Acta Orthopaedica, Volume 92, Issue 6 by Acta_Orthopaedica

6/21

Medical ACTA ORTHOPAEDICA

LONGER IMPLANT SURVIVAL. WITH THE RIGHT BONE CEMENT.

Real World Data show:

19%

lower revision risk* with PALACOS® R+G compared to other bone cements

* Calculated difference of cumulative revision rates in knee arthroplasty at 14 years of implantation NJR Data Supplier Feedback (summary reports); Cumulative revision rates (2007–2020) status May 2021. Current report accessible at http://herae.Us/njr-data :H WKDQN WKH SDWLHQWV DQG VWDII RI DOO WKH KRVSLWDOV LQ (QJODQG :DOHV 1RUWKHUQ ,UHODQG DQG WKH ,VOH RI 0DQ ZKR KDYH FRQWULEXWHG GDWD WR WKH 1DWLRQDO -RLQW 5HJLVWU\ :HbDUH JUDWHIXO WR WKH +HDOWKFDUH 4XDOLW\ ,PSURYHPHQW 3DUWQHUVKLS +4,3 WKH 1-5 6WHHULQJ &RPPLWWHH DQG VWDII DW WKH 1-5 &HQWUH IRU IDFLOLWDWLQJ WKLV ZRUN 7KH YLHZV H[SUHVVHG UHSUHVHQW WKRVH RI +HUDHXV 0HGLFDO *PE+ DQG GR QRW QHFHVVDULO\ UHƃHFW WKRVH RI WKH 1DWLRQDO -RLQW 5HJLVWU\ 6WHHULQJ &RPPLWWHH RU WKH +HDOWK 4XDOLW\ ,PSURYHPHQW 3DUWQHUVKLS +4,3 ZKR GR QRW YRXFK IRU KRZ WKH LQIRUPDWLRQ LV SUHVHQWHG

COVER.indd 1

10666

www.heraeus-medical.com

Vol. 92, No. 6, 2021 (pp. 633–765)

The element of success in joint replacement

Volume 92, Number 6, December 2021

27-10-2021 21:25:06

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

EDITORIAL OFFICE

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

THE FOUNDATION BOARD OF THE NORDIC ORTHOPAEDIC FEDERATION AND ACTA ORTHOPAEDICA

EDITOR

Anders Rydholm Lund, Sweden DEPUTY EDITOR

Peter A Frandsen Odense, Denmark CO-EDITORS

Li Felländer-Tsai Stockholm, Sweden Nils Hailer Uppsala, Sweden Ivan Hvid Oslo, Norway Søren Overgaard Copenhagen, Denmark Cecilia Rogmark Malmö, Sweden Urban Rydholm Lund, Sweden Bart A Swierstra Wageningen, The Netherlands Eivind Witsø Trondheim, Norway

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

WEB EDITOR

Magnus Tägil Lund, Sweden STATISTICAL EDITOR

Jonas Ranstam Lund, Sweden Philippe Wagner Västerås, Lund PRODUCTION MANAGER

Kaj Knutson Lund, Sweden Vol. 92, No. 6, 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 92, 2021

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. 6, December 2021

Guest editorial How to play the final chess match—or at least lose with dignity Different, yet strong together: the Nordic Arthroplasty Register Association (NARA)

633 635

C Rogmark and N Lynøe K Mäkelä and N P Hailer

Spine Return to work after lumbar disc herniation surgery: an occupational cohort study

638

R Laasik, P Lankinen, M Kivimäki, M H Neva, V Aalto, T Oksanen, J Vahtera, and K T Mäkelä

644

A N Fraser, B Bøe, T Fjalestad, J E Madsen, and S M Röhrl

651

C Bergdahl, D Wennergren, E Swensson-Backelin, J Ekelund, and M Möller

658

O Pakarinen, O Lainiala, A Reito, P Neuvonen, K Mäkelä, and A Eskelinen

665 673

V J Panula, K J Alakylä, M S Venäläinen, J J Haapakoski, A P Eskelinen, M J Manninen, J S Kettunen, A-P Puhto, A I Vasara, L L Elo, and K T Mäkelä M G Wadström, N P Hailer, and Y D Hailer A-S

678

A-S Neuts, H J Berkhout, A Hartog, and J H M Goosen

681

A Mørup-Petersen, S T Skou, C E Holm, P M Holm, C Varnum, M R Krogsgaard, M Laursen, and A Odgaard

689

D Bron, N Wolterbeek, R Poolman, D Kempen, and D Delawi

695

M Meyer, J Götz, L Parik, Trenkawitz, J Grifka, G Maderbacher, T Kappenschneider, and M Weber

701

H R Mohammad, A Judge, and D W Murray

709

H H Miozzari, C Barea, D Hannouche, and A Lübbeke

716

T Söderman, S Werner, M-L Wretling, M Hänni, C Mikkelsen, A Sundin, and A Shalabi C B Jensen, A Troelsen, P B Petersen, C C Jørgensen, H Kehlet, and K Gromov; Centre for Fast-Track Hip and Knee Replacement Collaborative Group

Shoulder Stable glenoid component of reverse total shoulder arthroplasty at 2 years as measured with model-based radiostereometric analysis (RSA) No change in reoperation rates despite shifting treatment trends: a population-based study of 4,070 proximal humeral fractures Hip Implant survival of 662 dual-mobility cups and 727 constrained liners in primary THA: small femoral head size increases the cumulative incidence of revision Risk factors for prosthetic joint infections following total hip arthroplasty based on 33,337 hips in the Finnish Arthroplasty Register from 2014 to 2018 No increased mortality after total hip arthroplasty in patients with a history of pediatric hip disease: a matched, population-based cohort study on 4,043 patients Bacteriophage therapy cures a recurrent Enterococcus faecalis infected total hip arthroplasty? A case report Measurement properties of UCLA Activity Scale for hip and knee arthroplasty patients and translation and cultural adaptation into Danish Resident training does not influence the complication risk in total knee and hip arthroplasty Postoperative delirium is a risk factor for complications and poor outcome after total hip and knee arthroplasty Knee A matched comparison of the patient-reported outcome measures of 38,716 total and unicompartmental knee replacements: an analysis of linked data from the National Joint Registry of England, Northern Ireland and Isle of Man and England’s National PROM collection programme History of previous surgery is associated with higher risk of revision after primary total knee arthroplasty: a cohort study from the Geneva Arthroplasty Registry Knee function 30 years after ACL reconstruction: a case series of 60 patients Influence of body mass index and age on day-of-surgery discharge, prolonged admission, and 90-day readmission after fast-track unicompartmental knee arthroplasty Fracture Nonoperative management of hip fractures in very frail elderly patients may lead to a predictable short survival as part of advance care planning Trochanteric stabilizing plate in the treatment of trochanteric fractures: a scoping review 30-day and 1-year mortality after skeletal fractures: a register study of 295,713 fractures at different locations Varia Effects of tourniquet inflation on peri- and postoperative cefuroxime concentrations in bone and tissue

722

728

H H Wijnen, P P Schmitz, H Es-Safraouy, L A Roovers, D G Taekema, and J L C Van Susante

733

C E Alm, J-E Gjertsen, T Basso, K Matre, S Rörhl, J E Madsen, and F Frihagen C Bergh, M Möller, J Ekelund, and H Brisby

739 746

P Hanberg, M Bue, J Kabel, A R Jørgensen, C Jessen, K Søballe, and M Stilling

Cat at home? Cat scratch disease with atypical presentations and aggressive radiological findings mimicking sarcoma, a potential diagnostic pitfall Correspondence Isometric hip strength impairments in patients with hip dysplasia are improved but not normalized 1 year after periacetabular osteotomy: a cohort study of 82 patients Errata Physical child abuse demands increased awareness during health and socioeconomic crises like COVID-19: a review and education material (Figure 1 in Supplementary data) Isometric hip strength impairments in patients with hip dysplasia are improved but not normalized 1 year after periacetabular osteotomy: a cohort study of 82 patients (Figure 2) Information to authors (see http://www.actaorthop.org/)

753

F Amerstorfer, J Igrec, T Valentin, A Leithner, L Leitner, M Glehr, J Friesenbichler, I Brcic, and M Bergovec

760

M Zhong and W Zhu versus J S Jacobsen, S S Jakobsen, K Søballe, Per Hölmich, and Kristian Thorborg

763

P Martinkevich, L L Larsen, T Græsholt-Knudsen, G Hesthaven, M B Hellfritzsch, K K Petersen, B Møller-Madsen, and J D Rölfing J S Jacobsen, S S Jakobsen, K Søballe, P Hölmich, and K Thorborg

765

Acta Orthopaedica 2021; 92 (6): 633–634

633

Guest editorial

How to play the final chess match—or at least lose with dignity In this issue of Acta Orthopaedica, Wijnen et al. (2021) report that after nonoperative treatment of hip fractures nine-tenths died during the first month although one-third lived independently before their fracture. This may be startling, but we cannot evade the topic, as we are experiencing a rapid increase of very old citizens who bring their generation’s way of thinking into the discussion of end-of-life care. An old person, a physician, an acute fracture, and death: a serious plot, leading to very different stories in different countries. Sometimes, other actors set the scene: legislation, tradition, family members. The rate of nonoperative treatment of hip fracture varies from 1% to 25% (SFR 2019) (Amrayev et al. 2018, Tomioka et al. 2020), with lowest rates in northern Europe, where non-surgery is regarded as a feature from the past (Jensen and Tøndevold 1980), even if the pandemics have challenged this picture (Mi et al. 2020). Therefore, the increasing interest in the Netherlands in nonoperative treatment is conspicuous (van de Ree et al. 2017, Joosse et al. 2019, van der Zwaard et al. 2020). How well does shared decision-making work in an acute setting? The elderly patient with a hip fracture is in an acute crisis, perhaps ready to choose death over the assumed consequences of a hip fracture (Salkeld et al. 2000). Being in pain in a stressful emergency room, she fears she will lose her independence. She might be so modest that she does not want to be a nuisance to the healthcare system or her family. Laypersons, i.e., family, will be more prone to see risks than benefit with acute surgery, wanting to spare their old family member such strain. Healthcare staff are supposed to respect competent patients’ right to participate in decision-making regarding treatment that concerns the patient—the autonomy principle (Gillon 2003). When in medical need, the patient should be offered treatment, but also has a right to decline treatment. This negative right of a competent patient should be respected. But a surgeon who considers surgery as a way to reduce pain and improve mobility during the remaining lifespan of the patient will find such a decline frustrating. To avoid decisionmaking based on prejudices regarding bad outcome, the physician has to listen carefully to the patient’s ideas, worries, expectations, prejudices, and preferences—without interrupting or arguing with the patient. After summarizing and acknowledging the input from the patient—the first step in patient-centered care (Hedberg and Lynøe 2013)—the physician should inform the patient of what is correct and incorrect regarding factual

Death playing chess. Mural painting by Albertus Pictor around 1480, Täby Church, Sweden

aspects, avoiding tacitly impregnating her or his own values in this presentation (Lynøe et al. 2018). The physician must also ensure that the patient actually understood the information regarding the consequences of the treatment alternatives. If, after the patient-centered procedure, the patient still declines surgery we have to respect a competent patient’s right to do so, when based on an informed and shared decision. When interpreting scientific results, we have to bear in mind basic differences between countries and their health-

© 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 10.1080/17453674.2021.1959159

634

care systems. Wijnen et al. (2021) report an extremely high mortality after non-surgical treatment. Even if 29% of these Dutch patients lived independently before their hip fracture, 87% died during the first month. Previous reports describe a 1-month mortality of 19–36% after nonoperative treatment (Hossain et al. 2009, Loggers et al. 2020), but the case-mix differs. The underlying philosophy of the current study seems to be that suffering a hip fracture can be taken as a possibility to shorten the life of an individual who “has no wish to prolong life.” In other countries, the endeavor is the contrary, for example as stated by Berry et al. (2018): “Surgical repair of a hip fracture was associated with lower mortality among nursing home residents with advanced dementia and should be considered together with the residents’ goals of care in management decisions.” It may be that our understanding of advance care planning differs more than we at first realize (van der Steen et al. 2016). Indeed, culture and legislation differ between counties, and the Netherlands has both leading research activity in advance care planning and a transparent practice of euthanasia (Onwuteaka-Philipsen et al. 2012, Rietjens et al. 2017). In the future we will encounter more patients with hip fractures, most certainly individuals who will express other requests for their final stage of life, and not just suffer in silence. The orthopedic surgeon should prepare by understanding that the attitudes to advance care planning vary, that patients may prefer not to prolong life and ask for palliative care. Others may have an unrealistic view of what could be done to prolong life. Regardless of direction, applying a patient-centered approach, information procedure, and shared decision-making might decrease the orthopedic surgeon’s frustration. Cecilia ROGMARK 1 and Niels LYNØE 2 1 Department of Orthopaedics, Lund University, Skane University Hospital, Malmö, Sweden E-mail: Cecialia.Rogmark@skane.se 2 Department of Learning, Informatics, Management and Ethics, Stockholm Centre for Healthcare Ethics, Karolinska Institutet, Stockholm, Sweden

Amrayev S, AbuJazar U, Stucinskas J, Smailys A, Tarasevicius S. Outcomes and mortality after hip fractures treated in Kazakhstan. HIP Int 2018; 28(2): 205-9. Berry S D, Rothbaum R R, Kiel D P, Lee Y, Mitchell S L. Association of clinical outcomes with surgical repair of hip fracture vs nonsurgical management in nursing home residents with advanced dementia. JAMA Int Med 2018; 178(6): 774-80.

Acta Orthopaedica 2021; 92 (6): 633–634

Gillon R. Ethics needs principles—four can encompass the rest—and respect for autonomy should be “first among equals”. J Med Ethics 2003; 29(5): 307-12. Hedberg C, Lynøe N. What is meant by patient-centredness being valuebased? Scand J Prim Health Care 2013; 31(4): 188-9. Hossain M, Neelapala V, Andrew J. Results of non-operative treatment following hip fracture compared to surgical intervention. Injury 2009; 40(4): 418-21. Jensen J S, Tøndevold E. A prognostic evaluation of the hospital resources required for the treatment of hip fractures. Acta Orthop Scand 1980; 51(16): 515-22. Joosse P, Loggers S A, Van de Ree C M, Van Balen R, Steens J, Zuurmond R G, Gosens T, Van Helden S H, Polinder S, Willems H C. The value of nonoperative versus operative treatment of frail institutionalized elderly patients with a proximal femoral fracture in the shade of life (FRAIL-HIP): protocol for a multicenter observational cohort study. BMC Geriatr 2019; 19(1): 1-12. Loggers S A, Van Lieshout E M, Joosse P, Verhofstad M H, Willems H C. Prognosis of nonoperative treatment in elderly patients with a hip fracture: a systematic review and meta-analysis. Injury 2020; 15(11): 2407-13. Lynøe N, Helgesson G, Juth N. Value-impregnated factual claims may undermine medical decision-making. Clinical Ethics 2018; 13(3): 151-8. Mi B, Chen L, Tong D, Panayi A C, Ji F, Guo J, Ou Z, Zhang Y, Xiong Y, Liu G. Delayed surgery versus nonoperative treatment for hip fractures in post-COVID-19 arena: a retrospective study of 145 patients. Acta Orthop 2020; 91(6): 639-643. Onwuteaka-Philipsen B D, Brinkman-Stoppelenburg A, Penning C, de Jong-Krul G J, van Delden J J, van der Heide A. Trends in end-of-life practices before and after the enactment of the euthanasia law in the Netherlands from 1990 to 2010: a repeated cross-sectional survey. Lancet 2012; 380(9845): 908-15. Rietjens J A, Sudore R L, Connolly M, van Delden J J, Drickamer M A, Droger M, van der Heide A, Heyland D K, Houttekier D, Janssen D J. Definition and recommendations for advance care planning: an international consensus supported by the European Association for Palliative Care. Lancet Onc 2017; 18(9): e543-e51. Salkeld G, Ameratunga S N, Cameron I, Cumming R, Easter S, Seymour J, Kurrle S, Quine S, Brown P M. Quality of life related to fear of falling and hip fracture in older women: a time trade off study commentary: older people’s perspectives on life after hip fractures. BMJ 2000; 320(7231): 341-6. SFR. Swedish Fracture Register (www.frakturregistret.se). Annual Report 2019 (in Swedish). https://www.doi.org/10.18158/BkOFY5vhL Tomioka S, Rosenberg M, Fushimi K, Matsuda S. An analysis of equity in treatment of hip fractures for older patients with dementia in acute care hospitals: observational study using nationwide hospital claims data in Japan. BMC Health Serv Res 2020; 20(1): 1-12. van de Ree C L, De Jongh M A, Peeters C M, de Munter L, Roukema J A, Gosens T. Hip fractures in elderly people: surgery or no surgery? A systematic review and meta-analysis. Ger Orthop Surg Rehab 2017; 8(3): 173-80. van der Steen J T, Galway K, Carter G, Brazil K. Initiating advance care planning on end-of-life issues in dementia: ambiguity among UK and Dutch physicians. Arch Geront Ger 2016; 65:225-30. van der Zwaard B C, Stein C E, Bootsma J E, van Geffen H J, Douw C M, Keijsers C J. Fewer patients undergo surgery when adding a comprehensive geriatric assessment in older patients with a hip fracture. Arch Orthop Trauma Surg 2020; 140(4): 487-92. Wijnen H H, Schmitz P P, Es-Safraouy H, Roovers E A, Taekema D G, van Susante J L C. Nonoperative management of hip fractures in very frail elderly patients may lead to a predictable short survival as part of advance care planning. Acta Orthop 2021. Epub ahead of print. DOI: 10.1080/17453674.2021.1959155

Acta Orthopaedica 2021; 92 (6): 635–637

635

Editorial

Different, yet strong together: the Nordic Arthroplasty Register Association (NARA) Development of NARA The Nordic Arthroplasty Register Association (NARA) was established in 2007 by hip and knee arthroplasty registry leaders from Denmark, Norway, and Sweden. Purportedly, the idea of combining national databases was presented and discussed while sitting in a bar. The 1st NARA manuscript by Havelin et al. (2009), published in Acta Orthopaedica, described differences in the demographics of patients receiving total hip arthroplasty (THA) among these 3 participating countries. The paper was well read in Finland, especially by author KM of this editorial, with disbelief and envy. It became obvious that the other Scandinavian registries were flourishing while the Finnish Arthroplasty Register (FAR) was suffering. The Finnish arthroplasty surgeons started to develop FAR in collaboration with the National Institute for Health and Welfare in Helsinki, Finland became a preliminary member at NARA meetings around 2010, and FAR data became electronic in 2014 (Finnish Arthroplasty Register 2021). Simultaneously, the scientific work of NARA reached high standards, concerning both quality and quantity (Havelin et al. 2009, Dale et al. 2012, Mäkelä et al. 2014a, Lazarinis et al. 2017). Frequent face-to-face meetings enabled mutual confidence building between collaborators. The Scandinavians allowed NARA to be under Finnish leadership for the time period 2014–2020. “NordForsk,” a funding agency under the Nordic Council of Ministers, financially supported NARA from 2014 to 2016, and currently all participating registries support NARA financially and logistically, but there is no central funding. All Nordic countries have similar state-funded public healthcare systems, but there are large dissimilarities between the participating countries when it comes to the practice of orthopedics, such as for example concerning fixation techniques and surgical approaches: the use of uncemented THA fixation is much more frequent in Denmark and Finland than in Norway and Sweden (Mäkelä 2014a), and the use of the posterior approach to the hip dominates in Denmark, whereas almost half of the Swedish exposures are by direct lateral approaches (Swedish Hip Arthroplasty Register 2019, Danish Hip Arthroplasty Register 2020). In total knee arthroplasty (TKA), the use of cemented fixation and of patellar resurfacing varies considerably between NARA countries (Irmola et

al. 2021). These differences result in what is termed a “natural experiment” by epidemiologists, such that country-wise outcomes can be investigated (Mäkelä et al. 2014b). Today, in June 2021, there are more than 50 scientific NARA publications, and many of them have influenced treatment practices, at least in the participating countries. For example, in 2014 a NARA paper concerning THA fixation methods in elderly patients showed that the proportion of THAs with uncemented implants had increased from 10% to 39% between 1995 and 2011, although survival for uncemented implants was much lower compared with cemented implants in patients aged ≥ 65 years (Mäkelä et al. 2014a). Gradually, cemented stems were favored in the elderly based on these and other similar data, especially in Denmark and Finland where uncemented fixation was common even in patients with hip fractures (Danish Hip Arthroplasty Register 2020, Finnish Arthroplasty Register 2021). The role of hydroxyapatite coating in THA has been mythbusted based on NARA data. Hydroxyapatite-coated femoral stems and cups of different brands did not reduce the risk of revision due to aseptic loosening (Hailer et al. 2015, Lazarinis et al. 2017). These, together with other similar findings, suggested that hydroxyapatite coating does not render primary hip implants more stable. Reducing dislocation rates is one of the major remaining challenges when developing modern THA further. Based on NARA data, Kreipke et al. (2019) showed that osteoarthritis patients operated on with a dual mobility cup (DMC) had a lower risk of revision due to dislocation, but a higher risk of revision caused by infection. Similarly, the use of a DMC in the treatment of patients with displaced femoral neck fractures was associated with a lower risk of revision, both for any reason and due to dislocation (Jobory et al. 2019). The total of 4,500 patients with DMC analyzed by use of the NARA database exceeded any cohort of DMC-treated hip fracture patients published previously. NARA has increased focus towards research on TKA: In agreement with data from outside the Nordic countries (Nugent et al. 2019), Niemeläinen et al. (2020) found that both cemented and hybrid TKA had excellent 10-year survival rates in patients below 65 years, and that cemented TKA still deserves the status of gold standard in working-age patients.

636

Another NARA paper indicated that totally uncemented TKA fixation was associated with an increased risk of revision when compared with cemented fixation in elderly TKA patients (Irmola et al. 2021). After the success of NARA research on THA and TKA, the Nordic shoulder registries also joined the NARA collaboration. Using the NARA shoulder database, Lehtimäki et al. (2018) found that the overall mid-term risk of revision after reverse shoulder arthroplasty performed for rotator cuff tear arthropathy was low, and they identified patient factors associated with an increased risk of revision.

NARA: the future National datasets can provide an impressive number of observations, but conclusions are still limited when investigating rare outcomes. For instance, 90-day mortality after THA is below 0.5% (Pedersen et al. 2020) and revision due to periprosthetic joint infection affects only about 1–2% of arthroplasty patients (Dale et al. 2012). Therefore, associations of any exposure of interest with these outcomes will be muddled by large estimation uncertainty if examined on a national level alone, at least in the Nordic countries with their small populations. The NARA database provides a better opportunity to investigate rare events such as periprosthetic joint infection and mortality. The seminal paper by Dale et al. (2012) describing the increased risk of periprosthetic joint infections in all Nordic countries after the turn of the century will soon be followed by a contemporary in-depth analysis of this issue, again based on data from all NARA countries. Designing data exchanges such as this on an international platform brings about its own difficulties in terms of jurisdiction and data governance, and administrative and regulatory problems related to sharing biometric data across borders are huge. Nonetheless, numerous challenges in modern orthopedics can only be addressed by joining forces across national borders. The already established collaboration of the Nordic countries within the NARA can provide the framework for such endeavors, even including joint ventures with other established national registries, such as the Australian Orthopaedic Association National Joint Replacement Registry (Ackerman et al. 2017) and the Dutch Arthroplasty Register (Van Steenbergen et al. 2021). The new EU Medical Device Regulation 2017/745 (MDR) concerns all participating NARA countries. MDR represents a keystone for future medical device management in Europe and worldwide, and NARA members are strongly committed to ensuring that implants and surgical procedures are both safe and efficient. Registry-based research has provided an enormous impetus to the advancement of modern orthopedics, and all modern standards of post-market surveillance and early detection of underperforming implants are based on the

Acta Orthopaedica 2021; 92 (6): 635–637

continuous collection of routine data by all involved national healthcare providers. Nonetheless, registry research is observational, by definition failing to meet the rigorous standards of a scientific experiment, and the conclusions of most registry research are thus limited by the fact that causal relationships between exposures and outcomes cannot be inferred. Given these well-known limitations of observational research it is not only reasonable but even necessary to explore the potential of registries to conduct hypothesis-testing, experimental research; thus, the term “registry-nested trial” was coined (James et al. 2015). Several Nordic national registries already provide the technical platforms needed to perform such registry-nested trials. Examples are 2 registry-based randomized controlled trials (RCT) on patients with femoral neck fractures that are nested within the Swedish Fracture Register (SFR) and the Swedish Hip Arthroplasty Register (SHAR), the “duality” and the “Hipsther” studies (Wolf et al. 2020a, 2020b). Both these studies recruit patients via a screening and randomization platform within the SFR, and the analysis of endpoints such as reoperations or implant revisions is performed via the SHAR and the National Patient Register. In Norway, a registry-nested RCT comparing the risk of periprosthetic joint infection after the use of either antibiotic-loaded or plain bone cement in the fixation of total knee arthroplasty has been initiated (Leta et al. 2021), and a Danish study comparing different regimes of perioperative antibiotics during THA surgery is in the pipeline. Future Nordic RCTs can be designed within the NARA framework, with screening and inclusion of patients performed in all 4 participating countries, and with registration of demographic and procedural details following the established routines within each participating national registry. Outcome analyses will be based either on the already established common NARA dataset, or on extended datasets, where appropriate. In all, the NARA provides a stable framework for observational arthroplasty research, it has changed practice, at least in the participating countries, and it is a perfect platform for future experimental studies. We look forward to further contributions from this Nordic goldmine, and are open to joint ventures with other arthroplasty registries. Keijo Mäkelä Past NARA Chairman, Department of Orthopaedics and Traumatology, Turku University Hospital and University of Turku, Turku, Finland Nils P. Hailer Co-Editor/NARA Chairman, Department of Surgical Sciences—Orthopaedics, Uppsala University, Uppsala, Sweden Correspondence: nils.hailer@surgsci.uu.se

Acta Orthopaedica 2021; 92 (6): 635–637

Ackerman I N, Bohensky M A, de Steiger R, Brand C A, Eskelinen A, Fenstad A M, Furnes O, Graves S E, Haapakoski J, Mäkelä K, Mehnert F, Nemes S, Overgaard S, Pedersen A B, Garellick G. Lifetime risk of primary total hip replacement surgery for osteoarthritis from 2003 to 2013: a multinational analysis using national registry data. Arthritis Care Res 2017; 69(11): 1659-67. doi: 10.1002/acr.23197. Dale H, Fenstad A M, Hallan G, Havelin L I, Furnes O, Overgaard S, Pedersen A B, Karrholm J, Garellick G, Pulkkinen P, Eskelinen A, Mäkelä K, Engesaeter L B. Increasing risk of prosthetic joint infection after total hip arthroplasty. Acta Orthop 2012; 83(5): 449-58. doi: 10.3109/17453674.2012.733918. Danish Hip Arthroplasty Register. National Annual Report; 2020. Finnish Arthroplasty Register. ENDOnet 2021. https://www.thl.fi/ far/#index Hailer N P, Lazarinis S, Mäkelä K T, Eskelinen A, Fenstad A M, Hallan G, Havelin L, Overgaard S, Pedersen A B, Mehnert F, Kärrholm J. Hydroxyapatite coating does not improve uncemented stem survival after total hip arthroplasty! Acta Orthop 2015; 86(1): 18-25. doi: 10.3109/17453674.2014.957088. Havelin L I, Fenstad A M, Salomonsson R, Mehnert F, Furnes O, Overgaard S, Pedersen A B, Herberts P, Kärrholm J, Garellick G. The Nordic Arthroplasty Register Association: a unique collaboration between 3 national hip arthroplasty registries with 280,201 THRs. Acta Orthop 2009; 80(4): 393-401. doi: 10.3109/17453670903039544. Irmola T, Ponkilainen V, Mäkelä K T, Robertsson O, W-Dahl A, Furnes O, Fenstad A M, Pedersen A B, Schroder H M, Eskelinen A, Niemeläinen M J. Association between fixation type and revision risk in total knee arthroplasty patients aged 65 years and older: a cohort study of 265,877 patients from the Nordic Arthroplasty Register Association 2000–2016. Acta Orthop 2021; 92(1): 91-6. doi: 10.1080/17453674.2020.1837422. James S, Rao S V, Granger C B. Registry-based randomized clinical trials: a new clinical trial paradigm. Nature Rev Cardiol 2015; 12(5): 312-6. doi: 10.1038/nrcardio.2015.33. 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. doi: 10.2106/JBJS.18.00614. 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 population-based study using the Nordic Arthroplasty Register Association database. J Bone Joint Surg Am 2019; 101(2): 169-76. doi: 10.2106/ JBJS.17.00841. Lazarinis S, Mäkelä K T, Eskelinen A, Havelin L, Hallan G, Overgaard S, Pedersen A B, Kärrholm J, Hailer N P. Does hydroxyapatite coating of uncemented cups improve long-term survival? An analysis of 28,605 primary total hip arthroplasty procedures from the Nordic Arthroplasty Register Association (NARA). Osteoarthritis Cartilage 2017; 25(12): 1980-7. doi: 10.1016/j.joca.2017.08.001. Lehtimäki K, Rasmussen J V, Mokka J, Salomonsson B, Hole R, Jensen S L, Aarimaa V. Risk and risk factors for revision after primary reverse shoulder arthroplasty for cuff tear arthropathy and osteoarthritis: a Nordic Arthroplasty Register Association study. J Shoulder Elbow Surg 2018; 27(9): 1596-601. doi: 10.1016/j.jse.2018.02.060.

637

Leta T H, Gjertsen J E, Dale H, Hallan G, Lygre S H L, Fenstad A M, Dyrhovden G S, Westberg M, Wik T S, Jakobsen R B, Aamodt A, Röhrl S M, Gøthesen O J, Lindalen E, Heir S, Ludvigsen J, Bruun T, Hansen A K, Aune K E M, Warholm M, Skjetne J P, Badawy M, Hovding P, Husby O S, Karlsen Ø E, Furnes O. Antibiotic-loaded bone cement in prevention of periprosthetic joint infections in primary total knee arthroplasty: a register-based multicentre randomised controlled non-inferiority trial (ALBA trial). BMJ Open 2021; 11(1): e041096. doi: 10.1136/bmjopen-2020-041096. 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 (Clin Res ed.) 2014a; 348:f7592. doi: 10.1136/bmj.f7592. Mäkelä K T, Matilainen M, Pulkkinen P, Fenstad A M, Havelin L I, Engesaeter L, Furnes O, Overgaard S, Pedersen A B, Kärrholm J, Malchau H, Garellick G, Ranstam J, Eskelinen A. Countrywise results of total hip replacement: an analysis of 438,733 hips based on the Nordic Arthroplasty Register Association database. Acta Orthop 2014b; 85(2): 107-16. doi: 10.3109/17453674.2014.893498. Niemeläinen M J, Mäkelä K T, Robertsson O, W-Dahl A, Furnes O, Fenstad A M, Pedersen A B, Schrøder H M, Reito A, Eskelinen A. The effect of fixation type on the survivorship of contemporary total knee arthroplasty in patients younger than 65 years of age: a register-based study of 115,177 knees in the Nordic Arthroplasty Register Association (NARA) 2000–2016. Acta Orthop 2020: 1-7. doi: 10.1080/17453674.2019.1710373. Nugent M, Wyatt M C, Frampton C M, Hooper G J. Despite improved survivorship of uncemented fixation in total knee arthroplasty for osteoarthritis, cemented fixation remains the gold standard: an analysis of a national joint registry. J Arthroplasty 2019; 34(8): 1626-33. doi: 10.1016/j. arth.2019.03.047. Pedersen A B, Mailhac A, Garland A, Overgaard S, Furnes O, Lie S A, Fenstad A M, Rogmark C, Kärrholm J, Rolfson O, Haapakoski J, Eskelinen A, Mäkelä K T, Hailer N P. Similar early mortality risk after cemented compared with cementless total hip arthroplasty for primary osteoarthritis: data from 188,606 surgeries in the Nordic Arthroplasty Register Association database. Acta Orthop 2020: 1-7. doi: 10.1080/17453674.2020.1842003. Swedish Hip Arthroplasty Register. Annual Report 2019. https://shpr.registercentrum.se/in-english/annual-reports/p/rkeyyeElz Van Steenbergen L N, Mäkelä K T, Kärrholm J, Rolfson O, Overgaard S, Furnes O, Pedersen A B, Eskelinen A, Hallan G, Schreurs B W, Nelissen R. Total hip arthroplasties in the Dutch Arthroplasty Register (LROI) and the Nordic Arthroplasty Register Association (NARA): comparison of patient and procedure characteristics in 475,685 cases. Acta Orthop 2021; 92(1): 15-22. doi: 10.1080/17453674.2020.1843875. Wolf O, Mukka S, Notini M, Möller M, Hailer N P, duality group. Study protocol: the DUALITY trial—a register-based, randomized controlled trial to investigate dual mobility cups in hip fracture patients. Acta Orthop 2020a: 1-8. doi: 10.1080/17453674.2020.1780059. 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 2020b; 20(1): 19. doi: 10.1186/s12877-020-1418-2.

638

Acta Orthopaedica 2021; 92 (6): 638–643

Return to work after lumbar disc herniation surgery: an occupational cohort study Raul LAASIK 1, Petteri LANKINEN 2,3, Mika KIVIMÄKI 4–6, Marko H NEVA 1, Ville AALTO 4, Tuula OKSANEN 4,7, Jussi VAHTERA 8, a, and Keijo T MÄKELÄ 2, a 1 Department

of Orthopaedics and Trauma, Tampere University Hospital, Tampere, Finland; 2 Department of Orthopedics and Traumatology, Turku University Hospital and University of Turku, Turku, Finland; 3 Satakunta Central Hospital, Pori, Finland; 4 Finnish Institute of Occupational Health, Helsinki, Finland; 5 Clinicum, Faculty of Medicine, University of Helsinki, Helsinki, Finland; 6 Department of Epidemiology and Public Health, University College London, London, UK; 7 Institute of Public Health and Clinical Nutrition, University of Eastern Finland, Kuopio, Finland; 8 Department of Public Health, University of Turku, and Centre for Population Health Research, University of Turku and Turku University Hospital; Turku, Finland a Joint senior authors Correspondence: keijo.makela@tyks.fi Submitted 2021-01-12. Accepted 2021-06-23.

Background and purpose — Lumbar disc herniation is a common surgically treated condition in the working-age population. We assessed health-related risk factors for return to work (RTW) after excision of lumbar disc herniation. Previous studies on the subject have had partly contradictory findings. Patients and methods — RTW of 389 (n = 111 male, n = 278 female; mean age 46 years, SD 8.9) employees who underwent excision of lumbar disc herniation was assessed based on the Finnish Public Sector Study (FPS). Baseline information on occupation, preceding health, and health-risk behaviors was derived from linkage to national health registers and FPS surveys before the operation. The likelihood of RTW was analyzed using Cox proportional hazard univariable and multivariable modelling. Results — 95% of the patients had returned to work at 12 months after surgery, after on average 78 days of sickness absence. Faster RTW in the univariable Cox model was associated with a small number of sick leave days (< 30 days) before operation (HR 1.3, 95% CI 1.1–1.6); high occupational position (HR 1.6, CI 1.2–2.1); and age under 40 years (HR 1.5, CI 1.1–1.9). RTW was not associated with sex or the health-related risk factors obesity, physical inactivity, smoking, heavy alcohol consumption, poor self-rated health, psychological distress, comorbid conditions, or purchases of pain or antidepressant medications in either the univariable or multivariable model. Interpretation — Almost all employees returned to work after excision of lumbar disc herniation. Older age, manual job, and prolonged sick leave before the excision of lumbar disc herniation were risk factors for delayed return to work after the surgery.

Return to work (RTW) is an important outcome of lumbar disc herniation surgery, and a key metric for its effectiveness, as it has profound implications for both individual patients and the economy at large. An early RTW is associated with beneficial effects on patients’ physical and mental health and social and economic benefits (Liang et al. 1986, Koenig et al. 2016, Khan et al. 2019). Favorable outcome of disc herniation surgery when compared with nonoperative treatment was already presented in the 1980s (Weber 1983). This finding has also been recently confirmed in an RCT setting (Bailey et al. 2020). The difference in outcomes between surgically and nonoperatively treated patients may diminish in longer follow-up (Österman et al. 2006). Long duration of leg pain and long preoperative sick leave increase the risk of not returning the work (Kotilainen et al. 1993, Nygaard et al. 2000, Khan et al. 2019). Furthermore, if a worker is on sick leave more than 6 months after the operation, the probability of not returning to work is as high as 50% (Frank et al. 1996). Therefore, the main indication for elective disc herniation surgery is fast relief of the symptoms to enable early RTW and prevent the development of permanent work disability. There are several suggested factors associated with prolonged sick leave such as postoperative leg pain, poor work motivation, and female sex (Graver et al. 1998, Puolakka et al. 2008, Huysmans et al. 2018, Khan et al. 2019). Identifying factors predicting RTW may help in patient selection and setting adequate goals for rehabilitation after the surgery. Previous studies on the subject have had partly contradictory findings. We therefore assessed health-related risk factors of RTW after lumbar disc herniation surgery, such as factors related to general health, health-risk behaviors, and socioeconomic status in a large cohort of public sector employees. This is a linkage study of national health registers and FPS surveys.

Acta Orthopaedica 2021; 92 (6): 638–643

Patients and methods Patients Patients were identified from the Finnish Public Sector (FPS) study, a nationwide register- and survey-based cohort among employees of 10 municipalities and 6 hospital districts covering a wide range of occupations—from city mayors and doctors to semiskilled cleaners, nurses, and teachers (Airaksinen et al. 2017, Laasik et al. 2019, Lankinen et al. 2019). The cohort members were employed for a minimum of 6 months in the participating organizations between 1991 and 2005 (n = 151,901). Since 1997/1998, repeated questionnaire data at 2- to 4-year intervals has been collected from all employees at work at the time of the survey. Further, the questionnaires were filled out between 1997 and 2005 and the patients underwent surgery between 1999 and 2010. Unfortunately, we do not know the exact time between completing the questionnaire and the surgery. We derived information on baseline characteristics before the surgery from the closest survey responses, the employers’ records, and national health registers. All participants were linked to data on lumbar disc herniation surgery from the National Hospital Discharge Register, maintained by the National Institute for Health and Welfare, as well as the National Sickness Absence Register, maintained by the Social Insurance Institution of Finland, where all sickness absence periods are medically certified and they are encoded to the register with start and end dates (Laasik et al. 2019, Lankinen et al. 2019). The linkage data were available until December 31, 2011. The ethics committee of the Hospital District of Helsinki and Uusimaa approved the study. Type of surgery and patient characteristics Of the FPS cohort members, 1,706 underwent excision of lumbar disc herniation between 1999 and 2010. Of these, 389 (n = 111 male, n = 278 female; mean age 46y, SD 8.9, range 22–64) had responded to a survey before the surgery and were included in the study. The type of surgery was defined as ABC16 or ABC26 (NOMESCO Classification of Surgical Procedures Version 1.14; the Nordic Medico-Statistical Committee). Return to work (RTW) RTW was determined as how long it took for the participants to return to work, if at all, after the surgery, that is the number of days between the date of discharge and the date of the end of the sick leave. The follow-up time for RTW was 12 months. All Finnish residents aged 16 to 67 years are legitimized to receive daily allowances due to medically certified sickness absence. After a qualifying period of the first 9 days of illness, compensation is paid based on salary for a maximum of 1 year. Overlapping and consecutive periods of sick leave were merged. At 12 months it was dichotomized as yes vs. no, depending on whether or not the patient had returned to work.

639

Risk factors for RTW The participants’ age, sex, and occupational title at the time of the surgery were obtained from the employers’ registers. To measure occupational status, occupational titles, coded according to the International Standard Classification of Occupation (ISCO), were categorized into three groups: higher-grade non-manual workers (e.g., teachers, physicians), lower-grade non-manual workers (e.g., registered nurses, technicians), and manual workers (e.g., cleaners, maintenance workers). Marital status (married or cohabiting vs. single, divorced, or widowed) was obtained from the baseline questionnaire. Age at the time of disc herniation surgery was categorized as age groups < 40 years, 40–50 years, > 50 years. Information on health and health behaviors was obtained from the baseline questionnaire and national health registers described in detail earlier. Briefly, physical activity was defined as average weekly hours of leisure-time physical activity, categorized into 2 groups, “low” (<1 4 MET/hours/ week) and “high” activity (> 14 MET/hours/week) (Kujala et al. 1998). Alcohol consumption was categorized according to the habitual frequencies of drinking beer, wine, and spirits as “none,” “moderate,” and “heavy” consumption. The cut-off for heavy alcohol consumption was set as 210 g/week (Rimm et al. 1999). Smoking status was dichotomized as “currently smoking” vs. “has quit or never smoked.” Self-reported bodyweight and height were used to calculate BMI, which was used to identify obese (BMI ≥ 30) and non-obese (BMI < 30) participants. Psychological distress was measured with the 12-item version of the General Health Questionnaire (GHQ) (Goldberg et al. 1997), using 3/4 positive responses as a cutoff point of psychological distress (“yes” vs. “no”). Participants rated their general health on a 5-point scale (from 1 = “good health” to 5 = “worst health”), and the self-rated health was then dichotomized by categorizing response scores 1 and 2 as good and scores 3 to 5 as poor self-rated health. Data on comorbid chronic conditions—diabetes, coronary heart disease, asthma, chronic obstructive pulmonary disease, and rheumatoid arthritis—was obtained from the Drug Reimbursement Register, which contains information on persons entitled to special reimbursement for treatment of severe chronic illnesses. The presence of comorbidity was then dichotomized as “yes” vs. “no.” Antidepressant and pain medication prescriptions within 100 days of surgery were obtained from Social Insurance Institution of Finland. Statistics The participants were followed from the date of the discharge between January 1, 1999 and December 31, 2010 to the date when an employee returned to work, was granted a disability pension, an old-age pension, died, or end of study, December 31, 2011, whichever came first. A univariable Cox proportional hazards regression model was used for estimation of possible risk factors and hazard ratios with 95% confidence intervals (CI) for RTW (Table). Directed acyclic graph (DAG)

640

Acta Orthopaedica 2021; 92 (6): 638–643

Sociodemographic factors Sex Age Occupation Preoperative sickness abscence

Health risk behaviors RTW Health

Directed acyclic graph (DAG). Age, sex, and occupational status were considered as sociodemographic factors in DAG, whereas alcohol consumption, smoking, and physical activity were considered as health risk behaviors. Further, obesity, self-rated health, psychological distress, comorbidities, antidepressant medication, and pain medication were considered as health in the DAG. RTW = return to work.

analysis on the risk factors for RTW was performed based on the literature and clinical practise to organize variables according to their supposed relation to RTW and to other variables (Figure). For all the variables with potential confounding bias we performed a multivariable analysis (Table). Age, sex, and occupational status were considered as sociodemographic factors in DAG, whereas alcohol consumption, smoking, and physical activity were considered as health-risk behaviors. Further, obesity, self-rated health, psychological distress, comorbidities, antidepressant medication, and pain medication were considered as health in the DAG. All analyses were performed using the SAS statistical software, version 9.1.3 (SAS Institute, Cary, NC, USA). Ethics, funding, and potential conflicts of interest Each author certifies that his or her institution has approved the human protocol and that all investigations were conducted in conformity with ethical principles of research. Conflicts of interest and source of funding: None.

Results The average age of the patients at the time of surgery was 46 years (SD 8.9, range 22–64) and 72% were women. One third of the patients were in manual work. Almost half had been on sick leave before the operation and 2 out of 3 had used pain medication. Among the patients, 12% had some chronic medical comorbidity, 33% rated their health as poor, and 29% were psychologically distressed. Moreover, 15% were obese, and over 20% were smoking or physically inactive. Purchased prescribed antidepressants were relatively rare (6%) prior to surgery (Table). Risk factors for RTW At 12-month follow-up after the surgery, 371 out of 389 patients (95%) had returned to work after being on sick leave for 78 (SD 77, range 13 to 366) days on average. Occupational status, sickness absence before the surgery, and age were asso-

Baseline characteristics of the patients (n = 389) and their associations with the rate of return to work after excision of lumbar disc herniation. Hazard ratios (HR) and their 95% confidence intervals (CI) are derived from Cox proportional hazard uni- and multivariable analyses Factor n (%)

Separately analyzed HR (CI)

Multivariable model HR (CI)

Age (missing n = 0): < 40 102 (26) 1.5 (1.1–1.9) 1.4 (1.1–1.9) 40–50 138 (36) 1.3 (1.0–1.7) 1.3 (1.0–1.7) > 50 149 (38) reference reference Sex (missing n = 0): Men 111 (29) 1.1 (0.8–1.3) 1.1 (0.8–1.4) Women 278 (72) reference reference Married or cohabiting (missing n = 0): No 100 (26) 1.2 (0.9–1.5) 1.1 (0.8–1.4) Yes 289 (74) reference reference Obese (BMI > 30) (missing n = 11): No 322 (85) 1.2 (0.9–1.6) 1.2 (0.9–1.7) Yes 56 (15) reference reference Comorbidities (missing n = 0) No 341 (88) 1.2 (0.9–1.7) 1.1 (0.7–1.5) Yes 48 (12) reference reference Current smoking (missing n = 4): No 301 (78) 1.2 (0.9–1.5) 1.0 (0.8–1.4) Yes 84 (22) reference reference High alcohol consumption (missing n = 2) No 349 (90) 1.0 (0.7–1.4) 0.9 (0.6–1.4) Yes 38 (10) reference reference Occupational status (missing n = 4): Non-manual Higher level 91 (24) 1.6 (1.2–2.1) 1.6 (1.2–2.1) Lower level 142 (37) 1.3 (1.1–1.7) 1.3 (1.0–1.7) Manual 152 (40) reference reference Self-rated health (missing n = 2): Good 261 (67) 1.1 (0.9–1.4) 1.0 (0.7–1.2) Poor 126 (33) reference reference Psychological distress (missing n = 1): No 277 (71) 1.1 (0.8–1.3) 1.1 (0.9–1.5) Yes 111 (29) reference reference Physically active (MET hours >14 h/week) (missing n = 4): Yes 295 (77) 1.0 (0.8–1.3) 1.0 (0.8–1.3) No 90 (23) reference reference Antidepressant medication purchase within 100 days (missing n = 0) No 365 (94) 1.1 (0.7–1.7) 1.0 (0.6–1.7) Yes 24 (6) reference reference Pain medication purchase within 100 days (missing n = 0): No 144 (37) 0.9 (0.8–1.1) 1.0 (0.8–1.3) Yes 245 (63) reference reference Preoperative sickness absence a (missing n = 0) No 221 (57) 1.3 (1.1–1.6) 1.3 (1.0–1.6) Yes 168 (43) reference reference a More

than 30 days of sickness absence within the 1-year period preceding the operation.

ciated with RTW in both a univariable and a multivariable model. Patients with higher level non-manual occupational status had a 1.6 (CI 1.2–2.1) times higher hazard of RTW than manual workers; in patients with ≤ 30 days’ sick leave before the surgery the corresponding hazard was 1.3-fold (CI 1.1– 1.6) compared with those with a longer sick leave. Patients aged < 40 years returned to work sooner than patients aged > 50 years (HR 1.5 (CI 1.1–1.9) (Table).

Acta Orthopaedica 2021; 92 (6): 638–643

In contrast, socioeconomic factors (sex and marital status), health behaviors (obesity, smoking, physical activity, alcohol consumption, self-related health), chronic medical comorbidities (asthma, diabetes mellitus, rheumatoid arthritis, and coronary artery disease), or psychological distress and purchase of pain or antidepressant medications were not associated with RTW (Table).

Discussion This occupational cohort study among 389 public sector employees who underwent lumbar disc herniation surgery showed that 95% of them returned to work within the 12-month follow-up, and on average 78 days (2.5 months) after the surgery. The average sick leave duration is comparable to those published earlier (Huysmans et al. 2018, Than et al. 2016). Younger age, a smaller number of sickness absence days before the surgery, and a higher occupational status were associated with the rate of return to work. In turn, demographic and socioeconomic factors (sex and marital status), health behaviors (smoking, physical inactivity, and excessive alcohol consumption), comorbidities, self-rated poor health, psychological distress, and antidepressant or pain medication prescriptions were not associated with RTW. Several studies have previously examined predictors of RTW after lumbar discectomy with partly contradictory findings. Than et al. (2016) assessed predictors of RTW at 3 months in 105 patients based on a US neurosurgical registry, and found that younger age was the only statistically significant predictor of postoperative RTW. Sex, BMI, smoking status, and comorbidity were not associated with RTW. In their study, 94% of the patients had returned to work at 12-month follow-up (average 67 days). Also, Paulsen et al. (2020) found in a Danish study of 146 patients that sex, BMI, and smoking status were not associated with RTW. Our results are in accordance with these previous studies, although Than et al. (2016) did not examine occupational status or preoperative sickness absence. Paulsen et al. (2020) stated further that preoperative sick leave was not associated with the duration of postoperative sick leave, whereas occupational status was associated. The latter is in line with our findings, as well as that of Nygaard et al. (2000) who stated that patients with sick leave of more than 7 months before the surgery for lumbar disc herniation surgery were at higher risk of not returning to work. Contrary to our findings and those of Than et al. (2016) and Paulsen et al. (2020), Sørlie et al. (2012) stated that smoking may be a predictor of prolonged RTW after lumbar microdiscectomy. They assessed 178 Norwegian patients and 25 of the smokers received sickness benefit after 12 months, compared with 19 of the non-smokers. However, their main interest was residual back pain, and potential confounders such as occupational status were not addressed, which may cause bias. Further evidence on the harmful effect of smoking on RTW was pre-

641

sented by Dewing et al. (2008) who claimed that smokers had a lower return to full active military duty after lumbar microdiscectomy (84% had returned to unrestricted duty at 3 years). However, this very young (mean age 27 years) and active population differs basically from that of ours (mean age 46 years, public sector workers), which may explain the difference. Schade et al. (1999) studied 46 patients undergoing lumbar discectomy with pain relief, reduction of disability in daily activities, and RTW as outcomes at 2 years. MRI-identified nerve root compromise and social support from the spouse were independent predictors of pain relief, whereas RTW at 2 years after surgery was best predicted by preoperative depression and work stress. They stated that the most important finding of the study was that RTW was not influenced by any clinical findings or MR-identified morphological alterations, but solely by psychological factors (i.e., depression) and psychological aspects of work (i.e., work stress). Our results are not consistent with these findings. In our study, psychological distress or antidepressant medication were not associated with RTW. It should be noted that compared with our study, their sample size was remarkably smaller (46 vs. 389 patients), which may have influenced results and caused bias due to chance findings or selection of patients. Vucetic et al. (1999) found in 156 patients that factors predicting RTW at 2 years after surgery for lumbar disc herniation were: no preoperative comorbidity, duration of sciatica less than 7 months, higher education, age younger than 41 years, male sex, and no previous non-spinal surgery. In our study, we did not have data on duration of symptoms, but we considered the length of the preoperative sickness absence as a proxy. Further, we measured socioeconomic status from occupational status and not from education. Longer time on preoperative sick leave was also a predictor for decreased working capacity after lumbar disc herniation surgery in the Swedish population (Silverplats et al. 2010). Also, Schoeggl et al. (2003) found that patients with strenuous occupations had a decreased RTW compared with patients with less strenuous or sedentary occupations. Sex was not associated with RTW in our study, nor in the study of Than et al. (2016) or Paulsen et al. (2020). Female sex was associated with not returning to work in the study of Jensdottir et al. (2007), but their patients had been operated on in the early 1980s in Iceland, which may explain the finding. Comorbidity was not an issue based on our data, contrary to that of Vucetic et al. (1999). Our study patients were solely public sector workers (no farmers, construction workers, or soldiers), which may reduce the generalizability of the findings to other populations. Our results need to be interpreted with some strengths and limitations in mind. Compared with previous studies, our cohort size was the second largest we are aware of after that of Schoeggl et al. (2003). The studied sample came from a wellcharacterized occupational cohort and represented a wide range of occupations. Comprehensive data had been gathered before the surgery on health and health-risk behaviors. All

642

data was linked to reliable national health registers including detailed information regarding the date of the operation and the beginning and end dates of all periods of sickness absence, enabling accurate estimation of the timing of return to work. Many predictors of RTW, such as occupational status, sickness absences, antidepressant and pain medications before the operation, and comorbid medical conditions, were measured objectively from the registers. As a limitation, the generalizability of our findings may be affected by the differences in welfare, pension, and workers’ compensation schemes in different countries (Scott et al. 2017). The cohort studied was limited to employees in the public sector predominated by women in a Nordic welfare state. We were not able to address changes in compensation and economic conditions in Finland. However, the terms of employment in the public sector were relatively stable during the time of the study, offering high job security to the employees and no major changes in sickness absence compensation. No data on workplace adjustments before or after the surgery was available. We did not have data on surgical techniques, so we were not able to assess whether less invasive surgical techniques were found to result in increased RTW compared with more invasive techniques. No data on radiographic imaging, complications, or repeated surgery was available. Another limitation is that we were not able to assess patient satisfaction reports or functional outcome scores. RTW may also be influenced by patients’ interactions with healthcare professionals as well as patients’ pre-surgery expectations (Bardgett et al. 2016). One limitation of our study was also that we were not able to separate those who had quit smoking from those who have never smoked. Conclusion In this occupational cohort of 389 employees, 95% of the participants had returned to work at 12 months after excision of lumbar disc herniation. Older patients in manual work with prolonged sick leave before the excision of lumbar disc herniation were at increased risk of a slower return to work after the surgery.

RL, PL, KM, and JV designed the study. RL, PL, VA, TO, MK, KM, and JV contributed to the acquisition of data or analysis of data. RL, PL, MN, KM, and JV drafted the manuscript. All authors contributed to the interpretation of the data, and to revising the manuscript. Acta thanks Hans Möller and Carl Mellner for help with peer review of this study

Airaksinen J, Jokela M, Virtanen M, Oksanen T, Pentti J, Vahtera J, Koskenvuo M, Kawachi I, Batty G D, Kivimaki M. Development and validation of a risk prediction model for work disability: multicohort study. Sci Rep 2017; 7(1): 13578,017–13892-1.

Acta Orthopaedica 2021; 92 (6): 638–643

Bailey C S, Rasoulinejad P, Taylor D, Sequeira K, Miller T, Watson J, Rosedale R, Bailey S I, Gurr K R, Siddiqi F, Glennie A, Urquhart J C. Surgery versus conservative care for persistent sciatica lasting 4 to 12 months. N Engl J Med 2020; 382(12): 1093-1102. Bardgett M, Lally J, Malviya A, Deehan D. Return to work after knee replacement: a qualitative study of patient experiences. BMJ Open 2016; 6 (2): e007912,2015–007912. Dewing C B, Provencher M T, Riffenburgh R H, Kerr S, Manos R E. The outcomes of lumbar microdiscectomy in a young, active population: correlation by herniation type and level. Spine 2008; 33(1): 33-8. Frank J W, Kerr M S, Brooker A S, DeMaio S E, Maetzel A, Shannon H S, Sullivan T J, Norman R W, Wells R P. Disability resulting from occupational low back pain, Part I: What do we know about primary prevention? A review of the scientific evidence on prevention before disability begins. Spine 1996; 21(24): 2908-17. Goldberg D P, Gater R, Sartorius N, Ustun T P, Piccinelli M, Gureje O, Rutter C. The validity of two versions of the GHQ in the WHO study of mental illness in general health care. Psychol Med 1997; 27(1): 191-7. Graver V, Ljunggren A E, Loeb M, Haaland A K, Lie H, Magnaes B. Background variables (medical history, anthropometric and biological factors) in relation to the outcome of lumbar disc surgery. Scand J Rehabil Med 1998; 30(4): 221-5. Huysmans E, Goudman L, Van Belleghem G, De Jaeger M, Moens M, Nijs J, Ickmans K, Buyl R, Vanroelen C, Putman K. Return to work following surgery for lumbar radiculopathy: a systematic review. Spine J 2018; 18(9): 1694-1714. Jensdottir M, Gudmundsson K, Hannesson B, Gudmunsson G. 20 years follow-up after the first microsurgical lumbar discectomies in Iceland. Acta Neurochir (Wien) 2007; 149(1): 51-8; discussion 57-8. Khan I, Bydon M, Archer K R, Sivaganesan A, Asher A M, Alvi M A, Kerezoudis P, Knightly J J, Foley K T, Bisson E F, Shaffrey C, Asher A L, Spengler D M, Devin C J. Impact of occupational characteristics on return to work for employed patients after elective lumbar spine surgery. Spine J 2019; 19(12): 1969-76. Koenig L, Zhang Q, Austin M S, Demiralp B, Fehring T K, Feng C, Mather 3rd R C, Nguyen J T, Saavoss A, Springer B D, Yates Jr A J. Estimating the societal benefits of THA after accounting for work status and productivity: a Markov model approach. Clin Orthop Relat Res 2016; 474(12): 2645-54. Kotilainen E, Valtonen S, Carlson C A. Microsurgical treatment of lumbar disc herniation: follow-up of 237 patients. Acta Neurochir (Wien) 1993; 120(3-4): 143-9. Kujala U M, Kaprio J, Sarna S, Koskenvuo M. Relationship of leisuretime physical activity and mortality: the Finnish twin cohort. JAMA 1998; 279(6): 440-4. Laasik R, Lankinen P, Kivimäki M, Aalto V, Saltychev M, Mäkelä K, Vahtera J. Return to work after primary total hip arthroplasty: a nationwide cohort study. Acta Orthop 2019; 90(3): 209-13. Lankinen P, Laasik R, Kivimäki M, Aalto V, Saltychev M, Vahtera J, Mäkelä K. Are patient-related pre-operative factors influencing return to work after total knee arthroplasty. Knee 2019; 26(4): 853-60. Liang M H, Cullen K E, Larson M G, Thompson M S, Schwartz J A, Fossel A H, Roberts W N, Sledge C B. Cost-effectiveness of total joint arthroplasty in osteoarthritis. Arthritis Rheum 1986; 29(8): 937-43. Nygaard O P, Kloster R, Solberg T. Duration of leg pain as a predictor of outcome after surgery for lumbar disc herniation: a prospective cohort study with 1-year follow up. J Neurosurg 2000; 9:(2 Suppl.): 131-4. Österman H, Seitsalo S, Karppinen J, Malmivaara A. Effectiveness of microdiscectomy for lumbar disc herniation: a randomized controlled trial with 2 years of follow-up. Spine 2006; 31(21): 2409-14. Paulsen R T, Rasmussen J, Carreon L Y, Andersen M O. Return to work after surgery for lumbar disc herniation, secondary analyses from a randomized controlled trial comparing supervised rehabilitation versus home exercises. Spine J 2020; 20(1): 41-7.

