (Della Valle and Paprosky 2004) by at least 2 researchers (YT, NM, or CH). The assessors were not aware of the inserted stem or the outcome when assessing the radiographs of the patients. The Paprosky classification (Table 1, see Supplementary data) was chosen because it offers substantial inter- and intraobserver reliability (Landis and Koch 1977, Brown et al. 2014), and is the classification most often used in previous studies comparing uncemented and cemented revision stems that include assessment of femoral bone defect size (Schmale et al. 2000, Iorio et al. 2008, Hernigou et al. 2015). Inter-observer reliability assessed using weighted Cohen’s kappa ranged between 0.65 and 0.90 in our study, and the intra-observer reliability was 0.90, which is assessed as substantial to almost perfect, according to the criteria of Landis and Koch (1977). In cases where classification by the 2 researchers did not agree, consensus was met under the guidance of the senior author (NH). Patients who had unclassifiable radiographs (for example due to osteosynthesis as a result of prior periprosthetic fracture), radiographs dating back over 2 years prior to index procedure, or unavailability of preoperative radiographs, were excluded. Bone impaction grafting and surgical approach All surgical notes were retrieved and assessed by 1 of 2 authors (YT or CH). We searched for the terms “bone impaction grafting (of the femur),” or any detailed description thereof, the terms “endofemoral approach,” “transfemoral approach,” or “extended trochanteric osteotomy”, or any detailed description thereof. We also compared the preoperative assessment in the surgical notes with the information in the register database in order to ascertain that index revisions were correctly classified as being due to aseptic loosening and whether they were indeed first-time revisions. Only 2% of the procedures were incorrectly classified as aseptic loosening or first-time revisions, and these were excluded. Outcomes Primary outcome was re-revision of any component for any reason, i.e., only cup, head, or stem revision, any combination thereof, or extraction of the prosthesis. Secondary outcomes were percentage of preoperative femoral bone defect sizes in both fixation groups, re-revision of the stem for any reason, and reasons for re-revision.

Acta Orthopaedica 2021; 92 (2): 143–150 145

Confounders The confounders age at index surgery, sex, and femoral bone defect size were included in multivariable regression models. We chose confounders by using a strict epidemiological approach in which only those factors that we believed would affect both exposure and outcome were deemed confounders. To illustrate our thought process, we constructed a directed acyclic graph (Figure 1, see Supplementary data) with the use of the dagitty.net online software. Use of surgical approach or of femoral bone impaction grafting were considered mediators, concomitant cup revision, and femoral head size were considered as having their own causal pathway, and all these variables were thus not defined as confounders. Since comorbidity as measured by ASA grade was first introduced in 2008 in the register and therefore not complete in our dataset, comorbidity was adjusted for only in a subgroup of patients. Bone defect size as confounder was divided into 3 groups: group 1 (Paprosky I and II), group 2 (Paprosky IIIA), and group 3 (Paprosky IIIB and IV), in order to obtain sufficient numbers of patients in each group. Statistics Demographics including femoral bone defect size were described with percentages, means, and standard deviations. Unadjusted implant survival was estimated using the Kaplan–Meier method. Risks of re-revision of any component or of the stem were expressed as hazard ratios (HR) with 95% confidence intervals (CI) and estimated by fitting multivariable Cox regression models adjusted for the confounders described above. Since the unadjusted cumulative hazard function for the endpoint re-revision for any reason deviated considerably from the assumption of proportionality, the follow-up time was divided into 2 time periods at the point where the curves began to converge. The maximum difference between the survival curves was calculated to 2.95 years and for reasons of readability 3 years was chosen as the dividing point. Thus, the 1st time period consisted of the 1st 3 years after index surgery and the 2nd time period ranged from the 4th to the 8th year after index surgery. Schoenfeld residuals were calculated in order to assess whether model assumptions were met. In order to assess the risk of re-revision due to aseptic loosening, dislocation, periprosthetic fracture, infection, or other reasons, cumulative incidence functions and subdistribution HRs with 95% CI were calculated using a competing risks model according to Fine and Gray (1999). All reasons for rerevision and death were considered competing events. Thorough sensitivity analyses were conducted (see Supplementary data). R Statistical Software (R version 3.4.3, 2017-11-30, R Foundation for Statistical Computing, Vienna, Austria), packages: haven, pastecs, knitr, Epi, rms, Gmisc, magrittr, tidyverse, irr, lpSolve, kableExtra, ggplot2, cmprsk, crrSC, scales, and reshape2 were used for the calculations (R Core Team 2017).

Stem revisions due to aseptic loosening reported to the Swedish Hip Arthroplasty register 2006–2016 n = 3,018 Excluded (n = 2,151): – not first-time revisions, 661 – second revised hip, 140 – stem types with < 100 observations, 37 – non-selected hospitals, 1,208 – cement-in-cement revisions, 22 – unclassifiable or missing radiographs, 69 – in fact not first-time revisions or due to aseptic loosening, 14 Included patients with first-time stem revision due to aseptic loosening n = 867

Figure 2. Flow chart of included patients.

Ethics, funding, and potential conflicts of interest Ethical approval was obtained from the Regional Ethics Review Board in Uppsala, Sweden (decision 2018/076). Financial support was received from the Health Care Committees in Region Uppsala and Region Västra Götaland. No competing interests were declared.

Results A cohort of 867 patients was included in this study (Figure 2). The study cohort was divided into 2 groups of patients, 1 operated on with uncemented (n = 601, 69%) and 1 with cemented revision stems (n = 266, 31%). The use of uncemented revision stems was slightly higher in the second half of the study period (Figure 3, see Supplementary data). The mean follow-up time was 4.5 (SD 3.0) years and the mean age at index surgery was 72 (SD 10) years. The groups had similar sex distribution, but patients who received uncemented revision stems were on average younger at the time of surgery and had a shorter mean follow-up time. Most patients in both fixation groups had medium size femoral bone defects prior to surgery, but fewer of those who received uncemented stems had large bone defects (8%) compared with patients with cemented stems (15%) prior to index surgery (Table 2). In the cemented group, bone impaction grafting was performed equally often when large bone defects were present as cementation alone (Table 3, see Supplementary data). Of the clinics that performed cemented fixation, most clinics tended to use bone impaction grafting in either all or none of the cases (Figure 4, see Supplementary data). Re-revision of any component for any reason The unadjusted 10-year implant survival with re-revision of any component for any reason was lower after use of uncemented revision stems compared with cemented stems (83%,

146

Acta Orthopaedica 2021; 92 (2): 143–150

Table 2. Characteristics of the study population. Values are count (%) unless otherwise specified

Implant survival (%) with endpoint re-revision of any component for any reason 100

Factor

Cemented Uncemented n = 266 (31) n = 601 (69)

Mean age (SD) 74 (9) 72 (10) Men 138 (52) 318 (53) Mean follow-up, years (SD) 5 (3) 4 (3) Diagnosis at primary THR Osteoarthritis 212 (80) 462 (77) Fracture 14 (5) 34 (6) Other 40 (15) 105 (17) Type of revision Cup and stem revision 219 (82) 504 (85) Stem revision only 47 (18) 90 (15) Stem brand MP 0 (0) 291 (48) Restoration 0 (0) 162 (27) Wagner 0 (0) 78 (13) Revitan 0 (0) 70 (12) Lubinus SPII 123 (46) 0 (0) Exeter 94 (35) 0 (0) Spectron 49 (18) 0 (0) Femoral bone defect size I 3 (1) 8 (1) II 51 (19) 96 (16) IIIA 170 (64) 449 (75) IIIB 17 (6) 25 (4) IV 25 (9) 23 (4) Bone impaction grafting 125 (47) 23 (4) Surgical approach Endofemoral 263 (99) 382 (64) Transfemoral 3 (1) 219 (36)

Cemented

Uncemented

50 0

At risk

Years to event

Cemented 266 210 162 116 58 16 266 210 162 116 58 16 Uncemented 601 434 304 173 68 24 Cemented 601

434

304

173

24 Uncemented

Figure 5. Unadjusted implant survival with endpoint re-revision of any component.

Aseptic loosening

CI 77–99 versus 89%, CI 83–95, Figure 5), but with overlapping CIs. During the first 3 years after index surgery, we attained a higher estimate for the adjusted risk of re-revision of any component for any reason in patients with uncemented compared with those with cemented revision stems (HR 4, CI 2–9). Between the 4th and 8th year, the adjusted risk of rerevision of any component for any reason was similar in both groups (HR 0.5, CI 0.2–1.4). Stem re-revision for any reason The unadjusted 10-year implant survival with stem re-revision for any reason was lower after use of uncemented revision stems compared with cemented stems (90%, CI 87–93 versus 94%, CI 90–98), but with overlapping CIs. The adjusted risk of stem re-revision for any reason was higher for patients with uncemented compared with those with cemented revision stem fixation during the first 3 years after index surgery (HR 6, CI 2–15). Between the 4th and 8th year, the adjusted risk of stem re-revision for any reason was lower after uncemented compared with cemented revision stem fixation (HR 0.3, CI 0.1–1.0). Risk of re-revision for different reasons Re-revisions of any component after the use of uncemented revision stem fixation were most often performed due to dis-

Infection

Periprosthetic fracture

Dislocation Cemented

Other

Uncemented 0

Distribution (%) of reasons for re-revision

Figure 6. Reasons for re-revision.

location (35%), whereas re-revisions after cemented revision stem fixation were most often performed due to aseptic loosening (30%) or dislocation (30%, Figure 6). The risk of rerevision due to dislocation was higher for uncemented compared with cemented revision stems, both during the first 3 years and during the 4th to 8th year after index surgery (HR 5, CI 1.2–23 and HR 3, CI 1.0–9). There was no statistically significant difference in risk of re-revision due to aseptic loosening, deep infection, fracture, or other reasons, between uncemented and cemented revision stems (Table 4).

Acta Orthopaedica 2021; 92 (2): 143–150 147

Table 4. Subdistribution hazard ratios (HR) for re-revision of uncemented compared with cemented revision stems, estimated with the Fine and Gray method Reason for re-revision

Years after index surgery 0–3 4–8 HR (95% CI) HR (95% CI)

Aseptic loosening Deep infection Periprosthetic fracture Dislocation Other reasons

2 (0.5–9) 3 (0.5–11) Too few events 5 (1.2–23) 6 (0.8–51)

1 (0.5–4) 2 (0.5–5) 3 (0.3–28) 3 (1–9) 3 (0.7–15)

Aseptic loosening, deep infection, periprosthetic fracture, dislocation, other reasons for re-revision, and death were considered competing events. The regression was adjusted for age, sex, and femoral bone defect size (aggregating defects into 3 main groups along Paprosky classes I + II, IIIA, IIIB + IV)

Discussion Uncemented stems are gradually replacing cemented stems in hip revision surgery in Sweden (Swedish Hip Arthroplasty Register [SHAR] 2015) but there is no study comparing the 2 fixation concepts with a sample size above 209 patients including assessment of femoral bone defect size (Iorio et al. 2008, Weiss et al. 2011, Hernigou et al. 2015, Tyson et al. 2019). Our study is the largest comparative study on uncemented and cemented revision stems with stringent assessment of preoperative femoral bone defect size, and our main finding is that although both concepts have satisfactory medium-term outcomes, the risk of re-revision for any reason is considerably higher after the use of uncemented revision stems during the first 3 years. For the remainder of the observation period the risk seems similar. Even though the choice of fixation method is ultimately surgeon-dependent, the increased use of uncemented revision stems might be due to the fact that modular stems offer the option of distal anchoring within intact bone, and several opportunities exist to vary the proximal part in order to achieve optimal soft tissue tension, anteversion, and offset (Weiss et al. 2009). Another explanation could be that primary cemented stems have higher risk of aseptic loosening compared with uncemented (Mäkelä et al. 2008, 2011), which is why it would be logical to assume that uncemented revision stems would decrease the risk of subsequent revisions in the case of aseptic loosening, a phenomenon that has been observed among revision stems in previous studies (Weiss et al. 2011, Tyson et al. 2019). Previous register-based studies state that uncemented revision stems fail more frequently during the early postoperative period but might confer a lower risk of loosening in the long term (Weiss et al. 2011, Tyson et al. 2019). Our present study supports that uncemented revision stems fail more frequently during the early postoperative years, and it adds to the cited

studies in that we were able to adjust for femoral bone defect size. In fact, fewer patients with uncemented stems had large femoral bone defects prior to surgery compared with patients receiving cemented stems, thus bone defect size seems not to be the main reason for difference in implant survival, which has been suggested (Weiss et al. 2011). No particular stem model within each fixation group deviated considerably from the mean risk of re-revision within each group (data not shown). One could question our results given that Swedish surgeons are more accustomed to using cemented primary implants. However, in our previous study we stratified our results based on the year of implantation as a proxy for becoming accustomed to the uncemented fixation technique, and no major differences were observed (Tyson et al. 2019). In a cohort study with preoperative femoral bone defect assessment that comprised 209 patients, 21% of the uncemented revision stems were re-revised at 5 years, in contrast to none of the cemented revision stems (Hernigou et al. 2015). However, the risk of cardio-pulmonary events after surgery but not re-revision was the primary outcome in that study. In an observational, matched cohort study including 86 patients, uncemented and cemented revision stems are comparable in terms of implant survival at 5 years (Iorio et al. 2008). However, only lowgrade femoral bone defects corresponding to Paprosky classes I to II were studied, which limits the generalizability of these findings. Although a study by Schmale et al. (2000) demonstrated superior survivorship of uncemented stems, the comparator group was pre-coated cemented stems, which are no longer used due to increased risk of early failure (Ong et al. 2002). Previous studies have suggested that uncemented revision stems should be used in younger patients, but cemented ones in the elderly (Weiss et al. 2011, Tyson et al. 2019). Our study does not support this assumption; however, the events in each age group were few in number, thus limiting the conclusions to be drawn, and one can argue that the elderly with short life expectancy would benefit from the decreased risk of early complications associated with cemented stem fixation. Further, our results are applicable only to revisions caused by aseptic loosening, and further research is necessary to investigate outcomes after revisions due to infection, periprosthetic fracture, dislocation, and other causes. We have previously reported similar results concerning reasons for re-revision after revision THR in a register-based cohort, but without assessment of femoral bone defect sizes (Tyson et al. 2019). Our study lends support to the theory that the different failure mechanisms observed in uncemented and cemented revision stems are at least in part analogous to the failure mechanisms of these 2 fixation principles in primary THR, where dislocation as an early complication is more frequently observed after uncemented stem fixation (Hailer et al. 2010, Pedersen et al. 2014, Gromov et al. 2015, Tyson et al. 2019). The increased incidence of re-revisions due to dislocation in patients who received uncemented revision stems may

148

in part be the result of stem subsidence, but without evaluation of serial radiographs up to the re-revision such a connection cannot be confirmed (Malchau et al. 1997, Paprosky et al. 1999, Mardones et al. 2005, Ström et al. 2006, Lakstein et al. 2010, Hernigou et al. 2015, Klein et al. 2019). Dislocation is associated with smaller head sizes (Hailer et al. 2012a); however, in our study uncemented stems on average had larger head sizes, which is why this phenomenon does not explain our observation. Further, without any follow-up visits we could not evaluate whether abduction impairment differed between the groups. It should be emphasized that we did not record the rate of dislocations but only the rate of re-revisions due to dislocation, and it might be that the threshold to re-revise the proximal part of a modular uncemented revision stem is lower than before re-revising a well-cemented revision stem. Aseptic stem loosening is a late complication, and after primary THR loosening is more common after the use of cemented than after that of uncemented stems (Mäkelä et al. 2008, 2011, Pedersen et al. 2014, Gromov et al. 2015). The increased risk of aseptic loosening among cemented revision stems could be due to longer follow-up of this group. It is also suggested that aseptic loosening of cemented stems results from debonding of the cement from the femoral canal (Sundfeldt et al. 2006), and the risk of its occurrence could increase with increasing magnitude of bone defects, but this phenomenon was not observed in our study. The technique of bone impaction grafting was introduced to restore bone stock and to facilitate cementation in a femoral canal devoid of trabecular bone, and modern bone impaction techniques yield promising results (Ornstein et al. 2009, Howie et al. 2010, Wilson et al. 2016, Stigbrand and Ullmark 2017). In our study bone impaction grafting in combination with cementation did not reduce the risk of re-revision compared with cementation alone, even though the use of bone impaction grafting was performed equally often in patients with large bone defects as was cementation alone. Further, when we restricted the cohort to patients that did not receive bone impaction grafting, there was no statistically significant difference when compared with the estimates derived from the main analysis. However, since we only have short- to medium-term follow-up, the full effect of bone impaction grafting might not have been observed in our study. It should be noted that although cemented revision stems were more often re-revised due to aseptic loosening than uncemented stems, there was no statistically significant difference in the risk of re-revision due to aseptic loosening between the 2 fixation methods in the competing risk analysis. However, a lack of statistically significant differences between groups, with confidence intervals that include 1.0, does not imply evidence of absence of differences between groups. The use of dual mobility cups at index surgery protected against re-revision due to dislocation among uncemented revision stems in our study, a phenomenon reported by other authors as well, in both the primary and revision setting (Hailer et al. 2012b, Mohaddes et al. 2017, Jobory et al. 2019,

Acta Orthopaedica 2021; 92 (2): 143–150

Kreipke et al. 2019). However, to study dual mobility cups was not the primary aim of our study so these results should be interpreted with some caution. There was no statistically significant difference in the incidence of periprosthetic fracture, even though previous authors report increased risk of periprosthetic fracture after use of uncemented stems (Hailer et al. 2010, Mäkelä et al. 2014, Thien et al. 2014, Gromov et al. 2015). One explanation for this could be the use of bone impaction grafting in conjunction with cementation; bone impaction grafting has been reported as a risk factor for perioperative periprosthetic fracture (Ornstein et al. 2002). In our study we did not investigate perioperative complications, only subsequent re-revisions. Further, in our study, most patients had the primary diagnosis osteoarthritis, but in a subgroup analysis on patients with a diagnosis of femoral neck fracture at primary arthroplasty surgery (n = 48) no patient was rerevised due to periprosthetic fracture. Strengths and limitations To our knowledge, our study represents the largest comparison of uncemented and cemented revision stems with stringent assessment of femoral bone defect size at the time of revision surgery. We performed several sensitivity analyses that supported our main results, but the study has its limitations. 1st, the follow-up time was only medium-term, thus possibly underestimating the incidence of aseptic loosening, which could positively bias the outcome after cemented revision stem fixation, while similarly underestimating the potentially beneficial effects of bone impaction grafting in the long-term. 2nd, bone defects in the more advanced stages of the Paprosky classification were scarce, which could imply that our conclusions are mainly valid for small to medium sized femoral bone defects. Although bone defect size was controlled for in the main analyses and sensitivity analyses stratified on bone defect size indicated similar results, one should be cautious when interpreting our results. It cannot be concluded, based on our data, that bone impaction grafting or uncemented stems should not be used in extensive femoral bone defects as we lack both statistical precision and long-term data for these scenarios. 3rd, the generalizability of our study can be questioned given that only half of the revision patients operated on within the time frame of this study were included, also excluding patients from low-volume institutions. On the other hand, outcomes after first-time revisions due to aseptic loosening in all revisions in Sweden but without assessment of preoperative radiographs were very similar (Tyson et al. 2019), indicating that that the results of this present study are valid. 4th, given that Sweden is a “cementation country,” surgeons’ learning curves could possibly influence the results after uncemented stems, for example if the surgeon is afraid to use a wider uncemented stem to achieve more stable fixation that prevents subsidence. However, in a previous study by the present authors on revision THR and fixation technique (Tyson et

Acta Orthopaedica 2021; 92 (2): 143–150 149

al. 2019), we evaluated whether year of implantation as proxy for learning a new technique influenced the results, and no influence was observed. 5th, even though bone impaction grafting was a variable collected from the surgical notes, bone impaction grafting is a surgical technique with certain principles, and it is difficult to know to what extent these principles were adhered to. In addition, as with all study designs, the results and conclusions are more applicable to the studied population; however, with an assessment of preoperative bone defect size, we believe the conclusions may be somewhat more generalizable. On the other hand, depending on their surgical experience with cementation, the present results may not apply to all countries. Finally, the validity of register data can always be questioned, but completeness is estimated to be 92% for revision THR in the SHAR (Swedish Hip Arthroplasty Register [SHAR] 2018). Further, variables such as bone impaction grafting and surgical approach were directly collected from the surgical notes, and surgical notes are routinely sent to the register to assess the correctness of data. In our study, only 2% of the patients had incorrectly classified data and these were excluded. Conclusions This large, register-based study with stringent assessment of femoral bone defect sizes indicates that uncemented revision stems are not used in more complex revision situations, but that they still confer an increased early risk of re-revision after revision THR, mostly due to dislocations occurring during the first postoperative years. Cemented stem fixation is a good option in hip revision surgery, and our findings do not support the declining use of this fixation technique. We advocate the use of cemented fixation in older patients with short life expectancy who will benefit from reduced risk of early complications, even though the ultimate choice of stem must be individualized. The different reasons for failure should be considered when counselling patients prior to these procedures, and measures should be taken to avoid the specific failure mechanisms associated with the different fixation methods. Supplementary data Table 1 and 3 and Figures 1, 3, 4, 7, and 8 are available in the online version of this article, as well as methodological description and results of the sensitivity analyses, http:// dx.doi.org/10.1080/17453674.2020.1846956

YT, MM, OR, JK, and NH conceived and designed the study. YT, CH, NM, and OS participated in data collection and assessment. YT, NM, and NH performed the statistical analyses. All authors participated in the interpretation of the results. YT wrote the manuscript, and all authors participated in the revision of the manuscript.

The authors would like to thank Eva Freyhult at Uppsala SciLifeLab for her statistical expertise. They also thank the Swedish orthopedic surgeons and administrators reporting to the Swedish Hip Arthroplasty Register, and all the nurses and secretaries involved in the retrieval of radiographs and surgical notes from different units. Acta thanks Ralf Bieger, Martin Buttaro, and Søren Overgaard for help with peer review of this study.

Berry D J, Harmsen W S, Ilstrup D, Lewallen D G, Cabanela M E. Survivorship of uncemented proximally porous-coated femoral components. Clin Orthop Relat Res 1995; (319): 168-77. Brown N M, Foran J R, Valle C J, Moric M, Sporer S M, Levine B R, Paprosky W G. The inter-observer and intra-observer reliability of the Paprosky femoral bone loss classification system. J Arthroplasty 2014; 29(7): 1482-4. Cnudde P H, Kärrholm J, Rolfson O, Timperley A J, Mohaddes M. Cementin-cement revision of the femoral stem: analysis of 1179 first-time revisions in the Swedish Hip Arthroplasty Register. Bone Joint J 2017; 99-B(4 Suppl. B): 27-32. Della Valle C J, Paprosky W G. The femur in revision total hip arthroplasty evaluation and classification. Clin Orthop Relat Res 2004; (420): 55-62. Fine J P, Gray R J. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: 496–509. Glassou E N, Hansen T B, Mäkelä K, Havelin L I, Furnes O, Badawy M, Kärrholm J, Garellick G, Eskelinen A, Pedersen A B. Association between hospital procedure volume and risk of revision after total hip arthroplasty: a population-based study within the Nordic Arthroplasty Register Association database. Osteoarthritis Cartilage 2016; 24(3): 419-26. Gromov K, Pedersen A B, Overgaard S, Gebuhr P, Malchau H, Troelsen A. Do rerevision rates differ after first-time revision of primary THA with a cemented and cementless femoral component? Clin Orthop Relat Res 2015; 473(11): 3391-8. Hailer N P, Garellick G, Kärrholm J. Uncemented and cemented primary total hip arthroplasty in the Swedish Hip Arthroplasty Register. Acta Orthop 2010; 81(1): 34-41. 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: an analysis of 78,098 operations in the Swedish Hip Arthroplasty Register. Acta Orthop 2012a; 83(5): 442-8. Hailer N P, Weiss R J, Stark A, Kärrholm J. Dual-mobility cups for revision due to instability are associated with a low rate of re-revisions due to dislocation: 228 patients from the Swedish Hip Arthroplasty Register. Acta Orthop 2012b; 83(6): 566-71. Herberts P, Ahnfelt L, Malchau H, Strömberg C, Andersson G B. Multicenter clinical trials and their value in assessing total joint arthroplasty. Clin Orthop Relat Res 1989: (249): 48-55. Hernigou P, Dupuys N, Delambre J, Guissou I, Poignard A, Allain J, Flouzat Lachaniette C H. Long, titanium, cemented stems decreased late periprosthetic fractures and revisions in patients with severe bone loss and previous revision. Int Orthop 2015; 39(4): 639-44. Howie D W, Callary S A, McGee M A, Russell N C, Solomon L B. Reduced femoral component subsidence with improved impaction grafting at revision hip arthroplasty. Clin Orthop Relat Res 2010; 468(12): 3314-21. Iorio R, Healy W L, Presutti A H. A prospective outcomes analysis of femoral component fixation in revision total hip arthroplasty. J Arthroplasty 2008; 23(5): 662-9. Jobory A, Kärrholm J, Overgaard S, Becic Pedersen A, Hallan G, Gjertsen J E, Mäkelä K, Rogmark C. Reduced revision risk for dual-mobility cup in total hip replacement due to hip fracture: a matched-pair analysis of 9,040 cases from the Nordic Arthroplasty Register Association (NARA). J Bone Joint Surg Am 2019; 101(14): 1278-85.

150

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. Kreipke R, Rogmark C, Pedersen A B, Kärrholm J, Hallan G, Havelin L I, Mäkelä K, Overgaard S. Dual mobility cups: effect on risk of revision of primary total hip arthroplasty due to osteoarthritis: a matched populationbased study using the Nordic Arthroplasty Register Association database. J Bone Joint Surg Am 2019; 101(2): 169-76. 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. Lakstein D, Backstein D, Safir O, Kosashvili Y, Gross A E. Revision total hip arthroplasty with a porous-coated modular stem: 5 to 10 years follow-up. Clin Orthop Relat Res 2010; 468(5): 1310-5. Landis J R, Koch G G. The measurement of observer agreement for categorical data. Biometrics 1977; 33(1): 159-74. Mäkelä K T, Eskelinen A, Pulkkinen P, Paavolainen P, Remes V. Total hip arthroplasty for primary osteoarthritis in patients fifty-five years of age or older: an analysis of the Finnish Arthroplasty Registry. J Bone Joint Surg Am 2008; 90(10): 2160-70. Mäkelä K T, Eskelinen A, Pulkkinen P, Paavolainen P, Remes V. Results of 3,668 primary total hip replacements for primary osteoarthritis in patients under the age of 55 years. Acta Orthop 2011; 82(5): 521-9. Mäkelä K T, Matilainen M, Pulkkinen P, Fenstad A M, Havelin L, Engesaeter L, Furnes O, Pedersen A B, Overgaard S, Kärrholm J, Malchau H, Garellick G, Ranstam J, Eskelinen A. Failure rate of cemented and uncemented total hip replacements: register study of combined Nordic database of four nations. BMJ 2014; 348: f7592. Malchau H, Wang Y X, Kärrholm J, Herberts P. Scandinavian multicenter porous coated anatomic total hip arthroplasty study. Clinical and radiographic results with 7- to 10-year follow-up evaluation. J Arthroplasty 1997; 12(2): 133-48. Mardones R, Gonzalez C, Cabanela M E, Trousdale R T, Berry D J. Extended femoral osteotomy for revision of hip arthroplasty: results and complications. J Arthroplasty 2005; 20(1): 79-83. Mohaddes M, Cnudde P, Rolfson O, Wall A, Kärrholm J. Use of dual-mobility cup in revision hip arthroplasty reduces the risk for further dislocation: analysis of seven hundred and ninety one first-time revisions performed due to dislocation, reported to the Swedish Hip Arthroplasty Register. Int Orthop 2017; 41(3): 583-8. Ong A, Wong K L, Lai M, Garino J P, Steinberg M E. Early failure of precoated femoral components in primary total hip arthroplasty. J Bone Joint Surg Am 2002; 84(5): 786-92. Ornstein E, Atroshi I, Franzen H, Johnsson R, Sandquist P, Sundberg M. Early complications after one hundred and forty-four consecutive hip revisions with impacted morselized allograft bone and cement. J Bone Joint Surg Am 2002; 84(8): 1323-8. Ornstein E, Linder L, Ranstam J, Lewold S, Eisler T, Torper M. Femoral impaction bone grafting with the Exeter stem: the Swedish experience: survivorship analysis of 1305 revisions performed between 1989 and 2002. J Bone Joint Surg Br 2009; 91(4): 441-6. Paprosky W G, Greidanus N V, Antoniou J. Minimum 10-year-results of extensively porous-coated stems in revision hip arthroplasty. Clin Orthop Relat Res 1999; (369): 230-42.

Acta Orthopaedica 2021; 92 (2): 143–150

Pedersen A B, Mehnert F, Havelin L I, Furnes O, Herberts P, Karrholm J, Garellick G, Makela K, Eskelinen A, Overgaard S. Association between fixation technique and revision risk in total hip arthroplasty patients younger than 55 years of age: results from the Nordic Arthroplasty Register Association. Osteoarthritis Cartilage 2014; 22(5): 659-67. Pekkarinen J, Alho A, Lepisto J, Ylikoski M, Ylinen P, Paavilainen T. Impaction bone grafting in revision hip surgery: a high incidence of complications. J Bone Joint Surg Br 2000; 82(1): 103-7. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2017. Rajaee S S, Campbell J C, Mirocha J, Paiement G D. Increasing burden of total hip arthroplasty revisions in patients between 45 and 64 years of age. J Bone Joint Surg Am 2018; 100(6): 449-58. Schmale G A, Lachiewicz P F, Kelley S S. Early failure of revision total hip arthroplasty with cemented precoated femoral components: comparison with uncemented components at 2 to 8 years. J Arthroplasty 2000; 15(6): 718-29. Stigbrand H, Ullmark G. A 3- to 18-year follow-up of revision total hip arthroplasty with impacted bone allografts and cemented Lubinus SP II stem: clinical, radiographic, and survivorship analysis with comparison to the literature. J Arthroplasty 2017; 32(9): 2810-4. 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. Sundfeldt M, Carlsson L V, Johansson C B, Thomsen P, Gretzer C. Aseptic loosening, not only a question of wear: a review of different theories. Acta Orthop 2006; 77(2): 177-97. Swedish Hip Arthroplasty Register (SHAR). SHAR Annual Report 2014. Gothenburg; 2015. Swedish Hip Arthroplasty Register (SHAR). SHAR Annual Report 2017. Gothenburg; 2018. Ten Have B L, Brouwer R W, van Biezen F C, Verhaar J A. Femoral revision surgery with impaction bone grafting: 31 hips followed prospectively for ten to 15 years. J Bone Joint Surg Br 2012; 94(5): 615-8. Thien T M, Chatziagorou G, Garellick G, Furnes O, Havelin L I, Mäkelä K, Overgaard S, Pedersen A, Eskelinen A, Pulkkinen P, Kärrholm J. Periprosthetic femoral fracture within two years after total hip replacement: analysis of 437,629 operations in the nordic arthroplasty register association database. J Bone Joint Surg Am 2014; 96(19): e167. Tyson Y, Rolfson O, Karrholm J, Hailer N P, Mohaddes M. Uncemented or cemented revision stems? Analysis of 2,296 first-time hip revision arthroplasties performed due to aseptic loosening, reported to the Swedish Hip Arthroplasty Register. Acta Orthop 2019; 90(5): 421-6. Weiss R J, Strömwall F, Beckman M O, Hansson K A, Stark A. Distal femoral stem–bone anchorage of a cementless revision total hip arthroplasty: evaluation of 14 patients by CT. Acta Orthop 2009; 80(3): 298-302. Weiss R J, Stark A, Kärrholm J. A modular cementless stem vs. cemented long-stem prostheses in revision surgery of the hip: a population-based study from the Swedish Hip Arthroplasty Register. Acta Orthop 2011; 82(2): 136-42. Wilson M J, Hook S, Whitehouse S L, Timperley A J, Gie G A. Femoral impaction bone grafting in revision hip arthroplasty: 705 cases from the originating centre. Bone Joint J 2016; 98-B(12): 1611-9.

Acta Orthopaedica 2021; 92 (2): 151–155 151

Reduced wear in vitamin E-infused highly cross-linked polyethylene cups: 5-year results of a randomized controlled trial Goulven ROCHCONGAR, Matthieu REMAZEILLES, Emeline BOURROUX, Julien DUNET, Valentin CHAPUS, Matthieu FERON, César PRAZ, Geoffrey BUIA, and Christophe HULET

Caen Normandy University Hospital Centre, Department of Orthopaedic and Trauma Surgery, 14000, Caen, France Correspondence: praz-c@chu-caen.fr Submitted 2020-05-29. Accepted 2020-10-22

Background and purpose — Vitamin E-infused polyethylene is a relatively new material in joint arthroplasty; there are no long-term reports, and only few mid-term results. Using radiostereometric analysis (RSA), we primarily determined whether vitamin E-infused highly cross-linked polyethylene (HXLPE/VitE) acetabular cups show less wear than ultra-high molecular weight polyethylene (UHMWPE) acetabular cups at 5 years after total hip arthroplasty (THA). We also assessed whether wear rates correlate with increasing cup inclination angles or cup sizes. Patients and methods — This is a 5-year follow-up of our previously reported randomized controlled trial of 62 patients with 3 years’ follow-up, who received THA with either an HXLPE/VitE or a UHMWPE acetabular cup. At 5 years, 40 patients were analyzed (22 in the HXLPE/VitE and 18 in the UHMWPE group). Results — HXLPE/VitE cups continued to show less cumulative femoral head penetration than UHMWPE cups (HXLPE/VitE: 0.24 mm, UHMWPE: 0.45 mm; p < 0.001). Distribution of wear was also more even with HXLPE/VitE cups than with UHMWPE cups (p = 0.002). Moreover, the difference in PE wear between 1 and 5 years in both groups showed no statistically significant correlation with increasing cup inclination angles or cup sizes. Finally, no osteolysis and implant loosening occurred, and no revision surgeries were required. Interpretation — Wear rates continue to be lower in HXLPE/VitE cups than in UHMWPE cups at 5 years of follow-up without correlation with increasing cup inclination angles or cup sizes. Finally, HXLPE/VitE cups may have the potential to prevent osteolysis and implant loosening.

Wear of the polyethylene (PE) component of total hip arthroplasties (THA) may result in osteolysis (Callary et al. 2015). Therefore, attempts such as cross-linking using irradiation and addition of vitamin E have been made to improve the wear properties of PE (Galea et al. 2019). Vitamin-E infused highly cross-linked polyethylene (HXLPE/VitE) has been developed to reduce the number of free radicals without compromising the mechanical properties. There are 2 methods of incorporating vitamin E into PE. The 1st is to blend vitamin E with PE powder before consolidation and. Once consolidated, the blend can be irradiated for sterilization or cross-linking. The 2nd is diffusion of vitamin E into the PE after radiation crosslinking: after PE is irradiated for cross-linking, it is diffused with vitamin E, then machined into its final form and gamma sterilized (Oral et al. 2005). Gamma irradiation causes crosslinking of UHMWPE, which changes its property from the original. However, it causes reduction in tensile strength and elongation of UHMWPE, and leads to long-lived free radicals that react with oxygen (Oral et al. 2007). As HXLPE/VitE is a relatively new material in orthopedic surgery, studies on its wear properties with longer follow-up periods are still limited (Nebergall et al. 2016, 2017, Shareghi et al. 2017, Galea et al. 2019). Our initial 3-year data showed less wear with HXLPE/VitE, which may prevent osteolysis, implant loosening, and eventually revision surgery (Rochcongar et al. 2018). We have now investigated clinical and radiographic outcomes of our previously reported patient cohort at 5-year follow-up. The primary objective is to know whether HXLPE/VitE acetabular cups continue to show less PE wear than ultra-high molecular weight polyethylene (UHMWPE) acetabular cups at 5 years. The secondary objective is to evaluate the correlation between PE wear rates with cup inclination angles or cup sizes, in addition to reporting clinical outcomes.

152

Acta Orthopaedica 2021; 92 (2): 151–155

Assessed for eligibility n = 494

Patients and methods

Excluded n = 432

Trial design This single-center, randomized controlled trial (RCT) was undertaken as a stratified, parallel-group RCT. Patients Inclusion and exclusion criteria were the same as described previously (Rochcongar et al. 2018). The study was performed at Caen University Hospital, a major referral hospital. Randomization and blinding Randomization and blinding were carried out as described previously (Rochcongar et al. 2018). Interventions On the acetabular side, patients received either an HXLPE/ VitE cup (RM Pressfit vitamys, Mathys Ltd, Bettlach, Switzerland) or a UHMWPE cup (RM Pressfit, Mathys Ltd, Bett lach, Switzerland). All procedures were performed by attending surgeons and fellows under supervision. Outcomes The primary and secondary outcomes were set a priori and measured as described previously (Rochcongar et al. 2018). As the primary outcome, the femoral head penetration using model-based RSA with the patient standing on both legs (Callary et al. 2015) was measured. To obtain suitable images of the hip, ceiling-mounted and mobile radiographic tubes were used simultaneously, with a calibration cage behind the patient. RSA measurements prior to the study were validated by 3 blinded investigators who performed the measurements of specially manufactured liners with different concentric wear rates. The precision was to be 0.072 mm and accuracy to be 0.034 mm, which is similar to the values reported in previous studies (Pineau et al. 2010). The RSA exam was performed 7 days after surgery (baseline) and then again at 6 months and at 1, 2, 3, and 5 years later. RSA images were analyzed using Medis Specials medical imaging software (Medis, Leiden, the Netherlands). All images were processed using contour detection software (Model-Based RSA [MBRSA], version 3.2; Medis) as described previously (Garling et al. 2005). The 3D contour of each implant was projected on each view. Wear was calculated by taking the 12-month mark as the baseline for the measurements, and all changes in relative distance between the head and cup were assumed to be due to creep and possible wear. The initial distance between the center of the cup and the center of the head was defined as the minimal head penetration into the liner. 3D femoral head penetration was calculated as the vectorial sum of medial (x), proximal (y), and anterior translation (z) (Callary et al. 2013). The results

Randomized n = 62 Allocated to HXLPE/VitE (n = 33) Received allocated intervention (n = 33): – baseline RSA not usable, 3 – analyzed, 30 Lost to follow-up, not available n=2 1-year follow-up (n = 28): – excluded from analysis no RSA or poor quality, 6 – analyzed, 22 Lost to follow-up, not available n=2 3-year follow-up (n = 26): – excluded from analysis no RSA or poor quality, 3 – analyzed, 23 Lost to follow-up n=0

5-year follow-up (n = 26): – excluded from analysis no RSA or poor quality, 4 – analyzed, 22

Allocated to UHMWPE (n = 29) Received allocated intervention (n = 29): – baseline RSA not usable, 2 – analyzed, 27 Lost to follow-up, reoperated n=1 1-year follow-up (n = 26): – excluded from analysis poor quality, 1 – analyzed, 25 Lost to follow-up, reoperated or not available n=2 3-year follow-up (n = 24): – excluded from analysis no RSA or poor quality, 4 – analyzed, 20 Lost to follow-up, reoperated or not available n=3 5-year follow-up (n = 21): – excluded from analysis no RSA or poor quality, 3 – analyzed, 18

Figure 1. Study enrollment. The reasons for the losses to follow-up—if known—are indicated in the corresponding boxes. HXLPE/VitE: Vitamin E-infused highly cross-linked polyethylene, RSA: radiostereometric analysis, UHMWPE: ultra-high molecular weight polyethylene.

were expressed as the global femoral head penetration rather than separating the 3 vectors as described by Önsten et al. (1998), with radiographs made with the patient in the standing position. Linear motion was our main interest and volumetric wear was not considered. As secondary outcomes, clinical scores were assessed preoperatively as well as postoperatively whenever RSA was performed; they included the Harris Hip Score (HHS) and Merle d’Aubigné and Postel (MAP) score. Radiographically, the mean cup inclination angle on the anteroposterior pelvic radiograph was measured postoperatively and during the follow-up period. PE wear was measured at 1 and 5 years in correlation with the cup inclination angles (ranging from 20º to 60º) and cup sizes (ranging from 48 to 62 mm). Postoperative complications were reported. Statistics The investigation was constructed as a superiority study. Statistical analysis was performed as described previously (Rochcongar et al. 2018). All patients were included in the analysis, regardless of the actual surgery performed, according to the intention-to-treat principle. We used the Mann–Whitney U-test to compare the endpoints and measured the Spearman coefficient to detect

Acta Orthopaedica 2021; 92 (2): 151–155 153

Table 1. Baseline demographics for the 2 patient groups Factor

HXLPE/VitE UHMWPE (n = 33) (n = 29)

Mean age (SD) 61 (6.5) 61 (7.8) Female sex 16 17 BMI (SD) 27 (4.1) 27 (3.7) Indication Primary or secondary osteoarthritis 31 26 Osteonecrosis 2 3 Surgical approach Anterolateral approach (modified Harding) 18 14 Posterior approach (Moore) 13 14 Trochanter osteotomy 2 1 Preoperative clinical scores (SD) Harris Hip Score 52 (11) 53 (12) Merle d’Aubigné and Postel score 13 (1.7) 12 (3) HXLPE/VitE: Vitamin E-infused highly cross-linked polyethylene. UHMWPE: ultra-high molecular weight polyethylene.

a correlation between PE wear and cup inclination angles or cup sizes. For participants who were withdrawn from the study before completion, data from the last observation was carried forward (imputed). All statistical analyses were performed with the statistical software StatView (SAS Institute, Cary, NC, USA). P-values of less than 0.05 were considered as significant. Ethics, registration, funding, and potential conflicts of interest All enrolled patients gave informed consent to participate in the study. This study was approved by the North West French regional ethics committee (Comité de Protection des Personnes Nord-Ouest III; study number: 2009-A00948-49). An independent administrative body (Commission Nationale de l’Informatique et des Libertés) ensured data protection. This study is in compliance with the latest recommendations in the Helsinki Declaration and Public Health Act No. 2004-806. It was registered at clinicaltrials.gov (identifier: NCT02524587). Caen University Hospital received funding from Mathys Ltd Bettlach to finance part of the study costs, which included RSA analysis and data collection. Mathys Ltd Bettlach had no role in the design or execution of the study, the analysis or interpretation of the data, or the decision to submit results. CH was paid speaker for Mathys. No other authors declare any conflict of interest in connection with the submitted article.

Results Patients 494 patients were enrolled in the study between January 2010 and November 2011. After exclusion of 432 patients, 62 of them were randomized into the 2 groups, with 29 patients in the UHMWPE group and 33 patients in the HXLPE/VitE

Table 2. Cumulative femoral head penetration. Values are mean (SD) mm Years after total hip arthroplasty Group 1 3 5 HXLPE/VitE UHMWPE) P-value a

0.16 (0.03) 0.20 (0.05) 0.004

0.20 (0.03) 0.32 (0.07) < 0.001

0.24 (0.04) 0.45 (0.13) < 0.001

HXLPE/VitE: Vitamin E-infused highly cross-linked polyethylene. UHMWPE: ultra-high molecular weight polyethylene. a Mann–Whitney U-test.

Penetration (mm) 0.5 Creep Creep + wear

Wear

UHMWPE

0.4

0.3 HXLPE/VitE p < 0.0001

0.2 p < 0.0001 p < 0.0001 p = 0.004

0.1

0.0

0.5

Years from index operation

Figure 2. Creep and wear behavior of UHMWPE (n = 18, red line) and HXLPE/VitE (n = 22, blue line) over the first 5 years after implantation. HXLPE/VitE: Vitamin E-infused highly cross-linked polyethylene, UHMWPE: ultra-high molecular weight polyethylene.

group (Figure 1). Finally, 40 patients were analyzed at 5 years, which included 22 patients in the HXLPE/VitE group and 18 patients in the UHMWPE group. Most patients underwent THA for osteoarthritis. At baseline, the 2 groups were similar with respect to patient age, sex, BMI, or any of the clinical scores (Table 1). Primary outcome: femoral head penetration 5 years after surgery, the cumulative femoral head penetration was significantly lower in the HXLPE/VitE group than in the UHMWPE group (p < 0.001) (Table 2, Figure 2). From 1 to 5 years after surgery, the mean femoral head penetration increased 0.08 mm in HXLPE/VitE cups, while it increased 0.2 mm in UHMWPE cups. The wear rate averaged 0.02 mm/ year in the HXLPE/VitE group compared with 0.06 mm/year in the UHMWPE group. The estimated steady-state rate of wear was thus approximately 66% lower in the HXLPE/VitE group than in the UHMWPE group (p < 0.001). HXLPE/VitE cups also showed a more even distribution of wear than UHMWPE cups (p = 0.002). At 5 years after surgery, the coefficient of variation, calculated by the standard deviation divided by the mean, was 28 in HXLPE/VitE cups

154

and 18 in UHMWPE cups, which translated to an approximately 50% lower variation of wear in HXLPE/VitE cups than in UHMWPE cups. Secondary outcomes: clinical results and complications From preoperative values to 5 years after surgery, both the HHS and the MAP score improved in both groups (p < 0.001). 5 years after surgery, none of the mean clinical scores differed statistically significantly between the HXLPE/VitE group and the UHMWPE group (HHS 97 [SD 8] versus 99 [SD 3], p = 0.4; MAP score 18 [SD 1] versus 18 [SD 0.4], p = 0.3). At 5 years, the mean cup inclination angle was similar in both groups (HXLPE/VitE 48° [SD 7°], UHMWPE 46° [SD 6°]; p = 0.3) and remained stable over the entire follow-up period (HXLPE/VitE range, 48° to 49°, UHMWPE range, 46° to 48°). Moreover, the difference in PE wear between 1 and 5 years in both groups showed no significant correlation with increasing cup inclination angles (HXLPE/VitE r = 0.2, p = 0.5; UHMWPE r = 0.1, p = 0.8) or cup sizes (HXLPE/VitE r = 0.02, p = 1.0; UHMWPE r = –0.1, p = 0.7). No complications occurred during the first 5 years of follow-up.

Discussion We confirmed continued good clinical and radiographic outcomes with HXLPE/VitE cups at 5 years after THA. From 1 to 5 years, femoral head penetration increased in both groups; however, the wear rate was 66% lower in HXLPE/VitE cups than in UHMWPE cups. This means that the difference in wear rate between the two groups remained highly significant at 5 years (p < 0.001). Based on the linearity of the curves between 1 and 5 years (Figure 2), one would expect the same trend to continue over time, further increasing the difference between the two groups. The mean femoral head penetration of HXLPE/VitE cups, from 1 to 3 years after surgery, increased 0.04 mm compared with 0.12 mm in UHMWPE cups (Rochcongar et al. 2018). Overall, the wear rate averaged 0.020 mm/year in the HXLPE/ VitE group and 0.058 mm/year in the UHMWPE group. The estimated steady-state rate of wear was thus 65% lower in the HXLPE/VitE group than in the UHMWPE group (p < 0.001). The same trend continued at 5 years. PE wear is associated with the onset of osteolysis after THA (Dumbleton et al. 2002, Bitar and Parvizi 2015). According to Elke and Rieker (2018), a wear rate of 0.1 mm/year for any femoral head size correlates with an osteolysis-free survival of less than 20 years. Dowd et al. (2000) found that osteolysis did not develop in patients after THA using non-cross-linked PE devices with a wear rate of < 0.1 mm/year, while Dumbleton et al. (2002) suggested that osteolysis was almost absent with a wear rate of 0.05 mm/year. Our wear rate of 0.02 and the more even distribution of wear in the HXLPE/VitE group is

Acta Orthopaedica 2021; 92 (2): 151–155

therefore promising for clinical outcomes, such as osteolysis over longer follow-up periods. Clinical evidence suggests that cup inclination angles and cup sizes are correlated with PE wear, with inclination angles of ≥ 45º and cup sizes of ≥ 58 mm leading to increased PE wear with time (Bono et al. 1994, Little et al. 2009, Tian et al. 2017 and 2018). Although the ideal cup inclination angle has not yet been established, it is accepted that suboptimal acetabular positioning can lead to accelerated wear (Little et al. 2009, Tian et al. 2017, 2018). In our study, both HXLPE/ VitE and UHMWPE cups showed no significant correlation of the PE wear rate with increasing cup inclination angles. A biomechanical study comparing the wear rates of HXLPE/VitE and UHMWPE cups found that the wear rates of HXLPE/ VitE cups remained similar at standard (45º) and at the highest possible inclination angle (80º), suggesting that HXLPE/ VitE cups accommodate implant malorientation better than UHMWPE cups (Halma et al. 2014). Teeter et al. (2018) also found no correlation between cup inclination angle and PE wear of HXLPE acetabular cups: PE wear of cups within the target inclination angle (40º) and outside of it (47.5º) was the same (0.05 mm/year), suggesting that HXLPE is a forgiving bearing material in terms of wear. Additionally, in our study the PE wear rate showed no statistically significant correlation with increasing cup sizes. In the monoblock implants used, the PE thickness depends on the cup size, which means that PE wear was not correlated with the PE thickness. So far, studies have reported outcomes of vitamin E-stabilized PE liners (Berend et al. 2015, Lindalen et al. 2015, Salemyr et al. 2015, Shareghi et al. 2015, Sillesen et al. 2015, Nebergall et al. 2016, Sillesen et al. 2016, Nebergall et al. 2017, Scemama et al. 2017, Shareghi et al. 2017, Wyatt et al. 2017, Busch et al. 2019, Galea et al. 2019). There are the limitations of our study. We were unable to isolate the effect of vitamin-E stabilization from irradiationinduced cross-linking, because the two cups were built from differently cross-linked PEs. It is known that cross-linking has a beneficial effect on wear, while vitamin E is expected to hinder aging (Lindalen et al. 2015, Salemyr et al. 2015, Nebergall et al. 2016, 2017, Scemama et al. 2017, Galea et al. 2019). However, we assessed none of these effects. Therefore, further clinical trials with identically cross-linked PEs and longer follow-up periods are required to estimate the true benefits of vitamin-E infusion. Overall, 10 patients were lost to follow-up, which can be expected in any long-term followup study. Finally, the small sample size may have limited us in detecting highly significant correlations of PE wear and varying cup inclination angles or cup sizes. In conclusion, the wear rate continues to be lower in HXLPE/VitE cups than in UHMWPE cups at 5 years of follow-up. The steady-state wear rate for HXLPE/VitE cups was more than 5 times below the critical value reported as leading to osteolysis. Additionally, wear rates had no correlation with increasing cup inclination angles or cup sizes. Therefore, this

Acta Orthopaedica 2021; 92 (2): 151–155 155

study confirms that HXLPE/VitE cups have the potential to prevent osteolysis, implant loosening, and eventually revision surgery in the future. All authors contributed equally to this study. The authors would like to thank Medical Minds GmbH for help with english translation. Acta thanks André Busch and Dennis Janssen for help with peer review of this study. Berend K R, Adams J B, Morris M J, Lombardi A V, Jr. Early experience with a new porous hemispheric acetabular component. Surg Technol Int 2015; 27: 263-7. Bitar D, Parvizi J. Biological response to prosthetic debris. World J Orthop 2015; 6(2): 172-89. doi: 10.5312/wjo.v6.i2.172. Bono J V, Sanford L, Toussaint J T. Severe polyethylene wear in total hip arthroplasty. Observations from retrieved AML PLUS hip implants with an ACS polyethylene liner. J Arthroplasty 1994; 9(2): 119-25. Busch A, Jager M, Wegner A, Haversath M. Vitamin E-blended versus conventional polyethylene liners in prostheses: prospective, randomized trial with 3-year follow-up. Orthopade 2019; Nov 6. doi: 10.1007/s00132-019-03830-6. Callary S A, Field J R, Campbell D G. Low wear of a second-generation highly crosslinked polyethylene liner: a 5-year radiostereometric analysis study. Clin Orthop Relat Res 2013; 471(11): 3596-600. doi: 10.1007/ s11999-013-3188-z. Callary S A, Solomon L B, Holubowycz O T, Campbell D G, Munn Z, Howie DW. Wear of highly crosslinked polyethylene acetabular components. Acta Orthop 2015; 86(2): 159-68. doi: 10.3109/17453674.2014.972890. Dowd J E, Sychterz C J, Young A M, Engh C A. Characterization of long-term femoral-head-penetration rates: association with and prediction of osteolysis. J Bone Joint Surg Am 2000; 82-A(8): 1102-7. 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. doi: 10.1054/arth.2002.33664. Elke R, Rieker C B. Estimating the osteolysis-free life of a total hip prosthesis depending on the linear wear rate and head size. Proc Inst Mech Eng H 2018; 232(8): 753-8. doi: 10.1177/0954411918784982. Galea V P, Rojanasopondist P, Laursen M, Muratoglu O K, Malchau H, Bragdon C. Evaluation of vitamin E-diffused highly crosslinked polyethylene wear and porous titanium-coated shell stability: a seven-year randomized control trial using radiostereometric analysis. Bone Joint J 2019; 101-b(7): 760-7. doi: 10.1302/0301-620x.101b7.bjj-2019-0268.r1. Garling E H, Kaptein B L, Geleijns K, Nelissen R G, Valstar E R. Marker configuration model-based roentgen fluoroscopic analysis. J Biomech 2005; 38(4): 893-901. doi: 10.1016/j.jbiomech.2004.04.026. Halma J J, Senaris J, Delfosse D, Lerf R, Oberbach T, van Gaalen S M, de Gast A. Edge loading does not increase wear rates of ceramic-on-ceramic and metal-on-polyethylene articulations. J. Biomed. Mater. Res. Part B Appl. Biomater 2014; 102(8): 1627-38. doi: 10.1002/jbm.b.33147. Lindalen E, Nordsletten L, Hovik O, Rohrl S M. E-vitamin infused highly cross-linked polyethylene: RSA results from a randomised controlled trial using 32 mm and 36 mm ceramic heads. Hip Int 2015; 25(1): 50-5. doi: 10.5301/hipint.5000195. Little N J, Busch C A, Gallagher J A, Rorabeck C H, Bourne R B. Acetabular polyethylene wear and acetabular inclination and femoral offset. Clin Orthop Relat Res 2009; 467(11): 2895-900. doi: 10.1007/s11999-0090845-3.

Nebergall A K, Troelsen A, Rubash H E, Malchau H, Rolfson O, Greene M E. Five-year experience of Vitamin E-diffused highly cross-linked polyethylene wear in total hip arthroplasty assessed by radiostereometric analysis. J. Arthroplasty 2016; 31(6): 1251-5. doi: 10.1016/j.arth.2015.12.023. Nebergall A K, 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. doi: 10.1302/0301620x.99b5.37521. Önsten I, Carlsson A S, Besjakov J. Wear in uncemented porous and cemented polyethylene sockets: a randomised, radiostereometric study. Br J Surg 1998; 80(2): 345-50. Oral E, Greenbaum E S, Malhi A S, Harris W H, Muratoglu O K. Characterization of irradiated blends of alpha-tocopherol and UHMWPE. Biomaterials 2005; 26(33): 6657-63. doi: 10.1016/j.biomaterials.2005.04.026. Oral E, Wannomae K K, Rowell S L, Muratoglu O K. Diffusion of vitamin E in ultra-high molecular weight polyethylene. Biomaterials 2007; 28(35): 5225-37. doi: 10.1016/j.biomaterials.2007.08.025. Pineau V, Lebel B, Gouzy S, Dutheil J J, Vielpeau C. Dual mobility hip arthroplasty wear measurement: experimental accuracy assessment using radiostereometric analysis (RSA). Orthop Traumatol Surg Res 2010; 96(6): 609-15. doi: 10.1016/j.otsr.2010.04.007. 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: a prospective randomized controlled trial. J Bone Joint Surg Am 2018; 100(2): 107-14. doi: 10.2106/jbjs.16.01379. Salemyr M, Muren O, Ahl T, Bodén H, Chammout G, Stark A, Sköldenberg O. Vitamin-E diffused highly cross-linked polyethylene liner compared to standard liners in total hip arthroplasty: a randomized, controlled trial. Int Orthop 2015; 39(8): 1499-505. doi: 10.1007/s00264-015-2680-3. 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): 1113-8. doi: 10.1007/s00264-016-3320-2. Shareghi B, Johanson P E, Karrholm J. Femoral head penetration of Vitamin E-infused highly cross-linked polyethylene liners: a randomized radiostereometric study of seventy hips followed for two years. J Bone Joint Surg Am 2015; 97(16): 1366-71. doi: 10.2106/jbjs.n.00595. Shareghi B, Johanson P E, Karrholm J. Wear of Vitamin E-infused highly cross-linked polyethylene at five years. J Bone Joint Surg Am 2017; 99(17): 1447-52. doi: 10.2106/jbjs.16.00691. Sillesen N H, Greene M E, Nebergall A K, Nielsen P T, Laursen M B, Troelsen A, Malchau H. Three year RSA evaluation of Vitamin E diffused highly cross-linked polyethylene liners and cup stability. J Arthroplasty 2015; 30(7): 1260-4. doi: 10.1016/j.arth.2015.02.018. Sillesen N H, Greene M E, Nebergall A K, Huddleston J I, Emerson R, Gebuhr P, Troelsen A, Malchau H. 3-year follow-up of a long-term registry-based multicentre study on vitamin E diffused polyethylene in total hip replacement. Hip Int 2016; 26(1): 97-103. doi: 10.5301/hipint.5000297. Teeter M G, Lanting B A, Naudie D D, McCalden R W, Howard J L, MacDonald S J. Highly crosslinked polyethylene wear rates and acetabular component orientation: a minimum ten-year follow-up. Bone Joint J 2018; 100-b(7): 891-7. doi: 10.1302/0301-620x.100b7.bjj-2017-1457.r3. Tian J L, Sun L, Hu R Y, Han W, Tian X B. Correlation of cup inclination angle with liner wear for metal-on-polyethylene in hip primary arthroplasty. Orthop Surg 2017; 9(2): 186-90. doi: 10.1111/os.12337. Tian J, Sun L, Hu R, Han W, Tian X. Long-term results of primary hip arthroplasty with cup inclination angle bigger than fifty degrees. J Clin Orthop Trauma 2018; 9(2): 133-6. doi: 10.1016/j.jcot.2017.03.007. Wyatt M, Weidner J, Pfluger D, Beck M. The RM Pressfit vitamys: 5-year Swiss experience of the first 100 cups. Hip Int 2017; 27(4): 368-72. doi: 10.5301/hipint.5000469.

156

Acta Orthopaedica 2021; 92 (2): 156–162

The PROMISE study protocol: a multicenter prospective study of process optimization with interdisciplinary and cross-sectoral care for German patients receiving hip and knee endoprostheses Ulrich BETZ 1, Laura LANGANKI 1, Florian HEID 2, Jan SPIELBERGER 3, Lukas SCHOLLENBERGER 4, Kai KRONFELD 4, Matthias BÜTTNER 5, Britta BÜCHLER 5, Markus GOLDHOFER 6, Lukas ECKHARD 7, and Philipp DREES 7; the PROMISE GROUP a 1 Institute

of Physical Therapy, Prevention and Rehabilitation, University Medical Center of the Johannes Gutenberg University, Mainz; 2 Department of Anaesthesiology, University Medical Center of the Johannes Gutenberg University, Mainz; 3 Department of Anaesthesiology and Intensive Care Medicine, St Josef Hospital, Vienna, Austria; 4 Interdisciplinary Center for Clinical Trials, University Medical Center of the Johannes Gutenberg University, Mainz; 5 Institute of Medical Biostatistics, Epidemiology and Informatics, University Medical Center of the Johannes Gutenberg University, Mainz; 6 Department of Trauma and Orthopedic Surgery, Hunsrück Hospital Kreuznacher Diakonie, Simmern/Hunsrück; 7 Department of Orthopedics and Traumatology, University Medical Center of the Johannes Gutenberg University, Mainz, Germany a Michael CLARIUS, Vulpius Klinik GmbH Bad Rappenau; Manfred KRIEGER, Health and Care Center (GPR) Rüsselsheim; Christoph MEISTER, Health and Care Center (GPR) Rüsselsheim; Birgit AYOSSO, MEDIAN Klaus Miehlke Klinik Wiesbaden; Andreas RITTER VON STOCKERT, MEDIAN Vesalius Klinik, Bad Rappenau; Johannes SCHROETER, MEDIAN Rehabilitation Clinic Aukammtal, Wiesbaden; Andreas SCHWARTING, ACURA Karl-Aschoff Rehabilitation Clinic, Bad Kreuznach; Peter LINDEMER, Ambulatory Rehabilitation and Health Center GmbH, Mainz-Mombach; Jörg WILTINK, Clinic for Psychosomatic Medicine and Psychotherapy, University Medical Center of the Johannes Gutenberg University, Mainz; Christian WERNER, Department of Anesthesiology, University Medical Center of the Johannes Gutenberg University, Mainz; Roland HARDT, Center for General Medicine and Geriatrics, University Medical Center of the Johannes Gutenberg University, Mainz; Stavros KONSTANTINIDES, Center for Thrombosis and Hemostasis (CTH), University Medical Center of the Johannes Gutenberg University, Mainz; Angelika SCHIFFMANN, Care Service, Department for Surgery and Neurosurgery, University Medical Center of the Johannes Gutenberg University, Mainz; Susanne SINGER, Institute of Medical Biostatistics, Epidemiology and Informatics, University Medical Center of the Johannes Gutenberg University, Mainz; Margit SCHMALHOFER, German Rheuma League Rhineland-Palatinate, Bad Kreuznach; Rolf SCHINDEL, Techniker Health Insurance (TK) Rhineland-Palatinate, Mainz; Birgit MEHL, Sandra SOIKE, Department of Orthopedics and Traumatology, University Medical Center of the Johannes Gutenberg University, Mainz; Tobias ENGELMANN, Frank ARNOLD Interdisciplinary Center for Clinical Trials, University Medical Center of the Johannes Gutenberg University, Mainz Correspondence: ulrich.betz@unimedizin-mainz.de Submitted 2020-10-16. Accepted 2020-10-23.

Background and purpose — Knee and hip replacement are common and increasing procedures, and an optimized care process that could be implemented in different settings would be useful. The PROMISE trial investigates whether a new care process works equally in different German settings and how the results compare with current non-standardized care. Patients and methods — This multi-center prospective mixed-method study includes 2,000 German patients receiving arthritis-related hip or knee endoprostheses. An interdisciplinary and cross-sectoral care process was developed and implemented in 3 German hospitals with different levels of care, and corresponding rehabilitation centers were included to bridge the gap after acute care. Duration and outcome — The PROMISE trial recruited patients between May 2018 and March 2020. Follow-up will end in February 2021. Assessments are performed at: examination on clinical indication, 1 week before surgery, on the day of surgery, at the end of hospitalization, end of the rehabilitation program, and 3 months, 6 months, and 12 months after surgery. Outcomes include patient-reported outcomes, medical examination findings, and routinely collected data

regarding the surgery and complications. Guideline-based interviews are conducted with selected patients and care partners. The primary endpoint is the presence of chronic pain at 12 months after surgery. Secondary endpoints are the number of recognized pre-existing conditions, physical activity at 12 months after surgery, use of medical services, quality of life, and interactions between care partners. Trial registration — The trial is registered with the German Clinical Trials Register (https://www.drks.de; DRKS00013972; March 23, 2018).

With demographic changes joint replacement is becoming one of the most frequently performed surgeries. In Germany, approximately 175,000 total hip arthroplasties (THA) (IQTIG 2018a) and 148,000 total knee arthroplasties (TKA) (IQTIG 2018b) are performed each year. These replacement procedures are associated with various risks and complications, including infection and thrombosis, as well as a considerable financial burden. Moreover, 7% to

Acta Orthopaedica 2021; 92 (2): 156–162 157

23% of patients with THA and 10% to 34% of patients with TKA report an unfavorable long-term outcome (Beswick et al. 2012). Due to the high number of these interventions, clinical, patient-centered, and economic results might be improved by even minor advancements in the treatment process. In Germany no generally applicable evidence-based treatment standard has yet been developed. Thus, the current care system does not achieve its full potential, including the cost–benefit ratio. The PROMISE trial aims to improve the care process in Germany based on the principles of the Enhanced Recovery after Surgery Society (ERAS). An ERAS path is defined to optimally prepare the patient for an intervention with minimized stress and stress reactions, maintained homeostasis, and avoiding catabolism that leads to loss of protein, muscle strength, and cellular dysfunction (Ljungqvist 2012). The ERAS approach has been applied to numerous elective procedures and provides approximately 30–50% reductions in the complication rate and length of stay (Ljungqvist et al. 2017). Because of a call for certified intersectoral centers to improve the quality of care (IGES Institut 2016), the PROMISE study involves 3 hospitals with different levels of care, including all required departments, as well as 5 inpatient and outpatient rehabilitation centers. Standard operating procedures defined by an interdisciplinary panel, and a central database for continuous evaluations, audits, and improvements (Ljungqvist et al. 2017) aim to minimize loss of potential effectiveness due to insufficient coordination, lack of common therapeutic goals, incomplete information flow, and separate data collection.

Patients and methods Study design The PROMISE trial is a prospective multi-center mixedmethod study, including patients with an indication to undergo surgery (THA or TKA) at 3 German hospitals (a regional hospital, an orthopedic-specialized hospital, and a tertiary referral university hospital). The PROMISE process involves standardized indication criteria for intervention (Schmitt et al. 2017), preoperative screening for psychological (Gylvin et al. 2016) and geriatric risk factors (Gronewold et al. 2017), blood management (Vaglio et al. 2016), preoperative patient education (Edwards et al. 2017), no preoperative fasting (Smith et al. 2011), postoperative nausea and vomiting prophylaxis (De Oliveira et al. 2013), maximum soft-tissue-sparing techniques (Ljungqvist et al. 2017), intraoperative bleeding and swelling management (Guler et al. 2016, Nielsen et al. 2016), avoidance of suction drainage (Kelly et al. 2014), bladder catheters (Huang et al. 2015), and intravenous catheters (Sharma et al. 2010), local infiltration analgesia (Yun et al. 2015), multimodal oral pain therapy (Khan et al. 2014), starting rehabilitation on the day of surgery (Okamoto et al.

2016), functional discharge criteria (Hansen 2017), and intensified rehabilitation (DRV 2016). All patients are followed for up to 12 months after surgery. Data are collected by medical staff at the time of examination on indication, during patient education, at surgery, at hospital discharge, at the end of rehabilitation, and at 12 months after surgery. Patient-reported outcome measurements are obtained before surgery, at hospital discharge, and at 3 months, 6 months, and 12 months after surgery. Regarding secondary endpoints, a control group will be used consisting of patients from a German health insurance database. Patients will be matched according to age, sex, and diagnosis. To limit the number of survey-based instruments and patient burden, we conducted guideline-based interviews with 10 randomly selected patients to identify aspects that are difficult to capture using a quantitative approach. Interviews have been also conducted with different care partners (e.g., physicians and physiotherapists) to identify and address potential burdens. Study subjects and eligibility criteria Patients indicated for joint replacement due to arthritis of the hip or knee are eligible for study participation. Patients are enrolled if they have met standardized criteria for surgery (Schmitt et al. 2017) and if they are able to understand the nature and extent of the study. Exclusion criteria are: life expectancy less than 1 year (e.g., advanced cancer), any conditions that might preclude elective surgical intervention, and medical or psychological factors that would prevent them from participating or providing informed written consent. Patient withdrawal Patients can withdraw their consent without giving reasons at any time during the trial without disadvantage. No additional data will be collected after that point, although any existing data will remain in the study database. Patients who withdraw are not replaced. Intervention The optimized PROMISE care process consists of interrelated measures (Figure 1). The ERAS Society guidelines (Wainwright et al. 2019) are evidence-based, although we complement those guidelines with non-evidence-based measures to avoid unnecessarily restricting the process optimization. These non-evidence-based measures are derived from a basic understanding of process optimization to avoid stress and promote activity. Indication for surgery The indication for surgery requires appropriate radiological findings, intolerable pain or suffering, restriction of activity, and exhaustion of nonoperative options (Schmitt et al. 2017). The patients define their activity or participation goals throughout the process.

158

Acta Orthopaedica 2021; 92 (2): 156–162

RESPONSIBILITY

WORKFLOW

Surgical outpatient department

Contact by the patient / referring physician

Surgeon

Medical history and clinical examination

Radiologist

Diagnostic imaging if necessary

Surgeon

Indication for surgery

Consultants

Preoperative risk screening; co-care by specialist if required

Patient management

Schedule, blood collection, MRSA screening

Interdisciplinary team

Patient school

Interdisciplinary team

Additional preparations

Surgeon

Prosthesis planning

Surgeon

Information on surgery

Anesthetist

Information on anesthesia

Surgeon

Preparation for surgery

(visit, clinical examination, surgical marking)

Surgeon, anesthetist, functional service

Just before, during and directly after surgery

Interdisciplinary team

Postoperative treatment and rehabilitation

Interdisciplinary team

Discharge from hospital

Rehabilitation team

Rehabilitation centre

Interdisciplinary team

12 months follow-up

(induction of anesthesia, patient positioning, intervention, postoperative monitoring)

Figure 1. Schematic representation of the treatment process. PROMISE-specific procedures are shown in the red boxes. MRSA: methicillinresistant Staphylococcus aureus.

Preoperative screening for concomitant previous conditions Anemia is identified using the laboratory values for hemoglobin, ferritin, transferrin saturation, blood cell counts, red blood cell morphology, glutamic pyruvic transaminase, glutamic oxaloacetic transaminase, bilirubin, C-reactive protein, and reticulocytes. If necessary, iron is administered (orally or intravenously). Screening is performed to identify seniors at risk (ISAR screening) and psychosomatic risk (Patient Health Questionnaire-4 [PHQ-4], Oslo Social Support Scale [OSSS], Somatic Symptom Disorder [SSD], Life-Orientation-Tests [LOT-R]). Preoperative screening for thrombosis and bleeding risk is based on previous deep vein thrombosis or pulmonary embolism, known coagulation disorder/thrombophilia, venous thromboembolism in immediate relatives, previous severe bleeding, cancer within the last 5 years, and varicose veins. Specialized co-care is permitted if required. Patient education Patients are educated preoperatively by all involved professional groups. This process also involves discussing the

patien’s tasks and co-responsibilities. Patients are also motivated to find a personal coach who supports the patient’s active role in the care process. Additional preparations Preoperative preparations include providing the patient with crutches and guidance regarding walking, using stairs, as well as instructions regarding preoperative fitness training. Patients also receive counseling regarding the rehabilitation options and programs (outpatient or inpatient). Immediately before surgery Medication for anxiolysis is avoided if possible. A light meal can be consumed up to 6 hours, and small amounts of clear liquids can be drunk up to 2 hours before anesthesia induction. Etoricoxib (90 mg) is administered 90 minutes and antibiotic (cefazolin 2 g i.v.) 30 minutes before the skin incision, followed by 1 g of tranexamic acid to reduce blood loss. During surgery General anesthesia and spinal anesthesia can be used. A single intraoperative steroid injection (e.g., 20 mg of dexamethasone or 125 mg of methylprednisolone) is used to attenuate the surgical stress response, improve analgesia, and reduce postoperative nausea and vomiting. The surgical technique aims to minimize trauma using the smallest possible incisions and natural muscle gaps. To prevent dislocation, the implanted THA must tolerate at least 30° internal rotation at 90° flexion and at least 70° external rotation in 0° extension. Local infiltration analgesia involves administration of ropivacaine (0.2%, 150 mL) with adrenaline (0.5mg/150mL) and an additional 50 mL of plain ropivacaine (0.2%) subcutaneously. Use of suction drainage and bladder/intravenous catheters is avoided whenever possible. Rehabilitation There are generally no restrictions regarding load bearing and range of motion. The patient begins independently leaving their bed and walking with crutches on the day of surgery, with support from the nursing staff or physiotherapist as necessary to achieve independent basic functions (changing position, personal hygiene, dressing and undressing, and eating at a table). Thromboprophylaxis is performed using low molecular weight heparins in weight- and risk-adapted dosage starting on the first postoperative day. In compliance with national guidelines thromboprophylaxis is continued for 35 (THA) or 14 (TKA) days postoperatively. Opioid use is minimized, and basic medication is used during days 3–8 (etoricoxib at 90 mg orally in the morning and metamizole 4 x 1 g). After day 8, the only medication is etoricoxib (90 mg orally in the morning). The patient is discharged to home or a rehabilitation facility when the wound is dry, pain is tolerable, and they can independently perform basic functions (change position, personal hygiene, dressing and undressing, walking > 150 m,

Acta Orthopaedica 2021; 92 (2): 156–162 159

and climb 10 steps). Patients who are discharged home also complete rehabilitation programs that comply with the rehabilitation therapy standards of the German Pension Insurance (DRV 2016). Endpoints Primary endpoint The primary endpoint is presence of chronic pain at 12 months after surgery, evaluated with the pain subscore from the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). Approximately 9% of patients report pain at 12 months after surgery (Beswick et al. 2012) and the evaluation is planned using a 2-sided binomial test. The WOMAC is calculated using the hip osteoarthritis outcome score (HOOS) (Klässbo et al. 2003) or the knee osteoarthritis outcome score (KOOS) (Roos et al. 2003). Secondary endpoints The secondary endpoints are number of recognized preexisting conditions (as proxy we use pre-existing anemia) at baseline, physical activity at 12 months after surgery, use of medical services during the 12 months after surgery, quality of life at 12 months after surgery, and interactions between care partners during the 12 months after surgery. Physical activity is evaluated using the HOOS/KOOS subscore HOOSPS/KOOS-PS (physical function short form) for THA/TKA cases, with data collected before the surgery and at 3 months, 6 months, and 12 months after the operation. Quality of life at 12 months after surgery is assessed using patient-reported scores from the EQ-5D-5L (Conner-Spady et al. 2015) and the EQ-5D visual analogue scale (VAS) on a 0–100 numeric scale. This data is also collected before surgery and at 3 months, 6 months, and 12 months after surgery. Type and frequency of use of medical service during the 12 months after surgery is evaluated using a questionnaire. The results will be compared with results from a control group of patients from a German health insurance database who are matched according to age, sex, and diagnosis/treatment history. Statistical power for those analyses will be improved by matching 1 PROMISE participant to 4 control individuals. Cost analyses will also be performed to compare the PROMISE and control groups, to identify potential savings. Interactions between care partners during the first 12 months after surgery are also considered to determine whether a stable plateau is reached in the number of intersectoral interactions for each patient. The interviews will be evaluated according to the qualitative content analysis described by Mayring (2000). Data collection Data is collected using an electronic case report form, and the data types, collection processes, and entry processes were jointly developed by the PROMISE group to ensure all participating institutions and functional areas are working to the same standard.

Data is recorded and stored pseudonymized in a central electronic database at a specialized interdisciplinary center. During the trial, the investigators and trial site staff receive system documentation, training, and support for the use of the database. Time-point specific questionnaires and forms are completed by the patient, or the attending physician/physiotherapist, or both. Data collection points (Table) are: 1. baseline visit, 2. 1 week before surgery, 3. day of surgery, 4. hospital discharge, 5. rehabilitation center discharge, 6. 3 months (±2 weeks) after hospital discharge, 7. 6 months (±2 weeks) after hospital discharge, 8. 12 months (±2 weeks) after hospital discharge, 9. and 12 months (±4 weeks) after hospitalization (clinical follow-up). Baseline data includes age, sex, comorbidities, initial diagnosis according to International Classification of Diseases (version 10), and socioeconomic data. The other collected data includes: • hip disability (HOOS), • knee disability (KOOS), • patient health quality (PHQ-4/SSD), • quality of life (EQ-5D-5L), • social support (Oslo Social Support Scale [OSSS]), • personal expectations (Hospital for Special Surgery [HSS] score), • rehabilitation success (Staffelstein score), • and optimism/pessimism (Life Orientation Test-Revised [LOT-R]). Data quality assurance On-site qualified study nurses ensure that all personnel understand the trial and follow its protocol, including adhering to the standardized surgery and postoperative procedures. Surgeries are performed at centers that are qualified for the Enhanced Recovery process. A steering committee supervises the trial progress and provides guidance to all participating treatment groups. Furthermore, the electronic database includes an interface to import quality-assured data that is sent by the participating hospitals to the Institute for Quality and Patient Safety. Estimated sample size and power The estimated sample size is based on the primary endpoint (chronic pain at 12 months), which will be evaluated using a 2-sided binomial test. A previous report has indicated that approximately 9% of patients report chronic pain after 1 year (Beswick et al. 2012). Thus, 1,900 patients are required to detect a 20% reduction (i.e., from 9% to 7.2%) based on an α level of 5% and power of 80%. Assuming 5% of patients will be lost to follow-up, the target sample size is 2,000 participants. This sample size also provides power of > 80% for

160

Acta Orthopaedica 2021; 92 (2): 156–162

Data collection, time-points, and instruments Item/score

A B C D E F G

Basic information X Comorbidities a X Staffelstein X X X X Indication X ASA X Preop. anemia assessment X HOOS/KOOS a X X EQ-5D 5L a X X X X PHQ-4/SSD a X X X X X OSSS, LOT-R a X HSS or INDICATE knee (preop) a X HSS or INDICATE knee (postop) a X ISAR a X Timed up & go and VAS/NRS X X X X Socioeconomic data a X X b Cost book a X X Functional goal a X Surgery X Implant details X Functional milestones and discharge X Complications X Rehabilitation X Thrombosis/ bleeding X X Follow-up a X b Clinical follow-up X A. Indication, 8 weeks preoperatively B. Preoperative assessment, 1 week preoperatively C. Operation D. Hospital stay E. End of rehabilitation F. Follow-up at 3, 6, and 12 months postoperatively G. Clinical follow-up at 12 months postoperatively a Patient-reported outcome. b Only at the 12-month follow-up ASA: American Society of Anesthesiologists score, HOOS: Hip disability and Osteoarthritis Outcome Score, KOOS: Knee injury and Osteoarthritis Outcome Score, EQ-5D 5L: EuroQol Group 5-level EQ-5D version, PHQ-4: Patient Health Questionnaire-4, SSD: somatic symptom disorders, OSSS: Online Social Support Scale, LOT-R: Life Orientation Test-Revised, HSS: Hospital for Special Surgery, ISAR: International Society of Arthroplasty Registries, VAS: visual analogue scale, NRS: numerical rating scale.

analyzing the 2 confirmatory secondary endpoints (number of anemia cases and physical activity after 1 year; values selected from the literature), which will be analyzed only if a significant improvement is observed in the primary endpoint. The secondary analyses are based on non-inferiority tests and previously reported values. The number of anemia cases will be tested to determine whether it is ≤ 2% below the expected reference value of 15.5% for Central Europe (McLean et al. 2009). Physical activity will be tested using the HOOS-PS/KOOS-PS to determine whether the patients achieve at least the minimum clinically important improve-

ment (23 points) (Paulsen et al. 2014). To comply with the α-error, confirmatory testing for these 2 secondary endpoints will be performed only if statistically significant improvement is observed for the primary endpoint. To evaluate the achievement of a plateau in the number of interactions between care partners, the implementation period is divided into 3 intervals. The number of interactions in intervals II and III will be tested for equality. The criterion for equality will be based on the difference between the 1st and 2nd intervals. Poisson models will be used to compare the numbers of healthcare interactions between PROMISE participants and the control group, with sub-analyses according to different cost categories (e.g., hospital stays vs. general practitioner visits). Factors that potentially influence healthcare utilization, interactions, and quality of life will be evaluated using hierarchical regression models with care partners considered as clusters. Ethics, registration, funding, conflicts of interests, and result presentation The trial is conducted in accordance with the latest versions of the Declaration of Helsinki, Good Epidemiological Practice, and local regulatory requirements, including the German Federal Data Protection Act (Bundesdatenschutzgesetz). The protocol was approved by the ethics committees of Rhineland-Palatinate [837.533.17 (11367)], Baden-Wuerttemberg [B-F-2018-042], and Hessen [MC 84/2018]. The protocol is registered with the German Clinical Trials Register (DRKS00013972). Written informed consent is obtained from all patients before enrolment. The trial is supported by a grant from the Federal Joint Committee (01NVF16015). None of the authors declare any conflicts of interest. Results will be presented at congresses and published in a peer-reviewed medical journal according to the guidelines of the International Committee of Medical Journal Editors. Study start and duration The 1st patient was recruited in May 2018, the last patient was recruited in March 2020, and follow-up will end in February 2021.

Discussion Evidence-based optimized perioperative processes have already been described for hip and knee replacements (Ibrahim et al. 2013, Sprowson et al. 2013). These programs can reduce the length of stay to 1–3 days (Kehlet 2013, Khan et al. 2014) without an increased risk of complications or readmission (Glassou et al. 2014), and have also reduced the 2-year mortality rate from 3.8% to 2.7% (Savaridas et al. 2013). However, most studies regarding process optimization for hip and knee replacement were performed in Scandinavia, the UK, and North America, which complicates generalizations to

Acta Orthopaedica 2021; 92 (2): 156–162 161

other countries with different healthcare systems. For example, Germany has more hospital beds per capita that Denmark (8 vs. 2.6 beds/1,000 population) (OECD 2017), the hospital reimbursement system is based on diagnosis-related groups, and insured individuals have a legal right to complete a rehabilitation program. Moreover, the median length of stay is 10 days in Germany (IQTIG 2018a, 2018b), versus only 2 days in Denmark (Husted et al. 2016, Knæalloplastikregister 2017). Thus, the setting in Germany can vary considerably from the setting in previous studies that examined optimized care processes, and we are not aware of any similar studies that were performed in Germany. The PROMISE trial is also designed to account for the fact that German patients participate in a rehabilitation program after their hospitalization. Therefore, we hope this prospective multi-center trial will help improve outcomes in the post-hospitalization period by incorporating hospitals that provide different levels of care as well as outpatient and inpatient rehabilitation centers. The resulting information may guide the development of an optimized process that accounts for the patient’s characteristics, expectations, and quality of life, as well as the relevant stress factors, functional outcomes, and economic costs up to 1 year after surgery. The strengths of the study we see in a prospective, multi-centric, intersectoral design, a relatively high number of participants, a very broad data collection at numerous points of time, an above-average follow-up period and an independent external evaluation. Unfortunately, it is not feasible to organize an optimized and a common treatment path at a single center, which intends randomization of the participants to an intervention and control group. We will at least partially address this issue by comparing the results with those from a matched cohort of control patients from a German health insurance database. Parameters that we cannot directly compare between the PROMISE and control groups will be evaluated based on previously published results. Also disadvantageous is that German funding agencies select the rehabilitation facility and we cannot ensure that all participants complete their rehabilitation in PROMISE-affiliated facilities. Although this may reduce the number of participants who complete the full PROMISE process, it will allow us to analyze statistically the benefits of cross-sectoral care. All authors are members of the PROMISE working group, which developed and wrote the study protocol together. UB is the main author and responsible for the introduction and discussion, LL, FH, JS, MG, LE, and PhD have described the intervention, LS and KK wrote the sections on general project description, BB and MB wrote the sections on data collection and evaluation. All authors reviewed the entire text.

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.

Conner-Spady B L, Marshall D A, Bohm E, Dunbar M J, Loucks L, Khudairy A A, Noseworthy T W. Reliability and validity of the EQ-5D-5L compared to the EQ-5D-3L in patients with osteoarthritis referred for hip and knee replacement. Qual Life Res 2015; 24(7): 1775-84. doi: 10.1007/s11136-014-0910-6. De Oliveira G S, Jr, Castro-Alves L J, Ahmad S, Kendall M C, McCarthy R J. Dexamethasone to prevent postoperative nausea and vomiting: an updated meta-analysis of randomized controlled trials. Anesth Analg 2013; 116(1): 58-74. doi: 10.1213/ANE.0b013e31826f0a0a. DRV. Deutsche Rentenversicherung. Reha Therapiestandards Hüft-und Knie-TEP 2016; Available from: http://www.deutsche-rentenversicherung.de/Allgemein/de/Navigation/3_Infos_fuer_Experten/01_Sozialmedizin_Forschung/02_reha_qualitaetssicherung/reha_therapiestandards/ indikationen/rts_hueft_knie_tep_node.html. (accessed September 25, 2020). Edwards P K, Mears S C, Lowry Barnes C. Preoperative education for hip and knee replacement: never stop learning. Curr Rev Musculoskelet Med 2017; 10(3): 356-64. doi: 10.1007/s12178-017-9417-4. Glassou E N, Pedersen A, Hansen T. Risk of re-admission, reoperation, and mortality within 90 days of total hip and knee arthroplasty in fast-track departments in Denmark from 2005 to 2011. Acta Orthop 2014; 85: 1-8. doi: 10.3109/17453674.2014.942586. Gronewold J, Dahlmann C, Jäger M, Hermann D M. Identification of hospitalized elderly patients at risk for adverse in-hospital outcomes in a university orthopedics and trauma surgery environment. PLoS One 2017; 12(11): e0187801-e. doi: 10.1371/journal.pone.0187801. Guler O, Mahirogullari M, Isyar M, Piskin A, Yalcin S, Mutlu S, Sahin B. Comparison of quadriceps muscle volume after unilateral total knee arthroplasty with and without tourniquet use. Knee Surg Sports Traumatol Arthrosc 2016; 24(8): 2595-605. doi: 10.1007/s00167-015-3872-5. Gylvin S H, Jørgensen C C, Fink-Jensen A, Kehlet H. Psychiatric disease as a risk factor in fast-track hip and knee replacement. Acta Orthop 2016; 87(5): 439-43. doi: 10.3109/17453674.2016.1151292. Hansen T B. Fast track in hip arthroplasty. EFORT Open Rev 2017; 2(5): 17988. doi: 10.1302/2058-5241.2.160060. Huang Z, Ma J, Shen B, Pei F. General anesthesia: to catheterize or not? A prospective randomized controlled study of patients undergoing total knee arthroplasty. J Arthroplasty 2015; 30(3): 502-6. doi: 10.1016/j. arth.2014.09.028. Husted H, Jorgensen C C, Gromov K, Kehlet H. Does BMI influence hospital stay and morbidity after fast-track hip and knee arthroplasty? Acta Orthop 2016; 87(5): 466-72. doi: 10.1080/17453674.2016.1203477. Ibrahim M S, Khan M A, Nizam I, Haddad F S. Peri-operative interventions producing better functional outcomes and enhanced recovery following total hip and knee arthroplasty: an evidence-based review. BMC Med 2013; 11:37-. doi: 10.1186/1741-7015-11-37. IGES-Institut. Infrastruktur und Gesundheit. Weißbuch Gelenkersatz – Versorgungssituation bei endoprothetischen Hüft- und Knieeingriffen in Deutschland 2016. Available from: https://www.iges.com/sites/iges.de/ myzms/content/e6666/e13520/e14425/e14438/e14439/attr_objs14757/ IGES_PK_Gelenkersatz_Praesentation_Haussler_07062016.pdf_ger.pdf (accessed September 25, 2020). IQTIG. Institut für Qualitätssicherung und Transparenz im Gesundheitswesen. Bundesauswertung zum Erfassungsjahr 2017. Hüftendoprothesenversorgung. Qualitätsindikatoren 2018a. Available from: https://www.qs-nrw. org/app/report/doc/bund_2017_hep.pdf. Day of access: September 25, 2020 IQTIG. Institut für Qualitätssicherung und Transparenz im Gesundheitswesen. Bundesauswertung zum Erfassungsjahr 2017. Knieendoprothesenversorgung.Qualitätsindikatoren 2018b. Available from: https://iqtig.org/ downloads/auswertung/2017/kep/QSKH_KEP_2017_BUAW_V02_201808-01.pdf (accessed September 25, 2020). Kehlet H. Fast-track hip and knee arthroplasty. Lancet 2013; 381(9878): 1600-2. doi: https://doi.org/10.1016/S0140-6736(13)61003-X. Kelly E G, Cashman JP, Imran F H, Conroy R, O’Byrne J. Systematic review and meta-analysis of closed suction drainage versus non-drainage in primary hip arthroplasty. Surg Technol Int 2014; 24: 295-301.

162

Khan S K, Malviya A, Muller S D, Carluke I, Partington P F, Emmerson K P, Reed M R. Reduced short-term complications and mortality following Enhanced Recovery primary hip and knee arthroplasty: results from 6,000 consecutive procedures. Acta Orthop 2014; 85(1): 26-31. doi: 10.3109/17453674.2013.874925. Klässbo M, Larsson E, Mannevik E. Hip disability and osteoarthritis outcome score: an extension of the Western Ontario and McMaster Universities Osteoarthritis Index. Scand J Rheumatol 2003; 32(1): 46-51. doi: 10.1080/03009740310000409. Knæalloplastikregister D. Årsrapport 2017; 2017. Available from: 2020https:// www.sundhed.dk/content/cms/99/4699_dkr-aarsrapport-2019_tiloffentliggoerelse.pdf (accessed September 25, 2020). Ljungqvist O. Jonathan E. Rhodes lecture 2011: Insulin resistance and enhanced recovery after surgery. JPEN J Parenter Enteral Nutr 2012; 36(4): 389-98. doi: 10.1177/0148607112445580. Ljungqvist O, Scott M, Fearon K C. Enhanced recovery after surgery: a review. JAMA Surg 2017; 152(3): 292-8. doi: 10.1001/jamasurg.2016.4952. Mayring P. Qualitative Inhaltsanalyse. Grundlagen und Techniken Weinheim: Deutscher Studien Verlag; 2000. McLean E, Cogswell M, Egli I, Wojdyla D, de Benoist B. Worldwide prevalence of anaemia, WHO Vitamin and Mineral Nutrition Information System, 1993–2005. Public Health Nutr 2009; 12(4): 444-54. doi: 10.1017/ s1368980008002401. Nielsen C S, Jans Ø, Ørsnes T, Foss N B, Troelsen A, Husted H. combined intra-articular and intravenous tranexamic acid reduces blood loss in total knee arthroplasty: a randomized, double-blind, placebo-controlled trial. J Bone Joint Surg Am 2016; 98(10): 835-41. doi: 10.2106/jbjs.15.00810. OECD. Hospital beds 2017. Available from: https://data.oecd.org/healtheqt/ hospital-beds.htm (accessed September 25, 2020). Okamoto T, Ridley R J, Edmondston S J, Visser M, Headford J, Yates P J. Dayof-surgery mobilization reduces the length of stay after elective hip arthroplasty. J Arthroplasty 2016; 31(10): 2227-30. doi: 10.1016/j.arth.2016.03.066. Paulsen A, Roos E M, Pedersen A B, Overgaard S. Minimal clinically important improvement (MCII) and patient-acceptable symptom state (PASS) in total hip arthroplasty (THA) patients 1 year postoperatively. Acta Orthop 2014; 85(1): 39-48. doi: 10.3109/17453674.2013.867782. Roos E M, Toksvig-Larsen S. Knee injury and Osteoarthritis Outcome Score (KOOS): validation and comparison to the WOMAC in total knee replacement. Health Qual Life Outcomes 2003; 1: 17. doi: 10.1186/1477-7525-1-17.

Acta Orthopaedica 2021; 92 (2): 156–162

Savaridas T, Serrano-Pedraza I, Khan S K, Martin K, Malviya A, Reed M R. Reduced medium-term mortality following primary total hip and knee arthroplasty with an enhanced recovery program: a study of 4,500 consecutive procedures. Acta Orthop 2013; 84(1): 40-3. doi: 10.3109/ 17453674.2013.771298. Schmitt J, Lange T, Günther K P, Kopkow C, Rataj E, Apfelbacher C, Aringer M, Böhle E, Bork H, Dreinhöfer K, Friederich N, Frosch K H, Gravius S, Gromnica-Ihle E, Heller K D, Kirschner S, Kladny B, Kohlhof H, Kremer M, Leuchten N, Lippmann M, Malzahn J, Meyer H, Sabatowski R, Scharf H P, Stoeve J, Wagner R, Lützner J. Indication criteria for total knee arthroplasty in patients with osteoarthritis: a multiperspective consensus study. Z Orthop Unfall 2017; 155(5): 539-48. doi: 10.1055/s-0043-115120. Sharma S, Iorio R, Specht L M, Davies-Lepie S, Healy W L. Complications of femoral nerve block for total knee arthroplasty. Clin Orthop Relat Res 2010; 468(1): 135-40. doi: 10.1007/s11999-009-1025-1. Smith I, Kranke P, Murat I, Smith A, O’Sullivan G, Søreide E, Spies C, in’t Veld B. Perioperative fasting in adults and children: guidelines from the European Society of Anaesthesiology. Eur J Anaesthesiol 2011; 28(8): 556-69. doi: 10.1097/EJA.0b013e3283495ba1. Sozialgesetzbuch V 2020. Fünftes Buch (SGB V), Gesetzliche Krankenversicherung. Available from: https://www.sozialgesetzbuch-sgb.de/sgbv/110a. html (accessed October 12, 2020). Sprowson A, McNamara I, Manktelow A. (v) Enhanced recovery pathway in hip and knee arthroplasty: “fast track” rehabilitation. Orthop Trauma 2013; 27(5): 296-302. doi: https://doi.org/10.1016/j.mporth.2013.09.002. Vaglio S, Prisco D, Biancofiore G, Rafanelli D, Antonioli P, Lisanti M, Andreani L, Basso L, Velati C, Grazzini G, Liumbruno G M. Recommendations for the implementation of a Patient Blood Management programme. Application to elective major orthopaedic surgery in adults. Blood Transfus 2016; 14(1): 23-65. doi: 10.2450/2015.0172-15. Wainwright T, Gill M, McDonald D, Middleton R, Reed M, Sahota O, Yates P, Ljungqvist O. Consensus statement for perioperative care in total hip replacement and total knee replacement surgery: Enhanced Recovery After Surgery (ERAS®) Society recommendations. Acta Orthop 2019; 91 (1): 3-19. doi 10.1080/17453674.2019.1683790 Yun X D, Yin X L, Jiang J, Teng Y J, Dong H T, An L P, Xia Y Y. Local infiltration analgesia versus femoral nerve block in total knee arthroplasty: a meta-analysis. Orthop Traumatol Surg Res 2015; 101(5): 565-9. doi: 10.1016/j.otsr.2015.03.015.

Acta Orthopaedica 2021; 92 (2): 163–169

163

Mortality and revision risk after femoral neck fracture: comparison of internal fixation for undisplaced fracture with arthroplasty for displaced fracture: a population-based study from Danish National Registries Bjarke VIBERG 1,2,4, Trine FRØSLEV 3, Søren OVERGAARD 4,5, Alma Becic PEDERSEN 3 1 Department

of Orthopaedic Surgery and Traumatology, Lillebaelt Hospital – University Hospital of Southern Denmark, Kolding; 2 Department of Regional Health Research, University of Southern Denmark, Odense; 3 Department of Clinical Epidemiology, Aarhus University Hospital; 4 Orthopaedics Research Unit, Department of Orthopaedic Surgery and Traumatology, Odense University Hospital, Odense; 5 Department of Clinical Research, University of Southern Denmark, Odense, Denmark Correspondence: bjarke.viberg@rsyd.dk Submitted 2020-07-03. Accepted 2020-10-27.

Background and purpose — Hemiarthroplasty has lower reoperation frequency and better mobilization compared with internal fixation (IF) in patients with undisplaced femoral neck fractures (FNF), which might translate into lower mortality. In this population-based cohort study we compare the risk of mortality and reoperation in undisplaced FNF treated with IF and displaced FNF treated with arthroplasty in patients older than 70 years old. We assume that, per se, there is no difference in mortality risk between patients with a displaced and an undisplaced FNF. Patients and methods — Hip fracture patients were identified in the Danish Multidisciplinary Hip Fracture Registry during 2005–2015. Data on medication, comorbidities, reoperation, and mortality were retrieved from other Danish medical databases. IF and arthroplasty patients were compared with regards to mortality and reoperation up to 5 years postoperatively. We calculated hazard ratios (HR) with 95 % confidence intervals (CI) adjusting for relevant confounders. Results — We included 19,260 FNF treated with arthroplasty and 10,337 FNF with IF. There was an increased risk of mortality for arthroplasty within 30 days, HR 1.3 (95% CI 1.3–1.4), compared with IF but not after 1 and 5 years. Arthroplasty patients had adjusted HRs for reoperation of 0.8 (0.8–0.9) within 1 year, 0.8 (0.7–0.9) within 2 years, and 0.8 (0.8–0.9) within 5 years postoperatively compared with IF. Interpretation — Patients treated for a displaced FNF with arthroplasty had a higher risk of 30-day mortality compared with patients who had an undisplaced FNF treated with IF. It has to be considered that there were baseline differences in the groups but there was no difference in mortality risk up to 5 years post-surgery. Concerning reoperation, patients with a displaced FNF treated with arthroplasty had a lower risk of reoperation compared with IF for undisplaced FNF.

The general consensus on treating an undisplaced femoral neck fracture (FNF) with internal fixation (IF) (Dansk Sygeplejeråd et al. 2008, National Institute for Health and Care Excellence 2011, updated 2017) has been questioned by a recent meta-analysis demonstrating that treatment with hemiarthroplasty may reduce the relative risk of reoperation by 70% when compared with IF (Richards et al. 2020). The meta-analysis included 2 randomized clinical trials (RCTs) that demonstrated a 5% reoperation frequency in the hemiarthroplasty group compared with 20–21% in the IF group (Lu et al. 2017, Dolatowski et al. 2019). There may be a lower reoperation rate but neither of the 2 RCTs (Lu et al. 2017, Dolatowski et al. 2019) found a difference in patient-reported outcome after 1 year. Dolatowski et al. (2019) did find a faster mobility (Timed-Up-And-Go) in the hemiarthroplasty (HA) group and better mobilization is also found when comparing arthroplasty with IF in displaced FNF (Gjertsen et al. 2008, Jiang et al. 2015). Better mobilization after hip fracture is important as it is associated with reduced mortality after surgery (Kristensen et al. 2016). Therefore, an arthroplasty for the undisplaced FNF may also reduce the mortality compared with IF. However, the current 2 RCTs were not large enough to address the mortality question. A difference in mortality could be investigated by assuming that there is no difference per se in mortality risk between patients with a displaced and undisplaced FNF. This assumption is supported by studies comparing IF for undisplaced FNF with arthroplasty for displaced FNF showing no difference in mortality (Mukka et al. 2020, Richards et al. 2020). We therefore compared the mortality and reoperation after treatment of patients operated on by IF for undisplaced FNF compared with arthroplasty for a displaced FNF in patients ≥ 70 years old.

164

Patients and method Study design This is a population-based cohort study on patients with a FNF during 2005–2015 (both years included) and reported to the Danish Multidisciplinary Hip Fracture Registry. The age cutoff of 70 years was used since the Danish guideline recommends arthroplasty for patients with a displaced FNF who are older than 70 years as well as surgery for all hip fracture patients (Dansk Sygeplejeråd et al. 2008). Reporting is performed according to the RECORD extension to the STROBE guidelines (Benchimol et al. 2015). Setting Denmark has approximately 5.8 million inhabitants and every Danish citizen is at birth issued a 10-digit Civil Personal Register number. This number allows unambiguous linkage between all Danish medical databases, and every person can therefore be traced until death or emigration. All Danish citizens are guaranteed free healthcare for any hospital treatment through the Danish National Health Service, which is why all patients will be treated in Denmark (Schmidt et al. 2014). Data sources The Danish Multidisciplinary Hip Fracture Registry is a population-based clinical-quality database. Collecting data and reporting is mandatory for all hospital units treating hip fracture patients. A number of preoperative and perioperative data are prospectively collected including data on quality of inpatient care (Mainz et al. 2004). Data from the Danish Multidisciplinary Hip Fracture Registry is collected from the Danish National Patient Registry for Charlson comorbidity index (CCI) and reoperations (Dansk Tværfagligt Register for Hoftenære Lårbensbrud 2017) and holds data on all hospital contacts including data on all surgical procedure dates and codes according to the Nordic Medico-Statistical Committee classification (Nordic Medico-Statistical Committee 2012). It is not mandatory to report BMI, which may explain reduced completeness. The completeness of the Danish National Patient Registry is considered to be 99.7% (Schmidt et al. 2015) and the positive predictive value of the hip fracture diagnosis is as high as 98% (Nymark et al. 2006, Hudson et al. 2013). Data from the Danish Multidisciplinary Hip Fracture Registry was also linked to the Danish Civil Registration System for vital status and migration for the entire Danish population (Schmidt et al. 2014). The Danish National Health Service Prescription Database has information on all prescriptions for reimbursed drugs dispensed by community pharmacies in Denmark and is recorded according to the Anatomical Therapeutic Chemical classification system (Johannesdottir et al. 2012). Study population We used the Danish Multidisciplinary Hip Fracture Regis-

Acta Orthopaedica 2021; 92 (2): 163–169

try to identify the study population. All patients admitted to a hospital in Denmark with a hip fracture diagnosis code (ICD-10 DS720, DS721, DS722), surgical procedure code, and laterality were included. Using procedure codes (Nordic Medico-Statistical Committee 2012) the patients were categorized into an internal fixation or arthroplasty group. Internal fixation was defined as screw fixation or sliding hip screw and arthroplasty as hemiarthroplasty or THA (Table 1, see Supplementary data). If patient had bilateral hip fracture, only the first hip fracture was included in the study population. In the Danish Multidisciplinary Hip Fracture Registry, a code for undisplaced and displaced fracture exists. However, not all patients have the code and it has not been validated. Our national guidelines recommend arthroplasty for all displaced FNF patients above 70 years and IF for all undisplaced FNF. All patients above 70 years old treated with IF were therefore deemed to have an undisplaced FNF and those with arthroplasty as a displaced FNF. Outcome Mortality was registered by date after surgery and collected from the Danish Civil Registration System. The follow-up for all patients was from surgery date to a maximum of 5 years or until end of follow-up. Reoperation was defined as any open procedure: deep infection, change of implant, open reduction, or operation due to a new (i.e., periprosthetic) fracture (Tables 2 and 3, see Supplementary data). Variables A priori, we identified potential confounders including age, sex, CCI, BMI, and medication. Age, sex, and BMI were retrieved from the Danish Multidisciplinary Hip Fracture Registry. From the Danish National Patient Registry (Schmidt et al. 2015) we retrieved information on comorbidity measured by CCI (Charlson et al. 1987) using discharge diagnoses up to 10 years prior to hip fracture operation. Both primary and secondary diagnoses, as well as diagnoses of inpatient and outpatient visits, were included. From the Danish National Health Service Prescription Database (Johannesdottir et al. 2012) data on several prescription medicine was retrieved that the author group a priori defined as potential confounders. NSAID, glucocorticoids, opioid, and antibiotics data was retrieved using prescriptions reimbursed within 90 days before operation. Data for antihypertensives, antidepressants, statins, and anticoagulants was retrieved using prescriptions reimbursed within 365 days before operation due to the larger packages of prescribed medicine. Age was categorized as 65–74, 75–84, and ≥ 85 years old. Using the WHO classification, patients were categorized as underweight if BMI was < 18.5, normal weight if BMI was ≤ 18.5–24.9, overweight if BMI was 25–29.9, and obese if BMI was ≥ 30 (WHO 2000). 3 comorbidity levels were defined using the CCI score: 0 (none), 1–2 (low), 3 or more (high).

Acta Orthopaedica 2021; 92 (2): 163–169

Medication use was categorized using dichotomous values (yes/no) at baseline into NSAIDs, antihypertensives, glucocorticoids, antidepressants, statins, anticoagulants, opioids, and antibiotics. Bias In Denmark, the commonly used approach to the hip is posterior; only 1 hospital uses the anterolateral approach routinely for patients with FNF. The posterior approach is associated with higher reoperation frequency compared with the lateral approach (van der Sijp et al. 2018) thereby possibly diminishing a difference in reoperation frequency between arthroplasty and IF. Study size The 1-year mortality was the primary outcome, which for hip fracture patients in Denmark is approximately 27% (Danish Multidisciplinary Registry for Hip Fracture 2019). We estimated a 2% difference between the 2 groups. A 2-sample proportion sample-size calculation was therefore performed using respectively 26% mortality in the arthroplasty group and 28% mortality in the IF group. This yielded a sample size of 7,734 in each group, using 0.05 for alpha and 0.80 for power. Data access, linkage, and cleaning methods The authors had complete access to data from the Danish Multidisciplinary Registry for Hip Fractures, the Danish National Patient Registry, the Danish National Database of Reimbursed Prescriptions, and the Danish Civil Registration System. This study draws on individual-level record linkage of data from nationwide medical registries using the unique Civil Personal Register number. The study was approved by the Danish Data Protection Agency (journal number 1-16-02-467-15) and the Danish Patient Safety Authority (case number 3-3013-1389/1). Statistics The study population was divided into patients with an undisplaced FNF treated with IF and patients with a displaced FNF treated with an arthroplasty. We describe the study population according to the distribution of patients’ characteristics, tabulating the number and percentage of patients. We used the Kaplan–Meier method to compute the mortality risk after surgery. Crude and adjusted Cox proportional hazards models were used to assess the choice of surgery impact on subsequent mortality up to 5 years after surgery. We adjusted for age at time of surgery, sex, BMI, comorbidity level, and medication, inclusive of NSAIDs, corticosteroids, antidepressants, opioids, and for reoperation the latter was included in the model as time-dependent variable. In the survival analyses of reoperation, patients were followed from the date of surgery to reoperation, death or end of study period. We plotted cumulative incidence curves for reoperation for the 2 groups, using death as a competing risk. Crude and adjusted proportional sub-distribution hazards

165

Proximal femoral fractures from Danish Multidisciplinary Registry for Hip Fractures Excluded: Pertrochanteric and subtrochanteric fractures Femoral neck fracures Excluded: Osteosynthesis other than screws and sliding hip screws, arthroplasty with erroneous coding Internal fixation for undisplaced, hemiarthroplasty or total hip arthroplasty for displaced femoral neck fractures Excluded: < 70 years old or operated in 2004 or 2016 Internal fixation for undisplaced, hemiarthroplasty or total hip arthroplasty for displaced femoral neck fractures in patients ≥ 70 years old and operated in 2005–2015

Figure 1. Flowchart of workup from the Danish Multidisciplinary Registry for Hip Fractures to study population of patients above 70 years treated with IF for undisplaced FNF and arthroplasty for displaced FNF.

models accounting for competing risk of death (Fine and Gray 1999) were tested for assessing the choice of surgery impact on subsequent reoperation risk for different time intervals after surgery. They were not different from the Cox proportional hazards models, which were therefore applied instead. The HRs for reoperation were adjusted for age at time of surgery, sex, BMI, comorbidity level, and medication, inclusive of NSAIDs, corticosteroids, antidepressants, and opioids. All hazard ratios (HR) were calculated with 95% confidence intervals (CI). All data management and analyses were conducted in SAS 9.4 (SAS Institute, Cary, NC, USA). Ethics, funding, and potential conflicts of interest This study has no direct implications for FNF patients. No funding was received for this study. There are no conflicts of interest related to this study. BV has, outside the study, received payment for lectures from Zimmer Biomet and Osmedic Swemac. SO has received a grant for research from Zimmer Biomet. AP and TF have no disclosures to declare.

Results During 2005–2015, there were 80,760 hip fracture operations in Denmark, of which 29,597 were in patients with an FNF over 70 years old treated with IF or arthroplasty (Figure 1). There were 10,337 patients with undisplaced FNF treated with IF and 19,260 displaced FNF treated with arthroplasty. The patient group with undisplaced FNF treated with IF were

166

Acta Orthopaedica 2021; 92 (2): 163–169

Table 4. Demographic data of the study population and divided by type of osteosynthesis. Values are count (%) Factor

Overall study Undisplaced Displaced FNF population FNF – IF – arthroplasty n = 29,597 n = 10,337 n = 19,260

Age at surgery 70–74 3,861 (13) 1,639 (16) 2,222 (12) 75–84 12,969 (44) 4,487 (43) 8,482 (44) ≥ 85 12,767 (43) 4,211 (41) 8,556 (44) Female sex 21,601 (73) 7,226 (70 14,375 (75) Charlson Comorbidity Index 0, none 11,901 (40) 3,998 (39) 7,903 (41) 1–2, low 12,061 (41) 4,280 (41) 7,781 (40) ≥ 3, high 5,635 (19) 2,059 (20) 3,576 (19) BMI Missing 6,066 (21) 2,231 (22) 3,835 (20) < 18.5 2,321 (7.8) 911 (8.8) 1,410 (7.3) 18.5–24 13,734 (46) 4,712 (46) 9,022 (47) 25–29 5,999 (20) 1,980 (19) 4,019 (21) ≥ 30 1,477 (5.0) 503 (4.9) 974 (5.1) Medication NSAID 2,993 (10) 1,091 (11) 1,902 (9.9) Antihypertensives 20,878 (71) 7,025 (68) 13,853 (72) Glucocorticoids 1,252 (4.2) 443 (4.3) 809 (4.2) Antidepressants 8,437 (29) 3,030 (29) 5,407 (28) Statins 5,451 (18) 1,730 (17) 3,721 (19) Anticoagulants 14,716 (50) 5,077 (49) 9,639 (50) Opioids 2,168 (7.3) 737 (7.1) 1,431 (7.4) Antibiotics 7,148 (24) 2,499 (24) 4,649 (24)

Table 5. Demographic data: patients with arthroplasty divided into hemiarthroplasty and total hip arthroplasty. Values are count (%) Factor

Hemiarthroplasty Total hip arthroplasty n = 16,437 n = 2,823

Age at surgery 70–74 1,597 (9.7 625 (22) 75–84 7,174 (44) 1,308 (46) ≥ 85 7,666 (47) 890 (32) Female sex 12,343 (75) 2,032 (72) Charlson Comorbidity Index 0, none 6,596 (40) 1,307 (46) 1–2, low 6,735 (41) 1,046 (37) ≥ 3, high 3,106 (19) 470 (17) BMI Missing 3,425 (21) 410 (15) < 18.5 1,206 (7.3) 204 (7.2) 18.5–24 4,712 (46) 9,022 (49) 25–29 7,656 (47) 1,366 (48) ≥ 30 3,340 (20) 679 (24) Medication NSAID 1,592 (9.7) 310 (11) Antihypertensives 11,866 (72) 1,987 (70) Glucocorticoids 676 (4.1) 133 (4.7) Antidepressants 5,446 (33) 840 (30) Statins 4,164 (25) 798 (28) Anticoagulants 8,315 (51) 1,324 (47) Opioids 737 (7.1) 1,431 (7.4) Antibiotics 3,984 (24) 665 (24)

FNF, femoral neck fracture. IF, internal fixation.

Table 6. Adjusted regression analysis for mortality concerning internal fixation (IF) for undisplaced FNF versus arthroplasty, hemiarthroplasty and total hip arthroplasty (THA) for displaced FNF Person Mortality Deaths time, Mortality Type of surgery n years risk (%) Day 0–30 IF Arthroplasty Hemiarthroplasty THA Year 0–1 IF Arthroplasty Hemiarthroplasty THA Year 0–5 IF Arthroplasty Hemiarthroplasty THA

908 2,017 1,817 200

805 1,480 1,259 222

2,980 5,354 4,773 581 6,249 11,089 9,764 1,325

Adjusted HR (95% CI) a

8.8 11 11 7.1

Reference 1.26 (1.17–1.37) 1.29 (1.19–1.40) 1.04 (0.89–1.22)

8,264 15,304 12,907 2,397

29 28 29 21

Reference 1.00 (0.96–1.05) 1.02 (0.98–1.07) 0.85 (0.77–0.93)

27,787 50,981 42,541 8,439

61 58 59 47

Reference 0.98 (0.95–1.01) 1.00 (0.97–1.03) 0.85 (0.80–0.90)

CI, confidence interval. FNF, femoral neck fracture, HR, hazard ratio. a Hazard ratios adjusted for age, sex, BMI, reoperation as a time dependent variable, comorbidity level, and medication, inclusive of NSAIDs, corticosteroids, antidepressants, opioids.

slightly younger, contained more males, and had a higher comorbidity level compared with the arthroplasty group (Table 4). The arthroplasty group consisted of 16,437 hemiarthroplasties and 2,823 THA. Patients receiving THA were younger and had a lower comorbidity level compared with hemiarthroplasty patients (Table 5). Mortality Within 30 days after surgery, the mortality was 11% in the arthroplasty group and 8.8% in the IF group. This corresponds to an adjusted HR of 1.3 (CI 1.2–1.4) for arthroplasty, 1.3 (1.2–1.4) for hemiarthroplasty while it was 1.0 (0.9–1.2) for THA when compared with IF (Figure 2 and Table 6). The mortality within 1 year was 28% in the arthroplasty group and 29% in the IF group. This corresponds to an adjusted HR of 1.0 (1.0–1.1) within 1 year for arthroplasty, 1.0 (1.0–1.1) for hemiarthroplasty, while it was 0.9 (0.8–0.9) for THA when compared with IF (Figure 2 and Table 6). Reoperation Within 1 year after surgery, the reoperation rate was 7.5% in the arthroplasty group and 9.3% in the IF group. This corresponds to an adjusted HR of 0.8 (0.8–0.9) when comparing arthroplasty with IF (Figure 3 and Table 7). Both hemiarthroplasty and THA had a lower reoperation risk than IF with similar results even after 2 and 5 years (Table 7).

Acta Orthopaedica 2021; 92 (2): 163–169

Table 7. Adjusted regression analysis for reoperation concerning internal fixation (IF) for undisplaced FNF versus arthroplasty, hemiarthroplasty and total hip arthroplasty (THA) for displaced FNF

167

KM patient survival

Cumulative revision rate

1.0

0.20 Arthroplasty Internal fixation

Arthroplasty Internal fixation

0.8

Follow-up Type of surgery

Cumulative incidence (%)

Year 0–1 IF Arthroplasty Hemiarthroplasty THA Year 0–2 IF Arthroplasty Hemiarthroplasty THA Year 0–5 IF Arthroplasty Hemiarthroplasty THA

Adjusted HR (95% CI) a

0.15

0.6 0.10

9.3 7.5 7.9 5.0

Reference 0.82 (0.75–0.89) 0.87 (0.80–0.95) 0.52 (0.44–0.62)

11 8.5 8.9 6.2

Reference 0.79 (0.73–0.85) 0.84 (0.77–0.90) 0.55 (0.46–0.64)

13 11 11 8.9

Reference 0.85 (0.79–0.91) 0.88 (0.82–0.95) 0.64 (0.56–0.74)

0.4 0.05 0.2

Years after index operation

Figure 2. Kaplan–Meier survival plot with 95% CI comparing internal fixation for undisplaced fracture and arthroplasty for displaced fracture.

CI, confidence interval. FNF, femoral neck fracture, HR, hazard ratio. a Hazard ratios adjusted for age, sex, BMI, comorbidity level, and medication, inclusive of NSAIDs, corticosteroids, antidepressants, opioids.

Discussion We assumed that there is no difference per se in mortality risk between patients with a displaced and undisplaced FNF. Patients receiving an arthroplasty for displaced FNF had a higher mortality after 30 days (11% vs. 8.8%) but not after 1 and 5 years compared with patients treated with IF for undisplaced FNF. However, patients with arthroplasty had, after 1 year, a 7.5% reoperation frequency compared with 9.3% in the IF group. The higher risk of 30-day mortality in the arthroplasty group may be due to selection bias. We see a baseline difference with higher age, more males, and higher comorbidity level in the hemiarthroplasty group compared with IF. All these factors are associated with higher mortality, thereby influencing the 30-day result. We also see a substantially lower age in the smaller THA group (15% of the arthroplasties), thereby demonstrating that THA is performed in primarily healthier patients. Another factor concerning the baseline differences could be bone cement implantation syndrome when using an arthroplasty, leading to increased perioperative mortality (Middleton et al. 2014). A possible confounder could be the degree of posterior tilt (posterior angulation of the femoral head in comparison with the femoral neck) introduced in 2009 (Palm et al. 2009). If an undisplaced FNF has more than 20 degrees’ posterior tilt, then the risk of reoperation when using IF is increased. This could lead to surgical bias, as a surgeon might be more prone to choose an IF for a 72-year-

Years after index operation

Figure 3. Cumulative incidence of reoperation for any reason over time of internal fixation for undisplaced fracture and arthroplasty for displaced fracture with 95% CI.

old patient with a 20-degree posterior tilt compared with an 82-year-old patient. This could explain some of the baseline difference, which, however, could also be due to an underlying confounding in the population sustaining a displaced or undisplaced FNF, but previous studies have not found a major difference between groups (Mukka et al. 2020, Richards et al. 2020). There is, however, no difference in the 1-year mortality between hemiarthroplasty and IF despite the baseline difference. This could be due to better mobilization, less pain, and lower reoperation frequency in patients treated with hemi arthroplasty. We found a 1-year reoperation frequency of 7.5% in the arthroplasty group compared with 9.3% in the IF group. The arthroplasty reoperation frequency is nearly double compared with the Norwegian and Swedish registers (Rikshöft 2017, Norwegian Hip Fracture Register 2018). This is probably due to the exclusive use in Denmark of the posterior approach, which is associated with higher reoperation frequency compared with the lateral and anterolateral approach (van der Sijp et al. 2018) that is primarily used in Norway and Sweden (Rikshöft 2017, Norwegian Hip Fracture Register 2018). By shifting to a different approach in Denmark the reoperation frequency could be lowered (Sköldenberg et al. 2010) and thereby perhaps mortality too. There has been 1 similar study published but it focuses on reoperation frequency and compares hemiarthroplasty with IF (Gjertsen et al. 2011). It demonstrates a reoperation frequency of 11% for IF in undisplaced FNF, which is higher in this study, but they included all reoperations including simple removal of implant, which we excluded. They report only 3% reoperation for hemiarthroplasty for displaced FNF, which is much lower than our 7.5% but could be due to the surgical approach. That study also demonstrated that hemiarthroplasty had the lowest degree of pain, and that patients were most satisfied and reported the highest quality of life. Our sub-

168

analysis comparing hemiarthroplasty with THA confirms this, as the 1-year mortality is 29% in the hemiarthroplasty group and 21% in the THA group. By not including THA, selection bias occurs when assessing mortality as some of the healthiest FNF patients receive a THA. This is an important aspect in our study. The RCT by Dolotowski et al. (2019) demonstrated a lower 24-month mortality in the hemiarthroplasty group but was not sufficiently powered to show a statistically significant difference. An ongoing Swedish RCT will give us more knowledge on mortality as it has calculated sample size on a composite variable of reoperation and mortality (Wolf et al. 2020). Concerning limitations there are, as discussed, baseline differences that may be due to the earlier explanations but there could also be unmeasured confounding or confounding by indication. We did not use the code for displaced/undisplaced fracture due to missing data and lack of validation, thereby not knowing the number of patients with a displaced FNF who were treated with IF and undisplaced FNF with arthroplasty. In addition, we did not measure the radiographs for displaced/ undisplaced, posterior tilt, and implant positioning, which could all have an impact on the results. However, we do not believe that these measures would have a great impact as the study uses the results of everyday practice. We do not have functional or patient-reported outcomes that potentially could demonstrate a difference. The study has a major strength in using national registries with complete follow-up that can follow patients on an individual level across the country. We have included several possible biases for adjustment such as comorbidity, medication, and BMI. Another strength is the large cohort, thereby enabling the possibility to demonstrate small differences. In conclusion, patients treated for a displaced FNF with arthroplasty had a higher risk of 30-day mortality compared with patients who had an undisplaced FNF treated with IF. It has to be considered that there were baseline differences in the groups but there was no difference in mortality risk up to 5 years post-surgery. Concerning reoperation, patients with a displaced FNF treated with arthroplasty had a lower risk of reoperation compared with IF for undisplaced FNF. Supplementary data Tables 1–3 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.20 20.1850940 BV, SO, and HP conceptualized the research idea and method, while TF conducted data curation and formal analyses. BV wrote the initial draft but all author performed critical review and accepted the final draft. Acta thanks Frede Frihagen and Olof Wolf for help with peer review of this study.