Acta Orthopaedica 2021; 92 (6): 638–643

Puolakka K, Ylinen J, Neva M H, Kautiainen H, Häkkinen A. Risk factors for back pain-related loss of working time after surgery for lumbar disc herniation: a 5-year follow-up study. Eur Spine J 2008; 17(3): 386-92. Rimm E B, Williams P, Fosher K, Criqui M, Stampfer M J. Moderate alcohol intake and lower risk of coronary heart disease: meta-analysis of effects on lipids and haemostatic factors. BMJ 1999; 319 (7224): 1523-8. Schade V, Semmer N, Main C J, Hora J, Boos N. The impact of clinical, morphological, psychosocial and work-related factors on the outcome of lumbar discectomy. Pain 1999; 80(1-2): 239-49. Schoeggl A, Reddy M, Matula C. Functional and economic outcome following microdiscectomy for lumbar disc herniation in 672 patients. J Spinal Disord Tech 2003; 16(2): 150-5. Scott C E H, Turnbull G S, MacDonald D, Breusch S J. Activity levels and return to work following total knee arthroplasty in patients under 65 years of age. Bone Joint J 2017; 99-B(8): 1037-46.

643

Silverplats K, Lind B, Zoëga B, Halldin K, Rutberg L, Gellerstedt M, Brisby H. Clinical factors of importance for outcome after lumbar disc herniation surgery: long-term follow-up. Eur Spine J 2010; 19(9): 145967. Sørlie A, Moholdt V, Kvistad K A, Nygaard O P, Ingebrigtsen T, Iversen T, Kloster R, Solberg T K. Modic type I changes and recovery of back pain after lumbar microdiscectomy. Eur Spine J 2012; 21(11): 2252-8. Than K D, Curran J N, Resnick D K, Shaffrey C I, Ghogawala Z, Mummaneni P V. How to predict return to work after lumbar discectomy: answers from the NeuroPoint-SD registry. J Neurosurg Spine 2016; 25(2): 181-6. Weber H. Lumbar disc herniation: a controlled, prospective study with ten years of observation. Spine 1983; 2:131-40. Vucetic N, Astrand P, Guntner P, Svensson O. Diagnosis and prognosis in lumbar disc herniation. Clin Orthop Relat Res 1999; (361): 116-22.

644

Acta Orthopaedica 2021; 92 (6): 644–650

Stable glenoid component of reverse total shoulder arthroplasty at 2 years as measured with model-based radiostereometric analysis (RSA) Alexander Nilsskog FRASER 2,3, Berte BØE 1, Tore FJALESTAD 1, Jan Erik MADSEN 1,2, and Stephan M RÖHRL 1 1 Division

of Orthopaedic Surgery, Oslo University Hospital; 2 Institute of Clinical Medicine; University of Oslo; 3 Diakonhjemmet Hospital, Norway Correspondence: a.n.fraser@medisin.uio.no Submitted 2021-01-27. Accepted 2021-05-14.

Background and purpose — Reverse total shoulder arthroplasty (TSA) is used for treating cuff arthropathy, displaced proximal humeral fractures (PHF), and in revision shoulder surgery, despite sparse evidence on long-term results. We assessed stability of the glenoid component in reverse TSA, using model-based RSA. Patients and methods — 20 patients (mean age 76 years, 17 female), operated on with reverse TSA at Oslo University Hospital, in 2015–2017 were included. Indications for surgeries were PHFs, malunion, cuff arthropathy, and chronic shoulder dislocation. RSA markers were placed in the scapular neck, the coracoid, and the acromion. RSA radiographs were conducted postoperatively, at 3 months, 1 year, and 2 years. RSA analysis was performed using RSAcore with Reversed Engineering (RE) modality, with clinical precision < 0.25 mm for all translations (x, y, z) and < 0.7° for rotations (x, z). Scapular “notching” was assessed in conventional radiographs. Results — 1 patient was excluded due to revision surgery. More than half of the patients displayed measurable migration at 2 years: 6 patients with linear translations below 1 mm and 8 patients who showed rotational migration. Except for one outlier, the measured rotations were below 2°. The migration pattern suggested implant stability at 2 years. 10 patients showed radiolographic signs of “notching”, and the mean Oxford Shoulder Score (OSS) at 2 years was 29 points (15–36 points). Interpretation — Stability analysis of the glenoid component of reversed total shoulder arthroplasty using reversed engineering (RE) model-based RSA indicated component stability at 2 years.

Reverse total shoulder arthroplasty (TSA) is a widely used procedure. It was originally intended for cuff arthropathy in elderly patients (Grammont and Baulot 1993), but is presently used for several indications, including acute proximal humeral fractures (PHFs) in the elderly, fracture malunions, chronic dislocations, and revision surgery (Clavert et al. 2019, Rugg et al. 2019, Malahias et al. 2020). For operative treatment of displaced 3- and 4-part PHFs in the elderly, reversed TSA has become the treatment of choice (Critchley et al. 2020), presently down to 60 years of age (Goldenberg et al. 2020). The increased use of reverse TSA has occurred despite sparse evidence concerning long-term clinical outcomes for the implant. However, short-term RSA may predict the longevity of implants (Valstar et al. 2005). For hips and knees, continuous micro-migration over 2 years has shown to be indicative of increased risk of implant loosening (Kärrholm et al. 1994, de Vries et al. 2014). To our knowledge, RSA stability analysis of the glenoid component of reverse TSA in patients has not previously been published. Much concern has been placed on the subject of “notching,” where the polyethylene liner of a reverse TSA over time erodes into the inferior scapular neck (Levigne et al. 2011). Several studies have related notching to poorer outcomes (Mollon et al. 2017, Simovitch et al. 2019), while others have voiced concerns about this causing instability and loosening of the glenoid component (Roche et al. 2013c, Huri et al. 2016). Model-based RSA has the advantage over traditional marker-based RSA of not having to alter implants by attaching markers, and the clinical precision of model-based RSA on the glenoid component is known (Fraser et al. 2018). With increased use, sparse long-term evidence, and with “notching” as the backdrop, we performed a stability analysis of the glenoid component of reversed TSA, using model-based RSA.

Acta Orthopaedica 2021; 92 (6): 644–650 645

Table 1. Baseline demographics in the 20 included patients. Values are count unless otherwise specified Sex (male/female) Mean age (SD) Median age (range) Living at home Diabetes (yes/no) Smoking (yes/no) Mean ASA group (SD) Operated arm (right/left) Indication for surgery Acute PHF Malunion PHF Delayed surgery PHF Cuff-tear arthropathy GH luxation (chronic) Oxford Shoulder Score (n = 5)

3/17 76 (5.7) 77 (66–85) 20 2/18 4/16 2.2 (0.4) 8/12 13 1 2 3 1 35

Figure 1. Radiograph of a patient operated with a reverse total shoulder arthroplasty postoperatively (A) and at 2-year follow-up (B), where “notching” grade 3 can be observed in the inferior scapular neck. RSA markers were implanted in the scapular neck, coracoid, and acromion.

Patients and methods In this clinical RSA study of the glenoid component of reverse TSA we have used model-based RSA technology, and specifically the reversed engineering (RE) RSA modality. A published method study (Fraser et al. 2018) supplied the framework, particularly with regard to how patients were positioned for RSA examinations, which RSA software modality was used, and the clinical precision of the RE model-based RSA on this implant, obtained by double examinations of 15 patients included in the current study. Inclusion 20 consecutive patients operated on with reverse TSA at Oslo University Hospital in the period from September 2015 to October 2017 were included. The majority of included patients had suffered an acute PHF, and a minority of patients were included on the basis of other indications (Table 1). Initially, we planned to include only fracture patients enrolled in a larger clinical trial comparing reverse TSA with plate fixation for displaced PHFs in the elderly (Fraser et al. 2020). The original inclusion criteria were patients aged 65–85 presenting with a displaced PHF type 11-B2 or 11-C2 (OTA/AO 2007 revision). All subgroups of B2 and C2 fractures were included, provided that the fractures were severely displaced, defined as > 45° valgus or > 30° varus in a true antero-posterior (AP) projection, > 45° angulation in the scapula Y-projection, or > 50% displacement of the humeral head against the metaphysis. Exclusion criteria were previous injury or illness of the injured or contralateral shoulder, concomitant injury to the ipsilateral or contralateral upper extremity, alcohol or other substance abuse, dementia or neurological disease, non-Norwegian speaking, glenoid fracture or deformity, or patients who were deemed non-compliant with rehabilitation. Head-split frac-

tures or fracture dislocations were not included. These criteria resulted in a slow inclusion rate, and we therefore changed the inclusion criteria to any patient destined for reverse TSA at Oslo University Hospital. Operative treatment Operative treatment with a reversed TSA (Delta Xtend Depuy Synthes, 700 Orthopaedic Drive, Warsaw, IN 46582, USA) was performed with all patients in the beach-chair position. In the 16 patients with PHF, acute or delayed, and 1 patient with a chronic glenohumeral dislocation, a deltopectoral approach was used. In the 3 patients with cuff arthropathy, a lateral transdeltoid approach was used. The tantalum markers were implanted in the glenoid, the acromion, and the coracoid process after surgical preparation of the glenoid, before implant insertion (Figure 1). Approximately 10 markers were used for each patient, adding about 15 minutes to the overall surgical procedure. RSA radiographs Paired RSA radiographs were obtained postoperatively (PO), at 3 months, 1 year, and 2 years. We strived to conduct PO radiographs within 1 week, and most PO examinations were performed on day 3 or 4 after surgery. RSA examinations were dual simultaneous radiographs, where the overhead X-ray tubes were focused on the implant at a mutual angle of 60°. The patient was positioned supine on the examination table, and approximately 20–30° tilted towards the operated side, with radiographic exposure in the sagittal plane. We named this the shoulder position (Figure 2) to distinguish it from traditional hip RSA where the radiographic plane is transverse. A uniplanar calibration cage (Cage 43, UmRSA Biomedical— RSA Biomedical, Umeå, Sweden) was positioned underneath the examination table.

646

Acta Orthopaedica 2021; 92 (6): 644–650

Table 2. Shoulder reference position and conversion table

Right shoulder: Clinical direction of migration/ rotation

Left shoulder: Conversion to right shoulder values

Peg down: Conversion to peg up

Translation x cranial (–), caudal (+) × (–1) × (–1) y lateral (–), medial (+) No difference × (–1) z posterior (–), anterior (+) No difference No difference Rotation x internal (–), external (+) No difference × (–1) y NP NP NP z superior tilt (–), × (–1) No difference inferior tilt (+)

Figure 2. Patient positioning when performing shoulder RSA. The radiographic tubes were centered on the glenoid implant of the right shoulder and aligned along the length axis of the patient, resulting in a sagittal radiographic plane.

Reference position We decided that the right shoulder with the peg of the glenoid implant pointing upwards on radiographs would be the reference position for RSA migration analyses. The clinical significance and direction of translation and rotation are listed on the left side of Table 2. For any left-sided implant, or for radiographs with the peg of the glenoid implant pointing downwards, certain conversions of direction would have to be applied before conducting the final migration analysis. These conversions are listed on the right side of Table 2, and basically convert all migration values to the decided reference position: right shoulder with peg pointing upwards. RSA analysis RSA radiographs were analyzed using RSAcore software (MBRSA 4.1, Leiden University Medical Centre, Leiden, NL), using reversed engineering (RE) (Figure 3). The RE model of the glenoid implant (Delta Xtend Metaglene and Glenosphere, Depuy Synthes, 700 Orthopaedic Drive, Warsaw, IN 46582, USA) was obtained by laser-scanning (RSAcore, Leiden University Medical Centre, Leiden, NL). The circumference of the glenoid component in the paired radiographs was matched with the virtual RE model of the implant. The RSA markers implanted in scapular bone were marked manually. Migration of the implant along each of the 3 axes (x, y, z) and rotation around the 2 measurable axes (x and z) were measured from point 0, corresponding with PO radiographs, to 2-year followup. The distribution of RSA bone markers was assessed using the condition number (CN) (Valstar et al. 2005). RSA radiographs with a CN > 120, indicating a narrow distribution of

The right shoulder with the peg up is the reference position, and the clinical significance of a negative or positive value for all translations and rotations are listed on the left side of this table. For the purpose of RSA migration analyses, left-sided implants or implants that are orientated with the peg down on RSA radiographs, must be converted to the reference position, as described on the right side of the table. NP= Not possible to analyze.

Figure 3. Radiostereometric analyses were performed with RSAcore software using reversed engineering (RE). The virtual glenoid implant (green) corresponds with the red demarcated outline of the actual glenoid implant shown on dual simultaneous radiographs of a right-sided shoulder with a reverse total shoulder arthroplasty. RSA markers are red, control markers are green, and fiducial markers are yellow.

markers, were omitted. RSA radiographs with a CN < 120, indicating a wider distribution of markers, were included. The mean error for rigid body fitting (ME) (Valstar et al. 2005), is an expression of marker stability, where the limit in this study was set to ME < 0.35. Any marker presenting with a ME > 0.35 would imply that the marker was unstable, and was therefore omitted. Clinical outcome Oxford Shoulder Score (OSS, 0–48 points, 0 = worst, 48 = best) was assessed at 2 years. Scapular notching was assessed on conventional shoulder radiographs obtained in conjunction with clinical follow-ups.

Acta Orthopaedica 2021; 92 (6): 644–650 647

20 Patients included (n = 20): – acute PHF, 13 – cuff-tear arthopathy, 3 – delayed surgery after PHF, 2 – malunion, 1 – chronic glenohumeral dislocation, 1 Postoperative follow-up (n = 20): – RSA data available, 19 – not valid RSA radiographs, 1 3-month follow-up (n = 20): – RSA data available, 19 – not valid RSA radiographs, 1

X-translation (mm)

Y-translation (mm)

Z-translation (mm)

0.2

0.3

0.4

0.1

0.2

0.3

0.0

0.1

0.2

–0.1

0.0

0.1

–0.2

–0.1

0.0

–0.1

–0.2

–0.3 0

Follow-up (months)

Dead (n = 2) 12-month follow-up (n = 18): – RSA data available, 15 – not valid RSA radiographs, 3 Excluded (n = 2): – revised for septic loosening, 1 – dead, 1

X-rotation (°)

Statistics The mean value with a confidence interval (CI) of ± 2 standard deviations (± 2SD) for linear migration and rotation was calculated separately for each degree of freedom. Statistical analyses were performed using IBM SPSS Statistics Version 24 (IBM Corp, Armonk, NY, USA). Ethics, funding, and potential conflicts of interest Ethical approval was obtained from the regional ethics committee (REK) on April 4, 2015, reference No. 2012/1606/ REK South-East, and patients gave signed consent after written and oral information. The project has received research funds from Sophies Minde Ortopedi AS, a subsidiary of Oslo University Hospital. The authors have no conflict of interest to declare.

Results 20 patients with a mean age of 76 (66–85) years were included, all living at home. All patients were treated operatively with reverse TSA, and in 15 of 20 cases the indication was a displaced proximal humeral fracture (Table 1). 1 patient was excluded from the final analysis because of septic implant loosening and revision surgery, 3 patients died before twoyear follow-up, and 1 patient was excluded from 24-month RSA analysis because of CN > 120 (Figure 4). Mean migration measurements showed linear translations along the x, y, and z axis, signifying caudal–cranial, lat-

Follow-up (months)

Z-rotation (°) 0.4

0.5

0.2

0.0

–0.2

–0.4 –1.0

D Figure 4. Flow of patients.

1.0

–0.5

24-month follow-up (n = 16): – RSA data available, 15 – not valid RSA radiographs, 1

Follow-up (months)

Figure 5. Mean values for (A) x-translations (negative values = cranial migration, positive values = caudal migration), (B) Y-translations (negative values = lateral migration, positive values= medial migration), (C) Z-translations (negative values = posterior migration, positive values = anterior migration) measured in mm, and (D) X-rotation (negative values = internal rotation, positive values = external rotation) and (E) Z-rotation (negative values = superior tilt, positive values = inferior tilt) measured in degrees.

eral–medial and anterior–posterior translations respectively. Rotational measurements around the x-axis and z-axis signified internal–external rotation and superior–inferior tilt of the implant (Figure 5). Rotation around the y-axis (anterior–posterior flexion) is not possible to measure with model-based RSA, as this implant is symmetrical around this axis. Individual migration measurements for each patient were similarly divided into linear translations and rotations, where the precision of the RSA method for each degree of freedom was marked with two stapled lines (Fraser et al. 2018) (Figure 6). Of the 19 patients included in the final analysis, 17 patients demonstrated measurable translation and/or rotation above the clinical precision of RSA on the glenoid component. Of these, no linear migrations exceeded 1.2 mm in any direction, and apart from 1 outlier (No. 8: internal rotation), the rotations measured were below 2°. 2 patients had a grade 1 notching, with a defect contained within the inferior pillar of the scapular neck; 5 patients had grade 2 notching where the erosion had reached the inferior screw; and 2 patients had grade 3 notching with erosion of bone beyond the lower fixation screw. The mean OSS at 2 years was 29 points (15–36). Of the 15 patients with acceptable RSA radiographs at 2 years, 9 patients displayed radio-

648

Acta Orthopaedica 2021; 92 (6): 644–650

Y-translation (mm)

X-translation (mm) 1.0

Table 3. Individual patient data at 2 years

Z-translation (mm)

1.0

1.5

0.5

1.0

0.0

0.5

–0.5

0.0

Patient Notching no. Indication grade OSS Dropout

0.5

0.0

–1.0

–0.5

Follow-up (months)

X-rotation (°)

–0.5 0

Follow-up (months)

Z-rotation (°)

2.0

1.0 1.0

0.0 –1.0

0.0

–2.0 –1.0

–3.0 –4.0

–2.0

Follow-up (months)

1 Malunion Dead 2 Acute PHF 33 3 Acute PHF Dead 4 Late PHF Dead 5 Late PHF 2 Revised 6 Acute PHF 3 35 7 Acute PHF 24 8 Arthrosis 1 29 9 Chronic dislocation 16 10 Acute PHF 21 11 Arthrosis 2 35 12 Acute PHF 33 13 Arthrosis 3 36 14 Acute PHF 2 33 15 Acute PHF 28 16 Acute PHF 2 36 17 Acute PHF 2 31 18 Acute PHF 1 15 19 Acute PHF 2 26 20 Acute PHF 36

Follow-up (months)

Figure 6. Individual patient RSA measurements for translations and rotations. For directions (A–E), see Figure 5. Stippled horizontal lines define the clinical precision of reverse-engineered model-based RSA on the glenoid component, where the area between the two stippled lines represent migration that is outside the resolution of this RSA method.

graphic notching, all of which showed measurable migration in 1 or more degrees of freedom. 6 patients did not show signs of notching at 2 years (Table 3).

Discussion Our study of model-based RSA on the glenoid component of reverse TSA has shown that approximately half of the patients displayed migration below the precision level of the RSA method. With the exception of 1 outlier, which displayed approximately 4º internal rotation at 2 years, the others had a measurable migration up to ~1 mm translation or 2° rotation over 2 years. When considering the individual RSA migration measurements (Figure 6), all lines—including those that present early measurable migration—seem to conform towards the horizontal. This migration pattern indicates a stable implant. Little RSA research has been published on this implant. Apart from a methodological study including precision measurements in patients (Fraser et al. 2018), our study seems to be the only clinical RSA stability study on reverse shoulder arthroplasty. Other shoulder RSA studies have involved a phantom glenoid component (Van de Kleut et al. 2018), ana-

OSS = Oxford Shoulder Score, PHF = proximal humeral fracture.

tomic shoulder arthroplasties or resurfacing implants (Nagels et al. 2002, Rahme et al. 2004, 2006, 2009, Nuttall et al. 2009, 2012, Sköldenberg and Odquist 2011, Stilling et al. 2012, Mechlenburg et al. 2014, Streit et al. 2015). These latter studies utilized marker-based RSA, and differ fundamentally from the current study with regard to implant design, fixation type, and RSA method. Comparisons therefore seem futile, and may even be misleading. An association between early migration and later implant loosening has not been established for reversed shoulder arthroplasties. For the hip, Pijls et al. (2012) conducted a systematic review to determine the association between early migration of acetabular cups and late aseptic revision, where proximal migration of < 0.2 mm was considered acceptable, while proximal migration > 1mm was considered unacceptable. A proximal migration of between 0.2 mm and 1.0 mm at 2 years carried a > 5% risk of revision at 10 years. In the context of model-based RSA, acetabular cups bear some resemblance to the glenoid implants in reverse TSA, in the sense that they are both small implants with a hemispherical shape. Even so, these implants are not readily comparable. Hip implants are subjected to different forces and weightbearing, and the acetabular cups in this study had a variety of fixation modalities, including cement fixation. Despite the obvious differences, it would be interesting to compare the thresholds (Pijls and Nelissen 2016) for increased risk of loosening. In our study, 13 of 15 patients were within ±0.22 mm caudal-cranial migration at 2 years, while 2 patients show

Acta Orthopaedica 2021; 92 (6): 644–650 649

measurable migration of less than 0.5 mm (Figure 6A). At 2 years, these patients had an OSS score of 35 and 26 points, and displayed notching grade 3 and 2 (Table 3). Whether a migration of 0.5 mm combined with notching represents an increased risk of implant loosening remains to be seen, and long-term follow-up is needed to establish threshold migration values for the glenoid component. Further, these thresholds may be compared with established threshold values of other implants. Conventional radiographic projections were obtained at every follow-up, and these were used to detect glenoid notching. 9 patients had signs of notching on 2-year AP radiographs: 6 with acute PHF and 3 with cuff arthropathy. This constituted half of the patients included in this study, and a higher fraction than expected, especially due to the fact that the operative technique has been altered to prevent this from happening (Roche et al. 2013a, 2013b). Except for 1 patient, all patients with notching displayed measurable migration in at least 1 degree of freedom (Table 3). The mean OSS was 29 points, which is substantially lower than the reverse TSA group in the DelPhi study, an RCT comparing reverse TSA with plate fixation for displaced PHFs (Fraser et al. 2020), where they scored 41 points at 2 years. However, the heterogenicity of indications in our study is a limitation, and makes OSS comparisons with other studies difficult. Besides acute PHFs, our study involved patients with operative indications that may have worsened the outcome, such as chronic dislocation, malunion, PHF delayed surgery, and cuff arthropathy. Furthermore, half the patients in our study showed radiological signs of notching, which is known to be associated with poorer outcomes (Mollon et al. 2017, Simovitch et al. 2019). 20 patients is a sparse number for most clinical trials, but arguably sufficient for an RSA study due to the high precision of the method (Valstar et al. 2005). Another limitation of our study was the variety of indications for reverse TSA. This makes some of the results more difficult to compare with other studies. Heterogenicity of indications, however, does not affect the overall result of what would be described as a stable glenoid component migration pattern. Furthermore, rotation around the y-axis, representing anteversion/retroversion of the glenoid implant, was not measurable with model-based RSA due to the implant being symmetrical around this axis. This also implied that maximal total point motion (MTPM) could not be calculated. One strength of our study was using an established RSA method previously tested in a methodological study (Fraser et al. 2018), where patient positioning, type of model-based RSA, and the clinical precision of the RSA method was established in advance. In conclusion, stability analysis of the glenoid component of reversed total shoulder arthroplasty using reversed engineering (RE) model-based RSA indicate component stability at 2 years.

The authors would like to thank the research team at the Center for Implant and Radiostereometric Research Oslo (CIRRO) with research coordinator Marte Traae Magnusson and radiographers Alexis Hinojosa, Silje Klausen, and Mona Risdal and the Department of Radiology, Oslo University Hospital, for conducting the RSA imaging. Physiotherapist Tone Wagle is acknowledged for clinical scoring of the fracture patients. Tom Ludvigsen conducted elective surgeries. Biostatistician Are Hugo Pripp (Oslo Center for Biostatistics and Epidemiology, Oslo University Hospital) is thanked for advice and help with the statistical analysis. ANF, SMR: conception, acquisition, surgery, follow-up, data sampling, analysis, interpretation of data, drafting, critical revision. BB: acquisition, surgery, follow-up, data sampling, critical revision. TF: conception, acquisition, surgery, follow-up, data sampling, critical revision. JEM: conception, acquisition, surgery, critical revision. Acta thanks Cyrus Brodén and Bo Sanderhoff Olsen for help with peer review of this study.

Clavert P, Kling A, Sirveaux F, Favard L, Mole D, Walch G, Boileau P. Reverse shoulder arthroplasty for instability arthropathy. Int Orthop 2019; 43(7): 1653-8. doi: 10.1007/s00264-018-4123-4. Critchley O, McLean A, Page R, Taylor F, Graves S, Lorimer M, Peng Y, Hatton A, Bain G. Reverse total shoulder arthroplasty compared to stemmed hemiarthroplasty for proximal humeral fractures: a registry analysis of 5946 patients. J Shoulder Elbow Surg 2020; 29(12): 2538-47. doi: 10.1016/j.jse.2020.04.005. de Vries L M, van der Weegen W, Pilot P, Stolarczyk P A, Sijbesma T, Hoffman E L. The predictive value of radiostereometric analysis for stem survival in total hip arthroplasty: a systematic review. Hip Int 2014; 24(3): 215-22. doi: 10.5301/hipint.5000102. Fraser A N, Tsukanaka M, Fjalestad T, Madsen J E, Rohrl S M. Modelbased RSA is suitable for clinical trials on the glenoid component of reverse total shoulder arthroplasty. J Orthop Res 2018; 36(12): 3299-307. doi: 10.1002/jor.24111. Fraser A N, Bjordal J, Wagle T M, Karlberg A C, Lien O A, Eilertsen L, Mader K, Apold H, Larsen L B, Madsen J E, Fjalestad T. Reverse shoulder arthroplasty is superior to plate fixation at 2 years for displaced proximal humeral fractures in the elderly: a multicenter randomized controlled trial. J Bone Joint Surg Am 2020; 102(6): 477-85. doi: 10.2106/JBJS.19.01071. Goldenberg B T, Samuelsen B T, Spratt J D, Dornan G J, Millett P J. Complications and implant survivorship following primary reverse total shoulder arthroplasty in patients younger than 65 years: a systematic review. J Shoulder Elbow Surg 2020; 29(8): 1703-11. doi: 10.1016/j.jse.2020.02.004. Grammont P M, Baulot E. Delta shoulder prosthesis for rotator cuff rupture. Orthopedics 1993; 16(1): 65-8. Huri G, Familiari F, Salari N, Petersen S A, Doral M N, McFarland E G. Prosthetic design of reverse shoulder arthroplasty contributes to scapular notching and instability. World J Orthop 2016; 7(11): 738-45. doi: 10.5312/ wjo.v7.i11.738. Kärrholm J, Borssén B, Löwenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 4–7-year stereoradiographic followup of 84 cemented prostheses. J Bone Joint Surg Br 1994; 76(6): 912-17. Levigne C, Garret J, Boileau P, Alami G, Favard L, Walch G. Scapular notching in reverse shoulder arthroplasty: is it important to avoid it and how? Clin Orthop Relat Res 2011; 469(9): 2512-20. doi: 10.1007/s11999010-1695-8. Malahias M A, Chytas D, Kostretzis L, Brilakis E, Fandridis E, Hantes M, Antonogiannakis E. Bone grafting in primary and revision reverse total shoulder arthroplasty for the management of glenoid bone loss: a systematic review. J Orthop 2020; 20:78-86. doi: 10.1016/j.jor.2019.12.005.

650

Mechlenburg I, Klebe T M, Dossing K V, Amstrup A, Søballe K, Stilling M. Evaluation of periprosthetic bone mineral density and postoperative migration of humeral head resurfacing implants: two-year results of a randomized controlled clinical trial. J Shoulder Elbow Surg 2014; 23(10): 1427-36. doi: 10.1016/j.jse.2014.05.012. Mollon B, Mahure S A, Roche C P, Zuckerman J D. Impact of scapular notching on clinical outcomes after reverse total shoulder arthroplasty: an analysis of 476 shoulders. J Shoulder Elbow Surg 2017; 26(7): 1253-61. doi: 10.1016/j.jse.2016.11.043. Nagels J, Valstar E R, Stokdijk M, Rozing P M. Patterns of loosening of the glenoid component. J Bone Joint Surg Br 2002; 84(1): 83-7. doi: 10.1302/0301-620x.84b1.11951. Nuttall D, Haines J F, Trail I A. The effect of the offset humeral head on the micromovement of pegged glenoid components: a comparative study using radiostereometric analysis. J Bone Joint Surg Br 2009; 91(6): 757-61. doi: 10.1302/0301-620X.91B6.22060. Nuttall D, Haines J F, Trail I A. The early migration of a partially cemented fluted pegged glenoid component using radiostereometric analysis. J Shoulder Elbow Surg 2012; 21(9): 1191-6. doi: 10.1016/j.jse.2011.07.028. Pijls B G, Nelissen R G H H. Correspondence: Strategy for RSA migration thresholds. Acta Orthop 2016; 87(4): 432-3. doi: 10.1080/17453674.2016. 1202464. Pijls B G, Nieuwenhuijse M J, Fiocco M, Plevier J W, Middeldorp S, Nelissen R G, Valstar E R. Early proximal migration of cups is associated with late revision in THA: a systematic review and meta-analysis of 26 RSA studies and 49 survival studies. Acta Orthop 2012; 83(6): 583-91. doi: 10.3109/17453674.2012.745353. Rahme H, Mattsson P, Larsson S. Stability of cemented all-polyethylene keeled glenoid components: a radiostereometric study with a two-year follow-up. J Bone Joint Surg Br 2004; 86(6): 856-60. doi: 10.1302/0301620x.86b6.14882. Rahme H, Mattsson P, Wikblad L, Larsson S. Cement and pressfit humeral stem fixation provides similar results in rheumatoid patients. Clin Orthop Relat Res 2006; 448: 28-32. doi: 10.1097/01. blo.0000224007.25636.85. Rahme H, Mattsson P, Wikblad L, Nowak J, Larsson S. Stability of cemented in-line pegged glenoid compared with keeled glenoid components in total shoulder arthroplasty. J Bone Joint Surg Am 2009; 91(8): 1965-72. doi: 10.2106/JBJS.H.00938.

Acta Orthopaedica 2021; 92 (6): 644–650

Roche C P, Marczuk Y, Wright T W, Flurin P H, Grey S, Jones R, Routman H D, Gilot G, Zuckerman J D. Scapular notching and osteophyte formation after reverse shoulder replacement: radiological analysis of implant position in male and female patients. Bone Joint J 2013a; 95-B(4): 530-5. doi: 10.1302/0301-620X.95B4.30442. Roche C P, Marczuk Y, Wright T W, Flurin P H, Grey S G, Jones R B, Routman H D, Gilot G J, Zuckerman J D. Scapular notching in reverse shoulder arthroplasty: validation of a computer impingement model. Bull Hosp Jt Dis (2013) 2013b; 71(4): 278-83. Roche C P, Stroud N J, Martin B L, Steiler C A, Flurin P H, Wright T W, DiPaola M J, Zuckerman J D. The impact of scapular notching on reverse shoulder glenoid fixation. J Shoulder Elbow Surg 2013c; 22(7): 963-70. doi: 10.1016/j.jse.2012.10.035. Rugg C M, Coughlan M J, Lansdown D A. Reverse total shoulder arthroplasty: biomechanics and indications. Curr Rev Musculoskelet Med 2019; 12(4): 542-53. doi: 10.1007/s12178-019-09586-y. Simovitch R, Flurin P H, Wright T W, Zuckerman J D, Roche C. Impact of scapular notching on reverse total shoulder arthroplasty midterm outcomes: 5-year minimum follow-up. J Shoulder Elbow Surg 2019; 28(12): 2301-7. doi: 10.1016/j.jse.2019.04.042. Sköldenberg O, Odquist M. Measurement of migration of a humeral head resurfacing prosthesis using radiostereometry without implant marking: an experimental study. Acta Orthop 2011; 82(2): 193-7. doi: 10.3109/17453674.2011.566133. Stilling M, Mechlenburg I, Amstrup A, Soballe K, Klebe T. Precision of novel radiological methods in relation to resurfacing humeral head implants: assessment by radiostereometric analysis, DXA, and geometrical analysis. Arch Orthop Trauma Surg 2012; 132(11): 1521-30. doi: 10.1007/s00402-012-1580-x. Streit J J, Shishani Y, Greene M E, Nebergall A K, Wanner J P, Bragdon C R, Malchau H, Gobezie R. Radiostereometric and radiographic analysis of glenoid component motion after total shoulder arthroplasty. Orthopedics 2015; 38(10): e891-7. doi: 10.3928/01477447-20151002-56. Valstar E R, Gill R, Ryd L, Flivik G, Borlin N, Kärrholm J. Guidelines for standardization of radiostereometry (RSA) of implants. Acta Orthop 2005; 76(4): 563-72. doi: 10.1080/17453670510041574. Van de Kleut M L, Yuan X, Athwal G S, Teeter M G. Validation of radiostereometric analysis in six degrees of freedom for use with reverse total shoulder arthroplasty. J Biomech 2018; 68:126-31. doi: 10.1016/j.jbiomech.2017.12.027.

Acta Orthopaedica 2021; 92 (6): 651–657

651

No change in reoperation rates despite shifting treatment trends: a population-based study of 4,070 proximal humeral fractures Carl BERGDAHL 1,2, David WENNERGREN 1,2, Eleonora SWENSSON-BACKELIN 1, Jan EKELUND 3, and Michael MÖLLER 1,2 1 Department

of Orthopaedics, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg; 2 Department of Orthopaedics, Sahlgrenska University Hospital, Gothenburg/Mölndal; 3 Centre of Registers, Western Healthcare Region, Gothenburg, Sweden Correspondence: carl.bergdahl@vgregion.se Submitted 2021-03-28. Accepted 2021-05-25.

Background and purpose — Clear and acknowledged treatment algorithms for proximal humeral fractures (PHFs) are lacking. Nevertheless, a change in treatment trends, including a change towards more reversed shoulder arthroplasties (RSA), has been observed during recent years. We examined the effect of these changes on reoperation rates. Patients and methods — Between 2011 and 2017, 4,070 PHFs treated at Sahlgrenska University Hospital were registered prospectively in the Swedish Fracture Register (SFR) and followed up until 2019 (mean follow-up of 4.5 years). Data on all reoperations were gathered from the SFR and from medical records. Results — The majority of PHFs were treated non-surgically and the proportion increased slightly, but not statistically significantly, during the study period (from 76% to 79%). Of the surgically treated fractures, the proportion fixed with a plate decreased from 47% to 25%, while the use of RSA increased 9-fold (from 2.0% to 19%). 221 patients underwent 302 reoperations. For those primarily treated surgically, the reoperation rate was 17%. Among treatment modalities, plate fixation was associated with the highest reoperation rate (21%). Rate of reoperations remained constant during the study period, both for the entire study cohort and for the surgically treated patients Interpretation — During the study period, treatment changes that are in accordance with recently published treatment recommendations were observed. However, these treatment changes did not affect the reoperation rate. Treatment with a plate, intramedullary nail, or hemiarthroplasty was associated with the highest reoperation rates. The fact that almost every 4th surgical procedure was a reoperation indicates a need for further improvement of modern treatment concepts for PHFs.

Fractures of the proximal humerus are common and are associated with a long-term negative impact on quality of life and excess mortality (Clement et al. 2014, Bergdahl et al. 2020). The optimal treatment is controversial. Displaced proximal humeral fractures (PHF) may result in poor outcome regardless of treatment modality and challenging revision procedures are common (Olerud et al. 2011a, 2011b, Lange et al. 2016). The treatment options for PHFs have evolved rapidly in recent years. The introduction of locking plates at the beginning of this century led to a sharp increase in the surgical fixation of PHFs (Bell et al. 2011, Sumrein et al. 2017). However, this trend was accompanied by an increased rate of complications and reoperations (Bell et al. 2011). In attempts to reduce the failure and revision rate, multiple modified surgical techniques evolved (Barlow et al. 2011, Boileau et al. 2013). Parallel to this development, reversed shoulder arthroplasties (RSAs) became increasingly popular (Han et al. 2016). Compared with the traditional fixation methods, RSA demonstrated lower complication rates (Klug et al. 2019). As a result, RSA use has increased markedly, while plate fixation has decreased (Han et al. 2016, Rajaee et al. 2017). According to the latest Cochrane review on PHF interventions there is a lack of data from randomized controlled trials (RCTs) to support one treatment over another (Handoll and Brorson 2015). The complication rate and need for revision surgery are therefore relevant measurements when evaluating PHF treatment. It is important to examine whether changes in treatment affect the need for repeat surgery. We evaluated trends in treatment methods for PHFs at a large Swedish orthopedic trauma unit and explored the rate of and risk factors for reoperations after primary treatment.

652

Patients and methods Study population All patients aged ≥ 16 years treated for a PHF at Sahlgrenska University Hospital (SUH) in Gothenburg in 2011–2017 were identified in the Swedish Fracture Register (SFR). Patients were followed until December 31, 2019, with an average follow-up time of 4.5 years (2–9). Data on patient demographics (age, sex), fracture characteristics according to the AO/OTA (Arbeitsgemeinschaft für Osteosynthesefragen/Orthopaedic Trauma Association) classification and treatment (primary treatment, cause, and type of late surgery/reoperation) were extracted from the SFR. By using the Swedish Tax Agency population register, all deaths during follow-up were identified. The implementation, design, validation, and registration process of the SFR have been described previously (Wennergren et al. 2015). SUH is the sole provider of orthopedic trauma care in Gothenburg and is also responsible for all sequalae related to fractures. To ensure that all reoperations were included, the digital surgical planning system at SUH was checked for all patients included in the study. Medical records were reviewed in search of absent registrations in the SFR and missed treatments were included in this study. The reporting of this observational study follows the STROBE guidelines. Treatment Indication for treatment was not standardized, but all PHFs were managed by a small group of experienced orthopedic trauma surgeons specialized in upper extremity trauma. Surgical management was prompted for severely displaced fractures in patients with high functional demands of their shoulder or when deemed compulsory. “Primary treatment” was defined as the treatment given within 30 days of the fracture date, as treatment can be altered for a PHF at an early (≤ 14 days) follow-up visit. As a result, nonoperatively treated patients who subsequently (within 30 days) underwent delayed surgery were classified as primarily surgically treated. Those treated non-surgically were immobilized in a sling for 2–4 weeks followed by physiotherapy. Surgical treatment was divided into 5 groups: locking plate, intramedullary nail (IM nail), hemi-arthroplasty (HA), reverse shoulder arthroplasty (RSA), and a combination of methods (combination). The combination methods included fixation with screws, cerclage wires, suture anchors, mini-plates, or a combination thereof typically used for displaced tuberosity fractures. Both cemented and uncemented HAs and RSAs were included. Trends in treatment were evaluated, and comparisons were made between the beginning (2011–2012) and the end of the study period (2016-2017). Reoperations All surgical interventions following an initial surgical treatment were referred to as a “reoperation.” In addition, a late

Acta Orthopaedica 2021; 92 (6): 651–657

surgical procedure (≥ 30 days) following an initial non-surgical treatment was regarded as a “reoperation.” Reasons for reoperations were analyzed (non-union, malunion, avascular necrosis [AVN] with collapse [Cruess grade 4 or 5], infection, implant failure, and reoperation due to patient demands). Secondary displacements with screw penetration, perioperative misplacement of implants, instability, and tuberosity absorption/displacement/malfunction were included in the “implant failure” group. Implant removal is not routinely performed at our institution and all reoperations were performed for symptomatic failures in agreement with the patient. In accordance with Olerud et al. (2011b), reoperations were further divided into major and minor reoperations. Major reoperations included all reoperations deemed compulsory, i.e., reoperations for all causes apart from reoperations due to patient demands. Statistics Changes in patient demographics and treatment between the beginning (2011–2012) and the end (2016–2017) of the study period were analyzed using chi-square and Student’s t-test for categorical and continuous variables, respectively. Statistical significance was set at p-values of < 0.05. Kaplan–Meier survival analysis was undertaken to illustrate the cumulative survival rate (i.e., time to reoperation) for the different treatment modalities. Date of death was used as censor and reoperation as event in the analyses, while the end of follow-up was December 31, 2019. Risk factors for reoperations were analyzed with a Cox proportional hazards regression model and expressed as hazard ratios (HR) with 95% confidence intervals (CI). IBM SPSS statistics version 25.0 (IBM Corp, Armonk, NY, USA) was used for all statistical analyses. Ethics, funding, data sharing, and potential conflicts of interest The study was conducted in accordance with the Helsinki Declaration and was approved by the Central Ethical Review Board in Gothenburg, Sweden (reference number T1137-18). In accordance with Swedish legislation, individual consent was not required. No grants from any public, commercial, or not-for-profit sector were received for this study. The data that supports the findings of this study is available from the corresponding author on reasonable request. The authors declare no competing interests.

Results During the study period, 4,009 patients with PHFs were registered in the SFR at SUH. 22 patients were excluded due to primary treatment elsewhere (n = 7), follow-up at another hospital (n = 13), and excision arthroplasty (n = 2), leaving 3,987 patients with 4,070 PHFs for the final analysis. The mean age at the time of a PHF was 68 years (16–104) and 72% of the

Acta Orthopaedica 2021; 92 (6): 651–657

653

Table 1. Frequency of treatment modalities with demographic data and distribution of fracture types for the study cohort Number of Fracture type Treatment fractures Age Female (AO/OTA b) n (%) c modality a n (%) mean (range) % A B C Non-surgical 3,117 (77) 68 (16–104) 73 1,653 (53) 1,299 (42) 165 (5.3) Surgical 953 (23) 65 (16–103) 68 347 (36) 307 (32) 299 (31) Plate 332 (8.2) 60 (16–99) 66 58 (17) 170 (51) 104 (31) IM nail 255 (6.3) 71 (19–103) 69 164 (64) 71 (28) 20 (7.8) Combination method 134 (3.3) 57 (20–102) 54 108 (81) 17 (13) 9 (6.7) HA 139 (3.4) 70 (35-92) 75 8 (5.8) 26 (19) 105 (76) RSA 93 (2.3) 75 (51–96) 85 9 (9.7) 23 (25) 61 (66) All treatments 4,070 (100) 68 (16–104) 72 2,000 (49) 1,606 (39) 464 (11) a IM nail, intramedullary nail; HA, hemiarthroplasty; RSA, reverse shoulder arthroplasty. b AO/OTA, Arbeitsgemeinschaft für Osteosynthesefragen/Orthopaedic Trauma Association fracture classification; A, fracture type A; B, fracture type B; C, fracture type C. c Percentage within the treatment group.

patients were female. According to the AO/OTA classification system, most fractures were classified as type A (49%), followed by type B (39%), and type C (11%). The majority were treated nonoperatively (77%), while 23% of PHFs were treated surgically (Table 1). Among surgical treatments, fixation with a locking plate was most common (35%), followed by fixation with an IM nail (27%). Most fractures treated with a combination method were of AO/OTA type A (81%), of which 101 fractures (94%) were isolated tuberosity fractures (AO/OTA 11-A1). Patients treated with an arthroplasty (HA and RSA) had the highest proportion of complex fractures (AO/OTA type C 76% and 66% respectively). No fractures were treated with percutaneous methods requiring mandatory removal of the fixation device. The completeness in the SFR for primary procedures and reoperations was 97% (n = 928/953) and 62% (n = 188/302) respectively when medical charts were reviewed. Change in treatment practice Throughout the 7-year study period, there were no statistically significant differences regarding patient demographics (age and sex) and fracture characteristics (AO/OTA groups). The proportion of surgically treated fractures did not differ statistically significantly during the study period (24% in 2011–2012 vs. 21% in 2016–2017; p = 0.06), but the distribution among the treatment modalities within the surgical group did. The proportion treated with a plate almost halved (n = 138/293; 47% in 2011–2012 vs. n = 61/237; 26% in 2016–2017; p < 0.001; Figure 1), while the proportion treated with IM nails substantially increased (n = 63/293; 22% in 2011–2012 vs. n = 71/237; 30% in 2016–2017; p = 0.03). However, the greatest proportional increase was seen in the group treated with RSA (n = 6/293; 2.0% in 2011–2012 vs. n = 44/237; 19% in 2016–2017; p < 0.001).

Annual method distribution 100 90 80 70 60 50 40 30 Plate IM nail Combination HA RSA

20 10 0

2011

2012

2013

2014

2015

2016

2017

Figure 1. Trends in treatment of surgically treated proximal humeral fractures (PHFs), presented by treatment modality as the proportion (%) of the total number of surgically treated PHFs for each year during the study period.

Reoperation/unplanned surgery During the 7-year study period, 221 patients underwent 302 reoperations. 1 of almost every 4 surgical interventions (302/1,257; 24%) for a PHF during the study period was a reoperation. The overall reoperation rate (including late surgery following initial non-surgical treatment) was 5.4% regardless of cause and 4.0% when considering only major reoperations (Table 2). Among the surgically treated fractures, 118 (12%) were subjected to a major reoperation and 163 (17%) to a reoperation of any cause. Despite the observed changes in treatment practice during the study period, the reoperation frequency remained similar between the beginning of the study and the end (all PHFs all reoperations 5.5% in 2011–2012 vs. 5.1% 2016–2017, all PHFs major reoperations 3.8% in 2011–2012 vs. 4.2% in 2016–2017, surgically treated PHFs all reoperations 18% in 2011–2012 vs. 16% in 2016–2017, surgically treated PHFs major reoperations 13% in 2011–2012 vs. 13% in 2016– 2017). The most common reason for the first reoperation was implant failure (35%) followed by reoperation on patient demands, most often due to impaired range of motion and/or pain (25%; Table 3). Reoperations due to infection were less common (6%). However, most infections required more than 1 surgical procedure, so infection was the underlying cause for 12% of all reoperations. Most reoperations occurred within 2 years of the initial treatment (Figure 2) and the treatment modality associated with the highest rate of reoperation was plate fixation (all cause reoperations 21% and major reoperations 15%). The majority of major reoperations following plate fixation were due to implant failure (43%) or AVN (33%). Among the surgical treatment modalities, patients treated with RSA had the lowest reoperation rate (all cause reoperations 6% and major reoperations 5%).

654

Acta Orthopaedica 2021; 92 (6): 651–657

Table 2. Number, frequency, and reason for the first reoperation by treatment modality. Values are count (%) Reason for first reoperation Treatment Total no. Fractures reoperated Implant Patient Number of modality of patients All a Major reop. a Nonunion Malunion AVN Infection failure demands Other b reoperations c Non-surgical 3,117 58 (1.9) 46 (1.5) 27 (0.87) 12 (0.38) 6 (0.19) 1 (0.03) – 12 (0.38) Surgical 953 163 (17) 118 (12) 2 (0.21) 4 (0.42) 22 (2.3) 13 (1.4) 77 (8.1) 44 (4.6) Plate 332 70 (21) 51 (15) 1 (0.30) 3 (0.90) 17 (5.1) 7 (2.1) 22 (6.6) 19 (5.7) IM nail 255 46 (18) 32 (13) – – 5 (2.0) – 27 (11) 14 (5.5) Combination method 134 18 (13) 11 (8.2) – 1 (0.75) 1 (0.75) 1 (0.75) 9 (6.7) 6 (4.5) HA 139 23 (17) 19 (14) – – – 3 (2.2) 16 (12) 4 (2.9) RSA 93 6 (6.5) 5 (5.4) – – – 2 (2.2) 3 (3.2) 1 (1.1) All treatments 4,070 221 (5.4) 164 (4.0) 9 (0.71) 16 (0.39) 28 (0.69) 14 (0.34) 77 (1.9) 56 (1.3) IM nail, intramedullary nail; HA = hemiarthroplasty; RSA, reverse shoulder arthroplasty; AVN, avascular necrosis. a Percentage within treatment group. b Reoperated with arthrodesis due to axillary nerve palsy sustained at initial trauma. c Percentage within all surgical procedures.

Table 3. Indication for reoperation: frequency, time to reoperation, and total number of reoperations Total number Days from of reoperated treatment to Total number Indication for fractures first reoperation of reoperations reoperation n (%) mean (range) n (%) Nonunion 29 (13) 194 (59–658) 45 (15) Malunion 16 (7.2) 651 (203–2,247) 20 (6.6) AVN 28 (13) 487 (76–1,946) 40 (13) Infection 14 (6.3) 171 (9–861) 37 (12) Implant failure 77 (35) 187 (1–727) 99 (33) Patient demands 56 (25) 500 (47–2,247) 60 (20) 1 (0.5) 158 1 (0.3) Other a Total 221 (100) 338 (1–2,247) 302 (100) AVN, avascular necrosis. a Reoperated with arthrodesis due to axillary nerve palsy sustained at initial trauma.

– 1 (0.10) 1 (0.30) –

78 (6.2) 224 (18) 103 (8.2) 51 (4.0)

– – – 1 (0.002)

24 (1.9) 32 (2.5) 14 (1.1) 302 (24)

Cumulative proportion reoperated (%) Non-surgical Plate IM nail Combination HA RSA

500

1000

1500

2000

2500

3000

Days from treatment Table 4. Cox proportional hazards regression of independent predictors of reoperation following treatment for a proximal humeral fracture All reoperations Major reoperations Risk factor p-value Exp(B) (95%CI) p-value Exp(B) (95% CI) Age (≤ 59 years as reference) 60–74 0.4 0.86 (0.63–1.2) 0.5 1.2 (0.78–1.7) 75–84 < 0.001 0.37 (0.23–0.59) 0.04 0.57 (0.34–0.96) ≥ 85 0.007 0.50 (0.30–0.83) 0.5 0.84 (0.49–1.5) Sex (male as reference) Female 0.8 1.0 (0.93–1.8) 0.8 0.95 (0.67–1.4) AO/OTA group (Group A as reference) Group B 0.1 1.3 (0.93–1.8) 0.02 1.6 (1.1–2.4) Group C < 0.001 4.9 (3.5–6.8) 0.001 5.7 (3.8–8.4) a Injury mechanism (high-energy trauma as reference) Low-energy trauma 0.2 0.71 (0.43–1.2) 0.3 0.71 (0.38–1.3) AO/OTA group, see Table 1 a 87 fractures excluded due to unknown (71) or inapplicable (16) trauma mechanism.

Figure 2. Kaplan–Meier curves, split by treatment modality, showing risk of reoperation over time.

Risk factors for reoperations The risk of a major reoperation increased with fracture complexity (HR 1.6 [CI 1.1–2.4] in AO/OTA type B and HR 5.7 [CI 3.6–8.4] in type C compared with type A; Table 4). Younger age (< 59 years) was an independent predictor of reoperation when all reoperations were analyzed. However, when patient-requested reoperations were excluded from the analysis, the increased risk with younger age became less apparent (Table 4). Neither sex nor injury mechanism (highor low-energy trauma) was associated with an increased risk of reoperation.

Acta Orthopaedica 2021; 92 (6): 651–657

Discussion This large prospectively collected but retrospectively validated and reviewed cohort study of PHF treatment in Gothenburg demonstrates that the proportion of patients treated surgically decreased slightly (although statistically non-significantly, p = 0.06) from 2011 to 2017, and that there was a significant change in surgical treatment modalities. The use of plate fixation decreased in favor of IM nails and RSAs. However, these changes in treatment have had no effect on the reoperation frequency. The reoperation frequency remained constant, 5% for all patients, 2% for nonoperatively treated patients, and 17% for surgically treated patients. Independent risk factors for reoperations were fracture complexity (AO/OTA type C) and younger age (< 59 years). PHFs are most commonly managed by non-surgical treatment. In our study, the proportion of non-surgical treatment was 76% in 2011/2012 and 79% in 2016/2017. After a decade of increasing rates of surgical treatment, a plateau appears to have been reached regarding surgical/non-surgical treatment. 2 studies from the United States reported a relative increase of 26% and 56% respectively of surgically managed PHFs from the late 1990s to 2005 (Bell et al. 2011; Petrigliano et al. 2014). A similar trend was reported in Sweden, with a doubling of the relative rate of surgical treatment for PHFs in 2000–2012 (Sumrein et al. 2017). The plateauing trend we observed might be the result of the numerous publications from 2010 and onwards, reporting a non-superior patientreported outcome following surgical treatment compared with non-surgical treatment for PHFs (Olerud et al. 2011a, 2011b, Rangan et al. 2015, Launonen et al. 2019). The decreased use of plate fixation and the increased use of IM nails and RSAs in this study are in accordance with previous reports from the western world. Both Rajaea et al. (2017) and Rosas et al. (2016) reported a decrease in the rate of plate fixation and a doubling in the rate of RSA use for PHFs in the US between 2011–2013 and 2009–2012 respectively. Several other studies have demonstrated the increasing role of RSA in the treatment of PHFs, especially in older patients with complex fractures. Registry data from the Nordic countries and New Zealand demonstrated a 5- to 6-fold increase in the incidence of RSA for PHFs between 2009 and 2016 (van der Merwe et al. 2017, Lehtimäki et al. 2020). Parallel to the growing popularity of RSA in recent years, an increased role for IM nails has been recognized, especially among elderly patients. An RCT from 2011 demonstrated lower complication rates associated with the modern version of IM nails compared with locking plates for two-part surgical-neck fractures (AO/OTA A2 and A3) (Zhu et al. 2011). Contrary to what we anticipated, the treatment changes we found did not render lower overall reoperation rates. One possible explanation might be the observation that reoperations following fixation with a plate, IM nail, or HAs remained

655

high during the study period. With recent treatment changes, including advances in surgical techniques and the preferred use of RSA instead of plate fixation in the elderly, as well as in complex fractures, a reduced failure rate/reoperation rate for these treatment modalities could be expected. However, no such changes were noted. Because plate fixation, IM nails, and HA still accounted for two-thirds (n = 159, 67%) of all surgical treatments at the end of the study period, the low reoperation rates following RSA did not result in lower overall reoperation rates. Considering the high rate of reported reoperations following plate, IM nail fixation, and HAs, these treatments must be questioned, especially as their superiority compared with nonoperative treatment has not been demonstrated (Olerud et al. 2011a, 2011b, Lange et al. 2016, Launonen et al. 2019). In fact, no surgical treatment modality has demonstrated superior functional and/or patient-reported outcome compared with non-surgical treatment for displaced PHFs, but surgery has been associated with substantially higher risks of secondary surgery with a risk ratio of 2.2 (95% CI 1.2–4.0) at 2 years post-fracture (Handoll and Brorson 2015, Lopiz et al. 2019). The fact that almost a quarter of all surgical procedures for PHFs in this study were unplanned reoperative procedures implies a need for additional research on how to select patients and fractures for the respective treatment in order to minimize the need for reoperations. RSA was associated with the lowest reoperation rates of all surgical treatment modalities in our study. However, this should be interpreted with caution because revision surgery following RSA is particularly demanding and few salvage procedures are available. Nevertheless, a recent Swedish study demonstrated superior clinical results in patients treated with RSA compared with HA and, taken together with the low revisions rates, these results indicate that RSA plays an important role in the surgical management of PHFs (Jonsson et al. 2021). Few previous studies have reported overall reoperation rates in a consecutive series of PHFs, regardless of treatment modality. We found no other study with such a large cohort of patients that had been individually controlled for reoperations. In a Swiss study, 192 consecutive patients with a PHF were followed up for a year and the reoperation rate of 11% was higher than the 5% reported in our study (Spross et al. 2019). In our study, reoperation was used as an indication of a complication or failure of the primary treatment. However, repeat surgery as an outcome measurement must always be interpreted with caution. The absence of reoperation does not necessarily represent an acceptable outcome. Many patients with complications or poor functional outcome are not subjected to reoperation (Amundsen et al. 2019, Barlow et al. 2020). Therefore the actual rate of complication or poor functional outcome in our study was probably substantially higher than the reported reoperation rates. Younger age and increasing fracture complexity were found to be independent predictors of reoperation, which are

656

in accordance with previous reports (Petrigliano et al. 2014, Barlow et al. 2020). Younger patients are probably more willing to undergo further surgery in order to avoid dysfunction and pain, which may be an explanation for the increased risk. This assumption is supported by the higher rate of patientrequested reoperations we found in patients ≤ 59 years old. Similar findings have been demonstrated for other surgically treated fractures (Wennergren et al. 2021). Elderly patients are generally reluctant to undergo additional surgery and probably more inclined to accept dysfunction. Barlow et al. (2020) found that older age was associated with increased radiological complications in a study of plate-fixated PHFs. Nevertheless, older patients were less likely to undergo reoperations. High validity for reoperations is a proven difficulty in register-based studies and underreporting leads to an underestimation of the risk of reoperation (Wennergren et al. 2021). To avoid reoperations not registered in the SFR remaining undetected, only patients primarily treated and eligible for followup at SUH were included. This enabled a medical-chart review in the search for reoperations. Reoperations performed outside SUH would consequently not be included. However, those are most likely few in numbers considering that SUH is the sole provider of treatment for PHFs and sequalae related to PHFs in the region. Strengths and limitations There are some limitations to this study. The data were collected prospectively but reviewed retrospectively. Thus, as previously mentioned, the risk of reoperations not being included in the study cannot be fully disregarded. On the other hand, the study design enabled the inclusion of more than 4,000 consecutive and individually reviewed PHFs, which is a considerable strength. This adds important knowledge to the evaluation of the overall treatment for PHFs outside the strict setting of RCTs. Another limitation always to be considered in research on PHFs is the limited interobserver reliability reported for PHF classifications (Wennergren et al. 2017). Lastly, the results are based on data from a single center, which may limit the generalizability. This was, however, a prerequisite in order to obtain a high level of completeness and validity regarding reoperations. Conclusion This study provides an evaluation of recent trends in PHF management with regards to reoperations. Although RSAs and IM nailing increased substantially, while plate fixation decreased, no effect was observed in relation to reoperation rates. Plate fixation, IM nailing, and HAs continue to be associated with high failure rates and 1 in every 4 surgical interventions for a PHF was a reoperation. These results highlight the need for a better treatment algorithm to optimize the care of patients with PHFs.

Acta Orthopaedica 2021; 92 (6): 651–657

CB and MM conceived the study idea and ESB and CB conducted the completion of data. CB and JE performed the statistical analyses and CB wrote the initial draft. All authors contributed to the interpretation of the data and revision of the manuscript. The authors wish to thank all the orthopedic surgeons at the affiliated department for entering detailed data on busy working days. Acta thanks Stig Brorson and Antti P Launonen for help with peer review of this study.