Acta Orthopaedica 2021; 92 (2): 163–169

Benchimol E I, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, Sorensen H T, von Elm E, Langan S M, Committee R W. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med 2015; 12(10): e1001885. doi: 10.1371/ journal.pmed.1001885. Charlson M E, Pompei P, Ales K L, MacKenzie C R. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40(5): 373-83. Danish Multidisciplinary Registry for Hip Fracture. National Annual Report, 2018. Danish Healthcare Service; 2019. p 1-149. Dansk Sygeplejeråd, Danske Fysioterapeuter, Dansk Ortopædisk Selskab. Referenceprogram for Patienter med Hoftebrud; 2008. Dansk Tværfagligt Register for Hoftenære Lårbensbrud. National Årsrapport; 2017. www.sundhed.dk; 2017. p 149. Dolatowski F C, Frihagen F, Bartels S, Opland V, Saltyte Benth J, Talsnes O, Hoelsbrekken S E, Utvag S E. Screw fixation versus hemiarthroplasty for nondisplaced femoral neck fractures in elderly patients: a multicenter randomized controlled trial. J Bone Joint Surg Am 2019; 101(2): 136-44. doi: 10.2106/JBJS.18.00316. Fine J, Gray R J. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: 496-509. Gjertsen J E, Vinje T, Lie S A, Engesaeter L B, Havelin L I, Furnes O, Fevang J M. Patient satisfaction, pain, and quality of life 4 months after displaced femoral neck fractures: a comparison of 663 fractures treated with internal fixation and 906 with bipolar hemiarthroplasty reported to the Norwegian Hip Fracture Register. Acta Orthop 2008; 79(5): 594-601. doi: 10.1080/17453670810016597. Gjertsen J E, Fevang J M, Matre K, Vinje T, Engesaeter L B. Clinical outcome after undisplaced femoral neck fractures. Acta Orthop 2011; 82(3): 268-74. doi: 10.3109/17453674.2011.588857. Hudson M, Avina-Zubieta A, Lacaille D, Bernatsky S, Lix L, Jean S. The validity of administrative data to identify hip fractures is high: a systematic review. J Clin Epidemiol 2013; 66(3): 278-85. doi: 10.1016/j. jclinepi.2012.10.004. Jiang J, Yang C H, Lin Q, Yun X D, Xia Y Y. Does arthroplasty provide better outcomes than internal fixation at mid- and long-term followup? A meta-analysis. Clin Orthop Relat Res 2015; 473(8): 2672-9. doi: 10.1007/ s11999-015-4345-3. Johannesdottir S A, Horvath-Puho E, Ehrenstein V, Schmidt M, Pedersen L, Sorensen H T. Existing data sources for clinical epidemiology: the Danish National Database of Reimbursed Prescriptions. Clin Epidemiol 2012; 4: 303-13. doi: 10.2147/CLEP.S37587. Kristensen P K, Thillemann T M, Soballe K, Johnsen S P. Are process performance measures associated with clinical outcomes among patients with hip fractures? A population-based cohort study. Int J Qual Health Care 2016; 28(6): 698-708. doi: 10.1093/intqhc/mzw093. Lu Q, Tang G, Zhao X, Guo S, Cai B, Li Q. Hemiarthroplasty versus internal fixation in super-aged patients with undisplaced femoral neck fractures: a 5-year follow-up of randomized controlled trial. Arch Orthop Trauma Surg 2017; 137(1): 27-35. doi: 10.1007/s00402-016-2591-9. Mainz J, Krog B R, Bjornshave B, Bartels P. Nationwide continuous quality improvement using clinical indicators: the Danish National Indicator Project. Int J Qual Health Care 2004; 16 (Suppl. 1): i45-50. doi: 10.1093/ intqhc/mzh031. Middleton R G, Uzoigwe C E, Young P S, Smith R, Gosal H S, Holt G. Perioperative mortality after hemiarthroplasty for fracture of the hip: does cement make a difference? Bone Joint J 2014; 96-B(9): 1185-91. doi: 10.1302/0301-620X.96B9.33935. Mukka S, Sjoholm P, Aziz A, Eisler T, Kadum B, Krupic F, Morberg P, SayedNoor A. A cohort study comparing internal fixation for undisplaced versus hip arthroplasty for displaced femoral neck fracture in the elderly: a pilot study for a clinical trial. Pilot Feasibility Stud 2020; 6: 98. doi: 10.1186/ s40814-020-00642-w. National Institute for Health and Care Excellence. Hip fracture: management. Clinical guideline [CG124]. National Institute for Health and Care Excellence; 2011, updated 2017.

Acta Orthopaedica 2021; 92 (2): 163–169

Norwegian Hip Fracture Register. Annual Report; 2018. www.nrlweb.ihelse. net › eng › Rapporter › Report2018_english Nordic Medico-Statistical Committee. NOMESCO Classification of Surgical Procedures; 2012. Nymark T, Lauritsen J M, Ovesen O, Rock N D, Jeune B. Decreasing incidence of hip fracture in the Funen County, Denmark. Acta Orthop 2006; 77(1): 109-13. doi: 10.1080/17453670610045777. Palm H, Gosvig K, Krasheninnikoff M, Jacobsen S, Gebuhr P. A new measurement for posterior tilt predicts reoperation in undisplaced femoral neck fractures: 113 consecutive patients treated by internal fixation and followed for 1 year. Acta Orthop 2009; 80(3): 303-7. doi: 10.3109/17453670902967281. Richards J T, Overmann A L, O’Hara N N, D’Alleyrand J C, Slobogean G P. Internal fixation versus arthroplasty for the treatment of nondisplaced femoral neck fractures in the elderly: a systematic review and meta-analysis. J Orthop Trauma 2020; 34(1): 42-8. doi: 10.1097/BOT.0000000000001656. Rikshöft. Annual report; 2017. www.rikshöft.se/forsking-och-utvärdering Schmidt M, Pedersen L, Sorensen H T. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 2014; 29(8): 541-9. doi: 10.1007/s10654-014-9930-3.

169

Schmidt M, Schmidt S A, Sandegaard J L, Ehrenstein V, Pedersen L, Sorensen H T. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7: 449-90. doi: 10.2147/ CLEP.S91125. Sköldenberg O, Ekman A, Salemyr M, Bodén H. Reduced dislocation rate after hip arthroplasty for femoral neck fractures when changing from posterolateral to anterolateral approach. Acta Orthop 2010; 81(5): 583-7. doi: 10.3109/17453674.2010.519170. van der Sijp M P L, van Delft D, Krijnen P, Niggebrugge A H P, Schipper I B. Surgical approaches and hemiarthroplasty outcomes for femoral neck fractures: a meta-analysis. J Arthroplasty 2018; 33(5): 1617-27 e9. doi: 10.1016/j.arth.2017.12.029. WHO. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000; 894: i-xii, 1-253. Wolf O, Sjöholm P, Hailer N P, Möller M, Mukka S. Study protocol: HipSTHeR—a register-based randomised controlled trial—hip screws or (total) hip replacement for undisplaced femoral neck fractures in older patients. BMC Geriatr 2020; 20(1): 19. doi: 10.1186/s12877-020-1418-2.

170

Acta Orthopaedica 2021; 92 (2): 170–175

Influence of day of surgery and prediction of LOS > 2 days after fasttrack hip and knee replacement Christoffer C JØRGENSEN 1,2, Kirill GROMOV 2,3, Pelle B PETERSEN 1,2, and Henrik KEHLET 1,2, on behalf of the Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Collaborative Group a 1 Section for Surgical Pathophysiology, Rigshospitalet, Copenhagen; 2 Lundbeck Foundation Centre for Fast-track Hip and Knee Arthroplasty; 3 Department of Orthopaedic Surgery, Clinical Orthopaedic Research Hvidovre (CORH), Copenhagen University Hospital Hvidovre, Denmark

Correspondence: christoffer.calov.joergensen@regionh.dk Submitted 2020-08-28. Accepted 2020-10-17. a Members

of the Lundbeck Foundation Centre for Fast-track Hip and Knee collaborative group: Frank Madsen, Department of Orthopedics, Aarhus University Hospital, Aarhus. Torben B Hansen, Department of Orthopedics, Regional Hospital Holstebro and University of Aarhus, Holstebro. Mogens Laursen, Aalborg University Hospital Northern Orthopaedic Division, Aalborg. Lars T Hansen, Sydvestjysk Hospital Grindsted. Per Kjærsgaard-Andersen, Department of Orthopedics, Vejle Hospital, Vejle. Soren Solgaard, Department of Orthopedics, Gentofte University Hospital, Copenhagen. Niels Harry Krarup, Department of Orthopedics, Viborg Hospital, Viborg. Jens Bagger, Department of Orthopaedic Surgery, Copenhagen University Hospital Bispebjerg, Copenhagen NV, Denmark.

Background and purpose — Enhanced recovery programs have reduced length of stay (LOS) after hip and knee arthroplasty (THA/TKA). Although risk factors disposing to prolonged LOS are well documented, there is limited information on the role of weekday of surgery. This study analyzed the role of weekday of surgery and other potential risk factors for LOS > 2 days. Patients and methods — We included 10,576 unselected consecutive procedures between January 2016 and August 2017 within a multicenter fast-track THA/TKA collaboration with prospective collection of preoperative characteristics. We used multiple regression analysis of potential risk factors for LOS > 2 days followed by construction of a simple risk score from 0 to 15 points based on the calculated odds ratios. Results — Mean LOS was 1.9 (SD 1.8) days, with 80% of patients having surgery from Monday to Wednesday. Of these, 17% (95% CI 16–18) had a LOS > 2 days vs. 19% (CI 17–21) in those operated on Thursday and Friday. Patients were scheduled evenly throughout the week regardless of risk of LOS > 2 days and despite the fact that 38% (CI 35–40) of patients with ≥ 6 points (16% of the total population) had a LOS > 2 days compared with 14% (CI 13–14) in those with < 6 points. In these “high-risk” patients, the fraction with LOS > 2 days increased when having surgery on Thursdays or Fridays (43% CI 38–49) compared with Monday to Wednesday (37% CI 34–39). Interpretation — A detailed preoperative risk assessment may be helpful to plan the weekday of surgery in order to decrease LOS and weekend hospitalization.

Recent developments in perioperative care have also led to enhanced recovery (ERAS) in hip (THA) and knee replacement (TKA) with a decrease in postoperative length of stay (LOS) to between 0 and 2 days in many centers (Wainwright and Kehlet 2019). These advances have led to several studies showing the feasibility of outpatient THA and TKA in selected patients (Vehmeijer et al. 2018). The positive effects of ERAS programs in THA and TKA remain indisputable, not only by reducing LOS, but also by lowering the risk of medical complications without an increase in readmissions (Wainwright and Kehlet 2019). However, challenges still exist to further improve outcome and where the strategy must be divided between optimizing preoperative comorbidities, perioperative care, and organizational issues, of which the latter has received less attention. Preoperative risk factors have been well assessed over several years in relation to short LOS and generally confirming increased age, obesity, diabetes, cardio-pulmonary diseases, and dependent functional status as risk factors for prolonged LOS (Jørgensen et al. 2016, Cram et al. 2018, Kim et al. 2018, Cizmic et al. 2019, Gkagkalis et al. 2019, Ziemba-Davis et al. 2019, Johnson et al. 2020). However, limited information is available from a fast-track setting on the role of weekday of surgery on LOS when adjusting for the above-mentioned risk factors. In this context, identification of patients unlikely to be discharged within 1–2 days and therefore to be scheduled for surgery at the start of the week may reduce the need for weekend hospitalization and transfer to other departments from otherwise well-functioning 5-day arthroplasty units or ambulatory arthroplasty centers.

Acta Orthopaedica 2021; 92 (2): 170–175 171

THA/TKA in the Danish National Patient Registry January 2010 to August 2017 n = 41,292 Excluded (n = 3,348): – fractures, non-elective procedures or previous surgery on the same joint within 90 days, 2,319 – severe congenital disorders, infection or cancer, 172 – simultaneous or staged bilateral procedures within 90 days, 816 – other (e.g., non-Danish citizens, revision surgery, and hemiarthroplasties), 41

Primary elective unilateral THA/TKA n = 37,944 Excluded (n = 27,368): – not in the Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement database, 1,009 – procedure performed prior to 2016, 26,359 Procedures performed between January 2016 and August 2017 n = 10,576 Excluded: missing data, n = 589 Procedures included in the multiple regression analysis of risk factors for length of hospital stay > 2 days n = 9,987

Figure 1. Study population. THA – total hip arthroplasty, TKA – total knee arthroplasty.

The Lundbeck Foundation Centre for Fast-track Hip and Knee Arthroplasty (www.fthk.dk) was founded as a multiinstitutional collaboration to improve care and outcome after THA and TKA and the most recent high-volume data have shown a median LOS of only 1 day in unselected patients (Petersen et al. 2019). The present study is a specific analysis of the role of preoperative risk factors and the weekday of surgery for a LOS > 2 days within the multicenter fast-track THA and TKA collaboration, in which unselected patients have been assessed in detail preoperatively and with complete registration of LOS in a socialized healthcare system where common practice in Denmark is discharge to home (Petersen et al. 2019).

Patients and methods Study design This was a descriptive multicenter cohort study based on the Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Database (LCDB). The LCDB is a prospective database that records information on patient characteristics using patient-reported questionnaires. These are completed within 1 month prior to surgery with assistance from staff if necessary and a completeness of about 95% (Jørgensen et al. 2016). Currently 9 different departments report to the LCDB, all of which are dedicated arthroplasty units with > 300 annual procedures and similar fast-track protocols including preference for spinal anesthesia, multimodal opioid-sparing anal-

gesia, in-hospital only thrombo-prophylaxis when LOS ≤ 5 days and early mobilization (≤ 6 hours postoperatively). Highdose methylprednisolone (125 mg) is standard in TKA and increasingly used in THA, while peripheral nerve blocks are only used at the discretion of the attending anesthesiologist. Readiness for discharge is evaluated using similar standardized functional discharge criteria including being able to get in and out of bed/chair, walking independently with an aid, and performing daily activities (Husted et al. 2010). There are no selection criteria for being included in the fast-track protocol as it is considered the standard of care in all participating departments (Petersen et al. 2019). Data from the LCDB are cross-referenced with the Danish National Patient Registry, which records all hospitalizations in Denmark regardless of geographic location. The accuracy of the DNPR with regards to capturing admissions is > 99%, while the accuracy of individual diagnostic codes varies (Schmidt et al. 2015). Patients We included consecutive unselected unilateral THA and TKA from the LCDB between January 2016 and August 2017, excluding patients with age < 18 years, simultaneous bilateral procedures, procedures due to cancer or severe congenital disorder, and patients with major surgery on the lower extremities before or after 90 days from the elective procedure (Figure 1). Statistics Comparison of categorical data was done using a chi-square test and reporting 95% confidence intervals (CI) for proportions. When testing for statistical significance a p-value of < 0.05 was used. Multiple logistic regression analysis was used for evaluating association between risk factors, weekday of surgery, and risk of LOS > 2 days. We included variables based on acyclic graphs to avoid mediation (Shrier and Platt 2008) and based on previous works (Jørgensen and Kehlet 2013b, Jørgensen et al. 2016). Finally, place of surgery was included as a random effect. We used complete case analysis as missing data was limited to about 5%. When evaluating the results of the risk score we included patients excluded from the regression model when possible. Model accuracy was reported as 83.3% using the Generalized Linear Mixed Model function in SPSS v. 25 (IBM Corp, Armonk, NY, USA). When constructing a risk score for LOS > 2 days we assigned points for each significant variable in the regression model depending on the calculated ORs. Variables with an OR between 1 and 1.8 were assigned 1 point, ORs 1.9–2.9 were assigned 2 points and ORs > 2.9 were assigned 3 points. Thus, a score ranging from 0 to 15 was constructed (see Results section). We evaluated the fit of the risk score using an ROC curve and based a cut-off for high-risk patients based on the calculated sensitivity and specificity, as well as consideration of clinical relevance.

172

Acta Orthopaedica 2021; 92 (2): 170–175

Table 1. Patient characteristics Values are count (%) unless otherwise specified

Monday Tuesday

Factor Age median (IQR) < 50 50–60 61–65 66–70 71–75 76–80 81–85 > 85 BMI median (IQR) < 18.5 18.5–24.9 25–29.9 30–34.9 35–39.9 > 39.9 Missing Female sex Walking aid Missing Living alone In institution Missing Smoking Missing Alcohol > 24 g/day Missing Total knee arthroplasty Psychiatric disease Missing Cardiac disease Missing Pulmonary disease Missing Hypertension Missing Diabetes mellitus Non-insulin dependent Insulin dependent Missing Anticoagulants Missing Preoperative anemia a Missing a

n (%) 70 (14) 514 (4.9) 1,671 (16) 1,384 (13) 2,055 (19) 2,218 (21) 1,639 (16) 806 (7.6) 289 (2.7) 27.4 (6.5) 94 (0.9) 4,110 (39) 2,968 (28) 2,237 (21) 788 (7.5) 279 (2.7) 100 (0.9) 6,213 (59) 2,391 (23) 195 (1.8) 3,632 (35) 75 (0.7) 80 (0.8) 1,347 (13) 89 (0.8) 790 (7.5) 91 (0.9) 4,448 (42) 1,520 (14) 0 (0.0) 1,439 (14) 106 (1.0) 945 (9.0) 64 (0.6) 5,911 (56) 0 (0.0) 912 (8.8) 199 (1.9) 61 (0.6) 821 (7.8) 0 (0.0) 2,522 (24) 155 (1.5)

Hemoglobin < 13 g/dL.

Ethics, registration, funding, and potential conflicts of interest As this is a non-interventional study, ethical approval was waived, but permission to collect and store data was obtained from the Danish Patient Safety authority (3-3013-56/2/EMJO) and the Danish Data Protection Agency (2012-58-0004). The LCDB is registered on clinicaltrials.gov (NCT01515670) as an ongoing registry study on outcomes after fast-track THA and TKA. The LCDB has been funded by a grant from the Lundbeck Foundation (R25-A2702). The Lundbeck Foundation is independent of the Lundbeck Pharmaceutical company and had no influence on any part of the study design or writing of the manuscript.

Wednesday Thursday Friday Saturday day of surgery

Sunday

day of discharge

Distribution according to weekday (%)

Figure 2. Distribution of procedures and day of discharge according to weekday. There were 11 (0.1%) procedures on Saturday and Sunday, all in patients with < 6 points.

CJ and PB declare no conflicts of interest. KG and HK are members of the board on “Rapid Recovery” by Zimmer Biomet.

Results Of 10,576 procedures, 9,987 (94%) could be included in the multiple regression analysis (Figure 1 and Table 1). Median LOS was 2 (IQR 1) days and mean 1.9 day (SD 1.8). The distribution of weekday of surgery was left skewed with > 50% of all procedures being performed on Mondays (28%) and Tuesdays (30%). Only 14% and 6% of procedures were performed on Thursdays and Fridays, respectively. Finally, there were 11 (0.1%) procedures performed during the weekend, all in patients later found to be at low risk (score < 6, see below) of having LOS > 2 days. These patients were subsequently excluded from the weekday analysis as they would probably have been selected to be performed on special personal indication (Figure 2). The number of THAs or TKAs performed on Monday to Wednesday (79% vs. 80%) or Thursday to Friday (21% vs. 20% in THA and TKA, respectively) was similar. The weekday of discharge was peaking on Thursdays, but with a considerable number of patients being discharged during the weekend (13%) (Figure 2). When analyzing the influence of weekday of surgery on risk of having a LOS > 2 days it was possible to include 9,987 (94%) procedures (Figure 1). Only surgery on a Friday was associated with a significantly higher risk of LOS > 2 days after adjusting for patient characteristics (Table 2). The proportion of patients who had their operation on Mondays, Tuesdays, or Wednesdays was 80%, of whom 17% (CI 17–18) had a LOS > 2 days compared with 19% (CI 17–21) with LOS > 2 days in those having surgery on Thursday to Friday (OR 1.1). Multiple logistic regression on preoperative risk factors associated with a LOS > 2 days was possible in 9,987 patients, and

Acta Orthopaedica 2021; 92 (2): 170–175 173

Table 2. Multiple-regression analysis of risk factors (n = 9,987) for LOS > 2 days and attributed risk-score points Variable Weekday of surgery Monday Tuesday Wednesday Thursday Friday Saturday Sunday Age group < 50 50–60 61–65 66–70 71–75 76–80 81–85 > 85 BMI < 18.5 18.5–24.9 25–29.9 30–34.9 35–39.9 > 39.9 Female sex Walking aid Living alone Smoking Alcohol > 24 g/day Total knee arthroplasty Psychiatric disease Cardiac disease Pulmonary disease Hypertension Diabetes mellitus Non-insulin dependent Insulin dependent Anticoagulants Preoperative anemia a

OR (95%CI) p-value points

Table 3. Distribution of points depending on number of risk factors for LOS > 2 days

1.3 (0.9–1.8) 1.2 (0.9–1.4) 1.0 (0.8–1.3) 1 (ref.) 1.2 (1.0–1.4) 1.4 (1.2–1.7) 2.0 (1.6–2.5) 3.4 (2.5–4.6)

0.08 0.1 0.9 – 0.1 < 0.01 < 0.01 < 0.01

1.6 (1.0–2.6) 1 (ref.) 1.0 (0.8–1.1) 1.0 (0.9–1.2) 1.2 (0.9–1.5) 1.3 (0.9–1.8) 1.3 (1.2–1.5) 1.9 (1.7–2.3) 1.6 (1.4–1.8) 1.1 (0.9–1.3) 0.8 (0.6–1.0) 1.9 (1.7–2.1) 1.6 (1.4–1.9) 1.2 (1.0–1.4) 1.3 (1.1–1.6) 1.0 (0.9–1.1)

0.07 – 0.7 1.0 0.2 0.1 < 0.01 < 0.01 < 0.01 0.5 0.06 < 0.01 < 0.01 0.04 < 0.01 0.8

1.3 (1.1–1.6) 2.0 (1.4–2.9) 1.2 (1.0–1.5) 1.5 (1.3–1.7)

0.02 < 0.01 0.09 < 0.01

1 2 0 1

1 2 3

0 0 0 1 2 1 0 0 1 1 1 1 0

LOS > 2 days LOS < 3 days 80

Number of points

1 (ref.) – 1.0 (0.9–1.2) 1.0 0.9 (0.7–1.0) 0.08 0.9 (0.8–1.1) 0.4 1.5 (1.2–2.0) < 0.01 2.9 (0.3–30.5) 0.4 1.4 (0.2–13.6) 0.8

Distribution of LOS (%) 100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

n (%) 898 (8.9) 1,278 (13) 1,759 (17) 1,902 (19) 1,497 (15) 1,107 (11) 727 (7.2) 449 (4.4) 293 (2.9) 134 (1.3) 68 (0.7) 14 (0.1) 3 (0.0) 0 (0) 0 (0) 0 (0)

N = 10,129 due to 447 cases with missing data for variables associated with LOS > 2 days (see Table 1).

OR: odds ratio. CI: confidence interval. a Hemoglobin < 13 g/dL.

found association with age (76–80: OR 1.4, 81–85: OR 2.0, and > 85: OR 3.4), female sex (OR 1.3), use of walking aids (OR 1.9), living alone (OR 1.6), TKA (OR 1.9), psychiatric disorder (OR 1.6), cardiac (OR 1.2) or pulmonary disease (OR 1.3), both non-insulin (OR 1.3) and insulin-dependent diabetes (OR 2.0), and preoperative anemia (OR 1.5) (Table 2). After assigning values to the significant risk factors for LOS > 2 days according to their respective ORs, a score ranging from 0 to 15 was constructed with increasing points indicating increasing risk of LOS > 2. However, only 3 patients scored 12 points and no patient scored more than 12 points (Table 3). An ROC curve was constructed for this model showing an area under the curve of 0.70 (CI 0.69–0.72). The fraction of patients with LOS > 2 days increased with increasing scores (Figure 3), and there was an even distribution of patients across weekdays regardless of score (Figure 4). When decid-

9 10 11 12

Number of points Figure 3. Distribution of patients with length of stay (LOS) > 2 days and < 3 days according to number of points based on odds ratios of relevant risk factors for LOS > 2 days.

Monday Tuesday Number of points 0 1 2 3 4 5 6 7 8 9 10 11 12

Wednesday Thursday Friday Saturday Sunday 0

Distribution according to weekday (%)

Figure 4. Distribution of procedures on each weekday according to number of points for LOS > 2 days.

ing on an adequate cut-off for at-risk patients, a threshold of ≥ 7 points would yield a sensitivity of 23% and a specificity of 93%. However, such patients attributed only 11% of the total population, severely reducing clinical relevance. Correspondingly, a threshold of ≥ 6 points would increase sensitivity (36%), decrease specificity (87%), and include 17% of the total population. Finally, sensitivity would increase to 51% and specificity decline to 78% but include 28% of all patients if using a threshold of ≥ 5 points. Consequently, a threshold of ≥ 6 points was chosen as this would yield acceptable sensitivity and specificity while including a clinically relevant and manageable number of patients. Of the “high-risk patients” with ≥ 6 points, 38% (n: 635) had a LOS > 2 and median LOS was 4 days (IQR 2). 18% had surgery on either Thursdays or Fridays of whom 43% (CI 38–49) stayed > 2 days compared with 37% (CI 34–39) who had surgery on Monday through Wednesday (OR 1.3, CI 1.0–1.7; p = 0.04). In contrast, 14% (CI 13–14) of patients with a risk score of < 6 had a LOS > 2 days and a median LOS of 3 days (IQR 1).

174

Discussion Although ERAS programs have been documented to decrease LOS after THA and TKA in many centers to median 0 to 2 days, limited information is available on the influence of specific risk factors and weekday of surgery on LOS. Thus, to our knowledge, only 1 previous retrospective single-center study on TKA has investigated the influence of weekday of surgery and found increased risk of LOS > 3 days when having surgery on Thursdays (Mathijssen et al. 2016). Consequently, we used prospective data collected from the well-established Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement database to study the association between preoperative risk factors, weekday of surgery and LOS > 2 days. The results of our analysis, which found anemia, use of walking aids, age > 75 years, TKA, female sex, living alone, psychiatric, pharmacologically treated cardiac and pulmonary disease, and type of diabetes to be important risk factors, are unsurprising as they have previously been demonstrated to influence LOS > 4 days in a fast-track setup (Jørgensen and Kehlet 2013b, Jans et al. 2014, Jørgensen et al. 2016, Cram et al. 2018, Johnson et al. 2020). One of the reasons that TKA was a risk factor for LOS > 2 days may be related to more extensive pain problems compared with THA, consequently prolonging time to achieving functional discharge criteria. However, TKA has also been found to be an independent predictor of potentially preventable “surgical” complications, which mostly occur after discharge, mainly prosthetic infections and manipulation under anesthesia (Jørgensen et al. 2016). That female sex was a risk factor has been found in some (Winemaker et al. 2015) but not in all studies (den Hartog et al. 2015, Jørgensen et al. 2016). A recent review specifically on TKA also found female sex to be a risk factor for increased LOS (Shah et al. 2019). However, although several of the included studies utilized “fast-track” programs, reported LOS was longer than in our study (3 days or more). In contrast, conventional risk factors such as BMI, smoking, and alcohol use have often been reported to influence complications and LOS (Belmont et al. 2014, Best et al. 2015, Winemaker et al. 2015, Jeschke et al. 2018, Sahota et al. 2018), but may be of reduced importance within a fasttrack setup (Jørgensen et al. 2013a, den Hartog et al. 2015, Husted et al. 2016). Importantly, as pointed out by Shah et al., the influence of a single risk factor is often clinically negligible, but the combination of several risk factors may increase patients’ disposition for extended LOS (Shah et al. 2019). This is further illustrated by the ideal threshold for defining “highrisk” patients being about 6 points on our risk score. Regarding the point score for identification of patients at high risk of having a LOS > 2 days the primary objective of our study was not to construct the ideal final risk-prediction model for LOS > 2 days. Rather, we wanted to investigate whether it would be possible to provide a simple algorithm for identification of patients who may benefit from having surgery early in

Acta Orthopaedica 2021; 92 (2): 170–175

the week. Consequently, we chose a pragmatic approach that included only significant risk factors and assigned increasing points according to increasing odds ratios. Thus, the presented THA/TKA risk score may likely benefit from further optimization to increase sensitivity and specificity, but bearing in mind that it should be effective, simple, and fast to use in clinical practice. In this context the clinical influence of advanced risk calculators such as the NSQUIP surgical-risk calculators on surgical planning and postoperative outcomes remains uncertain, potentially due to difficulties with implementation in clinical practice (Moonesinghe et al. 2013). The finding that patients with ≥ 6 points had an even higher risk of LOS > 2 days when having surgery on Thursday or Friday is not unexpected when considering the reduction in availability of staff resources and other providers (e.g., primary care providers etc.) during the weekend, although we have no data to confirm this suspicion. One could speculate that other logistic issues would influence whether patients were discharged earlier, i.e., premature discharge in the middle of the week or longer admission at the weekend due to beds being available. However, all departments use functional discharge criteria, and transfer to rehabilitation homes is extremely rare, occurring in < 7% of patients aged ≥ 85 years (Pitter et al. 2016). Thus, it seems unlikely that patients would be discharged prematurely in the middle of the week in order to free up beds. Furthermore, the proportion of patients being discharged on Saturdays largely reflected the number of surgeries on Thursdays, arguing against unnecessary prolongation of admission over the weekend. However, our results may be of clinical relevance, as the proposed risk score may be useful in planning weekday of surgery, potentially reducing the risk of LOS extending into the weekend. This would be of special value in ambulatory surgical centers or 5-day surgical units. Interestingly, although the Lundbeck Foundation Centre for Fast-track THA and TKA has been documented to be successful in reducing LOS (Petersen et al. 2019), apparently the present data analysis of risk factors vs. choice of weekday of surgery has not been included in the daily logistical preparation within the centers where about 20% of operations were performed on Thursdays and Fridays but with an even distribution of “high-” and “low”risk patients throughout the week. With regard to the definition of LOS, our study calculates LOS as postoperative nights in hospital and where the most recent data have shown a mean LOS of 1.9 days in unselected patients. (Petersen et al. 2019). In contrast, there are several studies where outpatient procedures are defined as less than 24 hours or less than 2 midnights (Vehmeijer et al. 2018, Johnson et al. 2020) or with use of nursing care facilities or rehabilitation homes (Cram et al. 2018, Ross et al. 2020), which may give a false impression of reduced LOS. In this context, a limitation of our study is the lack of detailed information on discharge destination although non-home discharge after fast-track THA and TKA is limited in Denmark and even in

Acta Orthopaedica 2021; 92 (2): 170–175 175

patients > 85 years been demonstrated to occur in only about 7% of patients (Pitter et al. 2016). Furthermore, we do not have detailed data on why patients were admitted for > 2 days or whether they were transferred to other departments. Such data would be of interest in order to define the exact number of patients remaining in the arthroplasty departments during the weekend, and how to prevent it. The strengths of our study are detailed prospective registration of risk factors, complete data on index LOS, and a large cohort of non-selected THA and TKA patients from a well-established enhanced recovery multicenter collaboration developed across the last decade. In summary, the planning of day of surgery in relation to relevant preoperative risk factors should receive more attention and may lead to an overall decrease in LOS and consequently risk of hospitalization into the weekend in high-risk patients. CJ contributed to the idea and study design, collection and analysis of data, writing of the primary draft, and revision and submission of the finished manuscript. KG contributed to the idea and study design, evaluation of data, and revision of the primary draft. PB contributed to collection and analysis of data and revision of the primary draft. HK contributed to idea and study design, evaluation of data, and writing and revision of the primary draft. Acta thanks David Houlihan-Burne and Michael Clarius for help with peer review of this study. Belmont P J Jr, Goodman G P, Waterman B R, Bader J O, Schoenfeld A J. Thirty-day postoperative complications and mortality following total knee arthroplasty: incidence and risk factors among a national sample of 15,321 patients. J Bone Joint Surg Am 2014; 96(1): 20-6 doi: JBJS.M.00018. Best M J, Buller L T, Gosthe R G, Klika A K, Barsoum W K. Alcohol misuse is an independent risk factor for poorer postoperative outcomes following primary total hip and total knee arthroplasty. J Arthroplasty 2015; 30(8): 1293-8. doi: 10.1016/j.arth.2015.02.028. Cizmic Z, Feng J E, Anoushiravani A A, Borzio R W, Schwarzkopf R, Slover J D. The risk assessment and prediction tool is less accurate in extended length of stay patients following total joint arthroplasty. J Arthroplasty 2019; 34(3): 418-21. doi: 10.1016/j.arth.2018.11.008. Cram P, Landon B E, Matelski J, Ling V, Stukel T A, Paterson J M, Gandhi R, Hawker G A, Ravi B. Utilization and short-term outcomes of primary total hip and knee arthroplasty in the United States and Canada: an analysis of New York and Ontario administrative data. Arthritis Rheumatol (Hoboken, NJ) 2018; 70(4): 547-54. doi: 10.1002/art.40407. den Hartog Y M, Mathijssen N M, Hannink G, Vehmeijer S B. Which patient characteristics influence length of hospital stay after primary total hip arthroplasty in a “fast-track” setting? Bone Joint J 2015; 97-B(1): 19-23. doi: 10.1302/0301-620X.97B1.33886. Gkagkalis G, Pereira L C, Fleury N, Luthi F, Lécureux E, Jolles B M. Are the Cumulated Ambulation Score and Risk Assessment and Prediction Tool useful for predicting discharge destination and length of stay following total knee arthroplasty? Eur J Phys Rehabil Med 2019; 55(6): 816-23. doi: 10.23736/s1973-9087.19.05568-0. Husted H, Solgaard S, Hansen T B, Soballe K, Kehlet H. Care principles at four fast-track arthroplasty departments in Denmark. Dan Med Bull 2010; 57(7): A4166. Husted H, Jørgensen C C, Gromov K, Kehlet H. Does BMI influence hospital stay and morbidity after fast-track hip and knee arthroplasty? Acta Orthop 2016; 87(5): 466-72. doi: 10.1080/17453674.2016.1203477 Jans O, Jørgensen C, Kehlet H, Johansson P I. Role of preoperative anemia for risk of transfusion and postoperative morbidity in fast-track hip and knee arthroplasty. Transfusion 2014; 54(3): 717-26. doi: 10.1111/trf.12332

Jeschke E, Citak M, Gunster C, Halder A M, Heller K D, Malzahn J, Niethard F U, Schrader P, Zacher J, Gehrke T. Obesity increases the risk of postoperative complications and revision rates following primary total hip arthroplasty: an analysis of 131,576 total hip arthroplasty cases. J Arthroplasty 2018; 33(7): 2287-92. doi: 10.1016/j.arth.2018.02.036. Johnson D J, Castle J P, Hartwell M J, D’Heurle A M, Manning D W. Risk factors for greater than 24-hour length of stay after primary total knee arthroplasty. J Arthroplasty 2020; 35(3): 633-7. doi: 10.1016/j.arth.2019.10.037. Jørgensen C C, Kehlet H. Outcomes in smokers and alcohol users after fasttrack hip and knee arthroplasty. Acta Anaesthesiol Scand 2013a; 57(5): 631-8. doi: 10.1111/aas.12086. Jørgensen C C, Kehlet H. Role of patient characteristics for fast-track hip and knee arthroplasty. Br J Anaesth 2013b; 110(6): 972-80. doi: 10.1093/bja/ aes505. Jørgensen C C, Petersen M A, Kehlet H. Preoperative prediction of potentially preventable morbidity after fast-track hip and knee arthroplasty: a detailed descriptive cohort study. BMJ Open 2016; 6(1): e009813. doi: 10.1136/ bmjopen-2015-009813. Kim K Y, Feng J E, Anoushiravani A A, Dranoff E, Davidovitch R I, Schwarzkopf R. Rapid discharge in total hip arthroplasty: utility of the outpatient arthroplasty risk assessment tool in predicting same-day and next-day discharge. J Arthroplasty 2018; 33(8): 2412-6. doi: 10.1016/j. arth.2018.03.025. Mathijssen N M, Verburg H, van Leeuwen C C, Molenaar T L, Hannink G. Factors influencing length of hospital stay after primary total knee arthroplasty in a fast-track setting. Knee Surg Sports Traumatol Arthrosc 2016; 24(8): 2692-6. doi: 10.1007/s00167-015-3932-x. Moonesinghe S R, Mythen M G, Das P, Rowan K M, Grocott M P. Risk stratification tools for predicting morbidity and mortality in adult patients undergoing major surgery: qualitative systematic review. Anesthesiology 2013; 119(4): 959-81. doi: 10.1097/ALN.0b013e3182a4e94d. Petersen P B, Jorgensen C C, Kehlet H. Temporal trends in length of stay and readmissions after fast-track hip and knee arthroplasty. Dan Med J 2019; 66(7): A5553. Pitter F T, Jorgensen C C, Lindberg-Larsen M, Kehlet H. Postoperative morbidity and discharge destinations after fast-track hip and knee arthroplasty in patients older than 85 years. Anesth Analg 2016; 122(6): 1807-15. doi: 10.1213/ANE.0000000000001190. Ross T D, Dvorani E, Saskin R, Khoshbin A, Atrey A, Ward S E. Temporal trends and predictors of thirty-day readmissions and emergency department visits following total knee arthroplasty in Ontario between 2003 and 2016. Journal Arthroplasty 2020; 35(2): 364-70. doi: 10.1016/j.arth.2019.09.015. Sahota S, Lovecchio F, Harold R E, Beal MD, Manning D W. The effect of smoking on thirty-day postoperative complications after total joint arthroplasty: a propensity score-matched analysis. J Arthroplasty 2018; 33(1): 30-5. doi: 10.1016/j.arth.2017.07.037. Schmidt M, Schmidt S A, Sandegaard J L, Ehrenstein V, Pedersen L, Sorensen H T. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7: 449-90. doi: 10.2147/clep.s91125. Shah A, Memon M, Kay J, Wood T J, Tushinski D M, Khanna V. preoperative patient factors affecting length of stay following total knee arthroplasty: a systematic review and meta-analysis. J Arthroplasty 2019; 34(9): 2124-65. e1. doi: 10.1016/j.arth.2019.04.048. 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. Vehmeijer S B W, Husted H, Kehlet H. Outpatient total hip and knee arthroplasty. Acta Orthop 2018; 89(2): 141-4. doi: 10.1080/17453674.2017.1410958. Wainwright T W, Kehlet H. Fast-track hip and knee arthroplasty: have we reached the goal? Acta Orthop 2019; 90(1): 3-5. doi: 10.1080/17453674. 2018.1550708. Winemaker M, Petruccelli D, Kabali C, de Beer J. Not all total joint replacement patients are created equal: preoperative factors and length of stay in hospital. Can J Surg 2015; 58(3): 160-6. doi: 10.1503/cjs.008214. Ziemba-Davis M, Caccavallo P, Meneghini R M. Outpatient joint arthroplastypatient selection: update on the outpatient arthroplasty risk assessment score. J Arthroplasty 2019; 34(7s): S40-s3. doi: 10.1016/j.arth.2019.01. 007.

176

Acta Orthopaedica 2021; 92 (2): 176–181

Increased rate of complications in myasthenia gravis patients following hip and knee arthroplasty: a nationwide database study in the PearlDiver Database on 257,707 patients William F SHERMAN, Victor J WU, Sione A OFA, Bailey J ROSS, Ian D SAVAGE-ELLIOTT, and Fernando L SANCHEZ

Department of Orthopaedic Surgery, Tulane University School of Medicine, New Orleans, LA, USA Correspondence: Swilliam1@tulane.edu Submitted 2020-07-27. Accepted 2020-10-13

Background and purpose — The increasing prevalence of total hip arthroplasty (THA) and total knee arthroplasty (TKA) within the growing elderly population is translating into a larger number of patients with neuromuscular conditions such as myasthenia gravis (MG) receiving arthroplasty. We compared systemic and joint complications following a THA or TKA between patients with MG and patients without MG. Patients and methods — Patient records were queried from PearlDiver (Pearl Diver Inc, Fort Wayne, IN, USA), an administrative claims database, using ICD-9/ICD-10 and Current Procedural Terminology codes. In-hospital and 90-day post-discharge rates of systemic and joint complications were compared between the 2 cohorts. Results — 372 patients with MG and 249,428 patients without MG who received a THA or TKA were included in the study. At 90 days post-discharge, MG patients exhibited exhibited between 1.6 and 15% higher rates of systemic complications, including cerebrovascular event, pneumonia, respiratory failure, sepsis, myocardial infarction, acute renal failure, anemia, and deep vein thrombosis (all p < 0.001). The same results were also found during the in-hospital time period. 90-day incidence of aseptic loosening was the only joint complication with significantly increased odds risk for the MG cohort (OR 5; 95% CI 2–12). Interpretation — Patients with MG exhibited significantly higher risk for multiple systemic complications during the index hospital stay and in the acute post-discharge setting.

Myasthenia gravis (MG) is an autoimmune neuromuscular disease that causes fluctuating muscle weakness throughout the body (Ciafaloni 2019). The estimated global incidence and prevalence of MG between 1950 and 2007 was between 5 and 77 per 105 persons respectively, and mortality was 0.1–0.9 per 105 persons (Carr et al. 2010). Thymectomy improves clinical outcomes, reduces the need for immunosuppressive therapy, and may be curative in a subset of patients (Wolfe et al. 2016, Remes-Troche et al. 2002). Advancements in the treatment have improved life expectancy and increased the demand for arthroplasty in these patient (Carr et al. 2010, Cichos et al. 2019). As the demand for primary hip and knee arthroplasty is projected to increase (Sloan et al. 2018), it is imperative that surgeons have access to data outlining clinical outcomes following arthroplasty in MG patients. Recent studies examining total joint replacement in patients with neuromuscular disorders have demonstrated increased rates of both prosthetic joint and systemic medical complications (Cichos et al. 2019). However, there remains a lack of research specifically aimed at identifying the risks and complications seen with arthroplasties in patients with MG. Prior studies examining postoperative complications in patients with MG have demonstrated increased risks of adverse medical events including re-intubation (Cichos et al. 2019), increased surgical site bleeding, myasthenic crisis, pneumonia, and septicemia (Marulli et al. 2013). This study compares rates of postoperative complications following total hip arthroplasty (THA) and total knee arthroplasty (TKA) in MG vs. non-MG patients, with the intent of helping guide care teams in developing and implementing optimal surgical management plans when performing TJR on patients with MG.

Acta Orthopaedica 2021; 92 (2): 176–181 177

Patients and methods This study used the Humana dataset, which contains the medical records of 25.4 million patients from 2007 to 2017 who were privately insured, commercially insured, or purchased their Medicare Advantage plans through Humana Health Insurance. Patient records were queried from PearlDiver (PearlDiver Inc, Fort Wayne, IN, USA), a commercially available administrative claims database, using ICD-9/ICD-10 and Current Procedural Terminology (CPT) codes. A retrospective cohort design was used to compare patients who had MG and underwent a THA or TKA vs. patients without MG who underwent a THA or TKA. Patients who had undergone a THA or TKA were identified using both ICD and CPT codes. Exclusion criteria included patients receiving arthroplasty for pathologic or traumatic fractures. Inclusion criteria for patients with MG included a diagnosis of MG, made by board-certified physicians and defined by ICD diagnosis codes in the database, at any time before the THA or TKA. To exclude patients who no longer were at risk for symptomatic MG, patients with a prior history of thymectomy were also excluded. The ICD codes that defined the study groups are provided in Appendix Table A1 (see Supplementary data). Each cohort was queried for basic demographic information, clinical characteristics, and hospital course data such as age, sex, hospital region, BMI, length of stay (LOS), 90-day readmission rate, discharge status, Charlson comorbidity index (CCI), and comorbidities. Regional data was categorized based on the United States Census Bureau classification of Northeast, Midwest, South, and West. Discharge status codes (DC) were classified into home, nursing, expired, or discontinued against medical advice (DC-07). Home discharge included self-care (DC-01, 21) and home health (DC06) codes. Expired discharge status included those with codes for expiration (DC-20, 40, 41). All other codes including discharge to or planned return to a short-term nursing facility, long-term care hospital, or other medical facility were classified as a nursing discharge status. Specific comorbidities were queried using ICD diagnosis codes from the database including the presence of diabetes mellitus, hypertension, chronic obstructive pulmonary disease, chronic kidney disease, congestive heart failure, coronary artery disease, rheumatoid arthritis, and tobacco use. Incidences of postoperative systemic and joint complications were queried for the 2 patient cohorts. Both systemic and local joint complications were examined during the surgical encounter before discharge, and at 90 days post-discharge. Systemic complications queried included malignant hyperthermia, cerebrovascular event (stroke, non-traumatic hemorrhage, occlusion of cerebral arteries), anemia (post-hemorrhagic, iron deficiency from blood loss), acute renal failure, myocardial infarction (MI), pneumonia, sepsis, deep vein

thrombosis (DVT), pulmonary embolism (PE), and respiratory failure. The codes used to define systemic complications are provided in Appendix Table A2 (see Supplementary data). Post-discharge joint complications queried included prosthetic joint infection (PJI), periprosthetic fracture, hip dislocation, and aseptic loosening. PJI was defined by procedural codes that indicated a surgical intervention for a deep joint infection to exclude superficial wound complications that would have been included in diagnosis codes for PJI. The codes used to define joint complications are provided in Appendix Table A3 (see Supplementary data) Statistics All data analyses were performed using the R statistical software (R Project for Statistical Computing, Vienna, Austria) integrated within PearlDiver with an α level set to 0.05. Multivariable logistic regression adjusting for patient sex, age, Charlson Comorbidity Index (CCI), and BMI was used to calculate odds ratios (OR) with corresponding 95% confidence intervals (CI) for rates of joint and systemic complications between the 2 cohorts. Proportions of various demographic and clinical characteristics in the MG and non-MG cohorts were compared using chi-square analysis to assess for baseline differences between the 2 patient populations. The null hypothesis was that different demographic proportions were the same between MG vs. non-MG cohorts. It is important to take into account the effect that age, sex, BMI, and CCI can have on outcomes of THA and TKA. Older patients experience higher risks of negative postoperative outcomes when undergoing either THA or TKA. In terms of in-hospital complications, examples of such poor outcomes include acute myocardial infarction, DVT or PE, surgical site infection, sepsis, hemorrhage, and mortality (Fang et al. 2015). Similarly, patients with a BMI classification of ≥ 30 are more likely to experience prosthetic failure and postoperative infection following TKA and THA (Boyce et al. 2019, Correa-Valderrama et al. 2019). For TKA, male patients have higher rates of revision surgery, mortality, hospital readmission, and wound infections compared with women, while female patients have increased risk of readmission, reoperation, and wound infection following THA (Singh et al. 2013, Patel et al. 2020). When evaluating comorbidity presence on mortality risk (CCI) for both THA and TKA, patients with a moderate (CCI score of 2) or high (CCI score of 3 or higher) comorbidity burden have a higher 90-day mortality risk in comparison with patients with a low comorbidity burden (CCI score of 1) (Glassou et al. 2017). The decision to control for age, sex, BMI, and CCI in the statistical analysis was because each variable has substantial potential for exerting confounding effects given the associated risks that each has on the outcomes following THA or TKA. Ethics, funding, and potential conflicts of interest Institutional review board exemption by Tulane University Human Research Protection Program was granted for this study

178

Acta Orthopaedica 2021; 92 (2): 176–181

Pearl Diver Humana Dataset n = 25.4 x 106 Primary THA or TKA n = 257,707 Excluded Pathological/traumatic fractures, history of thymectomy n = 7,907 THA or TKA included in the study (n = 249,800): – with myasthenia gravis, 372 – without myasthenia gravis, 249,428

Patients included in this study. THA, total hip arthroplasty; TKA, total knee arthroplasty.

on June 11, 2020, because provided data was deidentified and compliant with the Health Insurance Portability and Accountability Act. This study received no funding or financial support from outside sources. Author FLS receives research support as a Principal Investigator from Medacta International (USA), and serves on the Content Committee for the American Academy of Orthopaedic Surgeons (AAOS) Hip and Knee. Author IDSE serves as a Co-Editor for Orthopedics Today Grand Rounds. All other authors declare no conflict of interests.

Table 1. Demographics and clinical characteristic comparisons. Values are count (%) unless otherwise specified THA/TKA MG no MG Demographic variable (n = 372) (n = 249,428) Female sex Age < 65 65–79 ≥ 80 BMI a < 30 30–39 ≥ 40 CCI, mean (SD) Specific comorbidities: Diabetes mellitus Hypertension COPD Chronic kidney disease Congestive heart failure Coronary artery disease Rheumatoid arthritis Tobacco use

p-value

206 (55.4)

152,931 (61.3)

< 0.001

80 (21.5) 268 (72.0) 73 (19.6)

55,334 (22.2) 163,414 (65.5) 33,665 (13.5)

0.7 < 0.001 < 0.001

106 (38.7) 113 (41.2) 55 (20.1) 3.0 (2.8)

53,870 (34.9) 69,579 (45.1) 30,849 (20.0) 1.9 (2.3)

0.2 0.2 0.9 <0.001

217 (58.3) 332 (89.2) 152 (40.9) 0 (0) 64 (17.2) 148 (39.8) 38 (10.2) 84 (22.6)

104,304 (41.8) 199,152 (79.8) 68,945 (27.6) 287 (0.2) 24,965 (10.0) 69,686 (27.9) 15,443 (6.2) 47,351 (19.0)

< 0.001 < 0.001 < 0.001 NA < 0.001 < 0.001 < 0.001 0.1

BMI data was available for 75% of MG and 62% of non-MG patients. CCI, Charlson Comorbidity Index; COPD, chronic obstructive pulmonary disease; MG, myasthenia gravis; NA, not applicable; SD, standard deviation; THA, total hip arthroplasty; TKA, total knee arthroplasty.

Results Between 2007 and 2017 in the PearlDiver database, 80,727 primary THAs and 176,980 primary TKAs were performed for a total of 257,707 arthroplasty procedures. After applying exclusion criteria, 249,428 patients who received THA or TKA did not have MG and 372 patients had MG (Figure). As outlined in Table 1, the MG cohort had a greater proportion of males (45% vs. 39%, p < 0.001) and older patients (age ≥ 80: 20% vs. 14%, p < 0.001). MG patients also had a higher average burden of comorbidities (CCI 3.0 vs. 1.9, p < 0.001). Additionally, MG patients had an increased mean LOS (3.7 days vs. 3.5 days, p = 0.1), and a higher 90-day readmission rate (13.4% vs. 12.9%, p = 0.7). Regarding disposition status, MG patients were more likely to discharge home (78% vs. 65%, p < 0.001) and to a nursing facility (41% vs. 33%, p < 0.001) (Table 2). Systemic complications were more likely in patients with MG relative to non-MG patients in the 90-day post-discharge period (Table 3). Statistically significant findings included cerebrovascular event (OR 6.6), pneumonia (OR 6.1), respiratory failure (OR 5.5), sepsis (OR 4.4), pulmonary embolism (OR 3.4), myocardial infarction (OR 3.3), acute renal failure (OR 3.1), anemia (OR 2.1), and deep vein thrombosis (OR 1.9). Systemic complications were also more likely in the MG cohort during their inpatient hospital stay (Table 3). Complications that were statistically more likely for MG patients included pneumonia (OR 10), sepsis (OR 9.9), cerebrovascular event (OR 7.9), acute myocardial infarction (OR 7.4),

Table 2. Comparison of hospital region and course. Values are count (%) unless otherwise specified THA/TKA MG no MG Hospital course variable (n = 372) (n = 249,428) Region South Midwest Northeast West Length-of-stay, mean (SD) Discharge Home Nursing Expired Against medical advice 90-day readmission rate

p-value

230 (62) 105 (28) 5 (1.3) 32 (8.6) 3.7 (3.0)

143,403 (58) 73,557 (30) 6,426 (2.6) 26,186 (11) 3.5 (4.4)

0.1 0.6 0.1 0.2 0.1

290 (78) 152 (41) 1 (0.3) 0 (0) 50 (13)

162,656 (65) 82,108 (33) 196 (0.08) 66 (0.03) 32,307 (13)

< 0.001 < 0.001 0.2 NA 0.7

For abbreviations, see Table 1.

respiratory failure (OR 6.7), and acute renal failure (OR 2.1). With respect to joint-specific complications, almost no statistically significant differences were found between patients in the MG cohort and patients without MG for both the inpatient hospital stay duration and in the 90-day postoperative period (Table 4). The lone exception was aseptic loosening at 90 days post-discharge, which was found to be more likely in MG patients (90-day postop OR 5.1).

Acta Orthopaedica 2021; 92 (2): 176–181 179

Table 3. Comparison of systemic complications at 90 days postoperatively and during inpatient hospital stay. Values are count (%) and odds ratio (OR) with 95% confidence interval (CI) THA/TKA MG no MG Systemic complications (n = 372) (n = 249,428) Cerebrovascular event In-hospital 90-day Pneumonia In-hospital 90-day Respiratory failure In-hospital 90-day Anemia In-hospital 90-day Sepsis In-hospital 90-day Acute MI In-hospital 90-day Deep vein thrombosis In-hospital 90-day Acute renal failure In-hospital 90-day Pulmonary embolism In-hospital 90-day Malignant hyperthermia: In-hospital 90-day

OR (95% CI)

33 (8.9) 66 (18)

3,134 (1.3) 8,422 (3.4)

7.9 (5.4–11) 6.6 (4.9–8.7)

32 (8.6) 49 (13)

2,388 (1.0) 6,304 (2.5)

10 (6.8–14) 6.1 (4.5–8.3)

30 (8.1) 37 (9.9)

3,310 (1.3) 5,175 (2.1)

6.7 (4.5–9.6) 5.5 (3.8–7.6)

120 (32.3) 72,094 (28.9) 102 (27.4) 39,005 (15.6)

1.2 (0.9–1.5) 2.1 (1.6–2.6)

7 (1.9) 18 (4.8)

632 (0.3) 2,894 (1.2)

9.9 (4.7–18) 4.4 (2.6–6.9)

11 (3.0) 11 (3.0)

1,019 (0.4) 2,283 (0.9)

7.4 (3.8–13) 3.3 (1.7–5.7)

13 (3.5) 33 (8.9)

2,689 (1.1) 12,331 (4.9)

3.3 (1.8–5.6) 1.9 (1.3–2.6)

25 (6.7) 8,173 (3.3) 42 (11.3) 10,299 (4.1)

2.1 (1.4–3.2) 3.1 (2.2–4.2)

14 (3.8) 21 (5.6)

6.2 (3.5–10) 3.4 (2.1–5.2)

0 (0) 0 (0)

1,559 (0.6) 4,260 (1.7) 2 (< 0.1) 2 (< 0.1)

NA NA

For abbreviations, see Table 1.

Discussion The majority of systemic complications analyzed were statistically significantly more likely and clinically relevant for MG patients relative to patients without MG. This held true for both in-hospital and at 90 days post-discharge. Despite the association of malignant hyperthermia with individuals with a known myopathy or neuromuscular disease (Wedel 1992) there was no occurrence of this complication in either cohort. It is possible the study was underpowered to discern a difference for this very rare complication. Additionally, malignant hyperthermia has a higher odds risk in children compared with adults (Rosenberg and Fletcher 1994, Rosenberg et al. 2015). Because the majority of patients undergoing total joint replacement are adults, low rates of malignant hyperthermia in this population are expected. The increased occurrence of systemic complications in patients with MG aligns partially with the results of previous studies (Chang et al. 2017, Cichos et al. 2019). Similar to the findings by Chang et al. (2017), MG patients experienced

Table 4. 90 days postoperatively and during inpatient hospital stay comparison of joint-specific complications. Values are count (%) and odds ratio (OR) with 95% confidence interval (CI) THA/TKA MG no MG Local complications (n = 372) (n = 249,428) Prosthetic joint infection In-hospital 90-day Periprosthetic fracture In-hospital 90-day Aseptic loosening In-hospital 90-day Prosthetic joint dislocation In-hospital 90-day

OR (95% CI)

2 (0.5) 4 (1.1)

1,112 (0.5) 4,151 (1.7)

0.6 (0.1–1.6) 0.8 (0.3–1.8)

1 (0.3) 2 (0.5)

180 (0.7) 837 (0.3)

3.8 (0.2–17) 2.5 (0.6–6.5)

2 (0.5) 3 (0.8)

511 (0.2) 550 (0.2)

2.7 (0.4–8.4) 5.1 (1.6–12)

0 (0) 0 (0)

131 (< 0.1) 353 (0.1)

NA NA

For abbreviations, see Table 1.

higher rates of postoperative pneumonia and sepsis (Chang et al. 2017). Furthermore, these findings also support those by Cichos et al. (2019), which demonstrated higher rates of postoperative anemia and deep vein thrombosis in patients with MG (Cichos et al. 2019). Our study further expands the scope of systemic complications demonstrating MG patients are at a higher risk by also showing higher rates of cerebrovascular events, respiratory failure, acute MI, and acute renal failure. Chang et al. (2017) limited their examination of complications to 30 days post-discharge, and Cichos et al. (2019) limited their data collection to the perioperative event. Our data documented the increased occurrence of these complications during the hospital stay and within the 90-day post-discharge period. Given the similarities in findings to previous literature, our results suggest the presence of MG contributes to the increased odds risk for systemic complications after a TKA or THA. In addition to the systemic complications, the data demonstrated MG patients were generally older (age 65–79, 72% vs. 66%, and age ≥ 80, 20% vs. 14%), had higher levels of non-obese classifications (BMI < 30, 39% vs. 5%), and had an increased burden of medical comorbidities (CCI 3.0 vs. 1.9) when compared with non-MG patients. Given the presence of comorbidities is associated with poorer outcomes after undergoing surgical operation (Misra et al. 2020), we used multivariable logistic regression to adjust for select highimpact confounding variables in an attempt to more accurately evaluate an MG as an independent risk factor in the increased occurrence of systemic complications. The arthroplasty-specific complication found to be more likely for patients with MG was aseptic joint loosening within the 90-day post-discharge period (90-day postop OR 5.1; CI 1.6–12). The high odds ratio aligns with the findings by Cichos et al. (2019), who also demonstrated higher rates of

180

aseptic loosening. However, the reliability of comparisons between our study and prior work is limited, given that previous studies examined joint complications in patients with any neuromuscular disorder undergoing total joint replacement where this study specifically examined total joint arthroplasty in patients with MG. With TKA being moved to an outpatient designation by the Centers for Medicare and Medicaid Services (CMS), many total joint replacement procedures are being performed in surgery centers and centers without ICU capabilities. The increased systemic risks seen in this cohort of MG patients should guide surgeons who utilize outpatient surgery centers or centers without intensive care units to prepare for potential adverse events and account for the possibility of a systemic complication or need for sustained intubation postoperatively. The need for intubation of this cohort is a special consideration for resource management in the current COVID-19 pandemic. As elective cases are being reintroduced, arthroplasty in this vulnerable population should be closely examined to ensure appropriate support is available. We acknowledge several limitations to this study. Given that a total knee replacement has a reported survival rate of 98% at 10 years postoperatively (Argenson et al. 2013, Jauregui et al. 2015, Brockett et al. 2018, Batailler et al. 2020), and a total hip replacement reported 97% survival rates at 10 years postoperatively (Kremers et al. 2012, Fowler et al. 2019, Chaudhry et al. 2020), by only measuring joint complications through the 90-day post-discharge period, this study is limited to only short-term data and excludes long-term complications. Another limitation is the lack of data within PearlDiver regarding the type of anesthesia used on MG patients during the arthroplasties; specifically, the codes used to indicate a spinal/epidural block did not delineate whether or not a patient also received general anesthesia, which may increase the rate of complications when compared with spinal anesthesia alone (Warren et al. 2020). With the complex nature of medical billing and lack of standardized coding for a variety of conditions, there is a possibility of coding bias with the manual entry of diagnosis/procedural codes. These errors are inherent with database-driven studies using administrative claims information. Furthermore, because this study includes patient data prior to and after 2015, the diagnosis/procedural codes do not match exactly across ICD-9 and ICD-10. Accuracy was ensured with the use of a code translator to match corresponding codes. Despite the utilization of a large database with a heterogeneous patient sample, the generalizability may be limited because the data is derived from one private insurance provider, which has a greater representation in the Midwestern and Southern regions of the United States. Although we used multivariable logistic regression to diminish the confounding effect of certain potential confounding variables, it is possible that not all relevant confounding variables were identified and controlled for. Additionally, with the inclusion of each variable into our logistic regression models, adjustment bias

Acta Orthopaedica 2021; 92 (2): 176–181

was introduced into the analysis. Lastly, since the PearlDiver database provides data only on patients who retained Humana health insurance during the time period queried, sampling bias is present. Despite the regional limitation of the database used, the large size of this patient database allows for confidence in extrapolating the data to the general population. Additionally, this study is the first of its kind to examine in depth the complications seen specifically in myasthenia gravis patients when undergoing total joint replacement. Much of the prior literature has focused largely on complications seen in thymectomy operations, or on complications seen in patients with neuromuscular disorders, of which myasthenia gravis is its own subset of that group. Future research could further examine the relationship between total joint replacement and joint-specific complications for myasthenia gravis patients and other subgroups of neuromuscular disorders. In conclusion, this study suggests patients with MG experience higher rates of systemic complications in both the acute postoperative period and 90-day post-discharge period. These findings have important clinical implications for both the patient and the surgeon. Moving forward, it is recommended that patients are counseled on the higher risk of the procedure and the risks associated with systemic complications. Surgeons should also be aware of the increased risks and take the appropriate preventative measures to minimize the systemic risks associated with a THA or TKA. Supplementary data Appendix Tables A1–A3 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/ 17453674.2020.1865031 WFS, VJW, SAO, and BJR contributed to the ideas, helped in gathering data, and helped with writing and editing the paper. IDE and FLS contributed to the ideas, and helped with writing and editing the paper. Acta thanks Carlos Gonzalez and Jeppe Lange for help with peer review of this study.