Amundsen A, Rasmussen J V, Olsen B S, Brorson S. Low revision rate despite poor functional outcome after stemmed hemiarthroplasty for acute proximal humeral fractures: 2,750 cases reported to the Danish Shoulder Arthroplasty Registry. Acta Orthop 2019; 90(3): 196-201. Barlow J D, Sanchez-Sotelo J, Torchia M. Proximal humerus fractures in the elderly can be reliably fixed with a “hybrid” locked-plating technique. Clin Orthop Relat Res 2011; 469: 3281-91. Barlow J D, Logli A L, Steinmann S P, Sems S A, Cross W W, Yuan B J, Torchia M E, Sanchez-Sotelo J. Locking plate fixation of proximal humerus fractures in patients older than 60 years continues to be associated with a high complication rate. J Shoulder Elbow Surg 2020; 29: 1689-94. Bell J E, Leung B C, Spratt K F, Koval K J, Weinstein J D, Goodman D C, Tosteson A N. Trends and variation in incidence, surgical treatment, and repeat surgery of proximal humeral fractures in the elderly. J Bone Joint Surg Am 2011; 93: 121-31. Bergdahl C, Wennergren D, Ekelund J, Möller M. Mortality after a proximal humeral fracture. Bone Joint J 2020; 102-b: 1484-90. Boileau P, Winter M, Cikes A, Han Y, Carles M, Walch G, Schwartz D G. Can surgeons predict what makes a good hemiarthroplasty for fracture? J Shoulder Elbow Surg 2013; 22: 1495-506. Clement N D, Duckworth A D, McQueen M M, Court-Brown C M. The outcome of proximal humeral fractures in the elderly: predictors of mortality and function. Bone Joint J 2014; 96-b: 970-7. Han R J, Sing D C, Feeley B T, Ma C B, Zhang A L. Proximal humerus fragility fractures: recent trends in nonoperative and operative treatment in the Medicare population. J Shoulder Elbow Surg 2016; 25: 256-61. Handoll H H, Brorson S. Interventions for treating proximal humeral fractures in adults. Cochrane Database Syst Rev 2015; CD000434. Jonsson E Ö, Ekholm C, Salomonsson B, Demir Y, Olerud P. Reverse total shoulder arthroplasty provides better shoulder function than hemiarthroplasty for displaced 3- and 4-part proximal humeral fractures in patients over 70 years of age: a multicenter randomized controlled trial. J Shoulder Elbow Surg 2021; 30(5): 994-1006. Klug A, Wincheringer D, Harth J, Schmidt-Horlohé K, Hoffmann R, Gramlich Y. Complications after surgical treatment of proximal humerus fractures in the elderly: an analysis of complication patterns and risk factors for reverse shoulder arthroplasty and angular-stable plating. J Shoulder Elbow Surg 2019; 28: 1674-84. Lange M, Brandt D, Mittlmeier T, Gradl G. Proximal humeral fractures: non-operative treatment versus intramedullary nailing in 2-, 3- and 4-part fractures. Injury 2016; 47 Suppl 7: S14-s19. Launonen A P, Sumrein B O, Reito A, Lepola V, Paloneva J, Jonsson K B, Wolf O, Ström P, Berg H E, Felländer-Tsai L, Jansson K Å, Fell D, Mechlenburg I, Døssing K, Østergaard H, Märtson A, Laitinen M K, Mattila V M. Operative versus non-operative treatment for 2-part proximal humerus fracture: a multicenter randomized controlled trial. PLoS Med 2019; 16: e1002855. Lehtimäki K, J Rasmussen V, Kukkonen J, Salomonsson B, Arverud E D, Hole R, Fenstadt A M, Brorson S, Jensen S L, Äärimaa V. Low risk of revision after reverse shoulder arthroplasty for acute proximal humeral fractures. JSES Int 2020; 4: 151-5.

Acta Orthopaedica 2021; 92 (6): 651–657

Lopiz Y, Alcobia-Diaz B, Galan-Olleros M, Garcia-Fernandez C, Picado A L, Marco F. Reverse shoulder arthroplasty versus nonoperative treatment for 3- or 4-part proximal humeral fractures in elderly patients: a prospective randomized controlled trial. J Shoulder Elbow Surg 2019; 28(12): 2259-71. Olerud P, Ahrengart L, Ponzer S, Saving J, Tidermark J. Hemiarthroplasty versus nonoperative treatment of displaced 4-part proximal humeral fractures in elderly patients: a randomized controlled trial. J Shoulder Elbow Surg 2011a; 20: 1025-33. Olerud P, Ahrengart L, Ponzer S, Saving J, Tidermark J. Internal fixation versus nonoperative treatment of displaced 3-part proximal humeral fractures in elderly patients: a randomized controlled trial. J Shoulder Elbow Surg 2011b; 20: 747-55. Petrigliano F A, Bezrukov N, Gamradt S C, SooHoo N F. Factors predicting complication and reoperation rates following surgical fixation of proximal humeral fractures. J Bone Joint Surg Am 2014; 96: 1544-51. Rajaee S S, Yalamanchili D, Noori N, Debbi E, Mirocha J, Lin C A, Moon C N. Increasing use of reverse total shoulder arthroplasty for proximal humerus fractures in elderly patients. Orthopedics 2017; 40: e982-e89. Rangan A, Handoll H, Brealey S, Jefferson L, Keding A, Martin B C, Goodchild L, Chuang L H, Hewitt C, Torgerson D. Surgical vs nonsurgical treatment of adults with displaced fractures of the proximal humerus: the PROFHER randomized clinical trial. JAMA 2015; 313: 1037-47. Rosas S, Law T Y, Kurowicki J, Formaini N, Kalandiak S P, Levy J C. Trends in surgical management of proximal humeral fractures in the Medicare population: a nationwide study of records from 2009 to 2012. J Shoulder Elbow Surg 2016; 25: 608-13.

657

Spross C, Meester J, Mazzucchelli R A, Puskás G J, Zdravkovic V, Jost B. Evidence-based algorithm to treat patients with proximal humerus fractures: a prospective study with early clinical and overall performance results. J Shoulder Elbow Surg 2019; 28: 1022-32. Sumrein B O, Huttunen T T, Launonen A P, Berg H E, Fellander-Tsai L, Mattila V M. Proximal humeral fractures in Sweden: a registry-based study. Osteoporos Int 2017; 28: 901-07. van der Merwe M, Boyle M J, Frampton C M A, Ball C M. Reverse shoulder arthroplasty compared with hemiarthroplasty in the treatment of acute proximal humeral fractures. J Shoulder Elbow Surg 2017; 26: 1539-45. Wennergren D, Ekholm C, Sandelin A, Möller M. The Swedish fracture register: 103,000 fractures registered. BMC Musculoskeletal Disorders 2015; 16: 338. Wennergren D, Stjernstrom S, Moller M, Sundfeldt M, Ekholm C. Validity of humerus fracture classification in the Swedish fracture register. BMC Musculoskelet Disord 2017; 18: 251. Wennergren D, Bergdahl C, Selse A, Ekelund J, Sundfeldt M, Möller M. Treatment and re-operation rates in one thousand and three hundred tibial fractures from the Swedish Fracture Register. Eur J Orthop Surg Traumatol 2021; 31(1): 143-54. Zhu Y, Lu Y, Shen J, Zhang J, Jiang C. Locking intramedullary nails and locking plates in the treatment of two-part proximal humeral surgical neck fractures: a prospective randomized trial with a minimum of three years of follow-up. J Bone Joint Surg Am 2011; 93: 159-68.

658

Acta Orthopaedica 2021; 92 (6): 658–664

Implant survival of 662 dual-mobility cups and 727 constrained liners in primary THA: small femoral head size increases the cumulative incidence of revision Oskari PAKARINEN 1, Olli LAINIALA 1,2, Aleksi REITO 1, Perttu NEUVONEN 1, Keijo MÄKELÄ 3, and Antti ESKELINEN 1 1 Coxa

Hospital for Joint Replacement, and Faculty of Medicine and Health Technology, University of Tampere, Tampere; 2 Department of Radiology, Tampere University Hospital, Tampere; 3 Department of Orthopedics and Traumatology, Turku University Hospital, and University of Turku, Turku, Finland Correspondence: oskari.pakarinen@tuni.fi Submitted 2020-12-14. Accepted 2021-05-26.

Background and purpose — In total hip arthroplasty (THA), the risk for dislocation can be reduced using either dual-mobility cups (DMCs) or constrained liners (CLs). There are few studies comparing these concepts in primary THA. Therefore, we compared the cumulative incidence of revision in primary THA patients treated with DMC or CL with varying head sizes with conventional THA patients as reference group. Patients and methods — We performed a cohort study based on the Finnish arthroplasty register, comparing DMCs and CLs operated over the period 2000–2017. DMCs were divided into 2 groups based on the implant design: “DMC Trident” group (n = 399) and “DMC Others” group (n = 263). CLs were divided based on the femoral head size: “CL 36 mm” group (n = 425) and “CL < 36 mm” group (n = 302). All conventional primary THAs operated on in 2000–2017 with 28–36 mm femoral head were included as control group (“Conventional THA” group, n = 102,276). Implant survival was calculated by the corresponding cumulative incidence function with revision as the endpoint and death as competing event. Also, the prevalence of different reasons for revision was compared. Results — The 6-year cumulative incidence function estimates for the first revision were 6.9% (95% CI 4.0–9.7) for DMC Trident, 5.0% (CI 1.5–8.5) for DMC Others, 13% (CI 9.3–17) for CL < 36 mm, 6.3% (3.7–8.9) for CL 36 mm, and 4.7% (CI 4.5–4.8) for control group (conventional THA). The prevalence of dislocation revision was high (5.0%, CI 2.9– 8.2) in the CL < 36 mm group compared with other groups. Interpretation — The DMC and CL 36 mm groups had promising mid-term survival rates, comparable to those of primary conventional THA group. The revision rate of CLs with < 36 mm head was high, mostly due to high prevalence of dislocation revisions. Therefore, CLs with 36 mm femoral head should be preferred over smaller ones.

Dislocation is the most common complication after primary total hip arthroplasty (THA). Moreover, according to data from large national registers, dislocation is the most common reason for revision during the first postoperative year (Australian Orthopaedic Association 2019, National Joint Registry 2019). In primary THA, the prevalence of dislocation varies from 0.4% to 4.1% (Blom et al. 2008, Itokawa et al. 2013, Ravi et al. 2014, Klasan et al. 2019, Pakarinen et al. 2020, Hermansen et al. 2021). Recently, the role of implants that increase hip stability has been emphasized for patients who are at high risk of dislocation (Hernigou et al. 2010, 2016, Nessler et al. 2020). These implants use either larger femoral heads or have been specifically designed to prevent dislocations, as in dualmobility cups (DMC) and constrained acetabular liners (CL) (Guyen 2017, Van der Merwe 2018, Reina et al. 2019). The use of DMCs gained worldwide popularity during the 2010s (American Joint Replacement Registry 2018, Bloemheuvel et al. 2019). The advantage of DMCs with regard to hip stability is that larger femoral heads can be used. Despite the small femoral head in the inner bearing, a large mobile polyethylene liner can be used as an articulating head for the outer bearing (Terrier et al. 2017). Although CLs have differences in design, enhanced stability is achieved by more than a hemispheric coverage of the liner and a metallic reinforcement ring, which mechanically secures the head into the liner. The disadvantage of CLs is that the range of motion (ROM) of the hip joint is limited by their structure. This can lead to impingement, breakage of the locking mechanism, and increased wear (Burroughs et al. 2001). Therefore, CLs have traditionally been reserved for a very limited group of patients, especially those with abductor insufficiency (Herman et al. 2019). Even though the results of DMCs and CLs in both primary and revision THA have previously been studied separately, there is a scarcity of literature in which these concepts are

Acta Orthopaedica 2021; 92 (6): 658–664

659

Unlike the other designs that consist of a 1-piece metal cup articulating against a large dual-mobility head, the Trident is a classical modular uncemented cup, into Primary THA with a Primary THA with a which a metal liner is inserted. As Other primary THAs Constrained Liner Dual-Mobility Cup n = 132,097 a result, size of the dual-mobility n = 730 n = 662 head is smaller in Trident than in Excluded Excluded the 1-piece cup DMC designs. Missing femoral > 36 mm or < 28 mm or The Trident has also been associhead size missing femoral head size n=3 n = 29,821 ated with a risk of mal-insertion of the liner (Langdown et al. 2007, Romero et al. 2020). The “DMC 22 mm 32 mm 36 mm 28 mm Trident design Other design femoral femoral femoral femoral n = 399 n = 263 Trident group” included all Trihead head head head n = 64 n = 111 n = 127 n = 425 dent DMCs (n = 399), and the “DMC Others group” (n = 263) included the rest of the DMCs. The ‘DMC Trident The ‘DMC Others The ‘CL < 36 mm The ‘CL 36 mm The ‘Conventional group’ group’ group’ group’ THA group’ To assess the role of head size in n = 399 n = 263 n = 302 n = 425 n = 102,276 the survivorship of CLs, 2 differFigure 1. Flow diagram of the study. THA = total hip arthroplasty, DMC = dual-mobility cup, CL = con- ent groups were formed: the “CL strained liner. < 36 mm group” (n = 302), which included 22 mm (n = 64), 28 mm compared with each other in primary THA. Further, although (n = 111), and 32 mm (n = 127) femoral heads, and the “CL the effect of femoral head size on hip stability has been widely 36 mm group” that comprised 36 mm femoral heads only (n studied, this is not the case with CLs. In this study, we com- = 425). In the data, there were no CLs with femoral heads pared the cumulative incidence of revision in primary THA larger than 36 mm. The most commonly used DMCs and CLs patients treated with DMC or CL with varying head sizes with are listed in Table 1. The patients were operated on in 2012– 2017 in the DMC Trident group (median follow-up 2.4 years), conventional THA patients as reference group. 2007–2017 in the DMC Others group (3.3 years), 2000–2017 in the CL < 36 mm group (2.3 years), and 2005–2017 in the CL 36 mm group (2.4 years). We included all conventional Patients and methods primary THAs performed in 2000–2017 with a 28–36 mm Data sources sized femoral head as a control group (n = 102,276, median This study is based on data from the Finnish Arthroplasty follow-up 5.5 years). Operations with < 28 mm were excluded Register (FAR), which has 95% completeness of all primary from the control group because the smallest head sizes have THAs and 81% completeness of all revision THAs performed been associated with high risk of dislocation, and > 36 mm in Finland (Finnish Arthroplasty Register 2020). All orthope- heads were excluded because we assume that they mostly dic units are obligated to provide essential information to the consist of large metal-on-metal heads that are known to have Finnish National Institute of Health and Welfare. Death dates a high revision rate because of the adverse reaction to metal were obtained from the Population Information System main- debris (Berry et al. 2005, Lainiala et al. 2019). tained by the Finnish Population Register Centre. Statistics Study population Follow-up started on the day of primary THA and ended on We identified all primary THAs performed in Finland between the day of revision, death, or June 10, 2018 (the date of data January 1, 2000 and December 31, 2017 in which either a collection), whichever came first. Revision was defined as a DMC or CL cup was used. The final data included 662 hips new surgical procedure, including partial or complete removal with DMCs and 727 hips with uncemented cups with CLs, or exchange of any implant component. The indications for which represents 1.0% of the primary THA patients during revision specified in the FAR database were dislocation, perithat time period (Figure 1). We included all cemented and prosthetic femoral fracture (PFF), aseptic loosening, deep uncemented DMCs with either 22 mm or 28 mm metal or infection, pain, and other reasons. The overall mortality was high in all study groups compared ceramic inner femoral heads and larger outer dual-mobility liners of all sizes. DMCs were split into 2 groups, because with the endpoint of interest, i.e., the number of revisions the most common DMC in our data, i.e., the Trident (Stryker, (Table 2). Our original plan was to analyze the risk for revision Mahwah, NJ, USA), differs markedly from the other designs. using the Fine–Gray competing risk regression with first reviAll primary THAs performed in Finland in 2000–2017 n = 133,489

660

Acta Orthopaedica 2021; 92 (6): 658–664

Table 1. Cup and liner designs included in the study cohort

Dual-mobility (Trident)

Dual-mobility (Others)

Cup design (n) Trident (399) Restoration ADM (152) (Liner design %) (MDM 92%) (Restoration ADM 100%) (Restoration 8%) Novae E (80) (Novae E 100%) Avantage (15) (Avantage 100%) Others (16) Total 399 263

CL < 36 mm

CL 36 mm

Continuum (98) Vision Ringloc (293) (Continuum 100%) (Freedom 100%) Trident (80) Regenerex (72) (Trident 90%) (Freedom 100%) (Data missing 10%) Continuum (15) Pinnacle (44) (Continuum 100%) (Pinnacle 100%) G7 (12) Trabecular metal (35) (Freedom 100%) (Trabecular metal 72%) Pinnacle (11) (Vision Ringloc 17%) (Pinnacle 100%) (Continuum 9%) Others (22) Vision Ringloc (23) (Vision Ringloc 100%) Others (15) 302 425

Conventional THA Pinnacle (17,136) Trident (12,553) Contemporary (12,314) Continuum (11,895) R3 (7,515) Reflection (6,120) Vision Ringloc (4,941) ABG (3,885) Exeter (2,571) Exceed (2,047) Lubinus (1,572) G7 (1,183) Others (18,544) 102,276

DMC = dual-mobility cup, CL = constrained liner. Cup and liner designs listed with manufacturer: Trident (Stryker, Mahwah, NJ, USA); Restoration (Stryker); Novae E (Fischer Medical, Glostrup, Denmark); Avantage (Zimmer Biomet, Warsaw, IN, USA); Continuum (Zimmer Biomet); Pinnacle (Depuy Orthopaedics, Warsaw, IN, USA); Trabecular Metal (Zimmer Biomet); Vision RingLoc (Zimmer Biomet); Regenerex (Zimmer Biomet); G7 (Zimmer Biomet); Freedom (Zimmer Biomet); Contemporary (Stryker); R3 (Smith & Nephew, Memphis, TN, USA); Reflection (Smith & Nephew, London, UK); ABG (Stryker); Exeter (Stryker); Exceed (Zimmer Biomet); Lubinus (Link, Hamburg, Germany).

sion of the primary THA as the primary endpoint and death as the competing endpoint. Age, sex, primary diagnosis, femoral and acetabular fixation were supposed to be included in the analysis. However, multiple proportional hazards assumption violations were found after the inspection of the corresponding log-survival against log-time across categorized covariate levels, which would have made the interpretation of the results difficult. Moreover, it was evident that even after the adjustments there would be a substantial amount of unmeasured confounding as a result of selection bias. Therefore, we decided to calculate only the implant survival rate using corresponding cumulative incidence function (CIF) with patient death as a competing event and accept that the patient-related factors could not be reliably adjusted. The first sensitivity analysis was similar to the main one, but the endpoint was revision for any reason except PFF because the increased risk for this type of complication is mostly associated with the type of femoral stem and not the acetabular component (Thien et al. 2014). In the second sensitivity analysis, the different head sizes within the CL < 36 mm group were compared with each other. We calculated 95% confidence intervals (CI) for CIF graphs. CI for proportions were calculated using Wilson score interval. The analyses were performed using IBM SPSS 25.0 (IBM Corp, Armonk, NY, USA) and R statistical software (R Centre for Statistical Computing, Vienna, Austria). Ethics, funding, and potential conflicts of interest In accordance with Finnish regulations, informed patient consent was not required as the patients were not contacted. This work was supported by the competitive research funds of Pirkanmaa Hospital District, Tampere, Finland (representing governmental funding), Orion Research Foundation,

Vappu Uuspää Foundation, and Finnish Research Foundation for Orthopaedics and Traumatology. The sources of funding had no role at any stage of the study. Individual potential conflict of interests: OP, OL, AR, PN, KM: None. AE: Zimmer Biomet, paid lectures; Depuy Synthes and Zimmer Biomet, institutional research support (not related to current study).

Results Patient demographics of the 4 study groups and the control group of conventional THAs are summarized in Table 2. The specific primary reason for operation has been recorded in FAR data since 2014. Therefore, the more detailed indications for primary THA for patients operated on in 2014–2017 are presented in Table 3 (see Supplementary data). In the 4 study groups, 94 hips were revised during the follow-up (6.8%, CI 5.6–8.3). The most common reasons for revision were PFF (n = 25, 1.8%), deep infection (n = 25, 1.8%), and dislocation (n = 17, 1.2%). In the Conventional THA group, the overall revision rate was 5.9% (n = 6,069, CI 5.8–6.1), and the leading causes of revision were dislocation (n = 1,422, 1.4%) and aseptic loosening (n = 1,307, 1.3%). The 1-year postoperative mortality was 3.5% in the DMC Trident group, 5.7% in the DMC Others group, 11% in the CL < 36 mm group, 20% in the CL 36 mm group, and 1.8% in the Conventional THA group. The cumulative incidence of the first revision of the study groups is presented in Figure 2. At 6 years postoperatively, the CIF estimate of the first revision was 6.9% (CI 4.0–9.7) for DMC Trident, 5.0% (CI 1.5–8.5) for DMC Others, 13% (CI 9.3–17) for CL < 36 mm,

Acta Orthopaedica 2021; 92 (6): 658–664

661

Table 2. Patient demographics. Values are count (%) unless otherwise specified DMC DMC CL CL Conventional (Trident) (Others) < 36 mm 36 mm THA n = 399 n = 263 n = 302 n = 425 n = 102,276 Follow-up, years, 2.4 3.3 2.3 2.4 5.5 median (IQR) (1.1–3.9) (1.1–6.1) (0.9–4.0) (0.8–4.8) (2.6–10) Age, mean (SD) 71 (11) 69 (11) 73 (12) 71 (12) 67 (11) BMI, mean (SD) 27 (5.2) 26 (5.2) 26 (5.0) 27 (5.5) 28 (4.8) Sex Male 156 (39) 95 (36) 105 (35) 199 (47) 41,053 (40) Female 243 (61) 167 (64) 197 (65) 225 (53) 61,201 (60) Data missing 0 (0.0) 1 (0.4) 0 (0.0) 1 (0.2) 22 (0.0) Diagnosis Osteoarthritis 248 (62) 147 (56) 99 (33) 109 (26) 87,199 (85) Hip fracture 97 (24) 85 (32) 88 (29) 101 (24) 4,220 (4.1) Other 54 (14) 31 (12) 115 (38) 215 (51) 10,857 (11) a ASA score 1 8 (2.8) 1 (0.8) 3 (1.8) 5 (2.7) 3,836 (12) 2 91 (32) 37 (30) 30 (18) 33 (18) 15,179 (49) 3 175 (62) 79 (64) 110 (67) 111 (60) 11,367 (37) 4 9 (3.1) 7 (5.6) 21 (13) 36 (20) 493 (1.6) Approach a Posterior 198 (70) 106 (85) 165 (98) 166 (90) 24,603 (81) Anterolateral (mod. Hardinge) 86 (30) 19 (15) 4 (2.4) 19 (10) 5,892 (19) Cup fixation Uncemented 399 (100) 200 (76) 259 (86) 193 (45) 68,948 (67) Cemented 0 (0.0) 49 (19) 24 (7.9) 219 (52) 29,270 (29) Data missing 0 (0.0) 14 (5.3) 19 (6.3) 13 (3.1) 4,058 (4.0) Femoral fixation Uncemented 102 (34) 114 (43) 132 (44) 230 (54) 56,593 (55) Cemented 294 (74) 145 (55) 149 (49) 115 (27) 41,161 (41) Data missing 3 (0.8) 4 (1.5) 21 (7.0) 80 (19) 4,522 (4.4) Femoral head material Metal 323 (81) 181 (69) 270 (89) 408 (96) 68,739 (67) Ceramic 23 (5.8) 65 (25) 15 (5.0) 6 (1.4) 33,537 (33) Data missing 53 (13) 17 (6.5) 17 (5.6) 11 (2.6) 0 (0.0) Total mortality 34 (8.5) 38 (14) 99 (33) 212 (50) 22,180 (22) Revision surgery 23 (5.8) 9 (3.4) 39 (13) 23 (5.4) 6,069 (5.9) Reason for revision Aseptic loosening 2 (0.5) 1 (0.4) 2 (0.7) 3 (0.7) 1,307 (1.3) Deep infection 7 (1.8) 2 (0.8) 6 (2.0) 10 (2.4) 728 (0.7) PFF 7 (1.8) 3 (1.1) 11 (3.6) 4 (0.9) 769 (0.8) Dislocation 1 (0.3) 0 (0) 15 (5.0) 1 (0.2) 1,422 (1.4) Pain only 0 (0) 1 (0.4) 0 (0) 0 (0) 71 (0.0) Others 5 (1.3) 2 (0.8) 3 (1.0) 1 (0.2) 895 (0.9) Unknown reason 1 (0.3) 0 (0) 2 (0.7) 4 (0.9) 877 (0.9) DMC = dual-mobility cup, CL = constrained liner, IQR = interquartile range, BMI = body mass index, ASA score = American Society of Anesthesiologists score, PFF = periprosthetic femoral fracture. a Data available only for patients operated on after 2014. BMI coverage in data: 66% in DMC Trident group, 47% in DMC Others group, 47% in CL < 36 mm group, 33% in CL 36 mm group, and 28% in Conventional THA group.A

6.3% (CI 3.7–8.9) for CL 36 mm, and 4.7% (CI 4.5–4.8) for Conventional THA (Table 4). During the same 6-year period, the CIF estimate of death was 13% (CI 8.3–18) for DMC Trident, 17% (CI 11–23) for DMC Others, 37% (CI 29–44) for CL < 36mm, 54% (CI 48–60) for CL 36 mm, and 12% (CI 12–12) for Conventional THA. The CIF estimates are visually presented in Figure 3 (see Supplementary data). When revisions for PFF were excluded, the 6-year CIF estimate of the first revision was 5.1% (CI 2.5–7.7) for DMC Tri-

Cumulative incidence 0.20 Conventional THA DMC (other) DMC (Trident) CL < 36 mm CL 36 mm

0.15

0.10

0.05

0.00

Years after index operation Figure 2. Cumulative incidence of revision.

dent, 3.2% (CI 0.8–5.5) for DMC Others, 9.5% (CI 5.9–13) for CL < 36 mm, 5.3% (CI 2.9–7.7) for CL 36 mm, and 4.1% (CI 4.0–4.2) for Conventional THA (Table 4). The reasons for revision surgery are presented in Table 2. Dislocation was the most common reason for revision in the CL < 36 mm group (5.0% prevalence, CI 2.9–8.2), but a very rare reason for revision in the CL 36 mm (0.2%, CI 0.01–1.5), DMC Trident (0.3%, CI 0.01–1.6), and DMC Others (0.0%, CI 0.0–1.8) groups. In the Conventional THA group, the prevalence of dislocation revision was 1.4% (CI 1.3–1.5) (Table 2). When the 22 mm, 28 mm, and 32 mm head sizes within the CL < 36 mm group were compared, the cumulative incidence estimates of the first revision were similar regardless of femoral head size (Table 5, see Supplementary data).

Discussion

We found that the cumulative incidence of revision in 6-year follow-up was comparable with the conventional THA patients in the DMC Others group, and only slightly higher in the DMC Trident and CL 36 mm groups. However, in the CL < 36 mm group the revision rate was remarkably higher. Dislocation was a major cause of revision in CL < 36 mm group, but a rarity in other study groups. In total, DMCs and CLs represented 1.0% of the primary THAs implanted in Finland during the study period, indicating they were used in a selected population, presumably with high risk of dislocation. The survivorship of DMCs has been reported to be comparable to that of conventional THA in primary THA with

662

Acta Orthopaedica 2021; 92 (6): 658–664

be maximized when a constrained liner is used. The rationale for this is to increase the head-to DMC DMC CL CL Conventional neck ratio and lever-out distance (Trident) (Others) < 36 mm 36 mm THA and thus decrease the risk for impingement and dislocation All-cause revision Risk of revision (Soong et al. 2004, Brown et al. 1 year 4.6 (2.5–6.7) 1.5 (0.4–3.0) 9.3 (6.0–13) 3.3 (1.6–5.0) 2.3 (2.3–2.4) 2014). However, no prior studies 3 years 5.7 (3.3–8.0) 3.2 (0.8–5.5) 13 (9.3–17) 5.5 (3.2–7.7) 3.5 (3.4–3.6) have reported whether increasing 6 years 6.9 (4.0–9.7) 5.0 (1.5–8.5) 13 (9.3–17) 6.3 (3.7–8.9) 4.7 (4.5–4.8) Risk of death the head size with CLs actually 1 year 3.2 (1.4–5.0) 5.7 (2.8–8.6) 8.2 (5.0–11) 18 (14–22) 1.3 (1.2–1.4) results in better outcome. 3 years 9.0 (5.7–12) 10 (6.3–14) 20 (15–25) 34 (29–39) 4.6 (4.4–4.7) Recent studies have reported 6 years 13 (8.3–18) 17 (11–23) 37 (29–44) 54 (48–60) 12 (12–12) All-cause revision (PFF excluded) good survival rates for CLs in pri Risk of revision mary THA (Clave et al. 2016, Gill 1 year 2.8 (1.2–4.5) 1.1 (0.0–2.4) 6.7 (3.9–9.6) 2.9 (1.3–4.4) 2.0 (1.9–2.0) et al. 2016, Karvonen et al. 2020). 3 years 3.9 (1.9–5.9) 2.8 (0.5–5.0) 9.5 (6.0–13) 4.4 (2.3–6.5) 3.0 (2.9–3.1) 6 years 5.1 (2.5–7.7) 3.2 (0.8–5.5) 9.5 (5.9–13) 5.3 (2.9–7.7) 4.1 (4.0–4.2) In our study, the overall 6-year Risk of death survival rate of CLs with 36 mm 1 year 3.5 (1.6–5.4) 6.1 (3.1–9.1) 9.4 (5.9–13) 18 (15–22) 1.4 (1.3–1.5) head was also promising, but there 3 years 9.4 (6.0–13) 11 (6.6–15) 22 (16–27) 35 (30–40) 4.7 (4.6–4.8) 6 years 13 (8.6–18) 17 (12–23) 39 (31–47) 55 (49–61) 12 (12–12) were considerably more revi sions when CLs were used with < 36 mm head. This difference is mostly explained by higher rates mid-term follow-up (Kreipke et al. 2019), whereas DMCs of revisions for dislocation and PFF in the CL < 36 mm group. have been associated with low dislocation and revision rates These results may indicate that a large enough femoral head in primary THA with dislocation-prone patients (Harwin et with CLs allows a wide enough ROM that prevents impingeal. 2017, Jones et al. 2019). Some implant-related compli- ment and is therefore not as prone to dislocations. The revision cations, such as intraprosthetic dislocations, have, however, estimates in the CL 36 mm group were only marginally higher been reported (Addona et al. 2019). In our study, the mid- compared with conventional THA patients, even though the term survival rate of the DMC Others group was comparable patients in the former group were remarkably more morbid to the conventional THA patients. No dislocation revisions on average. Still, a failed THA with CL may predispose to occurred even though a third of the patients in the former recurrent revision surgeries (Hellman et al. 2018). Thus, more group were operated on for hip fracture. In the DMC Tri- studies are needed before the use of CLs can be recommended dent group the revision estimates were only slightly higher, to prevent dislocations in primary THA for patients who do and only 1 hip (0.3%) was revised for dislocation. Due to not have an obvious, strong predisposing factor for dislocathe register-based study setting, we were unable to verify the tion, such as abductor muscle deficiency, tumor resection, or reasons why surgeons chose either DMCs or CLs in primary femoral neck fracture. In patients with an increased risk of THAs. We can only assume that the reason was an anticipated dislocation but without abductor deficiency, DMC might be a high risk for dislocation in most of the cases. Nonetheless, the safer option as it provides better impingement-free ROM, and indications may have differed because both the mortality and thus has smaller risk for mechanical failure. the proportion of patients operated on for reasons other than In a recent study, the survivorship of the Freedom conosteoarthritis or hip fracture differ between the DMC and CL strained acetabular liner (Zimmer Biomet) was similarly comgroups. Since 2014, patients operated on for tumor comprised pared with conventional primary THA designs (Karvonen et a third of the patients in the CL 36 mm group (see Supple- al. 2020). In our study, 89% of the CLs with 36 mm heads mentary data 1), which partially explains the high mortality were Freedom as it is the most commonly used CL design in rate in this group. Regardless of the proportion of high-risk Finland. Freedom liners enable the use of a 36 mm head in patients in our data, our results are in line with other recent cup sizes as small as 50 mm, whereas the next smallest cup studies and do not oppose the idea of using DMCs in primary accepting a 36 mm head in our data is the Pinnacle 56 mm THA for patients who have a higher risk for dislocation. Still, (Karvonen et al. 2020). Because the survival rate in the CL 36 longer follow-up is needed to see how well these implants mm group was excellent irrespective of the CL cup design, it actually bear the test of time. seems that the larger head size, not the cup design itself, may The biggest advantage with larger femoral heads is the to be the key to success when CLs are used in primary THA. decreased risk of dislocation (Berry et al. 2005, Hailer et al. However, for some patients even the use of 50 mm diameter 2012, Howie et al. 2012, Kostensalo et al. 2013). Thoms and cup is impossible and therefore the use of CL with 36 mm Marwin (2008) have suggested that femoral head size ought to head is not an option. Table 4. Cumulative incidence function estimates at 1, 3, and 6 years for the first revision and death with 95% confidence intervals

Acta Orthopaedica 2021; 92 (6): 658–664

We acknowledge a few weaknesses in this study. Because not all DMCs and CLs have identical designs, there could be implant-related factors that have affected the risk for revision that we are not aware of. The rather short mean follow-up limits the interpretation of our long-term survivorship comparison. Because of the heterogeneity in the study population, we reported only unadjusted CIF estimates. Even after the available factors had been adjusted, the comparison would not have been equal because the indications for the use of CL or DMC in primary THA are different compared with conventional THA implants. There were differences in mortality, distribution of ASA score, and the primary reason for operation between the DMCs and CLs, implying that there is confounding by indication also between these groups (see Table 2). Because the mortality and the number of patients operated on for reasons other than osteoarthritis were highest in the CL 36 mm group, it is unlikely that the comorbidities would explain the inferior results in the CL < 36 mm group compared with the CL 36 mm group. Because of the limitations in the data, the impact of unmeasured confounding must be considered in the interpretation of the results. As this was a register study, we could not comprehensively assess the clinical outcome of the operations (e.g., patient-reported outcome measures). Conclusion The DMC Other group showed a comparable revision rate with conventional THA implants in 6-year follow-up, and the revision rate for the DMC Trident and CL 36 mm groups was only slightly higher. Conversely, the 6-year revision rate was clearly higher in the CL < 36 mm group. The difference was mostly explained by dislocations because revision for dislocation was a very rare event in both DMC groups and in the CL 36 mm group, whereas it was the most common type of revision in the CL < 36 mm group. The prevalence of PFF revision was also highest in the CL < 36 mm group. The good overall survival rate and low number of dislocation revisions with DMCs support the increased use of these devices over recent years. However, studies with long-term follow-up are still needed. According to our results, it seems that enlarging the femoral head with CLs enhances the survival rate of the implant. Therefore, we recommend that when a CL is used, a 36 mm femoral head should be preferred over a smaller head to avoid complications, especially dislocations. Supplementary data Tables 3 and 5 and Figure 3 are available as supplementary data in the online version of this article, http://dx.doi.org/10. 1080/17453674.2021.1939597 All authors: interpretation of results, manuscript editing and approval of the final version. OP: data collection, statistical analyses, writing of the first version of the manuscript. AR: statistical analyses. AE: study planning, project supervision.

663

The authors would like to thank Peter Heath MA for linguistic help. Acta thanks B Willem Schreurs and Claus Varnum for help with peer review of this study.

Addona J L, Gu A, De Martino I, Malahias M-A, Sculco T P, Sculco P K. High rate of early intraprosthetic dislocations of dual mobility implants: a single surgeon series of primary and revision total hip replacements. J Arthroplasty 2019; 34(11): 2793-8. American Joint Replacement Registry. Annual Report; 2018, http://www. ajrr.net/publications-data/annual-reports (Accessed December 11, 2020). Australian Orthopaedic Association. Annual Report; 2019, https://aoanjrr. sahmri.com/annual-reports-2019 (Accessed December 11, 2020). Berry D J, von Knoch M, Schleck C D, Harmsen W S. Effect of femoral head diameter and operative approach on risk of dislocation after primary total hip arthroplasty. J Bone Joint Surg Am 2005; 87(11): 2456-63. Bloemheuvel E M, van Steenbergen L N, Swierstra B A. Dual mobility cups in primary total hip arthroplasties: trend over time in use, patient characteristics, and mid-term revision in 3,038 cases in the Dutch Arthroplasty Register (2007–2016). Acta Orthop 2019; 90(1): 11-14. Blom A W, Rogers M, Taylor A H, Pattison G, Whitehouse S, Bannister G C. Dislocation following total hip replacement: the Avon Orthopaedic Centre experience. Ann R Coll Surg Engl 2008; 90(8): 658-62. Brown T D, Elkins J M, Pedersen D R, Callaghan J J. Impingement and dislocation in total hip arthroplasty: mechanisms and consequences. Iowa Orthop J 2014; 34: 1-15. Burroughs B R, Golladay G J, Hallstrom B, Harris W H. A novel constrained acetabular liner design with increased range of motion. J Arthroplasty 2001; 16(8 Suppl. 1): 31-6. Clave A, Maurer D, Tristan L, Dubrana F, Lefevre C, Pandit H. Midterm xurvivorship of the Lefevre Constrained Liner: a consecutive multisurgeon series of 166 cases. J Arthroplasty 2016; 31(9): 1970-8. Finnish Arthroplasty Register. https://www.thl.fi/far (Accessed December 11, 2020). Gill K, Whitehouse S L, Hubble M J W, Wilson M J. Short-term results with a constrained acetabular liner in patients at high risk of dislocation after primary total hip arthroplasty. Hip Int 2016; 26(6): 580-4. Guyen O. Constrained liners, dual mobility or large diameter heads to avoid dislocation in THA. EFORT Open Rev 2017; 1(5): 197-204. 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 2012; 83(5): 442-8. Harwin S F, Mistry J B, Chughtai M, Khlopas A, Gwam C, Newman J M, Higuera C A, Bonutti P M, Malkani A L, Kolisek F R, Delanois R E, Mont M A. Dual mobility acetabular cups in primary total hip arthroplasty in patients at high risk for dislocation. Surg Technol Int 2017; 30: 251-8. Hellman M D, Kaufman D J, Sporer S M, Paprosky W G, Levine B R, Della Valle C J. High rate of failure after revision of a constrained liner. J Arthroplasty 2018; 33(7S): S186-90. Herman A, Masri B A, Duncan C P, Greidanus N V, Garbuz D S. Multivariate analysis of risk factors for re-dislocation after revision for dislocation after total hip arthroplasty. Hip Int 2019; 1120700019831628. 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 2021; 103(4): 295-302 Hernigou P, Filippini P, Flouzat-Lachaniette C-H, Batista S U, Poignard A. Constrained liner in neurologic or cognitively impaired patients undergoing primary THA. Clin Orthop Relat Res 2010; 468(12): 3255-62.

664

Hernigou P, Trousselier M, Roubineau F, Bouthors C, Flouzat Lachaniette C H. Dual-mobility or constrained liners are more effective than preoperative bariatric surgery in prevention of THA dislocation. Clin Orthop Relat Res 2016; 474(10): 2202-10. Howie D W, Holubowycz O T, Middleton R. Large femoral heads decrease the incidence of dislocation after total hip arthroplasty: a randomized controlled trial. J Bone Joint Surg Am 2012; 94(12): 1095-102. Itokawa T, Nakashima Y, Yamamoto T, Motomura G, Ohishi M, Hamai S, Akiyama M, Hirata M, Hara D, Iwamoto Y. Late dislocation is associated with recurrence after total hip arthroplasty. Int Orthop 2013; 37(8): 1457-63. Jones C W, De Martino I, D’Apolito R, Nocon A A, Sculco P K, Sculco T P. The use of dual-mobility bearings in patients at high risk of dislocation. Bone Joint J 2019; 101-B(1_Supple_A): 41-5. Karvonen M, Laaksonen I, Pulkkinen P, Eskelinen A, Haapakoski J, Puhto A-P, Kettunen J, Manninen M, Makela K T. Implant survival of constrained acetabular device in primary total hip arthroplasty based on data from the Finnish Arthroplasty Register. J Arthroplasty 2020; 35(1): 219-23. Klasan A, Neri T, Oberkircher L, Malcherczyk D, Heyse T J, Bliemel C. Complications after direct anterior versus Watson–Jones approach in total hip arthroplasty: results from a matched pair analysis on 1408 patients. BMC Musculoskelet Disord 2019; 20(1): 77. Kostensalo I, Junnila M, Virolainen P, Remes V, Matilainen M, Vahlberg T, Pulkkinen P, Eskelinen A, Makela K T. Effect of femoral head size on risk of revision for dislocation after total hip arthroplasty: a populationbased analysis of 42,379 primary procedures from the Finnish Arthroplasty Register. Acta Orthop 2013; 84(4): 342-7. Kreipke R, Rogmark C, Pedersen A B, Karrholm J, Hallan G, Havelin L I, Makela K, Overgaard S. Dual mobility cups: effect on risk of revision of primary total hip arthroplasty due to osteoarthritis: a matched population-based study using the Nordic Arthroplasty Register Association database. J Bone Joint Surg Am 2019; 101(2): 169-76. Lainiala O S, Reito A P, Nieminen J J, Eskelinen A P. Declining revision burden of metal-on-metal hip arthroplasties. J Arthroplasty 2019; 34(9): 2058-64.e1. Langdown A J, Pickard R J, Hobbs C M, Clarke H J, Dalton D J N, Grover M L. Incomplete seating of the liner with the Trident acetabular system: a cause for concern? J Bone Joint Surg Br 2007; 89(3): 291-5.

Acta Orthopaedica 2021; 92 (6): 658–664

National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. Annual Report; 2019, https://reports.njrcentre.org.uk/Portals/0/PDFdownloads/NJR%2016th%20Annual%20Report%202019.pdf (Accessed December 11, 2020). Nessler J M, Malkani A L, Sachdeva S, Nessler J P, Westrich G, Harwin S F, Mayman D, Jerabek S. Use of dual mobility cups in patients undergoing primary total hip arthroplasty with prior lumbar spine fusion. Int Orthop 2020; 44(5): 857-62. Pakarinen O A, Neuvonen P S, Reito A R P, Eskelinen A P. Increased risk for dislocation after introduction of the Continuum cup system: lessons learnt from a cohort of 1,381 THRs after 1-year follow-up. Acta Orthop 2020; 91(3): 279-85. Ravi B, Jenkinson R, Austin P C, Croxford R, Wasserstein D, Escott B, Paterson J M, Kreder H, Hawker G A. Relation between surgeon volume and risk of complications after total hip arthroplasty: propensity score matched cohort study. BMJ 2014; 348: g3284. Reina N, Pareek A, Krych A J, Pagnano M W, Berry D J, Abdel M P. Dual-mobility constructs in primary and revision total hip arthroplasty: a systematic review of comparative studies. J Arthroplasty 2019; 34(3): 594603. Romero J, Wach A, Silberberg S, Chiu Y-F, Westrich G, Wright T M, Padgett D E. 2020 Otto Aufranc Award: Malseating of modular dual mobility liners. Bone Joint J 2020; 102-B(7_Supple_B): 20-6. Soong M, Rubash H E, Macaulay W. Dislocation after total hip arthroplasty. J Am Acad Orthop Surg 2004; 12(5): 314-21. Terrier A, Latypova A, Guillemin M, Parvex V, Guyen O. Dual mobility cups provide biomechanical advantages in situations at risk for dislocation: a finite element analysis. Int Orthop 2017; 41(3): 551-6. Thien T M, Chatziagorou G, Garellick G, Furnes O, Havelin L I, Makela 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. Thoms R J, Marwin S E. A unique failure mechanism of a constrained total hip arthroplasty: a brief review of the literature. J Arthroplasty 2008; 23(2): 293-8. Van der Merwe J M. Comprehensive review of current constraining devices in total hip arthroplasty. J Am Acad Orthop Surg 2018; 26(14): 479-88.

Acta Orthopaedica 2021; 92 (6): 665–672 665

Risk factors for prosthetic joint infections following total hip arthroplasty based on 33,337 hips in the Finnish Arthroplasty Register from 2014 to 2018 Valtteri J PANULA 1,a, Kasperi J ALAKYLÄ 1,a, Mikko S VENÄLÄINEN 2, Jaason J HAAPAKOSKI 3, Antti P ESKELINEN 4, Mikko J MANNINEN 5, Jukka S KETTUNEN 6, Ari-Pekka PUHTO 7, Anna I VASARA 8, Laura L ELO 2, and Keijo T MÄKELÄ 1 1 Department

of Orthopaedics and Traumatology, Turku University Hospital, and University of Turku, Turku; 2 Turku Bioscience Centre, University of Turku and Åbo Akademi University, Turku; 3 National Institute for Health and Welfare, Helsinki; 4 Coxa Hospital for Joint Replacement, Tampere; 5 Orton Hospital, Helsinki; 6 Department of Orthopaedics and Traumatology, Kuopio University Hospital, Kuopio; 7 Division of Operative Care, Department of Orthopaedic and Trauma Surgery, Oulu University Hospital, Oulu; 8 Helsinki University Hospital, Helsinki, Finland a Shared first authorship Correspondence: kasperi.j.alakyla@utu.fi Submitted 2020-04-08. Accepted 2021-06-01.

Background and purpose — Periprosthetic joint infection (PJI) is a devastating complication and more information on risk factors for PJI is required to find measures to prevent infections. Therefore, we assessed risk factors for PJI after primary total hip arthroplasty (THA) in a large patient cohort. Patients and methods — We analyzed 33,337 primary THAs performed between May 2014 and January 2018 based on the Finnish Arthroplasty Register (FAR). Cox proportional hazards regression was used to estimate hazard ratios with 95% confidence intervals (CI) for first PJI revision operation using 25 potential patient- and surgicalrelated risk factors as covariates. Results — 350 primary THAs were revised for the first time due to PJI during the study period. The hazard ratios for PJI revision in multivariable analysis were 2.0 (CI 1.3– 3.2) for ASA class II and 3.2 (2.0–5.1) for ASA class III–IV compared with ASA class I, 1.4 (1.1–1.7) for bleeding > 500 mL compared with < 500 mL, 0.4 (0.2–0.7) for ceramic-onceramic bearing couple compared with metal-on-polyethylene and for the first 3 postoperative weeks, 3.0 (1.6–5.6) for operation time of > 120 minutes compared with 45–59 minutes, and 2.6 (1.4–4.9) for simultaneous bilateral operation. In the univariable analysis, hazard ratios for PJI revision were 2.3 (1.7–3.3) for BMI of 31–35 and 5.0 (3.5–7.1) for BMI of > 35 compared with patients with BMI of 21–25. Interpretation — We found several modifiable risk factors associated with increased PJI revision risk after THA to which special attention should be paid preoperatively. In particular, high BMI may be an even more prominent risk factor for PJI than previously assessed.

Prosthetic joint infection (PJI) is currently the most common reason for revision surgery after THA (FAR 2016). A prior study based on Nordic data from 1995 to 2009 stated that 0.6% of all THAs were revised due to deep infection and that the risk was increasing towards the end of the study period (Dale et al. 2012). Cumulative incidence of PJI after primary THA from 1998 to 2009 in Finland was 0.92% (Huotari et al. 2015). PJI is a devastating complication; it can lead to reduced physical functioning, pain, poor quality of life (Cahill et al. 2008, Moore et al. 2015), and at worst even death of the patient. Thus, it is important to know PJI risk factors to be able to reduce PJIs. Risk factors for PJI can be divided to patient- and surgicalrelated factors. Previously known patient-related risk factors PJI include increased comorbidity, morbid obesity, male sex, and operative diagnosis, whereas long duration of operation is a surgery-related cause for infection (Pedersen et al. 2010, Kong et al. 2017, Smith et al. 2018). Risk factors for PJI after primary THA have not been previously assessed based on FAR data. FAR data contents were thoroughly updated in 2014 to include parameters such as BMI, ASA class, and duration of surgery (FAR 2016). We determined the risk factors for first PJI revision after primary THA.

Patients and methods FAR was established in 1980 and since then it has been compiling data on arthroplasty surgery in Finland (Paavolainen et al. 1991). It is mandatory for all Finnish private and public healthcare units to provide information of arthroplasty surgery

666

to the National Institute of Health and Welfare to maintain the FAR database (Puolakka et al. 2001). All Finnish citizens have a unique identification number that connects the person and the primary and possible revision THA. Reporting the patientand surgery-related data to FAR is performed using a standard online sheet that is completed during the operation. Dates of death are obtained from the Population Register Centre. Currently over 95% of all primary THAs and 81% of all revisions performed are reported to FAR (FAR 2016). Several new parameters were included to the FAR in May 2014. These were surgical approach, BMI, ASA class, intraoperative bleeding, duration of the operation, level of education of surgeon and assistant, mode of anesthesia, intraoperative complications, and previous operations on the same joint. The following 25 risk factors were considered as covariates based on previously reported associations with PJI and prior clinical knowledge: age group (≤ 55, 56–65, 66–75, ≥ 76 years), sex, simultaneous bilateral operation (yes, no), ASA class (I, II, III–IV), BMI (≤ 20, 21–25, 26–30, 31–35, > 35), diagnosis (primary osteoarthritis, fracture, inflammatory arthritis, other), hospital volume (low [< 240 THAs performed annually], medium [240–480], high [> 480]), level of education of the surgeon (specialist, resident), level of education of the first assistant (specialist, resident, other), surgical approach (posterior, anterolateral, anterior), bleeding (< 500mL, > 500mL), duration of the operation (< 45, 45–59, 60–89, 90–120, > 120 minutes), anesthesia mode (spinal, epidural, general, nerve block), local infiltrative anesthesia (LIA) (yes, no), perioperative complication during operation (no complication, calcar fracture, trochanteric fracture, femoral shaft fracture, acetabular fracture), previous operation on the same joint such as osteotomy or osteosynthesis (yes, no), antibiotic prophylaxis (cefuroxime, clindamycin, vancomycin, other, not used), antithrombotic prophylaxis (enoxaparin, rivaroxaban, tinzaparin, warfarin, other, not used), anticoagulant medications (tranexamic acid, no, other), mechanical antithrombotic prophylaxis (calf muscle pump, surgical stocking, not used), antimicrobial incise drape (yes, no), fixation method (cementless, cemented, hybrid, reverse hybrid), bearing couple used (ceramic-on-ceramic, ceramic-on-ultrahighly cross-linked polyethylene (UHXLPE), metal-on-UHXLPE, ceramized metal-on-UHXLPE, other) and femoral head size (28, 32, 36, > 36 mm). In addition, we tested potential association of operated side (right, left) with revision for PJI. We extracted data on 33,337 primary THAs and 350 revision operations due to PJI after the primary THA performed in Finland from May 2014 to January 2018 (Table 1, see Supplementary data). The survival endpoint was revision operation where at least 1 component was removed or exchanged due to PJI. Determining PJI as indication for revision operation was performed by the operating surgeon based on preoperative evaluation and clinical presentation. These evaluations should be based on recommended guidelines for diagnosing PJI (Parvizi et al. 2016). Unfortunately, FAR data contents do

Acta Orthopaedica 2021; 92 (6): 665–672

not include data on for example intraoperative bacterial cultures. Follow-up ranged between 0 and 3.7 years. The vast majority of PJI revisions (334 of 350) occurred during the first year after primary THA, but we decided to include the whole of the follow-up period and all of the cases. There were 2,839 patients with both hips operated (5,678 operations). In 456 patients both hips had been operated simultaneously. Bilateral THRs were treated as 2 independent observations, since bilaterality has been shown to have a negligible influence on the risk of revision for infection (Ranstam and Robertsson 2010). 2.4% of patients died during the study period. Although death can be considered as competing risk leading to potential overestimation of incidence of revision, we did not perform competing risk analysis, as our main focus was on the estimation of relative revision risks in which the Cox regression model has been reported to provide more accurate results (Ranstam and Robertsson 2017). Revisions performed for other reasons (fracture, dislocation etc.) were censored when they occurred. Statistics The unadjusted rate for revision due to PJI with 95% confidence intervals (CI) was first estimated with Kaplan–Meier analysis. Then univariable and multivariable Cox proportional hazards regression models were used for estimation of possible risk factors and hazard ratios with CIs for first infection revision operation (Tables 2 and 4, and Table 3, see Supplementary data). We performed a directed acyclic graph (DAG) analysis (Figure 1) based on the previous medical literature and the clinical practice to organize variables according to their supposed relation to PJI revision and to other variables. For all the variables in the univariable analysis with potential confounding bias, we performed multivariable analysis by choosing the adjusting variables based on the DAG. In the multivariable analysis the following 8 risk factors were adjusted with associated covariates identified in DAG: ASA class (adjusted for age), intraoperative bleeding (adjusted for BMI, previous contributing operations, complications during surgery and level of education [surgeon]), duration of operation (adjusted for previous contributing operations, level of education [surgeon], intraoperative bleeding, BMI and complications during surgery), anesthesia mode (adjusted for age and ASA class), bearing couple (adjusted for age and ASA class), fixation (adjusted for sex and age), simultaneous bilateral operation (adjusted for age and ASA class), and complications during the surgery (adjusted for BMI) (Tables 4 and 5). The proportional hazards assumption for Cox models were assessed from Kaplan–Meier curves graphically and by a statistical test based on scaled Schoenfeld residuals (Grambsch and Therneau 1994, Ranstam et al. 2011). A p-value < 0.05 for PH test indicated non-proportional hazards. In the multivariable analysis duration of the surgery > 120 minutes did not fulfil the proportional hazard assumption. Furthermore, LIA, simultaneous bilateral operation, and complication during surgery did not fulfil the assumption of proportional hazards in

Acta Orthopaedica 2021; 92 (6): 665–672 667

Hospital volume

Diagnosis Sex

Animicrobial incise drape

Fixation

Age

Education of surgeon

Revision for infection Anesthesia

Duration

Previous operations

BMI

ASA

Bleeding

Bilaterality Bearing couple

Antibiotic prophylaxis

Intraoperative complications

No expected causal effects on revision for infection Side

Education of assistent

Femoral head size

Surgical approach

Antithrombotic prophylaxis

Bearing couple

Anticoagulant medications

Figure 1. A directed acyclic graph (DAG) was constructed under the following assumptions: 1. THA “revision for infection” is dependent on “patient age,” “sex,” ‘bilaterality,” “ASA class,” “BMI,” “diagnosis,” “hospital volume,” “education of surgery,” “bleeding,” “duration,” “intraoperative complications,” “previous operations,” “antimicrobial incise drape,” “anesthesia,” “antibiotic prophylaxis,” and type of THA “fixation.” Choice of “side,” “education of assistant,” “surgical approach,” “bearing couple,” antithrombotic prophylaxis,” “anticoagulant medications,” and “femoral head size” are not expected to affect “revision for infection” due to clinical suspicion. 2. “Fixation” is dependent on “age” and “sex” because older and female patients have probably received a cemented or hybrid THA due to their poorer bone quality. “Bearing couple” may be dependent on age because surgeons have probably chosen ceramic-on-ceramic bearing couple in younger patients. “Bearing couple” may also be dependent on ASA class for the same reason. ASA class is partly dependent on age by definition. “Bilaterality” is dependent on “age” and “ASA class” because both hips are seldom operated on in elderly or high ASA class patients. 3. “BMI” may be affected by “duration” and “intraoperative complications” due to more difficult operation with high BMI. “Duration” may be dependent on “education of surgery” due to experience factor. “Bleeding,” “duration,” and “previous operations” may be dependent on clinical basis. 4.“Anesthesia” is dependent on “ASA class” and “age” because general anesthesia is usually avoided in elderly patients.

the univariable analysis. Therefore, we divided the follow-up time of these variables into suitable time intervals based on Kaplan–Meier analyses and performed uni- and multivariable analyses for these variables separately (Table 5). All the statistical analyses were carried out using R statistical computing environment version 3.4.1 (R Foundation for Statistical Computing, Vienna, Austria. URL https://www.Rproject.org/). A p-value < 0.05 was set as level of significance. Ethics, funding, and potential conflict of interest Ethical approval: Dnro THL/506/5.05.00/2016. Funding statement: This research received no specific grant from any funding agency in the public, commercial, or notfor-profit sectors. MSV reports funding from the Academy of Finland [grant number 322123]. The funders had no role in

study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors declare no conflicts of interest.

Results Most of the patients belonged to the age group from 66 to 75 years (37%), were women (57%), had an ASA class II (49%), and had a BMI of 26–30 (41%). The majority of patients received THA due to primary osteoarthritis (87%) and most of the operations lasted 60–89 minutes (49%). Most of the patients were operated on using spinal anesthesia (92%) and received THA with cementless fixation (62%), metal-on-ultrahighly crosslinked polyethylene (UHXLPE) bearing couple (50%), and 36 mm femoral head size (74%) (Table 1, see Supplementary data). The overall Kaplan–Meier probability of no PJI revision at the end of the study period with 0–3.7-year follow-up time was 98.8% (CI 98.7–98.9). Patients with advanced ASA class were associated with increased risk of revision for PJI in both univariable analysis (ASA class II vs. ASA class I HR 1.7 [CI 1.1–2.7] and ASA class III–IV vs. ASA class I HR 2.5 [CI 1.6–3.9]) (Table 2 and Table 3, see Supplementary data) and in multivariable analysis (ASA class II vs. ASA class I HR 2.0 [CI 1.3–3.2] and ASA class III–IV vs. ASA class I HR 3.2 [CI 2.0–5.1]) (Table 4). Intraoperative bleeding over 500 ml was associated with increased risk of revision for infection when compared to bleeding less than 500 ml in both univariable analysis (HR 1.5 [CI 1.2–1.9]) (Table 2 and Table 3, see Supplementary data) and in multivariable analysis (HR 1.4 [CI 1.1–1.7]) (Table 4). We found a decreased risk of revision for infection in univariable analysis for the patients with ceramic-on-ceramic bearing couple and also for the other group of bearing couples (ceramic-on-ceramic vs. metal-on-UHXLPE HR 0.4 [CI 0.2–0.7] and other vs. metal-on-UHXLPE HR 0.1 [CI 0.0–0.6]) (Table 2 and Table 3, see Supplementary data). The same association with ceramic-on-ceramic and other bearing couples was also found in multivariable analysis (ceramic-onceramic vs. metal-on-UHXLPE HR 0.4 [CI 0.2–0.7] and other vs. metal-on-UHXLPE HR 0.1 [CI 0.0–0.6]) (Table 4). The use of epidural and general anesthesia was associated with increased risk of revision for infection in both univariable analysis (HR 2.2 [CI 1.4–3.5] and HR 1.7 [CI 1.2–2.3], respectively) (Table 2 and Table 3, see Supplementary data) and in multivariable analysis (HR 2.1 [CI 1.3–3.4] and HR 1.6 [CI 1.2–2.3], respectively) (Table 3). The use of spinal anesthesia was associated with decreased risk of revision for infection in both univariable (HR 0.6 [CI 0.4–0.8]) (Table 2 and Table 3, see Supplementary data) and in multivariable analysis (HR 0.6 [CI 0.4–0.8]) (Table 4). Solely in the univariable analysis did we find an increased risk of revision for infection for the following parameters:

668

Acta Orthopaedica 2021; 92 (6): 665–672

Table 2. Univariable analysis of possible risk factors for revision for PJI Variable

Hazard ratio (95% CI)

Sex (reference male) Female 0.6 (0.5–0.7) ASA class (reference ASA I) ASA II 1.7 (1.1–2.7) ASA III–IV 2.5 (1.6–3.9) BMI (reference BMI 21–25) ≤ 20 0.7 (0.2–2.1) 26–30 1.3 (0.9–1.8) 31–35 2.3 (1.7–3.3) > 35 5.0 (3.5–7.1) Preoperative diagnosis (reference primary osteoarthritis) Fracture 1.0 (0.6–1.7) Inflammatory arthritis 1.3 (0.7–2.7) Other 1.6 (1.1–2.2) Intraoperative bleeding (reference < 500 mL) > 500 mL 1.5 (1.2–1.9) Anesthesia (spinal)(reference no) Yes 0.6 (0.4–0.8) Anesthesia (epidural)(reference no) Yes 2.2 (1.4–3.5) Anesthesia (general)(reference no) Yes 1.7 (1.2–2.3) Antithrombotic prophylaxis (reference enoxaparin) Warfarin 2.7 (0.9–8.4) Rivaroxaban 0.8 (0.6–1.0) Tinzaparin 0.6 (0.3–1.2) Not used 2.8 (1.5–5.3) Other 0.6 (0.3–1.2) Bearing couple (reference metal-on-UHXLPE) Ceramic-on-ceramic 0.4 (0.2–0.7) Ceramic-on-UHXLPE 0.9 (0.6–1.1) Ceramized metal-on-UHXLPE 0.9 (0.5–1.5) Other 0.1 (0.0–0.6) Femoral head size (reference 32 mm) 28 mm 2.8 (1.2–6.5) 36 mm 1.9 (1.4–2.6) > 36 mm 2.1 (0.7–5.7) UHXLPE = ultra-highly crosslinked polyethylene.

preoperative diagnosis (other) HR 1.6 (CI 1.1–2.2), high BMI (BMI 31–35 vs. BMI 21–25 HR 2.3 [CI 1.7–3.3] and BMI > 35 vs. BMI 21–25 HR 5.0 [CI 3.5–7.1]), high volume hospitals vs. low volume hospitals HR 1.3 (CI 1.0–1.7), previous contributing operations HR 1.8 (CI 1.0–3.2), antithrombotic prophylaxis not used HR 2.8 (CI 1.5–5.3), femoral head size 36 mm vs. 32 mm HR 1.9 (CI 1.4–2.6), and 28 mm vs. 32 mm heads HR 2.8 (CI 1.2–6.5). Females compared with males HR 0.6 (CI 0.5–0.7) were associated with decreased risk of revision for infection in the univariable analysis (Table 2 and Table 3, see Supplementary data). Simultaneous bilateral operation was associated with increased risk of PJI for the first 3 postoperative weeks in both univariable analysis HR 2.2 (CI 1.2–4.2) and in multivariable analysis HR 2.6 (CI 1.4–4.9) (Table 5). Further, duration of the operation over 120 minutes was associated with an increased risk of revision for infection for the first 3 postoperative weeks in both univariable analysis HR 3.3 (CI 1.8–6.0) and in multivariable analysis HR 3.0 (CI 1.6–5.6) (Table 5).

Table 4. Multivariable analysis for revision for PJI Variable

Hazard ratio (95% CI)

ASA class (reference ASA I) ASA II 2.0 (1.3–3.2) ASA III–IV 3.2 (2.0–5.1) Intraoperative bleeding (reference < 500 mL) > 500 mL 1.4 (1.1–1.7) Anesthesia (spinal)(reference no) Yes 0.6 (0.4–0.8) Anesthesia (epidural)(reference no) Yes 2.1 (1.3–3.4) Anesthesia (general)(reference no) Yes 1.6 (1.2–2.3) Bearing couple (reference metal-on-UHXLPE) Ceramic-on-ceramic 0.4 (0.2–0.7) Ceramic-on-UHXLPE 0.9 (0.7–1.2) Ceramized metal-on-UHXLPE 0.9 (0.5–1.6) Other 0.1 (0.0–0.6) Fixation (reference cementless) Cemented 1.1 (0.7–1.7) Hybrid 1.3 (0.9–1.7) Reverse hybrid 0.9 (0.5–1.5) UHXLPE = ultra-highly crosslinked polyethylene. ASA class was adjusted for age. Intraoperative bleeding was adjusted for BMI, previous contributing operations, complications during surgery, and level of education (surgeon). Spinal, epidural, and general anesthesia and bearing couples were adjusted for age and ASA class. Fixation was adjusted for sex and age.

Discussion We found that high BMI, advanced ASA class, bleeding over 500 mL and the use of epidural and general anesthesia increased the risk of revision for PJI, whereas ceramicon-ceramic bearing couple and spinal anesthesia decreased revision risk. Simultaneous bilateral operation and duration of operation over 120 minutes increased the risk of revision for PJI during the first 3 postoperative weeks. The cumulative rate of revision due to PJIs was 1.04%, which is slightly higher than published previously (Pedersen et al. 2010, Dale et al. 2011, 2012, Huotari et al. 2015, Kong et al. 2017, Lenguerrand et al. 2018). However, it is challenging to compare reported incidences of PJIs because of differences in definitions, time frame, surveillance systems, and completeness of reporting to registers (Wilson et al. 2007). We found that high ASA class was associated with increased risk of revision due to PJI. ASA class is a crude estimate of a patient’s medical condition, and has been associated with PJI risk in numerous previous reports (Dale et al. 2011, Namba et al. 2012, Kong et al. 2017, Lenguerrand et al. 2018, Smith et al. 2018). Another factor associated with increased risk of revision for infection in our multivariable analysis was bleeding over 500 mL. We are not aware of previous studies concerning intraoperative bleeding and PJI association, but blood transfusion

Acta Orthopaedica 2021; 92 (6): 665–672 669

Table 5. Uni- and multivariable analyses divided to suitable time intervals for the duration, simultaneous bilateral operation, anesthesia (LIA), and complications during surgery (fracture) due to not fulfilling the assumption of proportional hazards

Univariable Multivariable hazard ratio hazard ratio (95% CI) (95% CI)

Duration (minutes)(reference 45–59) Time interval 0–3 weeks < 45 1.0 (0.5–2.3) 1.1 (0.5–2.5) 60–89 1.4 (0.8–2.3) 1.3 (0.8–2.2) 90–120 1.4 (0.8–2.5) 1.3 (0.7–2.3) > 120 3.3 (1.8–6.0) 3.0 (1.6–5.6) Time interval > 3 weeks < 45 1.2 (0.4–3.7) 1.1 (0.3–3.4) 60–89 1.1 (0.5–2.2) 1.0 (0.5–2.2) 90–120 1.4 (0.6–3.1) 1.4 (0.6–3.1) > 120 0.6 (0.2–1.5) 0.5 (0.2–1.4) Simultaneous bilateral operation Time interval 0–3 weeks 2.2 (1.2–4.2) 2.6 (1.4–4.9) > 3 weeks 0.3 (0.07–1.0) 0.3 (0.07–1.0) Anesthesia (LIA) Time interval 0–3 weeks 0.7 (0.5–1.1) 0.7 (0.5–1.1) > 3 weeks 1.5 (0.9–2.6) 1.5 (0.8–2.5) Complications during surgery (fracture) Time interval 0–5 weeks 0.3 (0.04–2.2) 0.4 (0.05–2.6) > 5 weeks 8.8 (0.9–86.2) 8.6 (0.9–84) In the multivariable analysis simultaneous bilateral operations and local infiltrative anesthesia were adjusted for age and ASA classification. Complication during surgery (fracture) was adjusted for BMI. Duration was adjusted for previous contributing operations, level of education (surgeon), intraoperative bleeding, BMI, and complications during surgery.

and PJI have been associated previously (Kim et al. 2017). As intraoperative bleeding is a common indication for blood transfusion, we consider our finding to support the pre-existing evidence. In a comprehensive literature review it was stated that some association between intraoperative bleeding and PJI was found, but more quality studies are needed (Kwong et al. 2012). Male sex was a risk factor for revision due to PJI in our study, which is in accordance with previous studies (Pedersen et al. 2010, Dale et al. 2012, Lenguerrand et al. 2018, Smith et al. 2018). Our data also supports the magnitude of risk presented previously (1.2–1.7-fold). Only 1 study has stated that female sex was associated with higher risk of revision for PJI (Namba et al. 2012). Reason for the increased PJI risk for males is not clear but may lie in confounding factors that are not included in the FAR such as smoking and alcohol abuse, both more common among males (WHO 2015, 2018). Previously it has been stated that skin metabolism, hair growth, sebum production, skin pH, and skin thickness differ between males and females. These differences may predispose male patients to PJI compared with female patients. (Badawy et al. 2017). Detailed preoperative patient counseling should take

into account the increased PJI risk of male sex to manage modifiable surgery-related risks. Younger age was not a PJI revision risk factor in our study, which gives support to some previous findings (Pedersen et al. 2010, Smith et al. 2018). Comorbidities are more common with older age and older age can thus affect the risk of developing PJI. However, Lenguerrand et al. (2018) stated recently based on the largest register study so far (623,253 THAs, 2,705 PJI revisions), that the PJI risk decreases with increasing age. The authors considered that their finding could be due to increased follow-up time compared with previous studies. The correlation of obesity and risk of PJI is well documented in several prior studies and meta-analyses (Namba et al. 2012, Kunutsor et al. 2016, Kong et al. 2017, Kurtz et al. 2018, Lenguerrand et al. 2018, Smith et al. 2018, Triantafyllopoulos et al. 2018). Also, in our study BMI was associated with an increased risk of revision due to PJI. Patients with BMI of 30–35 and > 35 had a HR of 2.4 and 5.1, respectively. The PJI risk of those with BMI > 35 was even higher than that reported previously (OR 1.9 for BMI 35–40, OR 4 for BMI >40) (Smith et al. 2018). High BMI may be an even more prominent risk factor than assessed previously, special attention to which should be paid preoperatively. However, the effect of weight loss prior to THA on risk for PJI is not clear and more quality studies need to be done to clarify the subject (Lui et al. 2015, Li et al. 2019). We found that long duration of operation was associated with an increased risk of revision for PJI for the first 3 postoperative weeks. This finding supports previous evidence (Engesaeter et al. 2006, Pedersen et al. 2010). Similar to our findings, Pedersen et al. (2010) have stated that duration of 2 hours or more increased PJI rate. On the other hand, Namba et al. (2012) found that duration of operation was not an independent risk factor for PJI. Specializing in THA increases the numbers performed, which probably decreases operation time. Unfortunately, our data did not include surgeon volume data. High hospital volume in our study was associated with increased PJI rate. Bilateral operation was associated with increased risk of revision for PJI for the first 3 postoperative weeks; previous studies (Namba et al. 2012, Kong et al. 2017) have found the same association, but with no regard to time from operation. The risk of PJI should be considered in elective management of patients who require both hips to be operated on in the same operation. We found that spinal anesthesia was associated with lower risk of PJI, whereas epidural anesthesia and general anesthesia were associated with increased risk of revision due to PJI in comparison with other anesthesia options. It has been stated earlier that neuraxial anesthesia is associated with decreased PJI rate compared with general anesthesia (Helwani et al. 2015, Johnson et al. 2016, Lenguerrand et al. 2018, Memtsoudis et al. 2019, Scholten et al. 2019). We are unaware of such data showing that epidural anesthesia would be associated with increased risk of revision due to PJI. Epidural anes-

670

thesia is often used in patients with anticipated longer operation time and hence might be associated with increased risk of complications. Previous studies concerning bearing couples have found that ceramic-on-ceramic may be a protective factor for developing PJI (Lee et al. 2016, Pitto and Sedel 2016, Kurtz et al. 2017, Lenguerrand et al. 2018). Kurtz et al. (2017) stated that THA patients with ceramic-on-polyethylene and ceramic-on-ceramic bearings had reduced risk of infection relative to metal-on-polyethylene bearings (HR 0.9, HR 0.7 respectively). Lenguerrand et al. (2018) found that the risk of revision for PJI was influenced by the type of bearing couples and varied according to the time period. In the early postoperative period, no differences were observed. Ceramic-on-ceramic and ceramic-on-polyethylene surfaces were associated with a lower risk of long-term revision (from 12 months for ceramic-on-ceramic and 24 months for ceramic-on-polyethylene postoperation onwards) than metalon-polyethylene bearings (Lenguerrand et al. 2018). Contrary to previous studies, we found that ceramic-on-ceramic was associated with a lower rate of revision for PJI in the early time period, as our study did not include long-term infections. Further, ceramic-on-UHXLPE did not protect against PJI in our study. It is likely that this finding is affected by residual confounding as ceramic-on-ceramic population differs from other surface groups regarding patient-related factors. A ceramic-on-ceramic bearing couple tends to be used in younger and healthier patients with less comorbidity. Also, the surgeons using ceramic-on-ceramic may be more experienced. This residual confounding likely affects the results even after adjusting. Fixation method was not associated with PJI in our study. Previous reports have been contradictory. Lenguerrand et al. (2018) stated that in the early postoperative period patients who had undergone a cementless, hybrid, or reverse hybrid THA were at higher risk than those with cemented implants but from 3 to 24 months they were at lower or similar risk. Pedersen et al. (2010) found a tendency for increased risk of revision for patients who had received cemented THA without antibiotic and hybrid THA relative to patients with cementless implants. Kunutsor et al. (2019) stated in the meta-analysis that, in the first six months, cementless fixations were associated with increased PJI risk when compared with cemented fixation. Overall cemented fixation was associated with an increased PJI risk compared with uncemented THA. Most PJIs occur during the first postoperative year, and it seems that bone cement may protect fragile patients with cemented THAs from early infections. Antibiotics in the bone cement are not released later on, so the protective effect finishes. The fixation method is a variable that might be affected by both known and unknown confounders. For example, elderly patients are prone to have a cemented THA. On the other hand, antibiotics in the bone cement may protect from PJI. Surgical approach did not have an effect on PJI risk in the current study whereas previ-

Acta Orthopaedica 2021; 92 (6): 665–672

ous studies have been inconclusive on the subject (Namba et al. 2012, Lenguerrand et al. 2018, Smith et al. 2018, Triantafyllopoulos et al. 2018). High-volume hospital was associated with increased risk of revision, though preceding evidence has been contradictory. In study from the United States, no association between higher volume hospitals (> 200 THAs annually) and PJI revisions (Namba et al. 2012) was found. However, in a study from the UK risk of early infections was increased in THAs undertaken in high-volume hospitals (> 255 THAs in the previous 12 months) (Lenguerrand et al. 2018). In our study previous contributing operation was associated with increased risk of PJI in univariable analysis and a similar association has been presented before (Cordero-Ampuero and De Dios 2010). Previous reports concerning preoperative diagnosis and PJI risk after primary THA have often found an association (Pedersen et al. 2010, Namba et al. 2012, Lenguerrand et al. 2018). In our study, “other” preoperative diagnosis vs. primary osteoarthritis was associated with higher risk of PJI. Conditions that cause, e.g., avascular necrosis, such as steroid use or irradiation, cause immunosuppression and also predispose towards PJI. We acknowledge that our study has several limitations. Although prospectively collected, our data is observational. Further, FAR does not incorporate comprehensive data on possibly relevant patient-related factors such as socioeconomic status, smoking status, or comorbidities, although ASA class is a crude estimate of medical condition. Even though FAR has included new variables since 2014 there still might be some confounding factors not included in the FAR influencing our results, such as the lower risk of infection in ceramic-on-ceramic articulations. Furthermore, completeness of revision surgery of FAR is 81% compared with discharge register so we are missing some PJI revisions (FAR 2016). Those revision operations performed on call (PJI, fractures, dislocations) are probably slightly underreported compared with elective revisions (wear, metallosis). However, we do not think that this causes serious bias to our results. The mortality rate was low and thus we considered death not to be a significant competing event with PJI revision. The PJI diagnosis reflects a clinical judgment sufficient to lead the surgeon to conduct a revision operation. Our data is recorded in operating theatres based on clinical diagnosis, and is not complemented afterwards based on, e.g., microbiology data, which may lead to underestimation of the incidence of PJIs. The strength of this study is a large, unselected population-based register setting with prospectively collected data. In summary we found that high BMI, advanced ASA class, bleeding over 500 mL and the use of epidural and general anesthesia increased the risk of revision for PJI, whereas ceramicon-ceramic bearing couple and spinal anesthesia decreased revision risk. Simultaneous bilateral operation and duration of operation over 120 minutes increased the risk of revision for PJI during the first 3 postoperative weeks.

Acta Orthopaedica 2021; 92 (6): 665–672 671

Supplementary data Tables 1 and 3 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674. 2021.1944529 All authors participated in designing the protocol. MV and LE performed statistical analyses. All the authors participated in interpreting the results, and writing and revision of the manuscript. All the authors read and approved the final manuscript. Acta thanks Ove Furnes and Anna Stefánsdóttir for help with peer review of this study.

Badawy M, Espehaug B, Fenstad A M, Indrekvam K, Dale H, Havelin L I, Furnes O. Patient and surgical factors affecting procedure duration and revision risk due to deep infection in primary total knee arthroplasty. BMC Musculoskelet Disord 2017; 18(1): 544. Cahill J L, Shadbolt B, Scarvell J M, Smith P N. Quality of life after infection in total joint replacement. J Orthop Surg (Hong Kong) 2008; 16(1): 58-65. Cordero-Ampuero J, De Dios M. What are the risk factors for infection in hemiarthroplasties and total hip arthroplasties? Clin Orthop Relat Res 2010; 468(12): 3268-77. Dale H, Skråmm I, Løwer H L, Eriksen H M, Espehaug B, Furnes O, Skjeldestad F E, Havelin L I, Engesæter L B. Infection after primary hip arthroplasty: a comparison of 3 Norwegian health registers. Acta Orthop 2011; 82(6): 646-54. 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. Engesaeter L B, Espehaug B, Lie S A, Furnes O, Havelin L I. Does cement increase the risk of infection in primary total hip arthroplasty? Revision rates in 56,275 cemented and uncemented primary THAs followed for 0–16 years in the Norwegian Arthroplasty Register. Acta Orthop 2006; 77(3): 351-8. FAR. Finnish arthroplasty register. Finnish Inst Heal Welf [Internet]; 2016. Grambsch P M, Therneau T M. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994; 81(3): 515-26. Helwani M A, Avidan M S, Ben Abdallah A, Kaiser D J, Clohisy J C, Hall B L, Kaiser H A. Effects of regional versus general anesthesia on outcomes after total hip arthroplasty: a retrospective propensity-matched cohort study. J Bone Joint Surg Am 2015; 97(3): 186-93. Huotari K, Peltola M, Jämsen E. The incidence of late prosthetic joint infections. Acta Orthop 2015; 86(3): 321-5. Johnson R L, Kopp S L, Burkle C M, Duncan C M, Jacob A K, Erwin P J, Murad M H, Mantilla C B. Neuraxial vs general anaesthesia for total hip and total knee arthroplasty: a systematic review of comparativeeffectiveness research. Br J Anaesth 2016; 116(2): 163-76 Kim J L, Park J H, Han S B, Cho I Y, Jang K M. Allogeneic blood transfusion is a significant risk factor for surgical-site infection following total hip and knee arthroplasty: a meta-analysis. J Arthroplasty 2017; 32(1): 320-5. Kong L, Cao J, Zhang Y, Ding W, Shen Y. Risk factors for periprosthetic joint infection following primary total hip or knee arthroplasty: a metaanalysis. Int Wound J 2017; 14(3): 529-36. Kunutsor S K, Whitehouse M R, Blom A W, Beswick A D. Patient-related risk factors for periprosthetic joint infection after total joint arthroplasty: a systematic review and meta-analysis. Virdi AS, editor. PLoS One 2016; 11(3): e0150866. Kunutsor S K, Beswick A D, Whitehouse M R, Blom A W, Lenguerrand E. Implant fixation and risk of prosthetic joint infection following primary total hip replacement: meta-analysis of observational cohort and randomised intervention studies. J Clin Med 2019; 8(5): 722.

Kurtz S M, Lau E, Baykal D, Springer B D. Outcomes of ceramic bearings after primary total hip arthroplasty in the Medicare population. J Arthroplasty 2017; 32(3): 743-9. 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. Kwong L M, Kistler K D, Mills R, Wildgoose P, Klaskala W. Thromboprophylaxis, bleeding and post-operative prosthetic joint infection in total hip and knee arthroplasty: a comprehensive literature review. Expert Opin Pharmacother 2012; 13(3): 333-44. Lee Y K, Yoon B H, Choi Y S, Jo W L, Ha Y C, Koo K H. Metal on metal or ceramic on ceramic for cementless total hip arthroplasty: a meta-analysis. J. Arthroplasty 2016; 31(11): 2637-2645.e1. Lenguerrand E, Whitehouse M R, Beswick A D, Kunutsor S K, Burston B, Porter M, Blom A W. Risk factors associated with revision for prosthetic joint infection after hip replacement: a prospective observational cohort study. Lancet Infect Dis 2018; 18(9): 1004-14. Li S, Luo X, Sun H, Wang K, Zhang K, Sun X. Does prior bariatric surgery improve outcomes following total joint arthroplasty in the morbidly obese? A meta-analysis. J Arthroplasty 2019; 34(3): 577-85. Lui M, Jones C A, Westby M D. Effect of non-surgical, non-pharmacological weight loss interventions in patients who are obese prior to hip and knee arthroplasty surgery: a rapid review. Syst Rev 2015; 4: 121. Memtsoudis S G, Cozowicz C, Bekeris J, Bekere D, Liu J, Soffin E M, Mariano E R, Johnson R L, Hargett M J, Lee B H, Wendel P, Brouillette M, Go G, Kim S J, Baaklini L, Wetmore D, Hong G, Goto R, Jivanelli B, Argyra E, Barrington M J, Borgeat A, De Andres J, Elkassabany N M, Gautier P E, Gerner P, Gonzalez Della Valle A, Goytizolo E, Kessler P, Kopp S L, Lavand’Homme P, MacLean C H, Mantilla C B, MacIsaac D, McLawhorn A, Neal J M, Parks M, Parvizi J, Pichler L, Poeran J, Poultsides L A, Sites B D, Stundner O, Sun E C, Viscusi E R, Votta-Velis E G, Wu C L, Ya Deau J T, Sharrock N E. Anaesthetic care of patients undergoing primary hip and knee arthroplasty: consensus recommendations from the International Consensus on AnaesthesiaRelated Outcomes after Surgery group (ICAROS) based on a systematic review and meta-analysis. Br J Anaesth 2019; 123(3): 269-87. Moore A J, Blom A W, Whitehouse M R, Gooberman-Hill R. Deep prosthetic joint infection: a qualitative study of the impact on patients and their experiences of revision surgery. BMJ Open 2015; 5(12): e009495. Namba R S, Inacio M C S, Paxton E W. Risk factors associated with surgical site infection in 30 491 primary total hip replacements. J Bone Joint Surg 2012; 94-B(10): 1330-8. Paavolainen P, Hämäläinen M, Mustonen H, Slätis P. Registration of arthroplasties in Finland: a nationwide prospective project. Acta Orthop Scand 1991; (Suppl. 241): 27-30. 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. 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 2010; 81(5): 542-7. Pitto R P, Sedel L. Periprosthetic joint infection in hip arthroplasty: is there an association between infection and bearing surface type? Clin Orthop Relat Res 2016; 474(10): 2213-18. Puolakka T J, Pajamäki K J, Halonen P J, Pulkkinen P O, Paavolainen P, Nevalainen J K. The Finnish Arthroplasty Register: report of the hip register. Acta Orthop Scand 2001; 72(5): 433-41. Ranstam J, Robertsson O. Statistical analysis of arthroplasty register data. Acta Orthop 2010; 81(1): 10-14. 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. Acta Orthop 2011; 82(3): 258-67.

672

Scholten R, Leijtens B, Hannink G, Kamphuis E T, Somford M P, van Susante J L C. General anesthesia might be associated with early periprosthetic joint infection: an observational study of 3,909 arthroplasties. Acta Orthop 2019; 90(6): 554-8. Smith J O, Frampton C M A, Hooper G J, Young S W. The impact of patient and surgical factors on the rate of postoperative infection after total hip arthroplasty: a New Zealand Joint Registry Study. J Arthroplasty 2018; 33(6): 1884-90. Triantafyllopoulos G K, Soranoglou V G, Memtsoudis S G, Sculco T P, Poultsides L A. Rate and risk factors for periprosthetic joint infection

Acta Orthopaedica 2021; 92 (6): 665–672

among 36,494 primary total hip arthroplasties. J Arthroplasty 2018; 33(4): 1166-70. WHO. WHO global report on trends in prevalence of tobacco smoking 2015. Geneva: WHO; 2015. WHO. Global status report on alcohol and health 2018. Geneva: WHO; 2018. Wilson J, Ramboer I, Suetens C. Hospitals in Europe Link for Infection Control through Surveillance (HELICS): inter-country comparison of rates of surgical site infection—opportunities and limitations. J Hosp Infect 2007; 65(Suppl. 2): 165-70.

Acta Orthopaedica 2021; 92 (6): 673–677 673

No increased mortality after total hip arthroplasty in patients with a history of pediatric hip disease: a matched, population-based cohort study on 4,043 patients Miriam G WADSTRÖM, Nils P HAILER, and Yasmin D HAILER

Section of Orthopedics, Department of Surgical Sciences, Uppsala University, Sweden Correspondence: Miriam.wadstrom@surgsci.uu.se Submitted 2021-04-08. Accepted 2021-07-08.

Background and purpose — Patients with pediatric hip diseases are more comorbid than the general population and at risk of premature, secondary osteoarthritis, often leading to total hip arthroplasty (THA). We investigated whether THA confers an increased mortality in this cohort. Patients and methods — We identified 4,043 patients with a history of Legg–Calvé–Perthes disease (LCPD), slipped capital femoral epiphysis (SCFE), or developmental dysplasia of the hip (DDH) in the Swedish Hip Arthroplasty Register (SHAR) between 1992 and 2012. For each patient, we matched 5 controls from the general population for age, sex, and place of residence, and acquired information on all participants’ socioeconomic background and comorbidities. Mortality after THA was estimated according to Kaplan– Meier, and Cox proportional hazard models were fitted to estimate adjusted hazard ratios (HRs) for the risk of death. Results — Compared with unexposed individuals, patients exposed to a THA due to pediatric hip disease had lower incomes, lower educational levels, and a higher degree of comorbidity but a statistically non-significant attenuation of 90-day mortality (HR 0.9; 95% CI 0.4–2.0) and a lower risk of overall mortality (HR 0.8; CI 0.7–0.9). Interpretation — Patients exposed to THA due to a history of pediatric hip disease have a slightly lower mortality than unexposed individuals. THA seems not to confer increased mortality risks, even in these specific patients with numerous risk factors.

Altered morphology of the hip joint due to pediatric hip diseases, e.g., Legg–Calvé–Perthes disease (LCPD), slipped capital femoral epiphysis (SCFE), or developmental dysplasia of the hip (DDH) is closely linked to early-onset, secondary osteoarthritis (OA) (Jacobsen and Sonne-Holm 2005, Pun 2016) which may lead to total hip arthroplasty (THA) at a young age (Froberg et al. 2011). Thus, the mean age at THA surgery in patients with a history of pediatric hip disease ranges from 38 to 55 years (Traina et al. 2011, Engesæter et al. 2012), whereas it ranges from 65 to 70 years in patients with primary OA (Engesæter et al. 2012, Fang et al. 2015, Cnudde et al. 2018). Studies from Nordic countries report that between 4% and 9% of all primary THAs are due to pediatric hip disease (Engesæter et al. 2012). The long-term outcome and revision rates after THA in patients with previous pediatric hip disease have been studied (Thillemann et al. 2008, Traina et al. 2011), but 90-day mortality and overall mortality after THA in these patients have not yet been investigated. Comorbidities, such as attention deficit hyperactivity disorder (ADHD), depression, cardiovascular disease, hypothyroidism, obesity, and coagulation abnormalities are more common in patients with LCPD and SCFE (Hailer and Nilsson 2014, Perry et al. 2017, Hailer and Hailer 2018, Hailer 2020). In addition, patients with LCPD and SCFE have a higher overall mortality than the general population (Hailer and Nilsson 2014, Hailer 2020). Due to an increased comorbidity burden and possibly increased overall mortality one could therefore fear an increased mortality after THA surgery in these patients. We therefore investigated whether THA surgery in patients with a pediatric hip disease confers an increased 90-day and overall mortality when compared with the general population.

674

Patients and method Study design and setting This population-based longitudinal cohort study includes patients (exposed group) with a history of pediatric hip disease (LCPD, SCFE, DDH) who had unilateral or bilateral THA surgery between 1992 and 2012 and were registered in the SHAR (see below). For each patient, 5 matched controls from the general population were randomly selected based on age, sex, and place of residence (unexposed group). The matching variables seemed appropriate because these 3 pediatric orthopedic hip diseases have different “target populations.” Concerning age, all 3 pediatric orthopedic hip diseases occur at different ages and times with disease differences that might also affect the outcome variable “death.” Concerning sex, LCPD affects mainly male patients and has an underlying geographical variation with regards to the incidence, whereas DDH is more prevalent in female patients, also with a geographical variation. The sex-ratio of SCFE is quite equal. Swedish citizens have a personal identity number (PNR), which enables the collection of data in national quality registers, including the Swedish Population Register, with information on the individual’s birth, death, place of residence, and civil status. The Swedish National Patient Register (NPR) collects information about in- and outpatient care using the ICD code system and the Nordic Medico-Statistical Committee (NOMESCO) codes for the classification of treatments and surgical procedures. The NPR was established in 1964 with data limited to inpatient visits. In 1987, it became mandatory to report to the NPR nationwide. Since 2001, the register also covers outpatient care. The quality of the NPR is continuously controlled and improved by the Swedish National Board of Health and Welfare. Studies on the validity of the registry showed a high validity for the inpatient care (Ludvigsson et al. 2011). The Swedish Mortality Database (SMD) obtains information concerning the causes of death for all individuals in Sweden. The Swedish Hip Arthroplasty Register (SHAR) (Söderman 2000) is a national quality register that has been used since 1979. This register includes data on the patient’s age, sex, and diagnosis together with information regarding the surgical technique and type of implant. Patients were identified in the SHAR when exposed to primary THA as a result of either LCPD, SCFE, or DDH. Variables of interest were collected from the SHAR, NPR, Swedish Population Register, and SMD. Age was divided into 4 groups (<50, 50–59, 60–74, >74 years). The comorbidity level was expressed by the Charlson Comorbidity Index (CCI) classified into 3 categories (CCI = 0, CCI = 1–2, CCI > 2). Income levels were divided into quartiles, and the level of education was categorized into the 4 levels “less than primary school”, “primary school”, “high school”, or “university level.”

Acta Orthopaedica 2021; 92 (6): 673–677

Statistics The study includes patients from the beginning of 1992. Follow-up ended on December 31, 2012, emigration, or death, whichever occurred 1st. The endpoint mortality was noted after 1st surgery (exposed group). Continuous variables were described using means with (SD). Counts of categorical data were examined using the chi-square test. We investigated the 90-day and overall mortality of exposed patients compared with unexposed individuals. We used the Cox proportional hazards model rather than logistic regression models in order to account for the temporal progression of the event “death” within the 2 groups (McNutt et al. 2003). The assumption of proportionality was fulfilled, as investigated by visual control of cumulative incidence functions and by calculating Schoenfeld’s residuals. Cumulative unadjusted survival was estimated, and Cox proportional hazards models were fitted to estimate unadjusted and adjusted HRs with 95% confidence intervals (CI) for the risk of death within 90 days or the risk of death during the entire observation period as the outcomes. CCI, levels of income, and education were considered relevant confounders (Garland et al. 2015, Hailer et al. 2016, Weiss et al. 2019). Because the variable “place of residence” had more than 300 levels it was used as a strata variable in the analysis. Whether or not to adjust for matching variables (age groups, sex, and place of residence at time of surgery) in studies designed such as ours is debated. Both Bland and Altman (1994) and Sjölander and Greenland (2013) recommend adjusting for the matching variables in order to avoid bias in the presence of additional confounders, as in our study. Because we have a large population with access to the Swedish Population register, the number of matching variables is not a limitation in finding appropriate controls, which otherwise can be a disadvantage of matching according to Bland and Altman (1994). Considering this, we found it appropriate to include the matching variables in addition to the confounding variables when estimating the HR, but, as suggested during the review process of this study, we performed a supplementary analysis where HR was estimated after adjustment for confounders but not for matching variables. Here, we found that the obtained HR differed only marginally between models with and without adjustment for the matching variables. Statistical significance was set at p < 0.05. For analyses, R statistical software, version 3.5.3, package “survival” 2.43-3 and R studio version 1.2.1335 was used (R Foundation for Statistical Computing, Vienna, Austria). Ethics, funding, and potential conflicts of interest Ethical approval for the study was granted by the Regional Ethical Review Board in Gothenburg (reg nr 2013:360/13). This study was supported by a grant based on the ALF agreement (regional funds for participating in the education and training of doctors, implementing clinical research and devel-

Acta Orthopaedica 2021; 92 (6): 673–677 675

Table 1. Characteristics of the population. Values are count (%) Factor

Unexposed Exposed n = 19,388 n = 4,043

Dead 1,879 (10) 351 (9) Age < 50 7,519 (39) 1,570 (39) 50–59 5,835 (30) 1,220 (30) 60–74 4,758 (25) 989 (25) ≥ 75 1,276 (7) 264 (7) Female sex 12,488 (64) 2,600 (64) CCI 0 18,263 (94) 3662 (91) 1–2 974 (5) 345 (9) > 2 151 (1) 36 (1) Income 1st quarter 4,945 (26) 1,062 (26) 2nd quarter 4,893 (25) 1,047 (26) 3rd quarter 4,869 (25) 987 (24) 4th quarter 4,667 (24) 943 (23) Education None 284 (2) 75 (2) 9 years 5,302 (27) 1,111 (28) High school 8,387 (43) 1,824 (45) University 5,415 (28) 1,033 (26)

Survival probability

1.000

1.00

0.95 0.999 0.90

0.85

0.998

0.80 0.997 0.75

Unexposed Exposed 0.996 0.00

0.05

0.10

0.15

0.20

0.25

0.70

Unexposed Exposed 0

Years to event

Figure 1. 90-day survival analysis for THA patients (exposed, n = 4,043) and controls (unexposed, n = 19,388).

CCI = Charlson’s Comorbidity Index

oping health and medical care) and by a grant from the Swedish Research Council to NPH (VR 2018–02612). The authors report no conflicts of interest.

Results Characteristics of the study population (Table 1) The study population comprised 23,431 individuals: 4,043 patients with a history of THA surgery due to pediatric hip disease (exposed group), and 19,388 matched controls (unexposed group). Of the exposed group, 6% had a history of LCPD (38% females), 2% had SCFE (44% females), and 92% had DDH (67% females). 974 (24%) also had surgery to the contralateral hip during the observed period. The mean delay between the 1st and 2nd THA was 2.5 years. 72 patients had surgery to both hips within 90 days; 46 patients had 1-stage bilateral THA. No patient who had bilateral THA died within 90 days of surgery, and 47 bilaterally operated patients died during the observation period. 64% of the exposed individuals were female. 94% of the unexposed individuals and 91% of the exposed individuals were classified as CCI = 0. The exposed individuals had a higher degree of comorbidities according to the CCI. 6% of unexposed individuals had a CCI of > 1 and 9.5% of the exposed group had a CCI of > 1. In the exposed group the income levels were somewhat lower compared with the unexposed group. The mean age at the time of THA surgery for patients with unilateral THA was 54 years (SD 13). Patients who had bilat-

Years to event

Figure 2. Overall survival analysis for THA patients (exposed, n = 4,043) and controls (unexposed, n = 19,388).

Table 2. Overall mortality, adjusted for confounders (CCI, income, and education) and for matching variables (age group, sex, and place of residence) Item All cases LCPD SCFE DDH

Crude HR (95% CI)

Adj. HR (95% CI)

0.9 (0.8–1.0) 0.8 (0.6–1.3) 0.9 (0.4–1.7) 0.9 (0.8–1.0)

0.8 (0.7–0.9) 0.7 (0.4–1.3) 2.1 (0.8–5.7) 0.8 (0.7–0.9)

eral THA had a mean age at the time for the 1st THA surgery of 53 years (SD 11) and a mean age at the time for the 2nd THA surgery of 56 years (SD 12). The endpoint mortality was noted only after the 1st THA (exposed group). No patient who had bilateral THA died within 90 days after surgery. 90-day mortality after THA (Figure 1) Among the exposed individuals, 8 (0.20%) died within 90 days after THA surgery, as compared with 43 (0.24%) among unexposed individuals. The adjusted HR for death within 90 days after THA surgery was 0.9 (CI 0.4–2.0). Stratified analyses based on the underlying pediatric hip diagnoses (LCPD, SCFE, and DDH) revealed a risk reduction only in patients with DDH; however, this was not statistically significant. Risk of overall mortality after THA (Figure 2, Table 2) 351 (8.7%) THA patients died within the entire observation period, compared with 1,879 (9.7%) of the unexposed individuals. Of the deceased THA patients, 29 (8%) had a history of LCPD, 10 (3%) had SCFE, and 312 (89%) had DDH. Patients operated on with a THA as a result of pediatric hip disease had a lower cumulative mortality compared with unexposed controls, and Cox regression analyses indicated an attenuated adjusted HR for overall mortality among the exposed individuals (HR 0.8; CI 0.7–0.9). In the stratified analyses based

676

on the underlying pediatric hip diagnoses (LCPD, SCFE, and DDH), we observed a slightly attenuated overall mortality risk in patients with LCPD and DDH, but statistical significance was seen only in patients with DDH. Patients with a history of SCFE had an adjusted HR of 2.1 (CI 0.8–5.7). Causes of death The most common causes of death in exposed individuals were cardiovascular disease (43%) and cancer (27%), whereas 35% of the unexposed individuals died from cardiovascular disease and 34% from cancer.

Discussion In this nationwide cohort study, we found a slightly attenuated 90-day and overall mortality after THA in patients with previous LCPD, SCFE, or DDH, but the attenuation of 90-day mortality was not statistically significant. Despite a higher comorbidity and mortality risk in patients with LCPD and SCFE in general (Hailer and Nilsson 2014, Perry et al. 2017, Hailer and Hailer 2018, Hailer 2020) we found no notable alterations of the mortality risks after THA in such patients, even after stratifying the analyses on the underlying pediatric hip diseases. We found an 18% lower overall mortality risk in patients who had THA surgery after either LCPD, SCFE, or DDH in childhood when compared with an unexposed sample from the general population. 9% of patients with a history of pediatric hip disease and subsequent THA died during the observation period. This proportion is somewhat higher than in the study by Engesæter et al. (2012), who investigated a similar patient group and found that 7% of the patients with THA performed for pediatric hip disease died within the study period (1995–2009). In comparison, 18% of their patients with THA because of primary OA died during the study period 1995– 2009. The same pattern, with a slightly attenuated mortality risk compared with the general population, has been described in patients who have undergone THA surgery due to primary OA (Lie et al. 2000, Cnudde et al. 2018). A possible explanation for the lower mortality risk of patient with pediatric hip diseases compared with the general population could be that the majority of the exposed cohort were patients with DDH, in whom risk factors such as those associated with LCPD and SCFE have not been described. However, in our study, we did not see that patients with DDH were less comorbid than the patients with SCFE or LCPD. Further, the stratified analysis based on the underlying pediatric hip diagnosis did not reveal major differences in the mortality risks between these groups. Patients who received THA due to previous pediatric hip disease were relatively young (mean age 53 years) and healthy (94% had a CCI = 0), compared with patients who received THA because of primary OA, who had a median age of 70 years (16–100) and of whom 84% had a low CCI (Weiss et al. 2019). In general, young patients without

Acta Orthopaedica 2021; 92 (6): 673–677

comorbidities are expected to have a lower risk of mortality, both after surgical interventions and in general. Therefore, we decided to use age- and sex-matched controls from the general population rather than a control group who received THA surgery because of primary OA. Cardiovascular- and cancer-related deaths were the most prevalent in both groups. Nevertheless, in the group of patients with a pediatric hip disease leading to THA we found a higher proportion of cardiovascular deaths and a lower proportion of cancer-related deaths than in the group of unexposed individuals. The underlying mechanism remains unclear, but a plausible reason could be the increased incidence of cardiovascular diseases in patients with LCPD and SCFE described in previous studies (Hailer et al. 2010, Ucpunar et al. 2018). On the other hand, patients with primary OA of the hip or knee have an increased risk of cardiovascular deaths (e.g., chronic ischemic heart disease and heart failure) (Gordon et al. 2016, Turkiewicz et al. 2019). Cardiovascular disease is closely linked to lower activity levels, and patients with primary OA of the hip or knee often have a reduced level of activity due to pain (Hawker et al. 2014). This could also explain our observations, as pediatric hip disorder leads to early hip pain and subsequently physical inactivity early in life. Strength and limitations This is the first study investigating 90-day and overall mortality risk in patients with a history of pediatric hip diseases and THA surgery. A major strength is the large number of included participants and the nationwide observational design. But, like all registry-based studies, this study has many limitations, including uncertain quality of input data and various types of bias (e.g., selection, detection, and observational bias). The quality of the Swedish registers used in this study is generally good. The SHAR has been validated by Södermann (2000), showing 100% coverage and 95% completeness. Yet, the SHAR has not been validated for secondary OA as a consequence of pediatric hip disease. When registering this specific cause of OA the orthopedic surgeon most likely diagnosed the condition based on radiographs rather than medical journals, and it seems likely that the rate of false-positive pediatric hip diagnoses should be rather low, whereas false-negatives may be more common, i.e., patients classified as having primary OA instead of secondary OA due to a pediatric hip disease could be more common. The validity of the SMD also needs to be considered. According to the Swedish National Board of Health and Welfare the correctness of reported causes of death is highest in patients with malignant tumors, ischemic heart disease, and in people who died young. The cause of death in the elderly is not classified as accurately because of the difficulty in determining the exact cause of death in patients with many comorbidities. Furthermore, the physician’s external examination often determines the cause of death and an autopsy is seldom performed (Johansson et al. 2009).

Acta Orthopaedica 2021; 92 (6): 673–677 677

Most patients in our exposed cohort were female and had DDH; neither of these factors is known to be related to increased mortality. One factor that might have affected our results is that the patients selected for THA in the group with previous pediatric hip disease might be in good health because of regular contacts with the healthcare system since childhood. In addition, sicker patients are more commonly excluded from surgery, and we do not know to what extent such biases might have impacted our results. The option of selecting patients with primary OA as a control group instead of selecting controls from the general population can be reasonably argued. However, patients with primary OA are generally older than patients with secondary OA after a history of pediatric hip disease. We therefore decided to have a matched control group representing the general population to more clearly define the effect of THA and risk of mortality in patients with a history of pediatric hip disease. Conclusion THA in patients with a pediatric hip disease exposed seems not to confer increased early mortality, and the overall mortality of such patients also seems to be lower than that observed in the general population. Supplementary data Tables 3 and 4 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/1745 3674.2021.1963582 MGW: Data analysis and interpretation, drafting the manuscript. YDH: Study design and data analysis, revision of manuscript. NPH: Study design and revision of manuscript. The authors would like to thank Leslie Shaps for his contribution with language editing this manuscript and Eva Freyhult for reviewing the statistical analysis and data. Acta thanks Ross W Crawford and Kjeld Søballe for help with peer review of this study.

Bland J M, Altman D G. Matching. BMJ 1994; 309(6962): 1128. Cnudde P, Rolfson O, Timperley A J, Garland A, Kärrholm J, Garellick G, Nemes S. Do patients live longer after THA and is the relative survival diagnosis-specific? Clin Orthop Relat Res 2018; 476(6): 1166-75. Engesæter L B, Engesæter I Ø, Fenstad A M, Havelin L I, Kärrholm J, Garellick G, Pedersen A B, Overgaard S. Low revision rate after total hip arthroplasty in patients with pediatric hip diseases. Acta Orthop 2012; 83(5): 436-41. Fang M, Noiseux N, Linson E, Cram P. The effect of advancing age on total joint replacement outcomes. Geriatr Orthop Surg Rehabil 2015; 6(3): 173-9. Froberg L, Christensen F, Pedersen N W, Overgaard S. The need for total hip arthroplasty in Perthes disease: a long-term study. Clin Orthop Relat Res 2011; 469(4): 1134-40. Garland A, Rolfson O, Garellick G, Kärrholm J, Hailer N P. Early postoperative mortality after simultaneous or staged bilateral primary total hip arthroplasty: an observational register study from the Swedish Hip Arthroplasty Register. BMC Musculoskelet Disord 2015; 16: 77.

Gordon M, Rysinska A, Garland A, Rolfson O, Aspberg S, Eisler T, Garellick G, Stark A, Hailer N P, Sköldenberg O. Increased long-term cardiovascular risk after total hip arthroplasty. Medicine (Baltimore) 2016; 95(6): e2662. Hailer Y D. Fate of patients with slipped capital femoral epiphysis (SCFE) in later life: risk of obesity, hypothyroidism, and death in 2,564 patients with SCFE compared with 25,638 controls. Acta Orthop 2020; 91(4): 457-63. Hailer Y D, Hailer N P. Is Legg–Calvé–Perthes disease a local manifestation of a systemic condition? Clin Orthop Relat Res 2018; 476(5): 1055-64. Hailer Y D, Nilsson O. Legg–Calvé–Perthes disease and the risk of ADHD, depression, and mortality. Acta Orthop 2014; 85(5): 501-5. Hailer Y D, Montgomery S M, Ekbom A, Nilsson O S, Bahmanyar S. Legg-Calvé-Perthes disease and risks for cardiovascular diseases and blood diseases. Pediatrics 2010; 125(6): e1308-e1315. DOI: https://doi. org/10.1542/peds.2009-2935 Hailer N P, Garland A, Rogmark C, Garellick G, Kärrholm J. Early mortality and morbidity after total hip arthroplasty in patients with femoral neck fracture. Acta Orthop 2016; 87(6): 560-6. Hawker G A, Croxford R, Bierman A S, Harvey P J, Ravi B, Stanaitis I, Lipscombe L L. All-cause mortality and serious cardiovascular events in people with hip and knee osteoarthritis: a population based cohort study. PLoS ONE 2014; 9(3): e91286. Jacobsen S, Sonne-Holm S. Hip dysplasia: a significant risk factor for the development of hip osteoarthritis. A cross-sectional survey. Rheumatology (Oxford) 2005; 44(2): 211-18. Johansson L A, Björkenstam C, Westerling R. Unexplained differences between hospital and mortality data indicated mistakes in death certification: an investigation of 1,094 deaths in Sweden during 1995. J Clin Epidemiol 2009; 62(11): 1202-9. Lie S A, Engesaeter L B, Havelin L I, Gjessing H K, Vollset S E. Mortality after total hip replacement: 0–10-year follow-up of 39,543 patients in the Norwegian Arthroplasty Register. Acta Orthop Scand 2000; 71(1): 19-27. Ludvigsson J F, Andersson E, Ekbom A, Feychting M, Kim J-L, Reuterwall C, Heurgren M, Olausson P O. External review and validation of the Swedish national inpatient register. BMC Public Health 2011; 11:450. McNutt L-A, Wu C, Xue X, Hafner J P. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol 2003; 157(10): 940-3. Perry D C, Metcalfe D, Costa M L, Van Staa T. A nationwide cohort study of slipped capital femoral epiphysis. Arch Dis Child 2017; 102(12): 1132. Pun S. Hip dysplasia in the young adult caused by residual childhood and adolescent-onset dysplasia. Curr Rev Musculoskelet Med 2016; 9(4): 427-34. Sjölander A, Greenland S. Ignoring the matching variables in cohort studies: when is it valid and why? Stat Med 2013; 32(27): 4696-708. Söderman P. On the validity of the results from the Swedish National Total Hip Arthroplasty register. Acta Orthop Scand Suppl 2000; 71(296): 1-33. Thillemann T M, Pedersen A B, Johnsen S P, Søballe K, Danish Hip Arthroplasty Registry. Implant survival after primary total hip arthroplasty due to childhood hip disorders: results from the Danish Hip Arthroplasty Registry. Acta Orthop 2008; 79(6): 769-76. Traina F, De Fine M, Sudanese A, Calderoni P P, Tassinari E, Toni A. Long-term results of total hip replacement in patients with Legg–Calvé– Perthes disease. J Bone Joint Surg Am 2011; 93(7): e25. Turkiewicz A, Kiadaliri A A, Englund M. Cause-specific mortality in osteoarthritis of peripheral joints. Osteoarthritis Cartilage 2019; 27(6): 848-54. Ucpunar H, Camurcu I Y, Duman S, Ucpunar E, Sofu H, Bayhan A I. Obesity-related metabolic and endocrine disorders diagnosed during postoperative follow-up of slipped capital femoral epiphysis. Acta Orthop 2018; 89(3): 314-19. Weiss R J, Kärrholm J, Rolfson O, Hailer N P. Increased early mortality and morbidity after total hip arthroplasty in patients with socioeconomic disadvantage: a report from the Swedish Hip Arthroplasty Register. Acta Orthop 2019; 90(3): 264-9.

678

Acta Orthopaedica 2021; 92 (6): 678–680

Bacteriophage therapy cures a recurrent Enterococcus faecalis infected total hip arthroplasty? A case report Ann-Sophie NEUTS 1, Hanneke J BERKHOUT 2, Anita HARTOG 3, and Jon H M GOOSEN 4 1 Department of Orthopaedic Surgery, Sint Maartenskliniek, Nijmegen; 2 Department of Medical Microbiology and Infectious Diseases, Canisius-Wilhemina Hospital, Nijmegen; 3 NIZO, Ede; 4 Department of Orthopaedic Surgery, Sint Maartenskliniek, Nijmegen, The Netherlands Correspondence: a.neuts@maartenskliniek.nl Submitted 2021-01-19. Accepted 2021-07-10.

A 76-year-old male patient presented with osteoarthritis of the left hip. He had no relevant medical history and had been professionally active in agriculture. Cemented total hip arthroplasty (THA), performed in September 2015, was complicated by an extensive hematoma and wound leakage on the second postoperative day. Wound leakage persisted for another 8 days, and debridement was performed. All 6 tissue samples showed the presence of Enterococcus faecalis, and the patient was treated with intravenous teicoplanin 600 mg twice a day for 3 months. After this treatment, the infection recurred, and a 2-stage septic revision was scheduled. The prosthesis and all surrounding cement were removed in February 2016. All samples taken showed growth of E. faecalis. After susceptibility testing, teicoplanin was restarted and continued for 6 weeks. Reimplantation followed in May 2016 after an antibiotic-free period of more than 6 weeks. Tissue samples obtained at that time showed no bacterial growth. In December 2016, the patient complained of pain and discomfort. 2 separate joint aspirations showed no bacterial growth, but serial radiologi-

Figure 1. After explantation of total hip prosthesis and after final reimplantation of a new prosthesis.

cal images showed loosening of the femoral stem. The patient was planned for revision surgery in January 2017. In 5 out of 7 samples, E. faecalis was cultured. After susceptibility testing, teicoplanin was restarted for another 3 months at the same dose. In June 2017, an aspiration, performed due to persistent pain, turned out to be positive for E. faecalis again. Extraction of the hip prosthesis followed and teicoplanin was restarted for another 6 weeks. After an antibiotic-free period of 2 weeks, open biopsies were taken. They all showed no bacterial growth. 4 weeks later, the hip was reimplanted (Figure 1). At the time of reimplantation, E. faecalis was cultured in 4 out of 7 samples. Due to kidney failure during the last period of teicoplanin therapy, this antibiotic regime could not be restarted. Postoperatively, oral amoxicillin 1,000 mg was administered 4 times a day for 3 months. In the summer of 2018, a new sample was obtained because of pain. The known E. faecalis was again cultured. A Girdlestone procedure was not accepted by the patient, who opted for suppressive therapy with doxycycline 200 mg once a day. Due to gastrointestinal side effects, the doxycycline dose was reduced to 100 mg once a day in September 2018. Nausea, vomiting, and loss of appetite lasted until shortly after the antibiotics were stopped in December 2019. The option of bacteriophage therapy was discussed and started on the patient’s own initiative with help from the Eliava Institute of Bacteriophage, Microbiology and Virology in Tbilisi, Georgia. He collected the formerly obtained tissue samples and sent them to Georgia. Susceptibility was retested for different antibiotics and phages. In the received documentation from Georgia, the E. faecalis appeared to have 4+ susceptibility, which was the maximum obtainable score, to 2 different types of bacteriophage: Pyophage and IntestiPhage. None of the tested antibiotics (Table 1) reached the same “sensitivity score” as that determined for these 2 phages. The Enterococcus even appeared to be resistant to doxycycline, which the patient was using as a suppressive regimen. Our patient travelled to Georgia to retrieve his ampoules of Pyophages and IntestiPhages. They were delivered in 10-mL vials as an oral suspension. He started using them on June 13, 2018, and continued for 19 days. The Pyophages were taken in

Acta Orthopaedica 2021; 92 (6): 678–680 679

Table 1. List of tested bacteriophages and antibiotics Bacteriophage sensitivity Fersis Phage MonoStaph Encophage Pyophage IntestiPhage SES Phage Antimicrobial sensitivity Amibac (amikacin) Amoxicillin Ampicillin Ampiox (ampicillin) Aprid (ampicillin + sulbactam) Avelox (moxifloxacin) Biseptol (sulfamethoxazole + trimethoprim) Ciprofloxacin Claforan (cefotaxime) Clarithormycin Dalacin Doxycycline Erythromycin Floxan (levofloxacin) Fortum (ceftazidime) Gentamicin Meflocid (levofloxacin) Rifampicin Sumamed (azithromycin) Triaxon (ceftriaxon) Zinnat (cefuroxime)

R R R 4+ 4+ R R 2+ 1+ 3+ 2+ 3+ R 2+ R R R R R 2+ R R 2+ R R R R

the morning and Intestiphages in the evening. After a 2-week pause, the bacteriophage therapy was restarted for another 19 days. Unfortunately, the exact composition of these phages and their concentration, expressed as plaque-forming units (PFUs), remain unknown to us and cannot be deduced from the patient leaflet. As recommended in the available literature, our patient received daily antibiotics during the 2 short periods of phage therapy. During the first period of phage therapy, he used oral amoxicillin 1,000 mg 4 times a day. During the second period, he used oral doxycycline 200 mg once a day. In December 2019, all antibiotics were stopped. He had no hip complaints when we saw him in our outpatient clinic in February 2021. No new cultures have been obtained up to the time of writing (July 2021). A time schedule of the treatment is presented in Figure 2.

Discussion Enterococcus faecalis is a microorganism known for its ability to develop biofilms. These biofilms play a key role in antibiotic resistance and recurrent periprosthetic joints infections (PJIs), which makes Enterococcus-induced PJIs particularly notorious. Besides a comprehensive antibiotic cure, 2-stage revision with delayed reimplantation remains the gold standard treatment for PJIs, as described by Renz et al. (2019). Both 2-stage revisions in our patient were performed following the global guidelines, as was the surgical procedure, including the

2015 September: THA October: DAIR, E. faecalis October to January 2016: Teicoplanin 2 x 600 mg 2016 February: Extraction, E. faecalis February to April: Teicoplanin 2 x 600 mg during 6 weeks April: 6 weeks of drug holiday May: Reimplantation, all samples showed no bacterial growth December: 2 x cultures: no bacterial growth 2017 January: Revision, E. faecalis January to April: Teicoplanin 2 x 600 mg July: Culture: E. faecalis 2018 January: Extraction January to March: Teicoplanin 2 x 600 mg March: Open biopsies: all 6 showed no bacterial growth March: Reimplantation: E. faecalis March to April: Amoxycillin 6 x 1g IV April to October: Amoxycillin 4 x 1g oral June 13 to July 1: Phages July 16 to August 3: Phages July 25: Culture: E. faecalis August 10 to September 13: Doxycyclin 1 x 200 mg September 13 to December 16, 2019: Doxycyclin 1 x 100 mg 2019 September 13, 2018 to December 16: Doxycyclin 1 x 100 mg 2021 No complaints, no antibiotics

Figure 2. Time schedule of treatment.

removal of all cement residues and the duration of antibiotic treatment. The treatment of our patient involved a multidisciplinary approach, as recommended for PJI treatment, involving an infection specialist, a microbiologist, a pharmacist, and an orthopedic surgeon. By following an interdisciplinary standardized protocol, with successful extraction of the prosthesis and the surrounding cement, one should be able to eradicate 60–90% of PJIs (Akanda et al. 2018, Karczewski et al. 2019). In septic situations where a 1- or 2-stage revision is not an option, as in the case of multiresistant microorganisms, little bone stock, poor quality of the surrounding tissues, or more than 1 failed prior septic revision and/or severe comorbidity, less attractive treatment options must be considered, such as chronic suppressive antibiotics and/or definite extraction of the prosthesis or a Girdlestone procedure. When antibiotic and conventional surgical treatments fail, one option is bacteriophage therapy. Bacteriophages were discovered over a century ago, a decade before the first antibiotic. The United States and Western Europe focused on the further development of antimicrobial agents, whereas the former Soviet Union and Eastern Europe continued the use of phages. Due to the growing resistance to antibiotics, there is renewed interest in phage therapy. Nowadays, within the orthopedic department, they are mainly used in the treatment of multidrug-resistant bacterial infections (Lin et al. 2017, Akanda et al. 2018). Phage therapy is not currently approved in the Western Europe, at least not for human use. This is mainly because of the lack of literature, documentation (Lin et al. 2017), and a regulatory framework

680

(Fauconnier 2019). In bacteriophage therapy, viruses (phages) are used to treat bacterial infections by infiltrating these bacteria and inducing lysis. Phages are non-living biological entities containing RNA or DNA. This means they are dependent on their host, the bacteria, for survival and reproduction. In general, there are 2 types of phages: temperate phages, which integrate their material into the bacterial genome, and lytic phages, which take over the whole replication mechanism of the bacteria to produce more phages. These lytic phages have evident antibacterial activity, in which lysis of the bacteria occurs through the production of proteins, called endolysins (Sulakvelidze et al. 2001, Yilmaz et al. 2013, Lin et al. 2017, Akanda et al. 2018). After bacterial lysis, the phages are released and attack additional bacteria, thereby enhancing the effect. This implies that the synthesis of phages will be higher with a denser bacterial concentration, similar to that within a biofilm (Chaudry et al. 2017). Phages are host specific because they bond to 1 or a small number of receptors on the cell wall by recognizing bacterial surface proteins. This implies that they will attack only 1 type or a small variety of different bacteria (Sulakvelidze et al. 2001, Yilmaz et al. 2013, Lin et al. 2017, Akanda et al. 2018). The great advantage is that phageresistant bacteria remain susceptible to other phages of a similar target range (Sulakvelidze et al. 2001, Yilmaz et al. 2013, Chaudry et al. 2017). These new phages can be processed rapidly. By using several phages in combination, 1 can delay the development of resistance, as described by Sulakvelidze et al. (2001). Furthermore, phages can dissolve the bacterial biofilm (Ryan et al. 2012, Chaudry et al. 2017). In the English language, publications on in vivo bacteriophage therapy as a treatment for bone-related infections are scarce. Available data has shown a synergetic effect of phage therapy in combination with antibiotics (Yilmaz et al. 2013, Akanda et al. 2018). In an in vitro model, the pre-antibiotic administration of phages appeared to be the most efficient approach (Chaudry et al. 2017, Akanda et al. 2018). It is believed that use of phages will lower the required antibiotic concentration (Ryan et al. 2012, Akanda et al. 2018). This is why our patient took his previously prescribed antibiotics and his phage cocktails simultaneously. None of the published phase I studies reported any adverse outcomes (Sulakvelidze et al. 2001, Ryan et al. 2012, Yilmaz et al. 2013, Chaudry et al. 2017, Lin et al. 2017, Akanda et al. 2018). Similarly, our patient did not perceive any relevant side effects. The disease-causing bacterium must be identified before phage susceptibility can be determined and therapy can begin (Yilmaz et al. 2013). Therefore, phage therapy is not a viable treatment option for acute PJI. Another limitation is that published studies have only focused on the treatment of S. aureusand Pseudomonas-induced PJIs (El Helou et al. 2008, Ryan et

Acta Orthopaedica 2021; 92 (6): 678–680

al. 2012, Yilmaz et al. 2013, Chaudry et al. 2017, Akanda et al. 2018, Renz et al. 2019). In conclusion, to our knowledge, this is the first case report on bacteriophage therapy for an E. faecalis PJI. Due to increasing rates of resistance to currently used antibiotics or difficultto-treat PJIs, alternative treatment options are of high importance. One of these “new” options is bacteriophage therapy, which is interesting due to its ability to attack biofilms associated with PJIs. The available literature reports excellent safety profiles and promising results for combination therapy with regular antibiotics. However, many aspects of phage therapy remain unclear. Ethics, funding, and potential conflicts of interest Written informed consent was obtained from the patient for publication of this case report. No funding was obtained. The authors have no conflicts of interest to declare. AN wrote the manuscript and produced the figures; JG performed the surgery; HB performed the microbial culture; JG, HB, and AH were responsible for English editing and reviewing the manuscript. All authors reviewed the final version of the manuscript. Acta thanks Meritxell de Jesús Garcia Quintanilla for help with peer review of this study.