Argenson J N, Boisgard S, Parratte S, S Descamps, Bercovy M, Bonnevialle P, Briard J L, Brilhault J, Chouteau J, Nizard R, Saragaglia D, Servien E. Survival analysis of total knee arthroplasty at a minimum 10 years’ followup: a multicenter French nationwide study including 846 cases. Orthop Traumatol Surg Res 2013; 99: 385-90. Batailler C, Malemo Y, Demey G, Kenney R, Lustig S, Servien E. Cemented vs uncemented femoral components: a randomized, controlled trial at 10 years minimum follow-up. J Arthroplasty 2020; 35(8): 2090-6. Boyce L, Prasad A, Barrett M, Dawson-Bowling S, Millington S, Hanna S A, Achan P. The outcomes of total knee arthroplasty in morbidly obese patients: a systematic review of the literature. Arch Orthop Trauma Surg 2019; 139: 553-60. Brockett C L, Carbone S, Fisher J, Jennings L M. Influence of conformity on the wear of total knee replacement: an experimental study. Proc Inst Mech Eng 2018; 232: 127-34.

Acta Orthopaedica 2021; 92 (2): 176–181 181

Carr A S, Cardwell C R, McCarron P O, McConville J. A systematic review of population based epidemiological studies in myasthenia gravis. BMC Neurol 2010; 10: 46. Chang Y W, Chou Y C, Yeh C C, Hu C J, Hung C J, Lin C S, Chen T L, Liao C C. Outcomes after major surgery in patients with myasthenia gravis: A nationwide matched cohort study. PLoS One 2017; 12: e0180433. Chaudhry H, MacDonald S J, Howard J L, McCalden R W, Naudie D D, Vasarhelyi E M. Clinical outcomes and midterm survivorship of an uncemented primary total hip arthroplasty system. J Arthroplasty 2020; 35(6): 1662-6. Ciafaloni E. Myasthenia gravis and congenital myasthenic syndromes. Continuum (Minneap Minn) 2019; 25: 1767-84. Cichos K H, Lehtonen E J, McGwin Jr G, Ponce B A, Ghanem E S. Inhospital complications of patients with neuromuscular disorders undergoing total joint arthroplasty. J Am Acad Orthop Surg 2019; 27: e535-e43. Correa-Valderrama A, Stangl-Herrera W, Echeverry-Velez A, Cantor E, RonTranslateur T, Palacio-Villegas J .C Relationship between body mass index and complications during the first 45 days after primary total hip and knee replacement: a single-center study from South America. Clin Orthop Surg 2019; 11: 159-63. Fang M, Noiseux N, Linson E, Cram P. The effect of advancing age on total joint replacement outcomes. Geriatr Orthop Surg Rehabil 2015; 6: 173-9. Fowler A K, Gray A R, Gwynne-Jones D P. Hybrid fixation for total hip arthroplasty showed improved survival over cemented and uncemented fixation: a single-center survival analysis of 2156 hips at 12–18 years. J Arthroplasty 2019; 34: 2711-17. Glassou E N, Pedersen A B, Hansen T B. Is decreasing mortality in total hip and knee arthroplasty patients dependent on patients’ comorbidity? Acta Orthop 2017; 88: 288-93. Jauregui J J, Cherian J J, Pierce T P, Beaver W B, Issa K, Mont M A. Longterm survivorship and clinical outcomes following total knee arthroplasty. J Arthroplasty 2015; 30: 2164-6. Kremers H M, Howard J L, Loechler Y, Schleck C D, Harmsen W S, Berry D J, Cabanela M E, Hanssen A D, Pagnano M W, Trousdale R T, Lewallen D

G. Comparative long-term survivorship of uncemented acetabular components in revision total hip arthroplasty. J Bone Joint Surg Am 2012; 94: e82. Marulli G, Schiavon M, Perissinotto E, Bugana A, Di Chiara F, Rebusso A, Rea F. Surgical and neurologic outcomes after robotic thymectomy in 100 consecutive patients with myasthenia gravis. J Thorac Cardiovasc Surg 2013; 145: 730-5; discussion 35-6. Misra U K, Kalita J, Singh V K, Kumar S. A study of comorbidities in myasthenia gravis. Acta Neurol Belg 2020; 120: 59-64. Patel A P, Gronbeck C, Chambers M, Harrington M A, Halawi M J. Gender and total joint arthroplasty: variable outcomes by procedure type. Arthroplasty Today 2020; 6: 517-20. Remes-Troche J M, Tellez-Zenteno J F, Estanol B, Garduno-Espinoza J, Garcia-Ramos G. Thymectomy in myasthenia gravis: response, complications, and associated conditions. Arch Med Res 2002; 33: 545-51. Rosenberg H, Fletcher J E. An update on the malignant hyperthermia syndrome. Ann Acad Med Singapore 1994; 23: 84-97. Rosenberg H, N Pollock, A Schiemann, T Bulger, and K Stowell Malignant hyperthermia: a review Orphanet J Rare Dis 2015; 10: 93 Singh J A, Kwoh C K, Richardson D, Chen W, Ibrahim S A. Sex and surgical outcomes and mortality after primary total knee arthroplasty: a riskadjusted analysis. Arthritis Care Res (Hoboken) 2013; 65: 1095-102. Sloan M, Premkumar A, Sheth N P. Projected volume of primary total joint arthroplasty in the US, 2014 to 2030. J Bone Joint Surg Am 2018; 100: 1455-60. Warren J, Sundaram K, Anis H, Kamath A F, Mont M A, Higuera C A, Piuzzi N S. Spinal anesthesia is associated with decreased complications after total knee and hip arthroplasty. J Am Acad Orthop Surg 2020; 28: e213e21. Wedel D J. Malignant hyperthermia and neuromuscular disease. Neuromuscul Disord 1992; 2: 157-64. Wolfe G I, Kaminski H J, Sonnett J R, Aban I B, Kuo H C, Cutter G R. Randomized trial of thymectomy in myasthenia gravis. J Thorac Dis 2016; 8: E1782-E83.

182

Acta Orthopaedica 2021; 92 (2): 182–188

Variation and trends in reasons for knee replacement revision: a multi-registry study of revision burden Peter L LEWIS 1,4, Otto ROBERTSSON 2,4, Stephan E GRAVES 1, Elizabeth W PAXTON 3, Heather A PRENTICE 3 and Annette W-DAHL 2,4 1 Australian Orthopaedic Association National Joint Replacement Registry, Adelaide, Australia; 2 Swedish Knee Arthroplasty Register, Lund, Sweden; 3 Surgical Outcomes and Analysis, Kaiser Permanente, San Diego, CA, USA; 4 Lund University, Faculty of Medicine, Clinical Science Lund, Department

Orthopedics, Lund, Sweden Correspondence: plewis@aoanjrr.org.au Submitted 2020-08-10. Accepted 2020-10-14.

Background and purpose — Studies describing timerelated change in reasons for knee replacement revision have been limited to single regions or institutions, commonly analyze only 1st revisions, and may not reflect true caseloads or findings from other areas. We used revision procedure data from 3 arthroplasty registries to determine trends and differences in knee replacement revision diagnoses. Patients and methods — We obtained aggregated data for 78,151 revision knee replacement procedures recorded by the Swedish Knee Arthroplasty Register (SKAR), the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR), and the Kaiser Permanente Joint Replacement Registry (KPJRR) for the period 2003–2017. Equivalent diagnosis groups were created. We calculated the annual proportions of the most common reasons for revision. Results — Infection, loosening, and instability were among the 5 most common reasons for revision but magnitude and ranking varied between registries. Over time there were increases in proportions of revisions for infection and decreases in revisions for wear. There were inconsistent proportions and trends for the other reasons for revision. The incidence of revision for infection showed a uniform increase. Interpretation — Despite some differences in terminology, comparison of registry-recorded revision diagnoses is possible, but defining a single reason for revision is not always clear-cut. There were common increases in revision for infection and decreases in revision for wear, but variable changes in other categories. This may reflect regional practice differences and therefore generalizability of studies regarding reasons for revision is unwise.

Although the survivorship of knee arthroplasty has improved over the last 15 years, the increased volume of primary knee replacement has led to growing numbers of revision procedures (Kumar et al. 2015, Patel et al. 2015). A prior study we undertook outlined changes in the volume and incidence of revision rates in Sweden, Australia, and the Kaiser Permanente registry from the USA (Lewis et al. 2020b). Factors influencing revision change with time. Patient factors may affect the rate of primary procedures, such as rising patient and surgeon acceptance of knee replacement (Hamilton et al. 2015), increasing rates of osteoarthritis (Hunter and Bierma-Zeinstra 2019), growing use in younger patients (Leyland et al. 2016, Karas et al. 2019), and also survivorship, such as longer life expectancy, increasing obesity, and higher physical activity of those receiving a replacement (Hamilton et al. 2015). In addition, prosthesis designs change to improve perceived shortcomings such as wear, instability, and patellofemoral pain and tracking (Lewis et al. 2020a). Methods to improve surgical precision, such as computer navigation (Jones and Jerabek 2018), image-derived instrumentation (Kizaki et al. 2019), and robotic assistance (Jacofsky et al. 2016) may decrease revision requirements (Price et al. 2018) These changing factors alter the reasons for revision. Previous studies observed a decrease in revisions for wear and loosening (Sharkey et al. 2014, Thiele et al. 2015), and related this to improved prosthesis design and materials. Other studies note infection is now the most common reason for revision (Koh et al. 2017, Postler et al. 2018). Studies of changing knee replacement failure modes are limited by being derived from single institutions or regions and may not accurately reflect what is occurring elsewhere (Sharkey et al. 2014, Thiele et al. 2015, Dyrhovden et al. 2017, Koh et al. 2017, Lum et al. 2018,

Acta Orthopaedica 2021; 92 (2): 182–188 183

Postler et al. 2018). Additionally, these studies do not show the true revision burden as they are restricted to 1st revision procedures, or only revisions of previous total knee replacements (TKR), and do not include revisions of partial knee replacement procedures. Combining registry data can be difficult due to inconsistency in the definition of revision (Liebs et al. 2015), and lack of consensus in defining modes of failure, with different terminologies used (Niinimaki 2015, Siqueira et al. 2015). Some have attempted to overcome this by defining equivalent diagnoses (Havelin et al. 2011, Paxton et al. 2011, Rasmussen et al. 2016). We determined variations and trends in reasons for knee replacement revision using data on all knee arthroplasty revision procedures from the national registries of Sweden and Australia and the institutional registry of Kaiser Permanente in the USA by using equivalent diagnosis groups (Table 1, see Supplementary data).

Patients and methods We obtained data for the period January 1, 2003 until December 31, 2017 for all revision knee replacement procedures recorded in the Swedish Knee Arthroplasty Register (SKAR), Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR), and the Kaiser Permanente Joint Replacement Registry (KPJRR). Revision knee replacements included all revision procedures of a previous replacement where 1 or more components were added, removed, or exchanged, regardless of whether this was the 2nd or subsequent procedure in chronology. Revisions of all types of knee replacement were included irrespective of whether the arthroplasty was a partial or total knee replacement. Where knee revisions were bilateral, both knees were included and recorded separately. The capture rate or completeness of these registries exceeds 95% and loss to follow-up was less than 8% over the study period. Validation and quality control methods of these registries have been published (Paxton et al. 2010, Robertsson et al. 2014, AOANJRR 2019). In all registries the reason for revision was determined from the revision diagnosis selected by the surgeon at the time of the revision procedure from a predetermined list, or specifically added. Multiple reasons could be listed. In Sweden all operative reports were methodically read and from these the primary reason for revision was interpreted by registry staff. In the AOANJRR and KPJRR, when multiple reasons for revision were recorded, a diagnosis hierarchy was used to determine the most important reason for revision. In this study only one reason for revision was permitted for each revision procedure. We included 78,151 revision knee replacement procedures. The SKAR contributed 12,612 revision procedures, the AOANJRR 53,853 revisions, and the KPJRR 11,686 revisions.

Using the categories from the SKAR as a basis, a table of equivalent diagnoses was created. For each registry the reasons for revision were then reclassified according to the “harmonized diagnosis” category. Statistics Aggregated data regarding procedure numbers, patient age, and sex were obtained for each registry (Table 2, see Supplementary data). After categorization using the equivalent diagnosis method, the number of revisions for each of the 10 most common reasons was determined and the remainder classed as “other” (Table 3, see Supplementary data). The “other” category also included a small percentage of missing data (1.1% or 137 procedures) from Sweden. The “other” group from the KPJRR contained those with a recorded diagnosis of “failed TKR,” which contributed between 3.3% and 12% of all revisions each year. For all registries the annual proportions of each harmonized revision diagnosis were calculated. For further analysis of revision for infection, the incidence per 100,000 was calculated from population data obtained from Statistics Sweden and the Australian Bureau of Statistics, as well as the yearly active membership numbers from Kaiser Permanente. Ethics, funding, and conflicts of interest Ethics approval covering the SKAR data use was issued by the Ethics Board of Lund University (LU20-02). The AOANJRR is a declared Commonwealth of Australia Quality Assurance Activity under section 124X of the Health Insurance Act, 1973. All AOANJRR studies are conducted in accordance with the ethical principles of research (Helsinki Declaration II). Approval for inclusion of data from the Kaiser Permanente Joint Replacement Registry Institutional Review Board r(#5488) was granted on November 15, 2018. There was no funding. There are no conflicts of interest.

Results Considering all revisions during the entire time period, infection was the most frequent revision diagnosis in the SKAR and KPJRR while loosening was most common in the AOANJRR. Instability, patellar causes, progression of disease, wear, and pain showed variable proportions across the registries (Figure 1.) The number of revisions and yearly proportions for each of the 10 most common reasons for revision are given in Table 3 (see Supplementary data) and a graphical representation of the proportions to highlight trends is shown in Figure 2. In all registries, there was an increase in the proportion of revisions for infection through the study period rising from 20%, 16%, and 22% in the Swedish, Australian, and KP registries in 2003 to 35%, 30%, and 43% in 2017, respectively. To determine whether this was a true rise, not just a propor-

184

Acta Orthopaedica 2021; 92 (2): 182–188

Incidence of revision for infection per 105 7

Loosening Wear

Instability 5

Infection Patellar causes

Pain 3

Progression of disease Fracture

Implant breakage

Sweden Australia KPJRR

Stiffnes Other 0

Sweden Australia KPJRR

0 2003 2005 2007 2009 2011 2013 2015 2017

Revision diagnoses (%) by registry

Figure 1. Overall revision diagnoses shown as a proportion for each registry.

Figure 3. Yearly incidence of revision knee replacement for infection per 100,000 population for the SKAR, AOANJRR, and KPJRR.

tionate increase, the yearly incidence of revision procedures for infection was calculated. This also increased in all registries (Figure 3.) Revision for loosening fell from 41% in 2003 to 13% in 2017 in the AOANJRR but a smaller decline was seen in the SKAR (27% to 23%), while the proportion in the KPJRR fell from 27% in 2003 to 14% 2008 but then rose and remained around 20% from 2011 to 2017. There was a universal decrease in revisions for wear with the proportions

declining from 6.5% to 1.5% in Sweden, 13% to 5.3% in Australia, and 21% to 4.8% in the KPJRR. Instability as a revision diagnosis showed a trend for increase in Sweden and Australia, but fluctuated in the KPJRR. Revisions for patellar reasons contributed to a higher proportion of revisions in Sweden than Australia, showing a modest increase in these 2 countries while this diagnosis was infrequent in the KPJRR. Stiffness contributed proportionally more as a revision diagnosis in the

SWEDEN

AUSTRALIA

KPJRR

Loosening

Wear

Instability

Infection

Patellar causes

Patella causes

Pain

Progression of disease

Fracture

Implant breakage

Stiffnes

Other

Yearly proportion of revision diagnoses (%)

50 0

Yearly proportion of revision diagnoses (%)

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Yearly proportion of revision diagnoses (%)

Figure 2. Yearly proportions of knee replacement revision recorded in the SKAR, the AOANJRR, and the KPJRR, respectively.

Acta Orthopaedica 2021; 92 (2): 182–188 185

KPJRR, where this reason showed a small increase with time. There was a general tendency for fewer revisions for pain throughout all registries toward the end of the time period. Progression of disease decreased over time in both Sweden and the KPJRR while it increased in Australia as a reason for revision. Fracture and implant breakage were uncommon causes of revision in all registries.

Discussion We have previously shown a decrease in all-cause revision rates in all 3 of these registries, but the reasons for revision were not studied (Lewis et al. 2020b). In the present study, when considering the entire study period, infection, loosening, and instability were among the 5 most common reasons for revision in all 3 registries; however, ranking and proportions of these varied. Over time, reasons for knee replacement revision changed, and while there were some similarities in rising proportions of revisions for infection, and decreasing proportions for wear, there were also differences between registries in 8 of the 10 most common revision reasons. These findings suggest revision reasons are partially dependent on factors specific to each healthcare system, and while variation in prosthesis use may be a major cause, analysis of this aspect is the subject of a further study. A limitation of this study is that categorizing revision diagnoses can be subjective. While many diagnoses are self-evident, in a knee replacement with pronounced wear, loosening, instability, and prosthesis breakage it can be difficult to determine which is the main cause of failure. This choice may vary between surgeons. There may be differences in interpretation: where one surgeon may nominate “progression of disease” as the reason for revision, another may record “patella erosion” for the same clinical findings. These interpretive differences can exist both within and between registries. A technique to limit the effect of this would be to correlate the revision diagnosis with the revision procedure. Using the method of equivalent diagnoses, we created a “cross-walk” between reported reasons for revision in each of the registries. Most categorizing of revision reasons is straightforward but in a few instances creation of a format to compare registry results is also open to subjectivity. For example, the diagnosis of “inflammatory arthritis” in the KPJRR has been considered as “progression of disease” but may be the equivalent to the AOANJRR diagnosis of “synovitis,” which has been classed as “other.” While malalignment is a revision diagnosis in the AOANJRR, neither the SKAR nor the KPJRR record this specific diagnosis separately, and therefore these are included in the “other” category. Registries may also have “systematic” differences in ranking of relative importance where more than 1 diagnosis is reported. These classification and ranking issues are likely to have only a small effect on the overall results.

A further limitation is that while we included all knee revision procedures to compare revision burdens and changing reasons for revision with time, we could not determine whether these changes relate to the first or subsequent revisions. However, previous registry analyses have shown that 60–85% of annual revisions are first revisions (AOANJRR 2019). There was a universal increase in proportion and yearly incidence of revisions for infection in the 3 registries studied. The reason for this worrying widespread increase is not clear, but is consistent with the findings of others (Sharkey et al. 2014, Dyrhovden et al. 2017, Koh et al. 2017). It has been suggested that debridement, antibiotics, and implant retention with only polyethylene insert exchange (DAIR) is being increasingly and more aggressively used for the treatment of periprosthetic infection (Kunutsor et al. 2018). Increases in revisions for infection are even more concerning as registries under-report infection, particularly missing non-revision episodes of treatment that do not have a prosthetic component removed or replaced (Witsø 2015, Zhu et al. 2016). In the AOANJRR, where the reason for revision is recorded at the time of operation, there may be under-reporting of infection where delayed culture results are returned as positive and, similarly, there may be a small proportion of over-reporting where a suspicion of infection is not supported by microbiological results. This type of inaccuracy would be lower in the SKAR and KPJRR as these registries can postoperatively modify the recorded diagnosis of infection on the basis of microbiological results (SKAR 2019). Revisions for wear decreased in all 3 registries, which is also a finding reported by others (Le et al. 2014, Sharkey et al. 2014, Thiele et al. 2015). Proposed reasons for this decrease are improvements in polyethylene by modified sterilization and packaging methods (Faris et al. 2006), increased use of highly cross-linked polyethylene (de Steiger et al. 2015), increased bearing conformity (Zhang et al. 2019), altered knee kinematics with femoral component design changes (Gilbert et al. 2014), or decreased tibial baseplate roughness and improved polyethylene locking mechanisms (Sisko et al. 2017). Loosening decreased as a reason for revision in both the SKAR and AOANJRR but remained unchanged in the KPJRR. The SKAR can determine which components have loosened from the operative records, but in the other 2 registries this is not possible. While an impression may be obtained by correlation with the components changed in the revision procedures, this may not be precise as, for example, if tibial loosening alone is present, both major components may be revised to allow for increased stability in the revision prosthesis configuration. Late loosening is thought to be related to wear and its consequence of osteolysis (Holt et al. 2007) and would be expected to decrease as polyethylene wear decreases. Early loosening, in contrast, most likely relates to a lack of initial fixation and is greater where cementless prostheses are used with the intent of biological fixation (Aprato

186

et al. 2016). While our study did not explore prostheses attributes, the inter-registry differences in loosening may relate to the proportional use of cementless implants or factors such as different bone cements and cementing techniques or types of polyethylene inserts used. The Swedish and Australian registries showed an increase in proportion of revision for instability. While this finding supports previous reports (Thiele et al. 2015, Dyrhovden et al. 2017), it contrasts with another where a decrease has been shown (Sharkey et al. 2014). An explanation for this change could be an increase in recognition of instability, where revisions that were once diagnosed as pain of unknown origin have increasingly been interpreted as pain due to instability (Firestone and Eberle 2006, Grayson et al. 2016). Another possibility is the development of new knowledge, with the dissemination and acceptance of the concept of mid-flexion instability during the study period (Ramappa 2015, Longo et al. 2020). There may also be a link between instability revisions and the use of posterior cruciate substituting prostheses (Hino et al. 2013). Patellar causes for revision made up a consistently higher proportion of revisions in Sweden, followed by Australia and then the KPJRR. While revisions in this category predominantly involve secondary insertion of a patellar component in a previously un-resurfaced patella and much of this difference may relate to the use of patellar components at the time of primary surgery, it also includes patellar component revisions and even patellectomy. In 2018 in Sweden there was a 3% rate of primary patellar component use (SKAR 2019), in Australia the rate of use has climbed from 42% in 2005 to 69% in 2018 (AOANJRR 2019), while in the KPJRR patellar component use has been reported at 98% (Paxton et al. 2011). Leaving the patella unresurfaced allows the potential need for a secondary resurfacing procedure. Additionally, there may be differences relating to the prostheses used with respect to generation of anterior knee pain or other patellar complications such as maltracking. While there were no consistent trends in revision for progression of disease or for pain, these 2 categories are more difficult to understand. Revision for progression of disease was higher in Sweden than in the other 2 registries, and may, in part, be explained by the possible inclusion of patellar erosion or patellar degenerative change of an un-resurfaced patella as diagnoses in this category. The proportion of knees revised for progression of disease in Sweden decreased with time, and may mirror the fall in proportional use of unicompartmental knee replacement (from 13% of primary knee replacement in 2003 to 9% in 2017) (Lewis et al. 2020b) . However, these factors cannot explain the increase in revision for progression of disease in Australia, where there has been a decrease in use of unicompartmental knee replacement (from 15% of primary knee replacement in 2003 to 6% in 2017) with an increase in patellar component use (from 41% of primary TKR in 2005 to 67% in 2017) (AOANJRR 2019). Similarly,

Acta Orthopaedica 2021; 92 (2): 182–188

this cannot explain the decline in the KPJRR where unicompartmental knee use and patellar resurfacing remained constant (at 4% and 98% respectively) (Lewis et al. 2020b, Paxton et al. 2011). (The annual procedure numbers of partial and total knee replacement for each registry have been described in our previous paper—Lewis et al. 2020b). Other covert factors, such as the inclusion of revisions of knee replacements from the time prior to the commencement of this study where the proportions of unicompartmental or patellar prosthesis use are unknown, may contribute to these findings. The revision diagnoses of fracture, stiffness, and component breakage occurred infrequently. Fracture as a reason for revision showed a small increase, which is possibly related to a globally ageing and more osteoporotic knee replacement population (Johnson et al. 2019). Revision for fracture would understate the frequency of periprosthetic fracture, as many of these are treated by means other than revision, such as fracture fixation alone. Stiffness or true arthrofibrosis is rare, and there can be cultural differences in patients, and possibly even their surgeons, proceeding to revision surgery for this reason (Springer et al. 2012). Similar to fracture, registry data does not reflect the true incidence of stiffness, as non-revision treatment methods, such as manipulation under anesthetic, are not included. A decline in implant breakage may reflect improved component durability. Of concern is the “other” diagnosis category from the KPJRR, which included a diagnosis of “failed TKR.” The true reason for revision in these procedures is unclear, but the proportion in the “other” group decreased over the study period, indicating improving precision of revision diagnosis records in this registry. The influence of this is difficult to determine. In conclusion, we have shown that despite some differences in terminology it is possible to compare registry data regarding reasons for revision. Defining a single reason for knee replacement revision is not always clear-cut. While infection, loosening, and instability are within the 5 most common reasons for revision for all 3 registries studied, their magnitude and ranking varied through the period. There were consistent increases in revision for infection, and decreases in revision for wear, but variable changes in other categories. Findings from the 3 registries studied differed, which may reflect regional differences in patient, prosthesis, or technique characteristics, and further study is required to define these practice variations. Widespread generalizability of studies regarding reasons for knee replacement revision may not be prudent. There may also be a place for defining the revision diagnoses by an international consensus, in the method Kalson et al. (2016) used for arthrofibrosis, which would give clarity, consistency, and better understanding of this area. Supplementary data Tables 1–3 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.20 20.1853340

Acta Orthopaedica 2021; 92 (2): 182–188 187

PL: conception of study, statistical analysis, interpretation of data, manuscript preparation. AWD, OR, and HP: statistical analysis, interpretation of data, manuscript preparation. SG, EP: interpretation of data and manuscript preparation. The authors would like to thank the patients and surgeons of the registries involved for contributing their valuable data, without which studies such as this would not be possible. Acta thanks Ove Furnes, Hannu Miettinen, and Tuukka Niinimäki for help with peer review of this study. AOANJRR. Annual report: hip, knee & shoulder arthroplasty. AOANJRR; 2019. Available from https://aoanjrr.sahmri.com/annual-reports-2019. Aprato A, Risitano S, Sabatini L, Giachino M, Agati G, Massè A. Cementless total knee arthroplasty. Ann Transl Med 2016; 4(7): 129. doi: 10.21037/ atm.2016.01.34. de Steiger R N, Muratoglu O, Lorimer M, Cuthbert A R, Graves S E. Lower prosthesis-specific 10-year revision rate with crosslinked than with noncrosslinked polyethylene in primary total knee arthroplasty. Acta Orthop 2015; 86(6): 721-7. doi: 10.3109/17453674.2015.1065046. Dyrhovden G S, Lygre S H L, Badawy M, Gothesen O, Furnes O. Have the causes of revision for total and unicompartmental knee arthroplasties changed during the past two decades? Clin Orthop Relat Res 2017; 475(7): 1874-86. doi: 10.1007/s11999-017-5316-7. Faris P M, Ritter M A, Pierce A L, Davis K E, Faris G W. Polyethylene sterilization and production affects wear in total hip arthroplasties. Clin Orthop Relat Res 2006; 453: 305-8. doi: 10.1097/01.blo.0000229348.10458.79. Firestone T P, Eberle R W. Surgical management of symptomatic instability following failed primary total knee replacement. J Bone Joint Surg Am 2006; 88(Suppl. 4): 80-4. doi: 10.2106/JBJS.F.00825. Gilbert S L, Rana A J, Lipman J D, Wright T M, Westrich G H. Design changes improve contact patterns and articular surface damage in total knee arthroplasty. Knee 2014; 21(6): 1129-34. doi: 10.1016/j.knee.2014.07.022. Grayson C W, Warth L C, Ziemba-Davis M M, Michael Meneghini R. Functional improvement and expectations are diminished in total knee arthroplasty patients revised for flexion instability compared to aseptic loosening and infection. J Arthroplasty 2016; 31(10): 2241-6. doi: 10.1016/j. arth.2016.03.001. Hamilton D, Howie C, Burnett R, Simpson A, Patton J. Dealing with the predicted increase in demand for revision total knee arthroplasty: challenges, risks and opportunities. Bone Joint J 2015; 97-B(6): 723-8. doi: 10.1302/0301-620X.97B6. Havelin L I, Robertsson O, Fenstad A M, Overgaard S, Garellick G, Furnes O. A Scandinavian experience of register collaboration: the Nordic Arthroplasty Register Association (NARA). J bone Joint Surg Am 2011; 93(Suppl. 3): 13-19. doi: 10.2106/jbjs.K.00951. Hino K, Ishimaru M, Iseki Y, Watanabe S, Onishi Y, Miura H. Mid-flexion laxity is greater after posterior-stabilised total knee replacement than with cruciate-retaining procedures: a computer navigation study. Bone Joint J 2013; 95-B(4): 493-7. doi: 10.1302/0301-620X.95B4.30664. Holt G, Murnaghan C, Reilly J, Meek R M. The biology of aseptic osteolysis. Clin Orthop Relat Res 2007; 460: 240-52. doi: 10.1097/ BLO.0b013e31804b4147. Hunter D J, Bierma-Zeinstra S. Osteoarthritis. Lancet 2019; 393(10182): 1745-59. doi: 10.1016/s0140-6736(19)30417-9. Jacofsky D J, Allen M. Robotics in arthroplasty: a comprehensive review. J Arthroplasty 2016; 31(10): 2353-63. doi: 10.1016/j.arth.2016.05.026. Johnson R L, Abdel M P, Frank R D, Chamberlain A M, Habermann E B, Mantilla C B. Frailty index is associated with periprosthetic fracture and mortality after total knee arthroplasty. Orthopedics 2019; 42(6): 335-43. doi: 10.3928/01477447-20190812-05.

Jones C W, Jerabek S A. Current role of computer navigation in total knee arthroplasty. J Arthroplasty 2018; 33(7): 1989-93. doi: 10.1016/j. arth.2018.01.027. Karas V, Calkins T E, Bryan A J, Culvern C, Nam D, Berger R A, Rosenberg A G, Della Valle C J. Total knee arthroplasty in patients less than 50 years of age: results at a mean of 13 years. J Arthroplasty 2019. doi: 10.1016/j. arth.2019.05.018. Kalson N S, Borthwick L A, Mann D A, Deehan D J, Lewis P, Mann C, Mont M A, Morgan-Jones R, Oussedik S, Williams F M, Toms A, Argenson J N, Bellemans J, Bhave A, Furnes O, Gollwitzer H, Haddad F S, Hofmann S, Krenn V. International consensus on the definition and classification of fibrosis of the knee joint. Bone Joint J 2016; 98-b(11): 1479-88. doi: 10.1302/0301-620x.98b10.37957. Kizaki K, Shanmugaraj A, Yamashita F, Simunovic N, Duong A, Khanna V, Ayeni O R. Total knee arthroplasty using patient-specific instrumentation for osteoarthritis of the knee: a meta-analysis. BMC Musculoskelet Disord 2019; 20(1): 561. doi: 10.1186/s12891-019-2940-2. Koh C K, Zeng I, Ravi S, Zhu M, Vince K G, Young S W. Periprosthetic joint infection is the main cause of failure for modern knee arthroplasty: an analysis of 11,134 knees. Clin Orthop Relat Res 2017; 475(9): 2194-201. doi: 10.1007/s11999-017-5396-4. Kumar A, Tsai W C, Tan T S, Kung P T, Chiu L T, Ku M C. Temporal trends in primary and revision total knee and hip replacement in Taiwan. J Chin Med Assoc 2015; 78(9): 538-44. doi: 10.1016/j.jcma.2015.06.005. 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 2018; 77(6): 479-88. doi: 10.1016/j.jinf.2018.08.017. Le D H, Goodman S B, Maloney W J, Huddleston J I. Current modes of failure in TKA: infection, instability, and stiffness predominate. Clin Orthop Relat Res 2014; 472(7): 2197-200. doi: 10.1007/s11999-014-3540-y. Lewis P L, Graves S E, de Steiger R N, Campbell D G, Peng Y, Hatton A, Lorimer M. Does knee prosthesis survivorship improve when implant designs change? Findings from the Australian Orthopaedic Association National Joint Replacement Registry. Clin Orthop Relat Res 2020a; 478(6): 1156-72. doi: 10.1097/corr.0000000000001229. Lewis P L, Graves S E, Robertsson O, Sundberg M, Paxton E W, Prentice H A, A W D. Increases in the rates of primary and revision knee replacement are reducing: a 15-year registry study across 3 continents. Acta Orthop 2020b: 1-6. doi: 10.1080/17453674.2020.1749380. Leyland K M, Judge A, Javaid M K, Diez-Perez A, Carr A, Cooper C, Arden N K, Prieto-Alhambra D. Obesity and the relative risk of knee replacement surgery in patients with knee osteoarthritis: a prospective cohort study. Arthritis Rheumatol (Hoboken, NJ) 2016; 68(4): 817-25. doi: 10.1002/ art.39486. Liebs T R, Splietker F, Hassenpflug J. Is a revision a revision? An analysis of national arthroplasty registries’ definitions of revision. Clin Orthop Relat Res 2015; 473(11): 3421-30. doi: 10.1007/s11999-015-4255-4. Longo U G, Candela V, Pirato F, Hirschmann M T, Becker R, Denaro V. Midflexion instability in total knee arthroplasty: a systematic review. Knee Surg Sports Traumatol Arthrosc 2020; Mar 5. doi: 10.1007/s00167-02005909-6. Lum Z C, Shieh A K, Dorr L D. Why total knees fail: a modern perspective review. World J Orthop 2018; 9(4): 60-4. doi: 10.5312/wjo.v9.i4.60. Niinimaki T T. The reasons for knee arthroplasty revisions are incomparable in the different arthroplasty registries. Knee 2015; 22(2): 142-4. doi: 10.1016/j.knee.2014.12.007. Patel A, Pavlou G, Mujica-Mota R, Toms A. The epidemiology of revision total knee and hip arthroplasty in England and Wales: a comparative analysis with projections for the United States. A study using the national joint registry dataset. Bone Joint J 2015; 97-B(8): 1076-81. doi: 10.1302/0301620X.97B8. Paxton E W, Inacio M C, Khatod M, Yue E J, Namba R S. Kaiser Permanente National Total Joint Replacement Registry: aligning operations with information technology. Clin Orthop Relat Res 2010; 468(10): 2646-63. doi: 10.1007/s11999-010-1463-9.

188

Paxton E W, Furnes O, Namba R S, Inacio M C, Fenstad A M, Havelin L I. Comparison of the Norwegian knee arthroplasty register and a United States arthroplasty registry. J Bone Joint Surg Am 2011; 93(Suppl. 3): 20-30. doi: 10.2106/jbjs.K.01045. Postler A, Lutzner C, Beyer F, Tille E, Lutzner J. Analysis of total knee arthroplasty revision causes. BMC Musculoskelet Disord 2018; 19(1): 55. doi: 10.1186/s12891-018-1977-y. Price A J, Alvand A, Troelsen A, Katz J N, Hooper G, Gray A, Carr A, Beard D. Knee replacement. Lancet 2018; 392(10158): 1672-82. doi: 10.1016/ s0140-6736(18)32344-4. Ramappa M. Midflexion instability in primary total knee replacement: a review. SICOT J 2015; 1:24. doi: 10.1051/sicotj/2015020. Rasmussen J V, Brorson S, Hallan G, Dale H, Aarimaa V, Mokka J, Jensen S L, Fenstad A M, Salomonsson B. Is it feasible to merge data from national shoulder registries? A new collaboration within the Nordic Arthroplasty Register Association. J Shoulder Elbow Surg 2016; 25(12): e369-e77. doi: 10.1016/j.jse.2016.02.034. Robertsson O, Ranstam J, Sundberg M, A W D, Lidgren L. The Swedish Knee Arthroplasty Register: a review. Bone Joint Res 2014; 3(7): 217-22. doi: 10.1302/2046-3758.37.2000289. Sharkey P F, Lichstein P M, Shen C, Tokarski A T, Parvizi J. Why are total knee arthroplasties failing today: has anything changed after 10 years? J Arthroplasty 2014; 29(9): 1774-8. doi: 10.1016/j.arth.2013.07.024. Siqueira M B, Klika A K, Higuera CA, Barsoum W K. Modes of failure of total knee arthroplasty: registries and realities. J Knee Surg 2015; 28(2): 127-38. doi: 10.1055/s-0034-1396014.

Acta Orthopaedica 2021; 92 (2): 182–188

Sisko Z W, Teeter M G, Lanting B A, Howard J L, McCalden R W, Naudie D D, MacDonald S J, Vasarhelyi E M. Current total knee designs: does baseplate roughness or locking mechanism design affect polyethylene backside wear? Clin Orthop Relat Res 2017; 475(12): 2970-80. doi: 10.1007/ s11999-017-5494-3. SKAR. Annual Report 2019. SKAR; 2019. Available from http://myknee.se/ pdf/SVK_2019_1.0_Eng.pdf. Springer B D, Odum S M, Nagpal V S, Lombardi AV Jr, Berend K R, Kim R H, Dennis D A. Is socioeconomic status a risk factor for stiffness after total knee arthroplasty? A multicenter case-control study. Orthop Clin North Am 2012; 43(5): e1-7. doi: 10.1016/j.ocl.2012.07.001. Thiele K, Perka C, Matziolis G, Mayr H O, Sostheim M, Hube R. Current failure mechanisms after knee arthroplasty have changed: polyethylene wear is less common in revision surgery. J Bone Joint Surg Am 2015; 97(9): 715-20. doi: 10.2106/jbjs.M.01534. Witsø E. The rate of prosthetic joint infection is underestimated in the arthroplasty registers. Acta Orthop 2015; 86(3): 277-8. doi: 10.3109/17453674. 2015.1042320. Zhang Q, Chen Z, Zhang J, Hu J, Peng Y, Fan X, Jin Z. Insert conformity variation affects kinematics and wear performance of total knee replacements. Clin Biomech (Bristol, Avon) 2019; 65: 19-25. doi: 10.1016/j.clinbiomech.2019.03.016. Zhu M, Ravi S, Frampton C, Luey C, Young S. New Zealand Joint Registry data underestimates the rate of prosthetic joint infection. Acta Orthop 2016; 87(4): 346-50. doi: 10.3109/17453674.2016.1171639.

Acta Orthopaedica 2021; 92 (2): 189–193 189

Compensation claims after knee arthroplasty surgery in Norway 2008–2018 Per-Henrik RANDSBORG 1,6, Tommy Frøseth AAE 2, Ida Rashida Khan BUKHOLM 3, Anne Marie FENSTAD 4, Ove FURNES 4,7, and Rune Bruhn JAKOBSEN 1,5 1 Department

of Orthopaedic Surgery, Akershus University Hospital, Lørenskog, Norway; 2 Department of Orthopedic Surgery, Health Møre and Romsdal HF, Kristiansund Hospital, Kristiansund, Norway; 3 Norwegian System of Patient Injury Compensation, Oslo, Norway; 4 Norwegian Arthroplasty Register, Department of Orthopaedic Surgery, Haukeland University Hospital, Bergen, Norway; 5 Department of Health Management and Health Economics, Medical Faculty, University of Oslo, Oslo, Norway; 6 Sports Medicine Institute, Hospital for Special Surgery, New York, USA; 7 Department of Clinical Medicine, Faculty of Medicine, University of Bergen, Bergen, Norway Correspondence: pran@ahus.no Submitted 2020-09-24. Accepted 2020-10-15.

Background and purpose — Orthopedic surgery is one of the specialties with most compensation claims. We assessed the claims following knee arthroplasty surgery reported to the Norwegian System of Patient Injury Compensation (NPE) in light of institutional procedure volume. Patients and methods — We collected data from NPE and the Norwegian Arthroplasty Register (NAR) for the study period (2008–2018). Age, sex, type of claim, and reason for compensation were collected from NPE, while the number of arthroplasty surgeries was collected from NAR. The treating hospitals were grouped by quartiles according to annual procedure volume. The effect of hospital volume on the likelihood of an accepted claim was estimated. Results — NAR received 64,241 reports of arthroplasty procedures, of which 572 (0.9%) patients filed a claim for treatment injury. 55% of the claims were accepted, representing 0.5% of all knee arthroplasties. The most common reason for accepted claim was a hospital-acquired infection, in 28% of the patients, followed by misplaced implant (26%) and aseptic loosening (13%). The hospitals with the lowest annual volume (57 or fewer arthroplasties per year, first quarter) had a statistically significantly larger fraction of granted claims per procedures compared with other institutions. Interpretation — The overall risk of ending up with compensation due to treatment error following knee arthroplasty was 0.5%. The risk of accepted claim was greater for patients operated in the lowest volume hospitals.

The number of knee arthroplasty procedures in Norway has increased over the last decade and is now over 7,000 per year (Ackerman et al. 2017, NAR 2020). About 1 in 5 patients receiving a TKA remains dissatisfied with the result (Gunaratne et al. 2017). Although serious complications are rare, infections, implant loosening, misplaced implants, residual pain, and other complications do occur, with potential detrimental results. To monitor the safety of implants and define the epidemiology of the procedures, the Norwegian Arthroplasty Register (NAR) was established in 1987 (Havelin et al. 2000). NAR provides a comprehensive overview of knee arthroplasties taking place in Norway. Compliance with the registry is 97.6% for primary TKA and 93.2% for revisions (Wiik 2014). Patients who suffer an injury while receiving health services, within either the public or the private healthcare sector, can file a claim with the Norwegian System of Patient Injury Compensation (NPE). 3 criteria must be fulfilled for a claim to be accepted: 1. The injury must have been caused during health services (diagnosis, examination, treatment, care, or lack of such), even if no one is to blame. If the injury is severe and unexpected, compensation may be awarded even where no error or omission in treatment has occurred (for example if infection occurs despite adequate prophylaxis). 2. The injury must have caused financial loss to the patient, except if the injury leads to permanent medical impairment of more than 15%, in which case compensation might be awarded despite financial loss. This might be relevant for retired patients or for patients who can continue to work in spite of the disability.

190

Acta Orthopaedica 2021; 92 (2): 189–193

Table 1. Demography of knee arthroplasty procedures reported to the Norwegian Arthroplasty Registry (NAR) and compensation claims filed with the Norwegian System of Patient Injury Compensation (NPE) during 2008–2018 Knee procedures Compensation Accepted reported claims filed claims to NAR to NPE n = 312 Factor n = 64,241 n = 572 (54.5%)

Rejected claims n = 260 (45.5%) p-value a

Mean age (SD) range Females, n (%)

61 (11) 24–85 155 (60)

68 (9.7) 62 (10) 12–101 24–86 38,352 (60) 324 (57)

63 (9.9) 38–86 169 (54)

< 0.001

SD, standard deviation. a Chi-square test comparing proportion of women reported to NAR with proportion of women filing a complaint to NPE.

3. The patient must file a claim within a reasonable time (currently set at 3 years) after the patient realizes that the injury is caused by the treatment or lack of treatment received. The claim is filed with NPE at no cost to the patient. There is compelling evidence that low surgical volume increases the risk of complications and revision after knee arthroplasty surgery (Jaeschke et al. 1989, Badawy et al. 2013, Pamilo et al. 2015, Badawy et al. 2017). Whether this association is also true for injury compensation has not been studied. We evaluated the claims following primary and revision knee arthroplasty surgery filed with NPE and compared the findings with the results from NAR with a focus on annual hospital procedure volume.

Patients and methods Data from NAR was collected for the study period (2008 through 2018). The data was stratified by the number of arthroplasty procedures performed annually per hospital. The hospitals were then divided into quarters according to average annual procedure volume. The lowest quarter (Q1) represented 6 institutions with ≤ 57 knee arthroplasty procedures per year. The 2nd quarter (Q2) consisted of 8 institutions with an annual volume of 58–168 procedures, the 3rd quarter (Q3) included 8 institutions with an annual volume of 169–304 procedures, and finally the 4th quarter (Q4) contained 7 institutions with 305 or more knee arthroplasty procedures per year. All claims filed with NPE following knee arthroplasty surgeries that were performed during the study period were collected. The data were stratified by institution, the patient’s age, sex, type of complication, and any reoperations. The reason for the claim was recorded, together with the decision made by NPE (accepted or rejected claim). Statistics Continuous variables are presented as mean, median, 95% confidence interval (CI), range, and standard deviation (SD), while categorical data is presented in frequencies. Groups

Table 2. Distribution of implant type among 312 accepted compensation claims from the Norwegian System of Patient Injury Compensation during 2008– 2018 Implant type Primary total knee arthroplasty cemented non-cemented hybrid technique Primary unicondylar knee arthroplasty cemented non-cemented Primary patellofemoral arthroplasty cemented non-cemented Secondary total knee arthroplasty cemented hybrid technique Secondary patellofemoral arthroplasty cemented Other, unspecified

Number (%) 225 (72) 23 (7) 21 (7) 21 (7) 4 (1) 1 (0.3) 1 (0.3) 9 (3) 2 (0.6) 1 (0.3) 4 (1)

were hence compared using the 2-sample independent t-test or the chi-square test. We compared the institutions by procedure volume using ANOVA after asserting conditions were met, and p-values adjusted for multiple testing by Tukey’s comparison test. Associations were quantified by odds ratio. A p-value < 0.05 was considered statistically significant. The analysis was performed using the Statistical Package for Social Sciences (SPSS) version 25 (IBM Corp, Armonk, NY, USA). Ethics, funding, and potential conflicts of interest The Regional Ethical Committee (REK) has deemed approval not necessary as all data are based on already anonymized records (REK 15.10.10). This study received no external funding. The authors declare no conflicts of interests.

Results During the study period 2008–2018, 64,241 knee arthroplasty procedures were reported to NAR. There were 59,109 primary knee arthroplasties, of which 6,788 (12%) were unicompartmental knee arthroplasties (UKA). There were 5,132 (8%) revision arthroplasties. NPE received 572 claims for treatment injuries related to arthroplasties performed during these years, representing 0.9% of all knee arthroplasty procedures. The average age at the time of surgery for the claimants was 62 (range 24–86) years, and 57% of the claims were filed by women (Table 1). 312 (55%) claims were accepted, representing 0.5% of all knee arthroplasties reported to NAR in the period. 259 of the claims were accepted following a primary TKA (0.5% of all primary TKAs) and 25 claims were accepted following a UKA (0.4% of all UKAs, p = 0.2). 11 claims were accepted after revision arthroplasties (0.2% of all revisions) (Table 2).

Acta Orthopaedica 2021; 92 (2): 189–193 191

Table 3. Reasons for accepted claims for treatment injuries following knee arthroplasty surgeries in Norway during 2008–2018. Values are count (%)

Proportion of accepted claims (%) p = 0.03

2.5

p = 0.02 p = 0.04

Reason for accepted claim

All cases TKA UKA PFA n = 312 n = 284 n = 25 n = 3

Hospital-acquired infection 87 (28) Misplaced implant 81 (26) Early aseptic loosening a 40 (13) Wrong indication 27 (9) Wrong choice of implant 12 (4) Peroperative nerve injury 12 (4) Inadequate follow-up 10 (3) Peroperative vascular injury 10 (3) Peroperative tendon/ ligament injury 9 (3) Wrong or inadequate antithrombotic prophylaxis 4 (1) Peroperative fracture 4 (1) Nerve injury due from bandage/cuff 3 (1) Pain 3 (1.0) Delayed treatment 3 (1.0) Wrong technique 3 (1.0) Radial nerve damage from patient positioning 1 (0.3) Burn injury from diathermic plate 1 (0.3) Decubitus from inadequate postoperative care 1 (0.3) Other 1 (0.3)

2.0

75 (26) 77 (27) 37 (13) 23 (8) 10 (4) 11 (4) 10 (4) 10 (4)

11 4 3 3 2 - - -

1 1 1 -

1.5

8 (3)

4 (1) 4 (1)

- -

3 (1) 3 (1) 3 (1) 2 (0.5)

- - - 1

1 (0.3) 1 (0.3)

- -

1 (0.3) 1 (0.3)

- -

TKA, total knee arthroplasty; UKA, unicompartmental knee arthroplasty; PFA, patellofemoral arthroplasty. a Within 3 years.

The most common reason for accepted claim was a hospital-acquired infection, in 87 (28%) patients, followed by misplaced implant (81 patients, 26%), and aseptic loosening (40 patients, 13%). Nearly 9% of the claims were accepted due to the wrong indication (Table 3). 15 of 27 claims from private institutions were accepted, compared with 248 (46%) of 545 claims from public hospitals (p = 0.9). 2 claims involving fatalities were recorded, both related to thrombosis prophylaxis. A 78-year-old male undergoing a primary cemented TKA was given low molecular weight heparin while also taking celecoxib, and died from a bleeding gastric ulcer. A 64-year-old male with known increased thrombotic risk received a hybrid primary TKA. He did not receive guideline anticoagulation, and died of acute cerebral stroke. In both cases NPE granted compensation to the survivors. Hospital procedure volume The lowest volume hospitals (< 57 arthroplasties per year, Q1) had a statistically significantly larger fraction of accepted claims per procedures compared with other institutions (Figure). The odds ratio for receiving compensation after a knee arthroplasty performed in a low-volume hospital (Q1) was 3 (CI 2–5) compared with surgery performed in a hospital with higher procedure volume (Table 4). There were no sta-

p = 1.0 p = 1.0

p = 1.0

1.0

0.5

Procedure volume categories

Proportion of accepted claims by number of surgeries stratified by annual hospital procedure volume. The 4 categories represent quarters, see Table 4. P-values derived from ANOVA adjusted with Tukey’s comparison test. Table 4. Risk of accepted compensation claims from the Norwegian System of Patient Injury Compensation during 2008–2018 by annual procedure volume divided into quarters

Table 5. Proportion of accepted claims (accepted claims/total claims) for each quarter of annual procedure volume

Quarter (Q) Odds ratio (95% CI)

Accepted/ Quarter total claims (%)

Q1 vs, all others Q1 vs. Q2 Q1 vs. Q3 Q1 vs. Q4 Q2 vs. Q3 Q2 vs. Q4 Q3 vs. Q4

Q1 Q2 Q3 Q4 Total

3.0 (2.0–4.5) 2.7 (1.7–4.4) 2.6 (1.7–4.1) 3.4 (2.2–5.2) 1.0 (0.7–1.4) 1.3 (0.9–1.8) 1.3 (1.0–1.7)

Q1, ≤ 57 annual procedures; Q2, 58–168 annual procedures; Q3, 169–304 annual procedures; Q4, ≥ 305 annual procedures. CI, confidence interval.

26/41 (63) 47/112 (42) a 86/159 (54) 153/260 (59) 312/572 (55)

Q1–Q4, see Table 4 a Omnibus chi-square statistic was significant at p = 0.02. Analysis of adjusted standardized residuals revealed Q2 to be the main contributor with a z-score of 3.0.

tistically significant differences in accepted claims per annual procedure volume between the other 3 procedure volume quarters (Q2–Q4). When considering all claims filed, the lowest volume quarter (Q1) had a similar ratio of accepted claims compared with the 2 highest volume quarter (Q3 and Q4). The 2nd quarter (Q2) had a lower ratio of accepted claims compared with Q1 and Q3 (Table 5).

Discussion Our main finding is that patients operated on in the lowest volume hospitals have a 3-fold risk of being granted a compensation claim following knee arthroplasty surgery compared with patients being treated in higher volume institu-

192

tions. Overall, about 0.5% of knee arthroplasty patients were compensated by NPE due to treatment failure. The main reason for compensation was hospital-acquired infection. This is also the most common reason for accepted claims after other elective knee procedures, such as anterior cruciate ligament reconstruction and cartilage surgery (Randsborg et al. 2018, Aae et al. 2020). The proportion of accepted claims due to infection was particularly high for UKA compared with TKA (p = 0.06). Nearly 60% of knee arthroplasties reported to NAR are performed on women, yet only 57% of compensation claims were filed by women. This small but statistically significant difference might be explained by the fact that men have a 2-fold increased risk of revision due to deep infection (Badawy et al. 2017). Normally, an error in the healthcare provided is needed for a claim to be accepted. However, an exception is made for severe and rare complications, even if no treatment error has been identified. This explains the high rate of compensation following infection and early (within 3 years) aseptic loosening found in our study. From a surgical point of view, it is interesting that misplaced implant is the 2nd greatest reason for compensation, representing over a quarter of the accepted claims. More claims from patients treated at the lowest volume hospitals may relate to the fact that both surgical and hospital volume affects clinical results and complication rates (Soohoo et al. 2006, Paterson et al. 2010, Badawy et al. 2013, Pamilo et al. 2015). Furthermore, it is worth noting that nearly 9% of the claims were accepted due to the wrong indication. The relative success of knee arthroplasty must not lead clinicians to utilize a one-size-fits-all solution for knee pain. The proportion of claims accepted due to the wrong indication or wrong choice of implant was particularly high for UKA compared with TKA, but did not reach statistical significance. Nevertheless, our findings serve as a reminder that nonoperative measures and correlation between radiological findings and clinical symptoms remain cornerstones in the indication for knee arthroplasty (Schmitt et al. 2017). Unfortunately, avoidable treatment injuries such as wrong indication or inadequate follow-up remain major reasons for accepted claims in orthopedic surgery (Randsborg et al. 2018, Aae et al. 2020). These are modifiable factors and demonstrate how compensation claim reports can be useful for clinicians to learn from other people’s mistakes. By focusing on proper indication, surgical technique, and follow-up routine, the number of adverse outcomes and thus compensation claims will likely be reduced. Another interesting finding in our study is that the risk of compensation following primary UKAs was not higher than for primary TKAs. This came as some surprise to us, because the revision rate for UKAs is higher than for TKAs (Chawla et al. 2017, Arias-de la Torre et al. 2019, Jennings et al. 2019). However, the likelihood of compensation is not identical to the risk of complication. For a claim to be accepted, a treatment failure must have occurred. A revision of a UKA does not necessary indicate that the primary surgery was a treat-

Acta Orthopaedica 2021; 92 (2): 189–193

ment error. There is a lower threshold for surgeons to revise a painful UKA (Johnson et al. 2020), which may explain the higher rate of revision of UKA, but no difference in likelihood of treatment injury compensation. Procedure volume has been a hot topic since 1979, when Luft et al. asked the simple question: Should operations be regionalized? The effect on both hospital and surgeon volume on adverse outcome has been discussed across medical fields since then, and there is little doubt that very low volume increases risk of adverse outcome. For knee arthroplasties in particular, several authors have concluded that surgery performed in a lowvolume hospital increases the risk of adverse outcome (Soohoo et al. 2006, Paterson et al. 2010, Badawy et al. 2013, Pamilo et al. 2015). Our study confirms that the likelihood of compensation due to treatment injury following knee arthroplasty is also increased in low-volume institutions. In support of this, a report from Finland found that hospital volumes of less than 200 annual arthroplasty procedures were associated with more compensated treatment errors (Jarvelin et al. 2012). However, 200 procedures per year would in our study place the hospital in the 2nd highest quarter. Therefore, our study provides a more detailed analysis of medium institutional procedure volume. Notably, only the lowest quarter (< 57 procedures per year) had a significant increased likelihood of accepted claims. Limitations There are several limitations to this study. Some patients, who underwent surgery towards the end of the study period, may not yet have filed a compensation claim. There could be regional and institutional differences in the culture of claiming compensation or the information given to patients concerning the possibility of filing a complaint. It is also likely that some treatment errors were never reported to NPE. Furthermore, individual surgeons’ annual procedure volume was not available, which could influence results. There could be some variation in annual procedure volume during the study period causing fluctuation between quarters. Our data is collected from a single country, with a public compensation scheme based on the principle of no blame. This is similar to systems in other Nordic countries, but different from countries such as the United Kingdom, the United States, Italy, and Germany that have a tort-based system. This may limit the generalizability of our study. However, our purpose was not to compare different compensation schemes, but to analyze the causes and aspects of compensation following knee arthroplasty surgery. It is important to point out that this is not a study on complications following knee arthroplasty. Most complications will never be reported to NPE. A review of compensation claims investigates the quality of the healthcare provided, not the outcome of the medical procedures in question. In summary, the overall likelihood of ending up with compensation due to treatment error following knee arthroplasty was 0.5%. The likelihood was 3 times greater for patients operated on in the lowest volume hospitals.

Acta Orthopaedica 2021; 92 (2): 189–193 193

PHR: original idea, study design, interpretation of data, drafting of manuscript. TFA: original idea, interpretation of data, revision of manuscript. IRKB: data acquisition, revision of manuscript. AMF: data acquisition, data analysis, revision of manuscript. OF: collection of data, interpretation of data, revision of manuscript. RBJ: study design, collection and interpretation of data, revision of manuscript. Acta thanks Pelle Gustafson and Maiju Welling for help with peer review of this study.

Aae T F, Lian O B, Aroen A, et al. Compensation claims after knee cartilage surgery is rare: a registry-based study from Scandinavia from 2010 to 2015. BMC Musculoskelet Disord 2020; 21(1): 287. doi: 10.1186/s12891-02003311-4. Ackerman I N, Bohensky M A, de Steiger R, et al. Substantial rise in the lifetime risk of primary total knee replacement surgery for osteoarthritis from 2003 to 2013: an international, population-level analysis. Osteoarthritis Cartilage 2017; 25(4): 455-61. doi: 10.1016/j.joca.2016.11.005. Arias-de la Torre J, Valderas J M, Evans J P, et al., Catalan Arthroplasty Register Steering C. Differences in risk of revision and mortality between total and unicompartmental knee arthroplasty: the influence of hospital volume. J Arthroplasty 2019; 34(5): 865-71. doi: 10.1016/j.arth.2019.01.046. Badawy M, Espehaug B, Indrekvam K, et al. Influence of hospital volume on revision rate after total knee arthroplasty with cement. J Bone Joint Surg Am 2013; 95(18): e131. doi: 10.2106/JBJS.L.00943. Badawy M, Fenstad A M, Bartz-Johannessen C A, et al. Hospital volume and the risk of revision in Oxford unicompartmental knee arthroplasty in the Nordic countries: an observational study of 14,496 cases. BMC Musculoskelet Disord 2017; 18(1): 388. doi: 10.1186/s12891-0171750-7. Chawla H, van der List J P, Christ A B, et al. Annual revision rates of partial versus total knee arthroplasty: a comparative meta-analysis. Knee 2017; 24(2): 179-90. doi: 10.1016/j.knee.2016.11.006. Gunaratne R, Pratt D N, Banda J, et al. Patient dissatisfaction following total knee arthroplasty: a systematic review of the literature. J Arthroplasty 2017; 32(12): 3854-60. doi: 10.1016/j.arth.2017.07.021.