Akanda Z, Taha M, Abdelbary H. Current review: the rise of bacteriophage as a unique therapeutic platform in treating peri-prosthetic joint infections. J Orthop Res 2018; 36(4): 1051-60. doi: 10.1002/jor.23755 Chaudry W, Concepción-Acevedo J, Park T, Andleeb S, Bull J J, Levin B R. Synergy and order effects of antibiotics and phages in killing Pseudomonas aeruginosa biofilms. PLoS One 2017; 12: e0168615. El Helou O, Berbari E, Marculescu C, El Atrouni, W I, Razonable, R R, Steckelberg J M, Hanssen A D, Osmon D R. Outcome of enterococcal prosthetic joint infection: is combination systemic therapy superior to monotherapy? Clin Infect Dis 2008; 47: 903-9. Fauconnier A. Phage therapy regulation: from night to dawn. Viruses 2019; 11(4): 352. Karczewski D, Winkler T, Renz N, Trampuz A, Lieb E, Perka C, MüllerM. A standardized interdisciplinary algorithm for the treatment of prosthetic joint infections. Bone Joint J 2019; 101(2): 132-9. Lin D, Koskella B, Lin H. Phage therapy: an alternative to antibiotics in the age of multi-drug resistance. World J Gastrointest Pharmacol Ther 2017; 8(3): 162-173. Renz N, Trebse R, Akgün D, Perka C, Trampuz A. Enterococcal periprosthetic joint infection: clinical and microbiological findings from an 8-year retrospective cohort study. BMC Infect Dis 2019; 19: 1083. Ryan E, Alkawareek M, Donnelly R, Gilmore B F. Synergistic phage-antibiotic combinations for the control of Escherichia coli biofilms in vitro. FEMS Immunol Med Microbiol 2012; 65: 395-8. Sulakvelidze A, Alavidze Z, Morris J. Bacteriophage therapy. Antimicrob Agents Chemother 2001; 45: 649-659. Yilmaz C, Colak M, Yilmaz B C, Gulden E, Kutateladze M, Gozlugol M. Bacteriophage therapy in implant-related infections: an experimental study. J Bone Joint Surg Am 2013; 95: 117-25.

Acta Orthopaedica 2021; 92 (6): 681–688

681

Measurement properties of UCLA Activity Scale for hip and knee arthroplasty patients and translation and cultural adaptation into Danish Anne MØRUP-PETERSEN 1, Søren T SKOU 2, Christina E HOLM 3, Pætur M HOLM 2, Claus VARNUM 4, Michael R KROGSGAARD 5, Mogens LAURSEN 6, and Anders ODGAARD 1,7 1 Department

of Orthopedic Surgery, University of Copenhagen, Herlev and Gentofte Hospital, Copenhagen; 2 Research Unit for Musculoskeletal Function and Physiotherapy, Department of Sports Science and Clinical Biomechanics, University of Southern Denmark, Odense; Research unit PROgrez, Department of Physiotherapy and Occupational Therapy, Næstved-Slagelse-Ringsted Hospitals, Region Zealand; 3 Department of Orthopedic Surgery, Copenhagen University Hospital, Rigshospitalet; 4 Department of Orthopedic Surgery, Lillebaelt Hospital—Vejle, University Hospital of Southern Denmark; Department of Regional Health Research, Faculty of Health Science, University of Southern Denmark; 5 Section for Sports Traumatology, Bispebjerg and Frederiksberg Hospital, Copenhagen, University of Copenhagen; 6 Department of Orthopedic Surgery, Aalborg University Hospital, Aalborg & Farsø; 7 Department of Orthopaedic Surgery, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark; Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark Correspondence: morup.anne@gmail.com Submitted 2021-02-02. Accepted 2021-08-16.

Background and purpose — The UCLA Activity Scale (UCLA) is a questionnaire assessing physical activity level from 1 (low) to 10 (high) in patients undergoing hip or knee arthroplasty (HA/KA). After translation and cultural adaptation, we evaluated the measurement properties of the Danish UCLA. Patients and methods — After dual panel translation, cognitive interviews were performed among 55 HA/ KA patients. An orthopedic surgeon and a physiotherapist estimated UCLA scores for 80 KA patients based on short interviews. Measurement properties were evaluated in 130 HA and 134 KA patients preoperatively and 1-year postoperatively. Results — To suit Danish patients of today, several adaptations were required. Prior to interviews, 4 patients were excluded, and 11 misinterpreted the answer options. Examiners rated the remaining 65 patients (mean age 67 years) 0.2–1.6 UCLA levels lower than patients themselves. The 130 HA and 134 KA patients (mean age 71/68 years) changed from 4.3 (SD 1.9)/4.5 (1.8) preoperatively to 6.6 (1.8)/6.2 (1.0) at 1-year follow-up. 103 (79%) HA and 89 (66%) KA patients reported increased activity. Effect sizes were large (1.2/0.96). Knee patients reaching minimal important change (MIC, ≥ 8 Oxford Knee Score points) had higher 1-year UCLA scores than patients not reaching MIC. Interpretation — Original scale development was undocumented. Content validity was questionable, and there was discrepancy between patient and examiner estimates. UCLA appears valuable for measuring change in self-reported physical activity on a group level. 4 out of 5 HA patients and 2 out of 3 KA patients were more physically active 1 year after joint replacement surgery.

Hip and knee osteoarthritis (OA) strongly affect a person’s ability to be physically active (Price et al. 2018). When pain and functional impairment becomes so severe that joint replacement is considered, it is of interest for both patients and healthcare providers to know to what degree surgery can be expected to improve a patient’s opportunity to lead an active life. Yet, quantifying physical activity is complex. Accelerometers are often considered as the gold standard for measuring non-specific physical activity; however, being resource demanding they are often not a feasible option and, also, accelerometer results do not necessarily reflect the difficulty of the activities or how important an activity is to each patient (Shephard 2003). As an alternative, physical activity can be quantified using physical activity scales such as the UCLA Activity Scale (UCLA) from University of California, Los Angeles (Table 1) (Amstutz et al. 1984, Zahiri et al. 1998). UCLA is a single-item 10-level-scale, ranging from level 10, representing a highly physically active patient, to level 1, a patient who is dependent on others and unable to leave home. A description of the development process leading to UCLA has to our knowledge never been published (Amstutz et al. 1984). Originally, it appears to have been made for surgeons to assess activity levels of hip and knee arthroplasty patients (Zahiri et al. 1998). Today, UCLA is used as a patient-reported outcome measure (PROM), although it was probably not developed as such. A study comparing UCLA scores with accelerometer measurements of walking activity revealed a strong correlation but large measurement errors; patients reporting the same level of activity in UCLA varied up to a factor of 15 in average number of steps per day (Zahiri et al. 1998). Despite this, UCLA is widely used internationally and has been recommended as a useful physical activity PROM

682

Table 1. UCLA Activity Scale as first published by Amstutz et al. (1984)

Acta Orthopaedica 2021; 92 (6): 681–688

Step 1: Translation and cultural adaptation

Consensus version

Activity level

Translator

1 Wholly inactive: dependent on others; cannot leave residence 2 Mostly inactive: very restricted to minimum activities of daily living 3 Sometimes participates in mild activities such as walking, limited housework, and limited shopping 4 Regularly participates in mild activities 5 Sometimes participates in moderate activities such as swimming and can do unlimited housework or shopping 6 Regularly participates in moderate activities 7 Regularly participates in active events such as bicycling 8 Regularly participates in very active events such as bowling or golf 9 Sometimes participates in impact sports such as jogging, tennis, skiing, acrobatics, ballet, heavy labor, or backpacking 10 Regularly participates in impact sports

Orthopedic surgeon

In the later version (Zahiri et al. 1998), level 10 was presented first and a patient instruction was added: “Of the following options, which statement best describes your activity level?”

instruments in hip and knee arthroplasty (HA/KA) patient populations, mainly based on a positive rating of construct validity and high completion rates (Naal et al. 2009, Terwee et al. 2011, Rolfson et al. 2016). Its brevity and simplicity make it an attractive choice, especially when combined with other questionnaires. In Denmark, UCLA has been used in at least 2 different, unpublished versions. With a direct translation, cultural differences in bicycling habits led to a bimodal distribution of answers (Skou and Roos 2014). This study develops a Danish version of the UCLA through formal translation and cultural adaptation, and further tests the validity, reliability, and responsiveness of the translated questionnaire in relevant groups of hip and knee OA patients before and/or after arthroplasty.

Patients and methods The study was conducted in 4 parts (Figure 1): (1) translation and cultural adaptation, followed by evaluation of (2) correlation with external assessment (by healthcare professionals), and (3) test–retest reliability in KA patients, and (4) construct validity and responsiveness in a cohort of KA and HA patients. Study design was guided by the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) guidelines (Mokkink et al. 2018) and the Guidelines for Reporting Reliability and Agreement Studies (GRRAS) (Kottner et al. 2011). Translation and cultural adaptation A dual panel translation (Swaine-Verdier et al. 2004, McKenna and Doward 2005, Epstein et al. 2015) was made by a professional translator, a physiotherapist, and an orthopedic

Lay person feedback n = 22

Physiotherapist

Hip/knee patient feedback n = 55 Final version

Step 2 & 3: Correlation with external assessment & test-retest reliability Patient n = 76

Patient (retest)

Physiotherapist interview session

Orthopedic surgeon interview session

Same version Step 4: Responsiveness and interpretability Knee OA patients n = 134

Arthroplasty

1 year postoperative evaluation

Hip OA patients n = 130

Arthroplasty

1 year postoperative evaluation

Figure 1. Study overview.

surgeon (senior house officer, AM) in collaboration; all 3 were English–Danish bilingual with Danish mother-tongue. Each of the three prepared a Danish translation from the original American version (Table 1) and, subsequently, they met to discuss and agree on a consensus version. In case of disagreement, majority ruled. To ensure wording and cultural adaption, the questionnaire was presented to 3 different laymen panels of total 22 (10 males) heart and lung patients with a mean age of 72 years (SD 9), recruited at physiotherapy team training sessions. Participants completed the questionnaire while “thinking out loud,” and they commented on the questionnaire in plenary sessions. In turn, changes were made in layout, instructions, and activity examples. Subsequently, the revised version was presented in the same manner to target patients: 55 HA or KA patients (38 pre- and 17 postoperative [21 males]), mean age 70 years (SD 8) at Copenhagen University Hospital Gentofte. After 8 rounds of adjustments, the evaluations led to no further revisions and the development process was ended. Correlation with external assessment As UCLA was originally completed by surgeons, we intended to determine the degree of common understanding of activity levels between knee patients and healthcare professionals. During a 3-month period, KA patients (> 40 years, pre- and postoperative, primary and revision KA) were recruited at Naestved Hospital, Region Zealand. We excluded patients unable to read and understand the Danish language and patients with signs of dementia who failed a clock-drawing test (Mainland et al. 2014). Patients filled out UCLA without any help except for a short, written instruction, asking them to consider their physical activity level in the preceding 4 weeks. Age, sex, height, weight, and today’s knee pain level

Acta Orthopaedica 2021; 92 (6): 681–688

on a visual analogue scale (VAS 0–10, 0 “no pain,” 10 “worst pain imaginable”) was registered. Shortly after, patients were interviewed about daily physical activities for about 5 minutes by (1) a junior orthopedic surgeon (AM/CH) and (2) a physiotherapist (PH), separately and in random order. Each examiner then estimated the patient’s UCLA level (patients were instructed not to reveal their own reported UCLA score). Test–retest reliability After interviews, participants were given a prepaid envelope containing an extra UCLA questionnaire to complete 7–10 days later. On the front page, patients were asked if their physical activity habits had changed since the first test, since only those with unchanged habits were eligible for analysis. Patients awaiting or recovering from surgery (< 6 weeks) were excluded from retests as their activity level was expected to change rapidly. Construct validity and responsiveness UCLA score distribution and responsiveness (validity of change scores) were evaluated at Lillebaelt Hospital—Vejle, Region of Southern Denmark. As part of the normal clinical routine, through 5 months (inclusion March–July 2018), hip OA patients scheduled for HA and knee OA patients scheduled for total or medial unicompartmental KA completed electronic PROM questionnaires (forcing patients to choose only one answer option) before and 1 year after surgery (Procordo Software, Copenhagen). Patients who had revision surgery during year 1 were excluded. Paper versions were available for patients with no email address. Non-responders were reminded by mail and, if necessary, by phone. The PROM questionnaires included UCLA, and the well-established Oxford Hip or Knee Score (OHS, OKS) (Dawson et al. 1996, 1998, Murray et al. 2007, Hossain et al. 2015), the generic EQ-5D-5L and EQ-5D VAS (Jin et al. 2019), and an overall patient satisfaction question (“How satisfied are you with your hip/knee 1 year after surgery?”, 5 answer options, 1 neutral). Responsiveness was evaluated by use of the construct approach (de Vet et al. 2011), i.e., correlation of UCLA change with other PROM change scores and overall satisfaction. We expected only fair to moderate correlations (Naal et al. 2009) because (1) generic and joint-specific PROMs evaluate factors other than physical activity, (2) joint replacement may improve the ability to be physically active without changing the patients’ habits (because of, e.g., lack of motivation), and (3) perception of change may be influenced by preoperative expectations. A mean increase of 1–3 UCLA levels 1 year after surgery was expected (SooHoo et al. 2015, Ghomrawi et al. 2017, Scott et al. 2017), as was a 2-fold increase in the proportion of patients with UCLA score ≥ 6 (Scott et al. 2017). We also calculated the effect size, a traditional distribution-based measure to quantify responsiveness (Angst 2011).

683

Statistics UCLA scores were not expected to be equidistant or normally distributed, thus scores were treated as ordinal variables and analyzed using nonparametric statistical methods (Wilcoxon rank sum and Kruskal–Wallis test). To illustrate variations in results, means and standard deviations (SD) were reported as well, and (multiple) linear regression analyses were performed to check for score dependence on age, sex, and BMI. Paired tests (Wilcoxon signed rank test) were used to calculate within-patient differences in UCLA scores. Associations between 1-year (change) UCLA and reaching minimal important change (MIC) of 8 OHS/OKS points were assessed (Beard et al. 2015, Ingelsrud et al. 2018). For correlation with external assessment, agreements were estimated by mean difference, limits of agreement (LoA), weighted kappa coefficient (Landis and Koch 1977), and Spearman’s correlation coefficient (“very weak” [0–0.19], “weak” [0.20–0.39], “moderate” [0.40–0.59], “strong” [0.60– 0.79], and “very strong” [0.80–1.0]). Floor or ceiling effects were considered present if more than 15% of patients marked the lowest or highest score, respectively (Terwee et al. 2007) and effect sizes were calculated (mean UCLA change/SDbaseline) (Angst 2011, de Vet et al. 2011). Sample size was based on general recommendations (Terwee et al. 2007, de Vet et al. 2011). Statistical significance level was set at alpha level 0.05 (2-sided), and 95% confidence intervals (CI) were reported when relevant. Analyses were conducted in R (Rstudio) (RCoreTeam; R Foundation for Statistical Computing, Vienna, Austria). Ethics, data sharing, funding, and potential conflicts of interest The study was ethically approved by the National Committee of Health Research Ethics (Jr. no. 16030260). Data management was approved by the Danish Data Protection Agency (Jr. no. 2012-58-0004). Raw data is available upon request. The study was funded by the Health Research Fund of the Capital Region of Denmark. No authors had relevant conflicts of interest.

Results Translation and cultural adaptation The original American version of UCLA needed comprehensive changes in the aspiration of becoming a valid and patientrelevant measure of physical activity in Danish hip and knee replacement patients of today. For example, bicycling for transportation is very common in Denmark, even among the elderly. To prevent a bimodal score distribution (maximum at levels 4 and 7) as seen in a study based on a previous version (Skou and Roos 2014), the bicycling activity was split by intensity and frequency to cover levels 5–8 in the current version (appendix A [Danish] and B [English translation], see

684

Acta Orthopaedica 2021; 92 (6): 681–688

Table 2. Characteristics of interview participants

Examiners’ UCLA estimate 10

Patients (knee arthroplasty/knee OA), n Age, mean [median] (SD) Male sex, n BMI, mean (SD) (n = 60) Pain (VAS 0–10), mean [median] (SD) (n = 63)

65 (34/31) 66 [67] (11) 29 30.4 (5.8) 4.5 [5] (2.8)

Examiner Orthopedic surgeon Physiotherapist

Table 3. Results of interviews: correlation with external assessment of physical activity level

UCLA Activity scale

Absolute scores (total sample) Mean (SD) Median [IQR] Range Patient 5.0 (1.7) 5 [4–6] 2–10 Surgeon 4.4 (1.6) 4 [3–6] 1–8 Physiotherapist 3.8 (1.3) 4 [3–4] 2–8 Differences (within-patient) Patient minus Mean (CI) Median Range surgeon 0.6 (0.2–0.9) 0 –3 to 3 physiotherapist 1.2 (0.8–1.6) 1 –2 to 6 Reliability Agreement Correlation Patient minus weighted Kappa LoA Spearman’s rho surgeon 0.63 –2.0 to 3.1 0.65 physiotherapist 0.31 –2.0 to 4.4 0.47 Spearman’s rho correlation coefficient (–1 to 1) indicates the degree of linearity between measurement ranks. LoA = limits of agreement (mean ±2 SD). Differences are based on assessments within each patient.

Patients’ UCLA estimate

Figure 2. Correlation with external assessment of physical activity level: orthopedic surgeons’ and physiotherapists’ estimates of UCLA plotted against patients’ own estimates, with corresponding regression lines. Random variance (jitter) is added to prevent over-plotting. The red dotted line indicates perfect agreement. Patients’ UCLA estimate – examiners’ UCLA estimate

+2SD

2 Mean

–2SD

Supplementary data). Acrobatics, ballet, and bowling were exchanged for popular Danish activities, e.g., badminton and gymnastics/fitness. Examples were mentioned only once, but curly brackets illustrated how examples referred to two levels differing by frequency (“regularly” or “once in a while”). The questionnaire had to be self-explanatory, thus a short introductory text was added. During development, we found that thorough instructions led to patients skipping the introduction and misinterpreting the scale and marking multiple boxes or writing numbers instead. Cutting down instructions to a minimum led to fewer misunderstandings.

Figure 3. Patients’ UCLA estimate plotted against the difference between patients’ and examiners’ assessments (patient score minus mean of surgeons’ and physiotherapists’ scores) (modified Bland– Altman plot). The dotted lines indicate mean difference (blue) ±2 SD (limits of agreement, black) and hypothetical perfect agreement (red). Random variance (jitter) is added to prevent over-plotting.

Correlation with external assessment We invited 80 knee OA and KA patients for interviews. 2 were excluded due to poor language skills, and 2 because their paper questionnaires were lost. Of 76 patients, 11 (67 years) were excluded because they marked more than 1 answer. The remaining 65 patients (Table 2) overall rated their UCLA level higher than the examiners did (Table 3, Figure 2) and differences increased with UCLA level (Figure 3). In 32 cases, 1 or both examiners agreed perfectly with the patient or the patient’s score was between examiner estimates. The reliability of examiner assessment of patient activity level (Table 3) was “substantial” for surgeons and “fair” for the physiothera-

pist (Landis and Koch 1977). The corresponding correlations were “strong” and “moderate,” respectively. Patient UCLA was 4.8 (SD 1.7) in females and 5.3 (SD 1.6) in males (difference: CI –0.3 to 1.4). No association was observed between patient UCLA and age (–0.008 per year, CI –0.05 to 0.03) or current knee pain (–0.1 per increase in VAS, CI –0.3 to 0.03), but a small, negative association with BMI was detected (–0.08 per BMI unit, CI –0.15 to –0.01). Results were similar with multiple regression analysis. For examiner estimates, none of these factors were independently associated with activity level.

–2

–4

Patients’ UCLA estimate

Acta Orthopaedica 2021; 92 (6): 681–688

685

Count, hip patients

Count, knee patients

Count, hip patients

Count, knee patients 40

Baseline 1-year follow-up 40

Lower Unchanged Higher

–4

–2

UCLA Acitivity Scale

Figure 4. Distribution of pre- and 1-year postoperative UCLA scores in hip (left panel) and knee arthroplasty patients (right panel).

–4

–2

UCLA change at 1-year postoperative

Figure 5. Distribution of UCLA change scores (1-year follow-up minus preoperative score) in hip (left panel) and knee arthroplasty patients (right panel).

Table 4. Patient characteristics and UCLA values in hip and knee arthroplasty patients. Values are count (%) unless otherwise specified Factor Baseline Hips Male/female Age, mean (SD) UCLA a UCLA change a UCLA ≥ 6 Floor/ceiling c Knees Male/female Age, mean (SD) UCLA a UCLA change a UCLA ≥ 6 Floor/ceiling c

Patient satisfaction 1-year Very postoperative dissatisfied Dissatisfied Neither

130 (100) 130 (100) 62/68 (52/48) - 71 (9) - 4.3 [4] (1.9) 6.6 [6] (1.8) – 2.3 [2] (2.0) 29 (22) 96 (74) 6/3 (5/2) 0/13 (0/10) 134 (100) 134 (100) 61/73 (46/54) - 68 (9) - 4.5 [4] (1.8) 6.2 [6] (1.0) - 1.7 [1] (2.3) 26 (19) 85 (63) 2/2 (1/1) 0/10 (0/7)

a Values are mean [median] (SD). b Values not shown (1 patient only). c Floor/ceiling denotes the number (and

1 (1) 0/1 b b b b b 1 (1) 1/0 b b b b b

0 (0) 4 (3) 0/0 2/2 - 75 (6) - 6.3 [7] (2.1) - 2.5 [3] (1.9) - 2 (50) - 0/0 (0/0) 11 (8) 11 (8) 5/6 6/5 72 (6) 71 (8) 4.6 [4] (1.7) 5.1 [4] (2.2) 0.0 [0] (1.6) 0.0 [0] (2.0) 3 (27) 3 (27) 0/0 (0/0) 0/1 (0/10)

Satisfied

Very satisfied

16 (12) 9/7 76 (8) 6.2 [6] (2.3) 1.8 [2] (1.8) 8 (50) 0/2 (0/13) 40 (30) 21/19 66 (10) 6.0 [4] (1.6) 1.3 [1] (1.9) 24 (60) 0/1 (0/3)

109 (84) 51/58 70 (9) 6.7 [6] (1.8) 2.4 [2] (2.0) 86 (79) 0/11 (0/10) 71 (53) 28/43 68 (9) 6.8 [4] (1.9) 2.6 [2] (2.3) 54 (76) 0/8 (0/11)

percentage) of patients reporting level 1 or 10.

Test–retest reliability Retest questionnaires were returned by 43 of 53 patients. Exclusions were made for 2 who had completed the retest form on day 0, 1 who had marked multiple boxes and 2 returning blank forms. Of the remaining 38, 21 reported to have “unchanged exercise habits” after 8.3 days (range 1–25). In this group, 13 had perfect agreement with their initial score, 5 were 1 level apart, and 1 was 2 levels apart. Construct validity and responsiveness Completeness at 1 year reached 96% (HA, n = 130) and 95% (KA, n = 134), respectively. There were no statistically significant sex differences in scores (p = 0.7–1.0). UCLA typically improved from median level 4 to 6 in both groups (Figure 4, Table 4). Positive change in UCLA was reported by 103 (79%) hip and 89 (66%) knee patients (Figure 5). Patient satisfaction and change in other PROMs proved very weak to moderate

correlations with UCLA change scores (Table 5), and largest in KA patients. Knee patients reaching MIC (≥ 8 OKSchange) reported higher 1-year UCLA levels than patients not reaching MIC (1-year UCLA 6.4 and 5.2, respectively, p < 0.04) and had higher change scores (2.1 and –0.2, p < 0.001). In hip patients, the corresponding UCLA scores were 6.6 and 5.7 (p = 0.2) and change scores 2.4 and 1.5 (p = 0.3), respectively. Effect size was 1.2 in HA and 0.96 in KA patients; both “large” (≥ 0.8).

Discussion The UCLA Activity Scale (UCLA) was translated into Danish with several cultural adaptations required for the scale to be relevant to Danish hip and knee arthroplasty patients of today. Based on interviews, examiners rated patients lower on UCLA than did patients themselves. A 1-year postoperative increase

686

Acta Orthopaedica 2021; 92 (6): 681–688

Table 5. Change in relevant PROMs grouped by 1-year UCLA improvement. Values are mean (SD) unless otherwise specified Factor Hips, n (%) 1-year OHS ∆ OHS ∆ EQ-VAS Satisfied or very satisfied. n (%) Knee, n (%) 1-year OKS ∆ OKS ∆ EQ-VAS Satisfied or very satisfied, n (%)

UCLA improvement 1–2 3–4

≥ 5

Correlation Spearman’s rho

35 (27) 44 (5) 25 (8) 33 (22)

18 (14) 44 (6) 27 (9) 38 (27)

0.09 0.21 0.29

49 (98) 44 (33) 38 (9) 18 (8) 16 (18)

33 (94) 25 (19) 39 (5) 19 (7) 26 (18)

18 (100) 20 (15) 45 (4) 29 (6) 37 (26)

0.09 – 0.30 0.44 0.39

38 (86)

23 (92)

20 (100)

0.39

≤ –1

5 (4) 45 (2) 24 (5) 13 (26)

22 (17) 43 (6) 20 (7) 20 (15)

50 (38) 44 (5) 22 (8) 24 (27)

5 (100) 16 (12) 34 (10) 13 (11) 1 (26)

20 (91) 29 (22) 38 (9) 14 (9) 16 (25)

9 (56)

21 (72)

OHS/OKS: Oxford Hip/Knee Score (0–48, 48 best). ∆ (Delta): change scores from baseline to 1 year postoperatively (EQ-5D-5L results did not provide further valuable information, thus only EQ VAS results are reported). Correlations denote the non-parametrical correlation between the given parameter and UCLA change score (in “satisfaction,” all 5 levels were used in correlation analyses).

in UCLA was reported by four-fifths of hip and two-thirds of knee arthroplasty patients. We could not identify reports on the original development and purpose of UCLA. At the time when UCLA was first introduced, there was a need to determine the association between physical activity and polyethylene wear after joint replacement. Since then, polyethylene wear has come to play a smaller role in revisions and the interest in UCLA seems to have shifted towards evaluating the general health benefits of surgery. Despite involvement of patients in the current translation process, the uncertainty of patient involvement in the original scale development remains problematic. UCLA has no proven face or content validity, and this cannot be compensated for by good measurement properties (Mokkink et al. 2018). Interpretation of UCLA results involves obvious challenges. It encompasses several dimensions in one item: intensity, frequency, activity type, difficulty, and duration. This may be the price paid for brevity, but it can lead to large variations in individual perception of the scale, as the levels are neither mutually exclusive, nor exhaustive. For example, say you work hard at the gym once in a while but you are not able to kneel to do your usual garden work, which activity level should you indicate? To the best of our knowledge, agreement among patients’ and professionals’ UCLA estimates has not been evaluated before. The systematic differences of 0.2–1.6 points (examiners lower) and wide limits of agreement (95% LoA -2.2–4.4) underline that patient-reported outcomes and professional evaluation are not identical measures, and that interpretation of UCLA may be highly subjective. This, along with previous findings (Zahiri et al. 1998), suggests that comparison of individual patients’ UCLA levels should be made with great caution.

Despite several attempts to make the questionnaire self-explanatory (8 rounds of changes), 11 of 76 patients misunderstood the response options. With an electronic version allowing only 1 response, much of this problem is overcome as patients are guided towards a uniform response to the scale (Gudbergsen et al. 2011). Publications of previous versions of UCLA have not included histograms of score distributions. With the present version, both patients and examiners were more likely to choose levels 4, 6, 8, and 10 (where activities are performed “regularly”) than levels 3, 5, 7 and 9 (performed “sometimes”), perhaps because people tend to have regularity in their life. For example, skiing once a year can be considered a “regular” activity. Theoretically, a more even score distribution might be expected if the term “regularly” were replaced by “often”.

Interpretation and generalizability Danish mean scores were comparable to international results: e.g., baseline UCLA in Danish HA/KA patients were 4.3/4.5 corresponding to, e.g., 4.3/4.2 in California (SooHoo et al. 2015). Danish 1-year change from median score 4 to 6 was in accordance with a study of 261 British KA patients (59 years) (Scott et al. 2017). In that study by Scott et al., the number of KA patients reporting that they were very physically active (≥ level 6) increased from 37% to 72% after surgery. In our sample, numbers increased from 26 to 85 of 134 (KA) and from 29 to 96 of 130 (HA), thus the share of very active patients more than tripled in each group. Knee (but not hip) patients with postoperative clinically important improvement in Oxford scores had higher 1-year UCLA and UCLA change scores than others. It is not a given that all patients have a desire to become more physically active after a successful joint replacement, thus, as expected, UCLA correlated only poorly to moderately with other PROMs and overall patient satisfaction where, e.g., pain relief counts, too. Limitations and strengths No previous studies have addressed all measurement property aspects of UCLA or discussed its shortcomings in depth. Regarding reliability, we found no floor or ceiling effect, which was in accordance with previous studies (Naal et al. 2009). However, due to the low retest sample size (21 patients), we are reluctant to calculate weighted kappa or make conclusions about measurement error, an important aspect of reliability, which remains uncertain. The reported (lack of) association between UCLA and age, sex, pain, and BMI should be considered with caution, as the study was not powered to study these matters.

Acta Orthopaedica 2021; 92 (6): 681–688

Conclusion Based on the findings of this study, the UCLA Activity Scale (UCLA) probably cannot provide a valid measure of physical activity level in the individual patient, but the scale is useful on a group level. 1 year after joint replacement, 4 out of 5 hip patients and 2 out of 3 knee patients were more physically active. This information is relevant to hip and knee osteoarthritis patients considering joint replacement surgery. Authors recommend use of an electronic version of UCLA, if possible. Future reliability studies should include retests of more patients, and responsiveness studies should include a specific anchor question regarding change in physical activity to allow for an anchor-based calculation of minimal important change (MICUCLA). Validation of UCLA against other more comprehensive patient-reported activity scales, accelerometers, or performance-based measures would be interesting, though the underlying construct may be very different across measurement methods. Supplementary data Appendices are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.20 21.1977533 AM participated in all parts of the project and performed all statistical analyses. MK, ML, and AO took part in planning the study. STS participated in translation. CH and PH were involved in translation and conducted validity studies, while CV conducted responsiveness studies. All authors critically revised the final manuscript. The authors would like to thank all involved patients for their valuable time. They thank Henrik M. Schrøder and colleagues (nurses, secretaries, and surgeons) at Naestved Hospital and nurses and secretaries at Gentofte Hospital for good cooperation, Carsten Bogh Juhl for helpful advice during study planning, and Tobias Wirenfeldt-Klausen for statistical advice. Acta thanks Jesper Bie Larsen and David J Beard for help with peer review of this study.

Amstutz H C, Thomas B J, Jinnah R, Kim W, Grogan T, Yale C. Treatment of primary osteoarthritis of the hip: a comparison of total joint and surface replacement arthroplasty. J Bone Joint Surg Am 1984; 66(2): 228-41. Angst F. The new COSMIN guidelines confront traditional concepts of responsiveness. BMC Med Res Methodol 2011; 11: 152; author reply 152. Beard D J, Harris K, Dawson J, Doll H, Murray D W, Carr A J, Price A J. Meaningful changes for the Oxford hip and knee scores after joint replacement surgery. J Clin Epidemiol 2015; 68(1): 73-9. Dawson J, Fitzpatrick R, Carr A, Murray D. Questionnaire on the perceptions of patients about total hip replacement. J Bone Jt Surg - Ser B 1996; 78(2): 185-90. Dawson J, Fitzpatrick R, Murray D, Carr A. Questionnaire on the perceptions of patients about total knee replacement. J Bone Jt Surg Br 1998; 80(1): 63-9. de Vet H C W, Terwee C B, Mokkink L B, Knol D L. Measurement in medicine: a practical guide. Cambridge: Cambridge University Press; 2011. Epstein J, Santo R M, Guillemin F. A review of guidelines for cross-cultural adaptation of questionnaires could not bring out a consensus. J Clin Epidemiol 2015; 68(4): 435-41.

687

Ghomrawi H M K, Lee Y, Herrero C, Joseph A, Padgett D, Westrich G, Parks M, Lyman S. A crosswalk between UCLA and lower extremity activity scales. Clin Orthop Relat Res 2017; 475(2): 542-8. Gudbergsen H, Bartels E M, Krusager P, Waehrens E E, Christensen R, Danneskiold-Samsoe B, Bliddal H. Test–retest of computerized health status questionnaires frequently used in the monitoring of knee osteoarthritis: a randomized crossover trial. BMC Musculoskelet Disord 2011; 12: 190. Hossain F S, Konan S, Patel S, Rodriguez-Merchan E C, Haddad F S. The assessment of outcome after total knee arthroplasty: are we there yet? Bone Joint J 2015; 97-b(1): 3-9. Ingelsrud L H, Roos E M, Terluin B, Gromov K, Husted H, Troelsen A. Minimal important change values for the Oxford Knee Score and the Forgotten Joint Score at 1 year after total knee replacement. Acta Orthop 2018; 89(5): 541-7. Jin X, Al Sayah F, Ohinmaa A, Marshall D A, Smith C, Johnson J A. The EQ-5D-5L is superior to the -3L version in measuring health-related quality of life in patients awaiting THA or TKA. Clin Orthop Relat Res 2019; 477(7): 1632-44. Kottner J, Audige L, Brorson S, Donner A, Gajewski B J, Hrobjartsson A, Roberts C, Shoukri M, Streiner D L. Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were proposed. J Clin Epidemiol 2011; 64(1): 96-106. Landis J R, Koch G G. The measurement of observer agreement for categorical data. Biometrics 1977; 33(1): 159-74. Mainland B J, Amodeo S, Shulman K I. Multiple clock drawing scoring systems: simpler is better. Int J Geriatr Psychiatry 2014; 29(2): 127-36. McKenna S P, Doward L C. The translation and cultural adaptation of patient-reported outcome measures. Value Heal 2005; 8(2): 89-91. Mokkink L B, Prinsen C A C, Patrick D, Alonso J, Bouter L, de Vet H C W, Terwee C B. COSMIN methodology for systematic reviews of patient– reported outcome measures (PROMs). User manual, version 1.0; 2018. Murray D W, Fitzpatrick R, Rogers K, Pandit H, Beard D J, Carr A J, Dawson J. The use of the Oxford hip and knee scores. J Bone Joint Surg Br 2007; 89(8): 1010-4. Naal F D, Impellizzeri F M, Leunig M. Which is the best activity rating scale for patients undergoing total joint arthroplasty? Clin Orthop Relat Res 2009; 467(4): 958-65. 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. RCoreTeam. RStudio version 1.1.463. R: A Language and Environment for Statistical Computing. Available from: https://www.R-project.org/. Rolfson O, Bohm E, Franklin P, Lyman S, Denissen G, Dawson J, Dunn J, Eresian Chenok K, Dunbar M, Overgaard S, Garellick G, Lubbeke A, Patient-Reported Outcome Measures Working Group of the International Society of Arthroplasty R. Patient-reported outcome measures in arthroplasty registries. Report of the Patient-Reported Outcome Measures Working Group of the International Society of Arthroplasty Registries Part II. Recommendations for selection, administration, and analysis. Acta Orthop 2016; 87(Suppl. 1): 9-23. Scott C E H, Turnbull G S, MacDonald D, Breusch S J. Activity levels and return to work following total knee arthroplasty in patients under 65 years of age. Bone Joint J 2017; 99-b(8): 1037-46. Shephard R J. Limits to the measurement of habitual physical activity by questionnaires. Br J Sport Med 2003; 37(3): 197-206; discussion 206. Skou S T, Roos E. GLA:D (Good Life with osteoArthritis in Denmark) Annual Report 2014; 28. SooHoo N F, Li Z, Chenok K E, Bozic K J. Responsiveness of patient reported outcome measures in total joint arthroplasty patients. J Arthroplast 2015; 30(2): 176-91. Swaine-Verdier A, Doward L C, Hagell P, Thorsen H, McKenna S P. Adapting quality of life instruments. Value Heal 2004; 7(Suppl. 1): S27-30. Terwee C B, Bot S D, de Boer M R, van der Windt D A, Knol D L, Dekker J, Bouter L M, de Vet H C. Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol 2007; 60(1): 34-42.

688

Terwee C B, Bouwmeester W, van Elsland S L, de Vet H C, Dekker J. Instruments to assess physical activity in patients with osteoarthritis of the hip or knee: a systematic review of measurement properties. Osteoarthritis Cartilage 2011; 19(6): 620-33.

Acta Orthopaedica 2021; 92 (6): 681–688

Zahiri C A, Schmalzried T P, Szuszczewicz E S, Amstutz H C. Assessing activity in joint replacement patients. J Arthroplast 1998; 13(8): 890-5.

Acta Orthopaedica 2021; 92 (6): 689–694

689

Resident training does not influence the complication risk in total knee and hip arthroplasty Daphne M BRON 1, Nienke WOLTERBEEK 1, Rudolf W POOLMAN 2,3, Diederik H R KEMPEN 2, and Diyar DELAWI 1 1 Department 3 Department

of Orthopedic Surgery, St. Antonius Hospital, Nieuwegein; 2 Department of Orthopedic Surgery, JointResearch OLVG, Amsterdam; of Orthopedic Surgery, LUMC, Leiden, The Netherlands Correspondence: n.wolterbeek@antoniusziekenhuis.nl Submitted 2021-07-01. Accepted 2021-08-18.

Background and purpose — Gaining experience in the surgery room during residency is an important part of learning the skills needed to perform arthroplasties. However, in practice, patients are often not fully comfortable with trainee involvement in their own surgery. Therefore, we investigated complications, revision rates, mortality, and operative time of orthopedic surgeons and residents as primary surgeon performing total knee arthroplasties (TKAs) or total hip arthroplasties (THAs). Patients and methods — In this multi-center retrospective cohort study, 3,098 TKAs and 4,027 THAs performed between 2007 and 2013 were analyzed. Complications, revisions, mortality, and operative time were compared for patients operated on by the orthopedic surgeon or a resident as primary surgeon. An additional analysis was performed to determine whether the complication risk was affected by the postgraduate year of the resident. Results — Orthopedic complication rates were similar (TKA: orthopedic surgeon: 10%, resident: 11%; THA: 9% and 8%), revision rates (TKA: 3% and 2%, THA: 3% and 2%), or mortality rates (TKA: 0.1% and 0.3%, THA: 0.2% and 0.3%). For both procedures a higher non-orthopedic complication rate was found in the resident group (TKA: 8% and 10%; p = 0.03, THA: 8% and 10%; p = 0.01) and a slightly longer operative time (TKA: mean difference 9.0 minutes (8%); THA: 11.3 minutes (11%)). Interpretation — Complications, revisions, and mortality were similar in TKAs or THAs performed by the resident as primary surgeon compared with surgeries performed by an orthopedic surgeon. This data can be used in teaching hospitals and may help to reassure patients.

In teaching hospitals, surgeries are performed by a surgeon as primary surgeon or by a resident under direct supervision of a surgeon. Gaining experience in the surgical room during orthopedic residency is an important part of learning the skills needed to perform total hip (THA) and total knee arthroplasty (TKA). However, in practice, patients are often not fully comfortable with trainee involvement in their own surgery (Nahhas et al. 2019). A number of studies have been performed to evaluate the impact of resident involvement on complications following orthopedic surgical procedures (Schoenfeld et al. 2013, Edelstein et al. 2014, Haughom et al. 2014a, 2014b, Pugely et al. 2014, Cvetanovich et al. 2015, Bao et al. 2018). Most of these studies have used the same retrospective American National Surgical Quality Improvement Program database that collected data only up to 30 days postoperatively. Therefore, no implant-related complications such as dislocations or revisions after this time period were captured. Furthermore, the role of the resident during the procedure is missing in up to 79% of the surgeries (Edelstein et al. 2014, Haughom et al. 2014b, Cvetanovich et al. 2015, Neuwirth et al. 2018). This limits the level of evidence of the studies using this database, therefore conclusions are not reliable as to whether resident involvement in orthopedic surgical procedures is a risk factor for complications. Apart from these database studies, other studies on the effect of resident involvement were limited by small sample size or short-term follow-up. Therefore, we investigated complications, revision rates, mortality, and operation time of operations performed by an orthopedic surgeon or an orthopedic resident as primary surgeon in a large patient cohort validated with the national arthroplasty registry with over 95% completeness. We hypothesized that no clinically relevant difference in complications, revision rates, mortality, and operation time between orthopedic surgeon and orthopedic residents would be found.

690

Acta Orthopaedica 2021; 92 (6): 689–694

Total arthroplasties n = 7,123

Patients and methods This multi-center retrospective cohort study was conducted at 2 highly experienced teaching hospitals in the Netherlands. All patients who received a primary THA or TKA, in the orthopedic departments between January 2007 and December 2013, were included. There were no exclusion criteria. All individual patient records were reviewed and used to extract data. Recorded outcomes were complications and revisions during the complete follow-up and mortality within the first 90 days postoperatively. Complications were scored as either surgical site or systemic and were divided into subgroups for different kinds of complications. A patient with more than 1 complication could be scored in more than 1 subgroup. Collected risk factors were information on the surgeon (resident, postgraduate years of resident), duration of the surgery, age, sex, BMI, diagnosis, operation side, type of anesthesia, diabetes mellitus (DM), smoking history, and ASA score. The duration of the surgery was defined as the time between first incision until closure of the wound. All residents already had 2 years of mandatory pre-training in the general surgery department before they started with their orthopedic training, which took another 4 years. The postgraduate year (level of experience) of the resident was defined by the years in orthopedic residency so far, to determine whether an increase in experience of a resident would change the complication rates. To monitor resident training, the Entrustable Professional Activity concept is used, which allows surgeons to make competency-based decisions on the level of supervision required by residents. Four levels are used which are: assisting, operating under strict supervision, operating under limited supervision, and operating independently. Operating independently in arthroplasty is only allowed for final-year residents. Data was crosschecked with the complication registries of the departments and with the data in the Dutch Arthroplasty Register. This register has over 95% completeness (van Steenbergen et al. 2015). Statistics All statistical analyses were performed with SPSS (IBM SPSS statistics, version 24; IBM Corp, Armonk, NY, USA). Unpaired 2-tailed t-tests were used to compare continue baseline data. Categorical data were compared by chi-square tests, or if not allowed Fisher’s exact tests. Binary logistic regression models for surgical site and systemic complications were created using the variables age, sex, BMI, diabetes mellitus, ASA score, primary surgeon, smoking, type of anesthesia, and operation time. The same variables were used for the THA and TKA models, except for diagnosis, which was included only in the THA model. To determine whether experience level affects the complication risk within the resident group all resident cases were

Total knee arthroplasties n = 3,094

Orthopedic surgeon n = 2,218 (72%)

Resident n = 876 (28%)

Total hip arthroplasties n = 4,029

Orthopedic surgeon n = 2,878 (71%)

Resident n = 1,151 (29%)

Figure 1. Flowchart of the number of total arthroplasties in each group.

subdivided into 4 groups of increasing resident experience: first, second, third, and fourth postgraduate year. For both surgical site and systemic complications, these experience levels were compared using a chi-square test. Additionally, logistic regression analyses were done where every year in residency (risk factor) was compared with the complication rates of the orthopedic surgeon. All reported p-values were 2-tailed and for each analysis, p < 0.05 was considered significant. Odds ratios (OR) and 95% confidence intervals (CI) are reported. Ethics, funding, data sharing, and potential conflict of interest The study was conducted in accordance with the Helsinki Declaration and was approved by the Institutional Review Boards of both hospitals (trialregister.nl: NL8672). Individual consent was not required. No grants were received for this study. The data that supports the findings of this study is available from the corresponding author on reasonable request. The authors declare no competing interests.

Results 7123 total joint arthroplasties (TJA) (3,094 TKAs and 4,029 THAs) were included with a mean follow-up time of 97 months (52–136). In the TKA group 72% of patients were operated on by the orthopedic surgeon as primary surgeon and 28% by the orthopedic resident under supervision of the orthopedic surgeon. In the THA group 71% of patients were operated by the orthopedic surgeon and 29% by the resident (Figure 1). All THAs were performed using a posterolateral approach or straight lateral (Hardinge) approach. The fixation technique was uncemented, cemented, or hybrid. For the TKAs a medial parapatellar approach was used and all TKAs were cemented. In all patients, a low-vacuum wound drain was placed and removed after 24 hours. Total knee arthroplasty There were no relevant differences in age, operation side, diagnosis, DM, smoking and ASA score between groups (Table 1). There was a statistically significant difference in BMI; patients in the resident group had a slightly higher

Acta Orthopaedica 2021; 92 (6): 689–694

691

Table 1. Baseline characteristics of the patients included. For continuous data the mean and standard deviation (SD) are given, absolute and relative frequencies (%) are given for categorical data

Total knee arthroplasty (n = 3,094) Total hip arthroplasty (n = 4,029) Orthopedic Orthopedic surgeon Resident surgeon Resident (n = 2,218) (n = 876) p-value (n = 2,878) (n = 1,151) p-value

Age (SD) 68 (10) 69 (9) 0.2 66 (11) 71 (9) < 0.001 BMI (SD) 29 (5) 30 (5) 0.002 27 (5) 27 (5) 0.8 Sex 0.001 < 0.001 Male 693 (31) 222 (25) 1062 (37) 347 (30) Female 1,522 (69) 650 (75) 1,814 (63) 804 (70) Operation side 0.4 0.3 Right 1,162 (53) 446 (51) 1,587 (55) 615 (53) Left 1,052 (48) 430 (49) 1,286 (45) 536 (47) Diagnosis 0.1 0.04 Osteoarthritis 2,172 (99) 866 (99) 2,551 (89) 1,049 (92) Other 34 (2) 6 (1) 305 (11) 98 (9) Type of fixation < 0.001 Cemented 2,216 (100) 875 (100) – 1,275 (45) 775 (67) Uncemented – – 872 (31) 201 (18) Hybrid – – 404 (14) 171 (15) Resurfacing – – 277 (10) 2 (0) Diabetes mellitus 0.2 0.1 No 1,638 (84) 667 (82) 2,266 (91) 953 (89) Yes 310 (16) 147 (18) 225 (9) 118 (11) Smoking 0.9 0.9 No 1,657 (87) 704 (88) 1,,994 (82) 863 (82) Yes 243 (13) 101 (13) 439 (18) 187 (18) Anesthesia < 0.001 0.3 General 774 (36) 237 (28) 1949 (70) 753 (68) Regional 1,356 (64) 611 (72) 851 (30) 358 (32) ASA score 0.05 < 0.001 I 406 (21) 139 (17) 729 (29) 217 (20) II 1,307 (67) 560 (69) 1,480 (60) 703 (66) III–IV 236 (12) 114 (14) 269 (11) 152 (14)

Table 2. Complication rates (%) and Odds ratio with 95% confidence interval (CI) of orthopedic surgeons and residents for total knee arthroplasty (N = 3,094) Orthopedic surgeon Resident OddsComplications (n = 2,218) (n = 876) ratio (CI) Surgical site Deep infection Nerve palsy Intraoperative Reoperation Other Revision Systemic Delirium DVT or PE Pulmonary Urological Cardiac Gastrointestinal tract Cerebrovascular Other Death within 90-days

222 (10) 45 (2) 12 (0.5) 10 (0.5) 182 (8) 140 (6) 71 (3) 175 (8) 28 (1) 15 (0.7) 11 (0.5) 67 (3) 27 (1) 9 (0.4) 10 (0.5) 36 (2) 2 (0.1)

99 (11) 13 (2) 5 (0.6) 5 (0.6) 83 (10) 69 (8) 29 (3) 91 (10) 19 (2) 7 (0.8) 2 (0.2) 42 (5) 16 (2) 3 (0.3) 1 (0.1) 16 (2) 3 (0.3)

1.1 (0.9–1.5) 0.7 (0.4–1.4) 1.1 (0.4–3.0) 1.3 (0.4–3.7) 1.2 (0.9–1.5) 1.3 (0.9–1.7) 1.0 (0.7–1.6) 1.4 (1.0–1.8) 1.7 (1.0–3.1) 1.2 (0.5–2.9) 0.5 (0.1–2.1) 1.6 (1.1–2.4) 1.5 (0.8–2.8) 0.8 (0.2–3.1) 0.3 (0.0–2.0) 1.1 (0.6–2.0) 3.8 (0.6–23)

DVT or PE: deep vein thrombosis or pulmonary embolism.

BMI (30) than patients in the orthopedic surgeon group (29). In the residents’ group 72% of the surgeries were performed under regional anesthesia versus 64% in the orthopedic surgeon group (p < 0.001). The duration of surgery was longer in the resident group (118 [SD 27] minutes) compared with the orthopedic surgeons (109 [SD27] minutes) with a mean difference of 9 minutes between groups (p < 0.001).

Complications, revisions, and mortality 10% of the orthopedic surgeon group had at least 1 surgical site complication compared with 11% of the resident group (Table 2). Revision rates in both groups were 3%. There were 0.9% (peri-prosthetic) fractures in the resident group versus 0.2% in the surgeon group. There was a significant difference in systemic complications with 8% in the orthopedic surgeon group versus 10% in the resident group (p = 0.03). The difference in systemic complications was mainly seen in the number of urological complications, which consists of urinary retention and urinary tract infection. A urological complication was noted in respectively 3% versus 5% (p = 0.03). In all other subgroups, no differences were found. In the resident group, 0.2% of pneumonias were recorded versus 0.5% in the surgeon group. In the orthopedic surgeon group, 2 patients died within 90 days after surgery (1 patient with sudden death of unknown cause and 1 due to a brain stem cerebrovascular accident). In the resident group, in total, 3 patients died within 90 days after surgery (2 patients with sudden death of unknown cause and 1 from intestinal ischemia due to sepsis as a result of an infected total knee arthroplasty). Independent risk factors for complications Multivariate logistic regression analyses were performed to identify potential bias due to independent risk factors for both surgical site and systemic complications (Tables 3 and 4, see Supplementary data). Similar risks of complications were found in cases where the resident was the first surgeon. For surgical site complications, a higher age was associated with a reduction in the likelihood of exhibiting a complication (OR 1.0 for every year increase in age, p < 0.001) while for systemic complications increasing age predicted a higher complication rate (OR 1.1 for every year increase in age, p < 0.001). In addition, for surgical site complications ASA score was found to be an independent risk factor, where a higher ASA score predicted a higher complication rate (Table 3, see Supplementary data). Additional independent risk factors for

692

Acta Orthopaedica 2021; 92 (6): 689–694

Table 5. Complication rates (%) and Odds ratio with 95% confidence interval (CI) of orthopedic surgeons and residents for total hip arthroplasty (N = 4,029) Orthopedic surgeon Resident OddsComplications (n = 2,878) (n = 1,151) ratio (CI) Surgical site 248 (9) Deep Infection 50 (2) Dislocation 68 (2) Nerve palsy 25 (0.9) Intraoperative 29 (1) Reoperation 151 (5) Other 77 (3) Revision 69 (2) Systemic 216 (8) Delirium 50 (2) DVT or PE 12 (0.4) Pulmonary 23 (0.8) Urological 82 (3) Cardiac 41 (1) Gastrointestinal tract 12 (0.4) Cerebrovascular 6 (0.2) Other 33 (1) Death within 90-days 6 (0.2)

91 (8) 22 (2) 27 (2) 8 (0.7) 10 (0.9) 47 (4) 26 (2) 17 (2) 116 (10) 22 (2) 8 (0.7) 12 (1) 48 (4) 19 (2) 7 (0.6) 4 (0.3) 21 (2) 3 (0.3)

0.9 (0.7–1.2) 1.1 (0.7–1.8) 1.0 (0.6–1.6) 0.8 (0.4–1.8) 0.9 (0.4–1.8) 0.8 (0.6–1.1) 0.8 (0.5–1.3) 0.6 (0.4–1.0) 1.4 (1.1–1.8) 1.1 (0.7–1.8) 1.7 (0.7–4.1) 1.3 (0.6–2.6) 1.5 (1.0–2.1) 1.2 (0.7–2.0) 1.5 (0.6–3.7) 1.7 (0.5–5.9) 1.6 (0.9–2.8) 1.3 (0.3–5.0)

DVT or PE: deep vein thrombosis or pulmonary embolism.

systemic complications included sex: females had a lower complication rate. Diabetes and ASA score III/IV predicted a higher complication rate (Table 4, see Supplementary data). Total hip arthroplasty Complications, revisions, and mortality In the orthopedic surgeon group, 9% suffered at least 1 surgical site complication and in the resident group 8% (Table 5). There were 0.9% (peri-prosthetic) fractures in the resident group versus 0.4% in the surgeon group. No statistically significant differences were found when all surgical site complications were divided into subgroups. However, there were fewer systemic complications in the orthopedic surgeon group (8%) compared with the resident group (10%). Urinary tract complications (urinary retention and urinary tract infection) were more common in the resident group. In the resident group, 1% of pneumonias were recorded versus 0.6% in the surgeon group. 6 patients died in the first 90 days after surgery in the orthopedic surgeon group (1 patient with a rectus hematoma with abdominal compartment syndrome and renal failure; 1 sepsis possibly due to endocarditis; 1 outof-hospital death of unknown cause; 1 herniation syndrome due to undiagnosed brain tumor; 1 major bleeding from severe liver cirrhosis and bleeding disorder; and 1 myocardial infarction by hemoglobin decrease in a patient with a bleeding disorder and liver cirrhosis). 3 patients died in the resident group (1 patient with massive abdominal bleeding; 1 myocardial infarction; and 1 sepsis due to early THA infection). This difference in mortality rate was not significantly different (p = 0.7).

Table 8. Overall surgical site and systemic complication rates for total arthroplasty in the resident group split up for years in training compared to the surgeries performed by the orthopedic surgeon (N = 5,099) Experience level

Surgical site complication rate c Year 1 330 Year 2 482 Year 3 352 Year 4 859 Systemic complication rate d Year 1 330 Year 2 482 Year 3 352 Year 4 859

B (SE) a

OR (CI) b

–0.05 (0.2) –0.12 (0.2) 0.12 (0.2) 0.08 (0.1)

1.0 (0.6–1.4) 0.9 (0.6–1.3) 1.1 (0.8–1.6) 1.1 (0.9–1.4)

0.32 (0.2) 0.14 (0.2) 0.12 (0.2) 0.47 (0.1)

1.4 (1.0–2.0) 1.2 (0.8–1.6) 1.1 (0.8–1.7) 1.6 (1.3–2.0)

a B (SE): unstandardized regression weight with standard b OR (CI): odds ratio with 95% confidence interval c R2 < 0.001 (Nagelkerke). Model χ2(4) = 1.4, p = 0.8. d R2 = 0.005 (Nagelkerke). Model χ2(4) = 16.5, p = 0.002.

error.

Risk factors There were no relevant differences in BMI, DM, smoking, or type of anesthesia (Table 1). There was a statistically significant difference in age, sex, diagnosis, and the type of fixation. The difference in type of fixation can be explained by the fact that the orthopedic surgeon placed more uncemented THAs and almost all resurfacing prostheses. A significant difference was found in the ASA score, where the resident operated on more ASA II and ASA III patients. The duration of the surgery was statistically significantly longer in the resident group (115 [28] minutes) than in the group with orthopedic surgeons (103 [28] minutes) with a mean difference of 11 minutes between groups. Independent risk factors for complications The risk of surgical site and systemic complications was similar for the residents and orthopedic surgeons (Tables 6 and 7, see Supplementary data). For surgical site complications a longer operative time was found as an independent risk factor (OR 1.0 for each minute’s increase in operative time) and ASA score III– IV seemed to be a predictive factor, although significance was not achieved (OR 1.6, p = 0.06). For systemic complications independent risk factors included higher age, smoking, and a higher ASA score. In addition, male sex was an independent risk factor as women had a lower complication rate (OR 0.7). Experience level of the resident The TKA and THA group were analyzed together to maximize the group sizes. Odds ratios were calculated for all postgraduate years compared with the orthopedic surgeon group. The logistic regression model for surgical site complications was not statistically significant (Table 8). For systemic complications, patients operated on by residents who were 4 years in training were 1.6 times more likely to exhibit a systemic complication compared with patients operated on by the orthopedic surgeon. For all other postgraduate years, no significant

Acta Orthopaedica 2021; 92 (6): 689–694

Complication rate (%) 20

First year recident Second year recident Third year recident Fourth year recident Orthopedic surgeon

TKA surgical site TKA systemic THA surgical site THA systemic complication complication complication complication

Figure 2. Complication rates after total knee arthroplasty (TKA) and total hip arthroplasty (THA), stratified by experience level of the resident compared with orthopedic surgeons.

odds ratios were found. This model was statistically significant but explained only 0.5% of the variance in systemic complications (Table 8). No clinical differences in, respectively, surgical site and systemic complication rates were found when stratified by years in training (Figure 2).

Discussion Similar surgical site complications, revision rates, and mortality were found between surgeries performed by the orthopedic surgeon or resident as primary surgeon. However, more systemic complications were noted in the resident group for both THAs and TKAs. This seems to be caused by a larger number of urological complications. Residents operated on more patients with a higher ASA score and a higher ASA score was found to be an independent risk factor for systemic complications for both TKAs and THAs. Furthermore, THA patients in the resident group were older and a higher age in both procedures was found to be an independent risk factor for systemic complications. We demonstrated a prolonged operative time in the resident group for both TKAs and THAs. The mean difference in operative time of 9 minutes (8%) for TKA and 11 minutes (11%) for THA were slightly less than found in previous studies, where prolonged operative times between 9% and 15% were mentioned for TKA procedures and between 17% and 22% for THA procedures when a resident was involved during surgery (Haughom et al. 2014a, 2014b). This might be explained by the residency program as Dutch orthopedic residents already have 2 years of surgical experience after fulfilling a 2-year mandatory general surgery residency program.