Havelin L I, Engesaeter L B, Espehaug B, et al. The Norwegian Arthroplasty Register: 11 years and 73,000 arthroplasties. Acta Orthop Scand 2000; 71(4): 337-53. doi: 10.1080/000164700317393321. 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. doi: 10.1016/0197-2456(89)90005-6. Jarvelin J, Hakkinen U, Rosenqvist G, et al. Factors predisposing to claims and compensations for patient injuries following total hip and knee arthroplasty. Acta Orthop 2012; 83(2): 190-6. doi: 10.3109/17453674.2012.672089. Jennings J M, Kleeman-Forsthuber L T, Bolognesi M P. Medial unicompartmental arthroplasty of the knee. J Am Acad Orthop Surg 2019; 27(5): 16676. doi: 10.5435/JAAOS-D-17-00690. Johnson W B Jr, Engh C A Jr, Parks N L, et al. A lower threshold for revision of aseptic unicompartmental vs total knee arthroplasty. Bone Joint J 2020; 102-B(6_Supple_A): 91-5. doi: 10.1302/0301-620X.102B6.BJJ-20191538.R1. Luft H S, Bunker J P, Enthoven A C. Should operations be regionalized? The empirical relation between surgical volume and mortality. N Engl J Med 1979; 301(25): 1364-9. doi: 10.1056/NEJM197912203012503. NAR. Norwegian Arthroplasty Register. Annual Report; 2020. Pamilo K J, Peltola M, Paloneva J, et al. Hospital volume affects outcome after total knee arthroplasty. Acta Orthop 2015; 86(1): 41-7. doi: 10.3109/17453674.2014.977168. Paterson J M, Williams J I, Kreder H J, et al. Provider volumes and early outcomes of primary total joint replacement in Ontario. Can J Surg 2010; 53(3): 175-83. Randsborg P H, Bukholm I R K, Jakobsen R B. Compensation after treatment for anterior cruciate ligament injuries: a review of compensation claims in Norway from 2005 to 2015. Knee Surg Sports Traumatol Arthrosc 2018; 26(2): 628-33. doi: 10.1007/s00167-017-4809-y. Schmitt J, Lange T, Gunther K P, et al. Indication criteria for total knee arthroplasty in patients with osteoarthritis: a multi-perspective consensus study. Z Orthop Unfall 2017; 155(5): 539-48. doi: 10.1055/s-0043-115120. Soohoo N F, Zingmond D S, Lieberman J R, et al. Primary total knee arthroplasty in California 1991 to 2001: does hospital volume affect outcomes? J Arthroplasty 2006; 21(2): 199-205. doi: 10.1016/j.arth.2005.03.027. Wiik R B. Dekningsgradanalyse for kneproteseregisteret 2008–2012. Norwegian Arthroplasty Register: Helsedirektoratet; 2014

194

Acta Orthopaedica 2021; 92 (2): 194–198

Innervation of the distal part of the vastus medialis muscle is endangered by splitting its muscle fibers during total knee replacement: an anatomical study using modified Sihler’s technique Bettina PRETTERKLIEBER 1, Alfred UNGERSBÖCK 2, and Michael L PRETTERKLIEBER 1 1 Division

of Anatomy, Center for Anatomy and Cell Biology, Medical University of Vienna, Vienna; 2 Department of Orthopaedic and Trauma Surgery, Federal Hospital Neunkirchen, Neunkirchen, Austria Correspondence: bettina.pretterklieber@gmail.com Submitted 2020-09-14. Accepted 2020-11-02.

Background and purpose — The distal part of the vastus medialis muscle is an important stabilizer for the patella. Thus, knowledge of the intramuscular nerve course and branching pattern is important to estimate whether the muscle’s innervation is at risk if splitting the muscle. We determined the intramuscular course of the nerve branches supplying the distal part of the vastus medialis muscle to identify the surgical approach that best preserves its innervation. Material and methods — 8 vastus medialis muscles from embalmed anatomic specimens underwent Sihler’s procedure to make soft tissue translucent while staining the nerves to study their intramuscular course. After dissection under transillumination using magnification glasses all nerve branches were evaluated. Results — The terminal nerve branches were located in different layers of the muscle and ran mostly parallel but also transverse to the muscle fibers. In half of the cases, the latter formed 1 to 3 anastomoses and coursed close to the myotendinous junction. Additionally, most of the branches extended into the ventromedial part of the knee joint capsule. Interpretation — To preserve the innervation of the distal part of the vastus medialis muscle, any split of the muscle during surgical approaches to the knee joint should be avoided.

Based on the different orientation of the vastus medialis muscle fiber bundles, Lieb and Perry (1968) were the first authors who described 2 parts, i.e., the vastus medialis longus and the vastus medialis obliquus. Whether these 2 parts really exist as 2 individual muscles is debated (Speakman and Weisberg 1977, Hubbard et al. 1997, Peeler et al. 2005, Smith et al. 2009). In any event, these caudal oblique coursing muscle fiber bundles, which insert into the cranial part of the medial margin of the patella, are said to be essential for so-called patellar tracking (Goh et al. 1995, Toumi et al. 2007, Lin et al. 2008). The vastus medialis muscle is innervated by the femoral nerve. The branch for the distal portion of the vastus medialis muscle courses outside the adductor canal before it enters the muscle in the middle third of its belly. Some final sensory twigs of this branch have been reported to reach also the ventromedial part of the capsule of the knee joint, thus apparently being part of its proprioception (Thiranagama 1990, Horner and Dellon 1994, Nozic et al. 1997). Both these functions of these nerves seem to play an important role for the function of the knee joint. Therefore, these nerves and their terminal branches should be preserved as much as possible during medial approaches to the knee joint. Surgical incision into the knee joint for total knee replacement can be done in different ways: the classical medial parapatellar approach (Cooper et al. 1999), the subvastus approach (Erkes 1929, Halder et al. 2009), and as a compromise the midvastus approach. Here, the incision divides the distal part of the vastus medialis (Hube et al. 2009), which may interrupt its innervation. Parentis et al. (1999) and Kelly et al. (2006) have already reported abnormal electromyographical findings following this approach.

Acta Orthopaedica 2021; 92 (2): 194–198 195

We determined the intramuscular course and the final destination of the nerve branches within the distal part of the vastus medialis muscle to identify the approach that best preserves its innervation.

Material and methods We examined 8 vastus medialis muscles together with the adjacent part of the capsule of the knee joint. They were taken from the right leg of 3 male and 5 female embalmed anatomic specimens from our student dissection course. The age of the deceased individuals was on average 80 years (56–94). The bodies of the deceased persons were perfused and immerged with a low-percentage formalin-phenol solution. We used muscles only from sites without any signs of surgical intervention within the anterior femoral region and knee joint. Immediately after excising the muscles, we performed whole-mount nerve staining using the modified Sihler’s technique (Liu et al. 1997, Mu and Sanders 2010) to render soft tissue translucent or transparent while staining the nerves to study their intramuscular course and pattern. As the muscles originated from formalin-fixed specimens, we skipped the first step, i.e., fixation in 10% formalin. This modification has already successfully been done by Sekiya et al. (2005). To make the tissue transparent we macerated the muscles with a 3% potassium hydroxide (KOH) solution with 0.2 mL 3% hydrogen peroxide per 100 mL. The next step was decalcification in Sihler’s solution I (1 equivalent glacial acetic acid, 1 equivalent glycerin, 6 equivalents 1% aqueous chloral hydrate). After that, we stained the tissues using Sihler’s solution II (1 equivalent stock Ehrlich’s hematoxylin, 1 equivalent glycerin, 6 equivalents 1% aqueous chloral hydrate). To decolor the muscle fibers and connective tissue again, we destained the muscles again using Sihler’s solution I. Following neutralization in a 0.05 % lithium carbonate solution, we put the muscles into 50% aqueous glycerin for clearing. Finally, we stored them in 100% glycerin with a few thymol crystals as antiseptic agent. To record the course of the fine nerve twigs, which innervate the distal part of the vastus medialis muscle and the adjacent part of the capsule of the knee joint, we dissected the nerves under transillumination with a white light transilluminator using magnification glasses. We removed some of the superficial muscle fibers to make the whole muscle more transparent. To gain a better view of the course of the nerve branches, intramuscular arteries and veins were mostly resected. To compare the 2 medial minimal invasive approaches under discussion, we simulated them on 2 nonembalmed anatomic specimens. We took photographs with a digital reflex camera. Figure 1 shows an example of a vastus medialis muscle before and after Sihler’s procedure and further dissection.

Figure 1. Right vastus medialis muscle of a 56-year-old woman. The muscle was removed from a formalin-fixed specimen and marked with yellow yarn (a). The same muscle after Sihler’s procedure transilluminated by a white light transilluminator. The nerve branches are stained dark blue whereas the muscle fibers show a transparent lavender color after the destaining and clearing process. Some of the superficial muscle fibers were removed to show the whole course of all nerve branches. The ventrolateral part of the capsule of the knee joint appears amber (b).

Ethics and funding The bodies had been donated to medical education and research at our university. In addition to the informed consent of the body donors, approval was obtained from the ethics committee of our university (approval number: 1826/2017). No funding was received for this study.

Results In all 8 vastus medialis muscles, 4 to 8 nerve branches were recorded within the distal part of this muscle. Consistently, the most distal branch ran alongside the posterior margin of the muscle and gave rise to several other branches. All branches were located in different layers of the muscle. Although they coursed more or less parallel to the muscle fibers (Figure 2a), in 7 out of 8 cases, 1 to 7 of the branches also crossed the muscle fibers in a transverse direction (Figure 2b–d). These traversing branches often coursed close to the myotendinous junction. Due to further division, 4 to 9 branches extended into the ventromedial part of the fibrous capsule of the knee joint. In 4 cases, 1 to 3 anastomoses between 2 of the nerve branches were observed running perpendicular to the muscle

196

Acta Orthopaedica 2021; 92 (2): 194–198

(Mu and Sanders 2010). Thus, studies using this method are usually based on a small sample size, e.g., 3 human tongues (Doty et al. 2009), abdominal walls from 5 rats (Calguner et al. 2006), or the posterior cricoarytenoid muscle of 10 dogs (Drake et al. 1993). We used 8 specimens. Due to the use of this method, it was possible to trace the nerves without severing the topographical relationship to each other and to the muscle fibers. It revealed that the terminal nerve branches within all layers of the distal part of the vastus medialis muscle course parallel to the muscle fibers but also traverse them. Thereby they build anastomoses with each other. These traversing branches course close to the myotendinous junction. The results are a valuable addition to the findings of Ehler et al. (1959) and Jojima et al. (2004), who described the course of the nerves only by pure macroscopic dissection, preventing them from analyzing the branches in the detailed way we did by using Sihler’s procedure, i.e., the small anastomoses and traversing branches. As the innervation patterns of each vastus medialis muscle in detail seems to be prone to interindividual variation; one cannot predict if an intramuscular nerve branch will be severed by cutting through or between the muscle fibers. The anastomoses and the Figure 2. Detailed view of the transilluminated distal part of 4 right vastus medialis muscles transverse coursing branches seem thereby and the adjacent ventromedial part of the capsule of the knee joint: (a) from a 56-year-old especially vulnerable. woman, (b) from an 83-year-old woman, (c) from a 67-year-old woman, (d) from a 82-yearold man. The nerve branches are located in different layers of the muscle. They run parallel There are different medial approaches (single arrows) or transverse (double arrows) to the muscle fibers and finally reach the veninto the knee joint for total knee replacetromedial part of the capsule of the knee joint. In half of the subjects, they build 1 or more ment. Each has its advantages and disanastomoses (arrowhead) within the distal part of the vastus medialis muscle. advantages. During the standard medial parapatellar approach, the tendon of the fibers (Figure 2c–d). These anastomotic branches also fre- vastus medialis muscle is interrupted. This approach leads to quently coursed close to the myotendinous junction. In addi- a good overview of the joint and can be done in nearly all tion to this primary pattern, the nerve branches often formed patients. However, using this procedure, the extensor apparatus of the knee joint is severed, which was seen in former networks (Figure 2c). times as a reason for postoperative patellar tracking problems (Clayton and Thirupathi 1982, Cooper et al. 1999). Therefore, a lateral release is sometimes performed simultaneously Discussion (Keblish 2002). In addition, nowadays these problems are supOur study is the first to describe the intramuscular course of posed to be created, rather, by a malrotation of the implant the nerve branches innervating the vastus medialis muscle (van Rensch et al. 2020). The subvastus approach (Figure 3) using the modified Sihler’s technique. This technique has been preserves the integrity of the vastus medialis muscle (Erkes reported to be superior to microdissection, or 3D reconstruc- 1929). Hofmann et al. (1991) reported an equivalent exposure tion for observing motor nerve supply patterns in muscles, as compared with the parapatellar approach. However, some the 3D structure of the whole specimen can be preserved (Mu authors are convinced that this procedure is more difficult to and Sanders 2010). The modified Sihler’s technique is a time- perform and only provides diminished visibility of the joint consuming process, which requires several weeks or months surfaces (Keblish 2002, Halder et al. 2009). As a compromise,

Acta Orthopaedica 2021; 92 (2): 194–198 197

Figure 3. Simulation of the subvastus approach performed in a non-embalmed anatomic specimen. The vastus medialis muscle (vm) is preserved during the subvastus approach, as the dissection follows its caudal border (a, b). To gain more space, the vastus medialis muscle can be mobilized from the tendon of the adductor magnus by blunt dissection. The muscle together with the patella (P) can then be lateralized to get access to the joint surfaces (c).

neurosis but splitting the distal part of the muscle parallel to its fibers (Dalury and Jiranek 1999). This approach is also said to be more difficult than the parapatellar approach, due to the lesser reachability of the joint surfaces (Keblish 2002, Hube et al. 2009). Cooper et al. (1999) stated that this approach does not harm the innervation of the distal part of the vastus medialis muscle. They have postulated that one can cut 4 cm through the muscle and an additional 2 cm remain as a safe distance for blunt dissection. Apparently, the authors have only regarded the extramuscular course of the nerves supplying the distal part of the vastus medialis muscle. In contrast, our results suggest that the traversing branches can almost reach the myotendinous junction, indicating the safe zone of 4 cm stated by Cooper et al. (1999) to be inappropriate. Indeed, the midvastus approach is controversial. Parentis et al. (1999) reported abnormal postoperative electromyographical recordings in 9 of 21 knees undergoing the midvastus approach. In 2 of these 9 knees, these irregularities still exist after more than 5 years, even though Figure 4. Simulation of the midvastus approach performed in a non-embalmed anatomic speci- without discernible functional deficits men. The distal part of the vastus medialis muscle (vm) is split about 3 cm cranial from its caudal border (a, b, c). The cranial part of the muscle together with the patella (P) can be lateralized to (Kelly et al. 2006). In one-third of the gain access to the joint (d). midvastus group observed by Jojima et al. (2004) main nerve branches within the midvastus approach (Figure 4) has been developed. The the distal part of the vastus medialis had been disrupted. On cutting line is similar to the medial parapatellar approach leav- the other side, Dalury et al. (2008) found similar postoperaing the vastus medialis muscle in continuation with its apo- tive electromyographical recordings following the medial

198

parapatellar and the midvastus approaches. As they only cut sharp for a distance of 5 cm, they were convinced that blunt dissection is more harmful to the nerves. This is also in contrast to the study of Kelly et al. (2006), as the muscle split in their patients with persisting electromyographical irregularities were all performed by sharp dissection. According to our results, the nerves supplying the distal part of the vastus medialis muscle—especially the transverse coursing branches and the intramuscular anastomoses—are prone to be severed using the midvastus approach. However, further clinical studies are necessary to clarify whether such a possible denervation may lead to altered patellar tracking, which may result in long-term functional problems. In summary, the subvastus approach seems to offer the best possibility for a nerve-sparing way into the knee joint. Performing the medial parapatellar approach, no intramuscular nerve branches are harmed; only the sensory branches terminating within the ventromedial part of the capsule of the knee joint will likely be disrupted. Finally, using the midvastus approach, one cannot exclude the possibility of severing intramuscular nerve branches. As the approach performed depends on several factors, our results will provide additional valuable decision support. BP and MLP were responsible for the study design and conception. BP harvested the muscles, performed the Sihler’s technique, and wrote the first draft of the manuscript. BP and MLP evaluated the nerve patterns. AU simulated the surgical approaches and added valuable clinical information. All authors revised and approved the manuscript. The authors are indebted to all persons who voluntary donated their body for anatomic education and science. Without their altruism, studies like the present would be impossible. Special thanks are offered to Mag. Dr. Martin Schepelmann for critically reading the manuscript. Acta thanks Kaj Knutson and Kjell Nilsson for help with peer review of this study.

Calguner E, Erdogan D, Elmas C, Bahcelioglu M, Gozil R, Ayhan M S. Innervation of the rat anterior abdominal wall as shown by modified Sihler’s stain. Med Princ Pract 2006; 15(2): 98-101. doi: 10.1159/000090911. Clayton M L, Thirupathi R. Patellar complications after total condylar arthroplasty. Clin Orthop Rel Res 1982; 170: 152-5. Cooper R E Jr, Trinidad G, Buck W R. Midvastus approach in total knee arthroplasty: a description and a cadaveric study determining the distance of the popliteal artery from the patellar margin of the incision. J Arthroplasty 1999; 14(4): 505-8. doi: 10.1016/s0883-5403(99)90109-2. Dalury D F, Jiranek W A. A comparison of the midvastus and paramedian approaches for total knee arthroplasty. J Arthroplasty 1999; 14(1): 33-7. Dalury D F, Snow R G, Adams M J. Electromyographic evaluation of the midvastus approach. J Arthroplasty 2008; 23(1): 136-40. doi: 10.1016/j. arth.2007.01.020. Doty R L, Cummins D M, Shibanova A, Sanders I, Mu L. Lingual distribution of the human glossopharyngeal nerve. Acta Otolaryngol 2009; 129(1): 52-6. doi: 10.1080/00016480801998820. Drake W, 3rd, Li Y, Rothschild M A, Wu B L, Biller H F, Sanders I. A technique for displaying the entire nerve branching pattern of a whole muscle: results in 10 canine posterior cricoarytenoid muscles. Laryngoscope 1993; 103(2): 141-8. doi: 10.1002/lary.5541030204.

Acta Orthopaedica 2021; 92 (2): 194–198

Ehler E, Gärtner L, Schultz M, Schünke W. Zur Frage der Verteilung der Femoralisäste im M. vastus medialis mit besonderer Berücksichtigung der Kniegelenkäste. Anat Anz 1959; 107: 414-23. Erkes F. Weitere Erfahrungen mit physiologischer Schnittführung zur Eröffnung des Kniegelenks. Bruns’ Beitr Klin Chir 1929; 147: 221-32. Goh J C H, Lee P Y C, Bose K. A cadaver study of the function of the oblique part of vastus medialis. J Bone Joint Surg Br 1995; 77(2): 225-31. Halder A, Beier A, Neumann W. Mini-Subvastus-Zugang bei der Implantation von Knieendoprothesen. Oper Orthop Traumatol 2009; 21(1): 14-24. Hofmann A A, Plaster R L, Murdock L E. Subvastus (southern) approach for primary total knee arthroplasty. Clin Orthop Rel Res 1991; (269): 70-7. Horner G, Dellon A L. Innervation of the human knee joint and implications for surgery. Clin Orthop Rel Res 1994; (301): 221-6. Hubbard J K, Sampson H W, Elledge J R. Prevalence and morphology of the vastus medialis oblique muscle in human cadavers. Anat Rec 1997; 249: 135-42. Hube R, Keim M, Mayr H. Der Mini-Midvastus-Zugang zur Implantation von Kniegelenkendoprothesen. Oper Orthop Traumatol 2009; 21(1): 3-13. Jojima H, Whiteside L A, Ogata K. Anatomic consideration of nerve supply to the vastus medialis in knee surgery. Clin Orthop Rel Res 2004; (423): 157-60. Keblish P A. Alternate surgical approaches in mobile-bearing total knee arthroplasty. Orthopedics 2002; 25S(2): 257-64. Kelly M J, Rumi M N, Kothari M, Parentis M A, Bailey K J, Parrish W M, Pellegrini V D, Jr. Comparison of the vastus-splitting and median parapatellar approaches for primary total knee arthroplasty: a prospective, randomized study. J Bone Joint Surg Am 2006; 88(4): 715-20. doi: 10.2106/ jbjs.E.00107. Lieb F J, Perry J. Quadriceps function: an anatomical and mechanical study using amputated limbs. J Bone Joint Surg 1968; 50-A(8): 1535-49. Lin Y F, Lin J J, Jan M H, Wei T C, Shih H Y, Cheng C K. Role of the vastus medialis obliquus in repositioning the patella. Am J Sports Med 2008; 36(4): 741-6. Liu J, Kumar V P, Shen Y, Lau H-K, Pereira B P, Pho R W H. Modified Sihler’s technique for studying the distribution of intramuscular nerve branches in mammalian skeletal muscle. Anat Rec 1997; 247: 137-44. Mu L, Sanders I. Sihler’s whole mount nerve staining technique: a review. Biotech Histochem 2010; 85(1): 19-42. doi: https//doi. org/10.1080/10520290903048384. Nozic M, Mitchell J, de Klerk D. A comparison of the proximal and distal parts of the vastus medialis muscle. Aus J Physiother 1997; 43(4): 277-81. doi: 10.1016/s0004-9514(14)60416-5. Parentis M A, Rumi M N, Deol G S, Kothari M, Parrish W M, Pellegrini Jr D. A comparison of the vastus splitting and median parapatellar approaches in total knee arthroplasty. Clin Orthop Rel Res 1999; (367): 107-16. Peeler J, Cooper J, Porter M M, Thliveris J A, Anderson J E. Structural parameters of the vastus medialis muscle. Clin Anat 2005; 18: 281-9. Sekiya S, Suzuki R, Miyawaki M, Chiba S, Kumaki K. [Application of the modified Sihler’s stain technique to cadaveric peripheral nerves after medical students’ dissection course]. Kaibogaku Zasshi Jpn J Anat 2005; 80(3): 67-72. Smith T O, Nichols R, Harle D, Donell S T. Do the vastus medialis obliquus and vastus medialis longus really exist? A systematic review. Clin Anat 2009; 22: 183-99. Speakman H G B, Weisberg J. The vastus medialis controversy. Physiotherapy 1977; 63(8): 249-54. Thiranagama R. Nerve supply of the human vastus medialis muscle. J Anat 1990; 170: 193-8. Toumi H, Poumarat G, Benjamin M, Best T, F’Guyer S, Fairclough J. New insights into the function of the vastus medialis with clinical implications. Med Sci Sports Exerc 2007; 39(7): 1153-9. van Rensch P J H, Hannink G, Heesterbeek P J C, Wymenga A B, van Hellemondt G G. Long-term outcome following revision total knee arthroplasty is associated with indication for revision. J Arthroplasty 2020; 35(6): 16717. doi: https://doi.org/10.1016/j.arth.2020.01.053.

Acta Orthopaedica 2021; 92 (2): 199–203 199

National trends in lumbar spine decompression and fusion surgery in Finland, 1997–2018 Ville T PONKILAINEN 1, Tuomas T HUTTUNEN 2,4, Marko H NEVA 1, Liisa PEKKANEN 3, Jussi P REPO 3, and Ville M MATTILA 1,4,5 1 Department of Orthopaedics and Traumatology, Tampere University Hospital, Tampere; 2 Department of Anesthesiology, Tampere University Hospital, Tampere; 3 Department of Surgery, Central Finland Hospital District, Jyväskylä; 4 Faculty of Medicine and Health Technology, Tampere University, Tampere; 5 COXA Hospital for Joint Replacement, Tampere, Finland

Correspondence: ville.ponkilainen@tuni.fi Submitted 2020-07-12. Accepted 2020-10-08.

Background and purpose — During recent years, spine surgery techniques have advanced, the population has become older, and multiple high-quality randomized controlled trials that support surgical treatment for degenerative spinal stenosis and spondylolisthesis have been published. We assess the incidence and trends in spine fusion and decompression surgery in Finland between 1997 and 2018. Patients and methods — We used nationwide data from the Finnish nationwide National Hospital Discharge Register. The study population covered all patients aged 20 years or over in Finland (5.5 million inhabitants) during a 22-year period from 1997 through 2018. All patients who underwent spinal decompression were included. Patients with both decompression and fusion codes were analyzed as fusions. Results — 76,673 lumbar spine decompressions and fusions were performed during the study period. The incidence of lumbar spine decompressions increased from 33 (95% CI 23–45) per 100,000 person-years in 1997 to 77 (CI 61–95) per 100,000 person-years in 2018. The incidence of lumbar spine fusions increased from 9 (CI 5–17) per 100,000 person-years in 1997 to 30 (CI 21–43) per 100,000 personyears in 2018. The increase in incidence of lumbar spinal fusions was highest among women aged over 75 years, with a 4-fold increase. Interpretation — The incidence of lumbar spine fusions and decompressions increased between 1997 and 2018 in Finland. These findings may be the result of the emergence of advanced surgical techniques but may also be the result of an aging population and increased evidence supporting the surgical treatment of various spinal pathologies.

During recent years, there has been an increasing trend for the surgical treatment of degenerative spine pathologies (Gray et al. 2006). For example, more than 240 000 decompression procedures were performed in the United States in 2007. In 2004, the total cost of spine fusion and decompression surgeries in the United States was estimated to be 21 billion USD (Deyo 2007). Previous studies from the US have shown that the incidence of surgically treated degenerative spine pathologies slowly increased between 1980 and 1990 (Taylor et al. 1994). Thereafter, the rates rapidly increased from 1990 to 2010 (Deyo et al. 2005, Patil et al. 2005, Passias et al. 2017). According to a study by Seitsalo (1996), the incidence of lumbar spine fusion operations in Finland increased by 103% between 1987 and 1994. Conversely, they also reported that the incidence of lumbar spine decompression decreased by 12% during the same period. Recent trends in lumbar spine fusion and decompression surgery in Finland are not known, however. Spine surgery has undergone many evolutionary steps during the previous decades (Patil et al. 2005, Kim et al. 2011). Of these, new, less invasive techniques (Kim et al. 2011), better availability, higher quality of MRI (Patil et al. 2005), and a better understanding of degenerative spine pathologies (Patil et al. 2005, Passias et al. 2017) have all been presented as major factors behind the increasing incidence of spine surgery. Most importantly, multiple high-quality randomized controlled trials (RCT) supporting the operative treatment of degenerative spinal stenosis (Weinstein et al. 2008, 2010, Lurie et al. 2015) and spondylolisthesis (Weinstein et al. 2007, Ekman et al. 2009, Weinstein et al. 2009) have recently been published. By contrast, the effect of spinal fusion is debated (Försth et al. 2016).

200

Acta Orthopaedica 2021; 92 (2): 199–203

Taking into account the increasing trends in degenerative spine surgery and recent scientific evidence supporting surgery, we evaluated the changes in the incidence of spinal fusion and decompression surgery in Finland between 1997 and 2018.

Incidence (per 105 person-years) 100

Decompression Fusion

Patients and methods The data for this study was obtained from the Finnish National Hospital Discharge Register (NHDR). Patient characteristics, such as age, sex, diagnosis, and operations performed during the hospital stay, were obtained from the register. We included all patients aged 20 years or over. In Finland, it is mandatory for all public and private Finnish hospitals to collect data for the NHDR, and, as a result, the procedural coding has coverage and accuracy of more than 90% (Mattila et al. 2008, Sund 2012). However, the limitations of the NHDR are that it does not include the laterality of the surgery, comorbidities, or many clinically relevant risk factors, such as smoking or the use of alcohol (Sund 2012). The main outcome for this study was the number and incidence of spine surgeries in the NHDR performed with predefined procedures. We included hospitalizations with the following NOMESCO (Nordic Medico-Statistical Committee, Finnish version approved by the Finnish Institute for Health and Welfare) classification procedure codes: ABC36 (Decompression of lumbar nerve roots), ABC56 (Decompression of lumbar spinal canal and nerve roots), ABC66 (Decompression of lumbar spinal channel), NAG60 (Anterior fusion of lumbar spine with fixation), NAG61 (Posterior fusion of lumbar spine without fixation), NAG62 (Posterior fusion of lumbar spine with fixation, 2–3 vertebrae), NAG63 (Posterior fusion of lumbar spine with fixation, more than 3 vertebrae), NAG65 (Anterior and posterior fusion of lumbar spine), NAG66 (Posterior interbody fusion of lumbar spine, 2 vertebrae), and NAG67 (Posterior interbody fusion of lumbar spine, more than 2 vertebrae). The ABC codes listed above were defined as lumbar decompression surgery and the NAG60–67 codes were defined as lumbar fusion surgery. Patients with both decompression and fusion codes were analyzed as fusions. For those patients with the aforementioned lumbar spine surgeries, the International Classification of Diseases, Tenth Revision (ICD10) diagnosis codes were also evaluated, and patients with M00-M99 (Diseases of the musculoskeletal system and connective tissue) and G00-G99 (Diseases of the nervous system) codes were included. Patients with spine fracture (S1*, S2*, and S3*), sarcoma (C40–41 and C45–C49), and metastases (C76–79) were excluded. If a patient underwent multiple operations with the aforementioned diagnosis and operation codes during the followup, the first operation was considered to be the primary operation, and the time between the first and the second operations was calculated. All operations performed under the operation

0 1997

2000

2003

2006

2009

2012

2015

2018

Year

Figure 1. Incidence of lumbar spine decompression and fusion operations in Finland between 1997 and 2018.

and diagnosis codes described above were considered to be secondary operations. The patients were categorized into 3 age groups: between 20 and 54 years, between 55 and 74 years, and 75 years and older. Statistics The incidence rates were calculated using the annual mid-populations obtained from the national population register (Statistics Finland). The annual incidence rates (per 100,000 personyears) were based on the entire adult population of Finland and stratified according to sex and age. As the Finnish NHDR does not include death or other competing risks, we did not perform further survival analysis. 95% confidence intervals (CI) were calculated for incidence rates by using the Poisson exact method. Calculations were performed using R version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria). Ethics, funding, and potential conflicts of interest This study was performed as a retrospective register study, and therefore ethical approval was not required. The study received approval from the Finnish Institute for Health and Welfare Dnro: THL/1800/5.05.00/2019. This study has not received any external financial support. None of the authors have any conflicts of interests to declare.

Results 76,673 lumbar spine decompressions (69%) and fusions (31%) were performed in Finland between 1997 and 2018 according to the Finnish NHDR. The mean (SD) age of all patients was 63 (13) years. Females accounted for 55% (n = 42,454) of the operated patients. The annual population-based incidence of all lumbar decompressions and spine fusions increased by 155%, from 42 per 100,000 person-years in 1997 to 107 per 100,000 person-years in 2018 (Figure 1). From now on incidence refers to number per 100,000 person-years.

Acta Orthopaedica 2021; 92 (2): 199–203 201

Incidence (per 105 person-years) – fusion, age 20–54

Incidence (per 105 person-years) – fusion, age 55–75

Incidence (per 105 person-years) – fusion, age >75

240

Women Men

200

160

120

0 1997

2000

2003

2006

2009

2012

2015

2018

0 1997

2000

2003

2006

Year

2009

2012

2015

Women Men

0 1997

2018

2000

2003

2006

Year

2009

2012

2015

2018

Year

Incidence (per 105 person-years) – decompression, age 20–54

Incidence (per 105 person-years) – decompression, age 55–75

Incidence (per 105 person-years) – decompression, age >75

240

Women Men

200

160

120

0 1997

2000

2003

2006

2009

2012

2015

2018

0 1997

2000

2003

2006

Year

2009

2012

2015

Women Men

0 1997

2018

2000

2003

2006

Year

2009

2012

2015

2018

Year

Figure 2. Incidence of lumbar spine decompression and fusion operations by gender and age group in Finland between 1997 and 2018.

The incidence of lumbar spine decompressions increased by 133%, from 33 (CI 23–45) in 1997 to 77 (CI 61–95) in 2018 (Figure 1). The proportions of fusion and decompression operations remained similar throughout the study period, with decompressions accounting for 78% in 1997 and 72% in 2018. The mean (SD) age of the patients who underwent spinal decompression surgery was 64 (13) years, and the mean age for patients who underwent spinal fusion was 59 (14) years. The changes in the incidence of decompression surgery were similar among both males and females (Figure 2). The incidence of lumbar spine fusions increased by 289%, from 9 (CI 5–17) in 1997 to 35 (CI 24–48) in 2016 (Figure 1). After 2016, the incidence decreased and was 30 (CI 21–43) in 2018. The incidence of lumbar spine fusion surgery rose fastest among women aged over 75 years. In this group, incidence increased from 9 (CI 5–17) to 47 (CI 35–63) (Figure 2), an increase of 422%. The change was also notable among women aged between 55 and 74 years, with a 289% increase in incidence from 18 (CI 11–27) to 70 (CI 55–87) between 1997 and 2018. The incidence of lumbar spine fusion in women aged under 55 years remained low. However, incidence still increased by 140% in this group, from 10 (CI 5–17) in 1997 to 24 (CI 16–36) in 2018.

Cumulative reoperation rate (%) 25

Decompression Fusion

Years after index operation

Figure 3. Cumulative reoperation rate of lumbar spine decompression and fusion operations in Finland between 1997 and 2018.

The cumulative reoperation rate of decompression and spinal fusion was comparable for the first 10 years of followup (Figure 3). Thereafter, the reoperation rate of spinal fusion started to rise slightly faster than that of decompression, achieving a reoperation rate of 23%, whereas decompression achieved a reoperation rate of 21% after 22 years of follow-up.

202

Discussion Our main finding was that the nationwide population-based incidence of lumbar spine decompressions and fusions in Finland increased by 155% between 1997 and 2018. Lumbar spine fusion surgery increased by 289% until 2016, and then decreased slightly. Conversely, spinal decompression operations increased by 133% from 1997 to 2018. The most notable change in spinal fusion operations was among women over 75 years of age, as the incidence increased by 422% during our study period. The change was also notable among women aged between 55 and 74 years. The incidence in this group increased by 289%. This increasing trend in lumbar fusion surgery is in line with the findings of previous studies. An increase in lumbar spinal fusion surgery was reported in the United States in the 1990s, as the incidence doubled between 1979 and 1990 (Taylor et al. 1994). After 1990, the increase in lumbar spine fusions grew steeper, with a 220% increase in incidence between 1990 and 2001, reaching an incidence of 61 per 100,000 person-years (Deyo et al. 2005). The slope became even steeper after 1996 when the US Food and Drug Administration approved the use of intervertebral cages, a new type of spinal implant. In Finland, Seitsalo (1996) reported that the incidence of lumbar spine fusion operations increased by 103%, from 3.7 per 100,000 person-years in 1987 to 7.5 per 100,000 person-years in 1994. Moreover, Seitsalo also reported that the incidence of lumbar spine decompression (excluding lumbar discectomies) was 33 per 100,000 person-years in 1987 and 29 per 100,000 person-years in 1994, a decrease of 12%. During the late 1990s, the incidence of spinal fusion surgery in our data increased rapidly (289% vs. 103%), although the trend decreased slightly after 2016. The incidence of spinal decompression also began to rapidly increase during the same time period (133% vs. 12%). Patil et al. (2005) have suggested that advancements in diagnostics and the availability of MRI may be behind the increases in the incidence of cervical spinal operations in the United States (Patil et al. 2005). Due to increased knowledge, these pathologies have become more familiar to a larger group of physicians, and patients will therefore be more likely to be referred to a neurosurgeon or orthopedic spine surgeon. Additionally, the increase in both decompressions and fusions was most notable among patients aged 75 years and older. This change is probably partly explained by people living longer and the fact that older individuals in general are in better health and are more active than previous generations. The demands of older individuals might therefore be higher. Thus, there are more patients aged over 75 years who can benefit from surgery. Furthermore, the threshold for surgery may have also been lowered due to changes in surgical techniques and advancements in the perioperative care of the elderly. Along with the general advancements in medicine and sur-

Acta Orthopaedica 2021; 92 (2): 199–203

gery, more accessible resources in preoperative examinations, improvements in preoperative and postoperative care protocols, and enhancements in surgical practice, it is now much easier to offer operative treatment to older patients. As a result of the evolution of surgical practices, operation times have decreased markedly. Therefore, it is possible that the overall perioperative efficiency has increased the availability and thus the incidence of surgical treatment in Finland. Moreover, it is evident that there is more good quality scientific evidence supporting surgical treatment. Multiple RCTs were published between 2007 and 2015 that supported operative treatment for spinal stenosis (Malmivaara et al. 2007, Weinstein et al. 2008, 2010, Slätis et al. 2011, Lurie et al. 2015). 2 of the studies were conducted by a Finnish research group (Malmivaara et al. 2007, Slätis et al. 2011). These studies might have also affected the increasing trends in lumbar spine surgery. The incidence of spinal fusion surgery decreased after 2016, whereas the increasing trend in spinal decompression surgery continued. A possible explanation for this might be a Swedish RCT by Försth et al. that was published in 2016. The study compared decompression plus fusion surgery and decompression surgery alone in the treatment of spinal stenosis. Their most important finding was that decompression combined with fusion did not result in better pain relief or functional outcomes at 2 or 5 years when compared with decompression alone The strengths of their study were the high number of patients (n = 247) and long follow-up time (5 years) combined with a high-quality study setting. Thus, the conclusions of the study might have affected the decrease in spinal fusion surgery and the increase in spinal decompression surgery in Finland after 2016. Another explanation for the decrease in spinal fusion surgery is the low incidence of reoperations. In degenerative diseases, the degeneration of adjacent levels continues after the primary surgery and may result in adjacent level disease and reoperation. We analyzed only re-decompressions and re-fusions. Notably, the cumulative reoperation rates of spinal fusion and decompression were comparable for the first 10 years. After that, however, the reoperation rate of spinal fusion started to rise slightly faster than that of decompression, achieving a reoperation rate of 23%, whereas spinal decompression achieved a reoperation rate of 21% after 22 years of follow-up. These nationwide rates are notably lower than in the previous study by Irmola et al. (2018), partly due to our only including redecompressions and re-fusions. The incidence of spinal decompression combined with fusion was higher among women than men. It has previously been noted that spinal stenosis is slightly more common among women than men in northern countries (Lønne et al. 2019). One of the reasons for this difference is that women have lower density of the paraspinal muscles, which more often leads to low back pain and spinal degeneration (Bulcke et al. 1979, Kalichman et al. 2010) The density of the paraspinal muscles is also known to decrease with higher age

Acta Orthopaedica 2021; 92 (2): 199–203 203

(Kalichman et al. 2010). Both factors may have played a role in the observed increased incidence of operative treatment of symptomatic degenerative spinal pathologies among women (especially older women). A clear limitation of our study is that we were unable to identify whether a reoperation was performed on the same site in the spine. Additionally, due to the retrospective register nature of the study, we were unable to reliably assess the reason for reoperation. These limitations can also affect our reoperation rates because part of the operations categorized as reoperations might have been operations on another site. The strength of our study is the high-quality nationwide register data collected from the NHDR database. The database includes all operations performed in all private and public hospitals, and therefore the quality of the data has been shown to be excellent in terms of coverage and accuracy (Keskimaki 1991, Mattila et al. 2008, Sund 2012). In addition to the coverage, the strength of the data is that each individual patient can be followed by their personal ID numbers, and we were therefore able to observe all operations these patients underwent during the study period. Conclusions The annual population-based incidence of lumbar spine decompressions and fusions increased by 155% between 1997 and 2018. Moreover, lumbar spine fusion surgery increased rapidly until 2016, and then started to decrease slightly. Conversely, the number of spinal decompression operations increased continuously between 1997 and 2018. The most notable change in spinal fusion operations was among women aged 75 years and older, as the incidence increased by 422% during our study period. The reasons behind the changes may be the aging population, advances in diagnostics and surgical techniques, and increased evidence supporting the surgical treatment of various lumbar spine pathologies. All authors are responsible for conceptualizing the study. VP performed the statistical analyses and wrote the first draft of the manuscript. All authors critically reviewed the drafts and accepted the final version of the manuscript. Acta thanks Ivar Rossvoll and J L C van Susante for help with peer review of this study. Bulcke J, Termote J-L, Palmers Y, Crolla D. Computed tomography of the human skeletal muscular system. Neuroradiology 1979; 17(3): 127-36. Deyo R A. Back surgery—who needs it. N Engl J Med 2007; 356(22): 223943. Deyo R A, Gray D T, Kreuter W, Mirza S, Martin B I. United States trends in lumbar fusion surgery for degenerative conditions. Spine (Phila Pa 1976) 2005; 30(12): 1441-5. doi: 10.1097/01.brs.0000166503.37969.8a. Ekman P, Möller H, Shalabi A, Yu Y X, Hedlund R. A prospective randomised study on the long-term effect of lumbar fusion on adjacent disc degeneration. Eur Spine J 2009; 18(8): 1175-86.

Försth P, Ólafsson G, Carlsson T, Frost A, Borgström F, Fritzell P, Öhagen P, Michaëlsson K, Sandén B. A randomized, controlled trial of fusion surgery for lumbar spinal stenosis. N Engl J Med 2016; 374(15): 1413-23. Gray D T, Deyo R A, Kreuter W, Mirza S K, Heagerty P J, Comstock B A, Chan L. Population-based trends in volumes and rates of ambulatory lumbar spine surgery. Spine (Phila Pa 1976) 2006; 31(17): 1957-63. Irmola T M, Häkkinen A, Järvenpää S, Marttinen I, Vihtonen K, Neva M. Reoperation rates following instrumented lumbar spine fusion. Spine (Phila Pa 1976) 2018; 43(4): 295-301. Kalichman L, Hodges P, Li L, Guermazi A, Hunter D J. Changes in paraspinal muscles and their association with low back pain and spinal degeneration: CT study. Eur Spine J 2010; 19(7): 1136-44. Keskimaki I. Accuracy of data on diagnosis, procedures and accidents in the Finnish hospital register. Int J Health Sci 1991; 2:15-21. Kim C W, Siemionow K, Anderson D G, Phillips F M. The current state of minimally invasive spine surgery. Instr Course Lect 2011; 60: 353-70. Lurie J D, Tosteson T D, Tosteson A, Abdu W A, Zhao W, Morgan T S, Weinstein J N. Long-term outcomes of lumbar spinal stenosis: eight-year results of the Spine Patient Outcomes Research Trial (SPORT). Spine (Phila Pa 1976) 2015; 40(2): 63. Lønne G, Fritzell P, Hägg O, Nordvall D, Gerdhem P, Lagerbäck T, Andersen M, Eiskjaer S, Gehrchen M, Jacobs W. Lumbar spinal stenosis: comparison of surgical practice variation and clinical outcome in three national spine registries. Spine J 2019; 19(1): 41-9. Malmivaara A, Slätis P, Heliövaara M, Sainio P, Kinnunen H, Kankare J, Dalin-Hirvonen N, Seitsalo S, Herno A, Kortekangas P. Surgical or nonoperative treatment for lumbar spinal stenosis?: a randomized controlled trial. Spine (Phila Pa 1976) 2007; 32(1): 1-8. Mattila V M, Sillanpaa P, Iivonen T, Parkkari J, Kannus P, Pihlajamaki H. Coverage and accuracy of diagnosis of cruciate ligament injury in the Finnish National Hospital Discharge Register. Injury 2008; 39(12): 1373-6. Passias P G, Jalai C M, Worley N, Vira S, Marascalchi B, McClelland III S, Lafage V, Errico T J. Adult spinal deformity: national trends in the presentation, treatment, and perioperative outcomes from 2003 to 2010. Spine Deform 2017; 5(5): 342-50. Patil P G, Turner D A, Pietrobon R. National trends in surgical procedures for degenerative cervical spine disease: 1990–2000. Neurosurgery 2005; 57(4): 753-8. Seitsalo S. Back surgery rates in Finland 1987–1994: Increasing trends; regional variations. Fin J Orthop Traumatol 1996; 19:199-204. Slätis P, Malmivaara A, Heliövaara M, Sainio P, Herno A, Kankare J, Seitsalo S, Tallroth K, Turunen V, Knekt P. Long-term results of surgery for lumbar spinal stenosis: a randomised controlled trial. Eur Spine J 2011; 20(7): 1174-81. Sund R. Quality of the Finnish Hospital Discharge Register: a systematic review. Scand J Public Health 2012; 40(6): 505-15. doi: 10.1177/ 1403494812456637. Taylor V M, Deyo R A, Cherkin D C, Kreuter W. Low back pain hospitalization: recent United States trends and regional variations. Spine (Phila Pa 1976) 1994; 19(11): 1207-12; discussion 13. Weinstein J N, Lurie J D, Tosteson T D, Hanscom B, Tosteson A N, Blood E A, Birkmeyer N J, Hilibrand A S, Herkowitz H, Cammisa F P. Surgical versus nonsurgical treatment for lumbar degenerative spondylolisthesis. N Engl J Med 2007; 356(22): 2257-70. Weinstein J N, Tosteson T D, Lurie J D, Tosteson A N, Blood E, Hanscom B, Herkowitz H, Cammisa F, Albert T, Boden S D. Surgical versus nonsurgical therapy for lumbar spinal stenosis. N Engl J Med 2008; 358(8): 794-810. Weinstein J N, Lurie J D, Tosteson T D, Zhao W, Blood E A, Tosteson A N, Birkmeyer N, Herkowitz H, Longley M, Lenke L. Surgical compared with nonoperative treatment for lumbar degenerative spondylolisthesis: four-year results in the Spine Patient Outcomes Research Trial (SPORT) randomized and observational cohorts. J Bone Joint Surg Am 2009; 91(6): 1295. Weinstein J N, Tosteson T D, Lurie J D, Tosteson A, Blood E, Herkowitz H, Cammisa F, Albert T, Boden S D, Hilibrand A. Surgical versus non-operative treatment for lumbar spinal stenosis four-year results of the Spine Patient Outcomes Research Trial (SPORT). Spine (Phila Pa 1976) 2010; 35(14): 1329.

204

Acta Orthopaedica 2021; 92 (2): 204–207

Are conventional microbiological diagnostics sufficiently expedient in the era of rapid diagnostics? Evaluation of conventional microbiological diagnostics of orthopedic implant-associated infections (OIAI) Hege Vangstein AAMOT 1, J Christopher NOONE 1,2, Inge SKRÅMM 3 and Truls M LEEGAARD 1,4 1 Department

of Microbiology and Infection Control, Akershus University Hospital, Lørenskog, 2 Faculty of Medicine, University of Oslo, Oslo, 3 Orthopedic clinic, Akershus University Hospital, Lørenskog, 4 Institute of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway Correspondence: hege.vangstein.aamot@ahus.no Submitted 2020-05-26. Accepted 2020-10-16

Background and purpose — In a time when rapid diagnostics are increasingly sought, conventional procedures for detection of microbes causing orthopedic implant-associated infections (OIAI) seem extensive and time-consuming, but how extensive are they? We assessed time to (a) pathogen identification, (b) antibiotic susceptibility patterns, and (c) targeted antibiotic treatment using conventional microbiological diagnostics of OIAI in a consecutive series of patients. Patients and methods — Consecutive patients aged ≥18 years undergoing first revision surgery for acute OIAI, including prosthetic joints, fracture, and osteotomy implants, in 2017–2018 at Akershus University Hospital (Ahus), Norway were included. Information regarding microbiological diagnostics and clinical data was collected retrospectively from the hospital’s diagnostic and clinical databases. Results — 123 patients fulfilled the inclusion criteria. Median time to pathogen identification was 2.5 days and to antibiotic treatment recommendations was 3.5 days. The most common pathogens were S. aureus (52%) and S. epidermidis (15%). Cultures were inconclusive in 11% of the patients. Of the 109 patients with culture-positive results, antibiotic treatment was changed in 66 (61%) patients within a median of 4 days (0–24) after the recommendation was given. Interpretation — Conventional microbiological diagnostics of OIAI is time-consuming, taking days of culturing. Same-day diagnostics would vastly improve treatment efficacy, but is dependent on rapid implementation by clinicians of the treatment recommendations given by the microbiologist.

The majority of orthopedic procedures include the use of implants, which increase the risk of infection due to the reduced number of bacteria needed to establish an infection (Zimmerli et al. 1982). Orthopedic implant-associated infections (OIAI) are infrequent per se, with an overall surgical site infection rate following implant surgery of 3% (Skråmm et al. 2012). However, the number of patients undergoing orthopedic implant surgery is high and increasing (Norwegian National Advisory Unit on Arthroplasty and Hip Fractures 2020). A microbiological diagnosis is vital for providing the best treatment, with regards to both surgical options and providing targeted and narrow-spectrum antimicrobial therapy (Beam and Osmon 2018). Today’s conventional diagnostics include microbiological culturing of 5 biopsies from each infected patient on several different media for at least 5 days dependent on growing and dividing bacteria (Bergh et al. 2011, Osmon et al. 2013). More rapid diagnostic tools are being developed, but with varying degrees of sensitivity and specificity (Bonanzinga et al. 2017, Jun and Jianghua 2018, Aamot et al. 2019). We assessed time to (a) pathogen identification, (b) antibiotic susceptibility patterns, and (c) targeted antibiotic treatment using conventional microbiological diagnostics of OIAI in a consecutive series of patients.

Patients and methods This retrospective cohort study included all patients aged ≥ 18 years operated for acute OIAI (including prosthetic joint infections, fracture implants, and osteotomy implants) undergoing first revision surgery in 2017–2018 at Akershus University Hospital (Ahus), Norway. Ahus is Norway’s largest acute care hospital serving > 10% of the Norwegian population (5.4

Acta Orthopaedica 2021; 92 (2): 204–207 205

Table 1. Time to results after conventional microbiologic diagnostics of patients with orthopedic implant-associated infections (n = 123) Factor Hours to pathogen identification Hours to antibiotic recommendation Hours to final results Days from antibiotic recommendation to changed treatment (n = 66)

Median (range) 59 (16–238) 84 (36–238) 141 (59–298) 4 (0–24)

million inhabitants) and performs ~4,000 orthopedic implant surgeries annually. The microbiology laboratory is situated in the center of the hospital with short, indoor transportation of patient samples from the operating theatre. In the study period, laboratory opening hours were 07:30–16:30 on weekdays and 07:30–15:00 on weekends with an extension from November 2018 to 07:30–21:00 on weekdays. The criteria for an OIAI were based on the Modified Musculoskeletal Infection Society (MSIS) criteria described by Parvizi et al. (2011). Conventional culturing was performed by homogenizing up to 5 tissue samples individually with mortar and pistil in heart infusion broth (HIB) in a type 2 microbiological safety cabinet with subsequent seeding as previous described (Aamot et al. 2019). Incubation was terminated after 5 days following consensus (Parvizi et al. 2013) unless slow growing bacteria, such as Cutibacterium acnes, were suspected to be clinically relevant. The incubation period was prolonged to 14 days in such cases. Cultivation results have previously been published in 13 patients (Aamot et al. 2019). Microbe identification was performed by matrix-assisted laser desorption ionization time of flight (MALDI-TOF) using MALDI-TOF MS Biotyper (Bruker Daltonik GmbH, Bremen, Germany, MBT 6903 MSP Library, MBT Compass v4.1.70.1, Compass for flexControl v3.4). Antibiotic susceptibility testing was performed according to the guideline from the European Committee on Antimicrobial Susceptibility Testing EUCAST (EUCAST 2017a) and EUCAST breakpoints were utilized to categorize the isolate as sensitive (S), intermediate (I) (now susceptible, increased exposure), or resistant (R) (EUCAST 2017b). Information regarding microbiological diagnostics and clinical data was collected retrospectively from the hospital’s diagnostic and clinical databases. Time to results were defined as the time from biopsy to pathogen identification, the time to antibiotic treatment advice and the time to completed analyses including anaerobic cultivation. Confirmed infection, microbiologically, was defined as identification of the same microbe in 2 or more patient samples, whereas unconfirmed infection was defined as identifying a microbe in fewer than 2 patient samples (Parvizi et al. 2018). Empirical treatment was based on national guidelines (Norwegian Directorate of Health, 2020) and distributed intraoperatively after biopsy. For prosthetic joint infections (PJI), the

Table 2. Most common bacteria identified in orthopedic implantassociated infections Patients Identified bacteria n (%) Staphylococcus aureus Staphylococcus epidermidis Cutibacterium acnes Enterococcus faecalis Group B ß-hemolytic streptococci Staphylococcus capitis Staphylococcus lugdunensis Others a Culture negative/inconclusive a

64 (52) 18 (15) 9 (7) 11 (8) 8 (7) 4 (3) 5 (4) 23 (19) 14 (11)

Number of patients with monomicrobial/ polymicrobial infection 48/16 7/11 3/6 2/9 5/3 1/3 1/4 12/11

Pathogens found in 2 or fewer patients.

empirical treatment was vancomycin, ciproxin, and/or dicloxacillin. For other implant infections, the empirical treatment was penicillinase-resistant penicillin. 11 patients received non-empirical, targeted treatment prior to surgery due to previously diagnosed bloodstream infections or unrelated concurrent joint infections. Ethics, funding, and potential conflicts of interest. This study was approved by the Data Protection Officer (2018-105) at Akershus University Hospital. This study did not receive grants from public, commercial, or not-for-profit sectors. The authors report no conflict of interests.

Results Of the 123 patients included, 62 (50%) patients were female. The median age was 71 years (25–95). Time to microbiology results (Table 1) Pathogens were identified after a median of 59 hours (2.5 days) and antibiotic recommendations were available after a median of 84 hours (3.5 days). Culturing results were finalized within a median of 141 hours (6 days). Pathogens causing infections and culture-negative samples (Table 2) Confirmed infection, defined by positive cultivation results, was observed in 109/123 (89%) patients. The remaining 14/123 (11%) patients had inconclusive/negative cultivation, of whom 4 patients had received antibiotic treatment prior to revision surgery. Monomicrobial infections were most common, identified in 79/109 (72%) patients. S. aureus and S. epidermidis were the most frequent pathogens. None of the S. aureus isolates and 10/18 of the S. epidermidis isolates were resistant to methicillin. 8 of 76 patients undergoing surgery during the microbiology lab’s opening hours had culture-negative biopsies, whereas 6

206

of 47 patients undergoing surgery outside opening hours had culture-negative biopsies. Similar results were seen in patients’ biopsies requiring pre-cultivation in broth (6/76 versus 3/47). Change of treatment 111/123 (90%) patients were given empirical treatment. Of the remaining 12/123 (10%) patients, 11 patients received targeted treatment based on previous infections and 1 patient did not receive any antibiotic treatment prior to cultivation results as infection was considered unlikely. Of the 109 patients with culture-positive results, antibiotic treatment was changed to targeted and narrowed treatment in 66 (61%) patients within a median of 4 days (0–24) based on the antibiotic treatment recommendations.

Discussion Our study confirms that conventional microbial diagnostics of OIAIs is comprehensive and time-consuming with a median of 2.5 days to pathogen identification and a median of 3.5 days to antibiotic recommendation. In addition, we identified a delay of median 4 days from when antibiotic recommendations were given to clinicians to when treatment was changed. The lengthy time to results may be explained by a combination of several factors. The bacteria require time to multiply. In addition, 11% of the patients showed inconclusive or negative culturing, which involves 5 days of culturing before termination. Of the 109 patients with culture-positive results, 9 patients had positive samples only after pre-cultivation in broth. Pre-cultivation prolongs cultivation by 2 days. Lack of concurrence between the time of surgery and the opening hours of the microbiology lab may also prolong time to results. Biopsies taken after the lab’s opening hours were not cultivated until the following day in 47/123 patients. However, the concurrence between opening hours and time of surgery did not seem to affect the cultivation outcome. The frequency of inconclusive/negative results and positive results only after broth pre-cultivation did not differ among those patients with surgery performed before 16:00 compared with after 16:00. Staphylococci are the most frequently reported causes of orthopedic implant infections (Arciola et al. 2018), as was confirmed by our study. Of the 109 patients with culture-positive results, 66 received targeted treatment after receiving antibiotic resistance patterns. The majority of patients received better targeted antibiotic treatment, which may have led to more efficient treatment and reduced induction of antibiotic resistance. However, the response time from notification of antibiotic susceptibility results to the actual change of antibiotic treatment took a median of 4 days. This delayed response may negate the benefit of future rapid diagnostics. In our hospital, the microbio-

Acta Orthopaedica 2021; 92 (2): 204–207

logical results are sent electronically to the patient’s medical records immediately upon approval. Our continuous efforts to reduce the time to microbiological results will come to naught without clinicians reacting accordingly. Such optimization may also be relevant in diagnostics and treatment of other patient groups and types of infections. As a retrospective study, this work is limited by the information already registered in the patients’ journals at the time of care. Furthermore, this study was carried out in a country with a low prevalence of antibiotic resistance, so empirical treatment success may not be comparable to countries with a higher antibiotic resistance load. The study’s strengths lie in the number of patients included, as OIAI is infrequent, and the patients coming from an unselected patient population. As the majority of patients had culture-positive biopsies, more rapid diagnostics could improve time to targeted treatment and may potentially improve clinical outcome. Our study was not designed for patient-reported outcome measures (PROMS), but the obvious benefits would be faster diagnosis, and simpler, less resource-demanding care. An additional potential advantage is the reduction of antibiotic resistance development through more targeted and narrow-spectrum antibiotic treatment. This will require further investigation. In conclusion, in taking 2.5 days for pathogen identification and 3.5 days for targeted treatment advice, conventional microbiological diagnostics of OIAI are not sufficiently expedient. Same-day diagnostics may contribute to rapid targeted treatment and more favorable clinical outcomes, but the delayed response from clinicians on the treatment recommendations needs to be addressed. The authors would like to thank Tone Elin Langdal for help with extracting microbiological data. Preliminary results were presented at the European Conference for Clinical Microbiology and Infectious Diseases (ECCMID) in Amsterdam, the Netherlands 2019 (Poster #2515). Concept and design; HVA, IS, TML. Data extraction; HVA, IS. Data interpretation; HVA, JCN, IS, TML. Writing and revising manuscript; HVA, JCN, IS, TML. Acta thanks Jon Goosen and Thord von Schewelovfor help with peer review of this study.

Aamot H V, Johnsen B O, Skramm I. Rapid diagnostics of orthopedic implantassociated infections using Unyvero ITI implant and tissue infection application is not optimal for Staphylococcus species identification. BMC Res Notes 2019; 12(1): 725. doi: 10.1186/s13104-019-4755-5. Arciola C R, Campoccia D, Montanaro L. Implant infections: adhesion, biofilm formation and immune evasion. Nature Rev Microbiol 2018; 16(7): 397-409. doi: 10.1038/s41579-018-0019-y. Beam E, Osmon D. Prosthetic joint infection update. Infect Dis Clin North Am 2018; 32(4): 843-59. doi: 10.1016/j.idc.2018.06.005.

Acta Orthopaedica 2021; 92 (2): 204–207 207

Bergh K L, Leegaard T M, Steinbakk, M. Diagnostikk ved fremmedlegemerelaterte infeksjoner [Diagnostics of implant-associated infections]. In: Strategimøte nr 25; 2011. Ed Sandven P; 2011. Bonanzinga T, Zahar A, Dutsch M, Lausmann C, Kendoff D, Gehrke T. How reliable is the alpha-defensin immunoassay test for diagnosing periprosthetic joint infection? A prospective study. Clin Orthop Relat Res 2017; 475(2): 408-15. doi: 10.1007/s11999-016-4906-0. EUCAST ESoCMaID. Antimicrobial susceptibility tesitng EUCAST disk diffusion method, Version 6.0; 2017a. EUCAST. Breakpoint tables for interpretation of MICs and zone diameters, Version 7.1, 2017. http://www.eucast.org; 2017 b. Jun Y, Jianghua L. Diagnosis of periprosthetic joint infection using polymerase chain reaction: an updated systematic review and meta-analysis. Surg Infections 2018; 19(6): 555-65. doi: 10.1089/sur.2018.014. Norwegian Directorate of Health. Antibiotics in hospital. https://www.helsedirektoratet.no/retningslinjer/antibiotika-i-sykehus; 2020. Norwegian National Advisory Unit on Arthroplasty and Hip Fracture, Report. Helse Bergen HF, Ortopedisk klinikk, Haukeland universitetssjukehus; 2020. ISBN 978-82-91847-25-2; June 2020. Osmon D R, Berbari E F, Berendt A R, Lew D, Zimmerli W, Steckelberg J M, Rao N, Hanssen A, Wilson W R, Infectious Diseases Society of America.

Diagnosis and management of prosthetic joint infection: clinical practice guidelines by the Infectious Diseases Society of America. Clin Infect Dis 2013; 56(1): e1-e25. doi: 10.1093/cid/cis803. Parvizi J, Zmistowski B, Berbari E F, Bauer TW, Springer B D, Della Valle C J, Garvin K L, Mont M A, Wongworawat M D, Zalavras C G. New definition for periprosthetic joint infection: from the Workgroup of the Musculoskeletal Infection Society. Clin Orthop Relat Res 2011; 469(11): 2992-4. doi: 10.1007/s11999-011-2102-9. Parvizi J, Gehrke T, Chen A F. Proceedings of the International Consensus on Periprosthetic Joint Infection. Bone Joint J 2013; 95-b(11): 1450-2. doi: 10.1302/0301-620x.95b11.33135. Parvizi J, Tan T L, Goswami K, Higuera C, Della Valle C, Chen A F, Shohat N. The 2018 definition of periprosthetic hip and knee infection: an evidence-based and validated criteria. J Arthroplasty 2018; 33(5): 1309-14.e2. doi: 10.1016/j.arth.2018.02.078. Skråmm I, Saltyte Benth J, Bukholm G. Decreasing time trend in SSI incidence for orthopaedic procedures: surveillance matters! J Hosp Infect 2012; 82(4): 243-7. doi: 10.1016/j.jhin.2012.08.011. Zimmerli W, Waldvogel F A, Vaudaux P, Nydegger U E. Pathogenesis of foreign body infection: description and characteristics of an animal model. J Infect Dis 1982; 146(4): 487-97.