693

Experience level of the resident We found no relevant differences in complication rates between different levels of experience within the resident group. Only the systemic complication rate in the 4th-year resident group was higher but it explained only 0.5% of the variance in systemic complications. These results are in line with previous studies (Haughom et al. 2014a, 2014b). However, in these earlier studies, the role of the resident during surgery was not clear and the sample size was not reported or very small with only 53 patients in the junior resident group. Independent risk factors for complications We found that resident involvement was not a risk factor for either surgical site or systemic complications. For TKAs our results showed that higher age was associated with a decreased risk of surgical site complications, while a higher ASA score predicted an increased risk. For systemic complications, independent risk factors were male sex, DM, ASA III/IV, and higher age. Our results are partly in line with the results found in the literature (Haughom et al. 2014a). In the literature, increased operative time is found to be an independent risk factor for TKAs, however, this could not be confirmed in our study. Additionally, in the literature, higher age is indicated as a risk factor for complications (Pugely et al. 2013). This is confirmed only for systemic complications. For THAs a longer operative time was associated with a higher risk of surgical site complications. For systemic complications, higher age, male sex, smoking, and a higher ASA score were found to be independent risk factors. Male sex as an independent risk factor has not been found or noted in previous literature. A risk factor that has been found in literature, but not in our study, was obesity (Haughom et al. 2014b). This may be explained by the fact that that specific study included BMI as categorical variable instead of continuous. Despite the fact that the operative time in the resident group was longer and a longer operative time was associated with a higher risk of surgical site complications, no more surgical site complications were found. This is an important point for teaching hospitals. In our study, the resident was the first surgeon in only 29% of the TJA procedures. Given the fact that no more complications were found, it may be considered to have more TJA procedures performed by the resident, with due regard for the prolonged operative time. This may benefit the training of surgical skills during residency. Strengths and limitations In our study each individual patient chart was reviewed, and data was crosschecked with the complication registries of the departments and with the data in the Dutch Arthroplasty Register. Furthermore, each surgical report was reviewed to ascertain the role of the resident during surgery. The study has a large sample size and a long follow-up. As a result, a reliable number of complications have been captured and reliable conclusions can be drawn concerning the complication risk and risk of early

694

revisions. National revision rates are in the same ranges as the revision rates found in this study and the distribution of ASA score is more or less equal to the national distribution. Limitations might be the retrospective design of the study. Further, there is probably a selection bias due to the fact that the orthopedic surgeon will operate on the more difficult patients. Also, patients might have expressed their preference or might have refused a resident as primary surgeon. This has not been recorded in the patient files. No information is available on minimal clinically significant differences regarding differences in complications. A non-significant difference might be experienced as a clinically relevant difference by patients. In future research, preoperative radiographs could be used to classify the surgical difficulty of the operation to clarify possible selection bias. In addition, other results could be studied, such as patient-reported outcome measures, to analyze the effects of resident involvement. Conclusion Surgical site complications, revisions, and mortality were similar in TKAs or THAs performed by the resident as primary surgeon compared with surgeries performed by an orthopedic surgeon. This data can be used in teaching hospitals and may help to reassure patients. Supplementary data Tables 3–4 and 6–7 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2021.1979296

DB and NW had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors were involved with the concept and design, acquisition, analysis, and interpretation of data. Statistical analysis and drafting of the manuscript: DB and NW. Critical revision of the manuscript for important intellectual content: all authors. Supervision: RP, DK, DD

Acta Orthopaedica 2021; 92 (6): 689–694

Acta thanks Aare Märtson and Petri Virolainen for help with peer review of this study.

Bao M H, Keeney B J, Moschetti W E, Paddock N G, Jevsevar D S. Resident participation is not associated with worse outcomes after TKA. Clin Orthop Relat Res 2018; 476: 1375-90. Cvetanovich G L, Schairer W W, Haughom B D, Nicholson G P, Romeo A A. Does resident involvement have an impact on postoperative complications after total shoulder arthroplasty? An analysis of 1382 cases. J Shoulder Elbow Surg 2015; 24(10): 1567-73. Edelstein A I, Lovecchio F C, Saha S, Hsu W K, Kim J Y. Impact of resident involvement on orthopaedic surgery outcomes: an analysis of 30,628 patients from the American College of Surgeons national surgical quality improvement program database. J Bone Joint Surg Am 2014; 96(15): e131. Haughom B D, Schairer W W, Hellman M D, Yi P H, Levine B R. Does resident involvement impact post-operative complications following primary total knee arthroplasty? An analysis of 24,529 cases. J Arthroplasty 2014a; 29(7): 1468-72.e2. Haughom B D, Schairer W W, Hellman M D, Yi P H, Levine B R. Resident involvement does not influence complication after total hip arthroplasty: an analysis of 13,109 cases. J Arthroplasty 2014b; 29(10): 1919-24. Nahhas C R, Yi P H, Culvern C, Cross M B, Akhavan S, Johnson S R, Nunley R M, Bozic K J, Della Valle C J. Patient attitudes toward resident and fellow participation in orthopedic surgery. J Arthroplasty 2019; 34(9): 1884-8.e5. Neuwirth A L, Stitzlein R N, Neuwirth M G, Kelz R K, Mehta S. Resident participation in fixation of intertrochanteric hip fractures: analysis of the NSQIP database. J Bone Joint Surg Am 2018; 100(2): 155-64. Pugely A J, Martin C T, Gao Y, Mendoza-Lattes S, Callaghan J J. Differences in short-term complications between spinal and general anesthesia for primary total knee arthroplasty. J Bone Joint Surg Am 2013; 95(3): 193-9. Pugely A J, Gao Y, Martin C T, Callagh J J, Weinstein S L, Marsh J L. The effect of resident participation on short-term outcomes after orthopaedic surgery. Clin Orthop Relat Res 2014; 472(7): 2290-2300. Schoenfeld A J, Serrano J A, Waterman B R, Bader J O, Belmont P J Jr. The impact of resident involvement on post-operative morbidity and mortality following orthopaedic procedures: a study of 43,343 cases. Arch Orthop Trauma Surg 2013; 133(11): 1483-91. van Steenbergen L N, Denissen G A, Spooren A, van Rooden S M, van Oosterhout F J, Morrenhof J W, Nelissen R G H H. More than 95% completeness of reported procedures in the population-based Dutch Arthroplasty Register. Acta Orthop 2015; 86(4): 498-505.

Acta Orthopaedica 2021; 92 (6): 695–700 695

Postoperative delirium is a risk factor for complications and poor outcome after total hip and knee arthroplasty Matthias MEYER 1, Julia GÖTZ 1, Lukas PARIK 1, Tobias RENKAWITZ 2, Joachim GRIFKA 1, Günther MADERBACHER 1, Tobias KAPPENSCHNEIDER 1, and Markus WEBER 1 1 Department

of Orthopaedic Surgery, Regensburg University Hospital, Bad Abbach, Germany; 2 Heidelberg University Orthopedic Hospital, Heidelberg, Germany Correspondence: matthias.meyer@ukr.de Submitted 2021-05-17. Accepted 2021-08-19.

Background and purpose — Improving health care and demographic change have resulted in a steady increase in geriatric patients undergoing total hip (THA) and knee (TKA) arthroplasty. Postoperative delirium (POD) is a frequent and severe complication after major surgery. Therefore, we analyzed the impact of POD on outcome after THA and TKA. Patients and methods — In a consecutive series of 10,140 patients who had undergone elective THA or TKA between 2011 and 2020, rates of reoperation within 90 days, readmission within 90 days, complications, and responder rate as defined by the OMERACT-OARSI criteria were compared between patients with and without POD. Multivariable logistic regression models were used to assess the relationship between POD and other postoperative complications. Results — Patients with POD showed higher rates of reoperation (12% vs. 5%), readmission (15% vs. 5%), surgical complications (7% vs. 2%), non-surgical complications (8% vs. 4%), Clavien–Dindo IV° complications (10% vs. 2%) and transfusion (14% vs. 2%). POD led to lower responder rate (76% vs. 87%) 1 year after total joint replacement. All previous comparisons statistically significant. Multivariable logistic regression analyses revealed POD as an independent risk factor for reoperation (OR = 2; CI 1–3), readmission (OR = 2; CI 2–4) and Clavien–Dindo IV° complications (OR = 3; CI 2–5). Interpretation — POD is a serious problem in elective joint replacement. Affected patients suffer more complications and show poor patient-reported outcome 1 year postoperatively. Systematic prevention strategies and standardized therapy protocols are mandatory to avoid burden to patients and healthcare providers.

Numbers of primary total hip and knee arthroplasty (THA and TKA) are projected to grow 71% and 85%, respectively, by 2030 (Sloan et al. 2018). Demographic change as well as improvement in general living standards, health care, nutrition, and education result in a steady increase in geriatric patients undergoing major surgery, such as THA and TKA (United Nations et al. 2017). With an incidence of 0.7% to 2.2%, postoperative delirium (POD) is a common complication in patients undergoing elective THA and TKA (Petersen et al. 2017, Aziz et al. 2018, Weinstein et al. 2018, Yang et al. 2020). POD is considered to be a potent risk factor for adverse events, such as prolonged length of stay and discharge to nursing facility (Petersen et al. 2017, Aziz et al. 2018). Former studies showed that patients with POD had up to 3-fold increased length of stay and complication rates (Petersen et al. 2017, Aziz et al. 2018). Although delirium has been described in the medical literature for more than 2 millennials, especially hypoactive forms of POD are still frequently not recognized and managed appropriately in daily practice (Inouye et al. 2014). However, with the spread of orthogeriatric care models the clinical problem of postoperative delirium is being paid increasing attention. Elderly patients with hip fractures are especially vulnerable to POD, leading to poor functional recovery (Marcantonio et al. 2000). While patients with fragility fractures can already profit from interdisciplinary orthogeriatric care models, the topic still seems underrepresented in elective orthopedic surgery. Studies dealing with the relationship of POD to other complications after elective THA and TKA are rare (Aziz et al. 2018) and, to the best of our knowledge, no study has previously evaluated the impact of POD on patient-reported outcome measures (PROMs) after THA or TKA.

696

Therefore, we retrospectively evaluated the impact of postoperative delirium on complications and patient-reported outcome in a consecutive series of 10,140 patients who had undergone primary elective THA or TKA at a high-volume arthroplasty center. We hypothesized that patients who exhibit POD have higher rates of complications and worse PROM than patients without POD.

Patients and methods Study design and study population This is a retrospective study based on a database derived from the hospital information system and the department’s joint registry. From the database, all patients who had undergone primary elective THA and TKA between 2011 and 2020 were included, representing a consecutive series. As capture of patient-reported outcome measures (PROM) did not start until the establishment of a certified arthroplasty center in October 2012 and some patients were lost to follow-up 1 year postoperatively, PROMs were only available for a subgroup of patients. Reoperation within 90 days, readmission within 90 days, complications, and transfusion were analyzed. Complications were categorized into surgical (major bleeding with need for transfusion, periprosthetic fracture, wound healing disorder, wound infection, dislocation) and non-surgical (myocardial infarction, decompensated heart failure, cardiac arrhythmias, pneumonia, renal failure, urinary tract infection, collapse, thrombosis, pulmonary embolism, cerebrovascular accident). Furthermore, complications were categorized according to the Clavien–Dindo classification (Dindo et al. 2004). This classification system ranks complications in 5 grades, based on the therapy used for correction. Any deviation from the normal postoperative course without the need for pharmacological treatment or surgical, endoscopic, and radiological intervention represents a Grade I complication. Grade II complications require specific pharmacological treatment, whereas Grade III complications result in surgical, endoscopic, or radiological intervention. Grade IV complications are defined as life-threatening events requiring intensive care management. Grade V represents the death of a patient (Dindo et al. 2004). Follow up for the parameter “surgical complication” was limited to 30 days postoperatively. Non-surgical and Clavien–Dindo IV° complications could only be captured during hospital stay (mean 9 days postoperatively). Surgical techniques Indications for surgery were primary or secondary end-stage hip or knee osteoarthritis. All operations were performed in a single Department of Orthopedic Surgery at a University Medical Center. All patients received the same standardized treatment protocol for THA or TKA respectively. THA was performed under spinal anesthesia, whereas TKA was per-

Acta Orthopaedica 2021; 92 (6): 695–700

formed using anesthesia with perineural catheters. Only if there are contraindications for regional anesthesia or in the case of explicit patient desire was general anesthesia applied. Cementless THA was performed in the lateral decubitus position. A minimally invasive anterolateral approach was used. Cemented TKA was performed through a medial parapatellar approach. No patellar resurfacing was performed. Postoperative pain management included non-steroidal anti-inflammatory drugs and retarded opioids if needed. Mobilization was started on the first day following surgery under full or partial weight-bearing, according to the instructions of the orthopedic surgeon. Data collection (Figure 1, see Supplementary data) Diagnoses coded at the time of hospitalization and discharge were extracted from the hospital information system (ORBIS; Agfa Healthcare, Mortsel, Belgium) including corresponding ICD-10 codes. Diagnostic codes had been entered by professional clinical coders and were double-checked by physicians using information gathered from patients’ medical records. Postoperative delirium was coded when the Nursing Delirium Screening Scale (Nu-DESC) was greater or equal to 2 as proposed by Gaudreau et al. (2005). The Nu-DESCs were assessed daily by skilled nursing staff. Complications were assessed according to the ICD-10 codes at the time of discharge. Further available data from our clinical information system were age, sex, operative procedure, length of stay, transfusion, transfer to intensive care unit, reoperation, and readmission. Patientreported outcome were extracted from the department’s joint registry. The Western Ontario and McMaster Universities Arthritis index (WOMAC) was assessed preoperatively and 1 year postoperatively. Responders were differentiated from non-responders by means of the criteria of the Outcome Measures in Rheumatology and Osteoarthritis Research Society International consensus (OMERACT-OARSI) (Pham et al. 2004). The OMERACT-OARSI criteria identify patients as responders after THA or TKA if the WOMAC index shows an improvement in pain or function, either relatively by at least 50% or absolutely by at least 20 points. Alternatively, patients are defined as responders if 2 of the following criteria are met: Improvement of the pain subscore by at least 20% and at least 10 points, improvement of the function subscore by at least 20% and at least 10 points, or improvement in the global index by at least 20% and at least 10 points (Pham et al. 2004). Statistics Group comparisons were performed by 2-sided t-tests. Absolute and relative frequencies were given for categorical data and compared between groups by chi-square tests. The hypotheses of the study were tested on 5% significance level. Multivariable logistic regression analyses were conducted to provide risk factor estimates adjusted for confounding bias in terms of reoperation, readmission, and Clavien–Dindo IV° complications while controlling for other variables known to

Acta Orthopaedica 2021; 92 (6): 695–700 697

Table 1. Characteristics of study group and subgroup with available patient-reported outcome measures (PROMs). Values are percentage unless otherwise specified Demographics

Study Non-POD POD PROM group cohort cohort subgroup n = 10,140 n = 10,001 n = 139 n = 4,189

Age a 66 (11) 66 (11) 79 (7) 66 (10) Female sex 58 58 53 56 Total hip arthroplasty 55 57 58 55 Operative time, minutes a 77 (30) 76 (30) 81 (32) 75 (34) Length of stay, days a 9 (4) 9 (4) 10 (6) 9 (2) ASA classification 1 13 13 2 15 ASA classification 2 57 57 43 58 ASA classification 3 31 30 54 27 ASA classification 4 0.3 0.3 0.7 0.2 CCI a 0.5 (0.9) 0.4 (0.9) 2 (1) 0.4 (0.8) Dementia 0.1 0.1 1 0 Cerebrovascular disease 2 2 7 2 a

Values are mean (standard deviation). ASA = American Society of Anesthesiologists. CCI = Charlson Comorbidity Index.

be associated with adverse surgical outcomes. According to the literature, surgery site knee, long operative time, male sex, increasing age, and high ASA classification are known risk factors for complications after THA and TKA (Weber et al. 2018). Adjustment of covariates was performed based on considerations regarding cause–effect. The 6-step approach was used to prevent adjustment bias (Shrier and Platt 2008). The assumed cause–effect relation is shown in Figure 2 (see Supplementary data). Confidence level was defined at 95% and is presented as confidence interval (CI). IBM SPSS Statistics 25 (IBM Corp, Armonk, NY, USA) was used for analysis. Ethics, funding, and potential conflicts of interest This study was approved by the Ethics Committee of the University Hospital Regensburg, Germany (20-1821-104). Informed consent was not necessary for this type of study. No funding was received. The authors have no conflicts of interest to declare.

Table 2. Complications after total hip or knee arthroplasty in patients with and without postoperative delirium. Values are count (%) Adverse event

Non-POD cohort n = 10,001

Reoperation within 90 days Readmission within 90 days Surgical complications Non-surgical complications Clavien–Dindo IV° Transfusion PROM subgroup Responder rate

503 (5.0) 495 (4.9) 215 (2.1) 370 (3.7) 177 (1.8) 185 (1.8) 4,139 3,602 (87)

POD cohort n = 139 p-value 17 (12) 21 (15) 10 (7) 11 (8) 14 (10) 20 (14) 50 38 (76)

< 0.001 < 0.001 < 0.001 0.009 < 0.001 < 0.001 0.02

POD = postoperative delirium.

Results There were 5,575 and 4,565 patients who underwent THA or TKA, respectively, during the study period. Patient-reported outcome measures up to 1 year postoperatively were available for a subgroup of 4,189 patients (Table 1). Incidences of POD in the study group and the PROM subgroup were 1.4% and 1.2%, respectively. Mean age in the POD cohort was 79 years (SD 6.8) versus 66 years (SD 11) in the non-POD cohort. The distribution of POD in the study group according to patient age is shown in Figure 3. Patients who exhibited POD showed increased rates of reoperation (12% vs. 5%, p < 0.001) and readmission (15% vs. 5%, p < 0.001) compared with patients without POD (Table 2; Figure 4). The rates for surgical, non-surgical, and Clavien–Dindo grade IV complications were higher in the POD cohort, compared with the non-POD cohort (p < 0.01). Transfusion rate was 13% higher in patients with POD compared with patients without POD (p < 0.001; Table 2). Patients with POD showed a decreased responder rate compared with patients without POD (76% vs. 87%; p = 0.02) 1 year postoperatively (Table 2).

Frequency of POD (%) 8

Reoperation Readmission

Clavien-Dindo IV° 4

Surgical complication

Patients without POD Patients with POD

Nonsurgical complication

Transfusion 0

< 50

50–60

60–70

70–80

> 80

Agegroups

Figure 3. Age distribution of postoperative delirium (POD) in patients undergoing primary elective THA or TKA.

Frequency of complications (%)

Figure 4. Rates of complications in patients with and without postoperative delirium (POD) after primary elective total hip or knee arthroplasty.

698

Acta Orthopaedica 2021; 92 (6): 695–700

challenging, as study groups are heterogenic and a variety of assessment tools are used (Bin Abd Razak and Yung 2015). Even higher incidences up to 42% are reported for patients undergoing THA because of hip fractures (Mosk et al. 2017). As pre-existing dementia, comorbidities, and age are wellVariable OR (CI) p-value known risk factors, it is not surprising that patients with fraReoperation gility fractures are especially prone to POD. In our study, mild Postoperative delirium 2.0 (1.1–3.4) 0.02 and especially hypoactive forms of POD might not have been Surgery site (knee) 0.9 (0.8–1.1) 0.3 recognized appropriately (Inouye et al. 2014). Consequently, Age per 5 years 1.1 (1.0–1.1) 0.04 ASA classification 1.6 (1.4–1.9) 0.001 the true incidence of POD after elective THA and TKA might Operative time per 15 minutes 1.3 (1.2–1.3) 0.001 be higher than assumed and could only be determined in a Sex (male) 0.9 (0.7–1.1) 0.1 prospective controlled setting. Readmission Postoperative delirium 2.4 (1.5–4.0) 0.001 Whereas incidences and risk factors of POD are well Surgery site (knee) 1.1 (0.9–1.4) 0.2 reported, less is known about the associations between POD Age per 5 years 1.1 (1.0–1.2) 0.001 and other postoperative complications after THA and TKA. ASA classification 1.6 (1.4–1.9) 0.001 Operative time per 15 minutes 1.2 (1.2–1.3) 0.001 In our study, reoperation and readmission rate within 90 days Sex (male) 0.9 (0.8–1.1) 0.6 after surgery for patients with POD was 2 to 3 times higher Clavien–Dindo IV° than for patients without POD. The proportion of patients with Postoperative delirium 2.7 (1.5–5.0) 0.001 Surgery site (knee) 0.9 (0.6–1.2) 0.4 surgical complications, non-surgical complications, Clavien– Age per 5 years 1.4 (1.3–1.5) 0.001 Dindo IV° complications, and transfusion was even 2 to 7 times ASA classification 1.8 (1.4–2.3) 0.001 higher in the cohort with POD. A recent retrospective analysis Operative time per 15 minutes 1.1 (1.0–1.2) 0.01 Sex (male) 0.8 (0.6–1.2) 0.3 also found a 2- to 3-fold increase of surgical complications and a 2- to 4-fold increase of non-surgical complications after ASA = American Society of Anesthesiologists. primary elective THA (Aziz et al. 2018). Multivariable logistic For assumed cause–effect relation please see Figure 2. regression analyses showed independent associations of POD with 90-day reoperation, 90-day readmission, and Clavien– The results largely held true for subgroup analyses of THA Dindo IV° complications, while controlling for age, ASA and TKA (Tables 3 and 4, see Supplementary data). classification, and operative time. Our results are confirmed by Aziz et al. (2018), who found that POD was independently Multivariable logistic regression analysis associated with major (OR 2.0; CI 1.7–2.4; p < 0.001) and Multivariable logistic regression analysis identified POD as minor (OR 2.0; CI 1.7–2.4; p < 0.001) complications, while independent risk factors for reoperation (OR = 2, CI 1–3), controlling for age, sex, and number of comorbidities. The readmission (OR = 2, CI 2–4), and Clavien–Dindo IV° com- data show that exhibition of POD favors postoperative complications (OR = 3, CI 2–5; Table 5). plications. However, a reverse effect might also be present as major complications, especially those requiring ICU management, put patients at higher risk of POD (Aldecoa et al. 2017). Overall, assessment of causal relationships between POD and Discussion other complications is complex (Figure 2, see Supplementary We found that patients with POD showed higher rates of all data). Unfortunately, our database provides no information captured complications after THA and TKA. Furthermore, concerning the temporal relation between documentation of patients who exhibited POD were more likely to be non- POD and occurrence of complications. Demographic analysis responders compared with patients without POD up to one of our study cohorts showed higher comorbidity burden and year postoperatively. Multivariable logistic regression analy- incidences of dementia or cerebrovascular disease in the POD sis identified POD as risk factor for reoperation, readmission, cohort (Table 1). Hence, the occurrence of POD and postoperand Clavien–Dindo IV° complications. In our study the inci- ative complications could also be interpreted as a consequence dence of POD was 1.4%. This is consistent with the existing of underlying comorbidity (Meyer et al. 2020). We tried to literature as recent retrospective analyses found incidences of investigate the independent effect of POD on complications POD from 0.68% to 2.21% (Aziz et al. 2018, Weinstein et al. after THA and TKA by use of multivariable regression analy2018, Yang et al. 2020). A prospective study of 6,331 elec- sis. Controlling for comorbidity measured by ASA score, POD tive THA and TKA patients aged 70 years or older found an was independently associated with complications after THA incidence of 0.7% for POD (Petersen et al. 2017). In 2015, a and TKA. Nevertheless, due to the complex relations of POD, meta-analysis of prospective studies described incidences of comorbidity, and postoperative complications, the results of POD from 0% to 10% after total joint arthroplasty and con- this study have to be interpreted with caution. To the best of cluded that reliable estimation of the incidence of POD is our knowledge, this is the first study to evaluate PROMs in Table 5. Multivariable analysis with odds ratio (OR) and 95% confidence interval (CI) for total effect of postoperative delirium on reoperation, readmission, and Clavien–Dindo IV° complications after total joint arthroplasty

Acta Orthopaedica 2021; 92 (6): 695–700 699

patients with POD after THA and TKA. The overall responder rate as defined by the OMERACT-OARSI criteria was 91% and 82%, respectively. This is consistent with recent literature, where responder rates of 86% to 93% for THA and 73% to 86% for TKA are described (Judge et al. 2010, Weber et al. 2018, Overgaard et al. 2019). We found a significantly worse responder rate in patients with POD compared with patients without POD. Comparative data are hardly available. Abelha et al. (2013) reported lower quality of life for patients who experienced POD during ICU treatment after major surgery at follow-up 6 months postoperatively. With regard to orthopedic surgery, Duppils and Wikblad (2000) found that patients with POD after hip fracture scored lower in PROMs at followup than those without delirium. Demographic analysis of our study cohorts showed higher age and increased comorbidity burden in the POD cohort (see Table 1). Hence, a possible influence of age and comorbidity on responder rate has to be taken in account. In a former study, comorbidity was independently associated with responder rate, whereas age was not (Weber et al. 2018). Although PROMs were available only for a subgroup of patients, we assume that the results of the subgroup are valid, as demographic characteristics of the study group and the subgroup with available PROMS were comparable (see Table 1). Furthermore, incidences of POD and overall responder rates in the subgroup were consistent with current literature (Aziz et al. 2018, Weber et al. 2018). Our findings underline the relevance of POD in primary elective THA and TKA. Patients who experience POD are prone to other major complications and have worse patientreported outcome up to 1 year postoperatively. Therefore, POD prophylaxis and thorough screening of at-risk patients is mandatory to reliably recognize POD and enable early initiation of adequate treatment. Our study has some limitations. Data acquisition was limited to the data available from the hospital information system and the institutional joint registry. Because of the limited number of POD patients in the study group our results need to be confirmed in larger cohorts. Mild and especially hypoactive forms of delirium might have not been recognized appropriately and the severity of POD was not captured. Due to the retrospective design, analysis of temporal relation of POD and occurrence of complications was not possible. Follow-up for the variable surgical complication was limited to 30 days postoperatively. Non-surgical and Clavien–Dindo IV° complications could only be captured during hospital stay (mean 9 days postoperatively). Some patients might have also been readmitted to other hospitals. Furthermore, other parameters with possible impact on outcome, such as BMI and psychosocial aspects, could not be captured. Another limitation is the mean length of stay in our cohort. Considering international standards, the comparatively long length of stay is related to the German healthcare system and represents the typical length of stay after THA and TKA during the study period. Despite these limitations, our results emphasize the relationship of POD to complications after THA

and TKA. Furthermore, our study is the first to demonstrate a negative impact of POD on patient-reported outcome after elective total joint arthroplasty. A further strength of our study is the high number of patients in a monocentric study design that guarantees standardized operative workflows and postoperative treatment protocols for THA and TKA. In this way, possible confounding factors were minimized. Future research should focus on the utility of orthogeriatric care models in prevention of POD and its negative consequences. Conclusion The risk of postoperative delirium should be taken seriously in elderly patients undergoing primary elective THA and TKA. Even though cause and effect are difficult to differentiate, affected patients obviously suffer more complications and show worse PROMs up to 1 year postoperatively. Prevention strategies, screening for at-risk patients, and standardized therapy protocols are mandatory to avoid burden on patients and healthcare providers. Supplementary data Figures 1–2 and Tables 3–4 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1 080/17453674.2021.1980676

GM, MW, TK, and MM originated the idea for the study and led on its design. MW, TK, and JG supervised the project. MW, MM, JUG, LP, and TK participated in the design of the study and were responsible for data acquisition. GM, TR, MW, and MM contributed to analysis and interpretation of data. MW provided statistical consultation. MM drafted the manuscript. TK, GM, JG, and MW revised the manuscript critically for important intellectual content. All authors read and approved the final version of the manuscript. Acta thanks Esa Jämsen and Annette W-Dahl for help with peer review of this study.

Abelha F J, Luís C, Veiga D, Parente D, Fernandes V, Santos P, et al. Outcome and quality of life in patients with postoperative delirium during an ICU stay following major surgery. Crit Care 2013; 17(5): R257. Aldecoa C, Bettelli G, Bilotta F, Sanders R D, Audisio R, Borozdina A, et al. European Society of Anaesthesiology evidence-based and consensusbased guideline on postoperative delirium. Eur J Anaesthesiol 2017; 34(4): 192-214. Aziz K T, Best M J, Naseer Z, Skolasky R L, Ponnusamy K E, Sterling R S, et al. The association of delirium with perioperative complications in primary elective total hip arthroplasty. Clin Orthop Surg 2018; 10(3): 286-91. Bin Abd Razak H R, Yung W Y A. Postoperative delirium in patients undergoing total joint arthroplasty: a systematic review. J Arthroplasty 2015; 30(8): 1414-7. Dindo D, Demartines N, Clavien P-A. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004; 240(2): 205-13. Duppils G S, Wikblad K. Acute confusional states in patients undergoing hip surgery:. a prospective observation study. Gerontology 2000; 46(1): 36-43.

700

Gaudreau J-D, Gagnon P, Harel F, Tremblay A, Roy M-A. Fast, systematic, and continuous delirium assessment in hospitalized patients: the nursing delirium screening scale. J Pain Symptom Manage 2005; 29(4): 368-75. Inouye S K, Westendorp R G J, Saczynski J S. Delirium in elderly people. Lancet 2014; 383(9920): 911-22. Judge A, Cooper C, Williams S, Dreinhoefer K, Dieppe P. Patient-reported outcomes one year after primary hip replacement in a European Collaborative Cohort. Arthritis Care Res 2010; 62(4): 480-8. Marcantonio E R, Flacker J M, Michaels M, Resnick N M. Delirium is independently associated with poor functional recovery after hip fracture. J Am Geriatr Soc 2000; 48(6): 618-24. Meyer M, Parik L, Leiß F, Renkawitz T, Grifka J, Weber M. Hospital Frailty Risk Score predicts adverse events in primary total hip and knee arthroplasty. J Arthroplasty 2020; S088354032030752X. Mosk C A, Mus M, Vroemen J P, van der Ploeg T, Vos D I, Elmans L H, van der Laan L. Dementia and delirium, the outcomes in elderly hip fracture patients. Clin Interv Aging 2017; 12: 421-30. Overgaard A, Lidgren L, Sundberg M, Robertsson O, W-Dahl A. Patientreported 1-year outcome not affected by body mass index in 3,327 total knee arthroplasty patients. Acta Orthop 2019; 90(4): 360-5. Petersen P B, Jørgensen C C, Kehlet H, Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Collaborative Group. Delirium after fast-track hip and knee arthroplasty: a cohort study of 6331 elderly patients. Acta Anaesthesiol Scand 2017; 61(7): 767-72.

Acta Orthopaedica 2021; 92 (6): 695–700

Pham T, van der Heijde D, Altman R D, Anderson J J, Bellamy N, Hochberg M, Simon L, Strand V, Woodworth T, Dougados M. OMERACTOARSI Initiative: Osteoarthritis Research Society International set of responder criteria for osteoarthritis clinical trials revisited. Osteoarthritis Cartilage 2004; 12(5): 389-99. Shrier I, Platt R W. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008; 8: 70. Sloan M, Premkumar A, Sheth N P. Projected volume of primary total joint arthroplasty in the U.S., 2014 to 2030: J Bone Joint Surg 2018; 100(17): 1455-60. United Nations, Department of Economic and Social Affairs, Population Division. World population ageing: 2017 highlights. New York: UN; 2017. Weber M, Craiovan B, Woerner M L, Schwarz T, Grifka J, Renkawitz T F. Predictors of outcome after primary total joint replacement. J Arthroplasty 2018; 33(2): 431-5. Weinstein S M, Poultsides L, Baaklini L R, Mörwald E E, Cozowicz C, Saleh J N, et al. Postoperative delirium in total knee and hip arthroplasty patients: a study of perioperative modifiable risk factors. Br J Anaesth 2018; 120(5): 999-1008. Yang Q, Wang J, Huang X, Xu Y, Zhang Y. Incidence and risk factors associated with postoperative delirium following primary elective total hip arthroplasty: a retrospective nationwide inpatient sample database study. BMC Psychiatry 2020; 20(1): 343.

Acta Orthopaedica 2021; 92 (6): 701–708

701

A matched comparison of the patient-reported outcome measures of 38,716 total and unicompartmental knee replacements: an analysis of linked data from the National Joint Registry of England, Northern Ireland and Isle of Man and England’s National PROM collection programme Hasan R MOHAMMAD 1,2, Andrew JUDGE 1,2, and David W MURRAY 1 1 Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of 2 Musculoskeletal Research Unit, Bristol Medical School, University of Bristol, Level 1 Learning and

Trym, Bristol, UK Correspondence: hasanmohammad@doctors.org.uk Submitted 2021-04-13. Accepted 2021-06-17.

Background and purpose — The surgical treatment options for severe knee osteoarthritis are unicompartmental (UKR) and total knee replacement (TKR). For patients, functional outcomes are more important than revision rate. We compared the patient-reported outcome measures (PROMs) of both implant types using a large PROMs dataset. Patients and methods — We analysed a propensitymatched comparison of 38,716 knee replacements (19,358 UKRs and 19,358 TKRs) enrolled in the National Joint Registry and the English National PROM collection programme. Subgroup analyses were performed in different age groups. Results — 6-month postoperative Oxford Knee Score (OKS) for UKR and TKR were 38 (SD 9.4) and 36 (SD 9.4) respectively. A higher proportion of UKRs had an excellent OKS (≥ 41) compared with TKR (47% vs 36%) and a lower proportion of poor OKS (< 27) scores (13% vs. 16%). The 6-month OKS was higher in all age groups for UKR compared with TKR, with the difference increasing in older age groups. The mean 6-month EQ-5D score was 0.78 (SD 0.25) and 0.75 (SD 0.25) respectively. The improvement in EQ-5D resulting from surgery was higher for UKR than TKR both overall and in all age groups. All comparisons were statistically significant (p < 0.05). Interpretation — UKR had a greater proportion of excellent OKS scores and lower proportion of poor scores than TKR. Additionally, the quality of life was higher for UKR compared with TKR. These factors should be balanced against the higher revision rate for UKR when choosing which procedure to perform.

Oxford, Oxford, UK; Research Building, Southmead Hospital, Westbury-on-

The main treatments for severe knee arthritis that has failed to respond to nonoperative management are total knee replacement (TKR) and unicompartmental knee replacement (UKR). UKR offers advantages over TKR including reduced mortality and medical complications (Liddle et al. 2014), and a faster recovery, but the registries report several times higher revision rates (National Joint Registry 2018, Australian Orthopaedic Association 2019, New Zealand Joint Registry 2019). Approximately 50% of knees needing replacement are appropriate for UKR (Willis-Owen et al. 2009), yet current usage is only 10% given the higher revision rates (National Joint Registry 2018). Although there is some evidence of better functional outcomes for UKR compared with TKR, all previous studies are limited by sample size, particularly for the UKR arm (Baker et al. 2012, Liddle et al. 2015, Beard et al. 2019, Wilson et al. 2019). In assessing risk, patients need more information than revision rate alone, which is the traditional metric for measuring joint replacement outcome (Goodfellow et al. 2010). In recent years there has been a drive towards more patient-directed outcomes. Goodman et al. (2020) found that what mattered most to patients following a knee replacement was relief of pain, restoration of function, and improved quality of life. We compared the functional outcomes and quality of life of matched TKRs and UKRs, both overall and in different age groups, using data from 3 national datasets.

702

Acta Orthopaedica 2021; 92 (6): 701–708

Unilateral knee replacements from the linked NJR and HES APC dataset from January 1, 2004 to December 31, 2018 n = 858, 725

HES PROMS dataset 2009–2018

Excluded (n = 115,567): – patellofemoral replacements and missing implant typ, 8,342 – not complete data for baseline demographics, 107,225

Cohort with complete baseline data (n = 743,158): – TKR, 687,910 – UKR, 55,248 Excluded (n = 468,456): – no preoperative and postoperative OKS available, 458,798 – no preoperative anxiety levels available, 9,658 Unmatched cohort (n = 274,702) – TKR, 254,355 – UKR, 20,347 Matched cohort (n = 38,716): – TKR, 19,358 – UKR, 19,358

Figure 1. Data flowchart of dataset cleaning and merging.

Patients and methods Data sources We performed a retrospective observational study using National Joint Registry for England Wales and Northern Ireland and Isle of Man (NJR) records linked to the Hospital Episodes Statistics Admitted Patient Care records (HESAPC) database and England’s National Patient Reported Outcome Measures (PROMs) database. The NJR was established in 2003 and is now the world’s largest arthroplasty register (National Joint Registry 2018). HES-APC records is a database of all admission episodes for patients being admitted to an NHS hospital in England (NHS Digital 2020a). From approximately 2009 onwards, NHS-funded knee replacements as part of the PROMs programme have both preoperative and 6-month postoperative PROMs recorded (NHS Digital 2020b). These include the Oxford Knee Score (OKS) (Murray et al. 2007) and quality of life index EuroQol 5 Domain index (EQ-5D) (Group 1990, Devlin et al. 2010). The choice of time intervals by the PROMs programme was a compromise between appropriate proximity to surgery (to provide timely feedback and to avoid influence of nonoperative factors) and sufficient follow-up for comparison whilst accounting for the postoperative recovery period. Research indicates most improvement in PROMs after joint replacement occurs in the first 6 months, with only minor improvement between 6 months and 1 year (Browne et al. 2013). Long-term studies of TKR and UKR have shown that PROMs remain relatively constant after this, at least up to the 10th postoperative year (Pandit et al. 2011, Breeman et al. 2013, Williams et al. 2013).

Data linkage Between January 1, 2004 and December 31, 2018, 687,910 TKRs and 55,248 UKRs from the NJR dataset (National Joint Registry 2018) were successfully linked to the HES APC dataset (NHS Digital 2020a) with a full set of baseline demographic and surgical factors needed for matching. Bilateral knee replacements were excluded to allow data linkage. This dataset was merged with the HES PROMs dataset, which started collecting data from approximately 2009 onwards (NHS Digital 2020b). All preoperative PROMs needed to be completed within 3 months prior to surgery or at the latest 1 month postoperatively to be regarded as robust for inclusion. Cases were excluded if either no preoperative anxiety score was available or there was not both a preoperative and postoperative OKS. The demographics of these patients excluded are summarised in Table 1 (see Supplementary data). There were 254,355 TKRs and 20,347 UKRs meeting the above criteria, making them eligible for inclusion (Figure 1). Datasets were linked using pseudo-anonymised identification numbers. Propensity matching There were substantial differences in baseline characteristics between TKR and UKR groups (Table 2). Logistic regression was used to generate a propensity score representing the probability that a patient received a UKR and were generated from patient demographics and surgical factors. All patient and surgical factors in Table 2 were used for matching, apart from BMI, which had a large proportion of missing data. This is a well-recognised approach (Matharu et al. 2017, Matharu et al. 2018, Mohammad et al. 2020a, 2020b). Surgical factors included surgeon caseload, defined as the average number of primary knee replacements performed per year as described previously (Liddle et al. 2016, Mohammad 2020a). The algorithm used matched 1:1 on the logit of the propensity score with a 0.02-SD calliper width. Greedy matching without replacement was used given its superior performance for estimating treatment effects (Austin 2009a). Standardized mean differences (SMDs) were examined both before and after matching to assess for any imbalance between groups, with SMDs of > 10% suggestive of covariate imbalance (Austin 2009b). After matching, 38,716 knee replacements (19,358 TKRs and 19,358 UKRs) were included for analysis. Outcomes of interest Outcomes of interest were: (1) preoperative OKS and EQ-5D scores, (2) 6-month postoperative OKS and EQ-5D scores, and (3) difference in OKS and EQ-5D scores postoperatively and preoperatively. Subgroup analyses were performed in 4 different age groups as per the NJR (National Joint Registry 2018); < 55 years, 55–64 years, 65–74 years, and ≥ 75 years. The OKS has 12 items relating to knee pain and function, presented as an overall score between 0 and 48 (Murray et al. 2007). Mean OKS scores are reported with the proportion attaining excellent (≥ 41), good (34–41), fair (27–33),

Acta Orthopaedica 2021; 92 (6): 701–708

703

Frequency (%)

Preoperative distribution (%)

Postoperative distribution (%)

6-month postopertive Oxford knee score

Preoperative UKR Preoperative TKR Postoperative UKR Postoperative TKR

UKR TKR Overlap

UKR TKR 40

30 32

16 10 2

Oxford knee score

Preoperative Oxford knee score

Figure 2. Distribution of OKS in TKR and UKR groups: (A) preoperative and postoperative line graph; Figure 3. LOWESS curve showing (B) preoperative histogram; (C) postoperative histogram. the relationship between preoperative and postoperative Oxford Knee Score for TKR and UKR groups.

and poor (< 27) results defined by Kalairajah et al. (2005). Various estimates of the minimal clinically important difference (MCID) for the OKS have been made; this is considered to be between 3 and 5 points (Beard et al. 2015). The EQ-5D comprises 5 questions concerning mobility, selfcare, activities of daily living, pain, and anxiety/depression. These answers can be presented as a weighted overall index from 1 (perfect health) to –0.594 (worst possible state) (Group 1990, Devlin 2010). Statistics Given that PROMs scores were not normally distributed, appropriate nonparametric tests were used. To compare preand postoperative scores within TKR and UKR groups we used the Wilcoxon signed rank test. To compare TKR and UKR scores the Mann–Whitney test was performed. Locally weighted scatterplot smoothing (LOWESS) curves (Cleveland 1979) were plotted to explore the relationship between preoperative and postoperative PROMs. For clarity purposes the scatter points are suppressed in the plots presented. The percentage of the possible change (PoPC) was calculated as described previously (Kiran et al. 2014). This expresses the actual change attained as a percentage of the possible change, for example for a preoperative OKS of 20 with postoperative score of 40. The actual change is 20 points and the possible change is 48–20 = 28. Therefore the PoPC is 20/28*100 = 71.4%. All statistical analyses were performed using Stata (Version 15.1; StataCorp, College Station, TX, USA) except propensity score matching, which was performed using R (Version 3.4.0; R Foundation for Statistical Computing, Vienna, Austria). P-values of < 0.05 were considered significant. Ethics, funding, and potential conflicts of interest This study was approved by the NJR Research subcommittee and had ethical approval from the South Central Oxford B Research Ethics Committee (19/SC/0292). The linkage of the

datasets was approved by the Confidentiality Advisory Group (19/CAG/0054). Financial support has been received from Zimmer Biomet. HRM was supported by the Henni Mester Scholarship at University College, Oxford University and the Royal College of Surgeons’ Research Fellowship. AJ was supported by the NIHR Biomedical Research Centre at the University Hospitals Bristol NHS Foundation Trust and the University of Bristol.

Results The unmatched cohort consisted of 254,355 TKRs and 20,347 UKRs with several statistically significant baseline differences between groups (Table 2). The matched study group consisted of 38,716 knee replacements (19,358 TKRs, 19,358 UKRs), which were well balanced (Table 2). The distribution of the OKS scores is illustrated in Figure 2. LOWESS curves showed as preoperative OKS increased as did the postoperative score, with similar gradients for both implants. A ceiling effect was visible for the higher preoperative scores (Figure 3). For any given preoperative score, the postoperative score was higher for UKR than TKR. Figure 4 shows how the PoPC was influenced by preoperative score through LOWESS curves. For all preoperative scores the PoPC was greater for UKR than TKR. The higher the preoperative score the larger the differences. The mean preoperative OKS for the TKR and UKR groups was similar, at 21 (SD 7.9) and 21 (SD 7.7). Both groups showed statistically significant improvements in their 6-month postoperative scores (p < 0.001) to 36 (SD 9.4) and 38 (SD 9.4) respectively. The UKR group had a statistically significantly higher (p < 0.001) 6-month postoperative score by 1.7 points. The TKR group gained 15 points (SD 9.8) postoperatively whereas the UKR group gained 17 points (SD 9.6), with the difference being statistically significant (p < 0.001) (Figure 5).

704

Acta Orthopaedica 2021; 92 (6): 701–708

Table 2. Baseline characteristics before matching TKRs and UKRs. Values are count (%) unless otherwise specified Covariate

Unmatched cohort Matched cohort TKR UKR TKR UKR (n = 254,355) (n = 20,347) SMD (n = 19,358) (n = 19,358) SMD

Admission type Elective 254,178 (100) 20,337 (100) 0.01 19,349 (100) 19,349 (100) 0.01 Emergency 163 (0) 10 (0) 8 (0) 9 (0) Other 14 (0) 0 (0) 1 (0) 0 (0) Sex Female 145,049 (57) 9,611 (47) 0.20 9,083 (47) 9,206 (48) 0.01 Male 109,306 (43) 10,736 (53) 10,275 (53) 10,152 (52) Age at surgery, mean (SD) 70.2 (8.8) 64.7(9.3) 0.60 64.9 (9.0) 65.1 (9.3) 0.02 BMI, n 192,787 15,919 14,815 15,059 mean (SD) 30.9 (5.4) 30.2 (4.9) 0.14 31.0 (5.3) 30.2 (4.9) 0.16 Primary diagnosis Primary OA 246,026 (97) 19,993 (98) 0.10 19,011 (98) 19,013 (98) 0.01 Primary OA and other 2,645 (1) 129 (1) 139 (1) 125 (1) Other 5,684 (2) 225 (1) 208 (1) 220 (1) Charlson comorbidity index None 177,003 (70) 14,997 (74) 0.10 14,343 (74) 14,245 (74) 0.02 Mild 54,018 (21) 3,963 (19) 3,778 (20) 3,787 (20) Moderate 16,160 (6) 972 (5) 857 (4) 921 (4) Severe 7,174 (3) 415 (2) 380 (2) 405 (2) Ethnicity White 243,425 (96) 19,759 (97) 0.09 18,781 (97) 18,788 (97) 0.01 Black (Caribbean) 1,291 (1) 70 (0) 70 (0) 68 (0) Black (African) 849 (0) 39 (0) 41 (0) 39 (0) Black (other) 379 (0) 19 (0) 17 (0) 18 (0) Indian 4,499 (2) 217 (1) 228 (1) 213 (1) Pakistani 1,280 (0) 51 (0) 45 (0) 50 (0) Bangladeshi 113 (0) 4 (0) 6 (0) 4 (0) Chinese 175 (0) 5 (0) 5 (0) 5 (0) Other 2,344 (1) 183 (1) 165 (1) 173 (1) Rural/urban classification Urban 187,601 (74) 14,044 (69) 0.12 13,520 (70) 13,451 (70) 0.008 Town/fringe 31,579 (12) 2,612 (13) 2,458 (13) 2,476 (13) Village/hamlet 35,175 (14) 3,691 (18) 3,380 (17) 3,431 (17) Indices of multiple deprivation Most deprived (20%) 34,627 (14) 1,961 (10) 0.18 1,906 (10) 1,917 (10) 0.006 More deprived (20–40%) 45,285 (18) 3,111 (15) 3,010 (16) 2,996 (16) Middle-point 56,721 (22) 4,540 (22) 4,295 (22) 4,324 (22) Less deprived (20–40%) 60,782 (24) 4,957 (24) 4,779 (24) 4,737 (24) Least deprived (20%) 56,940 (22) 5,778 (29) 5,368 (28) 5,384 (28) Surgeon caseload of primary knee surgery practice Cases/years, mean (SD) 80.5 (48.2) 97.6 (50.6) 0.35 96.7 (55.1) 96.7 (50.5) 0.001 Primary complexity Normal 254,328 (100) 20,344 (100) 0.004 19,354 (100) 19,355 (100) 0.004 Complex 27 (0) 3 (0) 4 (0) 3 (0) ASA grade 1 22,257 (9) 3,646 (18) 0.34 3,418 (17) 3,294 (17) 0.02 2 190,181 (75) 14,901 (73) 14,274 (74) 14,297 (74) 3 or above 41,917 (16) 1,800 (9) 1,666 (9) 1,767 (9) VTE prophylaxis—chemical LMWH (± other) 179,562 (71) 13,586 (67) 0.10 12,708 (66) 12,882 (67) 0.02 Aspirin only 12,338 (5) 1,370 (7) 1,316 (7) 1,283 (7) Other 55,739 (22) 4,895 (24) 4,855 (25) 4,704 (24) None 6,716 (2) 496 (2) 479 (2) 489 (2) VTE prophylaxis—mechanical Any 242,433 (95) 19,775 (97) 0.10 18,781 (97) 18,793 (97) 0.004 None 11,922 (5) 572 (3) 577 (3) 565 (3) Fixation Cemented 246,269 (97) 14,932 (74) 0.70 15,023 (78) 14,926 (77) 0.01 Cementless 7,209 (3) 4,950 (24) 3,920 (20) 4,014 (21) Hybrid 877 (0) 465 (2) 415 (2) 418 (2) VTE = venous thromboemolism

The UKR groups had a statistically significantly higher proportion of excellent OKS compared with TKR (47% vs. 36%, p < 0.001) 6 months postoperatively (Table 3). The UKR group had a statistically significantly lower proportion of poor scores than TKR at 6 months postoperatively (13% vs. 16%, p < 0.001) (Table 3). The mean preoperative EQ-5D index for the TKR and UKR groups was 0.47 (SD 0.30) and 0.47 (SD 0.30) respectively with this difference being non-statistically significant (p = 0.3). Both groups showed a statistically significant improvement in their 6-month scores (p < 0.001) to 0.75 (SD 0.25) and 0.78 (SD 0.25) respectively. The TKR group gained 0.28 (SD 0.32) points postoperatively whereas the UKR group gained 0.31 (SD 0.31) with the difference being statistically significant (p < 0.001) (Figure 6). Effect of age on PROMS in the matched cohort Age groups were stratified into 4 groups as per the NJR: (1) < 55 years (2,569 TKRs, 2,763 UKRs); (2) 55–64 years (6,581 TKRs, 6,256 UKRs); (3) 65–74 years (7,415 TKRs, 7,182 UKRs); and (4) ≥ 75 years (n = 2,793 TKRs, 3,157 UKRs). Preoperative OKS was similar between TKR and UKR across all age groups. Younger age groups had poorer OKS than older groups, reflecting the higher threshold to operate in younger patients (Table 4, see Supplementary data). For both TKR and UKR all age groups showed statistically significant improvements postoperatively compared with

Acta Orthopaedica 2021; 92 (6): 701–708

705

Table 2 continued

Possible change (%) 80

Covariate

Unmatched cohort Matched cohort TKR UKR TKR UKR (n = 254,355) (n = 20,347) SMD (n = 19,358) (n = 19,358) SMD

Year of surgery 2008 6 (0) 0 (0) 0.16 1 (0) 0 (0) 2009 14,016 (5) 1,051 (5) 1,089 (5) 1,041 (6) 2010 22,414 (9) 1,621 (8) 1,565 (8) 1,595 (8) 2011 24,678 (10) 1,583 (8) 1,796 (9) 1,563 (8) 2012 24,922 (10) 1,844 (9) 1,711 (9) 1,802 (9) 2013 27,176 (11) 1,870 (9) 1,883 (10) 1,803 (9) 2014 29,682 (12) 2,273 (11) 2,241 (12) 2,176 (11) 2015 29,154 (11) 2,223 (11) 2,335 (12) 2,088 (11) 2016 29,896 (12) 2,517 (13) 2,381 (12) 2,315 (12) 2017 27,687 (11) 2,647 (13) 2,255 (12) 2,445 (13) 2018 24,724 (9) 2,718 (13) 2,101 (11) 2,530 (13) Preoperative Oxford Knee Score mean (SD) 18.9 (7.7) 21.5 (7.7) 0.34 21.4 (7.9) 21.4 (7.7) Preoperative anxiety/depression status (anxious or depressed) Not 160,604 (63) 13,739 (67) 0.10 13,027 (67) 13,030 (67) Moderately 83,970 (33) 6,015 (30) 5,783 (30) 5,753 (30) Extremely 9,781 (4) 593 (3) 548 (3) 575 (3) Bone graft None 251,103 (100) 20,284 (100) 0.11 19,300 (100) 19,295 (100) Used 3,252 (0) 63 (0) 58 (0) 63 (0)

0.09

UKR TKR 60

Preoperative Oxford knee score

0.005

Figure 4. LOWESS curve of percentage of the possible change for TKR and UKR in relation to preoperative OKS.

0.009

0.005

SMD = Standardized mean difference

Table 3. Proportion of OKS (Kalairajah et al. classification) Factor

TKR (n = 19,358) UKR (n = 19,358) n (%) n (%) p-value

Preoperative OKS categorisation Poor 14,244 (74) 14,391 (74) Fair 3,780 (20) 3,820 (20) Good 1,223 (6) 1,070 (6) Excellent 111 (0) 77 (0) Postop OKS categorisation: Poor 3,115 (16) 2,538 (13) Fair 3,071 (16) 2,356 (12) Good 6,200 (32) 5,442 (28) Excellent 6,972 (36) 9,022 (47)

Mean (SD) Oxford knee score

Mean (SD) EQ5D index

1.0

UKR TKR

0.8

0.6

0.5 0.7 0.002 0.01 < 0.001 < 0.001 < 0.001 < 0.001

Comparisons between implant types were performed using the chi-square proportional test.

UKR TKR

0.4 0.2

–0.2

–0.4

Preopertive

6-month postoperative

Difference

Figure 5. Comparison of mean OKS in matched cohort of TKR and UKRs. Error bars represent SD.

preoperatively (p < 0.001) (Table 4, see Supplementary data). UKR gained more points postoperatively compared with TKR across all age groups (p < 0.001), with this difference increasing with age (Table 4, see Supplementary data). The 6-month OKS was higher in all age groups for UKR compared with TKR (p < 0.001) with the difference increasing in the older age groups (Table 4, see Supplementary data). Preoperatively the proportions of poor, fair, good, and excellent OKS were similar between TKR and UKR for all age groups (Table 5, see Supplementary data). At 6 months postoperatively there was a greater proportion of excellent and a

–0.596

Preopertive

6-month postoperative

Difference

Figure 6. Comparison of mean EQ-5D index in matched cohort of TKR and UKRs. Error bars represent SD.

lower proportion of poor scores for UKR compared with TKR across all age groups (Table 5, see Supplementary data). The proportion of excellent scores was 6%, 12%, 12%, and 10% higher in UKR compared with TKR for the < 55 years, 55–64 years, 65–74 years, and ≥ 75 years groups respectively. The proportion of poor OKS scores was 3–4% lower in UKR compared with TKR across all age groups. Preoperatively the EQ-5D score was similar between TKR and UKR across age groups (Table 6, see Supplementary data). For both TKR and UKR younger age groups had poorer EQ-5D scores than older groups. For both TKR and

706

UKR all age groups showed statistically significant improvements postoperatively compared with preoperatively (p < 0.001) although UKR gained more points postoperatively compared with TKR across all age groups (p < 0.001) except the < 55 group (p = 0.5) (Table 6, see Supplementary data). The 6-month EQ-5D was higher in all age groups for UKR compared with TKR (p < 0.001) (Table 6, see Supplementary data).

Discussion This is the largest study comparing the PROMs of TKR and UKR and helps provide answers to outcome metrics patients find most important (Goodman et al. 2020). After matching, a substantially higher proportion of UKRs had an excellent OKS compared with TKR (47% vs. 36%) and a lower proportion of poor scores (13% vs. 16%). This is important, given that currently 1 in 5 patients who undergo a total knee replacement are dissatisfied with their knee replacement (Beswick et al. 2012, Price et al. 2018). This number would likely be lower if more patients suitable for UKR had UKR surgery, which by some estimates could be up to 50% (WillisOwen et al. 2009). The 6-month postoperative OKS of UKR was higher than TKR by 2 points. This difference is similar to the TOPKAT (Beard et al. 2019) randomised control trial, suggesting that it is a real difference. Although the magnitude of the difference is below the suggested MCID for the OKS, given the skewed nature of the outcome scores (Figure 2), together with the ceiling effect of the OKS (Dawson et al. 2014), this does not mean that the difference is unimportant. Indeed, its importance is highlighted by finding that the relative risk (1.3) of having an excellent score is 30% higher following a UKR rather than a TKR and the relative risk (0.81) of having a poor score is 20% less following UKR. Additionally, for any given preoperative OKS, a greater PoPC postoperatively was observed for UKR than for TKR (Figure 4). This was particularly marked in patients with higher preoperative OKS. In all age groups except the < 55 years group, the average 6-month postoperative OKS was about 2 points greater with UKR than TKR and UKR were about 30% more likely to have an excellent OKS and 20% less likely to have a poor OKS. In the < 55 age group the difference in 6-month postoperative OKS between TKR and UKR groups was only 1.3 points. Also, in this age group UKR were 20% more likely to have an excellent outcome and 12% less likely to have a poor outcome. It is not clear why this is. It may be because the preoperative scores were much lower in this group, and, as can be seen in the graph comparing pre- and postoperative OKS (Figure 3), with lower preoperative scores the difference between UKR and TKR is smaller. It may also be that in this age group more UKR patients had early stages of osteoarthritis (without boneon-bone arthritis) and these patients tend not to do well (Ken-

Acta Orthopaedica 2021; 92 (6): 701–708

nedy et al. 2020a). Despite this, in all age groups UKR have better function than TKR, which justifies using UKR in all age groups. This is particularly important in the elderly as only 4% of knee replacements in patients older than 75 years are UKR (National Joint Registry 2018), when approximately one-third of knee replacements are appropriate for UKR (Kennedy et al. 2020b). Therefore approximately 10 times as many UKRs could be done in this age group. This study has also shown UKR offers better 6-month quality of life with EQ-5D scores both overall and on age subgroup analyses. Overall and in all age groups the EQ-5D index was between 0.02 and 0.03 points higher for UKR compared with TKR. This is close to the lower limit of the predicted MCID for the EQ-5D index, which is considered to range between 0.03 and 0.54 points (Coretti et al. 2014). When determining quality of life improvement the index is summated annually, so, over the time period the devices are implanted, the differences are likely to be appreciable and well above the MCID. Our results agree with those reported from Liddle et al. (2015) who found that UKR had higher 6-month OKS and EQ-5Ds than matched TKRs. In contrast they are different from those of Baker et al. (2012) who found no difference in the PROMs gained by UKR and TKR in an analysis adjusted for case-mix and preoperative score. However, there were only 505 UKRs in the Baker study, suggesting that this study was underpowered compared with the Liddle study (3,519 UKRs), which had about 7 times as many UKR, and our study, which had about 40 times as many UKR (n = 19,358). This is the largest study comparing the PROMs of matched UKR and TKR and the first to perform analyses in different age groups. The main study strengths are that we used an unselected registry sample, which reduces the chances of selection bias. By linking datasets various confounding factors were matched, allowing fair comparison. The main limitation is that this is a retrospective study with 6-month postoperative scores with some evidence existing (Browne et al. 2013) of slight further increases (below the MCID) in OKS between 6 and 12 months postoperatively. However, most improvement in PROMs after joint replacement occurs in the first 6 months (Pandit et al. 2011, Breeman et al. 2013, Browne et al. 2013, et al.Williams 2013). Additionally, matching can reduce the generalisability of findings, but given we were able to match virtually all the UKR to TKR this is unlikely to be an issue. Finally, we were only able to match using variables collected in the databases. There could be unaccounted variables that could lead to some residual confounding. Conclusion Surgeons have traditionally made the decision on which implant to use based on relative revision rates; however, patients are much more concerned about functionality (Goodman et al. 2020). This study shows that UKR offers superior functional outcomes and quality of life across all age groups.