208

Acta Orthopaedica 2021; 92 (2): 208–214

Increasing but levelling out risk of revision due to infection after total hip arthroplasty: a study on 108,854 primary THAs in the Norwegian Arthroplasty Register from 2005 to 2019 Håvard DALE 1,2, Pål HØVDING 1, Sindre M TVEIT 1, Julie B GRAFF 1, Olav LUTRO 3, Johannes C SCHRAMA 1, Tina S WIK 4, Inge SKRÅMM 5, Marianne WESTBERG 6, Anne Marie FENSTAD 1, Geir HALLAN 1,2, Lars B ENGESÆTER 1,2, and Ove FURNES 1,2 1 The

Norwegian Arthroplasty Register, Department of Orthopaedic Surgery, Haukeland University Hospital, Bergen; 2 Department of Clinical Medicine, University of Bergen, Bergen; 3 Department of Medicine, Stavanger University Hospital, Stavanger; 4 Department of Orthopaedic Surgery, St Olav University Hospital, Trondheim; 5 Department of Orthopaedic Surgery, Akershus University Hospital, Lørenskog; 6 Division of Orthopaedic Surgery, Oslo University Hospital, Oslo, Norway Correspondence: havard.dale@helse-bergen.no Submitted 2020-06-25. Accepted 2020-10-26.

Background and purpose — Focus on prevention, surveillance, and treatment of infection after total hip arthroplasty (THA) in the last decade has resulted in new knowledge and guidelines. Previous publications have suggested an increased incidence of surgical revisions due to infection after THA. We assessed whether there have been changes in the risk of revision due to deep infection after primary THAs reported to the Norwegian Arthroplasty Register (NAR) over the period 2005–2019. Patients and methods — Primary THAs reported to the NAR from January 1, 2005 to December 31, 2019 were included. Adjusted Cox regression analyses with the first revision due to deep infection after primary THA were performed. We investigated changes in the risk of revision as a function of time of primary THA. Time was stratified into 5-year periods. We studied the whole population of THAs, and the subgroups: all-cemented, all-uncemented, reverse hybrid (cemented cup), and hybrid THAs (cemented stem). In addition, we investigated factors that were associated with the risk of revision, and changes in the time span from primary THA to revision. Results — Of the 108,854 primary THAs that met the inclusion criteria, 1,365 (1.3%) were revised due to deep infection. The risk of revision due to infection, at any time after primary surgery, increased through the period studied. Compared with THAs implanted in 2005–2009, the relative risk of revision due to infection was 1.4 (95% CI 1.2–1.7) for 2010–2014, and 1.6 (1.1–1.9) for 2015–2019. We found an increased risk for all types of implant fixation. Compared to 2005–2009, for all THAs, the risk of revision due to infection 0–30 days postoperatively was 2.2 (1.8–2.8) for 2010– 2014 and 2.3 (1.8–2.9) for 2015–2019, 31–90 days postoperatively 1.0 (0.7–1.6) for 2010–2014 and 1.6 (1.0–2.5) for

2015–2019, and finally 91 days–1 year postoperatively 1.1 (0.7–1.8) for 2010–2014 and 1.6 (1.0–2.6) for 2015–2019. From 1 to 5 years postoperatively, the risk of revision due to infection was similar to 2005–2009 for both the subsequent time periods Interpretation — The risk of revision due to deep infection after THA increased throughout the period 2005–2019, but appears to have levelled out after 2010. The increase was mainly due to an increased risk of early revisions, and may partly have been caused by a change of practice rather than a change in the incidence of infection.

“Postoperative infection is the saddest of all complications…” John Charnley postulated in 1982 (Waugh and Charnley 1990). Despite advances in knowledge and awareness of prophylactic perioperative routines, there are indications that the incidence of infections after total hip arthroplasty (THA) is still increasing (Dale et al. 2012, Parvizi et al. 2013, Lenguerrand et al. 2017, Parvizi et al. 2017, Brochin et al. 2018, Kurtz et al. 2018). To disclose changes in the risk of infection we need a large number of primary THAs, registered in a uniform manner. The Norwegian Arthroplasty Register (NAR) found an increasing risk of deep infection after primary THA during the years 1987–2007. Over 10 years ago, Kurtz et al. (2007) projected a substantial demand for revisions due to infection in the coming decades. We have now assessed changes in the risk of surgical revision due to deep infection for THAs reported to the NAR during the years 2005 to 2019, as a follow-up of our previous study (Dale et al. 2009). In addition, we investigated factors that could be associated with revision, and the time span between primary and revision surgery.

Acta Orthopaedica 2021; 92 (2): 208–214 209

Patients and methods Since its inception in 1987, the NAR has registered detailed data on primary THAs and THA revisions. The data gathered includes information on patient characteristics, indication for THA, and surgery-related factors such as approach, type of implant, method of fixation, and duration of surgery. The unique identification number of each inhabitant of Norway is used to link the primary THA to any subsequent revision (Havelin et al. 2000). The data is validated, with 97% completeness of reporting of primary THAs, 93% reporting of revisions, 100% coverage of Norwegian hospitals, and 100% reporting of deaths (Furnes et al. 2019). Revision due to infection of the implant is defined as the removal or exchange of the whole or parts of the prosthesis, with deep infection reported as the reason for surgery. Isolated soft tissue debridement without the exchange of implant parts was not reported to the register until 2011, and was therefore not included. The surgeon completes the register form immediately after surgery, and the indication for the revision, infection or other, is based on perioperative assessment and evaluation. The diagnosis is not to be corrected in the NAR according to findings in peroperative bacterial samples. Due to this lack of validation, rate of revision due to infection will be only an approximation of the rate of true periprosthetic joint infection (PJI), as defined by Parvizi et al. (2018). The period of inclusion and observation in this study was January 1, 2005 to December 31, 2019. In this period, the NAR contained data on 116,779 primary THAs. 7,925 (7%) THAs were excluded due to missing information on covariates. 108,854 primary THAs had complete information and were eligible for analyses. All THAs were followed until their first revision due to infection or revision for other causes after the primary operation, death, or emigration, or until December 31, 2019. Thus, follow-up was 0–15 years. 3 time periods according to the year of primary THA were compared: 2005–2009, 2010–2014, and 2015–2019, with sub-analyses for the different THA fixation methods (all-cemented, all-uncemented, reverse hybrid [cemented cup], and hybrid [cemented stem]). Statistics Survival analyses were performed with Cox regression models, with year of primary THA as the main risk factor and date of revision due to deep infection as the endpoint. Revision hazard rate ratios (HRR) for the 2 later time periods relative to the 1st time period were calculated and presented as an expression of relative risk, with 95% confidence intervals (CI). We adjusted for the following: age (< 45, 45–54, 55–64, 65–74, 75–84, ≥ 85 years), sex, ASA class, indication for the primary THA (osteoarthritis, inflammatory disease, acute hip fracture, complications after hip fracture, complications after childhood hip disease, avascular necrosis of the femoral

head, other), surgical approach (anterior, anterolateral, lateral, posterolateral), duration of surgery (< 70, 70–99, 100–129, ≥ 130 minutes), and fixation (cemented, uncemented, reverse hybrid, or hybrid). Revisions due to infection in the case of monobloc THAs were not recorded if no implant parts were exchanged. We therefore adjusted for modularity of the prosthesis in the Cox analyses. In addition, we performed analyses where monobloc THAs (n = 3,936) were excluded. Monobloc implants were predominantly used early in the study period. More than 99.8% of the THA patients received perioperative antibiotic prophylaxis systemically, and for all cemented components antibiotic-loaded bone cement was used. We used Cox regression analyses, with time period as the stratification factor, to construct cumulative revision curves (1 minus cumulative survival) at mean values of the covariates, and to assess 5-year revision percentages. Analyses with follow-up restricted to 0–5 years for each period were performed, to assess for the effect of differences in follow-up. In addition, analyses were performed without THAs with metal-on-metal articulations (n = 300), since metal debris reactions may mimic infection. Further, we performed subanalyses on THAs due to osteoarthritis only (n = 83,770), as a more homogenous subgroup. We also performed separate Cox analyses on revision due to aseptic loosening as endpoint for all THAs, to be able to compare these with our findings of revision due to infection. Revision HRR due to infection as a function of year of the primary THA was studied, to give a graphical display of the relationship based on a generalized additive model for survival data (Hastie and Tibshirani 1990). These curves are presented with 95% CIs. HRRs were calculated for the different potential risk factors for the whole 15-year period adjusted for year of primary THA, to adjust for time-dependent confounding. The analyses were performed in accordance with the guidelines for statistical analyses of arthroplasty register data (Ranstam et al. 2011a and b). The proportional hazard assumptions of the Cox survival analyses were not completely fulfilled between the 3 time periods when tested by smoothed Schoenfeld residuals (see Figure 3). We therefore assessed the risk of revision due to infection 0–30 days, 31–90 days, 91 days–1 year, and 1–5 years postoperatively. A study from the Swedish Knee Arthroplasty Register has found that potential overestimation of incidence of revision through the effect of competing risks (death and revision) is negligible, and that Cox analyses are better than competing risk analyses in estimating revision risks in arthroplasty register data (Ranstam and Robertsson 2017). Based on this we chose to include results only from Cox analyses. Bilateral THAs are dependent observations, but the influence of bilaterality on the outcome has been found to be negligible, also in the case of infection (Lie et al. 2004, Ranstam et al. 2011b, Dale et al. 2012). Hence, patients with bilateral THAs were included, and considered independent.

210

Acta Orthopaedica 2021; 92 (2): 208–214

Table 1. Absolute annual number of primary THAs and revisions due to infection for the period 2005–2019

Hazard risk ratio

3.0

Year of primary THA

Number of THAs revised due to infection Number 0–30 31–90 91 days 1–5 of primary days days –1 year years Total (%) THAs

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019

6 10 23 30 34 53 71 54 58 47 60 65 64 72 n.a.

4 6 9 10 12 11 8 10 11 14 17 18 15 16 n.a.

8 6 4 10 13 10 7 13 11 11 14 17 20 12 n.a.

15 46 (0.9) 23 55 (1.2) 22 74 (1.2) 17 78 (1.4) 17 95 (1.5) 18 98 (1.6) 17 111 (1.3) 16 98 (1.3) 14 96 (1.3) 11 84 (1.1) n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

5,849 5,816 6,099 6,327 6,577 6,722 6,783 7,272 7,228 7,333 7,705 8,247 8,539 9,053 9,304

n.a. = not applicable due to incomplete follow-up

We calculated 95% CIs for survival rates and HRRs, and considered non-overlapping 95% CIs statistically significant. We used IBM SPSS 26.0 (IBM Corp, Armonk, NY, USA) and R statistical software packages for analyses (R Foundation for Statistical Computing, Vienna, Austria), and the study was performed in accordance with the STROBE and RECORD statements. Ethics, data sharing plan, funding, and potential conflicts of interests The registration of data and the study was performed confidentially on patient consent and according to Norwegian and EU data protection rules. Data may be accessible upon application to the NAR. The study was fully financed by the NAR, and no conflict of interest is declared.

Results 108,854 primary THAs in 91,621 patients met the inclusion criteria. 1,365 (1.3%) first revisions due to deep infection after primary THA were reported. Median follow-up was 5.4 (inter quartile range [IQR] 6.7) years, median age was 70 (IQR 14) years whereas mean ASA class for the THA patients was 2.0. Time trend of revision due to deep infection The annual number of revisions due to infection is presented in Table 1, whereas the annual increase in risk of revision due to infection presented in Figure 1. We found that the risk of revision due infection was higher for the periods 2010–2015 and 2015–2019 compared with 2005–2009. This finding was valid for all fixation methods. We found no difference between the 2 most recent periods, except for the cohort of uncemented

2.5

2.0

1.5

1.0

0.5

0 2005

2010

2015

2019

Year of primary THA

Figure 1. Relationship between year of primary surgery and risk of revision due to deep infection (with 95% confidence interval) for all THAs, adjusted for sex, age, ASA class, indication for primary THA, duration of surgery, surgical approach, and modularity of the THA. The broken line represents the HRR in 2005 (HRR = 1).

THAs, in which there was an increased risk for revision due to infection in 2015–2019 (CI 1.4 [1.1–1.7]), compared with 2010–2014, as well (Figure 2 and Table 2, see Supplementary data). The increase in risk of revision due to infection was most pronounced in the first 30 postoperative days, but for 2015– 2019 we found an increased risk for the whole first postoperative year (Figure 3 and Table 3, see Supplementary data). Excluding THAs with metal-on-metal articulation (n = 300), THAs due to other causes than osteoarthritis, and monobloc THAs did not alter our findings. Restricting follow-up for each period to 0–5 years also showed similar results. Factors associated with risk of revision due to infection The distribution of risk factors is presented in Table 4 (see Supplementary data). Patient-related factors such as age, sex, and indication for primary THA was stable throughout the period studied. There was more comorbidity in patients undergoing primary THA, from mean ASA class 1.8 (SD 0.7) in 2005 to 2.1 (SD 0.6) in 2019. The duration of surgery decreased slightly, the use of monobloc stems was terminated, and there was a shift towards uncemented fixation. Further, the use of posterolateral, anterolateral, and anterior surgical approaches increased at the expense of the direct lateral approach, which used to be the most common in our country. We assessed the impact of the different risk factors adjusted for in the Cox analyses, and the findings are presented in Table 5 (see Supplementary data). Male sex, advanced age (> 75 years), and comorbidity (ASA class > 1) were patient-related factors associated with increased risk of revision due to infection. THA due to complications after hip fracture surgery and due to avascular necrosis of the femoral head was associated with an increased risk of revision due to infection, whereas THA due to

Acta Orthopaedica 2021; 92 (2): 208–214 211

Percent revised – all THAs

Percent revised – cemented THAs

Percent revised – uncemented THAs

2.5

2.0

1.5

1.0

0.5

0.5 2005–2009 2010–2014 2015–2019

2005–2009 2010–2014 2015–2019

0 0

Figure 2. Curves of adjusted revisionpercentage due to deep infection, for all THAs, for cemented THAs, uncemented THAs, reverse hybrid THAs, and hybrid THAs, for 3 periods of primary surgery, adjusted for sex, age, ASA class, indication for primary THA, duration of surgery, surgical approach, and modularity of the THA.

Log HRR

–2

Percent revised – hybrid THAs 2.5

2.0

1.5

1.0

0.5

Year postoperatively

2005–2009 2010–2014 2015–2019

0.5

0 0

2005–2009 2015–2019 0

2005–2009 2010–2014 2015–2019

–2

2005–2009 2010–2014

Year postoperatively

Percent revised – reverse hybrid THAs

Log HRR 6

2.5

Year postoperatively

Figure 3. The relationship between HRR of revision due to infection and time span postoperatively after primary THAs for the period 2010–2014 (red line) and 2015–2019 (green line) compared with 2005–2009 (blue lines). Smoothed Schoenfeld residuals adjusted for sex, age, ASA class, indication for primary THA, duration of surgery, surgical approach, and modularity of the THA (solid lines) with 95% confidence intervals (broken lines).

complications after childhood hip disease was associated with a lower risk. Long duration of surgery (> 100 minutes), and anterolateral and lateral surgical approaches were associated with a slightly higher risk of revision due to infection. Uncemented and hybrid THAs had higher risk of revision due to infection than cemented THAs, whereas reverse hybrid

Year postoperatively

THAs had lower risk. Compared with modular THAs, monobloc THAs apparently had half the risk of revision due to infection, as expected, since debridement of these were not reported to the NAR as a revision, as explained earlier. Trends that may have contributed to the increased risk of revision for infection were a higher number of patients with substantial comorbidity, and more use of uncemented and hybrid THA (Tables 4 and 5, see Supplementary data). Trends that may have contributed to less increase in risk of revision for infection were less use of the lateral surgical approach and shorter duration of surgery (Tables 4 and 5, see Supplementary data). Time trend of revision due to aseptic loosening The risk of revision due to aseptic loosening, compared with 2005–2009, was decreasing, with 0.6 (CI 0.5–0.7) for 2010–2014, and 0.8 (CI 0.7–1.0) for 2015–2019 respectively.

Discussion Our main finding was an increased risk for revision due to deep infection after primary THA for the 2 consecutive 5-year periods after 2005–2009. The most pronounced increase was

212

for THAs performed before 2010, whereafter the risk of revision due to infection seems to have levelled out. Our findings confirm the trend from earlier studies from the NAR and the Nordic Arthroplasty Register Association (NARA) on the risk of revision due to infection (Dale et al. 2009, 2012). There are also other studies reporting an increased risk of PJI (Lenguerrand et al. 2017, Brochin et al. 2018, Kurtz et al. 2018). The finding that the increase in PJI is flattening out after 2010 is in concordance with findings from New York State (Perfetti et al. 2017). On the other hand, several infection surveillance registries report a trend of decreasing rates of surgical site infection (SSI) after THA, including both superficial and deep infections (Manniën et al. 2008, Choi et al. 2016, Sodhi et al. 2019). This was also the finding of the Norwegian Institute of Public Health’s SSI surveillance registry (Berg et al. 2019). The European Centre for Disease Prevention and Control (ECDC) reports a stable in-hospital incidence of SSI after total hip arthroplasty since 2011 (ECDC 2019), similar to what we found for that period. A review from 2015 reports increasing risk of SSI in several countries (Lamagni et al. 2015). This variety in trends might be explained by the differences in definitions and duration of observation between SSIs reported in regional or national surveillance systems, and revision due to infection, as reported to the arthroplasty registers. SSI is observed at discharge from hospital or at post-discharge surveillance by a general physician (30 days, 90 days, or 1 year postoperatively), in concordance with a specific set of diagnostic criteria and strict definition, and may be superficial or deep (ECDC 2017). In Norway, we have only 30 days’ surveillance of SSI after THA, but with good completeness since 2013 (Berg et al. 2019). In the NAR, however, the surgeon reports revision due to infection at any time after the primary THA. In our material, we found stable risk of revision due to infection in the first 30 days postoperatively for the period 2010–2019 (Table 3), in contrast to the slightly reduced incidence of SSIs from the Norwegian SSI surveillance program in 2013–2018 (Berg et al. 2019). Even if we do not have absolute numbers for direct comparison, and the definitions of infection are different, this may reflect a trend towards more revisions being performed due to superficial SSIs and prolonged wound drainage, which would, in that case, result in an apparent increase in risk of infection in our material. Surgeons may report to the NAR debridement for prolonged wound drainage as revision due to infection. This revision strategy has evolved because prolonged wound drainage and superficial SSIs are strongly associated with PJI (Zhu et al. 2015). Some of these cases will have negative cultures and will not fulfill criteria for a PJI. However, they remain registered as infections in the NAR. This may explain the discrepancies between the 2 registers, at least to some degree. For 2015–2019, we found an increased risk of revision due to infection in the first postoperative year. In contrast to our earlier findings, there was also an increased risk of revision

Acta Orthopaedica 2021; 92 (2): 208–214

later than 90 days postoperatively, which indicates a prolonged increase in risk of revision due to infection in recent years (Dale et al. 2012). The finding that the risk of revision increased more for the uncemented THAs compared with the other fixation methods may be explained by more frequent use of uncemented THAs in older patients with more comorbidity in recent years. The increased use of uncemented THA was most pronounced in 2005–2009, and as uncemented THA was found to be a risk factor associated with revision due to infection, this may partly explain our finding of increased risk of revision due to infection during this period. All patients with cemented components in the present study had antibiotic-loaded bone cement. This has been found to be beneficial as prophylaxis against postoperative infection (Parvizi et al. 2008, Dale et al. 2012, Wang et al. 2013). Hybrid THAs, although few, had higher risk of revision due to infection compared with cemented THAs. This is in concordance with our previous and others’ findings (Pedersen et al. 2010b, Dale et al. 2012, Leong et al. 2020). Compared with our previous study, we now had the benefit of being able to adjust for ASA class. However, this did not change the finding of increased risk of revision due to infection. Even if risk factors for revision due to infection may differ from risk factors for PJI, one possible explanation for the increased risk of infection is that THA is now being performed in frailer patients. We found an increase in mean ASA class for THA patients, and that ASA 2 and higher was associated with revision due to infection. The change in ASA class for THA patients was most pronounced between 2005–2009 and 2010–2014. Some comorbidities, such as obesity and diabetes with hyperglycemia, are found to be potent risk factors for postoperative infection and have an increasing incidence in the population (Pedersen et al. 2010a, Jämsen et al. 2012, Lamagni et al. 2015, Liu et al. 2015). These specific risk factors are not reported to the NAR, and might, if in our material as in the general population the prevalence increases, and if not fully captured by ASA class, contribute to an increased risk of infection. In the present study primary THA for an acute hip fracture was more used in 2015–2019 than in 2005–2009. Acute hip fracture was not found to be a risk factor for revision due to infection. Displaced femoral neck fractures are recommended to be treated with hemiarthroplasty in old and frail patients with limited life expectancy (Gjertsen 2019). However, in patients expected to live longer, THA may the recommended treatment in certain cases. This selection of THA for healthier patients may explain why acute hip fracture was not found to be a risk factor for revision due to infection, as may have been expected. There have been improvements in diagnostic procedures of PJI, and more standardized sampling, culturing, and analyzing techniques lead to fewer samples being false negative (Parvizi et al. 2016, Ting and Della Valle 2017). In addition, bacteria like Staphylococcus epidermidis and Cutibacterium acnes have emerged as important agents of implant infection (Zeller

Acta Orthopaedica 2021; 92 (2): 208–214 213

et al. 2018, Flurin et al. 2019). This may have resulted in more low-grade infections being correctly diagnosed, revised, and reported, where such infections earlier may have been overlooked or misdiagnosed as aseptic loosening. We found a trend towards more use of surgical approaches associated with lower risk of revision due to infection, and shorter duration of surgery. This should be considered beneficial and contribute to lower risk of revision due to infection as found by previous publications (Dale et al. 2009, Amlie et al. 2014, Mjaaland et al. 2017, Miller et al. 2018). Registers can provide a useful source of information on incidences and trends because of large numbers and long duration of observation. The NAR contain good quality, detailed information on THA patients and primary and revision surgery, gathered uniformly over a long period. Our data are prospective and with acceptable completeness on possible risk factors for primary and revision THA (Furnes et al. 2019). We therefore have an excellent base for a trend study on a relatively rare complication like PJI. Since a large number of THAs from a nationwide population were included, external validity is expected to be good. Register studies do have, however, inherent limitations (Varnum et al. 2019). Even if we adjusted for several important factors that could be associated with revision due to infection, there will be residual confounding. Such factors may be changes over time in reporting, revision threshold, diagnostics, surgeon awareness, selection of patients, and the virulence and resistance of bacteria causing infection. These factors may only partly be elucidated in a national arthroplasty register. Reported THA revision rates due to infection are not necessarily the same as rates of PJI, although we have reason to believe that this is a close approximation, since guidelines recommend revision in the case of suspected infection, and reporting of revisions to the NAR is good. Improved reporting of revision due to infection may explain our findings to some degree. However, compared with validation studies from Sweden and Denmark, and considering the 93% reporting of revisions to the NAR, our reporting of revision due to infection resembles the “true” incidence of PJI reported from similar registers (Lindgren et al. 2014, Gundtoft et al. 2015, Furnes et al. 2019). Yet another limitation is that an erroneous diagnosis of infection is not corrected in the NAR, when results from peroperative bacterial sampling presents. This may lead to both under-reporting of low-grade infections and over-reporting in the case of negative bacterial cultures. Misreporting will only influence our findings if it changes during the period studied. Based on the findings of a slight decrease in risk of revision due to aseptic loosening and the fact that there may have been improvements in diagnostics and surgeons’ awareness, there may have been a decrease in misreporting. Focus on the importance of thorough reporting has probably improved the reporting of revisions due to infection over the period studied. However, a time trend evaluation of this has not been performed.

In summary, our finding of increased risk of revision is probably multifactorial. Partly it may reflect changes over time in reporting, revision threshold, diagnostics, surgeon awareness, surgical volume and skills, and the virulence and resistance of bacteria causing infection. An increase in the risk of revision due to infection in such cases will not reflect an increase in the incidence of PJI. On the other hand, if THA is performed on patients with more comorbidity, higher age, or implants or techniques associated with a higher risk of infection are used, this would, as found in our study, contribute to an increased risk of infection after THA. To conclude, the risk of revision due to infection after THA increased by approximately 50% through the period 2005– 2019. However, the increase in risk of revision due to infection appears to have levelled out after 2010. The increase was mainly caused by more revisions during the first postoperative year. Uncemented THAs were increasingly used during the study period, and also in patients with comorbidity, and we found a corresponding increase in risk of revision due to infection for uncemented THAs. In addition, THA patients had more comorbidity within all groups of fixation, and this may have contributed to the increased risk of revision due to infection. Supplementary data Tables 2–5 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.20 20.1851533 HD performed the analyses of the data and wrote the manuscript. All authors contributed to the conception and design of the study, critical analyses of the data, interpretation of the findings, and critical revision of the manuscript through all stages of the study. The authors would like to thank the Norwegian surgeons for their contribution by thoroughly reporting to the NAR and the staff at the NAR for their meticulous punching and statistical advice. Acta thanks Esa Jämsen and Annette W-Dahl for help with peer review of this study. Amlie E, Havelin L I, Furnes O, Baste V, Nordsletten L, Hovik O, Dimmen S. Worse patient-reported outcome after lateral approach than after anterior and posterolateral approach in primary hip arthroplasty: a cross-sectional questionnaire study of 1,476 patients 1–3 years after surgery. Acta Orthop 2014; 85(5): 463-9. Berg T, Løwer H L, Alberg T, Eriksen H M. (Norwegian) Årsrapport 2018 Helsetjenesteassosierte infeksjoner, antibiotikabruk (NOIS), antibiotikaresistens (MSIS) og Verdens håndhygienedag [in Norwegian]. Norwegian Institute of Public Health; 2019. Brochin R L, Phan K, Poeran J, Zubizarreta N, Galatz L M, Moucha C S. Trends in periprosthetic hip infection and associated costs: a populationbased study assessing the impact of hospital factors using national data. J Arthroplasty 2018; 33(7): 233-8. Choi H J, Adiyani L, Sung J, Choi J Y, Kim H B, Kim Y K, Kwak Y G, Yoo H, Lee S O, Han S H, Kim S R, Kim T H, Lee H M, Chun H K, Kim J S, Yoo J D, Koo H S, Cho E H, Lee K W. Five-year decreased incidence of surgi-

214

cal site infections following gastrectomy and prosthetic joint replacement surgery through active surveillance by the Korean Nosocomial Infection Surveillance System. J Hosp Infect 2016; 93(4): 339-46. Dale H, Hallan G, Espehaug B, Havelin L I, Engesaeter L B. Increasing risk of revision due to deep infection after hip arthroplasty. Acta Orthop 2009; 80(6): 639-45. Dale H, Fenstad A M, Hallan G, Havelin L I, Furnes O, Overgaard S, Pedersen A B, Kärrholm J, Garellick G, Pulkkinen P, Eskelinen A, Mäkelä K, Engesæter L B. Increasing risk of prosthetic joint infection after total hip arthroplasty. Acta Orthop 2012; 83(5): 449-58. ECDC. Surveillance of surgical site infections and prevention indicators in European hospitals: HAISSI protocol. European Centre for Disease Prevention and Control; 2017. Available from https://www.ecdc.europa.eu/ en/publications-data/surveillance-surgical-site-infections-and-preventionindicators-european ECDC. Healthcare-associated infections: surgical site infections. Annual Epidemiological Report for 2017. European Centre for Disease Prevention and Control; 2019. Available from https://www.ecdc.europa.eu/sites/default/ files/documents/AER_for_2017-SSI.pdf. Flurin L, Greenwood-Quaintance K E, Patel R. Microbiology of polymicrobial prosthetic joint infection. Diagn Microbiol Infect Dis 2019; 94(3): 255-9. Furnes O, Hallan G, Gjertsen J E, Visnes H, Gundersen T, Fenstad A M, Dybvik E, Kroken G. The Norwegian Arthroplasty Register, Annual Report. Norwegian Arthroplasty Register; 2019. Available from http:// nrlweb.ihelse.net/eng/Rapporter/Report2019_english.pdf Gjertsen J E. Should total hip arthroplasty be used for hip fracture? N Engl J Med 2019; 381(23): 2261-2. 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. Acta Orthop 2015; 86(3): 326-34. doi: 10.3109/17453674.2015.1011983. Hastie T J, Tibshirani R J. Generalized additive models. London: Chapman & Hall; 1990. Havelin L I, Engesæter L B, Espehaug B, Furnes O, Lie S A, Vollset S E. The Norwegian Arthroplasty Register: 11 years and 73,000 arthroplasties. Acta Orthop Scand 2000; 71(4): 337-53. Jämsen E, Nevalainen P, Eskelinen A, Huotari K, Kalliovalkama J, Moilanen T. Obesity, diabetes, and preoperative hyperglycemia as predictors of periprosthetic joint infection: a single-center analysis of 7181 primary hip and knee replacements for osteoarthritis. J Bone Joint Surg Am 2012; 94(14): 101-9. Kurtz S M, Ong K L, Schmier J, Mowat F, Saleh K, Dybvik E, Kärrholm J, Garellick G, Havelin L I, Furnes O, Malchau H, Lau E. Future clinical and economic impact of revision total hip and knee arthroplasty. J Bone Joint Surg Am 2007; 89(Suppl. 3): 144-51. Kurtz S M, Lau E C, Son M S, Chang E T, Zimmerli W, Parvizi J. Are we winning or losing the battle with periprosthetic joint infection: trends in periprosthetic joint infection and mortality risk for the Medicare population. J Arthroplasty 2018; 33(10): 3238-45. Lamagni T, Elgohari S, Harrington P. Trends in surgical site infections following orthopaedic surgery. Curr Opin Infect Dis 2015; 2(2): 125-32. 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 2017; 6(6): 391-8. Leong J W, Cook M J, O’Neill T W, Board T N. Is the use of antibiotic-loaded bone cement associated with a lower risk of revision after primary total hip arthroplasty? Bone Joint J 2020; 102-b(8): 997-1002. Lie S A, Engesæter L B, Havelin L I, Gjessing H K, Vollset S E. Dependency issues in survival analyses of 55,782 primary hip replacements from 47,355 patients. Stat Med 2004; 23(20): 3227-40. Lindgren J V, Gordon M, Wretenberg P, Kärrholm J, Garellick G. Validation of reoperations due to infection in the Swedish Hip Arthroplasty Register. BMC Musculoskelet Disord 2014; 15: 384. Liu W, Wahafu T, Cheng M, Cheng T, Zhang Y, Zhang X. The influence of

Acta Orthopaedica 2021; 92 (2): 208–214

obesity on primary total hip arthroplasty outcomes: a meta-analysis of prospective cohort studies. Orthop Traumatol Surg Res 2015; 101(3): 289-96. Manniën J, van den H S, Muilwijk J, van den Broek P J, van Benthem B, Wille J C. Trends in the incidence of surgical site infection in the Netherlands. Infect Control Hosp Epidemiol 2008; 29(12): 1132-8. Miller L E, Gondusky J S, Kamath A F, Boettner F, Wright J, Bhattacharyya S. Influence of surgical approach on complication risk in primary total hip arthroplasty. Acta Orthop 2018; 89(3): 289-94. Mjaaland K E, Svenningsen S, Fenstad A M, Havelin L I, Furnes O, Nordsletten L. Implant Survival after minimally invasive anterior or anterolateral vs. conventional posterior or direct lateral approach: an analysis of 21,860 total hip arthroplasties from the Norwegian Arthroplasty Register (2008 to 2013). J Bone Joint Surg Am 2017; 99(10): 840-7. Parvizi J, Saleh K J, Ragland P S, Pour A E, Mont M A. Efficacy of antibioticimpregnated cement in total hip replacement. Acta Orthop 2008; 79(3): 335-41. Parvizi J, Gehrke T, Chen A F. Proceedings of the International Consensus on Periprosthetic Joint Infection. Bone Joint J 2013; 95-b(11): 1450-2. Parvizi J, Fassihi S C, Enayatollahi M A. Diagnosis of periprosthetic joint infection following hip and knee arthroplasty. Orthop Clin North Am 2016; 47(3): 505-15. Parvizi J, Shohat N, Gehrke T. Prevention of periprosthetic joint infection: new guidelines. Bone Joint J 2017; 99-b(4 Suppl. B): 3-10. Parvizi J, Tan T L, Goswami K, Higuera C, Della Valle C, Chen A F, Shohat N. The 2018 definition of periprosthetic hip and knee infection: an evidence-based and validated criteria. J Arthroplasty 2018; 33(5): 1309-14. Pedersen A B, Mehnert F, Johnsen S P, Sørensen H T. Risk of revision of a total hip replacement in patients with diabetes mellitus: a population-based follow up study. J Bone Joint Surg Br 2010a; 92(7): 929-34. Pedersen A B, Svendsson J E, Johnsen S P, Riis A, Overgaard S. Risk factors for revision due to infection after primary total hip arthroplasty: a population-based study of 80,756 primary procedures in the Danish Hip Arthroplasty Registry. Acta Orthop 2010b; 81(5): 542-7. Perfetti D C, Boylan M R, Naziri Q, Paulino C B, Kurtz S M, Mont M A. Have periprosthetic hip infection rates plateaued? J Arthroplasty 2017; 32(7): 2244-7. 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; 88(6): 578-80. Ranstam J, Kärrholm J, Pulkkinen P, Mäkelä K, Espehaug B, Pedersen A B, Mehnert F, Furnes O. Statistical analysis of arthroplasty data, I: Introduction and background. Acta Orthop 2011a; 82(3): 253-7. Ranstam J, Kärrholm J, Pulkkinen P, Mäkelä K, Espehaug B, Pedersen A B, Mehnert F, Furnes O. Statistical analysis of arthroplasty data, II: Guidelines. Acta Orthop 2011b; 82(3): 258-67. Sodhi N, Anis H K, Garbarino L J, Gold P A, Kurtz S M, Higuera C A, Hepinstall M S, Mont M A. Have we actually reduced our 30-day short-term surgical site infection rates in primary total hip arthroplasty in the United States? J Arthroplasty 2019; 34(9): 2102-6. Ting N T, Della Valle C J. Diagnosis of periprosthetic joint infection: an algorithm-based approach. J Arthroplasty 2017; 32(7): 2047-50. Varnum C, Pedersen A B, Gundtoft P H, Overgaard S. The what, when and how of orthopaedic registers: an introduction into register-based research. EFORT Open Reviews 2019; 4(6): 337-43. Wang J, Zhu C, Cheng T, Peng X, Zhang W, Qin H, Zhang X. A systematic review and meta-analysis of antibiotic-impregnated bone cement use in primary total hip or knee arthroplasty. PLoS One 2013; 8(12): e82745. Waugh H, Charnley J. The man and the hip. London: Springer Verlag; 1990; 153-69. Zeller V A, Letembet V A, Meyssonnier V A, Heym B, Ziza J M, Marmor S D. Cutibacterium (fFormerly Propionibacterium) avidum: a rare but avid agent of prosthetic hip infection. J Arthroplasty 2018; 33(7): 2246-50. Zhu Y, Zhang F, Chen W, Liu S, Zhang Q, Zhang Y. Risk factors for periprosthetic joint infection after total joint arthroplasty: a systematic review and meta-analysis. J Hosp Infect 2015; 89(2): 82-9.

Acta Orthopaedica 2021; 92 (2): 215–221

215

Hospital variation in the risk of infection after hip fracture surgery: a population-based cohort study including 29,598 patients from 2012–2017 Jeppe Damgren VESTERAGER 1, Pia Kjær KRISTENSEN 1,2, Irene PETERSEN 1,3, and Alma Becic PEDERSEN 1 1 Department

of Clinical Epidemiology, Aarhus University Hospital, Aarhus N, Denmark; 2 Department of Orthopedic Surgery, Regional Hospital Horsens, Horsens, Denmark; 3 Department of Primary Care and Population health, University College London, London, UK Correspondence: jdv@clin.au.dk Submitted 2020-06-02, Accepted 2020-11-09.

Background and purpose — Understanding the key drivers of hospital variation in postoperative infections after hip fracture surgery is important for directing quality improvements. Therefore, we investigated variation in the risk of any infection, and subgroups of infections including pneumonia and sepsis after hip fracture surgery. Methods — In this nationwide population-based cohort study, all Danish patients aged ≥ 65 undergoing surgery for an incident hip fracture from 2012 to 2017 were included. Risk of postoperative infections, based on data from hospital registration (hospital-treated infections) and antibiotic dispensing (community-treated infections), were calculated using multilevel Poisson regression analysis. Hospital variation was evaluated by intra-class coefficient (ICC) and median risk ratio (MRR). Results — The risk of hospital-treated infection was 15%. The risk of community-treated infection was 24%. The adjusted risk varied between hospitals from 7.8–25% for hospital-treated infection and 16–34% for communitytreated infection. The ICC indicated that 19% of the adjusted variance was due to hospital level for hospital-treated infection. The ICC for community-treated infections was 13%. The MRR showed a 2-fold increased risk for the average patient acquiring a hospital-treated infection at the highest risk hospital compared with the lowest risk hospital. For community-treated infection, the MRR was 1.4. Interpretation — Our results suggest that 20% of infections could be reduced by applying the top performing hospitals’ approach. Nearly a 5th of the variation was at the hospital level. This suggests a more standardized approach to avoid postoperative infection after hip fracture surgery.

Hip fracture is a leading cause of hospital admission among the elderly. The 30-day mortality following hip fracture surgery has been approximately 10% during the last few years in Denmark (Pedersen et al. 2017). Higher mortality after hip fracture has been associated with a range of hospital factors (Kristensen et al. 2016, Sheehan et al. 2016) and patient factors in observational studies (Roche et al. 2005). Furthermore, variation in 30-day mortality after hip fracture surgery has been observed between Danish hospitals, but not fully explained (Kristensen et al. 2019). Postoperative infection among hip fracture patients is associated with a 3-fold increase in mortality, within 30 days of operation, compared with non-infected patients (Kjørholt et al. 2019a). Additionally, postoperative infections adversely affect quality of life and hospital costs (Shander et al. 2011). The increased risk of infections after hip fracture surgery is a consequence of multiple patient-, surgery-, and hospitalrelated factors (Taylor and Oppenheim 1998, Poh and Lingaraj 2013). In the past decade, the 30-day cumulative incidence of postoperative infection after hip fracture has increased substantially in Denmark, reaching 14% in 2015–2016 (Kjørholt et al. 2019b), suggesting room for quality improvement. Postoperative infections could be a relevant quality performance measure for ranking hospitals as good treatment, rehabilitation, and care of hip fracture patients should reduce postoperative infections. No previous studies have investigated the hospital variation in postoperative infections among hip fracture patients. However, in order to interpret hospital variation in postoperative infections it is important to understand the relative contributions of patient and healthcare factors. Multilevel models can estimate and separate the relative contribution of the hospital context (hospital level) and patient characteristics (patient level) to the total between-hospital variation

216

in the infections. Thus, studying hospital variation in postoperative infections using multilevel models is an important step towards understanding the key drivers of high infection risk in general and implementation of targeted prevention strategies We investigated the variation between hospitals in the risk of infection within 30 days of hip fracture surgery.

Methods Study design The study was designed as a cohort study, with surgery-performing hospitals being the exposure and postoperative infection being the outcome. The study was based on data from prospectively collected nationwide population-based medical registries in Denmark. Setting and participants The medical registries used encompasses the entire Danish population. The Danish healthcare system is tax-supported with free access to care (Schmidt et al. 2019). All patients admitted to hospital with a hip fracture from January 1, 2013 until December 1, 2017 were included (n = 29,937). Patients with incorrect recording of time, meaning patients registered to be operated on before admission, were excluded (n = 127). To avoid imprecise estimates, hospitals that performed less than 15 hip fracture surgeries per year (7 hospitals and 74 patients) or no longer performed hip fracture surgery (1 hospital and 138 patients) were excluded. The final study cohort included 29,598 patients treated at 23 hospitals (Figure 1, see Supplementary data). Exposure was hospitals performing surgery, outcomes were any hospital-treated infection, sepsis, pneumonia, and community-treated hospital within 30 days of operation. Variables The primary outcome was hospital-treated infections, identified from the Danish National Patient Registry (DNPR) based on ICD-10 codes (Table 1, see Supplementary data). The list of infections included chronic and more rare infections, so as also to detect possible flare-up in already ongoing infections. We excluded urinary tract infections (UTI) because of the high risk of different registration praxis among hospitals. First, there is no economic benefit in the coding of UTI for the department. Second, elderly patients with UTI often have persistent symptomless bacteriuria, which may cause a positive urinary culture without any symptoms (Gavazzi and Krause 2002). The secondary outcomes were 2 of the most common subtypes of hospital-treated infection: pneumonia and sepsis. Table 2 (see Supplementary data) shows the frequency of all infections. Community-treated infections were identified from the Danish National Health Service Prescription Database (DNHSPD) based on ATC codes (Table 3, see Supplemen-

Acta Orthopaedica 2021; 92 (2): 215–221

tary data) and defined as at least 1 dispensing of any antibiotic within 30 days of surgery. Covariates To account for hospital case mix, we collected several wellestablished prognostic factors known to increase infection risk (Poh and Lingaraj 2013). Comorbidity was summarized according to the Charlson Comorbidity Index (CCI). From the DNPR, ICD-10 codes were used to identify CCI (Table 4, see Supplementary data). Data on age, sex, BMI, surgery delay, and surgery type was obtained from the Danish Multidisciplinary Hip Fracture Registry (DMHFR). See Figure 2 (in Supplementary data) for a chart of the study design. Data sources The Danish Civil Registration System (DCRS) has assigned to all residents in Denmark a unique 10-digit personal identiﬁcation number at birth or upon immigration since 1968. This number encodes age, sex, and date of birth. It is recorded at all contacts with the healthcare system. Therefore, an unambiguous record linkage between all medical registers in the population is possible (Schmidt et al. 2019). The DMHFR is a nationwide clinical quality registry on all patients aged ≥ 65 years operated on at Danish hospitals with a medial (S720), pertrochanteric (S721), or subtrochanteric (S722) femoral fracture since 2003 (Kristensen et al. 2020, Hjelholt et al. 2020). The DNPR has registered all non-psychiatric inpatient hospital admissions since 1977 and all hospital outpatient and emergency room visits since 1995. The DNPR contains records of dates of admission and discharge, discharge diagnoses, and up to 20 secondary discharge diagnosis codes according to the ICD-10 (Schmidt et al. 2015). The DNHSPD has registered all redeemed prescriptions from pharmacies in Denmark since 2004. The treatments are coded according to the Anatomical Therapeutic Chemical (ATC) classification (Johannesdottir et al. 2012). Statistics An appropriate statistical method to evaluate hospitals’ performance is multilevel models (Abel and Elliott 2019). We used such models to account for the fact that patients were nested within hospitals. Thereby, any unexplained variation in infection was divided into patient-specific variation and hospitalspecific variation. Differences among patients were considered by adjusting for the patient-level characteristics: age, sex, comorbidity, BMI, surgery delay, and surgery type. For each outcome we performed a multilevel Poisson regression analysis, including a random effect to account for the within-cluster correlation between hospitals. Furthermore, an offset term was included for the time parameter; therefore the outcome can be interpreted as a rate (Austin et al. 2018).

Acta Orthopaedica 2021; 92 (2): 215–221

League tables for each outcome were created. The league tables show the ranking of hospitals by risk of acquiring a postoperative infection. The crude tables were adjusted for hospital level, while the adjusted tables have taken individual covariates into account. Hospitals with less than 5 outcomes were not shown in the league tables due to Danish data protection agency rules regarding personally identifiable data. However, all hospitals were included in the analyses. We evaluated hospital variation from the intra-class coefficient (ICC) and the median rate ratio (MRR). ICC denotes the proportion of hospital variance compared with the total variance that is unexplained by the already defined covariates. An ICC value for hospital of 100% denotes all unexplained variation is due to hospital level, while an ICC of 0% denotes all unexplained variation is at patient level. MRR denotes the median relative change in the rate of the outcome between 2 patients with identical characteristics from different hospitals, comparing the highest risk hospital with the lowest risk hospital. An MRR of 1 is equal to no hospital variance. Confidence intervals for ICC and MRR were estimated with bootstrapping using 100 iterations. For BMI, 17% of data was missing. We applied a multiple imputation strategy, using ordered logistic regression, to impute BMI, assuming data was missing at random, and computed 17 imputations. Sensitivity analysis A series of sensitivity analyses assessed the robustness of our estimates and accounted for variability in clinical practice. 1st, to investigate whether loss to follow-up due to death would introduce bias, we calculated the risk of infection and mortality as a combined outcome. 2nd, patients might already have been infected at admission. Therefore, we repeated the analyses, excluding all patients who had redeemed any antibiotic prescription 14 days prior to surgery date. 3rd, hospitals may have different strategies to identify infections before discharge. Therefore, we investigated whether patients were discharged with infection after the primary hospitalization for hip fracture or readmitted with an infection. 4th, infections may go undetected at the hospital, but later be detected by a general practitioner. Therefore, we combined hospital-treated infection and community-treated infection to a single outcome and repeated the analysis. 5th, to ensure identical follow-up time for community-treated infection, we repeated the analyses starting follow-up at discharge. All analyses were performed in STATA 15.1 or R version 3.6.1 (StataCorp, College Station, TX, USA). Ethics, funding, and potential conflicts of interest The study was approved by the Danish Data Protection Agency (journal number 2015-57-0002) and Aarhus University’s journal number 2016-051-000001 (record number 880). As there was no contact with patients and no study interventions were performed, permission from the scientific ethical committee

217

Table 5. Patient characteristics. Values are count (%) Variable

Hospital-treated infection Not infected Infected Total n = 25,066 n = 4,532 n = 29,598

Women 17,860 (71) 2,694 (59) 20,554 (69) Age 65–79 9,670 (38) 1,376 (30) 11,046 (37) 80–89 10,456 (42) 2,092 (46) 12,548 (43) > 89 4,940 (20) 1,064 (24) 6,004 (20) Charlson Comorbidity Index None (0 points) 9,869 (40) 1,243 (27) 11,112 (37) Low (1–2 points) 9,844 (39) 1,934 (43) 11,778 (40) High (> 3 points) 5,353 (21) 1,355 (30) 46,708 (23) BMI Underweight (< 18.5) 3,271 (13) 646 (14) 3,917 (13) Normal (18.5–24.9) 10,326 (41) 1,726 (38) 12,052 (41) Overweight (25–29.9) 5,650 (22) 973 (22) 6,623 (22) Obese (≥ 30) 1,690 (6.7) 314 (6.9) 2,004 (7) Missing 4,129 (17) 873 (19) 5,002 (17) Surgery delay < 24 h 17,404 (67) 3,002 (66) 20,406 (69) 24–48 h 6,080 (24) 1,230 (27) 7,310 (24) > 48 h 1,582 (6.3) 300 (6.6) 1,882 (7) Surgery type Osteosynthesis 16,291 (65) 2,782 (61) 19,073 (64) Total/hemiarthroplasty 8,775 (35) 1,750 (39) 10,525 (36)

was not necessary according to Danish law. The study was supported by a grant from the Novo Nordisk Foundation (reference number NNF190C0056429). The authors reported no conflicts of interest to declare.

Results 29,598 patients from 23 different hospitals were included, of whom the majority were women and aged 65–89 years. Overall, 15% of patients were diagnosed with a hospital-treated infection, whereas 10% were diagnosed with pneumonia, and 1.8% with sepsis. Additionally, 24% had a community-treated infection within 30 days of surgery (Table 5). Hospital-treated infections The average risk of any hospital-treated infections varied between 8.2% and 27% among hospitals. After adjustment for hospital case mix, the risk varied from 7.8% to 25% (Figure 3). The adjusted variance attributed to hospital level was 19% (95% CI 10–25). The risk of acquiring any hospital-treated infection at the highest risk hospital compared with the lowest risk hospital for a patient with identical characteristics was 2.0 (CI 1.6–2.3) (Table 6). Furthermore, increasing age and comorbidity were strongly associated with higher risk of hospital-treated infection. Men had an increased risk compared with women (RR = 1.6, CI 1.5–1.7). Underweight patients had a 21% higher risk compared with normal weight patients. Patients operated on with total/hemiarthroplasty had a 14% increased risk compared

218

Acta Orthopaedica 2021; 92 (2): 215–221

Hvidovre Bispebjerg Aarhus Nykøbing Falster Slagelse Køge Kolding Herlev Farsø Hjørring Nordsjælland Aalborg Aabenraa Holbæk Holsterbro Esbjerg Viborg Odense Horsens Thy Vejle Randers Bornholm

Hvidovre Bispebjerg Aarhus Nykøbing Falster Slagelse Køge Kolding Herlev Farsø Hjørring Nordsjælland Aalborg Aabenraa Holbæk Holsterbro Esbjerg Horsens Viborg Odense Thy Vejle Randers Bornholm 0

Lorem ipsum

Lorem ipsum11111

Lorem ipsum

Percent of patients with hospital-treated infection — crude Lorem ipsum

Lorem ipsum

Lorem ipsum11111

Lorem ipsum

Percent of patients with hospital-treated infection — adjusted Lorem ipsum

Lorem ipsum

Figure 3. League tables ranking hospitals for hospital-treated infections.

Table 6. Multilevel Poisson regression for hospital-treated infection and stratified for pneumonia and sepsis. Values are relative risk (95% confidence interval) Hospital-treated Individual variables infection Pneumonia Sepsis Sex (ref. Female) Male 1.59 (1.50–1.70) 1.81 (1.68–1.95) 2.23 (1.87–2.67) Age (ref. 65–79) 80–89 1.40 (1.31–1.50) 1.59 (1.46–1.73) 1.72 (1.40–2.12) > 89 1.55 (1.43–1.68) 1.84 (1.66–2.03) 1.83 (1.43–2.39) Charlson Comorbidity Index (ref. 0 points) Low (1–2 points) 1.36 (1.27–1.46) 1.48 (1.35–1.61) 1.31 (1.06–1.62) High (> 3 points) 1.60 (1.48–1.73) 1.68 (1.53–1.86) 1.69 (1.34–2.12) BMI (ref. 18.5–24.9) Underweight (< 18.5) 1.21 (1.11–1.31) 1.28 (1.16–1.41) 1.20 (0.95–1.52) Overweight (25–29.9) 1.00 (0.93–1.07) 1.00 (0.91–1.09) 0.86 (0.69–1.06) Obese (≥ 30) 1.11 (0.99–1.24) 1.04 (0.90–1.19) 0.78 (0.55–1.12) Surgery delay (ref. < 24 h) 24–48 h 1.09 (1.02–1.17) 1.15 (1.06–1.24) 1.09 (0.89–1.33) > 48 h 1.06 (0.94–1.19) 1.01 (0.86–1.17) 1.38 (1.00–1.89) Operation type (ref. Osteosynthesis) Total/hemiarthroplasty 1.14 (1.08–1.22) 1.14 (1.06–1.23) 0.93 (0.78–1.12) Hospital contextual effects ICC a hospital (%) 18.8 (10.0– 24.9) 12.1 (5.7–18.8) 1.8 (0.6–3.7) MRR b 1.96 (2.33–1.57) 2.08 (0.40–0.97) 1.82 (2.33–1.41) a ICC = intra-class coefficient. b MRR = median risk ratio.

with patients operated on with osteosynthesis. For the specific hospital-treated infections, pneumonia and sepsis, the results were similar except regarding surgery type (Table 6). The risk for pneumonia and sepsis varied across hospitals and was reduced after adjustment (Figures 5 and 6, see Supplementary data). The hospital variance for pneumonia and sepsis was similar to any hospital-treated infections. The amount of variation attributed to hospital level was 12% (CI 5.0–15) for pneumonia and 1.8 % (CI 0.6–3.7) for sepsis (Table 6).

Table 7. Multilevel Poisson regression for community-treated infection. Values are relative risk (95% confidence interval) Community-treated Individual variables infection Sex (ref. Female) Male 0.94 (0.89–0.99) Age (ref. 65–79) 80–89 1.29 (1.22–1.36) > 89 1.44 (1.35–1.53) Charlson Comorbidity Index (ref. 0 points) Low (1–2 points) 1.17 (1.10–1.23) High (> 3 points) 1.24 (1.16–1.32) BMI (ref. 18.5–24.9) Underweight (< 18.5) 1.02 (0.94–1.09) Overweight (25–29.9) 1.05 (0.99–1.11) Obese (≥ 30) 1.19 (1.10–1.30) Surgery delay (ref. < 24 h) 24–48 h 1.01 (0.96–1.07) > 48 h 1.04 (0.95– 1.15) Operation type (ref. Osteosynthesis) Total/hemiarthroplasty 0.97 (0.93–1.02) Hospital contextual effects ICC a hospital (%) 13.3 (6.0–20.5) MRR b 1.43 (1.56–1.25) a, b See

Table 6.

Community-treated infections The average risk for community-treated infection varied between 17% and 34% among hospitals. After adjustment for hospital case mix, the risk varied from 16% to 34% (Figure 4). The adjusted variance attributed to hospital level was 13% (CI 10–25). The risk of acquiring a community-treated infection at the highest risk hospital compared with the lowest risk hospital for a patient with identical characteristics was 1.4 (95% CI 1.3–1.6) (Table 7).

Acta Orthopaedica 2021; 92 (2): 215–221

219

Holsterbro Aabenraa Aarhus Bispebjerg Randers Kolding Odense Horsens Thy Hvidovre Viborg Hjørring Herlev Nordsjælland Slagelse Nykøbing Falster Bornholm Vejle Esbjerg Køge Farsø Aalborg Holbæk

Holsterbro Aabenraa Aarhus Bispebjerg Randers Kolding Horsens Odense Thy Hvidovre Viborg Nordsjælland Hjørring Herlev Slagelse Bornholm Nykøbing Falster Vejle Esbjerg Køge Farsø Aalborg Holbæk Lorem ipsum

Lorem ipsum11111

Lorem ipsum

Percent of patients with community-aquired infection — crude Lorem ipsum

Lorem ipsum

Lorem ipsum11111

Lorem ipsum

Percent of patients with community-aquired infection — adjusted Lorem ipsum

Lorem ipsum

Figure 4. League tables ranking hospitals for community-acquired infections.

Furthermore, increasing age and comorbidity were strongly associated with higher risk of community-treated infection. Obese patients had a 19% increased risk compared with normal weight patients. There were no differences in the risk of community-treated infection by surgery delay, type of surgery, or gender (Table 7). Sensitivity analysis 1st, when combining mortality and hospital-treated infection as a single outcome, the risk varied between 15% and 29% (Table 8, see Supplementary data) with 8.4% (CI 3.8– 12) of the variation due to hospital level. The MRR showed an increased risk for a patient operated on at the highest risk hospital of 1.5 (CI 1.3–1.6) compared with the lowest risk hospital. 2nd, excluding all patients who had redeemed a prescription for antibiotics < 14 days prior to surgery did not change the results considerably (Table 9, see Supplementary data). 3rd, three-quarters of hospital-treated infections were detected during primary hospitalization, with hospital variation between 50% and 85%. Hospitals with a high infection risk had more infections detected during primary hospitalization (Figure 7, see Supplementary data). 4th, when combining hospital-treated infection and community-treated infection, the risk of infection was 34%, varying from 25% to 46% between hospitals (Figure 8, see Supplementary data). Hospital level explained 11% (CI 4.1–16) of the variation (Table 10, see Supplementary data). 5th, when starting follow-up at discharge, community-treated infection varied between 15% and 29% (Figure 9, see Supplementary data). The MRR showed an increased risk of 1.3 (CI 1.2–1.5) between the lowest risk hospital and the highest risk hospital. The ICC indicated that 7.3% (CI 3.3–12) of the adjusted variance was due to hospital level (Table 10, see Supplementary data)

Discussion Our study is the first to examine the variation between hospitals in the risk of hospital-treated and community-treated infections following hip fracture surgery, and to quantify the hospital-level contribution to the variation using a nationwide population-based cohort design. We found a more than 3-fold difference in hospital-treated infections between hospitals, where 19% of the variation was attributed to hospital level. The variation was sustained when stratifying for pneumonia and sepsis. For community-treated infection, we found a 2-fold difference between hospitals, with 13% of the variation attributed to hospital level. Strength and limitations This study was based on a nationwide population-based cohort design, prospectively collected individual-level data, and complete follow-up of all patients. We included nearly 30,000 patients with free-of-charge and equal access to healthcare services, thereby reducing the risk of selection bias. When investigating death and hospital-treated infection as combined outcome, we found a minor decrease in variation. Therefore, we do not consider loss to follow-up from death to introduce any pertinent bias. A limitation of this study regards the validity of data, as this is collected by numerous clinicians as part of daily routine clinical work. We cannot exclude the possibility that variation in reporting practice between hospitals can overestimate or underestimate infections. We identified hospital-treated infections based on ICD-10 codes from the DNPR, which is known to have a high positive predictive value (PPV), in other patient groups (Holland-Bill et al. 2014). However, the PPV might vary between hospitals. Unfortunately, we did not have data on radiographs, changes in inflammatory markers etc. to confirm the diagnosis codes’ negative predictive value, and

220

thereby assess the amount of misclassification of infections. However, because infections do not clear spontaneously, we combined hospital-treated infections with community-treated infections and found a slightly lower variation due to hospital level. Additionally, we included only infection diagnoses, for which the hospitals receive payment based on their registration of diagnoses. We therefore assume that all patients treated for infection are registered. Furthermore, in the case of under-reporting infections at specific hospitals, we would have observed some hospitals with a negligible low infection risk, which was not the case. However, when analyzing specific infections such as pneumonia and sepsis we observed a lower variation, as well as a lower amount of variation attributed to hospital level. This points towards the hypothesis that the more severe the infection, the easier the infection becomes to detect, which may lead to less misclassification by hospital variation. Regarding the infections included in “any infections,” we found that chronic infections such as HIV were very few. The same applies for infections less relevant to hospital admission for hip fracture, such as ear and eye infections (Table 2, see Supplementary data)). Therefore, “any infections” may predominantly be interpreted as infections associated with hip fracture and hospital admission. Hip fracture patients in Denmark are admitted to the nearest hospital offering hip fracture surgery and are therefore not classified according to health status, fracture severity, or other characteristics. This minimizes the risk of confounding by indication. Finally, we adjusted for a range of well-established prognostic factors to reduce confounding, including the CCI, which comprised complete in-hospital comorbidity history. However, we did not have information on the severity of diseases in the CCI or full information on all factors exposing for infection. Therefore, we cannot exclude the possibility of residual confounding. Comparison with previous literature Previous studies on hospital variation in postoperative infections have primarily focused on cardiac surgery (Hirahara et al. 2019) or combined multiple surgical procedures (Wakeam et al. 2016). However, 1 study on elective hip and knee arthroplasties has shown a 4-fold difference in risk between hospitals in the United States (Bozic et al. 2014) for complications, including pneumonia. We found a 5-fold difference for pneumonia. However, our study population was acutely operated on, older, more frail, and had more comorbidities compared with patients undergoing elective hip arthroplasty. In addition, our absolute risk estimates were much higher, suggesting that more standardized and complex care of patients could contribute to mortality reduction. When looking at hospital variation attributed to hospital level in other outcomes, a Dutch study investigated the hospital variation in any-cause readmission within 30 days among

Acta Orthopaedica 2021; 92 (2): 215–221

patients operated on for a femoral neck fracture. They reported the risk to vary among hospitals between 2.2% and 11% (Hekkert et al. 2018). Moreover, the study found 2.3% of the variation explained by hospital level. We found a higher risk only of postoperative infections, probably due to our inclusion of infections detected during primary hospitalization. Furthermore, hospital level explained 19% of the variation in postoperative infections in our study. This suggest that variation due to hospital level for postoperative infections is more frequent than for any-cause readmission. Any-cause readmissions are thereby a less sensitive marker for hospital performance than postoperative infections. The same applies to mortality. This is supported by the fact that the ICC and MRR in our study is higher compared with a previous Danish variation study on 30-day mortality after hip fracture, which found that less than 1% of the variation in mortality was explained by hospital level (Kristensen et al. 2019). Clinical implications Our results imply that quality of in-hospital care for hip fracture patients is not homogeneous regarding postoperative infections. We found that patients predominantly had their infection detected during the primary hospitalization. We found nearly a 5th of the variation was explained by hospital-level factors, whereas the largest variation was due to individual-level factors. Previous studies have evaluated interventions to decrease postoperative pneumonia with success. As we showed the most common postoperative infection to be pneumonia, this should be the primary focus in such interventions. Kazaure et al. (2014) propose a standardized postoperative pneumonia program, including education of nursing staff, coughing and deep-breathing exercises, twicedaily oral hygiene, ambulation, and elevated head of the bed during meals. This intervention showed a 44% decreased rate of postoperative pneumonia among 4,099 American, noncardiac, surgical patients. Furthermore, a study from Taiwan included 240 hip fracture patients. They showed a pneumonia risk of 14%, which we regard as comparable to ours at 10% (Chang et al. 2018). Their study showed a decrease in postoperative pneumonia to 5.9% among an intervention group implemented with deep-breathing exercises, chest physiotherapy, and cough-assisted maneuvers. Since postoperative infections are associated with higher mortality, a decrease in postoperative infection would lead to decreased mortality. In conclusion, we advocate for improvement of national clinical guidelines to detect and treat infections during primary hospitalization. This may be as a standardized infection screening of all hip fracture patients or implantation of a standardized infection prevention program. Supplementary data Tables 1–4 and 8–10 and Figures 1–2 and 5–10 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.2020.1863688

Acta Orthopaedica 2021; 92 (2): 215–221

All authors contributed to the idea and design of the study. JDV was responsible for statistics and writing the first draft. PKK and IP supervised the statistics. ABP was the primary supervisor on epidemiology. All authors contributed to writing the final manuscript. Acta thanks Max Gordon and Per Kjærsgaard-Andersen for help with peer review of this study. Abel G, Elliott M N. Identifying and quantifying variation between healthcare organisations and geographical regions: using mixed-effects models. BMJ Qual Saf 2019; 28(12): 1032-8. doi: 10.1136/bmjqs-2018-009165. Austin P C, Stryhn H, Leckie G, Merlo J. Measures of clustering and heterogeneity in multilevel Poisson regression analyses of rates/count data. Stat Med 2018; 37(4): 572-89. doi: 10.1002/sim.7532. Bozic K J, Grosso L M, Lin Z, Parzynski C S, Suter L G, Krumholz H M, Lieberman J R, Berry D J, Bucholz R, Han L, Rapp M T, Bernheim S, Drye E E. Variation in hospital-level risk-standardized complication rates following elective primary total hip and knee arthroplasty. J Bone Joint Surg Am 2014; 96(8): 640-7. doi: 10.2106/JBJS.L.01639. Chang S C, Lai J I, Lu M C, Lin K H, Wang W S, Lo S S, Lai Y C. Reduction in the incidence of pneumonia in elderly patients after hip fracture surgery: an inpatient pulmonary rehabilitation program. Medicine (Baltimore) 2018; 97(33): e11845. doi: 10.1097/MD.0000000000011845. Gavazzi G, Krause K H. Ageing and infection. Lancet Infect Dis 2002; 2(11): 659-66. doi: 10.1016/s1473-3099(02)00437-1. Hekkert K, Kool R B, Rake E, Cihangir S, Borghans I, Atsma F, Westert G. To what degree can variations in readmission rates be explained on the level of the hospital? A multilevel study using a large Dutch database. BMC Health Serv Res 2018; 18(1): 999. doi: 10.1186/s12913-018-3761-y. Hirahara N, Miyata H, Motomura N, Kohsaka S, Nishimura T, Takamoto S. Procedure- and Hospital-level variation of deep sternal wound infection from All-Japan Registry. Ann Thorac Surg 2020; 109(2): 547-54. doi:10.1016/j.athoracsur.2019.05.076 Hjelholt T J, Edwards N M, Vesterager J D, Kristensen P K, Pedersen A B. The positive predictive value of hip fracture diagnoses and surgical procedure codes in the Danish Multidisciplinary Hip Fracture Registry and the Danish National Patient Registry. Clin Epidemiol 2020; 12:123-131. doi: 10.2147/CLEP.S238722. Holland-Bill L, Xu H, Sørensen H T, Acquavella J, Sværke C, Gammelager H, Ehrenstein V. Positive predictive value of primary inpatient discharge diagnoses of infection among cancer patients in the Danish National Registry of Patients. Ann Epidemiol 2014; 24(8): 593-7, 597.e1-18. doi: 10.1016/j. annepidem.2014.05.011. Johannesdottir S A, Horváth-Puhó E, Ehrenstein V, Schmidt M, Pedersen L, Sørensen H T. Existing data sources for clinical epidemiology: the Danish National Database of Reimbursed Prescriptions. Clin Epidemiol 2012; 4: 303-13. doi: 10.2147/CLEP.S37587. Kazaure H S, Martin M, Yoon J K, Wren S M. Long-term results of a postoperative pneumonia prevention program for the inpatient surgical ward. JAMA Surg 2014; 149(9): 914-8. doi: 10.1001/jamasurg.2014.

221

Kjørholt K E, Johnsen S P, Kristensen N R, Prieto-Alhambra D, Pedersen A B. Increasing risk of hospital-treated infections and community-based antibiotic use after hip fracture surgery: a nationwide study 2005–2016. J Bone Miner Res 2019b; 34(3): 437-46. doi: 10.1002/jbmr.3620. Kjørholt K E, Kristensen N R, Prieto-Alhambra D, Johnsen S P, Pedersen A B. Increased risk of mortality after postoperative infection in hip fracture patients. Bone 2019a; 127: 563-70. doi: 10.1016/j.bone.2019.07.023. Kristensen P K, Thillemann T M, Søballe K, Johnsen S P. Are process performance measures associated with clinical outcomes among patients with hip fractures? A population-based cohort study. Int J Qual Health Care 2016; 28(6): 698-708. doi: 10.1093/intqhc/mzw093. Kristensen P K, Merlo J, Ghith N, Leckie G, Johnsen S P. Hospital differences in mortality rates after hip fracture surgery in Denmark. Clin Epidemiol 2019; 11: 605-14. doi: 10.2147/CLEP.S213898. Kristensen P K, Röck N D, Christensen H C, Pedersen A B. The Danish Multidisciplinary Hip Fracture Registry 13-year results from a populationbased cohort of hip fracture patients. Clin Epidemiol 2020; 12: 9-21. doi: 10.2147/CLEP.S231578. Pedersen A B, Ehrenstein V, Szépligeti SK, Lunde A, Lagerros Y T, Westerlund A, Tell G S, Sørensen H T. Thirty-five-year trends in first-time hospitalization for hip fracture, 1-year mortality, and the prognostic impact of comorbidity: a Danish nationwide cohort study, 1980–2014. Epidemiology 2017; 28(6): 898-905. doi: 10.1097/EDE.0000000000000729. Poh K S, Lingaraj K. Complications and their risk factors following hip fracture surgery. J Orthop Surg (Hong Kong) 2013; 21(2): 154-7. doi: 10.1177/230949901302100207. Roche J J, Wenn RT, Sahota O, Moran C G. Effect of comorbidities and postoperative complications on mortality after hip fracture in elderly people: prospective observational cohort study. BMJ 2005; 331(7529): 1374. doi: 10.1136/bmj.38643.663843.55. Schmidt M, Schmidt S A, Sandegaard J L, Ehrenstein V, Pedersen L, Sørensen H T. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7: 449-90. doi: 10.2147/CLEP.S91125. Schmidt M, Schmidt S A J, Adelborg K, Sundbøll J, Laugesen K, Ehrenstein V, Sørensen H T. The Danish health care system and epidemiological research: from health care contacts to database records. Clin Epidemiol 2019; 11: 563-91. doi: 10.2147/CLEP.S179083. Shander A, Fleisher L A, Barie P S, Bigatello L M, Sladen R N, Watson C B. Clinical and economic burden of postoperative pulmonary complications: patient safety summit on definition, risk-reducing interventions, and preventive strategies. Crit Care Med 2011; 39(9): 2163-72. doi: 10.1097/ CCM.0b013e31821f0522. Sheehan K J, Sobolev B, Guy P, Kuramoto L, Morin S N, Sutherland J M, Beaupre L, Griesdale D, Dunbar M, Bohm E, Harvey E; Canadian Collaborative Study of Hip Fractures. In-hospital mortality after hip fracture by treatment setting. CMAJ 2016; 188(17-18): 1219-25. doi: 10.1503/ cmaj.160522. Taylor M E, Oppenheim B A. Hospital-acquired infection in elderly patients. J Hosp Infect 1998; 38(4): 245-60. doi: 10.1016/s0195-6701(98)90074-3. Wakeam E, Hyder J A, Lipsitz S R, Cohen M E, Orgill D P, Zinner M J, Ko C Y, Hall B L, Finlayson S R. Hospital-level variation in secondary complications after surgery. Ann Surg 2016; 263(3): 493-501. doi: 10.1097/ SLA.0000000000001227.

222

Acta Orthopaedica 2021; 92 (2): 222–227

Knee and foot contracture occur earliest in children with cerebral palsy: a longitudinal analysis of 2,693 children Erika CLOODT 1,2, Philippe WAGNER 3, Henrik LAUGE-PEDERSEN 1, and Elisabet RODBY-BOUSQUET 1,3 1 Department of Clinical Sciences Lund, Orthopaedics, Lund University, Lund; 2 Department of Research 3 Centre for Clinical Research Västerås, Uppsala University-Region Västmanland, Västerås, Sweden

and Development, Region Kronoberg, Växjö;

Correspondence: erika.cloodt@med.lu.se Submitted 2020-06-17. Accepted 2020-10-17.

Background and purpose — Joint contracture is a common problem among children with cerebral palsy (CP). To prevent severe contracture and its effects on adjacent joints, it is crucial to identify children with a reduced range of motion (ROM) early. We examined whether significant hip, knee, or foot contracture occurs earliest in children with CP. Patients and methods — This was a longitudinal study involving 27,230 measurements obtained for 2,693 children (59% boys, 41% girls) with CP born 1990 to 2018 and registered before 5 years of age in the Swedish surveillance program for CP. The analysis was based on 4,751 legs followed up for an average of 5.0 years. Separate Kaplan–Meier (KM) curves were drawn for each ROM to illustrate the proportions of contracture-free legs at a given time during the follow-up. Using a clustered bootstrap method and considering the child as the unit of clustering, 95% pointwise confidence intervals were generated for equally spaced time points every 2.5 years for each KM curve. Results — Contracture developed in 34% of all legs, and the median time to the first contracture was 10 years from the first examination. Contracture was most common in children with a higher Gross Motor Function Classification System (GMFCS) level. The first contracture was a flexion contracture preventing dorsiflexion in children with GMFCS level I or II and preventing knee extension in children with GMFCS level III to V. Interpretation — Early interventions to prevent knee and foot contractures in children with CP should be considered.

Joint contracture is a common problem in children with cerebral palsy (CP) (Rosenbaum et al. 2007). Spasticity, muscle imbalance, inability to move, and muscle pathology constrain normal muscle growth and lead to dynamic contracture followed by static joint contracture over time because of tight muscles surrounding the joints (Barrett and Lichtwark 2010). Children with CP exhibit increased sarcomere length and reduced number of satellite cells, both of which affect the ability to maintain muscle length during development and bone growth (Barrett and Lichtwark 2010, Smith et al. 2013). The risk of contracture increases with age and higher level on the Gross Motor Function Classification System (GMFCS) (Nordmark et al. 2009). Some 60% of adults with CP experience contracture in the lower limbs (Agustsson et al. 2018). Joint contracture prevents mechanical alignment of the joints, which affects standing and lying positions, and the quality and energy cost of gait (Raja et al. 2007). Contracture of the foot, knee, or hip joint may also affect adjacent joints and lead to severe postural asymmetries, windswept hips, and scoliosis (Agustsson et al. 2017, 2018, Pettersson et al. 2020). Contracture is often associated with pain, which occurs most frequently in the lower limbs of children with CP (AlrikssonSchmidt and Hägglund 2016, Blackman et al. 2018). To prevent severe joint contracture and reduce its effects on adjacent joints, it is crucial to identify children with reduced range of motion (ROM) to begin targeted treatment early (Chan and Miller 2014). We analyzed whether contracture preventing hip extension, knee extension, or foot dorsiflexion occurs first in children with CP with GMFCS level I to V.

Acta Orthopaedica 2021; 92 (2): 222–227 223

Reports of spasticity-reducing surgery, such as that to insert an ITB or SDR, and the date of surgery or pump insertion were extracted. Surgery of the lower limbs was recorded in the database and was grouped into either soft tissue surgery or bony surgery for the hip, knee, or foot along with the date of surgery for the left and right leg. Examples of the soft tissue surgeries performed were adducFigure 1. Guidelines for clinical examination and hip radiograph within the Swedish Ceretor tenotomy, hamstring or Achilles tendon bral Palsy Follow-up Program (CPUP). lengthening, and tendon or muscles transfer. Examples of the bony surgeries reported were osteotomy, physiodesis, or arthrodesis. Surgery on the upper extremity or the spine, fracture, extraction of osteosynthesis material, or treatment with botulinum toxin injection were not Patients and methods censored and data for children who received these treatments This was a prospective study based on register data from the were retained in the analyses. Swedish Cerebral Palsy Follow-up Program (CPUP), which includes > 95% of all children with CP in Sweden. We ana- Statistics lyzed all measurements reported from the start of the program Hip extension, knee extension, and foot dorsiflexion were in October 1994 until the end of June 2018 and included all dichotomized into 2 groups: no contracture and contracture. children born 1990 to 2018 and registered in the CPUP before Both legs for each child were then followed up from inclu5 years of age. Children registered at 5 years of age or later sion in the CPUP surveillance program until the last examinawere excluded. Children included in the CPUP were examined tion or the date of surgery. The follow-up time was calculated every 6 months, once a year, or every other year, depending on for each leg for each child, and separate Kaplan–Meier (KM) their age and GMFCS level (Figure 1). curves were drawn for each leg’s ROM to illustrate the proporThe systematic follow-up includes several variables such as tions of contracture-free legs at a given time during the followreports of surgery, CP subtype, and clinical examinations of up. For children with unilateral spastic CP, only the affected passive ROM and gross motor function. The full protocol is side was included in the analyses. Using a clustered bootstrap available at: https://cpup.se/in-english/manuals-and-evalua- method and considering the child as the unit of clustering, 95% pointwise confidence intervals (CI) were generated for equally tion-forms/. Gross motor function was classified by the child’s phys- spaced time points every 2.5 years for each KM curve. IBM iotherapist into level I to V according to the expanded and SPSS Statistics (version 25.0; IBM Corp, Armonk, NY, USA), revised version of the GMFCS (Palisano et al. 2008). Passive STATA (version 14, Stata-Corp, College Station, TX, USA), hip extension, knee extension, and dorsiflexion were assessed and R (R Foundation for Statistical Computing, Vienna, Ausby goniometric measurements in standardized positions. Hip tria) were used for the statistical analyses. Categorical variextension was measured with the child in the prone position ables are described by frequency (n) and percentage (%). with legs over the end of the examining table and the pelvis straight. Knee extension and foot dorsiflexion were measured Ethics, funding, and potential conflicts of interest with the child in the supine position with the hip and knee The study was approved by the Medical Research Ethics extended. Committee in Lund (383/2007, 443-99), and permission was Contracture was defined as hip and knee flexion contracture obtained to extract data from the CPUP registry. The study of 10° or more and plantarflexion contracture of at least 10° received funding from Stiftelsen för bistånd åt rörelsehindrade in extension of the hip, knee, or foot. Children with unilateral i Skåne, Promobilia, Forte and Region Kronoberg. The fundspastic CP were identified, and only the affected side was used ing sources had no decision-making role or influence on the in the analyses. For children with ataxic, dyskinetic, or spastic study design, data collection, data analysis, data interpretabilateral CP, both legs were included. tion, or writing of the report. The authors declare that they Each leg that underwent soft tissue surgery or bony surgery have no conflicts of interest. and both legs of children with an intrathecal baclofen pump (ITB) or who underwent a selective dorsal rhizotomy (SDR) operation before the onset of the first contracture were cenResults sored from the analyses at the date of surgery. If information on which leg was operated on was missing, both legs were 2,693 children (1,598 boys, 1,095 girls) with an examination before the age of 5 years were reported between 1994 and censored at the date of surgery.

224

Acta Orthopaedica 2021; 92 (2): 222–227

Table 2. Age at first hip, knee, and foot operation

Table 1. Contracture events and number of person-years according to the Gross Motor Function Classification System I–V (GMFCS) GMFCS level n (%) Contracture, legs (%) I 1,901 (40) 237 (13) II 765 (16) 183 (24) III 540 (11) 247 (46) IV 752 (16) 436 (58) V 793 (17) 499 (63) Total 4,751 1,602 (34)

Age at first operation Type of surgery Median 1st quartile 3rd quartile

Person-years (mean)

Hip, soft tissue Hip, bony Knee, soft tissu Knee, bony Foot, soft tissu Foot, bony

11,450 (6.0) 4,495 (5.9) 2,250 (4.2) 3,166 (4.2) 2,406 (3.0) 23,768 (5.0)

4.0 6.5 9.0 10 7.0 11

3.0 4.0 7.0 5.5 5.0 8.5

6.0 10 12 12 10 14

Proportion of operation-free legs

Proportion of legs without contracture

1.0

0.8

0.6

0.4

0.4 GMFCS 1 2 3 4 5

0.2

0 0

GMFCS I Hip Knee Foot

0.2

Figure 2. Proportions of operation-free legs from the time of the first measurement for each GMFCS level.

0 0

Years from first follow-up

GMFCS II Hip Knee Foot

0.2

Years from first follow-up

Proportion of legs without contracture

1.0

0.8

2018. The analysis was based on 4,751 legs followed up for an average of 5.0 years (only the 0.6 affected leg was included for children with unilateral CP). Contractures were in general most 0.4 common at higher GMFCS levels (Table 1). There were 27,230 measurement occasions. GMFCS III 0.2 Hip Within the 10 years of follow-up, 937 legs Knee (20%) had been operated on. The most common Foot 0 operation was soft tissue surgery of the hip (316 0 5 10 15 Years from first follow-up legs, 6.7%), followed by soft tissue surgery of the foot (201 legs, 4.2%), and bony surgery (osteFigure 3. Proportions of legs free from otomy) of the hip (151 legs 3.2%). The median hip, knee, and foot contracture, stratified by GMFCS level and 95% pointwise ages were 4 years for soft tissue surgery of the confidence intervals for equally spaced hip, 7 years for soft tissue surgery of the foot, and time points every 2.5 years during the 6.5 years for bony surgery of the hip (Table 2). follow-up. The proportions of legs not operated on in children with different GMFCS levels are presented in Figure 2. 1,602 contractures were recorded. A contracture developed in 34% of all legs, and the median time to the first contracture was 10 years from the first examination within the follow-up program. Contracture occurred most frequently in children with a higher GMFCS level. The first contracture to occur in children with GMFCS level I or II was a foot or ankle con-

0.6

0.4

GMFCS IV Hip Knee Foot

0.2

0 0

Years from first follow-up

Proportion of legs without contracture 1.0

0.8

0.6

0.4

GMFCS V Hip Knee Foot

0.2

0 0

Years from first follow-up

Acta Orthopaedica 2021; 92 (2): 222–227 225

Table 3. Proportions of legs free from hip, knee, and foot contracture, stratified by GMFCS level, and 95% pointwise confidence intervals after 10 years of follow-up GMFCS level I II III IV V

Hip

Knee

Ankle

1.0 (0.9–1.0) 0.9 (0.8–0.9) 0.9 (0.8–0.9) 0.8 (0.8–0.8) 0.8 (0.7–0.8)

1.0 (0.9–1.0) 0.9 (0.9–0.9) 0.5 (0.4–0.6) 0.3 (0.3–0.4) 0.2 (0.2–0.3)

0.9 (0.9–0.9) 0.8 (0.8–0.9) 0.7 (0.6–0.8) 0.7 (0.7–0.8) 0.6 (0.6–0.7)

tracture, and a knee contracture was the first contracture to develop in children with GMFCS level III to V. Proportions of legs free from hip, knee, and foot contracture in each GMFCS level are presented in Table 3 and with separate KM curves in Figure 3.

Discussion We identified that the first joint contracture to occur in the lower limb involved foot contracture for children with GMFCS level I or II and knee contracture for children with GMFCS level III to V. Children classified with GMFCS level I and II are ambulant whereas those classified with GMFCS level III to V in general rely on wheeled mobility and spend more time sitting with flexed knees. This could explain the sequence in which the contracture presents. This seems to follow the same pattern as previously reported for pain localization in children with CP, in which children with GMFCS level I or II report pain primarily in the feet and those with GMFCS III to V pain in the knees and hips (Alriksson-Schmidt and Hägglund 2016). These results are consistent with those of a previous study (Nordmark et al. 2009) that reported decreasing ROM from 2 years of age in all lower limb joints. Together, these findings suggest that contracture should be treated early given that ROM seems to decrease over time. Cloodt et al. (2018) found that hamstring length, measured as the unilateral popliteal angle, and dorsiflexion of the foot are strongly associated with the development of knee contracture, whereas spasticity had a significantly smaller effect. Contracture in the lower limbs affects the mechanical position of both the affected and adjacent joints. Preventing normal movements and forces around the joint may increase the risk for additional contractures. Our study excluded children whose first measurement was at 5 years of age or later. The reason for this was to increase the opportunity of identifying the first contracture and to reflect the natural development of contracture as much as possible because information on treatment and surgery before enrollment in the CPUP was not available in these children. Among the children included, their access to early service and follow-up during their first years of life should be considered. All children with CP in Sweden have access to free health care

and interventions from multiprofessional habilitation centers (Alriksson-Schmidt et al. 2017). Surgery or treatment with SDR and ITB can affect the development of contracture in a specific joint as well as in adjacent joints (McGinley et al. 2012). There are several strengths and limitations of our study. Classification of subtype was missing in several cases and therefore not included. The contractures were recorded by goniometric measurement, which is a standard measurement in clinical settings but whose reliability varies according to the joint and position (Hancock et al. 2018, Kim et al. 2018). The results included in our study were based on repeated measurements taken by many different examiners, and this may have introduced information bias or measurement errors. In survival analysis, as we used, measurement error can introduce bias. However, this bias has been shown to be small when differences between groups are moderate in terms of hazard ratios (Oh et al. 2018). Furthermore, in some situations in which bias was found to be substantial, bias attenuates observed differences between groups. For our study, this would mean an underestimation of differences in risk of having a first contracture in a specific joint compared with another joint. We validated the data for incorrectly reported measurements to reduce the risk of incorrect outliers. Examiners in the CPUP are encouraged to practice and learn to perform the register’s standardized measurements, which has been shown to be important for reliability (Fosang et al. 2003). One limitation of our study was the cutoff values for defining contracture. We chose –10° as the cutoff for all joints to be consistent with the reference values used in the surveillance program. Hip or knee extension or foot dorsiflexion of –10° or less causes functional limitations and affects mechanical alignments. To evaluate the effects of this cutoff, we also ran the statistical analyses using less than 0° as the cutoff and found a similar outcome. For children with GMFCS level III to V, the statistical analyses indicated that the conclusion of this study was insensitive to the choice of cutoff values for contracture, that is, 0° or –10°. It is more difficult to draw a conclusion from the trend for GMFCS level I and II because the fewer contractures at these levels of motor function (Nordmark et al. 2009) make it harder to detect statistically significant differences. Another limitation was the age of the children censored because of surgery. In the CPUP surveillance program, children at risk for hip dislocation have their first soft tissue surgery of the hip early (median age 4 years), which precedes the first foot and knee surgery by 3 and 5 years, respectively. In some cases, hip surgery is likely to occur before confirmed contracture of the hip because of the indication for surgery based solely on lateralization on radiographs (Hägglund et al. 2005). In our study, the operated leg was censored from analyses at the time of the first operation. This may have affected the results, especially for children with GMFCS level IV or V, because many of these children receive their hip operation early and therefore were not included in the analysis.

226

Botulinum toxin A injection is a common treatment for spasticity in children with CP in Sweden (Franzén et al. 2017) and it could be argued that this interfered with our results. However, given the large number of treated children and the fact that the effect of botulinum toxin is not permanent, these children were included. The intervals between examinations in the CPUP are based on GMFCS level and age, from twice a year up to the age of 6 years and then every year (GMFCS II–V) or every second year (GMFCS I) (Figure 1). This indicates that some differences between GMFCS levels may stem from differences in detection, probably because of the different number of examinations. Although this may explain some of the differences in the proportion of contractures, it does not explain the fact that contractures occurred in a different sequence, with dorsiflexion first in children with GMFCS level I or II and knee extension first in children with GMFCS level III to V. Our longitudinal study covers a time period of 24 years and during this period several interventions have changed. Physiotherapy interventions have changed from hands-on sessions to a greater focus on activity and participation. The use of assistive devices and orthoses has increased, and interventions are primarily integrated into the child’s everyday life. There are regional differences in pediatric physiotherapy interventions in Sweden (Degerstedt et al. 2020). More intensive treatment has been associated with better outcome for the children (Storvold et al. 2020). Surgeries are more proactive rather than reactive today and the amount of surgeries has decreased (Hägglund et al. 2005). Our study indicates that contractures in the lower limb occur early. It seems possible to avoid major surgeries by monitoring patients with CP and treating contractures early on with physiotherapy, orthoses, and botulinum toxin (Hägglund 2005, Novak et al. 2013). Knowledge regarding the occurrence of the first contracture is important to be able to provide early treatment. An ankle–foot orthosis (AFO) is used widely to facilitate dorsiflexion and to improve walking ability, gait pattern, or foot alignment when standing. In Sweden, 50% of children with CP use an AFO, and the use of AFO starts early and increases with age up to 5 years (Wingstrand et al. 2014). A similar trend has been reported for the development of spasticity of the gastrosoleus muscle, which peaks at age 5–6 years in children with CP (Lindén et al. 2019). Knee orthoses can also be used to increase ROM, but this has not been evaluated widely (Laessker-Alkema and Eek 2016). Most children with CP in Sweden who require a standing device use individually molded hip–knee–ankle–foot orthoses for mechanical alignment of all the segments of the lower extremities. Botulinum toxin is frequently used in children with CP in Sweden (Franzén et al. 2017). It is most frequently injected into the gastrocnemius muscle to improve gait and increase ROM. The treatment is usually followed by use of orthosis or in some cases serial casting. Botulinum toxin is more fre-

Acta Orthopaedica 2021; 92 (2): 222–227

quently used in younger children (Franzén et al. 2017) and its use corresponds to the development of spasticity that increases over the first 5 years of life and then decreases after 6 years of age (Lindén et al. 2019) and GMFCS level I or II. Our results are consistent with this observation that foot contracture occurred first in children with GMFCS levels I–II. Botulinum toxin injection into the hamstring muscles is most common in older children with GMFCS level IV or V (Franzén et al. 2017). Our study showed that knee contracture occurs first in children with GMFCS level IV or V but that contracture occurs at an early age. It is reasonable to think that a contracture in a joint affects the adjacent joints, which are then at risk for developing contracture. With a knee joint contracture, the hip will not be extended and flexion contracture in connection with abduction or adduction is likely to occur. Knowledge concerning the sequence of the development of contracture in children at different GMFCS levels is crucial for treating the joints early on and preventing the sequence of contractures following the first one. In conclusion, lower limb contracture occurs first in the foot of children with GMFCS level I and II and in the knee in children with GMFCS level III to V. Early interventions to prevent knee and foot contractures in children with CP should be considered. Abbreviations AFO: ankle–foot orthosis; CP: cerebral palsy; CPUP: Cerebral Palsy Follow-up Program (Sweden); GMFCS: Gross Motor Function Classification System; ITB: intrathecal baclofen pump; KM: Kaplan–Meier; ROM: range of motion; SDR: selective dorsal rhizotomy. Study design: EC, PW, HLP, ERB. Data collection: EC, ERB. Data analysis: EC, PW, ERB. Manuscript preparation: EC, PW, HLP, ERB. The authors thank Anna Lindgren for statistical support. Acta thanks Marek Jozwiak and Wade M Shrader for help with peer review of this study. Agustsson A, Sveinsson T, Rodby-Bousquet E. The effect of asymmetrical limited hip flexion on seating posture, scoliosis and windswept hip distortion. Res Dev Disabil 2017; 71: 18-23. doi: 10.1016/j.ridd.2017.09.019. Agustsson A, Sveinsson T, Pope P, Rodby-Bousquet E. Preferred posture in lying and its association with scoliosis and windswept hips in adults with cerebral palsy. Disabil Rehabil 2018: 1-5. doi: 10.1080/09638288.2018.1492032. Alriksson-Schmidt A, Hägglund G. Pain in children and adolescents with cerebral palsy: a population-based registry study. Acta Paediatr 2016; 105(6): 665-70. doi: 10.1111/apa.13368. Alriksson-Schmidt A I, Arner M, Westbom L, Krumlinde-Sundholm L, Nordmark E, Rodby-Bousquet E, Hägglund G. A combined surveillance program and quality register improves management of childhood disability. Disabil Rehabil 2017; 39(8): 830-6. doi: 10.3109/09638288.2016.1161843.

Acta Orthopaedica 2021; 92 (2): 222–227 227

Barrett R S, Lichtwark G A. Gross muscle morphology and structure in spastic cerebral palsy: a systematic review. Dev Med Child Neurol 2010; 52(9): 794-804. doi: 10.1111/j.1469-8749.2010.03686.x. Blackman J A, Svensson C I, Marchand S. Pathophysiology of chronic pain in cerebral palsy: implications for pharmacological treatment and research. Dev Med Child Neurol 2018; 60(9): 861-5. doi: 10.1111/dmcn.13930. Chan G, Miller F. Assessment and treatment of children with cerebral palsy. Orthop Clin North Am 2014; 45(3): 313-25. doi: 10.1016/j.ocl.2014.03.003. Cloodt E, Rosenblad A, Rodby-Bousquet E. Demographic and modifiable factors associated with knee contracture in children with cerebral palsy. Dev Med Child Neurol 2018; 60(4): 391-6. doi: 10.1111/dmcn.13659. Degerstedt F, Enberg B, Keisu B I, Björklund M. Inequity in physiotherapeutic interventions for children with cerebral palsy in Sweden: a national registry study. Acta Paediatr 2020; 109(4): 774-82. doi: 10.1111/apa.14980. Fosang A L, Galea M P, McCoy A T, Reddihough D S, Story I. Measures of muscle and joint performance in the lower limb of children with cerebral palsy. Dev Med Child Neurol 2003; 45(10): 664-70. doi: 10.1017/ s0012162203001245. Franzén M, Hägglund G, Alriksson-Schmidt A. Treatment with Botulinum toxin A in a total population of children with cerebral palsy: a retrospective cohort registry study. BMC Musculoskelet Disord 2017; 18(1): 520. doi: 10.1186/s12891-017-1880-y. Hägglund G, Andersson S, Duppe H, Lauge-Pedersen H, Nordmark E, Westbom L. Prevention of severe contractures might replace multilevel surgery in cerebral palsy: results of a population-based health care programme and new techniques to reduce spasticity. J Pediatr Orthop B 2005; 14(4): 269-73. Hancock G E, Hepworth T, Wembridge K. Accuracy and reliability of knee goniometry methods. J Exp Orthop 2018; 5(1): 46. doi: 10.1186/s40634018-0161-5. Kim D H, An D H, Yoo W G. Validity and reliability of ankle dorsiflexion measures in children with cerebral palsy. J Back Musculoskelet Rehabil 2018; 31(3): 465-8. doi: 10.3233/BMR-170862. Laessker-Alkema K, Eek M N. Effect of knee orthoses on hamstring contracture in children with cerebral palsy: multiple single-subject study. Pediatr Phys Ther 2016; 28(3): 347-53. doi: 10.1097/PEP.0000000000000267. Lindén O, Hägglund G, Rodby-Bousquet E, Wagner P. The development of spasticity with age in 4,162 children with cerebral palsy: a registerbased prospective cohort study. Acta Orthop 2019; 90(3): 286-91. doi: 10.1080/17453674.2019.1590769.

McGinley J L, Dobson F, Ganeshalingam R, Shore B J, Rutz E, Graham H K. Single-event multilevel surgery for children with cerebral palsy: a systematic review. Dev Med Child Neurol 2012; 54(2): 117-28. doi: 10.1111/j.1469-8749.2011.04143.x. Nordmark E, Hägglund G, Lauge-Pedersen H, Wagner P, Westbom L. Development of lower limb range of motion from early childhood to adolescence in cerebral palsy: a population-based study. BMC Med 2009; 7: 65. doi: 10.1186/1741-7015-7-65. Novak I, McIntyre S, Morgan C, Campbell L, Dark L, Morton N, Stumbles E, Wilson S A, Goldsmith S. A systematic review of interventions for children with cerebral palsy: state of the evidence. Dev Med Child Neurol 2013; 55(10): 885-910. doi: 10.1111/dmcn.12246. Oh E J, Shepherd B E, Lumley T, Shaw P A. Considerations for analysis of time-to-event outcomes measured with error: bias and correction with SIMEX. Stat Med 2018; 37(8): 1276-89. doi: 10.1002/sim.7554. Palisano R J, Rosenbaum P, Bartlett D, Livingston M H. Content validity of the expanded and revised Gross Motor Function Classification System. Dev Med Child Neurol 2008; 50(10): 744-50. doi: 10.1111/j.14698749.2008.03089.x. Pettersson K, Wagner P, Rodby-Bousquet E. Development of a risk score for scoliosis in children with cerebral palsy. Acta Orthop 2020: 91(2): 203-8 doi: 10.1080/17453674.2020.1711621. Raja K, Joseph B, Benjamin S, Minocha V, Rana B. Physiological cost index in cerebral palsy: its role in evaluating the efficiency of ambulation. J Pediatr Orthop 2007; 27(2): 130-6. doi: 10.1097/01.bpb.0000242440.96434.26. Rosenbaum P, Paneth N, Leviton A, Goldstein M, Bax M, Damiano D, Dan B, Jacobsson B. A report: the definition and classification of cerebral palsy April 2006. Dev Med Child Neurol 2007; 109(Suppl.): 8-14. Smith L R, Chambers H G, Lieber R L. Reduced satellite cell population may lead to contractures in children with cerebral palsy. Dev Med Child Neurol 2013; 55(3): 264-70. doi: 10.1111/dmcn.12027. Storvold G V, Jahnsen R B, Evensen K A I, Bratberg G H. Is more frequent physical therapy associated with increased gross motor improvement in children with cerebral palsy? A national prospective cohort study. Disabil Rehabil 2020; 42(10): 1430-8. doi: 10.1080/09638288.2018.1528635. Wingstrand M, Hägglund G, Rodby-Bousquet E. Ankle–foot orthoses in children with cerebral palsy: a cross sectional population based study of 2200 children. BMC Musculoskelet Disord 2014; 15: 327. doi: 10.1186/14712474-15-327.

228

Acta Orthopaedica 2021; 92 (2): 228–234

Risk factors for implant-related fractures after proximal femoral osteotomy in children with developmental dysplasia of the hip: a case-control study Jing DING a, Zhen-Zhen DAI a, Zhu LIU , Zhen-Kai WU , Zi-Ming ZHANG , and Hai LI

Department of Pediatric Orthopedics, Xin Hua Hospital affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China a Shared first authorship Correspondence: lihai@xinhuamed.com.cn Submitted 2020-02-02. Accepted 2020-10-19.

Background and purpose — Proximal femoral osteotomy (PFO) is commonly performed to treat children with developmental dysplasia of the hip (DDH). Implant-related femoral fractures after osteotomy are sometimes reported, but the potential risk factors for these fractures remain unclear. We investigated the association of implant-related fractures with PFO and potential risk factors for these fractures. Patients and methods — We retrospectively reviewed 1,385 children undergoing PFO for DDH in our institution from 2009 to 2016 after obtaining institutional review board (IRB) approval and identified 27 children (28 hips, fracture group) with implant-related femoral fractures after PFO. We selected 137 children (218 hips, control group) without fractures who matched the children in the fracture group by age, weight, surgeon, and surgical period. Relevant clinical data were collected and compared between the 2 groups. Multiple analyses of risk factors for implant-related fractures were conducted by logistic regression with the stepwise regression method. Results — The occurrence rate of implant-related fractures was 1.9% (27/1,385). Compared with the control group, the fracture group more commonly exhibited bilateral involvement (74% vs. 53%, p = 0.04), used a spica orthosis for immobilization after osteotomy (43% vs 21%, p = 0.01) and exhibited mild remodeling at the osteotomy site (46% vs. 19%, p = 0.003), and less commonly required capsulotomy during osteotomy (61% vs. 79%, p = 0.03). According to the multiple regression analysis, the only factor identified as an independent risk factor for implant-related fractures was mild remodeling at the osteotomy site (OR = 3.2, 95% CI 1.4–7.5). Remodeling at the osteotomy site was significantly associated with varus osteotomy (coefficient = 1.4, CI 1.03– 1.8). The fracture occurred at a mean of 12 months (2.2–25) after osteotomy or 3.3 months (0–12) after implant removal. In children undergoing implant removal, the fractures mostly

occurred at the osteotomy site (n = 13/15), while in those with the implant remaining, the fractures mostly occurred in the screw hole (n = 8/13). Interpretation — The type of PFO performed is not associated with implant-related fractures in children with DDH. Children with mild remodeling at the osteotomy site should be closely followed up, regardless of whether the hardware is removed, and high-intensity activity should not be permitted until moderate or extensive remodeling is confirmed. After PFO, the implants should be removed when solid union is achieved at the osteotomy site.

Proximal femoral osteotomy (PFO) is commonly performed to correct proximal femoral deformities in individuals with developmental dysplasia of the hip (DDH), and types of PFO include femoral shortening, varus osteotomy, and derotation osteotomy. Internal fixation implants, such as a blade plate or locking compress plate, are used to maintain the stability of the osteotomy site (Papavasiliou and Papavasiliou 2005, Sharpe et al. 2006, Shaw et al. 2016). Implant-related complications or fractures after osteotomy have been reported, with a prevalence rate of 0.3% to 3.6% (Jain et al. 2012, 2016). Although the rate is low, these complications or fractures prolong immobilization in children and sometimes require additional surgery. Few studies have investigated relations between implant-related fractures and sites of fractures or types of plates (Becker et al. 2012, Jain et al. 2012, 2016, Chung et al. 2018). Jain et al. (2012) reported that the femur is more likely to incur an implant-related fracture than are other bones, and suggested that the level of stress exerted by an implant can be high over short anatomic distances in the proximal femur. However, the authors did not stratify the results by the indications for PFO, such as DDH, Perthes disease, and cerebral palsy. Varus in

Acta Orthopaedica 2021; 92 (2): 228–234

229

All of the children treated for DDH from 2009 to 2016 n = 1,385 Child with implantrelated fracture

Child without implantrelated fracture

Inclusion criteria

Matched on age, weight, surgeon, and the period: 1:5 Fracture group n = 27

Control group n = 125

1-year (2012) patient group n = 68

Figure 1. Flow chart of patients reviewed and selected.

PFO may increase the level of stress on the implant, which in turn leads to stress shielding at the osteotomy site; therefore, it is presumable that changes in both the anatomy of the proximal femur and stress on the implant and osteotomy site may increase the probability of implant-related fractures. However, the relation of PFO itself to implant-related fractures after DDH has not been clarified, and the potential risk factors for implant-related fractures remain unclear. Therefore, we investigated the association of implant-related fractures with PFO and possible risk factors for these fractures.

Patients and methods We retrospectively reviewed all 1,385 children who were younger than 14 years old, did not have pathological or metabolic diseases or cerebral palsy, and had undergone PFO at our hospital from 2009 to 2016. From this initial patient population, we identified every child who had sustained an implantrelated fracture. For each of these children, we selected 5 children who did not have a fracture but were matched in terms of the surgeon who operated on him or her, age (difference < 6 months), weight (difference < 2 kg), and duration since osteotomy (within 1 month) to form a control group. Furthermore, to determine whether the selection of the control group was biased, we collected the demographic data of the children included in the initial patient population (2009–2016) over the period of 1 whole year (January 2012 to December 2012) and compared the data with those of the control group (Figure 1). The inclusion criteria of the fracture group were as follows: (a) a femoral fracture adjacent to the osteotomy or implant site; (b) a fracture that occurred within 2 years after PFO, regardless of whether the implant was removed; (c) the absence of a history of severe trauma, such as a fall from a height or car accident; and (d) complete medical data spanning a follow-up period of more than 2 years after PFO. The inclusion criteria of the control group were (a) the absence of a femoral fracture within 2 years after PFO and (b)

Figure 2. Immobilization types: A = spica cast. B = spica orthosis.

complete medical data spanning a follow-up period of more than 2 years after PFO. In our hospital, PFO is an elective procedure performed in combination with pelvic osteotomy in children older than 18 months old with DDH by the senior surgeon according to the method described by Weinstein and Flynn (2014). During PFO, we usually perform femoral derotation and/or varus or shortening osteotomy. Basically, we decrease the anteversion angle to no less than 30°, decrease the neck–shaft angle to no less than 120°, and shorten the femur for complete dislocation. Moreover, capsulotomy is usually performed in children with Tönnis grade III/IV and sometimes those with Tönnis grade II to enable reduction and capsulorrhaphy. The extent of dislocation of the femoral head, neck–shaft angle, and anteversion angle in the proximal femur are evaluated by preoperative radiography and CT. Postoperative immobilization by a spica cast or spica orthosis (Figure 2) is continued for 6–8 weeks, and active nonweight-bearing exercises throughout a range of motion are performed for approximately 3–4 weeks. Thereafter, gradual weight-bearing is permitted when union of the osteotomy site is confirmed by radiography. The implant is removed routinely at approximately 6–12 months after PFO when solid union has occurred. For children with bilateral DDH, we usually perform osteotomy on each side with an interval of 4–6 months between surgeries. We collected the following demographic and clinical data: age, sex, weight, side (unilateral or bilateral), severity of DDH (dysplasia or dislocation), degree of derotation or varus, length of shortening in PFO, types of implant, whether capsulotomy existed, implant removal status, remodeling condition at the osteotomy site, time to implant removal, follow-up time, fracture site (osteotomy site, screw hole or others), and time from osteotomy to fracture or implant removal. The time at which the fracture occurred in the fracture group was set as the endpoint of the follow-up period for the children

230

Acta Orthopaedica 2021; 92 (2): 228–234

Table 1. Patient characteristics stratified by occurrence of implantrelated fractures. Values are count (%) or mean (range) All subjects Characteristics (N = 152)

Fracture (n = 27)

Non-fracture (n = 125) p-value

Female sex 129 (82) 22 (81) 107 (82) 0.6 Age, year 4.6 (2–14) 5.2 (2–14) 4.5 (2–13) 0.3 Weight, kg 19.6 (8–66) 21.4 (9.5–66) 19.2 (9.4–54) 0.4 Side 0.04 Unilateral 66 (43) 7 59 (47) Bilateral 86 (57) 20 66 (53)

Figure 3. Evaluation of the remodeling condition at the osteotomy site. A, B: Proportion of residual trace line (red arrow) located at the osteotomy site to diameter of the medullary cavity (red dotted line, vertical line represents the center point of the diameter of medullary cavity) was less than ½, so the remodeling condition of the osteotomy site was defined as moderate or extensive. C, D: Proportion of residual trace line (red arrow) located at the osteotomy site to diameter of the medullary cavity (red dotted line, vertical line represents the center point of the diameter of medullary cavity) was more than ½, so the remodeling condition of the osteotomy site was defined as mild.

in the control group. Whether the implant was still in situ was also determined at the endpoint. The remodeling condition at the osteotomy site was also evaluated at the endpoint. According to the methods described by Davids et al. (2013), for the convenience of statistical analysis, we classified remodeling radiographically as the proportion of the residual trace line located at the osteotomy site to the diameter of the medullary cavity (Figure 3). If the proportion was smaller than 1/2, a moderate or extensive remodeling condition existed at the osteotomy site; if the proportion was larger than 1/2, mild remodeling existed.

Statistics The categorical variables were assessed by chi-square test and Fisher’s exact test, and the continuous variables were computed by t-tests. Possible relations between all kinds of factors, such as demographics (age, sex, weight), diseases (severity of DDH, side), surgical factors (degree of derotation or varus, length of shortening in PFO, types of implant, whether a capsulotomy existed) and postoperative factors (immobilization, remodeling condition at the osteotomy site, implant removal status), were assessed by Spearman rank correlation test and logistic regression. Multiple analyses of risk factors for implant-related fractures were evaluated by logistic regression, and odds ratios (ORs) with their 95% confidence intervals (CIs) were also obtained. The potential risk factors included in the multiple analyses were the type of PFO, factors considered in previous studies (types of implant, implant removal status) (Jain et al. 2012, Chung et al. 2018) and the positive factors identified in the crude analysis that were assumed to have a cause–effect relation. The stepwise regression method was used to find the most appropriate logistical model. Statistical analyses were carried out with the statistical software Stata/SE for Windows (version 15.0; StataCorp LLC, College Station, TX, USA), and all statistical tests were 2-tailed; p-values < 0.05 were considered significant. Ethics, funding, and potential conflicts of interest The study was conducted according to the ethical principles stated in the Declaration of Helsinki. The study received approval from the Institutional Review Board/Ethics Committee of Xin Hua Hospital (reference number: XHEC-D2019-011). This work was supported by the Shanghai Collaborative Innovation Center for Translational Medicine (grant TM201712) and the Clinical Research Unit, Xin Hua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine (grant 16CR3100B). No conflicts of interest were declared by the authors.

Results We identified 27 children (28 hips) with fractures (mean age 5.2 years [2–14]), accounting for a fracture rate of 1.9%

Acta Orthopaedica 2021; 92 (2): 228–234

231

Table 2. Comparison of patient characteristics between control group and 1-year patient group. Values are count (%) or mean (range)

Months between osteotomies

Mean age

12 p = 0.6

Characteristics

Control 1-year patient group group (n = 125) (n = 68) p-value

Female sex 107 (82) 53 (78) 0.2 Age, year 4.5 (2–13) 4.6 (2–14) 0.8 Weight, kg 19.2 (9.4–54) 18.0 (8–50) 0.3 Side 0.1 Unilateral 59 (47) 24 (35) Bilateral 66 (53) 44 (65) Dislocation 86 (69) 52 (76) 0.3

All subjects (N = 218)

Fracture (n = 28)

a 1 means 5°. b In children who

had the implants removed.

(27/1,385). The 125 children (190 hips) in the control group had a mean age of 4.5 years (2–13). The 2 groups were similar in terms of age, sex, and weight (Table 1). The children in the 1-year group selected from the initial study population and children in the control group were similar in terms of sex, age, weight, side, and dislocation (Table 2). Compared with the control group, the fracture group more commonly had bilateral involvement (74% vs. 53%, p = 0.04, Table 1), used a spica orthosis for immobilization after osteotomy (43% vs. 21%, p = 0.01), and exhibited mild remodeling at the osteotomy site (46% vs. 19%, p = 0.003) but less commonly required capsulotomy during osteotomy (61% vs. 79%, p = 0.03). However, no significant differences between

Control group

Fracture group

Spica cast

Spica orthosis

Months osteotomy–fracture

p = 0.4

15 p = 0.6

p = 0.2

10 5

p < 0.0001

Age stratified fracture rate (%)

Non-fracture (n = 190) p-value

Severity of DDH 1.0 Dysplasia 69 (32) 9 60 (32) Dislocation 149 (68) 19 130 (68) Derotation a 3.3 (0–8) 3.6 (0–6) 3.2 (0–8) 0.2 Varus a 4.5 (0–8) 4.8 (1–7) 4.4 (0–8) 0.2 Shortening, cm 1.3 (0–5) 1.3 (0–4) 1.3 (0–5) 1.0 Hardware 0.8 Locking plate 167 (77) 21 146 (77) Blade plate 51 (23) 7 44 (23) Immobilization 0.01 Spica cast 166 (76) 16 150 (79) Spica orthosis 52 (24) 12 40 (21) Capsulotomy 168 (77) 17 151 (79) 0.03 Implant removed 0.4 No 117 (54) 13 104 (55) Yes 101 (46) 15 86 (45) Remodeling condition at the osteotomy site 0.003 Moderate or extensive 169 (78) 15 154 (81) Mild 49 (22) 13 36 (19) Months to implant removal b 10 (5–21) 11 (7–21) 10 (5–20) 0.4

Table 3. Clinical characteristics of the hip stratified by occurrence of implant-related fractures. Values are count (%) or mean (range) Characteristics

≤4

≤7

Age groups

Spica cast

Spica orthosis

Figure 4. A: Comparison of interval from separate osteotomy surgery in the children with bilateral DDH between fracture and control groups (6.6 vs. 7 months, p = 0.6). B: Comparison of age of children by immobilization type (4 vs. 6.4 years old, p < 0.0001). C: Fracture rate according to age group (12% vs. 14%, p = 0.6; 11.8% vs. 19%, p = 0.2). D: Comparison of time from osteotomy to fracture between children with different immobilization types (11.4 [SD 5.2] vs. 13.2 [SD 6.4] months, p = 0.4).

Table 4. Multiple analyses of factors related to immobilization type by logistic regression Related factors Age Side Weight Dislocation

Coefficient (95% CI) 1.4 (1.1–1.7) 0.7 (0.4–1.6) 1.0 (1.0–1.1) 1.6 (0.7–3.6)

CI = conﬁdence interval.

the two groups were found in other clinical factors, such as dislocation, derotation, varus, shortening osteotomy in the proximal femur, implant type, implant side, or time to implant removal (Table 3) For the children with bilateral involvement, the time to the separate osteotomy surgery did not statistically significantly differ between the children with a fracture and those without a fracture (Figure 4A). We also found that age was significantly related to the immobilization type selected (coefficient = 1.4, Table 4), and the children using a spica orthosis were younger than those using a spica cast (average age, 4 vs. 7 years old, Figure 4B). The rate of implant-related fractures was similar between the 4- and 7-year-old groups (Figure 4C). There were

232

Acta Orthopaedica 2021; 92 (2): 228–234

Figure 6. A: A 2-year-old girl with right DDH. She underwent PFO, open reduction, and pelvic osteotomy. B: Image taken at 5 months postoperatively. C, D: She had an implant-related fracture in the screw hole at 10.4 months postoperatively when she was walking and was immobilized with an orthotic. Figure 5. A: A 4-year-old girl with DDH. She underwent PFO, open reduction, and pelvic osteotomy. B: Image taken at 7.5 months postoperatively. C: Implant was removed at 13 months postoperatively. D: She had an implant-related fracture at the osteotomy site at 3.5 months after implant removal when she was walking.

no differences in the time from osteotomy to fracture between the children immobilized by an orthosis and those immobilized by a cast (Figure 4D). Capsulotomy was significantly related to dislocation (coefficient = 2.4), age (coefficient = 0.6), and shortening osteotomy (coefficient = 6.4, Table 5). Varus osteotomy was found to be an independent factor for the remodeling condition at the osteotomy site (coefficient = 1.4, Table 6). We finally incorporated the following factors into the multifactorial analysis of implant-related fractures (Table 4): age, side, severity of DDH (dysplasia or dislocation), degree of derotation or varus, length of shortening in PFO, types of implant, implant removal status, and remodeling condition at the osteotomy site. According to the stepwise regression results, the factor identified as an independent risk factor for implant-related fractures was mild remodeling at the osteotomy site (OR = 3.2, Table 7). For the children in the fracture group, the average time from osteotomy to fracture was 12 months (2.2–25 months). It was 9.6 months (2.2–24 months) for the children with implants remaining and 15 months (8.7–25) for the children who

underwent implant removal. The average time from implant removal to fracture was 3.3 months (0–12 months) (Table 8). A fracture occurred at the osteotomy site in 16/28 children. The distribution of fracture sites significantly differed between the children who did and did not undergo hardware removal (p = 0.001, Table 8). In the children who underwent implant removal, the fractures mostly occurred at the osteotomy site (13/15) (case shown in Figure 5), while in those who still had the implant the fractures mostly occurred in the screw hole (8/13, Table 8) (case shown in Figure 6).

Discussion We did not find any relations between the type of PFO and implant-related fractures in children with DDH. Only mild remodeling at the osteotomy site was identified as an independent risk factor for these fractures. However, varus osteotomy was found to be related to the remodeling condition. The distribution of fracture sites differed between the children who did and did not undergo hardware removal. No clear definitions of implant-related fractures exist; generally, these fractures include peri-implant fractures that occur within 6 months after hardware removal without trauma (Busam et al. 2006, Chung et al. 2018). The occurrence rate

Acta Orthopaedica 2021; 92 (2): 228–234

233

stress in the proximal femur or the implant. Our study showed that mild remodeling at the osteotomy site may be a risk factor for implantrelated fractures. PFO is a kind of “end-to-end” technique with a transverse osteotomy at the subtrochanteric level, after which the proximal segment is abducted to some degree (varus osteotomy) and aligned with the distal shaft segment (Davids et al. 2013). After the union of the osteotomy site, the medial cortex of the proximal segment becomes the residual line in the medullary cavity, as seen in anterior–posterior radiographs, which gradually disappears through remodeling within 2–3 years, according to our observations (Figure 7). The process of bone remodeling follows Wolff’s Law but may be affected by the alignment of the osteotomy site. We found severe varus osteotomy to be related to the remodeling condition at the osteotomy site (Table 6). We think that remodeling at the Figure 7. Remodeling process at the osteotomy site in a 4-year-old girl with PFO. A: End-to-end align- osteotomy site may be affected not ment at the osteotomy site immediately after operation (red arrow). B: mild remodeling at the osteotomy only by the alignment of the ostesite (union already) with a distinct residual line (red arrow) from medial cortex of proximal segment in the medullar cavity (6 months after PFO). C: mild remodeling with obscure residual line (red arrow) (15 otomy site during surgery but also months after PFO with hardware removal). D: moderate remodeling at the osteotomy site with obscure by subsequent bone healing and and shorter residual line (red arrow) in the medullar cavity (27 months after PFO). E: extensive remodremodeling. Therefore, we do not eling with obscure and short residual line in the medullar cavity (red dotted box) (33 months after PFO). think that the condition of remodeling at the osteotomy site can be of implant-related fractures has been reported to be 0.3–3.6% used to measure the extent of varus osteotomy performed. by other authors (Jain et al. 2016, Chung et al. 2018) and However, how the mechanical characteristics of the proximal was 1.9% (27/1,385) in our study. The rate varied by loca- femur change with remodeling conditions remains unclear. tion and disease. Jain et al. (2012) reported that the femur is Whether mild remodeling at the osteotomy site weakens the more likely to incur an implant-related fracture than are other mechanical strength of the proximal femur, making it susbones and suggested that the level of stress shielding exerted ceptible to fracture, needs to be researched further. We also by the implant can be high in the proximal femur. However, suggest that children with mild remodeling at the osteotomy we did not find varus, derotation, or shortening osteotomy in site, regardless of whether the hardware is removed, should PFO to be associated with implant-related fractures. Although be followed up closely and that high-intensity activity should locking plates are thought to reduce the level of stress at the not be permitted until moderate or extensive remodeling at the osteotomy site (Bottlang et al. 2010, Becker et al. 2012), nei- osteotomy site is confirmed. ther the occurrence of an implant-related fracture nor the fracIn our institution, implant removal after PFO in children ture site was found to be related to plate type in our study. with DDH is performed routinely because Chinese parents Children with DDH always have a larger neck–shaft angle or usually prefer not to leave any metal implants in their child’s anteversion angle in the proximal femur, especially those with body. In addition, implant removal facilitates future surgeries, complete dislocation. Therefore, in children with DDH, PFO if needed (Jain et al. 2012, 2016). In our study, whether the is performed to restore the anatomy of the proximal femur to hardware was retained or removed was not identified as a risk a relatively normal state and, in theory, should not increase the factor for implant-related fractures. However, it was found to

234

be related to the location of the fracture site (Table 8), which can be explained as “stress shielding,” as suggested by previous research (Lovell et al. 1999, Hanson et al. 2008, Jain et al. 2012). Therefore, the implants should be removed after PFO when solid union is observed at the osteotomy site. Our study has several limitations. First, because of its retrospective nature, there may have been discrepancies in the characteristics of the patients between groups (Chung et al. 2018). The case-control design could have decreased the effects of confounders, including age, sex, surgeon, and surgical period. Moreover, the children in the control group did not differ in age, weight, side, or dislocation from those in the one-year group, indicating that the control group is a representative sample of the initial study population. Second, we cannot determine whether clinical factors other than those we recorded, such as the postoperative range of motion in the hip, are related to fractures after osteotomy. In conclusion, PFO is not associated with implant-related fractures in children with DDH. We suggest that children with mild remodeling at the osteotomy site are followed up closely, regardless of whether the hardware is removed, and that high-intensity activity is not permitted until moderate or extensive remodeling at the osteotomy site is confirmed. After PFO, the implants should be removed when solid union at the osteotomy site occurs. Supplementary data Tables 5–8 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.20 20.1848315 All authors helped design the study, interpreted and analyzed the data, and reviewed the manuscript. Li H and Ding J designed the study and drafted the manuscript. Dai ZZ, Liu Z, Zhang ZM, and Wu ZK reviewed the patient records and analyzed the data, and Li H and Dai ZZ wrote the manuscript. All authors read and approved the final manuscript. Acta thanks Martin Gottliebsen and Aare Märtson for help with peer review of this study.

Acta Orthopaedica 2021; 92 (2): 228–234

Becker T, Weigl D, Mercado E, Katz K, Bar-On E. Fractures and refractures after femoral locking compression plate fixation in children and adolescents. J Pediatr Orthop 2012; 32(7): e40-6. doi: 10.1097/ BPO.0b013e318264496a. Bottlang M, Doornink J, Lujan T J, Fitzpatrick D C, Marsh J L, Augat P, von Rechenberg B, Lesser M, Madey S M. Effects of construct stiffness on healing of fractures stabilized with locking plates. J Bone Joint Surg Am 2010; 92(Suppl. 2): 12-22. doi: 10.2106/jbjs.J.00780. Busam M L, Esther R J, Obremskey W T. Hardware removal: indications and expectations. J Am Acad Orthop Surg 2006; 14(2): 113-20. Chung M K, Kwon S S, Cho B C, Lee G W, Kim J, Moon S J, Lee J W, Chung C Y, Sung K H, Lee K M, Park M S. Incidence and risk factors of hardware-related complications after proximal femoral osteotomy in children and adolescents. J Pediatr Orthop B 2018; 27(3): 264-70. doi: 10.1097/ BPB.0000000000000448. Davids J R, Gibson T W, Pugh LI, Hardin JW. Proximal femoral geometry before and after varus rotational osteotomy in children with cerebral palsy and neuromuscular hip dysplasia. J Pediatr Orthop 2013; 33(2): 182-9. doi: 10.1097/BPO.0b013e318274541a. Hanson B, van der Werken C, Stengel D. Surgeons’ beliefs and perceptions about removal of orthopaedic implants. BMC Musculoskelet Disord 2008; 9: 73. doi: Artn 7310.1186/1471-2474-9-73. Jain A, Erkula G, Leet A I, Ain M C, Sponseller P D. Implant-related fractures in children: a 15-year review. J Pediatr Orthop 2012; 32(5): 547-52. doi: 10.1097/BPO.0b013e318259fe75. Jain A, Thompson J M, Brooks J T, Ain M C, Sponseller P D. Implantrelated fractures in children with proximal femoral osteotomy: blade plate versus screw-side plate constructs. J Pediatr Orthop 2016; 36(1): e1-5. doi: 10.1097/BPO.0000000000000481. Lovell M E, Galasko C S, Wright N B. Removal of orthopedic implants in children: morbidity and postoperative radiologic changes. J Pediatr Orthop B 1999; 8(2): 144-6. Papavasiliou V A, Papavasiliou A V. Surgical treatment of developmental dysplasia of the hip in the periadolescent period. J Orthop Sci 2005; 10(1): 15-21. doi: 10.1007/s00776-004-0850-z. Sharpe P, Mulpuri K, Chan A, Cundy PJ. Differences in risk factors between early and late diagnosed developmental dysplasia of the hip. Arch Dis Child Fetal Neonatal Ed 2006; 91(3): F158-F162. doi: 10.1136/adc.2004.070870. Shaw B A, Segal L S, Section On Orthopaedics. Evaluation and referral for developmental dysplasia of the hip in infants. Pediatrics 2016; 138(6): e20163107. doi: 10.1542/peds.2016-3107. Weinstein S L, Flynn J M. Developmental Hip dysplasia and dislocation. Lovell and Winter’s pediatric orthopaedics. 7th ed. Philadelphia: Lippincott Williams & Wilkins 2014. pp: 1024-32.