Acta Orthopaedica 2021; 92 (6): 701–708

Although the absolute mean difference in OKS is below the MCID, the likelihood of an excellent OKS is about 30% higher for UKR and the likelihood of a poor OKS is about 20% lower for UKR. We recommend that the findings of this study are discussed with patients alongside the increased risk of UKR revision to help patients make more informed decisions about their care. Supplementary data Tables 1 and 4–6 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2021.1956744

HRM, AJ, and DWM designed the study. HRM and DWM analysed the data with statistical support from AJ. HRM, AJ, and DWM helped with data interpretation. HRM wrote the initial manuscript draft, which was then revised appropriately by all authors. The authors would like to thank the patients and staff of all the hospitals in England, Wales, Northern Ireland and Isle of Man who have contributed data to the National Joint Registry. They are grateful to the Healthcare Quality Improvement Partnership (HQIP), the NJR Research Sub-committee, and staff at the NJR Centre for facilitating this work. The authors have conformed to the NJR’s standard protocol for data access and publication. The views expressed represent those of the authors and do not necessarily reflect those of the National Joint Registry Steering Committee or the Healthcare Quality Improvement Partnership (HQIP), who do not vouch for how the information is presented. Acta thanks Margareta Hedström for help with peer review of this study.

Austin P C. Some methods of propensity–score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations. Biom J 2009a; 51(1): 171-84. Austin P C. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity–score matched samples. Stat Med 2009b; 28(25): 3083-107. Australian Orthopaedic Association. Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR). Hip, Knee & Shoulder Arthroplasty 2019. Baker P, Petheram T, Jameson S, Avery P, Reed M, Gregg P, Deehan D. Comparison of patient-reported outcome measures following total and unicondylar knee replacement. Bone Joint J 2012; 94(7): 919-27. Beard D J, Harris K, Dawson J, Doll H, Murray D W, Carr A J, Price A J. Meaningful changes for the Oxford hip and knee scores after joint replacement surgery. J Clin Epidemiol 2015; 68(1): 73-9. Beard D J, Davies L J, Cook J A, MacLennan G, Price A, Kent S, Hudson J, Carr A, Leal J, Campbell H. The clinical and cost-effectiveness of total versus partial knee replacement in patients with medial compartment osteoarthritis (TOPKAT): 5-year outcomes of a randomised controlled trial. Lancet 2019; 394(10200): 746-56. 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). Breeman S, Campbell M, Dakin H, Fiddian N, Fitzpatrick R, Grant A, Gray A, Johnston L, MacLennan G, Morris R. Five-year results of a randomised controlled trial comparing mobile and fixed bearings in total knee replacement. Bone Joint J 2013; 95(4): 486-92.

707

Browne J P, Bastaki H, Dawson J. What is the optimal time point to assess patient-reported recovery after hip and knee replacement? A systematic review and analysis of routinely reported outcome data from the English patient-reported outcome measures programme. Health Qual Life Outcomes 2013; 11(1): 128. Cleveland W S. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979; 74(368): 829-36. Coretti S, Ruggeri M, McNamee P. The minimum clinically important difference for EQ-5D index: a critical review. Expert Rev Pharmacoecon Outcomes Res 2014; 14(2): 221-33. Dawson J, Beard D J, McKibbin H, Harris K, Jenkinson C, Price A. Development of a patient-reported outcome measure of activity and participation (the OKS-APQ) to supplement the Oxford knee score. Bone Joint J 2014; 96(3): 332-8. Devlin N J, Parkin D, Browne J. Patient–reported outcome measures in the NHS: new methods for analysing and reporting EQ–5D data. Health Econ 2010; 19(8): 886-905. Goodfellow J, O’Connor J, Murray D. A critique of revision rate as an outcome measure: re-interpretation of knee joint registry data. Bone Joint J 2010; 92(12): 1628-31. Goodman S M, Mehta B, Mirza S Z, Figgie M P, Alexiades M, Rodriguez J, Sculco P, Parks M, Singh J A. Patients’ perspectives of outcomes after total knee and total hip arthroplasty: a nominal group study. BMC Rheumatol 2020; 4(1): 3. Group T E. EuroQol—a new facility for the measurement of health-related quality of life. Health Policy 1990; 16(3): 199-208. Kalairajah Y, Azurza K, Hulme C, Molloy S, Drabu K J. Health outcome measures in the evaluation of total hip arthroplasties: a comparison between the Harris hip score and the Oxford hip score. J Arthroplasty 2005; 20(8): 1037-41. Kennedy J A, Palan J, Mellon S J, Esler C, Dodd C A, Pandit H G, Murray D W. Most unicompartmental knee replacement revisions could be avoided: a radiographic evaluation of revised Oxford knees in the National Joint Registry. Knee Surg, Sports Traumatol, Arthroscopy 2020a: 1-9. Kennedy J A, Mellon S J, Lombardi A V, Berend K R, Hamilton T W, Murray D W. Candidacy for medial unicompartmental knee replacement declines with age. Orthop Traumatol Surg Res 2020b; 106(3): 443. Kiran A, Hunter D J, Judge A, Field R E, Javaid M K, Cooper C, Arden N K, Committee ONMEUW. A novel methodological approach for measuring symptomatic change following total joint arthroplasty. J Arthroplasty 2014; 29(11): 2140-5. Liddle A D, Judge A, Pandit H, Murray D W. Adverse outcomes after total and unicompartmental knee replacement in 101 330 matched patients: a study of data from the National Joint Registry for England and Wales. Lancet 2014; 384(9952): 1437-45. Liddle A, Pandit H, Judge A, Murray D. Patient-reported outcomes after total and unicompartmental knee arthroplasty: a study of 14 076 matched patients from the National Joint Registry for England and Wales. Bone Joint J 2015; 97(6): 793-801. Liddle A D, Pandit H, Judge A, Murray D W. Effect of surgical caseload on revision rate following total and unicompartmental knee replacement. J Bone Surg Am 2016; 98(1): 1-8. Matharu G S, Judge A, Murray D W, Pandit H G. Trabecular metal acetabular components reduce the risk of revision following primary total hip arthroplasty: a propensity score matched study from the National Joint Registry for England and Wales. J Arthroplasty 2017; 33(2): 447. Matharu G S, Judge A, Murray D W, Pandit H G. Outcomes after metal-on-metal hip revision surgery depend on the reason for failure: a propensity score-matched study. Clin Orthop Relat Res 2018; 476(2): 245-58. Mohammad H R, Matharu G S, Judge A, Murray D W. Comparison of the 10-year outcomes of cemented and cementless unicompartmental knee replacements: data from the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. Acta Orthop 2020a; 91(1): 76-81.

708

Mohammad H R, Matharu G S, Judge A, Murray D W. The effect of surgeon caseload on the relative revision rate of cemented and cementless unicompartmental knee replacements: an analysis from the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. J Bone Joint Surg Am 2020b; 102(8): 644. Murray D, Fitzpatrick R, Rogers K, Pandit H, Beard D, Carr A, Dawson J. The use of the Oxford hip and knee scores. Bone Joint J 2007; 89(8): 1010-4. National Joint Registry. National Joint Registry 15th Annual Report. National Joint Registry for England, Wales, Northern Ireland and Isle of Man; 2018. New Zealand Joint Registry. Twenty Year Report January 1999 to December 2018. New Zealand Joint Registry; 2019. NHS Digital. Hospital Admitted Patient Care Activity 2019-20. Accessed 4/10/2020; 2020a. Available from: https://digital.nhs.uk/data-and-information/publications/statistical/hospital-admitted-patient-care-activity/ 2019-20.

Acta Orthopaedica 2021; 92 (6): 701–708

NHS Digital. Patient reported outcome measures (PROMs). Accessed 4/10/2020; 2020b. Available from: https://digital.nhs.uk/data-and-information/data-tools-and-services/data-services/patient-reported-outcomemeasures-proms. Pandit H, Jenkins C, Gill H, Barker K, Dodd C, Murray D. Minimally invasive Oxford phase 3 unicompartmental knee replacement: results of 1000 cases. Bone Joint J 2011; 93(2): 198-204. 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. Williams D, Blakey C, Hadfield S, Murray D, Price A, Field R. Long-term trends in the Oxford knee score following total knee replacement. Bone Joint J 2013; 95(1): 45-51. Willis-Owen C A, Brust K, Alsop H, Miraldo M, Cobb J. Unicondylar knee arthroplasty in the UK National Health Service: an analysis of candidacy, outcome and cost efficacy. Knee 2009; 16(6): 473-8.

Acta Orthopaedica 2021; 92 (6): 709–715 709

History of previous surgery is associated with higher risk of revision after primary total knee arthroplasty: a cohort study from the Geneva Arthroplasty Registry Hermes H MIOZZARI, Christophe BAREA, Didier HANNOUCHE, and Anne LÜBBEKE

Division of Orthopaedics and Trauma Surgery, Geneva University Hospitals, University of Geneva, Faculty of Medicine, Geneva, Switzerland Correspondence: hermes.miozzari@hcuge.ch Submitted 2021-03-12. Accepted 2021-07-09.

Background and purpose — Prior to primary total knee arthroplasty (pTKA), 6–34% of patients have undergone surgical procedure(s) of their knee. We investigated whether history of previous surgeries influences the risk of revision of pTKA, the risk according to the type of previous surgery, and how previous surgery influences specific causes of revision and the time of revision. Patients and methods — This is a prospective cohort study from the Geneva Arthroplasty Registry. All pTKA between 2000 and 2016 were included and followed until December 31, 2019. Outcomes were risk of revision, evaluated using Kaplan–Meier survival and Cox and competing risks regression, the specific causes, and time of revision. Results — Of 3,945 pTKA included (mean age 71 years, 68% women), 21% had a history of previous surgery, with 8.3% revisions vs. 4.3%, at 3–20 years’ follow-up (mean 8.6). 5- and 10-year cumulative failure by previous surgery (yes vs. no) were 6.6% (95% CI 5.1–8.5) vs. 3.3% (CI 2.7– 4.0), and 8.4% (CI 6.6–10.6) vs. 4.5% (CI 3.8–5.4). Baseline differences explained only part of the higher risk (adjusted HR 1.5, CI 1.1–2.1). The risk of failure was higher for all causes of revision considered. Patients in the previous surgery group had a higher risk of an early revision. Interpretation — A history of previous surgery adversely affected the outcome with a 1.5 times higher cumulative risk of all-cause revision over the course of up to 20 years after index surgery. The increased risk was seen for all causes of revision and was highest in the first years.

The proportion of patients with a history of previous knee surgery before pTKA is documented in most but not all European national and a few local arthroplasty registries (Lübbeke et al. 2018), varying from 6% in Finland to 34% in Switzerland (Table 1). If previous surgery seems to affect both the age and interval for the need for pTKA, with patients undergoing arthroplasty at a significantly younger age (Brophy et al. 2014), it remains unclear whether this plays a substantial role in its outcome. According to recent data, 10-year pTKA survival rate seems not to be affected by arthroscopy (Viste et al. 2017), while the opposite was observed (87% vs. 98%) in an older study (Piedade et al. 2009). Data from the Norwegian (Badawy et al. 2015) knee arthroplasty registry did not show differences in survival rates in patients undergoing pTKA after high tibial osteotomy (HTO), while publications from the Swedish (Robertsson and W-Dahl 2015) and New Zealand (Pearse et al. 2012) registries showed a higher risk of revision. To our knowledge, there is only 1 study considering history of any given previous knee surgery (Lim et al. 2016), showing no difference in terms of revision. We compared revision rates up to 20 years after pTKA in a prospective cohort with and without previous surgeries. Our specific questions were: (i) Does history of previous surgeries influence the risk of revision of pTKA? (ii) What is the risk of revision according to the type of previous surgery? (iii) How does previous surgery influence specific causes of revision and the time of revision?

Patients and methods Study design and setting We performed a prospective cohort study based on the local arthroplasty registry (Geneva Arthroplasty Registry, GAR). Since 1998, all patients undergoing partial, primary total, and © 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 10.1080/17453674.2021.1970322

710

Acta Orthopaedica 2021; 92 (6): 709–715

Table 1. History of previous surgery in publicly available national joint registry reports and present study (GAR) Country Registry Period Published Belgium Finland Germany Italy Netherlands Norway Portugal Sweden Switzerland New Zealand Geneva/Switzerland

ORTHOpride FAR EPRD RIAP LROI NAR PAR SKAR SIRIS NZJR GAR

History of previous surgery (%)

2014–2018 10/2019 (29) 2014–2020 2020 4,300/70,288 (6) 2010–2018 10/2019 (8) 2006–2017 02/2018 (12) 2014–2018 Online (31) 1994–2018 06/2019 1,449/6,905 (24) 2009–2013 (7) 1975–2019 01/2020 (18) 2012–2018 2019 (34) 1999–2018 12/2019 15,376/110,079 (14) 1998–2019 – (21)

either a second-generation cephalosporin or vancomycin, in the case of known allergy or bacterial resistance. Since 2017, the dosage has been doubled for patients with either BMI ≥ 35 or bodyweight ≥ 100 kg. Aftercare Full weight-bearing with use of crutches was allowed from day 1, with full range of motion. Deep vein thrombosis prophylaxis was initiated immediately postoperatively and discontinued at 8 weeks. Physiotherapy was prescribed for at least 3 months.

Variables, outcome measures, data sources, and bias Data collection in the registry is done prospectively. Previous surgeries are documented in the electronic healthcare system and recorded in the registry in pre-specified categories as follows: arthroscopy, meniscectomy, meniscectomy external, meniscectomy internal, osteosynthesis, osteotomy, ligamentoplasty, others, or none. For ease, all meniscectomies were grouped in a single category as “meniscectomy.” The following covariates were assessed: age, sex, BMI, ASA score, smoking status, diagnosis, patellar resurfacing, type of constraint (PS), type of tibial plateau (fixed bearing), year of surgery, and surgery duration. All the information on baseline characteristics and operation, including previous surgery, are routinely recorded on specifically designed data collection forms by the operating surgeon at the time of surgery. Information on comorbidities (BMI, ASA score, smoking status) was collected from the anesthesiology chart. All data are routinely double-checked for completeness by the physician in charge (AL) to assure the quality of the registry. The outcome all-cause revision was subdivided into revision due to aseptic loosening, infection, femoro-patellar problem, pain, arthrofibrosis, periprosthetic fracture, instability, and other causes. A revision is by definition any surgery with (partial) implant exchange or component extraction or resurfacing of the patella. Therefore, reoperation for manipulation under anesthesia, open/arthroscopic synovectomy, or any hardware removal did not account for the endpoint of the present study.

Source: ISAR (International Society of Arthroplasty Registries (https://www.isarhome. org/members). ORTHOpride, Belgian National Arthroplasty Register (https://www.ehealth.fgov.be/file/ view/AXDOTDE0mTlaOSp4Nmeq?filename=Orthopride_Annual_Report_2018.pdf) FAR, Finnish Arthroplasty Registry (https://www.thl.fi/far/#data/cphd) EPRD, Endoprothesenregister Deutschland (https://www.eprd.de/fileadmin/user_upload/ Dateien/Publikationen/Berichte/EPRD_Jahresbericht_2019_2.0.pdf) RIAP, Italian Arthroplasty Registry (http://riap.iss.it/riap/en/activities/reports/2020/05/13/ report-2018-english-addendum/) LROI, Dutch Arthroplasty Register (https://www.lroi-rapportage.nl/media/pdf/PDF%20 Online%20LROI%20annual%20report%202019.pdf) NAR, Norwegian Arthroplasty Register (http://nrlweb.ihelse.net/Rapporter/Rapport2020. pdf) PAR, Portuguese Arthroplasty Registry (http://www.rpa.spot.pt/getdoc/c3d0a244-c0564949-a50b-07d0fdeac2b9/RPA-Report-2013.aspx) SKAR, Swedish Knee Arthroplasty Register (http://myknee.se/pdf/SVK_2019_1.0_Eng. pdf) SIRIS, Swiss National Joint Registry (https://www.siris-implant.ch/fr/ Downloads&category=16) NZJR, New Zealand Joint Registry (https://nzoa.org.nz/system/files/DH8328_ NZJR_2019_Report_v4_7Nov19.pdf)

revision knee arthroplasties are prospectively enrolled in the GAR, which was approved by the local ethics committee and is a member of the International Society of Arthroplasty Registries (ISAR). Participants/study subjects All consecutive patients who underwent elective pTKA for any indication between January 2000 and December 2016 at the Geneva University Hospitals were included and followed up until December 31, 2019. The minimum follow-up was 3 years. The exposure of interest was the presence of previous surgery (yes/no). The outcome of interest was all-cause revision after pTKA. Surgery The vast majority of the patients (99%) underwent pTKA by a standard medial parapatellar approach, with mechanical alignment, mostly with a postero-stabilized (PS) design. Patellar resurfacing has been performed in 68% of cases. Routine component fixation was by antibiotic-loaded cement (Palacos R+G, Heraeus Medical GmbH, Wehrheim, Germany). Preoperative single-shot antibiotic prophylaxis was done with

Statistics To assess the influence of previous surgeries on the risk of revision of pTKA, we first compared patient characteristics at baseline between the group who had undergone previous surgery and the group who had not. Information on previous surgery had been collected for all patients operated on during

Acta Orthopaedica 2021; 92 (6): 709–715 711

Table 2. Baseline characteristics according to previous surgery yes/ no (all primary TKAs). Values are count (%) unless otherwise specified Factor

Previous surgery yes no n = 844 n = 3,101 (21%) (79%) p-value

Women 439 (52) 2,233 (72) Age at operation, mean (SD) 67.3 (9.3) 72.5 (9.2) BMI, mean (SD) a 29 (5.1) 30 (5.7) BMI categories < 24.9 206 (24) 578 (19) 25–29.9 321 (38) 1,106 (36) 30–34.9 227 (27) 829 (27) ≥ 35 89 (11) 572 (19) ASA score 3–4 b 158 (19) 879 (28) c Ever smoker 740 (35) 286 (24) Bilateral primary TKA 88 (10) 625 (20) Diagnosis primary OA 545 (65) 2,848 (92) Previous knee surgery Arthroscopy 127 (15) – Meniscectomy 400 (47) – Osteosynthesis 75 (9) – Osteotomy 129 (15) – Ligamentoplasty 43 (5) – Other 70 (8) – Implant-related information Patellar resurfacing, yes 527 (62) 2,135 (69) Posterior-stabilized, yes 677 (80) 2,496 (81) Fixed-bearing, yes 780 (92) 2,974 (96) Surgery duration, min (SD) 124 (31) 119 (28)

< 0.001 < 0.001 < 0.001

< 0.001 < 0.001 < 0.001 < 0.001

< 0.001 0.9 < 0.001 < 0.001

a BMI was missing in 17 cases (0.4%). b ASA score was missing in 4 cases (0.1%). c Smoking status was missing in 73 cases (1.9%).

stratified by previous surgery yes/no. Moreover, smoothed hazard estimates were obtained with their 95% CIs to evaluate whether the timing of the revisions differed between the 2 groups (Tanner and Wong 1983). Hazard estimates quantify the immediate risk, in this case of all-cause revision, attached to an individual known to be alive at time t. The statistical analyses were performed using the statistical packages IBM SPSS statistics version 25 (IBM Corp, Armonk, NY, USA) and STATA version 15 (StataCorp, College Station, TX, USA). Ethics, funding, data sharing, and potential conflicts of interest This study was approved by the local ethics committee (CCER Geneva, Switzerland). All the data used in this study is retrieved from the Geneva Arthroplasty Registry. The Division of Orthopaedic Surgery received financial institutional support from the “Fondation pour la recherche ostéo-articulaire” for the knee arthroplasty registry. The funding source had no role in the collection, analysis, or interpretation of the data, in the preparation of the manuscript, or its submission for publication. The Geneva Arthroplasty Registry obtains patient consent for data collection and protects access to the data. Patients gave consent to future sharing of data only upon request by other research institutions. Interested researchers may request access to data from DH. None of the authors report any conflict of interest.

Results the inclusion period. Missing values on patient characteristics were: BMI in 17 cases (0.4%), ASA score in 4 (0.1%), and smoking status in 73 cases (1.9%). P-values were obtained using Student’s t-test for continuous variables and Pearson’s chi-square test for categorical variables. To assess the risk of all-cause revision, cumulative failure analysis with all-cause revision as an endpoint was performed using the Kaplan–Meier method. Person-time at risk was determined as the length of the interval between date of surgery for the pTKA and the date of either revision for any reason, death, leaving the area of residency, or end of follow-up (December 31, 2019). We also performed both Cox regression and competing-risks analyses (with death as a competing event) and estimated unadjusted and adjusted hazard ratios (HR) with 95% confidence intervals (CI) (Fine and Gray 1999). Final adjustment variables included age, sex, ASA score, and year of surgery. We evaluated the primary outcome in all pTKAs and in only the first pTKA. Finally, we performed a subgroup analysis including only patients with pTKA for primary osteoarthritis. To investigate whether the causes of revision were related to a history of previous surgery, the revision risk over the entire follow-up period was calculated for specific causes and

Study population (Table 2) 3,945 pTKAs (mean age 71 years, 68% women) were performed during the inclusion period, all enrolled in the registry and included in the final analysis. The mean follow-up time was 8.6 years (3–20 years). In the group of patients with previous surgery, the percentage of women (52% vs. 72%) was statistically significantly smaller, as well as the mean BMI (29 vs. 30), and the percentage of ASA ≥ 3 (19 vs. 28), whereas the proportion of ever smokers was significantly higher (35 vs. 24%). The percentage of patients with patellar resurfacing (69%) and fixed bearing (96%) was statistically significantly higher in the group without previous surgeries, whereas the amount of PS constraint was similar. Does history of previous surgeries influence the risk of revision of primary total knee arthroplasty? Of 3,945 pTKAs included in the study, 844 (21%) had a history of previous surgery, mostly meniscectomies (47%), followed by osteotomies (15%) and arthroscopies (15%) (see Table 2). At an average follow-up of 8.6 years, there were 204 revisions, 70 (8.3%) in patients with previous surgery and 134 (4.3%) in patients without. 5-year cumulative failure by previous surgery (yes vs. no) was 6.6% (CI 5.1–8.5) vs. 3.3% (CI 2.7–4.0).

712

Acta Orthopaedica 2021; 92 (6): 709–715

Table 3. Hazard ratios (HR) for all-cause revision according to previous surgery yes/no Previous surgery (events / total number) Included cases yes no All TKAs Only first TKA a Adjusted

70 / 844 67 / 756

134 / 3,101 109 / 2,476

Cox regression HR (CI) aHR (CI) a 1.9 (1.4–2.6) 2.0 (1.5–2.8)

2.0 (1.5–2.6) 2.1 (1.6–2.8)

Table 4. 5-year cumulative failure rates (and 95% CI) by type of previous surgery procedure (all TKAs)

Kaplan–Meier failure estimate (%) Previous surgery No previous surgery

Factor

Total Number 5-year cumulative number of events failure rate (%)

No previous procedure Previous procedure Arthroscopy Meniscectomy Osteosynthesis Osteotomy Ligamentoplasty Other

Number at risk No previous surgery 3,101 Previous surgery 844

1.6 (1.2–2.1) 1.6 (1.2–2.2)

for age, sex, ASA score and year of surgery using Cox regression and competing-risks regression

1.5 (1.1–2.1) 1.6 (1.2–2.2)

Competing-risks regression HR (CI) aHR (CI) a

3,101 844 127 400 75 129 43 70

134 70 15 29 7 7 3 9

3.3 (2.7–4.0) 6.6 (5.1–8.5) 7.9 (4.4–14) 5.7 (3.8–8.5) 8.3 (3.8–18) 4.1 (1.7–9.5) 7.1 (2.4–21) 12 (6.1–22)

Years after index operation 2,371 627

1,086 329

306 101

0 0

Figure 1. 5- and 10-year cumulative failure for patients with and without history of previous surgery. 5-year cumulative failure (yes vs. no) was 6.6% (CI 5.1–8.5) vs. 3.3% (CI 2.7–4.0). 10-year cumulative failure was 8.4% (CI 6.6–11) vs. 4.5% (CI 3.8–5.4).

10-year cumulative failure was 8.4% (CI 6.6–11) vs. 4.5% (CI 3.8–5.4) (Figure 1). Including all pTKAs, the unadjusted HR (Cox regression) for revision was 1.9 (CI 1.4–2.6) and 1.5 (CI 1.1–2.1) when adjusted for age, sex, ASA score, and year of surgery. Considering only the first pTKA implanted, the unadjusted HR was 2.0 (CI 1.5–2.8), and 1.6 (CI 1.2–2.2) when adjusted (Table 3). Corresponding estimates obtained with competing risks regression were similar. When restricting the inclusion to the diagnosis primary OA (n = 3,393) only, Cox regression estimates were only slightly higher than in the previous analyses. For all pTKA the unadjusted HR was 2.0 (CI 1.4–2.8) and the adjusted HR 1.7 (CI 1.2–2.4). Including only the first pTKA the unadjusted HR was 2.1 (CI 1.5–3.0) and the adjusted HR 1.7 (CI 1.2–2.5). What is the risk of revision according to the type of previous surgery? The 5-year cumulative failure rates according to the type of previous surgery varied between 12% and 5.7 % (Table 4).

Table 5. Revision risk overall and for specific causes according to previous surgery, yes/no (all TKAs). Values are count (%) Revision cause

Previous surgery yes no n = 844 n = 3,101

Aseptic loosening Infection Femoropatellar problem Pain Arthrofibrosis Periprosthetic fracture Instability Other

18 (2.1) 16 (1.9) 9 (1.1) 8 (0.9) 6 (0.7) 5 (0.6) 3 (0.4) 5 (0.6)

29 (0.9) 36 (1.2) 11 (0.4) 17 (0.5) 8 (0.3) 11 (0.4) 5 (0.2) 17 (0.6)

All causes

70 (8.3)

134 (4.3)

How does previous surgery influence specific causes of revision and the time of revision? The risk of revision in patients with vs. without previous surgery was almost twice as high for any specific causes, more often related to aseptic loosening (2.1% vs. 0.9%) or infection (1.9% vs. 1.2%) (Table 5). In terms of main procedures, all component revision was more common in patients without previous surgery whereas partial revision was mostly performed in patients with previous surgery (Table 6). The timing of revision differed between the 2 groups. The risk of revision was substantially higher in the short term as evidenced by the distinct, non-overlapping confidence intervals of the

Acta Orthopaedica 2021; 92 (6): 709–715 713

Table 6. Categories of revision according to previous surgery, yes/no. Values are count (%) Revision

Previous surgery yes no n = 70 n = 134

All component revision 21 (30) Partial revision a 32 (46) Poly exchange 15 (21) Arthrodesis 0 TKA extraction 0 Other 2 (3)

57 (42) 41 (31) 26 (19) 2 (2) 5 (4) 3 (2)

Smoothed hazard estimate 0.25 Previous surgery No previous surgery 0.20

0.15

0.10

0.05

a Partial

revision: revision of either component (femoral, tibial, or patellar).

Years after index operation

smoothed hazard estimates in Figure 2. The curves overlapped completely in the mid-term and partly in the long term.

Discussion Background and rationale The proportion of patients with a history of previous surgery before pTKA reported in registries is highly variable (6–34%) (Table 1). Overestimation, due to multiple counting, underestimation, due to patient recall bias, incomplete chart fill or insufficient anamnesis, different current practices from one country to other, and different time periods included might all explain this variability. Nevertheless, it is not clear how a history of previous surgery influences the outcome after pTKA. In our study, patients in the group with previous surgery had primary arthroplasty at a younger age and showed a 1.5 times higher risk of subsequent revision. The risk did not substantially change when restricting the inclusion to primary OA. The difference in implant failure at 5 and 10 years was notable: about twice the risk at both time points (6.6% vs. 3.3 and 8.4% vs. 4.5%, respectively). The 2 groups’ baseline differences only partly explained the increased risk of revision, which was higher for any specific causes (from aseptic loosening to infection, etc.). The timing of revision differed between the 2 groups and was substantially higher in the short term in patients with pre-dating surgeries. Limitations This study has several limitations. First, no analysis was carried out for single vs. multiple previous surgeries. The number of patients in whom it was clearly stated that they had more than 1 surgery at different points in time was too small to enable any further analysis (60 patients among 844). Other patients had more than 1 procedure noted but it was not possible to discriminate between different procedures for the same surgery or several surgeries at different time intervals. When more than 1 procedure was performed we considered the type of

Figure 2. Smoothed hazard estimates of all-cause revision for patients with and without history of previous surgery. The risk of revision was substantially higher in the short term as evidenced by the distinct, nonoverlapping confidence intervals.

previous surgery that was first documented by the surgeon for calculation of failure by type of previous surgery. Second, we did not consider any surgery where the components were left untouched, such as manipulation under anesthesia, and open or arthroscopic synovectomy. Our results might underestimate the risk of reoperation. Nevertheless, our definition complies with the definition commonly used in registries, therefore allowing for comparison with the results of further studies. Third, in our analysis we could not differentiate between previous open and closed surgical procedures because our charts were not complete enough in this regard. This might be an important issue considering that past open surgical procedures are among the most influential risk factors for periprosthetic joint infection (Tan et al. 2018). Any past longitudinal scar included in the approach for pTKA might be unnoticed in the case of patient recall bias and if not reported in the charts at the time of arthroplasty. Finally, arthroscopy has been in use for diagnostic purposes, but it might well be that data entry for arthroscopy in the registry included a meniscectomy (external, internal) without mentioning this. Therefore, there is a risk of overestimation for arthroscopy and an underestimation for meniscectomy as a risk factor. Does history of previous surgeries influence the risk of revision of primary total knee arthroplasty? The crude risk of all-cause revision after pTKA among those with a history of previous knee surgery was about twice as high as among those without (8.3 vs. 4.3%). Baseline differences in age, sex, ASA score, BMI, smoking status, patellar resurfacing, type of tibial plateau, and surgery duration partly explained the higher risk; it was, however, still 1.5 times greater after adjusting for the baseline imbalances. A subgroup analysis considering only the first pTKA implanted revealed similar results. Patients who had previous surgery were sub-

714

stantially younger, more often men, had fewer comorbidities including obesity, and were more often ever smokers. Similarly, Lim et al. (2016) highlighted that pTKA after previous surgery was performed at a younger age (61 vs. 66 years), even younger than ours (67 vs. 73 years). However, besides the age difference and a similar BMI of 27, we do not know if their groups differed in baseline characteristics compared with our cohort. Indeed, the etiology of osteoarthritis might be different. Our 2 groups were composed of different patient populations: younger, more active, vs. older, sicker patients, less active, and more obese. Work and sports-related accidents and the kind of work itself are more common risk factors in the first group than in the second where obesity prevails, for instance. The chance of having previous surgery such as ligament reconstruction, meniscal repair, osteotomies, or fracture repair is therefore higher in the first group. What is the risk of revision according to the type of previous surgery? In this study, the risk of revision varied according to the type of previous surgery and it was lowest, with 4.1% (CI 1.7–9.5) 5-year cumulative failure rate in the case of previous osteotomy, and higher in the case of ligamentoplasty (7.1%), arthroscopy (7.9%), or previous osteosynthesis (8.3%). However, the confidence intervals around the estimates for the different types were large and overlapped considerably. The kind of surgery might alter knee mechanics. Typically, previous osteotomies around the knee, or posttraumatic conditions, make TKA technically more challenging in terms of implant positioning and ligament balancing. But their effect on the revision risk is not evident. A study from the New Zealand Joint Registry (Pearse et al. 2012), showed a 3-fold increased risk of early revision in patients with a history of osteotomies around the knee, compared with pTKA without previous surgery, but the risk was not adjusted. In a more recent study from the Danish Knee Arthroplasty Registry (El-Galaly et al. 2018), 10-year survival of pTKA after HTO was inferior (91% vs. 94%), although this could be explained by lower age and male sex rather than the osteotomy (adjusted HR of 1.2 vs. a crude HR of 1.7). Another study from the same group reported an increased risk of early and mid-term revision of pTKA in the setting of OA after fractures around the knee (El-Galaly et al. 2017). How does previous surgery influence specific causes of revision and the time of revision? The risk of revision after pTKA with previous surgery was about twice as high for any specific diagnosis, with aseptic loosening (2.1%) and infection (1.9%) being the most frequent (Table 5). The vast majority of patients in our cohort were homogeneously treated (80% PS, > 90% fixed bearing, 100% cemented pTKA), making implant-related factors unlikely to explain the difference in revision rates due to aseptic loosening. Both younger age (Khan et al. 2016) and a

Acta Orthopaedica 2021; 92 (6): 709–715

BMI over 35 (Abdel et al. 2015, Zingg et al. 2016) are known patient-related risk factors for revision, due to high activity levels and a higher mechanical load across the bone–cement interface, respectively. Nevertheless, our 2 groups were differently affected by these factors (those with previous surgery were younger, those without were more often obese), thus underlining the importance to control for previous surgery when comparing revision rates due to aseptic loosening. The higher risk of infection encountered in patients with a history of previous surgery might be explained by an intrinsic risk due to previous interventions, as reported in a recent meta-analysis, with a RR of 3.0 (CI 1.5–5.9) (Kunutsor et al. 2016), especially with open surgical procedures (Tan et al. 2018), as well as a history of resolved septic arthritis following surgery or prolonged surgical time. The amount of patellar resurfacing was lower in the group of patients with previous surgeries (62 vs. 69%) and partly explains the higher percentage of revision for femoro-patellar conflict in this group, as shown through the adjustment. A fair number of patients were revised because of a “painful” pTKA, with no obvious cause of failure identified. Residual pain after pTKA is not unusual, but high patient expectations, years-long chronic pain situation, and social/economic pressure to resume work might all play a central role. The higher prevalence of revision for instability in our series might be a consequence of post-traumatic conditions, consistent with a recent study from the Danish Knee Arthroplasty Register (El-Galaly et al. 2017). Concerning the timing of revision, their results are in line with our findings, with substantially more short-term revisions in the previous surgery group and no difference in the mid-term. In the long term, there might be a higher number of revisions in those with previous surgery. However, confidence intervals were overlapping. Conclusions In this large cohort, 21% of patients undergoing pTKA had a history of previous surgery. The difference in implant failure at 5 and 10 years was notable, and the 2 groups’ baseline differences only partly explained the increased risk of revision. It is important to advise patients that their knee history adversely influences the outcome of pTKA, with a 1.5 times higher risk of revision, especially in the short term. It would be of interest to know if data from other registries, including those with less frequent previous surgeries, supports our results. Future studies should analyze whether 1 vs. multiple surgeries prior to pTKA influences the survival differently and should focus on what causes of revision are related to a specific previous surgery, in an attempt to understand why this is and change our practice.

All authors were involved in planning the study. CB and AL retrieved and analyzed the data. HM and AL wrote the initial manuscript, and all authors revised the manuscript.

Acta Orthopaedica 2021; 92 (6): 709–715 715

The authors would like to thank Carole Bandi, Lamia Blatter-Sellak, and Heloise Corpataux for their continuous effort to keep the registry updated, all orthopedic surgeons who contributed to the knee registry since its creation in 1998 and Professor Thomas Perneger for his precious advice on methods. Acta thanks Annette W-Dahl for help with peer review of this study.

Abdel M P, Bonadurer G F, 3rd, Jennings M T, Hanssen A D. Increased aseptic tibial failures in patients with a BMI ≥35 and well-aligned total knee arthroplasties. J Arthroplasty 2015; 30(12): 2181-4. doi: 10.1016/j.arth.2015.06.057. Badawy M, Fenstad A M, Indrekvam K, Havelin L I, Furnes O. The risk of revision in total knee arthroplasty is not affected by previous high tibial osteotomy. Acta Orthop 2015; 86(6): 734-9. doi: 10.3109/17453674.2015.1060402. Brophy R H, Gray B L, Nunley R M, Barrack R L, Clohisy J C. Total knee arthroplasty after previous knee surgery: expected interval and the effect on patient age. J Bone Joint Surg Am 2014; 96(10): 801-5. doi: 10.2106/ JBJS.M.00105. El-Galaly A, Haldrup S, Pedersen A B, Kappel A, Jensen M U, Nielsen P T. Increased risk of early and medium-term revision after postfracture total knee arthroplasty. Acta Orthop 2017; 88(3): 263-8. doi: 10.1080/17453674.2017.1290479. El-Galaly A, Nielsen P T, Jensen S L, Kappel A. Prior high tibial osteotomy does not affect the survival of total knee arthroplasties: results from the Danish Knee Arthroplasty Registry. J Arthroplasty 2018; 33(7): 2131-5 e1. doi: 10.1016/j.arth.2018.02.076. Fine J P, Gray R J. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94(446): 496-509. doi: 10.1080/01621459.1999.10474144. Khan M, Osman K, Green G, Haddad F S. The epidemiology of failure in total knee arthroplasty: avoiding your next revision. Bone Joint J 2016; 98-B(1 Suppl A): 105-12. doi: 10.1302/0301-620x.98b1.36293.

Kunutsor S K, Whitehouse M R, Blom A W, Beswick A D. Patient-related risk factors for periprosthetic joint infection after total joint arthroplasty: a systematic review and meta-analysis. PLoS One 2016; 11(3): e0150866. doi: 10.1371/journal.pone.0150866. Lim J B, Loh B, Chong H C, Tan A H. History of previous knee surgery does not affect the clinical outcomes of primary total knee arthroplasty in an Asian population. Ann Transl Med 2016; 4(16): 303. doi: 10.21037/ atm.2016.08.15. Lübbeke A, Silman A J, Barea C, Prieto-Alhambra D, Carr A J. Mapping existing hip and knee replacement registries in Europe. Health Policy 2018; 122(5): 548-57. doi: 10.1016/j.healthpol.2018.03.010. Pearse A J, Hooper G J, Rothwell A G, Frampton C. Osteotomy and unicompartmental knee arthroplasty converted to total knee arthroplasty: data from the New Zealand Joint Registry. J Arthroplasty 2012; 27(10): 182731. doi: 10.1016/j.arth.2012.05.031. Piedade S R, Pinaroli A, Servien E, Neyret P. Is previous knee arthroscopy related to worse results in primary total knee arthroplasty? Knee Surg Sports Traumatol Arthrosc 2009; 17(4): 328-33. doi: 10.1007/s00167-0080669-9. Robertsson O, W-Dahl A. The risk of revision after TKA is affected by previous HTO or UKA. Clin Orthop Relat Res 2015; 473(1): 90-3. doi: 10.1007/s11999-014-3712-9. Tan T L, Maltenfort M G, Chen A F, Shahi A, Higuera C A, Siqueira M, Parvizi J. Development and evaluation of a preoperative risk calculator for periprosthetic joint infection following total joint arthroplasty. J Bone Joint Surg Am 2018; 100(9): 777-85. doi: 10.2106/ JBJS.16.01435. Tanner M A, Wong W H. The estimation of the hazard function from randomly censored data by the kernel method. Ann Stat 1983; 11:989-93. Viste A, Abdel M P, Ollivier M, Mara K C, Krych A J, Berry D J. Prior knee arthroscopy does not influence long-term total knee arthroplasty outcomes and survivorship. J Arthroplasty 2017; 32(12): 3626-31. doi: 10.1016/j.arth.2017.06.052. Zingg M, Miozzari H H, Fritschy D, Hoffmeyer P, Lübbeke A. Influence of body mass index on revision rates after primary total knee arthroplasty. Int Orthop 2016; 40(4): 723-9. doi: 10.1007/s00264-015-3031-0.

716

Acta Orthopaedica 2021; 92 (6): 716–721

Knee function 30 years after ACL reconstruction: a case series of 60 patients Thomas SÖDERMAN 1, Suzanne WERNER 3, Marie-Louise WRETLING 2, Mari HÄNNI 1, Christina MIKKELSEN 3, Anders SUNDIN 1, and Adel SHALABI 1 1 Department

of Radiology, Institution of Surgical Sciences, Uppsala University, Akademiska Hospital, Uppsala; 2 Department of Radiology, Karolinska University Hospital, Stockholm; 3 Department of Molecular Medicine and Surgery, Stockholm Sports Trauma Research Center, Karolinska Institutet, Stockholm, Sweden Correspondence: thomas.soderman@radiol.uu.se Submitted 2021-03-24. Accepted 2021-06-24.

Background and purpose — Until now, there have been no studies beyond 30 years after anterior cruciate ligament (ACL) reconstruction. We report knee function a mean 31 years after ACL reconstruction. Patients and methods — This cohort comprised a case series of 60 patients with a mean follow-up of 31 years (28– 33) after ACL reconstruction. Patients were evaluated with the International Knee Documentation Committee (IKDC) objective assessment, Knee injury Osteoarthritis Outcome Score (KOOS), Tegner Activity Scale, radiography, and MRI. Results — 30 patients showed an intact ACL graft and 30 a ruptured or missing ACL graft. 40 patients had osteoarthritis in the tibiofemoral compartment and 24 patients in the patellofemoral compartment. Patients with intact ACL grafts scored higher than those with ruptured or missing ACL grafts when it comes to KOOS Sport/Rec. The Hodges Lehmann estimated median difference between groups was 15 (95% CI 0–35). The KOOS scores were lower in the group with ruptured or missing ACL grafts when compared with a healthy-knee reference group of males in terms of Pain, mean difference –8 (CI –15 to –1), Symptoms, mean difference –18 (CI –27 to –9), and Sport/Rec, mean difference –21 (CI –34 to –8). In the group with intact ACL grafts, the KOOS score was lower than a healthy-knee reference group of males in terms of Symptoms, mean difference –12 (CI –21 to –3). Scores for all subgroups of KOOS were higher in patients without osteoarthritis. The IKDC overall clinical assessment outcome was worse in patients with a ruptured or missing ACL graft. The Hodges Lehmann estimated median difference between groups was 1 (CI 0–1). Interpretation — Patients with an intact ACL graft reported higher sports activity and recreation, as measured with KOOS, than patients with a ruptured or missing ACL graft. Patients with severe osteoarthritis reported lower sports activity and recreation, as measured with KOOS.

The anterior cruciate ligament (ACL) has a reported injury rate between 78 and 81/100,000 per individual and year (Nordenvall et al. 2012). In approximately half of the patients osteoarthritis (OA) is present 10–20 years after the ACL injury (Lohmander et al. 2007). It seems as if the development of OA does not differ whether patients have undergone ACL reconstruction or nonoperative treatment (Nordenvall et al. 2012, Smith et al. 2014, Lien-Iversen et al. 2019). The development of OA in the injured joint is caused by intra-articular pathogenic processes initiated at the time of injury, combined with long-term changes in dynamic joint loading (Lohmander et al. 2007). Patients with ACL disruption sustain an acute chondral lesion that affects the overall cartilage homeostasis, resulting in a global degradation of the joint (Potter et al. 2012). There is increased anterior tibial subluxation in patients with failed ACL reconstructions compared with patients with acute ACL disruptions (Tanaka et al. 2013). Furthermore, abnormal tibiofemoral relationships alter the knee joint cartilage’s stress distributions, predisposing to degenerative changes (Andriacchi et al. 2004). Moreover, meniscal tears occur frequently in the setting of anterior cruciate ligament ACL ruptures, where medial meniscal tears are of most importance because the posterior horn of the medial meniscus is an essential secondary stabilizer against anterior tibial translation (Papageorgiou et al. 2001). A study by Shelbourne et al. (2017) is one of a few studies beyond 20 years after ACL reconstruction reporting radiographic, physical, and patient-reported outcomes. We report the outcome of ACL reconstruction, approximately 30 years after surgery, in terms of knee function evaluated with clinical assessments, radiological examinations, and patient self-reported outcomes. The hypothesis was that patients with ruptured or missing ACL grafts would report lower outcome scores as measured with KOOS than patients with intact ACL grafts. Furthermore, we hypothesized that

Acta Orthopaedica 2021; 92 (6): 716–721 717

Patients with ACL reconstruction 1968–1973 n = 134 Dropouts (n = 30): – no response to letters, 14 – dead, 11 – moved out of the country, 4 – multiple knee injuries, 1 Follow-up 2001–2002 n = 104 Excluded (n = 44): – no clinical follow-up, KOOS only, 15 – surgery to the other knee, 13 – radiographs missing, 10 – knee arthroplasty, 3 – ACL grafts ungradable, 2 – clinical data missing, 1 Patients analyzed n = 60

Figure 1. Flowchart of study patients.

Figure 2. Schematic illustration of how the drill channel through the tibia was designed and how the tendon was drawn through it. Courtesy Ejnar Eriksson, MD.

patients with radiological OA would report lower outcome scores as measured with KOOS than those without OA.

Patients and methods This retrospective case series on clinical and radiological assessments was performed 28–33 years after ACL reconstruction. Patients 134 consecutive patients, who underwent ACL reconstruction at the Karolinska University Hospital between 1968 and 1973, were included. Primarily, all patients had been treated nonoperatively for their ACL rupture but were later referred to surgery because their knee joint instability prevented them from returning to their desired physical activities. Some of the patients showed previous injuries to their knee and entered the study diagnosed with medial collateral ligament injury, meniscal tears, and/or OA. In the first follow-up of this cohort in 1978, 87 patients were included, and the time from original injury to reconstructive surgery varied from 1 to 245 months (median 29 months; mean 41 months) (Johnson et al. 1984). At the present follow-up 2001/2002, 89 out of the original 134 patients were evaluated at the Karolinska University Hospital. 29 of these patients were excluded, and 60 patients (55 men) remained for the present assessment (Figure 1). The mean age was 27 years (SD = 6) at index surgery and 58 years (SD = 6) at follow-up. 35 patients had injured their right knee and 25 their left knee. Surgical technique The ACL reconstruction was performed according to a standardized method developed at the Karolinska University Hos-

Figure 3. 4 to 5 weeks postoperatively, a partly mobile cast brace was applied that allowed 40° of flexion, from 20° to 60°. Courtesy Ejnar Eriksson, MD.

pital by Gillquist et al. in 1966 (Gillquist et al. 1971). They used the medial third of the patellar tendon, dissecting off two-thirds of the ligament that covered the patella, the patellar retinaculum. A hole was made from the tibial tubercle into the knee joint (close to the normal tibial attachment of the ACL), and the tendon flap was pulled into the knee joint. By a separate incision over the outside of the lateral condyle, two holes for sutures were drilled. A shallow hole at the normal posterior insertion of the ACL was also excavated. The sutures through the ligament were placed at varying sites on the ligament to facilitate stretching of the ligament flap when the sutures were tied one by one (Figure 2). Each bundle of sutures was pulled out using an extra thread, and the operation was then completed. In those cases where associated surgery was performed simultaneously with index surgery, meniscus resection and repair of the medial collateral ligament were performed (Gillquist et al. 1971). Postoperative treatment A posterior plaster sprint was used for 1 week. The splint was then changed for a partially mobile cast-brace, allowing the knee to flex 20°–60°, and the patients were allowed to put weight on the operated leg (Figure 3). The cast-brace was used for 4 to 5 weeks postoperatively. Physiotherapy started immediately after surgery and continued for 6 continuous weeks while the cast-brace was still in place. The goal of rehabilitation was to improve the range of motion of the knee joint and increase thigh muscle strength, including balance and coordination exercises (Eriksson 1976). Clinical outcome measures Objective evaluations were performed using the 2000 International Knee Documentation Committee (IKDC) objective

718

form (Hefti et al. 1993). Clinical evaluations were conducted by an orthopedic surgeon with extensive experience in performing objective measures of the knee joint. Subjective evaluations utilized the KOOS (Roos et al. 1998) and Tegner Activity Scale (Tegner and Lysholm 1985). The Tegner Activity Scale was used in a somewhat modified version due to the inclusion of certain new sports such as floorball and martial arts that had become popular since 1985 when the Tegner Activity Scale was published. The KOOS subscale Quality of Life will be reported elsewhere. Radiological examinations At the present follow-up 2001/2002, all patients underwent radiographic and MRI examinations of the index knee. Weightbearing radiographs in anteroposterior, lateral, and a skyline view of the patellofemoral compartment were acquired (Ahlbäck and Rydberg 1980). MRI was performed on a low field, 0.2 T scanner (Arthroscan; Esaote Group, Genoa, Italy), to obtain sagittal (T2 STIR 5 mm, T2 5 mm, T1 5 mm, T1 4 mm, and T1 3 mm) and coronal (T2 STIR 5 mm and T1 5 mm) images. The radiological examinations were read in consensus by 2 senior consultants in musculoskeletal radiology. The radiological examinations were assessed to determine OA degree using the Kellgren–Lawrence classification. Radiographic tibiofemoral and patellofemoral OA was defined as Kellgren–Lawrence ≥ 2 in any of the compartments. MRI was used to assess the menisci and the structural integrity of the ACL graft. The menisci were classified as 1 (normal), 2 (small/defect), 3 (rupture), or 4 (missing). The ACL graft was graded as intact, ruptured, missing, or impossible to evaluate due to artifacts. Statistics Data was analyzed using GraphPad Prism version 7.00 for Macintosh (GraphPad Software, LaJolla, CA, USA; www. graphpad.com). Descriptive statistics such as frequency, mean (SD), median (range) were employed. A Mann–Whitney U-test was used for group comparisons in terms of intact and ruptured or missing ACL grafts, respectively. Mann–Whitney was also used for comparison between groups with OA and without OA. To define the magnitude of effects, an estimate of the median difference between groups, together with the 95% confidence interval (CI), was calculated with the use of the Hodges–Lehmann approach. Based on KOOS, Student’s t-test was used comparing knee function of the study cohort with a healthy Swedish population. To define the magnitude of effects, the mean difference between groups, together with the 95% confidence interval, was calculated. Spearman’s test was used for correlation analysis. Statistical significance was set at a p-value of < 0.05. 3 patients had missing data in single variables due to incomplete filling out of forms (n = 1) and clinical examination results not reported (n = 2). 4 patients had missing data in terms of the Tegner Activity Scale. The existing values were

Acta Orthopaedica 2021; 92 (6): 716–721

used in the statistical analysis. No mathematical correction was made for multiple comparisons. Ethics, funding, and potential conflicts of interest Ethical approval was obtained from the Regional Ethics Committee in Stockholm, Sweden (Dnr 2016/2108-31/4). All patients gave informed consent to participate. No funding has been obtained for this study. The authors declare no potential conflicts of interest.

Results International Knee Documentation Committee The results of IKDC were available for all 60 patients (Table 1). The IKDC overall clinical assessment outcome was worse in patients with a ruptured or missing ACL graft. The Hodges Lehmann estimated median difference between groups was 1 (CI 0–1). There was no statistical group difference regarding extension, flexion, or effusion. Anterior knee laxity based on the Lachman test was higher in patients with ruptured or missing ACL grafts than those with intact ACL grafts. The Hodges Lehmann estimated median difference between groups was 1 (CI 0–2). Knee injury Osteoarthritis Outcome Score All 60 patients completed the KOOS (Table 2). Patients with intact ACL grafts scored higher than those with ruptured or missing ACL grafts when it comes to KOOS Sport/Rec. The Hodges Lehmann estimated median difference between groups was 15 (CI 0–35). There was no statistical group difference in terms of Pain, Symptoms, or ADL. The KOOS scores were lower in the group with ruptured or missing ACL grafts when compared with a healthy-knee reference group of males (Paradowski et al. 2006) in terms of Pain, mean difference –8 (CI –15 to –1), Symptoms, mean difference –18 (CI –27 to –9), and Sport/Rec, mean difference –21 (CI –34 to –8). The KOOS scores were lower in the group with intact ACL grafts than a healthy-knee reference group of males (Paradowski et al. 2006) in terms of Symptoms, mean difference –12 (CI –21 to –3). On the other subscales of KOOS, there were no statistical differences. When the patients were divided into those with OA (n = 40) and without OA (n = 20), in the tibiofemoral compartment, patients without OA scored higher in terms of Pain, median difference 8 (CI 3–17), Symptoms, median difference 17 (CI 7–29), ADL median difference 4 (CI 0–9), and Sport/ Rec, median difference 20 (CI 5–35). When the patients were divided into those with OA (n = 24) and without OA (n = 36) in the patellofemoral compartment, Pain, median difference 6 (CI 0–19), Symptoms, median difference 14 (CI 4–29), ADL, median difference 4 (CI 0–12), and Sport/Rec, median difference 20 (CI 5–35) were higher in the group without OA.

Acta Orthopaedica 2021; 92 (6): 716–721 719

Table 1. Clinical outcome of knee function according to IKDC Nearly Severely Factor Normal normal Abnormal abnormal Effusion, n = 59 Intact ACL 22 6 0 Ruptured or missing ACL 19 10 0 Passive motion deficit extension, n = 59 Intact ACL 11 13 4 Ruptured or missing ACL 12 7 7 Passive motion deficit flexion, n = 60 Intact ACL 10 15 5 Ruptured or missing ACL 6 13 10 a Ligament examination, n = 60 Intact ACL 22 1 7 Ruptured or missing ACL 9 2 19 IKDC overall, n = 60 Intact ACL 4 14 10 Ruptured or missing ACL 1 4 18 a

1 1 2 3 0 1 0 0 2 7

Lachman manual max, 25° flexion

Table 2. KOOS scores for the groups with intact and ruptured or missing ACL grafts. Values are mean (SD) KOOS subscales

Intact ACL (n = 30)

Ruptured a (n = 30)

Pain Symptoms ADL Sport/Rec

86 (18) 76 (22) 92 (13) 68 (24)

80 (16) 70 (22) 87.5 (14) 52 (32)

Reference group b p-value c 88 (17) 88 (17) 86 (19) 73 (30)

0.06 0.2 0.2 0.04

a or missing b Reference

ACL population (n = 88 except for Sport/Rec with 1 missing value) of 55- to 74-year-old men added for comparison from Paradowski et al. 2006. c Intact ACL graft versus ruptured or missing ACL graft.

A weak association was found between Sport/Rec and OA in the medial tibiofemoral compartment (r = –0.25, p = 0.05), lateral tibiofemoral compartment (r = –0.26, p = 0.04), and the patellofemoral compartment (r = –0.35, p = 0.007). Tegner Activity Scale 56 patients completed the Tegner Activity Scale, while 4 did not (Table 3). The median activity level was 4, ranging from 2 to 7 for patients with intact ACL graft and 3 to 7 for patients with a ruptured or missing ACL graft. 11 of the patients with intact ACL graft and 14 of those with ruptured or missing ACL graft reported an activity level of recreational sports or higher (≥ level 5). Radiographic assessment OA changes were assessed in the medial tibiofemoral compartment, the lateral tibiofemoral compartment, and the patellofemoral compartment (Table 4). 40 patients had OA in the tibiofemoral compartment and 24 patients in the patellofemoral compartment. In patients with

Table 3. Tegner Activity Scale for patients with intact and ruptured or missing ACL grafts Activity Intact ACL graft level (n = 27) 9–10 7–8 5–6 3–4 1–2 0

Ruptured or missing ACL graft (n = 29)

0 4 7 13 3 0

0 5 9 15 0 0

Table 4. Distribution of osteoarthritis for the groups with intact and ruptured or missing ACL grafts using the Kellgren–Lawrence classification Intact ACL graft (n = 30) K–L grade: 0 and 1 2 3 4 Patellofemoral Medial tibiofemoral Lateral tibiofemoral

21 4 5 0 20 3 7 0 22 4 3 1

Ruptured or missing ACL graft (n = 30) 0 and 1 2 3 4 15 11 3 1 7 8 8 7 19 6 1 4

intact ACL grafts, 16 had OA in the tibiofemoral compartment and 9 in the patellofemoral compartment. In patients with ruptured or missing ACL grafts, 23 had OA in the tibiofemoral compartment and 15 in the patellofemoral compartment. MRI assessment Of the 60 patients, 30 had intact ACL grafts, and in 30 patients the ACL graft was either ruptured or missing. Before or in connection with the index surgery, 15 out of 60 patients had been treated with meniscus resection of the medial meniscus and 4 of the lateral meniscus (Table 5).

Discussion Our main finding a mean 31 years after ACL reconstruction was that patients with an intact ACL graft reported higher sports activity and recreation, as measured with KOOS, than patients with a ruptured or missing ACL graft. Remarkably low graft survival and that patients with ruptured ACL grafts showed substantially more OA of the medial tibiofemoral compartment than in those with an intact graft has been reported previously (Söderman et al. 2019). In a recently published 20-year follow-up study (van Yperen et al. 2018), similar KOOS results were reported for operative and nonoperative treatment of ACL ruptures. However, the clinical outcome of that study assessed with IKDC was better for surgical treatment. In our study, with exclusively surgical patients, higher KOOS Sport/Rec scores and better IKDC

720

Acta Orthopaedica 2021; 92 (6): 716–721

Table 5. Distribution of meniscal injuries for the groups with intact and ruptured or missing ACL grafts Ruptured or Meniscal Intact ACL graft (n = 30) missing ACL graft (n = 30) injury Medial Lateral Medial Lateral grade meniscus meniscus meniscus meniscus 1 2 3 4

9 9 5 7

17 2 9 2

2 19 1 8

14 3 11 2

values were found for patients with intact ACL grafts than those with ruptured or missing ACL grafts, which may point to the importance of an intact ACL graft. This is supported by Granan et al. (2015), who reported an association between impaired knee function measured with KOOS and a prospective ACL graft failure. Furthermore, the functional results assessed with IKDC are supported by Kessler et al. (2008), who showed a clear clinical outcome advantage in favor of ACL reconstruction compared with nonoperative treatment. It must be considered that the IKDC total is determined by the worst individual parameter, which in our study was the ligament test. A large number of the patients in the study by Kessler et al. with evident objective anteroposterior instability (IKDC B, C, D) were also almost symptom-free and physically active at a high level. When interpreting patient-reported outcomes, it is essential to do so in a clinically meaningful manner. The measurement of the patient’s acceptable symptom state (PASS) is an accepted way of doing that (Svantesson et al. 2020). Half of the patients in the group with intact ACL graft and one-third of the patients with ruptured ACL grafts scored KOOS Sport/ Rec as 75 or more, which constitutes a previously identified threshold for the patient acceptable symptom state 1 to 6 years after ACL reconstruction (Muller et al. 2016). The remainder of our patients scored below 75. This is in line with a previous long-term follow-up mean 16 years after ACL reconstruction (Hamrin Senorski et al. 2019), where half of the patients had a PASS value above the cutoff value. Interestingly, in patients with intact ACL grafts, the proportion of patients with PASS value above the cutoff value was nearly the same as in the study by Hamrin et al. (2019), even though the follow-up time was significantly longer in our study and the mean age was higher. The reason for this is unclear, but it could partly be explained by the fact that the cutoff value for PASS in our study was KOOS but IKDC in the study by Hamrin et al. Moreover, the PASS cutoff value could possibly vary with age due to different activity levels in different ages (Briggs et al. 2009). It has been suggested that an 8–10-point change in KOOS constitutes a clinically relevant difference (Roos and Lohmander 2003). In the subscales Symptoms and Sport/Rec in the group with ruptured or missing ACL graft, the mean KOOS

scores were more than 10 points lower than in the male reference group of KOOS (Paradowski et al. 2006). Furthermore, the difference between KOOS scores for healthy controls and for patients with intact ACL grafts was less than 10 points in all KOOS subgroups besides Symptoms. This further emphasizes the importance of an intact ACL graft. Patients with radiologic OA reported lower KOOS scores than those without OA. This is in line with previous findings of associations between severe OA and KOOS scores in terms of Pain, Symptoms, ADL, Sport/Rec, and QOL after ACL reconstruction (Oiestad et al. 2011). It has been shown that loss of knee extension and/or loss of flexion increases the odds of developing moderate to severe OA (Shelbourne et al. 2017). This is in line with our results, where 36 of 60 patients had abnormal knee extension, and 40 of 60 had OA in the tibiofemoral compartment. Patients with intact and ruptured or missing ACL grafts showed similar median Tegner activity levels (4), the reasons for which are somewhat unclear. Taken together, the majority of the patients had an activity level between 3 and 6, corresponding to activities during daily living or recreational sports, and the average age at follow-up was 56 years, which may imply that age is essential or even determines activity level. This is consistent with a previous study, which showed that healthy persons’ Tegner activity level was inversely correlated with age (Briggs et al. 2009). The following study limitations should be mentioned. These are the low rate—60/134—of follow-up, the heterogeneous cohort, and that the clinical evaluations were performed almost 20 years ago with the risk of lost data and the development of evaluation methods. Furthermore, some of the patients had previous injuries to their knee and entered the study diagnosed with collateral ligament injury, meniscal tears, and/or OA. Another limitation is the lack of a control group; all our patients had had surgery. A further limitation is that MRI was performed with a low field strength magnet (0.2 T). The strength of the study is that it presents data from a long-term follow-up after ACL reconstruction, mean 31 years, which to our knowledge is unique. Furthermore, the clinical assessments were unbiased to some extent because they were done by an orthopedic surgeon from another institution who was not involved in the ACL reconstruction. The interpretation of our results cannot without reservation be generalized and transformed to today’s population of ACL-reconstructed knees, as surgical techniques have been developed and improved continuously since the beginning of this study. Current surgery is performed arthroscopically and the graft is fully resected instead of leaving the insertion of the tendon in the tibial tuberosity intact. Anatomical ACL reconstructions are increasingly performed. However, it is also noteworthy that over time different grafts are being used, and it cannot be excluded that this might lead to a different result. In addition, the studied population is a group where nonoperative treatment was not sufficient to regain knee joint

Acta Orthopaedica 2021; 92 (6): 716–721 721

stability and desired knee function. The time between injury and surgery might have been longer than in current clinical practice. Therefore, it is not unlikely that future long-term outcomes after surgery applying modern techniques will result in a better clinical outcome. In conclusion, a mean 31 years after ACL reconstruction, patients with an intact ACL graft reported higher sports activity and recreation than those with a ruptured or missing ACL graft, and patients with severe OA reported lower sports activity and recreation (KOOS).