Acta Orthopaedica 2021; 92 (2): 235–239 235

Most surgeons still prefer to reduce overriding distal radius fractures in children Topi LAAKSONEN 1, Jani PUHAKKA 2, Jussi KOSOLA 2, Antti STENROOS 2, Matti AHONEN 1, and Yrjänä NIETOSVAARA 1 1 Department

of Pediatric Orthopedics and Traumatology, Helsinki New Children’s Hospital; 2 Department of Orthopedics and Traumatology, Töölö Hospital, Helsinki University Hospital, Finland Correspondence: topi.laaksonen@hus.fi Submitted 2020-08-11. Accepted 2020-10-29

Background and purpose — Traditionally, overriding distal radius fractures in children have been reduced and immobilized with a cast or treated with percutaneous pin fixation. There is recent evidence that these fractures heal well if immobilized in the bayonet position without reduction. We evaluated the present treatment of these fractures. Methods — A questionnaire including AP and lateral radiographs of overriding distal radius fractures in 3 prepubertal children was answered by 213 surgeons from 28 countries. The surgeons were asked to choose their preferred method of treatment (no reduction, reduction, reduction and osteosynthesis), type and length of cast immobilization, and the number of clinical and radiographic follow-ups. Results — Of the 213 participating surgeons, 176 (83%) would have reduced all 3 presented fractures, whereas 4 (2%) would have treated all 3 children with cast immobilization without reduction. Most reductions (77%) would have been done under general anesthesia. Over half (54%) of the surgeons who preferred anesthesia would have fixed (pins 99%, plate 1%) the fractures. An above-elbow splint or circular cast was chosen in 84% of responses, and the most popular (44%) length of immobilization was 4 weeks. Surgeons from the Nordic countries were more eager to fix the fractures (54% vs. 31%, p < 0.001) and preferred shorter immobilization and follow-up times and less frequent clinical and radiological follow-ups compared with their colleagues from the USA. Interpretation — Most of the participating surgeons prefer to reduce overriding distal radius fractures in children under anesthesia. There is substantial lack of agreement on the indications for osteosynthesis, type of cast, length of immobilization, and follow-up protocol.

Most authors recommend reduction of overriding distal radius fractures in children (McLauchlan et al. 2002, Miller et al. 2005, Zamzam and Khoshhal 2005, Wendling-Keim et al. 2014). Routine percutaneous pin fixation has also been advocated because these fractures have a high risk of re-displacement after reduction (McLauchlan et al. 2002, Zamzam and Khoshhal 2005, Alemdaroğlu et al. 2008, Hang et al. 2011). On the other hand, Do et al. (2003) and Crawford et al. (2012) have reported good results after cast immobilization without fracture reduction regardless of fracture displacement. We assessed current treatment preferences in overriding distal radius fractures in pre-pubertal children. The secondary aim was to record the proposed types of cast, the length of immobilization, and the number of clinical and radiological follow-ups from different institutions.

Methods This survey was designed to assess current opinions and practices in the treatment of overriding distal metaphyseal fractures of the radius in less than 10-year-old children. Participation was voluntary, and no compensation was given. The SurveyMonkey™ (San Mateo, CA, USA) website served as a platform to collect and store responses. 3 otherwise healthy, aged in accordance with study criterion, children’s overriding (complete displacement and shortening) distal metaphyseal radius fractures (Figures 1–3) with different types of distal ulnar fractures were presented. Age, mechanism of injury, and both anteroposterior (AP) and lateral radiographs were shown. The participants were asked to choose their preferred method of treatment, type and length

236

Figure 1. Case 1: a 5-year-old girl who fell of a swing and sustained a completely displaced distal radius fracture with shortening and a nondisplaced but slightly angulated fracture of the distal ulna.

Acta Orthopaedica 2021; 92 (2): 235–239

Figure 2. Case 2: a 7-year-old boy who fell off a climbing frame and sustained a completely displaced distal radius fracture with shortening and a subtotally displaced fracture of the distal ulna with some angulation.

Table 1. Survey questions and reply alternatives Method of treatment Cast immobilization without reduction Alignment adjustment during casting Reduction in local anesthesia Reduction under anesthesia Percutaneous pin fixation Open reduction and pin fixation Plate fixation Type of cast None Dorsal forearm splint Dorsal and volar forearm splint Above-elbow dorsal splint Above-elbow dorsal and volar forearm splint Circular forearm cast Circular above-elbow cast Weeks of immobilization None, 1, 2, 3, 4, 5, 6 or >6 Number of radiological follow-ups None, 1, 2, 3, 4, 5, 6 or >6 clinical follow-ups None, 1, 2, 3, 4, 5, 6 or >6

of cast immobilization, and the number of clinical and radiographic follow-ups (Table 1). Results were analyzed based on treatment method in 3 groups: (1) no reduction, (2) reduction, and (3) reduction and osteosynthesis (Table 2). A query was sent to heads of several pediatric orthopedic departments in Europe (n = 25) and in North America (n = 9) to be circulated to attending surgeons and registrars treating children’s fractures. An additional 48 respondents who claimed to treat children’s fracture in their practice were recruited from the European Pediatric Orthopedic Society (EPOS) meeting in Tel Aviv 2019. Respondents’ experience in the field of pediatric orthopedics and their country of practice were registered anonymously. The survey was completed by 213 surgeons from 21 countries (Table 3). Among the respondents, 110 (52%) had more than 5 years of experience in pedi-

Figure 3. Case 3: a 5-year-old boy who fell off a climbing frame sustaining completely displaced and shortened fractures of both radius and ulna with angulation.

Table 2. Distribution of all 639 responses given by the 213 surgeons participating in the survey Case no. No reduction Cast immobilization only Alignment adjustment during casting Total Reduction with local anesthesia under anesthesia Total Reduction and osteosynthesis Percutaneous pin fixation Open reduction and pin fixation Plate fixation Total

22 10 20 17 7 19 39 17 39

Total 52 43 95

16 6 28 50 93 93 42 228 109 99 70 278 58 86 49 193 6 10 55 71 1 1 0 2 65 97 104 266

Case 1: a non-displaced but slightly angulated fracture of the distal ulna (Figure 1). Case 2: a sub-totally displaced fracture of the distal ulna with some angulation (Figure 2). Case 3: a completely displaced and shortened fracture of the ulna with angulation (Figure 3).

atric orthopedics. There were 94 respondents from the Nordic countries and 50 from the USA, with a similar profile of pediatric orthopedic experience. The response rate concerning the e-mailed queries is unknown, but only 1 of 49 surgeons asked to participate in the survey at the EPOS meeting declined. Statistics The response distribution for each individual question was analyzed, and the agreement between surgeons was determined using Cronbach’s α with a cutoff value of 0.8. Values below 0.5 are considered unacceptable (Cortina 1993). Binary logistic regression analysis was performed to determine which parameter (ulnar fracture type, surgeon’s experience in pedi-

Acta Orthopaedica 2021; 92 (2): 235–239 237

Table 3. Pediatric orthopedic experience and country of the 213 participants Factor Experience < 1 year 1–5 years 6–10 years > 10 years Country Finland USA Sweden Australia Austria UK Estonia Norway Switzerland France Russia Germany Israel Other a

n (%) 47 (22) 56 (26) 34 (16) 76 (36) 73 (34) 50 (24) 14 (7) 9 (4) 7 (3) 7 (3) 6 (3) 6 (3) 5 (2) 5 (2) 5 (2) 4 (2) 4 (2) 18 (9)

Figure 4. A–C: all 3 patients’ fractures were immobilized in bayonet position (A) without reduction with synthetic dorsal above-elbow and volar below-elbow splints applied in finger-trap traction without anesthesia in the emergency department. Splints were removed at 4 weeks. Radiographs at 3.5 years from the injury in case 1 (B) and case 3 (C). Parents of Case 2 (A) did not want any follow-up radiographs taken, because their son had made full functional recovery and he had no pain.

a Fewer than 3 respondents from Armenia, China, Denmark, Japan, Northern Ireland, Mexico, Portugal, and Singapore.

atric orthopedics, country of origin) was of greatest and most independent significance for the prediction of osteosynthesis. Statistical analysis among countries was done only between Nordic countries (Finland, Denmark, Norway, and Sweden) and the USA due to the small number of respondents from other countries. Demographic data were explored using a chisquare test and Pearson’s correlation with a p-value of < 0.05 to define statistical significance. All analyses were performed using SPSS for Windows (IBM Statistics for Windows, Version 22.0, released 2013, IBM Corp, Armonk, NY, USA). Ethics, funding, and potential conflicts of interest The institutional research ethics committee approved the study, and the principle of implied consent was applied, thus formal consent was not required. Study and consent details were clearly communicated before respondents began the questionnaire. The study was supported by Finska Läkaresällskapet (the Medical Society of Finland). There were no potential conflicts of interest.

Results Of the 213 participating surgeons, 176 (83%) would have reduced all the fractures with or without fixation, whereas 4 respondents (2%) would have treated all 3 children with cast immobilization without reduction. General anesthesia would have been performed in 77% of the fracture reductions. Sur-

geons who would have liked to stabilize these fractures opted for pins in 99% of cases. There was no difference between respondents from the Nordic countries and those from the USA concerning the rate of reduction (87% vs. 89%), but surgeons from the Nordic countries were more eager to use pins to fix the fractures compared with their American colleagues (54% vs. 31%, p < 0.001). The majority (84%) of all surgeons chose an above-elbow splint or cast to immobilize the presented fractures. Splints were more popular than circular casts, especially for surgeons who advocated reduction and percutaneous pin fixation (61% vs. 39%, p < 0.001). The responses concerning the length of immobilization varied widely, with the following distribution: no immobilization 0.8%, 2 weeks 1.4%, 3 weeks 13%, 4 weeks 44%, 5 weeks 14%, 6 weeks 26%, and > 6 weeks 0.8%. The median length of cast immobilization chosen by the Nordic surgeons was 4 weeks, whereas the American surgeons preferred 6 weeks. There was also a wide variation in the preferred number of outpatient visits and follow-up radiographs. The median number of follow-up visits suggested by Nordic surgeons was 2 compared with the 3 recommended by surgeons from the USA (p = 0.003). Follow-up of longer than 2 months was suggested by 16% of the Nordic respondents and by 32% of the Americans. Table 2 presents the distribution of answers in each patient case and the lack of agreement on the method of treatment (α = 0.12). Overall, participants showed internal consistency in the method of treatment (α = 0.86). There was no correlation with the type of ulnar fracture, the surgeon’s pediatric orthopedic experience, the preferred method of treatment, type and length of immobilization, or with the number of clinical and radiographic follow-ups that were suggested (p > 0.1). When

238

all variables were analyzed by binary logistic regression analysis, the only attribute that correlated positively with pin fixation was Finland as the surgeon’s country of practice (OR 5.3, 95% CI 1.5–19). Conversely, the type of ulnar fracture (Case 1 – no displacement and no shortening) (OR 2.1, CI 1.1–3.7) and the length (> 10 years) of respondents’ pediatric orthopedic experience (OR 2.1, CI 1.3–3.2) correlated positively with nonoperative treatment.

Discussion Do et al. (2003) and Crawford et al. (2012) have reported that overriding distal radius fractures in children can be treated by letting the fractures unite in the bayonet position in a cast, and uneventful remodeling will follow. According to our survey, a significant change in the traditional thinking that these fractures should be reduced has not taken place. The 3 fractures presented in the survey were healed in a bayonet position by 4 weeks (Figure 4) and were radiologically completely remodeled in 1 year (Figure 4). Full function was evident by 6 months. A recent Cochrane analysis (Handoll et al. 2018) outlines the need for high-quality studies on whether cast immobilization has better results with no formal reduction or with closed reduction and percutaneous pin fixation of distal displaced forearm fractures in children. AO Surgery Reference Guidelines for treatment of displaced distal metaphyseal forearm fractures do not give an upper limit for angulation or shortening in children aged under 10 years. In other words, AO Surgery Reference Guidelines, used worldwide, do not give clear recommendations for children under 10 years of age. Could this be one of the reasons why most surgeons still reduce these fractures in all age groups? In Bernthal et al. (2015) an internet-based survey on the management of pediatric distal radius fractures reveals that fewer than 10% of American surgeons recommend cast immobilization without reduction of overriding fractures as the primary treatment. Interestingly, the overriding fracture position was accepted by approximately half of the pediatric orthopedic surgeons at the 1-week follow-up. Several previous studies have recommended percutaneous pin fixation to avoid redisplacement (Gibbons et al. 1994, McLauchlan et al. 2002, Miller et al. 2005, Colaris et al. 2013). Our study results were similar to those of Bernthal et al. (2015): only a minority of surgeons would leave overriding distal radius fractures in children unreduced. Percutaneous pin fixation appears to be especially popular in the Nordic countries for no evident reason. Above-elbow casts do not seem to retain alignment of distal forearm fractures in children any better than below-elbow casts, according to earlier reports (Bohm et al. 2006, Paneru et al. 2010). Nevertheless, most surgeons who participated in this survey preferred an above-elbow cast. We could not find any evidence in the literature on the superiority of either

Acta Orthopaedica 2021; 92 (2): 235–239

circular casts or splints in immobilizing distal forearm fractures in children, which aligns with the results of our study, with a relatively even distribution of circular casts and splints. Nordic surgeons seem to prefer splints in contrast to American surgeons, who favor circular casts. There are no clear guidelines on the length of cast immobilization in children’s distal forearm fractures, which has varied from 4 to 6 weeks in previous studies (McLauchlan et al. 2002, Miller et al. 2005, Bohm et al. 2006, Paneru et al. 2010, Crawford et al. 2012, Colaris et al. 2013) In our study, 84% of responses fell into the 4–6-week category. Again, there was regional variability concerning the length of cast immobilization, as Nordic surgeons would generally remove casts at 4 weeks, 2 weeks earlier than their American colleagues. Rockwood and Wilkins (Flynn et al. 2015) recommend repeated follow-ups weekly for the first 3 weeks to monitor alignment of distal forearm fractures in children, but they give no recommendations concerning the length of the follow-up. In addition, the guidelines from the recent Cochrane review (Handoll et al. 2018) are no better. Malunited distal forearm fractures in children are completely remodeled within 3 to 12 months with few exceptions (Do et al. 2003, Crawford et al. 2012). Therefore, routine radiographic controls and long-term follow-up seem unnecessary. Most respondents would have nevertheless arranged at least 2 or 3 clinical and radiographic follow-up examinations, presumably at least partially to monitor fracture alignment. Conversely, more than 80% of the participating surgeons would have discontinued follow-ups of their patients by 3 months. The number of suggested outpatient appointments and the length of the follow-up appear to be shorter in Nordic countries than in the USA, which could be partially explained by the higher rate of pin fixation in the Nordic countries. In the USA, longer and more frequent clinical and radiographic follow-up may represent defensive medicine regarding malpractice litigation. The results of this study should be interpreted with caution because the respondents comprise only a small fraction of all surgeons treating pediatric fractures. Second, treatment decisions in this survey were based on radiographs and a short patient history. Actual bedside decisions could be different. Third, according to responses from some North American respondents, closed reduction with conscious sedation in the emergency room is a common method of treatment, which was not included as an option. Fourth, we did not present an overriding distal radius fracture with an intact ulna, which might have propelled more surgeons to choose the options that did not involve formal reduction of the fractures. Conclusion Based on our survey, the most common treatment method of overriding distal radius fractures in < 10-year-old children is reduction under anesthesia and immobilization with an aboveelbow cast. Percutaneous pin fixation is popular in the Nordic countries. Very few surgeons would treat these fractures with-

Acta Orthopaedica 2021; 92 (2): 235–239 239

out reduction. There is no consensus regarding the type of cast, the length of immobilization, or the number of follow-up examinations. The reports of Do et al. (2003) and Crawford et al. (2012) have thus not led to a change in treatment praxis. We have therefore started a non-inferiority randomized controlled treatment trial registered in Clinical Trials (Casting in finger trap traction without reduction and percutaneous pin fixation of dorsally displaced, overriding distal forearm fractures in children under 11 years old, ClinicalTrials.gov Identifier: NCT04323410). Acta thanks Pepijn Bisseling and Klaus Dieter Parsch for help with peer review of this study.

Alemdaroğlu K B, İltar S, Çimen O, Uysal M, Alagöz E, Atlıhan D. Risk factors in redisplacement of distal radial fractures in children. J Bone Joint Surg 2008; 90(6): 1224-30. Bernthal N M, Mitchell S, Bales J G, Benhaim P, Silva M. Variation in practice habits in the treatment of pediatric distal radius fractures. J Pediatr Orthop B 2015; 24(5): 400-7. Bohm E R, Bubbar V, Yong Hing K, Dzus AJ. Above and below-the-elbow plaster casts for distal forearm fractures in children: a randomized controlled trial. J Bone Joint Surg Am 2006; 88(1): 1-8. Colaris J W, Allema J H, Biter L U, de Vries M R, van de Ven CP, Bloem RM, et al. Re-displacement of stable distal both-bone forearm fractures in children: a randomised controlled multicentre trial. Injury 2013; 44(4): 498-503.

Cortina J M. What is coefficient alpha? An examination of theory and applications. J Appl Psychol 1993; 78(1): 98-104. Crawford S N, Lee L S, Izuka B H. Closed treatment of overriding distal radial fractures without reduction in children. J Bone Joint Surg Am 2012; 94(3): 246-52. Do T T, Strub W M, Foad S L, Mehlman C T, Crawford A H. Reduction versus remodeling in pediatric distal forearm fractures: a preliminary cost analysis. J Pediatr Orthop B 2003; 12(2): 109-15. Flynn J M, Skaggs D L, David L, Waters P M. Rockwood and Wilkins: Fractures in children. 8th ed. Philadelphia: Wolters Kluver Health; 2015. Gibbons C L M H, Woods D A, Pailthorpe C, Carr A J, Worlock P. The management of isolated distal radius fractures in children. J Pediatr Orthop 1994; 14(2): 207-10. Handoll H H, Elliott J, Iheozor-Ejiofor Z, Hunter J, Karantana A. Interventions for treating wrist fractures in children. Cochrane Database Syst Rev 2018; 12(12): CD012470. doi: 10.1002/14651858.CD012470.pub2. Hang J R, Hutchinson A F, Hau R C. Risk factors associated with loss of position after closed reduction of distal radial fractures in children. J Pediatr Orthop 2011; 31(5): 501-6. McLauchlan G J, Cowan B, Annan I H, Robb J E. Management of completely displaced metaphyseal fractures of the distal radius in children. J Bone Joint Surg Br 2002; 84-B(3): 413-17. Miller B S, Taylor B, Widmann R F, Bae D S, Snyder B D, Waters P M. Cast immobilization versus percutaneous pin fixation of displaced distal radius fractures in children. J Pediatr Orthop 2005; 25(4): 490-4. Paneru S R, Rijal R, Shrestha B P, Nepal P, Khanal G P, Karn N K, et al. Randomized controlled trial comparing above- and below-elbow plaster casts for distal forearm fractures in children. J Child Orthop 2010; 4(3): 233-7. Wendling-Keim D S, Wieser B, Dietz H-G. Closed reduction and immobilization of displaced distal radial fractures: method of choice for the treatment of children? Eur J Trauma Emerg Surg 2014; 41(4): 421-8. Zamzam M M, Khoshhal K I. Displaced fracture of the distal radius in children. J Bone Joint Surg Br 2005; 87-B(6): 841-3.

240

Acta Orthopaedica 2021; 92 (2): 240–243

Emotional tones in scientific writing: comparison of commercially funded studies and non-commercially funded orthopedic studies Anath N V STEFFENS 1, David W G LANGERHUIZEN 1, Job N DOORNBERG 2, David RING 3, and Stein J JANSSEN 1 1 Department 2 Department

of Orthopaedic Surgery, Amsterdam, Amsterdam Movement Sciences (AMS), Amsterdam University Medical Centre, The Netherlands; of Orthopaedic & Trauma Surgery, Flinders University, Adelaide, Australia, Flinders Medical Centre; 3 Department of Surgery and Perioperative Care, Dell Medical School, The University of Texas at Austin, Austin, TX, USA Correspondence: david.langerhuizen@gmail.com Submitted 2020-10-05. Accepted 2020-11-09.

Background and purpose — There is ongoing debate as to whether commercial funding influences reporting of medical studies. We asked: Is there a difference in reported tones between abstracts, introductions, and discussions of orthopedic journal studies that were commercially funded and those that were not commercially funded? Methods — We conducted a systematic PubMed search to identify commercially funded studies published in 20 orthopedic journals between January 1, 2000 and December 1, 2019. We identified commercial funding of studies by including in our search the names of 10 medical device companies with the largest revenue in 2019. Commercial funding was designated when either the study or 1 or more of the authors received funding from a medical device company directly related to the content of the study. We matched 138 commercially funded articles 1 to 1 with 138 non-commercially funded articles with the same study design, published in the same journal, within a time range of 5 years. The IBM Watson Tone Analyzer was used to determine emotional tones (anger, fear, joy, and sadness) and language style (analytical, confident, and tentative). Results — For abstract and introduction sections, we found no differences in reported tones between commercially funded and non-commercially funded studies. Fear tones (non-commercially funded studies 5.1%, commercially funded studies 0.7%, p = 0.04), and analytical tones (noncommercially funded studies 95%, commercially funded studies 88%, p = 0.03) were more common in discussions of studies that were not commercially funded. Interpretation — Commercially funded studies have comparable tones to non-commercially funded studies in the abstract and introduction. In contrast, the discussion of non-commercially funded studies demonstrated more fear and analytical tones, suggesting them to be more tentative, accepting of uncertainty, and dispassionate. As text analysis tools become more sophisticated and mainstream, it might help to discern commercial bias in scientific reports.

There is ongoing debate as to whether commercial funding influences reporting of medical studies. While some studies of industry funding find differences in author conclusions, others do not (Clifford et al. 2002, Kjaergard and Als-Nielsen 2002). For instance, Lundh et al. (2018) included 75 papers that compared primary research studies sponsored by industry with studies with other sources of sponsorship. They found that industry sponsored studies present more favorable results (RR 1.3, 95% CI 1.2–1.4) and conclusions (RR 1.3, CI 1.2–1.5) as compared with studies that are not sponsored by industry. Conversely, Clifford et al. (2002) included 100 randomized controlled trials, of which 66% received funding, in whole or in part, from industry. They did not find a statistically significant association (p = 0.5) between funding source and trial outcome. However, this study may not be generalizable since it focused on recent publications of the top 5 general medical journals. Lower tier journals and specialty journals might not be as good at editing out bias. Furthermore, 100 studies might provide inadequate power for a small influence of funding. Machine-learning-based tone analyzers are increasingly used to provide psycholinguistic analysis of text. For instance, the IBM Watson Tone Analyzer measures tones such as confidence and joy that might be more common in commercially funded studies if they are more promotional (Cloud 2019). Prior evidence suggests medical studies that use words such as “unique” and “novel” are more likely to be cited. Furthermore, there is evidence that men frame their studies more positively than women; women are more dispassionate in their writing (Lerchenmueller et al. 2019). As such, it is of interest to evaluate whether a tone analyzer could help identify a difference in reported tones between commercially and non-commercially funded studies. This study addressed the primary null hypothesis that there is no difference in reported tones between abstracts of studies that were commercially funded and those that were not in 20 orthopedic journals, as analyzed by the IBM Watson Tone Analyzer. Secondarily, we addressed differences in tones in the introduction and discussion sections of the full paper.

Acta Orthopaedica 2021; 92 (2): 240–243 241

Methods We conducted a systematic PubMed search to identify commercially funded studies published in 20 orthopedic journals between January 1, 2000 and December 1, 2019 (Appendix 1). The top 20 orthopedic journals were selected based on ranked impact factors according to Clarivate, a non-profit organization maintaining a website where journal statistics including impact factor are reported, on December 12, 2019. Commercial funding of studies was identified by including the names of 10 medical device companies with the largest revenue in 2019 in our search. We excluded letters to the editor, review articles, conference abstracts, animal and cadaveric studies, and studies not published in English. A research fellow (ANVS) independently reviewed the conflict of interest (COI) statement for each study to confirm whether it was commercially funded or not. Commercial funding was designated when either the study or 1 or more of the authors received funding from a medical device company directly related to the content of the study. All COI statements were reviewed by a second research fellow (DWGL) to assess funding and confirm study eligibility. We excluded articles in which the COI statement did not mention funding, articles without a COI statement, and articles lacking an introduction or discussion. For every selected commercially funded article, a similar article without commercial funding was matched 1 to 1 based on orthopedic journal, study design, and timeframe. A noncommercially—as indicated in the COI statement—funded study needed to have the same study design (e.g., randomized controlled trials, prospective cohort, retrospective cohort, case-series, case-control) as the commercially funded study, and to have been published within a time range of 5 years. The PubMed search yielded 753 articles. We excluded 106 citations without a COI statement and 26 publications lacking an introduction or a discussion. Of the remaining 621 articles, we retained 138 commercially funded studies that could be matched with 138 non-commercially funded studies. IBM Watson Tone Analyzer We used the IBM Watson Tone Analyzer to determine the reported tones of each article. The tone analyzer is based on the theory of psycholinguistics, wherein the relationship between behavior and psychological theories is explored. The IBM Watson Tone Analyzer is a machine-learning based model that has been trained on 96,000 customer-service Twitter conversations, rated by 5 annotators. According to IBM, the analyzer’s performance showed high accuracy against benchmark data. However, no reliability statistic or actual number has been reported to measure its performance (Cloud 2019). The tone analyzer reports emotional tones (anger, fear, joy, and sadness) as well as language style (analytical, confident, and tentative) (Table 1). In this study, every abstract, intro-

Table 1. Definition of dominant tones Tones Definition Anger

Fear

Joy

Sadness

Analytical Confident Tentative

Evoked due to injustice, conflict, humiliation, negligence, or betrayal. If anger is active, the individual attacks the target, verbally or physically. If anger is passive, the person silently sulks and feels tension and hostility A response to impending danger. It is a survival mechanism that is a reaction to some negative stimulus. It may be a mild caution or an extreme phobia Joy or happiness has shades of enjoyment, satisfaction, and pleasure. There is a sense of well-being, inner peace, safety, and contentment Indicates a feeling of loss and disadvantage. When a person can be observed to be quiet, less energetic, and withdrawn, it may be inferred that sadness exists A person’s reasoning and analytical attitude about things A person’s degree of certainty A person’s degree of inhibition

From IBM Cloud Docs. Personality Insights. Available at: https:// console.bluemix.net/docs/services/personality-insights/models. html#models. Accessed October 28, 2020.

duction, and discussion was copied separately into the tone analyzer. The reported scores vary between 0 and 1, in which < 0.5 means no tone, 0.5–0.75 means there is a tone detected, and > 0.75 means a strong tone is detected (Cloud 2019). Statistics The continuous data obtained by the IBM Watson Tone Analyzer was categorized into 2 groups: no tone (reported score < 0.5) and tone (reported score > 0.5). We used a McNemar test to compare dominant tones between commercially funded and non-commercially funded articles. A 2-tailed p-value less than 0.05 was considered statistically significant. All statistical analyses were performed using Stata® 15.0 (StataCorp LP, College Station, TX, USA). To identify a difference with an effect size of 0.05 per dominant tone with sufficient power, we needed at least 84 papers per group, therefore 164 papers in total (α = 0.05, b = 0.10). Ethics, funding, and potential conflicts of interest This study was exempt from institutional review board approval because it involves open-source data. We did not receive financial support for this study. All authors declare no conflicts of interest.

Results There were a similar number of tones that met the 0.5 threshold for commercially funded and non-commercially funded studies in the abstract and the introduction sections (Tables 2 and 3); for example, analytical tone was detected in 72%

242

Acta Orthopaedica 2021; 92 (2): 240–243

Table 2. Tone prevalence between non-commercially and commercially funded orthopedic journal articles — abstracts Tones Anger Fear Joy Sadness Analytical Confident Tentative

Non-commercially Commercially funded funded article (n = 138) article (n = 138) n (%) n (%) p-value 1 (0.7) 9 (7.0) 43 (31) 30 (22) 104 (75) 20 (14) 13 (9.4)

1 (0.7) 8 (5.8) 40 (29) 30 (22) 100 (72) 18 (13) 14 (10)

1.0 0.8 0.9 1.0 0.6 0.7 0.8

Table 3. Tone prevalence between non-commercially and commercially funded orthopedic journal articles — introduction Tones Anger Fear Joy Sadness Analytical Confident Tentative

Non-commercially Commercially funded funded article (n = 138) article (n = 138) n (%) n (%) p-value – 6 (4.4) 27 (20) 43 (31) 121 (88) 4 (2.9) 43 (31)

– 11 (8.0) 27 (20) 37 (27) 115 (84) 9 (6.6) 30 (22)

– 0.2 1.0 0.4 0.3 0.2 0.09

– indicates no tones detected.

Table 4. Tone prevalence between non-commercially and commercially funded orthopedic journal articles — discussion Tones Anger Fear Joy Sadness Analytical Confident Tentative

Non-commercially Commercially funded funded article (n = 138) article (n = 138) n (%) n (%) p-value – 7 (5.1) 42 (31) 67 (49) 130 (95) 3 (2.2) 51 (37)

– 1 (0.7) 56 (41) 73 (53) 120 (88) 1 (0.7) 49 (36)

– 0.04 0.09 0.5 0.03 0.3 0.8

– indicates no tones detected.

of study abstracts that were commercially funded, and 75% of study abstracts that were not commercially funded. In the introduction, analytical tone was detected in 84% of the studies that were commercially funded, and in 88% of the studies that were not commercially funded. There was a difference in number of tones that met the 0.5 threshold in the discussion section (Table 4). Fear tones (noncommercially funded studies: 5.1%, commercially funded studies: 0.7%, p = 0.04) and analytical tones (non-commercially funded studies: 95%, commercially funded studies: 88%, p = 0.03) were more common in unfunded studies.

Discussion Tone analyzers may help determine whether there is bias in commercially funded studies. We found only limited difference in tone between commercially and non-commercially funded studies in orthopedic journals. This study has several limitations. 1st, the relative infrequency of tones greater than 0.5 meant that we had to categorize tones as detected or not detected and could not analyze tone on its continuum. This might have introduced information bias. However, including only tones greater than 0.5 (leaving out studies in which no tone was detected) also leads to loss of information and therefore information bias. We felt that dichotomizing tone was the most adequate solution, as this method allowed for inclusion of all papers. 2nd, the IBM tone analyzer was trained on a large Twitter customer-support dataset. Although previously used in the context of medical studies, the reliability of the analyzer for medical journals is untested (Ottenhoff et al. 2018, Rajesh et al. 2018, Bakker et al. 2019, Karacic et al. 2019, Black et al. 2020, Langerhuizen et al. 2020). 3rd, we included only studies in high-impact orthopedic journals. This may potentially have introduced selection bias. However, we included studies from 20 different journals, and therefore consider this risk to be low. The observation that the abstract and introduction were similar among commercially and non-commercially funded studies suggests that these sections of orthopedic studies are reported with comparable sentiment. The high percentage of abstracts and introductions that demonstrate analytical tone (72–88%) suggests that both may be reported in relatively dispassionate scientific language. This finding is in line with prior evidence demonstrating that there is no correlation between industry funding and more favorable reporting of a specific treatment in abstracts of orthopedic randomized controlled trials (Boutron et al. 2010, Arthur et al. 2020). In contrast, another study—evaluating randomized controlled trials in five orthopedic journals—demonstrated a substantially higher likelihood of presenting favorable outcomes in industry-funded studies (Khan et al. 2008). In addition, studies from ophthalmology and emergency medicine journals found that industry-funded trials were more likely to present the results as positive. In a study of topical prostaglandin research in ophthalmology, the conclusion presented in the abstract was not consistent with the statistical results in 18 of 29 of the commercially funded studies and none of the 10 non-commercial studies (Alasbali et al. 2009). In a review of emergency medicine clinical trial abstracts with no significant differences, a positive spin (i.e., selective reporting of significant differences, promotion of non-significant differences, favorable interpretation of non-significant results, or claimed benefit in spite of no significant difference) was present in 15 of 21 industry-funded trials compared with 35 of 93 non-industry funded trials (Reynolds-Vaughn et al. 2019).

Acta Orthopaedica 2021; 92 (2): 240–243 243

These are 2 of the many studies presenting evidence that industry-funded trials are associated with positive, proindustry study findings (Kjaergard and Als-Nielsen 2002, Bhandari et al. 2004, Alasbali et al. 2009, Boutron et al. 2010, Lundh et al. 2018, Lerchenmueller et al. 2019, Arthur et al. 2020). Among 186 registered randomized controlled trials comparing generic and brand-name drugs, only 46% were published within 4 years of completing the trial: 71% sponsored by a company with financial gains from both the generic and brand-name drugs, 28% comparing drugs from competing companies, and 46% with a non-profit sponsor (Flacco et al. 2016). Our finding that fear and analytical tones were slightly more common in the discussion of non-commercially funded studies suggests the authors of non-commercially funded studies might be addressing uncertainty more directly, while also being less promotional and more dispassionate and analytical. Although we found a statistically significant difference, we consider this finding not to be a large clinically relevant difference. Our finding that commercially funded studies have tones comparable to non-commercially funded studies in the abstract and introduction, but not in the discussion, suggests that abstracts and introductions might be more carefully edited to remove self-promotion than the discussion section of the paper. The discussion section of non-commercially funded studies has more fear and analytical tones, suggesting they might be more tentative, accepting of uncertainty, and dispassionate. As text analysis becomes more sophisticated, it might be able to discern commercial bias in scientific reports. Acta thanks Jon A Tsai for help with peer review of this study.

Alasbali T, Smith M, Geffen N, Trope G E, Flanagan J G, Jin Y, Buys Y M. Discrepancy between results and abstract conclusions in industry- vs nonindustry-funded studies comparing topical prostaglandins. Am J Ophthalmol 2009; 147(1): 33-8 e2. doi: 10.1016/j.ajo.2008.07.005. Arthur W, Zaaza Z, Checketts J X, Johnson A L, Middlemist K, Basener C, Jellison S, Wayant C, Vassar M. Analyzing spin in abstracts of orthopaedic randomized controlled trials with statistically insignificant primary endpoints. Arthroscopy 2020: 1443-50 e1. doi: 10.1016/j.arthro.2019.12.025. Bakker D, Ottenhoff J S E, Ring D. Factors associated with the quality of online information about scapholunate interosseous ligament insufficiency. J Hand Microsurg 2019; 11(2): 94-9. doi: 10.1055/s-0038-1675887.

Bhandari M, Busse J W, Jackowski D, Montori V M, Schunemann H, Sprague S, Mears D, Schemitsch E H, Heels-Ansdell D, Devereaux P J. Association between industry funding and statistically significant pro-industry findings in medical and surgical randomized trials. CMAJ 2004; 170(4): 477-80. Black C K, Fan K L, Economides J M, Camden R C, Del Corral G A. Analysis of chest masculinization surgery results in female-to-male transgender patients: demonstrating high satisfaction beyond aesthetic outcomes using advanced linguistic analyzer technology and social media. Plast Reconstr Surg Glob Open 2020; 8(1): e2356. doi: 10.1097/GOX.0000000000002356. Boutron I, Dutton S, Ravaud P, Altman D G. Reporting and interpretation of randomized controlled trials with statistically nonsignificant results for primary outcomes. JAMA 2010; 303(20): 2058-64. doi: 10.1001/ jama.2010.651. Clifford T J, Barrowman N J, Moher D. Funding source, trial outcome and reporting quality: are they related? Results of a pilot study. BMC Health Serv Res 2002; 2(1): 18. doi: 10.1186/1472-6963-2-18. Cloud I W D. Tone Analyzer. Available from: https://cloudibmcom/docs/services/tone-analyzer (accessed July 7, 2019). Flacco M E, Manzoli L, Boccia S, Puggina A, Rosso A, Marzuillo C, Scaioli G, Gualano M R, Ricciardi W, Villari P, Ioannidis J P. Registered randomized trials comparing generic and brand-name drugs: a survey. Mayo Clin Proc 2016; 91(8): 1021-34. doi: 10.1016/j.mayocp.2016.04.032. Karacic J, Dondio P, Buljan I, Hren D, Marusic A. Languages for different health information readers: multitrait-multimethod content analysis of Cochrane systematic reviews textual summary formats. BMC Med Res Methodol 2019; 19(1): 75. doi: 10.1186/s12874-019-0716-x. Khan S N, Mermer M J, Myers E, Sandhu H S. The roles of funding source, clinical trial outcome, and quality of reporting in orthopedic surgery literature. Am J Orthop (Belle Mead, NJ) 2008; 37(12): E205-12; discussion E12. Kjaergard L L, Als-Nielsen B. Association between competing interests and authors’ conclusions: epidemiological study of randomised clinical trials published in the BMJ. BMJ (Clinical Research ed.) 2002; 325(7358): 249. doi: 10.1136/bmj.325.7358.249. Langerhuizen D W G, Brown L E, Doornberg J N, Ring D, Kerkhoffs G, Janssen S J. Analysis of online reviews of orthopaedic surgeons and orthopaedic practices using natural language processing. J Am Acad Orthop Surg 2020. doi: 10.5435/JAAOS-D-20-00288. [Online ahead of print] Lerchenmueller M J, Sorenson O, Jena A B. Gender differences in how scientists present the importance of their research: observational study. BMJ (Clinical Research ed.) 2019; 367: l6573. doi: 10.1136/bmj.l6573. Lundh A, Lexchin J, Mintzes B, Schroll J B, Bero L. Industry sponsorship and research outcome: systematic review with meta-analysis. Intensive Care Med 2018; 44(10): 1603-12. doi: 10.1007/s00134-018-5293-7. Ottenhoff J S E, Kortlever J T P, Teunis T, Ring D. Factors associated with quality of online information on trapeziometacarpal arthritis. J Hand Surg Am 2018; 43(10): 889-96 e5. doi: 10.1016/j.jhsa.2018.08.004. Rajesh K, Crijns T J, Ring D. Themes in published obituaries of people who have died of opioid overdose. J Addict Dis 2018; 37(3-4): 151-6. doi: 10.1080/10550887.2019.1639485. Reynolds-Vaughn V, Riddle J, Brown J, Schiesel M, Wayant C, Vassar M. Evaluation of spin in the abstracts of emergency medicine randomized controlled trials. Ann Emerg Med 2019; 14: 423-31. doi: 10.1016/j.annemergmed.2019.03.011.

244

Acta Orthopaedica 2021; 92 (2): 244–246

Correspondence

Antibiotics should not be used for back/leg pain Acta Orthop 2021; 92 (1): 1–3. DOI 10.1080/17453674.2020.1855561

Sir,—It is rather depressing to read the article addressing the topic above in Acta Orthopaedica (Fritzell et al. 2021). By using the definitive title: ”Antibiotics should not be used to treat back pain” supported by a highly selective and extremely limited reference list, Fritzell et al. attempt to shut down the low virulent infection hypothesis leading to Modic changes (MC) and chronic low back pain (Fritzell et al. 2021). Readers have to plow through a long and basically irrelevant introduction (Albert et al. 2008, Fritzell et al. 2021) prior to encountering the article’s point of contention. Are antibiotics effective for the treatment of patients with MCs? 1. The authors’ own publication, in which circa 50% of a child cohort complete a questionnaire 13 years after inclusion is likely just an abstraction for most readers as regards making the case for or against the usage of antibiotic treatment for patients with back pain. 2. Fritzell et al. (2019) also refer to their own biopsy article in which they found no evidence of bacteria in the disc material from several patients and in which they conclude that if bacteria were found in an individual patient that this would be due to a contamination process due to the biopsy itself. Many studies (20+) have disproven the contamination hypothesis during the years (Capoor et al. 2019, Manniche and O’Neill 2019. Pradip et al. 2020). And by using fluorescence in situ hybridization microscopy C. acnes bacteria can be seen in aggregates and biofilms in human disc material which has initiated a local inflammatory response (Capoor et al. 2017, Ohrt-Nissen et al. 2018). Several leading experts in this field have criticized the results of this study and point to several methodological problems as the reason for their lack of bacterial identification (Capoor et al. 2017). The interesting considerations in the article are hidden in the last 20 lines (Fritzell et al. 2021). Fritzell describes how a controversial RCT from Bråten et al. (2019) tested whether it was possible to reproduce the same large effect with antibiotic treatment on patients with MCs as the Danish trial (Albert et al. 2013). The Norwegian trial concluded that they did not find a “significant” effect (Bråten et al. 2019). Bråten et al. chose – despite reviewer objections (BMJ 2019) – to mix results for MC type 1 with MC type 2 in their analyses. This is the equivalent of mixing cold and warm water!

The Tables in the Bråten et al. article’s supplementary appendix (2019) told a different story when data for patients with MC1 was presented distinctly from patients with MC2. Data for MC1 patients demonstrated a statistically significant difference and meaningful improvement rarely seen in RCTs involving supervised exercise, manipulation or even spinal surgery for chronic back pain. Patients with MC2 did less well than the placebo group! Already at the start of the reviewer process and since then a large number of back pain experts have written in different forums about the methodological weaknesses of the study (including the mixing up of data sets) in the Norwegian trial and highlighted the misleading conclusions based upon mixed MC1 and 2 patients (BMJ 2019, Albert 2019, Creaney 2019, Fairbanks 2019, Joffe 2019, Lambert 2019). The Norwegian authors have recently published a new sub-group analysis of their RCT (Kristoffersen et al. 2020). The conclusions have been significantly modified. A substantial subgroup of patients with MC1 and oedema seen on STIR sequences demonstrated a large difference between the actively treated group and the placebo group as measured by the Roland Morris Disability Questionnaire (RMDQ), the primary outcome measure; –5.1 RMDQ points; 95% CI –8.2 to –1.9; p = 0.008). The clinical improvements were already seen at 3 months and were consistent at 1-year follow-up which showed that 27% of the actively treated group experienced improvements of more than 75%! The Number Needed to Treat (NNT) in this subgroup was 3.1. The overall conclusion can be that the disc low grade infection hypothesis is a most interesting area of research. This patient group which suffers from longstanding and severe back pain worldwide is deserving of more than simplistic attempts to block further research in this space. Claus Manniche Department of Occupational and Environmental Medicine, Odense University Hospital, and Institute of Clinical Research, University of Southern Denmark, Odense, Denmark E-mail: Claus.Manniche2@rsyd.dk Conflict of interest: Shares in Persica Pharmaceutical Ltd., London

Acta Orthopaedica 2021; 92 (2): 244–246 245

Sir,—Claus Manniche’s comment to our paper (Fritzell et al. 2021) contains several incorrect claims which taken together create the idea that ordinary back pain could be an infection and the remedy antibiotics. Misconception 1: Modic Type 1 change (MC1) on MRI is a cause of chronic low back pain The implication of Modic changes has engaged spine surgeons and radiologists for more than 30 years (Modic et al. 2008a and b). The conclusion of the Danish long-term followup (Udby et al. 2019) concurs with 2 recent reviews stating that “There is no conclusive evidence on the causative role of MC in chronic low back pain (LBP) or any influence on the long-term outcome in patients with LBP or lumbar disc herniations” (Viswanathan et al. 2020) and “the association between MCs and LBP-related outcomes are inconsistent” (Herlin et al. 2019). Misconception 2: Infection of the intervertebral disc with Cutibacterium acnes (C. acnes) in cases of MC1 is confirmed by “biofilm” and “inflammatory response” Several studies have reported this bacterial finding as Manniche states. C. acnes is a commensal of the skin flora with the potential to contaminate sampling procedures. That is what the Swedish study (Fritzell et al. 2019) produced evidence of. Manniche claims that the presence of “aggregates and biofilm” and “local inflammatory response” disproves the evidence. First, aggregates and biofilm are not specific of a disc infection. They can occur in the skin and hair follicles (Jahns et al. 2012). So, a contamination may be in the appearance of bacteria, aggregated bacteria or with a biofilm. The presence of biofilm solely demonstrates that C. acnes is capable of this presentation. Biofilm appears to be more related to phylotype of the bacteria than to site or tissue (Kuehnast et al. 2018). Second, the presence of inflammatory response is what you would expect in a case of MC1, since the implication of this phenomenon usually is considered to be just that – an inflammation. If the MRI demonstrates an inflammation above and/ or below the degenerated disc, an inflammation in the disc itself is quite feasible. All in all, inflammation is to be expected in MC1. Biofilm is an optional presentation of C. acnes. Misconception 3: The Norwegian AIM-study (Bråten et al. 2019) obscures the superior effect of antibiotic treatment by presenting the combined effect of MC1 and MC2 This is wrong. In fact, the analyses of the combined group and the MC1 group separately, both demonstrate statistically significant effects in favor of antibiotic treatment. With the outcome measure RMDQ (Roland and Morris 1983, Ogura et al. 2019) the difference in favor of antibiotic treatment is 1.6 units (p = 0.04) in the combined group. In the MC1 group the difference in favor of antibiotics is 2.3 units (p = 0.02). The problem is that the minimal clinically important difference

(MCID) of RMDQ is 4–5 units (Maughan and Lewis 2010, Ogura et al. 2019) and the measurement error makes it unable to detect a change smaller than 4–8 units (Stratford et al. 1996, Grotle et al. 2003, Chiarotto et al. 2016). So, a difference of 1.6 or 2.3 is a non-value or a “nonsense value” when it comes to clinical practice. It is like trying to measure a millimeter difference with a centimeter measuring stick. One should not be dazzled by p-values without clinical relevance. We note that those who questioned the conclusion of the AIM-study were nearly all shareholders in Persica, a pharmaceutical company founded to promote the use of antibiotics in back pain. Misconception 4: A recent subgroup analysis in the Norwegian AIM-study (Kristoffersen et al. 2020) supports the idea of a low-grade disc infection As the authors themselves conclude, there are several reasons of concern in the interpretation of the analyses. A large number of tests result in a probability of false positive results (Milojevic et al. 2020). They find a difference, exceeding the MCID, in favor of antibiotics for a post-hoc constructed variable constituting 22% of the original study population for one of three outcome measures (RMDQ). Neither back pain (NRS) nor function (Oswestry Disability Index) show clinically relevant or statistically significant differences. This study does not give support to the idea of low-grade disc infection. It regenerates a hypothesis already unsubstantiated and questioned. So far, the only study that claims the effect of antibiotic treatment of MC1 is published by Manniche himself et al. (Albert et al. 2013). It is a study with a strange and unexplained asymmetrical randomization procedure and a dose response construction which is not reported. Conclusion: MC1 is not a condition, a disorder, or a disease.It is a finding on MRI. Present scientific evidence does not support prescription of antibiotics for that. Peter Fritzell 1, Olle Hägg 2, Tomas Bergström 3, Bodil Jönsson 4, Siv G E Andersson 5, Mikael Skorpil 6, Peter Muhareb Udby 7, and Mikkel Andersen 8 1 RKC

Centre for Spine Surgery in Stockholm, Sweden Center Göteborg, Västra Frölunda, Sweden Email: olle.hagg@spinecenter.se 3 Department of Infectious Diseases, Institute of Biomedicine, University of Gothenburg, Sweden 4 Sahlgrenska University Hospital, Göteborg, Sweden 5 Department of Cell and Molecular Biology, Biomedical Center, Uppsala University, Sweden 6 Department of Neuroradiology, Karolinska University Hospital, Stockholm, Sweden 7 Spine Unit, Ortopædkirurgisk Afdeling, Sjællands Universitetshospital, Køge, Denmark 8 Spine Center of Southern Denmark, Lillebaelt Hospital, Middelfart, Denmark No author has any conflict of interest 2 Spine

246

Albert H B. Re: Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019; 367: l5654. https://www. bmj.com/content/367/bmj.l5654/rr-5. Albert H B, Kjaer P, Jensen T S, et al. Modic changes, possible causes and relation to low back pain. Med Hypotheses 2008; 70(2): 361-8. doi: 10.1016/j.mehy.2007.05.014. Albert H B, Sorensen J S, Christensen B S, et al. Antibiotic treatment in patients with chronic low back pain and vertebral bone edema (Modic type 1 changes): a double-blind randomized clinical controlled trial of efficacy. Eur Spine J 2013; 22: 697-707.doi: 10.1007/s00586-013-2675-y. BMJ 2019 Peer review. Decision on Manuscript ID BMJ-050262: Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ Open 14 2020; published online Sept 8. https:// www.bmj.com/sites/default/files/attachments/bmj-article/pre-pub-history/ first_decision_9.6.19_0.pdf. Bråten L C H, Rolfsen M P, Espeland A, et al. Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019 367: l5654. doi: 10.1136/bmj.l5654. Capoor M N, Ruzicka F, Schmitz J E, et al. Propionibacterium acnes biofilm is present in intervertebral discs of patients undergoing microdiscectomy. PLoS ONE 2017; 12: e0174518. doi: 10.1371/journal.pone.0174518. Capoor M N, Birkenmaier C, Wang J C, et al. A review of microscopy-based evidence for the association of Propionibacterium acnes biofilms in degenerative disc disease and other diseased human tissue. Eur Spine J 2019; 28: 2951–71. https://doi.org/10.1007/s00586-019-06086-019-06086-y. Capoor M N, McDowell A, Birkenmaier C, et al. Letter to the Editor concerning “Bacteria: back pain, leg pain and Modic sign: a surgical multicenter comparative study” by Fritzell, P., Welinder-Olsson, C., Jönsson, B. et al. Eur Spine J 2020; 29: 628–630. doi: 10.1007/s00586-019-06237-1. Chiarotto A, Maxwell L J, Terwee C B, et al. Roland-Morris Disability Questionnaire and Oswestry Disability Index: which has better measurement properties for measuring physical functioning in nonspecific low back pain? Systematic review and meta-analysis. Phys Ther 2016; 96(10): 162037. Creaney L C. Re: Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019; 367: l5654. https://www. bmj.com/content/367/bmj.l5654/rr-1. Fairbanks J. Re: Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019; 367: l5654. https://www. bmj.com/content/367/bmj.l5654/rr-3. Fritzell P, Welinder-Olsson C, Jönsson B, et al. Bacteria: back pain, leg pain and Modic sign—a surgical multicentre comparative study. Eur Spine J 2019; 28(12): 2981-9. doi: 10.1007/s00586-019-06164-1 Fritzell P, Bergström T, Jönsson B, et al. Antibiotics should not be used for back/leg pain. Acta Orthop 2021; 92(1): 1-3. doi: 10.1080/17453674. 2020.1855561. Epub ahead of print. Grotle M, Brox J I, Völlestad N K. Cross-cultural adaptation of the Norwegian versions of the Roland-Morris Disability Questionnaire and the Oswestry Disability Index. J Rehabil Med 2003; 35: 241-7. Herlin C, Kjaer P, Espeland A, et al. Modic changes –their associations with low back pain and activity limitation: A systematic literature review and meta-analysis. PLoS One 2018; 13: 1–27.

Acta Orthopaedica 2021; 92 (2): 244–246

Joffe P. Re: Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019; 367: l5654. https://www.bmj. com/content/367/bmj.l5654/rr-4. Jahns A C, Lundskog B, Ganceviciene R, et al. An increased incidence of Propionibacterium acnes biofilms in acne vulgaris: a case-control study. Br J Dermatol 2012; 167(1): 50-8. Kuehnast T, Cakar F, Weinhäupl T, et al. Comparative analyses of biofilm formation among different Cutibacterium acnes isolates. Int J Med Microbiol 2018; 308(8): 1027-35. Kristoffersen P M, Bråten L C H, Vetti N, et al. Oedema on STIR modified the effect of amoxicillin as treatment for chronic low back pain with Modic changes—subgroup analysis of a randomized trial. Eur Radiol 2020. https://doi.org/10.1007/s00330-020-07542-w. Online ahead of print. Lambert P E. Re: Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019; 367: l5654. https://www. bmj.com/content/367/bmj.l5654/rr. Manniche C, O’Neill S. New insights link low-virulent disc infections to the etiology of severe disc degeneration and Modic changes. Future Sci OA 2019; 5(5): FSO389. doi: 10.2144/fsoa-2019-0022 Maughan E F, Lewis J S. Outcome measures in chronic low back pain. Eur Spine J 2010; 19(9): 1484-94. Milojevic M, Nikolic A, Jüni P, et al. A statistical primer on subgroup analyses. Interact Cardiovasc Thorac Surg 2020; 30(6): 839-45. Modic M T, Steinberg P M, Ross J S, et al. Degenerative disk disease: assessment of changes in vertebral body marrow with MR imaging. Radiology 1988a; 166: 193e9. Modic M T, Masaryk T J, Ross J S, et al. Imaging of degenerative disk disease. Radiology 1988b; 168: 177e86. Ogura Y, Ogura K, Kobayashi Y, et al. Minimally clinically important differences for the Japanese Orthopaedic Association Back Pain Evaluation Questionnaire (JOABPEQ) following decompression surgery for lumbar spinal stenosis. J Clin Neurosci 2019; 69: 93-6. Ohrt-Nissen S, Fritz B G, Walbom J, et al. Bacterial biofilms: a possible mechanism for chronic infection in patients with lumbar disc herniation – a prospective proof-of-concept study using fluorescence in situ hybridization. APMIS Acta Pathol. Microbiol Immunol Scand 2018; 126: 440-7. doi: 10.1111/apm.12841 Pradip I, Dilip Chand Raja S, Rajasekaran S, et al. Presence of preoperative Modic changes and severity of endplate damage score are independent risk factors for developing postoperative surgical site infection: a retrospective case-control study of 1124 patients. Eur Spine J 2020. https://doi. org/10.1007/s00586-020-06581-7. Epub ahead of print. Roland M O, Morris R W. A study of the natural history of back pain. Part 1: Development of a reliable and sensitive measure of disability in low back pain. Spine 1983; 8: 141-4. Stratford P W, Binkley J, Solomon P, et al. Defining the minimum level of detectable change for the Roland-Morris questionnaire. Phys Ther 1996; 76(4): 359-65; discussion 366-8. Udby P M, Bendix T, Ohrt-Nissen S, et al. Modic changes are not associated with long-term pain and disability: A cohort study with 13-year follow-up. Spine 2019; 44: 1186-92. doi: 10.1097/BRS.0000000000003051 Viswanathan V K, Shetty A P, Rajasekaran S. Modic changes – an evidencebased, narrative review on its patho-physiology, clinical significance and role in chronic low back pain. J Clin Orthop Trauma 2020; 11(5): 761-9.

Acta Orthopaedica 2021; 92 (2): 247–248 247

Correspondence

Preparation for the next COVID-19 wave: The European Hip Society and European Knee Associates recommendations Knee Surgery, Sport Traumatol Arthrosc 2020; 28(9): 2747–55. doi: 10.1007/s00167-020-06213-z

Sir,—We are writing regarding the publication of “Preparation for the next COVID-19 wave: The European Hip Society en European Knee Associates recommendations” by Simon T Donell et al. (2020). Hospitals throughout the world are making tough choices regarding the care that can be provided during the COVID-19 pandemic, therefore we have read this article with interest. However, we have encountered some issues of concern. COVID-19 has impacted the orthopedic practice by the measurements undertaken to control the virus and its disease. Today with a second phase of COVID-19 with almost no elective orthopedic surgery, evidence-based information is important and much sought after. During a search for literature to support guidelines for the Dutch Orthopaedic Society to restart elective orthopedic surgery in the presence of SARSCoV-2, we came across the article of Donell et al. (2020). The article describes a systematic review and can be found as such in PubMed (PMID: 32803277). However, the article does not follow PRISMA guidance. Even though 61 articles were said to be found, essential items such as study selection, data collection process or a flow-chart, study characteristics and risk of bias assessments weren’t described. At the end of the Materials and methods section it is concluded that: “…any recommendations would be based on expert opinion without any robust independent evidence to support them.” A Result section follows even though it is unclear how the authors came to the content, as it is not at all described in the Materials and methods section. Of the 19 references reported in the Results section, 6 references endorsed the information the authors were referring to (refs 4, 7, 9, 10, 14, 26). The other references did not contain the information as stated in the article. Furthermore, the article contains several sentences that are written as statements which are not corroborated with references. For example: “The possibility of a new slowdown of elective surgery needs to be discussed with the patients, in particular in the most severe ones where delay will lead to a worse outcome” and “Postponing total joint arthroplasty leads to an increase in the use of medication and more unsatisfactory overall outcome. The prolonged time of pain and social isolation, because of immobilisation, risks their mental health.” First of all, to our

knowledge it has never been convincingly shown that a delay in arthroplasty leads to an impaired outcome. Secondly, these statements should be corroborated with evidence or should not be stated as factual. From our standpoint, the conflict of interest statement should contain the fact that several authors of this article occupy an active function within the European Society for Sports Traumatology, Knee Surgery and Arthroscopy (ESSKA) or its affiliated or partner societies (i.e. board member, chair), of which the Journal of Knee Surgery, Sports Traumatology, Arthroscopy (KSSTA) is the official scientific journal. Conflict of interest: The authors certify that they have no affiliations with or involvement in any organisation or entity with any financial interest, or nonfinancial interest in the subject matter or materials discussed in this manuscript. Babette C van der Zwaard 1, Wai-Yan Liu 2,3, Judith Sprengers 1, Nico Verschoor 1, Ruud P van Hove 1 1 Jeroen

Bosch Ziekenhuis, Department of Orthopaedic Surgery, ’s-Hertogenbosch, the Netherlands 2 Máxima Medical Center, Department of Orthopaedic Surgery, Eindhoven, the Netherlands 3 Catharina Hospital, Department of Orthopaedic Surgery, Eindhoven, the Netherlands Email: b.v.d.zwaard@jbz.nl

Sir,—Thank you for the opportunity to respond to this letter which was originally sent to the KSSTA journal but rejected. In essence the complaint is that we have undertaken a systematic review without reporting to the PRISMA guidance. This is correct. However, we would make the following points: 1. The article does not use the term “Systematic review” (SR) in the Title nor the Key words. 2. The SR performed was part of an iterative process to suggest Recommendations to the readers. 3. The SR showed that there was no supportive literature to inform on the recommendations. This can be read in the

248

Abstract. It therefore demonstrated that there was no scientific evidence to support the Recommendations, and that therefore these were based solely on expert opinion. 4. There are space constraints in publishing the Recommendations. To report details of the SR when it concludes that there is no literature to report is pointless. 5. The fact that it is available on PubMed (PMID: 32803277) is nothing to do with the authors but suggests someone feels that listing it is worthwhile. As for mistaking the SR as informing the Recommendations, the key to the Results is the section immediately before headed Consensus. We agree that we did not report the actual process of reaching agreement on the Recommendations, but experts suggesting recommendations and then passing round to the rest for agreement, and then the process of collecting these into a coherent document, is neither interesting nor important. There was not time to undertake a more formal process such as the Delphi method. It should be noted that the time from the first wave to the second was short, and yet the whole process from conception to publication was achieved before the second wave started. The section on the effect of delays on the outcome of arthroplasty presumes that the SR was used to support the views. We

Acta Orthopaedica 2021; 92 (2): 247–248

leave it to the readers to decide whether there has never been any evidence to show “that a delay in arthroplasty leads to an impaired outcome.” KSSTA is the official journal of ESSKA of which the European Knee Associates is a section. It is perfectly reasonable to publish recommendations; conflict of interest is irrelevant. We look forward to the correspondents reporting their recommendations for the Dutch Orthopaedic Society on guidelines on restarting elective surgery in the presence of SARSCoV-2. Simon Donell Norwich Medical School, University of East Anglia, Norwich, UK Email: s.donell@uea.ac.uk

Donell S T, Thaler M, Budhiparama N C, et al. Preparation for the next COVID-19 wave: The European Hip Society and European Knee Associates recommendations. Knee Surgery, Sport Traumatol Arthrosc 2020; 28(9): 2747–55. doi: 10.1007/s00167-020-06213-z

2/21 ACTA ORTHOPAEDICA

Medical

IMPROVE THE CHANCES

REDUCE RISK FOR INFECTION Reduction of infection risk* using dual antibiotic-loaded bone cement in high risk patients

in aseptic revision TKA * as reported in study results

10453

www.heraeus-medical.com

in fractured neck of femur

Vol. 92, No. 2, 2021 (pp. 127–248)

34 % 69 % 57 %

in primary hip & knee arthroplasty

Volume 92, Number 2, April 2021

2102_20479_AZ_COPAL_RFI_Acta_Orthopaedica_Europe_215x280_EN.indd 1 TF-IORT210050.indd 1

23.02.21 08:49

29-04-2021 19:41:39