The authors would like to thank professor Ejnar Eriksson, MD, who initiated the present study in 1968–1973, for supplying them with the original data of this cohort. All authors took part in the planning, performance, and reporting of this study. Acta thanks Martin Lind and Asbjørn Årøen for help with peer review of this study.

Ahlbäck S, Rydberg J. [X-ray classification and examination technics in gonarthrosis]. Läkartidningen 1980; 77(22): 2091-3, 6. Andriacchi T P, Mündermann A, Smith R L, Alexander E J, Dyrby C O, Koo S. A framework for the in vivo pathomechanics of osteoarthritis at the knee. Ann Biomed Eng 2004; 32(3): 447-57. doi: 10.1023/b:abme.00000 17541.82498.37. Briggs K K, Steadman J R, Hay C J, Hines S L. Lysholm score and Tegner activity level in individuals with normal knees. Am J Sports Med 2009; 37(5): 898-901. doi: 10.1177/0363546508330149. Eriksson E. Sports injuries of the knee ligaments: their diagnosis, treatment, rehabilitation, and prevention. Med Sci Sports 1976; 8(3): 133-44. Gillquist J, Liljedahl S O, Lindvall H. Reconstruction for old rupture of the anterior cruciate ligament: a follow-up study. Injury 1971; 2(4): 271-8. doi: 10.1016/s0020-1383(71)80076-1. Granan L P, Baste V, Engebretsen L, Inacio M C. Associations between inadequate knee function detected by KOOS and prospective graft failure in an anterior cruciate ligament-reconstructed knee. Knee Surg Sports Traumatol Arthrosc 2015; 23(4): 1135-40. doi: 10.1007/s00167-014-2925-5. Hamrin Senorski E, Sundemo D, Svantesson E, Sernert N, Kartus J T, Karlsson J, Samuelsson K. Preoperative and intraoperative predictors of long-term acceptable knee function and osteoarthritis after anterior cruciate ligament reconstruction: an analysis based on 2 randomized controlled trials. Arthroscopy 2019; 35(2): 489-99. doi: 10.1016/j.arthro.2018.07.009. Hefti F, Müller W, Jakob R P, Stäubli H U. Evaluation of knee ligament injuries with the IKDC form. Knee Surg Sports Traumatol Arthrosc 1993; 1(3-4): 226-34. doi: 10.1007/bf01560215. Johnson R J, Eriksson E, Häggmark T, Pope M H. Five- to ten-year follow-up evaluation after reconstruction of the anterior cruciate ligament. Clin Orthop Relat Res 1984; (183): 122-40. Kessler M A, Behrend H, Henz S, Stutz G, Rukavina A, Kuster M S. Function, osteoarthritis and activity after ACL-rupture: 11 years follow-up results of conservative versus reconstructive treatment. Knee Surg Sports Traumatol Arthrosc 2008; 16(5): 442-8. doi: 10.1007/s00167-008-0498-x. Lien-Iversen T, Morgan D B, Jensen C, Risberg M A, Engebretsen L, Viberg B. Does surgery reduce knee osteoarthritis, meniscal injury and subsequent complications compared with non-surgery after ACL rupture with at least 10 years follow-up? A systematic review and meta-analysis. Br J Sports Med 2019. doi: 10.1136/bjsports-2019-100765.

Lohmander L S, Englund P M, Dahl L L, Roos E M. The long-term consequence of anterior cruciate ligament and meniscus injuries: osteoarthritis. Am J Sports Med 2007; 35(10): 1756-69. doi: 10.1177/0363546507307396. Muller B, Yabroudi M A, Lynch A, Lai C L, van Dijk C N, Fu F H, Irrgang J J. Defining thresholds for the patient acceptable symptom state for the IKDC Subjective Knee Form and KOOS for patients who underwent ACL reconstruction. Am J Sports Med 2016; 44(11): 2820-6. doi: 10.1177/0363546516652888. Nordenvall R, Bahmanyar S, Adami J, Stenros C, Wredmark T, Fellander-Tsai L. A population-based nationwide study of cruciate ligament injury in Sweden, 2001–2009: incidence, treatment, and sex differences. Am J Sports Med 2012; 40(8): 1808-13. doi: 10.1177/0363546512449306. Oiestad B E, Holm I, Engebretsen L, Risberg M A. The association between radiographic knee osteoarthritis and knee symptoms, function and quality of life 10–15 years after anterior cruciate ligament reconstruction. Br J Sports Med 2011; 45(7): 583-8. doi: 10.1136/bjsm.2010.073130. Papageorgiou C D, Gil J E, Kanamori A, Fenwick J A, Woo S L, Fu F H. The biomechanical interdependence between the anterior cruciate ligament replacement graft and the medial meniscus. Am J Sports Med 2001; 29(2): 226-31. doi: 10.1177/03635465010290021801. Paradowski P T, Bergman S, Sunden-Lundius A, Lohmander L S, Roos E M. Knee complaints vary with age and gender in the adult population: population-based reference data for the Knee injury and Osteoarthritis Outcome Score (KOOS). BMC Musculoskelet Disord 2006; 7: 38. doi: 10.1186/1471-2474-7-38. Potter H G, Jain S K, Ma Y, Black B R, Fung S, Lyman S. Cartilage injury after acute, isolated anterior cruciate ligament tear: immediate and longitudinal effect with clinical/MRI follow-up. Am J Sports Med 2012; 40(2): 276-85. doi: 10.1177/0363546511423380. Roos E M, Roos H P, Lohmander L S, Ekdahl C, Beynnon B D. Knee Injury and Osteoarthritis Outcome Score (KOOS): development of a selfadministered outcome measure. J Orthop Sports Phys Ther 1998; 28(2): 88-96. doi: 10.2519/jospt.1998.28.2.88. Roos E M, Lohmander L S. The Knee injury and Osteoarthritis Outcome Score (KOOS): from joint injury to osteoarthritis. Health Qual Life Outcomes 2003; 1: 64. doi: 10.1186/1477-7525-1-64. Shelbourne K D, Benner R W, Gray T. Results of anterior cruciate ligament reconstruction with patellar tendon autografts: objective factors associated with the development of osteoarthritis at 20 to 33 years after surgery. Am J Sports Med 2017; 45(12): 2730-8. doi: 10.1177/0363546517718827. Smith T O, Postle K, Penny F, McNamara I, Mann C J. Is reconstruction the best management strategy for anterior cruciate ligament rupture? A systematic review and meta-analysis comparing anterior cruciate ligament reconstruction versus non-operative treatment. Knee 2014; 21(2): 462-70. doi: 10.1016/j.knee.2013.10.009. Söderman T, Wretling M L, Hanni M, Mikkelsen C, Johnson R J, Werner S, Sundin A, Shalabi A. Higher frequency of osteoarthritis in patients with ACL graft rupture than in those with intact ACL grafts 30 years after reconstruction. Knee Surg Sports Traumatol Arthrosc 2019. doi: 10.1007/ s00167-019-05726-6. Svantesson E, Hamrin Senorski E, Webster K E, Karlsson J, Diermeier T, Rothrauff B B, Meredith S J, Rauer T, Irrgang J J, Spindler K P, Ma C B, Musahl V. Clinical outcomes after anterior cruciate ligament injury: Panther Symposium ACL injury clinical outcomes consensus group. Knee Surg Sports Traumatol Arthrosc 2020; 28(8): 2415-34. doi: 10.1007/ s00167-020-06061-x. Tanaka M J, Jones K J, Gargiulo A M, Delos D, Wickiewicz T L, Potter H G, Pearle A D. Passive anterior tibial subluxation in anterior cruciate ligament-deficient knees. Am J Sports Med 2013; 41(10): 2347-52. doi: 10.1177/0363546513498995. Tegner Y, Lysholm J. Rating systems in the evaluation of knee ligament injuries. Clin Orthop Relat Res 1985; (198): 43-9. van Yperen D T, Reijman M, van Es E M, Bierma-Zeinstra S M A, Meuffels D E. Twenty-year follow-up study comparing operative versus nonoperative treatment of anterior cruciate ligament ruptures in high-level athletes. Am J Sports Med 2018; 46(5): 1129-36. doi: 10.1177/0363546517751683.

722

Acta Orthopaedica 2021; 92 (6): 722–727

Influence of body mass index and age on day-of-surgery discharge, prolonged admission, and 90-day readmission after fast-track unicompartmental knee arthroplasty Christian Bredgaard JENSEN 1, Anders TROELSEN 1, Pelle Baggesgaard PETERSEN 2,3, Christoffer Calov JØRGENSEN 2,3, Henrik KEHLET 2,3, and Kirill GROMOV 1,3, on behalf of the Centre for FastTrack Hip and Knee Replacement Collaborative Group a 1 Department of Orthopaedic Surgery, Clinical Orthopaedic Research Hvidovre, Copenhagen University Hospital Hvidovre, Hvidovre 2650, Denmark; 2 Section for Surgical Pathophysiology, Rigshospitalet, Copenhagen 2100, Denmark; 3 Centre for Fast-track Hip and Knee Arthroplasty, Rigshospitalet,

Copenhagen 2100, Denmark Centre for Fast-track Hip and Knee Replacement Collaborative Group: Frank MADSEN, Department of Orthopedics, Aarhus University Hospital, Aarhus, Denmark; Torben Bæk HANSEN, Department of Orthopedics, Regional Hospital Holstebro and University of Aarhus, Holstebro, Denmark; Mogens LAURSEN, Aalborg University Hospital Northern Orthopaedic Division, Aalborg, Denmark; Lars Tambour HANSEN, Department of Orthopedics, Sydvestjysk Hospital Esbjerg/Grindsted, Grindsted, Denmark; Per KJÆRSGAARD‑ANDERSEN, Department of Orthopedics, Vejle Hospital, Vejle, Denmark; Mikkel Rathsach ANDERSEN, Department of Orthopedics, Gentofte University Hospital, Copenhagen, Denmark; Niels Harry KRARUP , Department of Orthopedics, Viborg Hospital, Viborg, Denmark; and Henrik PALM, Department of Orthopaedic Surgery, Copenhagen University Hospital Bispebjerg, Copenhagen, NV, Denmark. Correspondence: christian.bredgaard.jensen@regionh.dk / chrisbredgaard@hotmail.com Submitted 2021-06-15. Accepted 2021-07-21. a The

Background and purpose — The indications for unicompartmental knee arthroplasty (UKA) have become less restrictive and, today, high age and high BMI are not considered contraindications by many surgeons. While the influence of these patient characteristics on total knee arthroplasty is well documented, evidence on UKA is lacking. We investigated the effect of BMI and age on day of surgery (DOS) discharge, prolonged admission, and 90-day readmission following UKA surgery. Patients and methods — This retrospective cohort study included 3,897 UKA patients operated on between 2010 and 2018 in 8 fast-track arthroplasty centers. Patients were divided into 5 BMI groups and 5 age groups. Differences between groups in the occurrence of DOS discharge, prolonged admission > 2 days, and 90-day readmission was investigated using a chi-square test and mixed-effect models adjusted for patient characteristics using surgical center as a random effect. Results — Median LOS was 1 day. DOS discharge was achieved in 26% of patients with no statistically significant differences between BMI groups. DOS discharge was less likely in UKA patients aged > 70 years (age 71–80; odds ratio [OR] 0.7 [95% CI 0.6–0.9]). Prolonged admission was not affected by BMI or age in the adjusted analysis. 90-day readmission was more likely in patients with BMI > 35 (OR 1.9 [CI 1.1–3.1]) and patients aged 71–80 (OR 1.5 [CI 1.1–2.1]).

Interpretation — Age > 70 years decreased the likelihood of DOS discharge after UKA. High BMI as well as advanced age increased the likelihood of 90-day readmission. This should be noted by surgeons operating on patients with high BMI and age.

The indications for unicompartmental knee arthroplasty (UKA) as treatment for osteoarthritis (OA) have become less restrictive in terms of age and weight. Early contraindications included age < 60 years and weight > 82 kg (Kozinn and Scott 1989). However, recent studies report that revision rates and patient-reported outcomes are not worse in such patients (Pandit et al. 2011, van der List et al. 2016, Hamilton et al. 2017). Current indications focus solely on the pathoanatomy of the knee OA (Goodfellow et al. 1988, Hamilton et al. 2017). Despite being informed of increased risk of certain postoperative complications in high BMI patients and young/old patients, these patients are increasingly undergoing UKA surgery as well as knee arthroplasty in general (Price et al. 2018, Henkel et al. 2019). While length of stay (LOS) and readmissions are not the primary factors when determining indications/contraindications of arthroplasty procedures, they do affect patient satisfaction, logistics and cost-effectiveness (Reilly et al. 2005, Molloy et al. 2017).

Acta Orthopaedica 2021; 92 (6): 722–727 723

The few studies investigating the effect of BMI and age on LOS and readmission after UKA have varying conclusions. Some studies have associated higher BMI with increased risk of prolonged admission as well as short-term complications, while others find no such association (Otero et al. 2016, Plate et al. 2017, Sephton et al. 2020). Likewise, day of surgery (DOS) discharge is reported to be less likely in older patients while some do not find this association (Haughom et al. 2015, Matsumoto et al. 2020). Due to the usage of UKA in patients with high BMI and advanced age, it is important to investigate the effect of BMI and age on the postoperative course after UKA. We therefore investigated the association between BMI and age and the proportion of UKA patients with DOS discharge, prolonged admission, and readmission within 90 days of surgery in a prospective unselected multicenter fast-track setup.

Patients and methods This study was an observational cohort study investigating primary UKA procedures. All data was acquired from the Lundbeck Foundation Centre for Fast-track Hip and Knee Replacement Database (LCDB). A previous study regarding this cohort of UKA patients from LCDB as well as multiple studies describing the data methodology within the LCDB have been published (Jørgensen and Kehlet 2013, Gromov et al. 2020). Data regarding UKA patients from the LCDB is gathered from 8 surgical centers in Denmark between 2010 and 2018. All included centers have implemented a fast-track setup with intended spinal anesthesia, multimodal opioid-sparing analgesia, high-dose preoperative corticosteroids (Kehlet and Lindberg-Larsen 2018), preoperative intravenous tranexamic acid, no drains, early mobilization with full weight-bearing, and discharge to the patient’s own home. All centers use functional discharge criteria (Husted 2012). Data on length of stay and readmissions from the LCDB was obtained from the Danish National Patient Registry (DNPR), from which such data is considered complete (> 99%) (Schmidt et al. 2015), as well as chart-review in case of LOS > 4 days, 90-day readmission, or death. Preoperative comorbidity was evaluated using preoperative nurse-assisted patient-reported questionnaires and data from the Danish National Database of Reimbursed Prescriptions (DNDRP) (Johannesdottir et al. 2012), which provided information on use of anticoagulants, diabetic medication, antihypertensive treatment, and psychotropic treatment (Jørgensen et al. 2016). Length of stay was calculated as the number of nights spent in hospital. DOS discharge was defined as LOS = 0 days and prolonged admission as LOS > 2 days. Readmissions were defined as unplanned admissions with a LOS ≥ 1 day within 90 days surgery. The exclusion criteria for the LCDB were age < 18, nonelective procedures, simultaneous bilateral procedures, addi-

tional major arthroplasty within 90 days, and surgery due to congenital disorders, infections, or cancer. Patients who did not answer the preoperative questionnaire were also excluded. 3,927 primary medial or lateral UKA procedures in 3,623 patients performed between 2010 and 2018 with a completed questionnaire on demographics and comorbidities were identified within the LCDB. Patients were sorted into groups based on their BMI in accordance with the suggestions by the World Health Organization: BMI < 18.5 as underweight, BMI ≥ 18.5 and < 25 as normal weight, BMI ≥ 25 and < 30 as overweight, BMI ≥ 30 and < 35 as obese, BMI ≥35 and < 40 as very obese, BMI ≥ 40 as morbidly obese (World Health Organization n.d.). However, since only 8 patients (0.2%) were underweight (BMI < 18.5) the underweight group was included in the normal weight group (BMI < 25). Patients were also divided into 5 age groups: age < 50, age ≥ 50 and ≤ 60, age > 60 and ≤ 70, age > 70 and ≤ 80, age > 80. Statistics Categorical data is displayed as n (%) with 95% confidence intervals (CI). Continuous data is reported as mean and standard deviation (SD) or median and interquartile range (IQR) for normal and non-normal distributions of data, respectively. Normality was investigated using histograms and Q–Q plots. Differences in proportions of UKA patients within each BMI group with DOS discharge, prolonged admission, or 90-day readmission are compared using a chi-square test with BMI < 25 as reference category. Differences in proportions of UKA patients within each age group with DOS discharge, prolonged admission, or 90-day readmission are compared using a chisquare test with 61–70 years of age as reference category. Adjusted analyses using multivariable mixed-effect models were conducted in order to determine the effect of age and BMI on DOS discharge, prolonged admission, and 90-day readmissions. The adjusted analyses included the BMI groups; age groups; sex; living situation (with others, alone, institution); use of walking aid; preoperative anemia; diabetes mellitus (insulin dependent [IDDM] or non-insulin dependent [NIDDM]); hypertension; use of potent anticoagulants; and pharmacologically treated cardiac disease, pulmonary disease, and psychiatric disorder. The surgical center was included as a random effect in the adjusted analyses to account for differences between centers (Figure 1, see Supplementary data). Only patients with complete data on the above-mentioned patient characteristics were included in the adjusted analyses. A p-value < 0.05 was considered significant. Statistical analyses were conducted using R-studio and R v 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria). Ethics, funding, and conflicts of interest Approval from an ethical committee was not required as this is an observational study. Data retrieval for the LCDB was permitted by the Danish National Board of Health (3-3013-56/2/

724

Acta Orthopaedica 2021; 92 (6): 722–727

Table 3. Distribution of BMI and age groups among all included patients (N = 3,897). Values are count (%) unless otherwise specified

BMI, n 3,897 Median [IQR] Normal: < 25 840 (22) Overweight: 25.0–29.9 1,677 (43) Obese: 30.0–34.9 996 (26) Very obese: 35.0–39.9 294 (7.5) Morbidly obese ≥ 40 90 (2.3) Age, n 3,897 Mean (SD) [range] < 50 168 (4.3) 50–60 901 (23) 61–70 1,441 (37) 71–80 1,198 (31) > 80 189 (4.9)

28 [25–32]

66 (9.4) [26–97]

EMJO). Storing of data for the LCDB was approved by the Danish Data Protection Agency (P-2019-709). No funding was received for this study. The authors declare no conflicts of interest directly related to this study.

Results Due to missing data on BMI only 3,897 UKA procedures (99%) were included in the study, and of these 3,727 (96%) were medial UKA. Due to missing data in variables included in the adjusted analyses only 3,596 UKA procedures (92%)

were included in these analyses. Differences in endpoint measures between included and excluded procedures regarding the adjusted analyses are displayed in Tables 1 and 2 (see Supplementary data). Within the included cohort the median BMI was 28 (IQR 25–32), mean age was 66 (SD 9.4), years, and 54% were female (Table 3). The median LOS within all BMI and age groups was 1 day (age ≤ 70 and BMI < 35, IQR 0–1; age 71–80 and BMI 35–39.9, IQR 1–1; age > 80 and BMI ≥ 40, IQR 1–2). Day of surgery discharge DOS discharge was achieved in 26% (n = 992) of included patients. The proportion of DOS discharge in UKA patients with normal BMI was 26% versus 22% in morbidly obese patients. Individual analyses of the proportion of DOS discharge within each BMI group showed no statistically significant difference compared with the reference group with BMI < 25 (Table 4). Statistically significant differences in proportion of DOS discharge were present between age groups, with the reference group aged 61–70 years being discharged on DOS in 28% of cases compared with 21% of cases aged 71–80 years and 7.4% of cases aged > 80 years (Table 5). In the adjusted analysis the 2 oldest age groups were also found to be less likely to be discharged on DOS compared with the reference group (age 71–80, odds ratio [OR] = 0.71 [CI 0.58–0.88]; age > 80, OR = 0.18 [CI 0.10–0.34]) (Table 6); while BMI had no effect on DOS discharge (Table 7). UKA patients aged 50–60 years were more likely to be discharged on DOS (OR = 1.3 [CI 1–1.6]).

Table 4. Unadjusted analyses of difference in DOS discharge, LOS >2 days and 90-day readmission between BMI groups BMI groups (n = 3,897) Normal: < 25 Overweight: 25.0–29.9 Obese: 30.0–34.9 Very obese: 35.0–39.9 Morbidly obese ≥ 40

DOS discharge n (%) [95%CI] p-value 220 (26) [23–29] 423 (25) [23–27] 257 (26) [23–29) 72 (25) [20–29] 20 (22) [14–31]

Ref. 0.6 0.9 0.6 0.4

LOS > 2 days n (%) [95%CI] p-value 62 (7.4) [5.6–9.2] 112 (6.7) [5.5–7.9] 80 (8.0) [6.3–9.7) 23 (7.8) [4.8–11] 15 (17) [9.0–24]

Ref. 0.5 0.6 0.8 0.002

90-day readmission n (%) [95%CI] p-value 54 (6.4) [4.8–8.1] 107 (6.4) [5.2–7.6] 65 (6.5) [5.0–8.1) 32 (11) [7.3–14] 11 (12) [5.5–19]

Ref. 1.0 0.9 0.01 0.04

DOS = day of surgery. Ref. = reference group. LOS = length of stay.

Table 5. Unadjusted analyses of difference in DOS discharge, LOS >2 days and 90-day readmission between age groups Age groups (n = 3,897) < 50 50–60 61–70 71–80 > 80

DOS discharge n (%) [95%CI] p-value 47 (28) [21–35] 276 (31) [28–34] 406 (28) [26–31] 249 (21) [19–23] 14 (7.4) [3.7–11]

LOS > 2 days n (%) [95%CI] p-value

1.0 15 (8.9) [4.6–13] 0.2 66 (7.3) [5.6–9.0] Ref. 100 (6.9) [5.6–8.3] <0.001 89 (7.4) [5.9–8.9] <0.001 22 (12) [7.1–16]

DOS = day of surgery. Ref. = reference group. LOS = length of stay.

0.3 0.7 Ref. 0.6 0.02

90-day readmission n (%) [95%CI] p-value 10 (6.0) [2.4–9.5] 48 (5.3) [3.9–6.8] 82 (5.7) [4.5–6.9] 108 (9.0) [7.4–11] 21 (11) [6.6–16]

0.9 0.7 Ref. 0.001 0.004

Acta Orthopaedica 2021; 92 (6): 722–727 725

Table 6. Adjusted analysis of occurrence of DOS discharge, LOS > 2 days and 90-day readmission in age groups Age groups (n = 3,596)

DOS discharge OR (95% CI)

LOS > 2 days 90-day readmission OR (95% CI) OR (95% CI)

< 50 50–60 61–70 71–80 > 80

0.85 (0.55–1.3) 1.3 (1.0–1.6) Ref. 0.71 (0.58–0.88) 0.18 (0.10–0.34)

1.2 (0.68–2.3) 1.2 (0.83–1.7) Ref. 0.95 (0.69–1.3) 1.3 (0.73–2.2)

1.2 (0.59–2.6) 1.0 (0.70–1.6) Ref. 1.5 (1.1–2.1) 1.6 (0.89–2.8)

DOS = day of surgery. Ref. = reference group. LOS = length of stay. OR = odds-ratio.

age patients did not increase readmission rate compared with patients with LOS > 0 days (very obese with DOS discharge, 5.6% vs. very obese with LOS ≥ 1 day, 13%; age > 80 years with DOS discharge, 7.1% vs. age > 80 years with LOS ≥ 1 day = 13%) (Figures 2 and 3, see Supplementary data).

Discussion

UKA patients in our study had a normal BMI in roughly one-fifth of cases, 0.2% were underTable 7. Adjusted analysis of occurrence of DOS discharge, LOS > 2 days and 90-day weight, almost half of the patients categorized readmission in BMI groups as overweight (43%), and 2.3% were morbidly obese. Patients were mostly aged between 61 DOS discharge LOS > 2 days 90-day readmission and 70 years (37%). Age > 70 years reduced BMI groups (n = 3,596) OR (95% CI) OR (95% CI) OR (95% CI) the likelihood of DOS discharge. While age and BMI did not affect prolonged admission, Normal: < 25 Ref. Ref. Ref. Overweight: 25.0–29.9 0.96 (0.76–1.2) 0.97 (0.69–1.4) 1.0 (0.73–1.5) high BMI and advanced age increased the Obese: 30.0–34.9 1.0 (0.78–1.3) 1.1 (0.76–1.6) 0.99 (0.65–1.5) likelihood of readmission within 90 days of Very obese: 35.0–39.9 0.91 (0.62–1.3) 1.0 (0.60–1.8) 1.9 (1.1–3.1) surgery. Morbidly obese ≥ 40 1.1 (0.59–2.0) 1.5 (0.71–3.0) 2.1 (0.98–4.5) A study from the American College of SurDOS = day of surgery. Ref. = reference group. LOS = length of stay. OR = odds-ratio. geons National Quality Improvement Program (ACS-NSQIP) reported that most UKA patients were obese and 9.2% were morbidly Prolonged admission obese (Sundaram et al. 2019). Differences in national baseProlonged admission beyond 2 days occurred in 7.5% (n = line demographics might explain the differences between 292) of included patients. Prolonged admission was more these results and ours, as the baseline rate of obesity among frequent in morbidly obese patients (17%) as well as patients adults in Denmark was 17% in 2017 compared with 42% in aged > 80 years (12%) compared with patients with normal the Unites States of America in the same year (OECD 2019, BMI (7.4%) and patients aged 61–70 years (6.9%), respec- Hales et al. 2020). A previous study investigating BMI in TKA tively (Tables 4 and 5). However, in the analysis adjusted for patients from the LCDB found a somewhat similar distribuother patient characteristics and preoperative comorbidity tion of BMI, while a larger proportion of TKA patients were there was no statistically significant difference in LOS > 2 very or morbidly obese (very obese, 11% vs. 7.5%; morbidly obese, 4.4% vs. 2.3%) (Husted et al. 2016). This coincides days between BMI and age groups (Tables 6 and 7). with multiple studies reporting differences in patient characReadmission within 90 days of surgery teristics between UKA and TKA patients (Liddle et al. 2014, The overall readmission rate within the cohort was 6.9% (n = Drager et al. 2016). We found that UKA patients > 70 years of age were less likely 269). Readmission within 90 days of surgery occurred more frequently in patients with a BMI ≥ 35 (very obese 11%; mor- to be discharged on DOS compared with younger patients, bidly obese 12% [p = 0.07]) compared with patients with while BMI did not affect DOS discharge. This was similar to a normal BMI (6.4%) (Table 4). Readmissions within 90 days study by Matsumoto et al. (2020), who also reported advanced were also more frequent in patients aged > 70 years (aged age to be a barrier in achieving DOS discharge and while BMI 71–80 years, 9.0%; aged > 80 years, 11%) compared with did not affect DOS discharge. In that study BMI did, however, patients aged 61–70 years (5.7%) (Table 5). In the adjusted appear to be higher in patients discharged on DOS. Likewise, analysis patients with BMI ≥ 35 (very obese, OR = 1.9 [CI a study by Plate et al. (2017) found no correlation between 1.1 – 3.1]) (Table 7); and aged > 70 years (age 71–80, OR = BMI and LOS after robotic-assisted UKA surgery. 1.5 [CI 1.1–2.1]) remained more likely to be readmitted within Our study found morbid obesity and age > 80 years to be 90 days compared with the BMI and age reference groups, associated with prolonged hospitalization beyond 2 days after respectively (Table 6). However morbid obesity (OR = 2.1 [CI UKA in unadjusted analyses, but not in the adjusted analy1.0–4.5], p = 0.06) and age > 80 years (OR = 1.6 [CI 0.9–2.8]) sis. Therefore, other patient characteristics and preoperative were not statistically significant factors in the adjusted anal- comorbidity beyond high BMI and age might influence the ysis (Tables 6 and 7). DOS discharge in high BMI or high- risk of prolonged admission. In contrast, other studies have

726

reported higher BMI to be associated with prolonged admission (Haughom et al. 2015, Sephton et al. 2020). Differences in variables included in the adjusted analyses as well as the adherence to fast-track principles could potentially contribute to the above-mentioned discrepancies. In comparison, obesity and high BMI may increase LOS after TKA (Piuzzi et al. 2019, Shah et al. 2019). Readmissions within 90 days of UKA surgery occurred more frequently in patients > 70 years of age and BMI > 35 (but not statistically significant in the morbidly obese patients and patients > 80 years of age in the adjusted analysis). However, other patient characteristics, preoperative comorbidity, and missing data in the adjusted analysis might account for readmissions in extreme BMI and age groups. Using data from ACS NSQIP from 2005–2012, Haughom et al. (2015) (n = 2,316) reported that obesity increased the risk of 30-day readmissions and complications. Using data from ACS NSQIP from 2008–2016, Sundaram et al. (2019) (n = 8,029) found overweight patients to have a lower risk of 30-day readmissions and complications compared with normal-weight patients, while morbidly obese patients had an increased risk of superficial surgical site infections. It is plausible that the difference between these studies could be explained by temporal improvements in perioperative care and surgical technique positively influencing the overall risk associated with UKA surgery from 2005 to 2016 (Gromov et al. 2020). Despite the potential increased risk of readmissions in high BMI and highage patients, we find that DOS discharge in these patients does not increase the readmission rate. This coincides with other studies investigating safety of DOS discharge after UKA and knee arthroplasty surgery in general (Otero et al. 2016, Pollock et al. 2016). Some studies investigating recovery and complications after surgery, as well as arthroplasty surgery, report better outcome in overweight/obese patients compared with normal-weight patients (Shaparin et al. 2016, Smith et al. 2020). This is labelled the “obesity paradox.” The findings of Sundaram et al. (2019) mentioned above also reported fewer complications in overweight patients. A study by Katakam et al. (2021) investigating LOS in patients undergoing total hip arthroplasty and TKA found that being either underweight or morbidly obese resulted in increased LOS. It is suggested that the relationship between BMI and various outcomes is not linear but rather quadratic, with slight overweight in elderly patients countering sarcopenia and malnutrition (Shaparin et al. 2016, Smith et al. 2020). However, patient selection is also proposed as the reason for these findings, as surgeons might refrain from operating on obese patients with additional comorbidity, thus creating a healthier cohort (Zhang et al. 2018). Strengths and limitations Despite extensive research on LOS and readmission rates after TKA, the literature on these subjects in UKA patients is scarce. Data from the LCDB has a high quality due to the use of the

Acta Orthopaedica 2021; 92 (6): 722–727

DNRP as well as chart review regarding readmission and the use of prospectively collected patient questionnaires crossreferenced to the DNDRP regarding prescribed drugs, thus ensuring complete follow-up for all patients. Data is collected from 8 surgical centers and differences in perioperative practice between these centers could be present, although centers reporting to the LCDB are required to have implemented fasttrack principles with a median LOS < 3 days for hip and knee arthroplasty surgery. The indications regarding UKA surgery could differ between surgical centers, and some surgeons could potentially use high BMI as a relative contraindication. Also, 304 included patients (8.5%) had more than one UKA procedure within the study period (> 90 days from first procedure). While this may limit the external validity of the study (Bryant et al. 2006), excluding these patients with multiple procedures could also introduce bias (Ravi et al. 2013). Despite having included a high number of patients compared with other cohort studies investigating UKA patients, our study is not powered to investigate underweight patients or differences in the specific complications leading to readmission. In conclusion, UKA patients > 70 years are less likely to be discharged on DOS, while BMI did not affect DOS discharge. Advanced age and high BMI did not increase the likelihood of prolonged admission beyond 2 days but did increase likelihood of 90-day readmission. However, DOS discharge could still be considered safe in patients with advanced age and high BMI, as it did not increase the readmission rate in these patients. This data may be considered as part of the shared decision-making process and should be noted by surgeons operating on patients with high BMI and advanced age. Supplementary data Figures 1–3 and Tables 1 and 2 are available as supplementary data in the online version of this article, http://dx.doi.org/10. 1080/17453674.2021.1968727 CBJ and KG had full access to all data in the study and all authors take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: CBJ, AT, and KG. Acquisition, analysis, and interpretation of data: all authors. Drafting of the manuscript: CBJ and KG. Critical revision of the manuscript: all authors. Acta thanks Michael Clarius and David Houlihan-Burne for help with peer review of this study.

Bryant D, Havey T C, Roberts R, Guyatt G. How many patients? How many limbs? Analysis of patients or limbs in the orthopaedic literature: a systematic review. J Bone Joint Surg Am 2006; 88: 41-5. Drager J, Hart A, Khalil J A, Zukor D J, Bergeron S G, Antoniou J. Shorter hospital stay and lower 30-day readmission after unicondylar knee arthroplasty compared to total knee arthroplasty. J Arthroplasty 2016; 31: 356-61. Goodfellow J W, Kershaw C J, Benson M K, O’Connor J J. The Oxford Knee for unicompartmental osteoarthritis: the first 103 cases. J Bone Joint Surg Br 1988; 70: 692-701.

Acta Orthopaedica 2021; 92 (6): 722–727 727

Gromov K, Petersen P B, Jørgensen C C, Troelsen A, Kehlet H. Unicompartmental knee arthroplasty undertaken using a fast-track protocol. Bone Joint J 2020; 102-b: 1167-75. Hales C M, Carroll M D, Fryar C D, Ogden C L. Prevalence of obesity and severe obesity among adults: United States, 2017–2018. NCHS Data Brief 2020: 1-8. Hamilton T W, Pandit H G, Jenkins C, Mellon S J, Dodd C A F, Murray D W. Evidence-based indications for mobile-bearing unicompartmental knee arthroplasty in a consecutive cohort of thousand knees. J Arthroplasty 2017; 32: 1779-85. Haughom B D, Schairer W W, Hellman M D, Nwachukwu B U, Levine B R. An analysis of risk factors for short-term complication rates and increased length of stay following unicompartmental knee arthroplasty. HSS J 2015; 11: 112-16. Henkel C, Mikkelsen M, Pedersen A B, Rasmussen L E, Gromov K, Price A, Troelsen A. Medial unicompartmental knee arthroplasty: increasingly uniform patient demographics despite differences in surgical volume and usage—a descriptive study of 8,501 cases from the Danish Knee Arthroplasty Registry. Acta Orthop 2019; 90: 354-9. Husted H. Fast-track hip and knee arthroplasty: clinical and organizational aspects. Acta Orthop Suppl 2012; 83: 1-39. 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: 466-72. 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. Jørgensen C C, Kehlet H. Role of patient characteristics for fast-track hip and knee arthroplasty. Br J Anaesth 2013; 110: 972-80. 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: e009813. Katakam A, Melnic C M, Bragdon C R, Sauder N, Collins A K, Bedair H S. Low body mass index is a predictor for mortality and increased length of stay following total joint arthroplasty. J Arthroplasty 2021; 36: 72-7. Kehlet H, Lindberg-Larsen V. High-dose glucocorticoid before hip and knee arthroplasty: to use or not to use—that’s the question. Acta Orthop 2018; 89: 477-9. Kozinn S C, Scott R. Unicondylar knee arthroplasty. J Bone Joint Surg Am 1989; 71: 145-50. Liddle A D, Judge A, Pandit H, Murray D W. Adverse outcomes after total and unicompartmental knee replacement in 101,330 matched patients: a study of data from the National Joint Registry for England and Wales. Lancet 2014; 384: 1437-45. Matsumoto M, Saito S, Andrews S, Mathews K, Morikawa L, Nakasone C. Barriers to achieving same day discharge following unilateral unicompartmental knee arthroplasty. Knee 2020; 27: 1365-9. Molloy I B, Martin B I, Moschetti W E, Jevsevar D S. Effects of the length of stay on the cost of total knee and total hip arthroplasty from 2002 to 2013. J Bone Joint Surg Am 2017; 99: 402-07. OECD. Systems E O o H, Policies; 2019. Denmark: Country Health Profile 2019. Otero J E, Gholson J J, Pugely A J, Gao Y, Bedard N A, Callaghan J J. Length of hospitalization after joint arthroplasty: does early discharge affect complications and readmission rates? J Arthroplasty 2016; 31: 271425.

Pandit H, Jenkins C, Gill H S, Smith G, Price A J, Dodd C A, Murray D W. Unnecessary contraindications for mobile-bearing unicompartmental knee replacement, J Bone Joint Surg Br 2011; 93: 622-8. Piuzzi N S, Strnad G J, Sakr Esa W A, Barsoum W K, Bloomfield M R, Brooks P J, Higuera-Rueda C A, Joyce M J, Kattan M W, Klika A A, Krebs V, Mesko N W , Mont M A , Murray T G, Muschler G F, Nickodem R J, Patel P D, Schaffer J L, Spindler K P, Stearns K L, Suarez J C, Zajichek A, Molloy R M. The main predictors of length of stay after total knee arthroplasty: patient-related or procedure-related risk factors. J Bone Joint Surg Am 2019; 101: 1093-101. Plate J F, Augart M A, Seyler T M, Bracey D N, Hoggard A, Akbar M, Jinnah R H, Poehling G G. Obesity has no effect on outcomes following unicompartmental knee arthroplasty. Knee Surg Sports Traumatol Arthrosc 2017; 25: 645-51. Pollock M, Somerville L, Firth A, Lanting B. Outpatient total hip arthroplasty, total knee arthroplasty, unicompartmental knee arthroplasty: a systematic review of the literature. JBJS Rev 2016; 4(12). Price A J, Alvand A, Troelsen A, Katz J N, Hooper G, Gray A, Carr A, Beard D. Knee replacement. Lancet 2018; 392: 1672-82. Ravi B, Croxford R, Hawker G. Exclusion of patients with sequential primary total joint arthroplasties from arthroplasty outcome studies biases outcome estimates: a retrospective cohort study. Osteoarthritis Cartilage 2013; 21: 1841-8. Reilly K A, Beard D J, Barker K L, Dodd C A, Price A J, Murray D W. Efficacy of an accelerated recovery protocol for Oxford unicompartmental knee arthroplasty: a randomised controlled trial. Knee 2005; 12: 351-7. 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, research potential. Clin Epidemiol 2015; 7: 449-90. Sephton B M, Bakhshayesh P, Edwards T C, Ali A, Kumar Singh V, Nathwani D. Predictors of extended length of stay after unicompartmental knee arthroplasty. J Clin Orthop Trauma 2020; 11: S239-s45. 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: 2124-65.e1. Shaparin N, Widyn J, Nair S, Kho I, Geller D, Delphin E. Does the obesity paradox apply to early postoperative complications after hip surgery? A retrospective chart review. J Clin Anesth 2016; 32: 84-91. Smith E L, Shahien A A, Chung M, Stoker G, Niu R, Schwarzkopf R. The obesity paradox: body mass index complication rates vary by gender and age among primary total hip arthroplasty patients. J Arthroplasty 2020; 35: 2658-65. Sundaram K, Warren J, Anis H, George J, Murray T, Higuera C A, Piuzzi N S. An increased body mass index was not associated with higher rates of 30-day postoperative complications after unicompartmental knee arthroplasty. Knee 2019; 26: 720-8. van der List J P, Chawla H, Zuiderbaan H A, Pearle A D. The role of preoperative patient characteristics on outcomes of unicompartmental knee arthroplasty: a meta-analysis critique. J Arthroplasty 2016; 31: 2617-27. World Health Organization. Body mass index—BMI. Available from: https://www.euro.who.int/en/health-topics/disease-prevention/nutrition/ahealthy-lifestyle/body-mass-index-bmi. Zhang J C, Matelski J, Gandhi R, Jackson T, Urbach D, Cram P. Can patient selection explain the obesity paradox in orthopaedic hip surgery? An analysis of the ACS-NSQIP Registry. Clin Orthop Relat Res 2018; 476: 964-73.

728

Acta Orthopaedica 2021; 92 (6): 728–732

Nonoperative management of hip fractures in very frail elderly patients may lead to a predictable short survival as part of advance care planning Hugo H WIJNEN 1, Peter P SCHMITZ 2, Houda ES-SAFRAOUY 1, Lian A ROOVERS 3, Diana G TAEKEMA 1, and Job L C VAN SUSANTE 2 1 Department

of Clinical Geriatrics, Rijnstate, Arnhem; 2 Department of Orthopedics, Rijnstate, Arnhem; 3 Clinical Research Department, Rijnstate, Arnhem, the Netherlands Correspondence: PSchmitz@rijnstate.nl Submitted 2021-01-24. Accepted 2021-06-22.

Background and purpose — Surgical treatment is still the mainstay of care even in very frail elderly hip fracture patients. However, one may argue whether surgery is in the best interest of all patients. We elucidated mortality rates of nonoperative management (NOM) of a hip fracture after shared decision-making in a cohort of very frail elderly patients. Patients and methods — Orthogeriatric patients (age > 70 years) admitted with a hip fracture between 2011 and 2019 were included. In the presence of fragility features the motivation for surgery or NOM was supported by advance care planning (ACP) and shared decision-making through geriatric assessment. Mortality rates after NOM were assessed and also presented for the remaining surgical group for reference. Results — In 1,279 out of 3,467 patients, geriatric assessment was indicated and subsequently 1,188 (93%) had surgery versus 91 (7%) NOM. The motivation for NOM was based on patient and family preferences in only 20% of patients, medical grounds in 54%, and a combination of both in 26%. The 30-day and 1-year mortality in the frail NOM group was 87% and 99% respectively, whereas this was 7% and 28% in the surgery group. No statistical comparison between groups was performed due to profound bias by indication. Interpretation — This study provides further insight into the predictable and high short-term mortality after NOM in carefully selected very frail elderly hip fracture patients. This information may help to consider NOM as an alternative treatment option to surgery when no significant gain from surgery is anticipated.

The incidence of hip fractures in frail patients is rising due to an increase in life expectancy and cumulative comorbidities (Kanis 2002, Johnell and Kanis 2006, Ferrucci et al. 2016). In particular, frail elderly patients experience a substantial decrease in quality of life and mobility in the 12 months after hip fracture surgery (Amarilla-Donoso et al. 2020). Deliberations whether to operate or not in these frail elderly patients are common in daily practice (Dunn et al. 2016, van der Zwaard et al. 2020, Rietjens et al. 2021). Surgery is still the mainstay of treatment because it results in a better mobility and survival (van de Ree et al. 2017, Berry et al. 2018) and nonoperative management (NOM) is often characterized by problematic after-treatment with substantial patient discomfort. One may argue, however, whether surgery is in the best interest of all patients. It may be that the time has come to re-evaluate to what extent frail patients should always be treated surgically. Frailty is negatively associated with quality of life after a hip fracture and may require tailored treatment, especially in patients with a short life expectancy and anticipated postoperative functional decline (van de Ree et al. 2019, Kanters et al. 2020). A hip fracture is a life-threatening condition in these frail patients and as such may be considered an opportunity to discuss end-of-life care, personal goals, and the preference for surgery or NOM with their potential pros and cons (Dunn et al. 2016) This concept is also known as “advance care planning” (ACP) (Teno et al. 1994, Johnston et al. 2018). The literature on life expectancy after nonoperative management (NOM) is limited. For example, 30-day and 1-year mortality rates after NOM ranging from 5–65% and 30–95% have been reported (Loggers et al. 2020). These wide ranges of early mortality rates reflect differences in patient character-

Acta Orthopaedica 2021; 92 (6): 728–732

istics and fracture types between studies and extrapolation of information from these studies towards shared decision-making in clinical practice is difficult. There is increasing awareness that operative treatment of a hip fracture may not always be the best option in all patients and that in a subgroup of frail elderly patients NOM should be considered and may also be patients’ preference. For this purpose it is important to obtain valid information on what can be expected in terms of survival after NOM in a well-selected subgroup of very frail patients. In this way informed consent and shared decision-making on treatment options for fragility hip fractures can be improved. We elucidated mortality rates in a consecutive group of 91 out of 1,279 patients with a fragility hip fracture treated with NOM after ACP and shared decision-making. Survival for the remaining surgically treated patients is also presented for reference.

Patients and methods Study population We performed a retrospective study to elucidate the course of mortality after NOM of a hip fracture. All consecutive hip fracture patients aged 70 years and older treated on the orthogeriatric ward were included between January 2011 and June 2019. Indication for admittance to the orthogeriatric ward was set by the orthogeriatric team after a comprehensive geriatric assessment on the somatic, psychiatric, functional, or social domain in order to identify fragility features. Patients with 1 or more fragility features were admitted to the orthogeriatric ward, whereas relatively vital and mostly younger patients were admitted to the surgical ward. The latter group of patients was not part of this study (Figure 1). Exclusion criteria were pathological hip fractures and periprosthetic hip fractures due to their impact on mortality. The decision for surgery or NOM was made as a shared decision-making consultation with the patient and relatives. ACP was applied by mutually exploring treatment preferences, mobility, quality of life, goals in life, and survival in a comprehensive geriatric assessment (van der Zwaard et al. 2020). Surgery consisted of either hemiarthroplasty or internal fixation. The NOM approach focused on optimal comfort for the patient and mainly consisted of treatment of symptoms, in particular pain with analgesics. Patients returned to their homes or residential care facilities if feasible or were discharged to a nursing home with hospice care facilities. Study variables The following parameters were retrospectively collected from the patient records: age, sex, Charlson Comorbidity Index (CCI), mobility, living situation, cognitive status (dementia), type of fracture, and type of surgery. Reports from the shared-decision consultation resulting in a choice for NOM were reviewed. The reason for NOM was

729

Hip fracture patients 2011–2019 n = 3,467 Trauma ward n = 2,188 Orthogeriatric ward n = 1,279

Surgery n = 1,188

Non-operative management n = 91

Figure 1. Flowchart of patient selection.

scored independently by 2 authors (HW and DT) as (1) based on patient and family preferences, (2) obvious medical grounds with unacceptable high perioperative risk, or (3) a combination of both. In the case of disagreement consensus was subsequently achieved by discussion between both authors. If applicable, date of death was obtained from the medical records or from the Dutch Personal Records Database, which contains personal data of people who live in the Netherlands. Subsequently 30-day mortality and 1-year mortality rates were determined. Statistics From a profound bias by indication together with a relatively small group on NOM no statistical analysis on differences between groups was performed. Ethics, funding, and potential conﬂicts of interest Ethical approval for the study was granted by the Institutional Review Board (decision 2016-938). No funding was obtained and the authors have no conflicts of interest to declare.

Results Baseline characteristics (Table) Between 2011 and 2019, 3,467 hip fracture patients were admitted to our hospital of which 1,279 frail elderly hip fracture patients (37%) were admitted to the orthogeriatric ward. From this study group 1,188 patients received surgical treatment whereas for 91 NOM was chosen. As such NOM was offered to 7.1% (91/1,279) of patients admitted to the orthogeriatric ward and to 2.6% (91/3,467) of all patients (Figure 1). As anticipated, the 1,188 patients in the surgery group were younger than the 91 patients in the NOM group—84 (SD 6.7) years versus 87 (SD 6.3) years, respectively. Pre-fracture mobility was better in patients who were operated on compared with those who were not, reflected by 45% of surgery patients being able to walk independently compared with 12% of patients in the NOM group. Further, dementia and living situation were different in the 2 groups, which is probably a reflection of the confounding bias by indication (NICE 2011).

730

Acta Orthopaedica 2021; 92 (6): 728–732

Patients’ characteristics at baseline. Values are count (%) unless otherwise specified Population variable

Surgery Nonoperative n = 1,188 n = 91

Age (SD) 84 (6.7) 87 Sex (female) 877 (74)59 Type of hip fracture Femoral neck 639 (54) 53 Trochanteric 444 (37) 35 Other 105 (9) 3 Charlson Comorbidity Index < 3 877 (74) 59 ≥ 3 311 (26) 32 Living situation Independent 725 (61) 26 Sheltered care 149 (13) 19 Nursing home 200 (17) 46 Missing 114 (9) Dementia 317 (27) 48 Mobility Without assistance 511 (43) 9 Cane/walker 614 (52) 53 Wheelchair/bedridden 24 (2) 13 Missing 39 (3) 16

Motivation for nonoperative management In 18 of 91 patients the decision for NOM was clearly based on patient and family preference whereas medical grounds were not evident. In 49 patients NOM was almost completely dictated on medical grounds (comorbidity and high perioperative risk) and in 24 patients this was a combination of both. Mortality rates after nonoperative management and surgery The 30-day and 1-year mortality rates in the frail NOM group were 87% and 99%, respectively (Figure 2). Of the 91 patients in the NOM group only 21 survived after 2 weeks, 12 were alive after 30 days, and only 3 after 3 months. Mean survival was 0.7 months in the nonoperatively managed patients. The 30-day and 1-year mortality rates in the surgery group were 7% and 28%, respectively (Figure 2). Mean survival was 36 months in the surgically managed patients.

Discussion This study was primarily conducted to provide more insight into the course of NOM in a selected group of very frail elderly hip fracture patients in whom ACP with patients and relatives had resulted in a shared decision not to operate to prolong life. The 30-day and 1-year mortality rates in this group were 87% and 99%, respectively. In contrast the reference 30-day and 1-year mortality rates in the surgery group were 7% and 28%, respectively. The choice for NOM was at least partly based on patient or family preference in almost half of these

Cumulative survival 1.0 Nonoperative management Surgery 0.8

0.6

0.4

0.2

Years after fracture

Figure 2. Kaplan–Meier survival analysis after nonoperative management (91 patients) and surgery (1,188 patients).

patients (46%), whereas the other half was treated nonoperatively mainly based on medical grounds (54%). We found a very high 30-day mortality of 87% in nonoperatively managed hip fracture patients. These findings differ from other previously published studies with a 30-day mortality ranging from 5% to 65% (Chlebeck et al. 2019, Loggers et al. 2020). A recent review has already pointed out that these large differences in mortality rates are most probably caused by differences in patient characteristics such as pre-fracture mobility, prevalence of dementia, and living situation (Loggers et al. 2020). The clinical dilemma whether to decide for NOM and subsequent palliative care mainly applies to very frail elderly hip fracture patients. As such, it should be noted that the nonoperatively managed patients in our study were typically identified on the basis of their frailty with a low functional pre-fracture mobility who have not much to gain from surgery. In contrast, patients from available studies so far were less frail and had a better pre-fracture mobility, resulting in a treatment focus on active mobilization to regain function and prolong life (Loggers et al. 2020). For example, Berry et al. (2018) found a lower mortality rate of 54% at 6 months in a retrospective cohort study of 468 nonoperatively treated nursing home patients with advanced dementia and a hip fracture and Hossain et al. (2009) reported a mortality rate of 19% at 1 month in a retrospective cohort study of 21 nonoperatively treated patients. Again, we feel this can be explained by a different inclusion of patients for NOM. For example, Berry et al. also included patients with pelvic fractures and palliative care was initiated in only 34% of the patients, whereas in our study this accounted for all patients (Berry et al. 2018). In turn, Hossain et al. seemed to have used a different selection of patients as well, as a substantial number of non-displaced and impacted femoral neck fractures were included in which case NOM is an established approach to achieve healing of the fracture (Hossain et al. 2009). Again, this is an important confounder because nonoperative fracture treatment in non-

Acta Orthopaedica 2021; 92 (6): 728–732

displaced femoral neck fractures is entirely different from the palliative care approach in our study. The choice of NOM in very frail elderly patients with a hip fracture is delicate and ethical, cultural, and legal issues apply, which can be different for countries. We believe that there is an increasing awareness that, in spite of the fact that—in general—surgery increases survival and the chances of regaining mobility, not all patients prefer surgical treatment. Instead, we should aim to improve recognition of this small group of very frail patients who have not much to gain from surgery and often do not wish to prolong life. This phenomenon is also well illustrated by a recent article of the Dutch association for medical doctors in which half of the nursing home patients indicated a preference for NOM above surgery should they be admitted to hospital with a hip fracture, irrespective of possible shorter survival (Stavenuiter et al. 2018). A retrospective evaluation of the reason for NOM in our study confirmed that patient preference is an important reason for NOM in almost half of the patients. A thorough decision-making process is essential to recognize this specific group of very frail elderly patients in which NOM of a hip fracture may be considered. Frail elderly patients require ACP where multiple comorbidities, short life expectancy, anticipated postoperative functional decline, and low quality of life are discussed, together with patient preference personal goals and end-of-life wishes (van de Ree et al. 2019, Kanters et al. 2020). In this ACP the frail patients, their relatives, and healthcare professionals participate and decide together on NOM or surgery. The concept of patient-centered tailored treatment, ACP, was first conceptualized in the United States (Teno et al. 1994) with the purpose to receive medical care consistent with one’s preferences (Dunn et al. 2016, Johnston et al. 2018, Rietjens et al. 2021). ACP is increasingly integrated in daily care for frail elderly patients with a hip fracture in European countries, although there is still room for improvement (Evans et al. 2013). This study has its limitations. The most important limitation is the recognized “confounding bias by indication” because the decision for NOM is typically made in a subgroup of frail patients with more comorbidities, higher age, lesser mobility, and dementia. In spite of the fact that these confounders can be corrected for in statistical models, we feel that this would oversimplify the problem and that profound bias would remain. One may argue whether this confounding is truly relevant with regards to the interpretation of the results. We decided to simply present the survival curves for NOM and surgery as reference, because it is beyond the scope of this study to compare mortality rates in NOM and operative treatment of hip fractures in elderly patients. Moreover, this study aimed to gain insight into the course of NOM with palliative care so that this option can be incorporated in balanced (shared) decision-making. Perhaps more importantly, from the mortality observed in the NOM group it appears that a rapid decline can be expected. Second, the inclusion of patients was limited to those patients

731

admitted to the orthogeriatric ward, also indicating selection bias. However, these frail elderly patients in particular, selected by a comprehensive geriatric assessment, reflect the population of concern in the clinical dilemma of NOM of hip fractures. We feel that this clear selection bias may also reflect the high 30-day and 1-year mortality rates encountered in this study as compared with the available literature. Besides limitations, important strengths also apply. In this study a rather homogeneous group of 91 very frail hip fracture patients were treated nonoperatively. This provided important information on the course of treatment and mortality, which is lacking in the available literature with a smaller number of patients and more importantly inhomogeneous fragility features (van de Ree et al. 2017). This information is important as it can be applied directly towards the clinical setting where we encounter the dilemma of whether surgery is indeed in the best interests of and in accordance with the preference of a patient with a fragility hip fracture. In conclusion, this study revealed a predictable short survival after NOM of hip fractures in a group of very frail elderly patients where prolonged survival was not considered to be the primary goal of treatment. These results may reassure clinicians, patients, and their relatives that NOM can be regarded as a relevant treatment option with a predictable outcome in very frail hip fracture patients. In the case of limited life expectancy, information regarding nonoperative supportive management and well-managed pain in a palliative setting might help patients and their families to come to a wellfounded decision in line with their wishes, goals, and endof-life expectations. Therefore, in our opinion, NOM should gain more attention and ACP should be part of standardized preoperative hip fracture care in the frailest patients.

HW, DT, and HE designed the study; HW and HE collected the data; ER carried out data analyses; HW and PS wrote the manuscript; HW, PS, HE, ER, DT, and JS contributed to revision of the manuscript. Acta thanks Margareta Hedström and Cecilia Rogmark for help with peer review of this study.

Amarilla-Donoso F J, Roncero-Martin R, Lavado-Garcia J M, ToribioFelipe R, Moran-Garcia J M, Lopez-Espuela F. Quality of life after hip fracture: a 12-month prospective study. Peer J 2020; 8: e9215. doi: 10.7717/peerj.9215. Berry S D, Rothbaum R R, Kiel D P, Lee Y, Mitchell S L. Association of clinical outcomes with surgical repair of hip fracture vs nonsurgical management in nursing home residents with advanced dementia. JAMA Intern Med 2018; 178(6): 774-80. doi: 10.1001/jamainternmed.2018.0743. Chlebeck J D, Birch C E, Blankstein M, Kristiansen T, Bartlett C S, Schottel P C. Nonoperative geriatric hip fracture treatment is associated with increased mortality: a matched cohort study. J Orthop Trauma 2019; 33(7): 346-50. doi: 10.1097/bot.0000000000001460. Dunn R H, Ahn J, Bernstein J. End-of-life care planning and fragility fractures of the hip: are we missing a valuable opportunity? Clin Orthop Relat Res 2016; 474(7): 1736-9. doi: 10.1007/s11999-015-4675-1.

732

Evans N, Pasman H R, Vega Alonso T, Van den Block L, Miccinesi G, Van Casteren V, Donker G, Bertolissi S, Zurriaga O, Deliens L, Onwuteaka-Philipsen B. End-of-life decisions: a cross-national study of treatment preference discussions and surrogate decision-maker appointments. PLoS One 2013; 8(3): e57965. doi: 10.1371/journal. pone.0057965. Ferrucci L, Cooper R, Shardell M, Simonsick E M, Schrack J A, Kuh D. Age-related change in mobility: perspectives from life course epidemiology and geroscience. J Gerontol A Biol Sci Med Sci 2016; 71(9): 1184-94. doi: 10.1093/gerona/glw043. Hossain M, Neelapala V, Andrew J G. Results of non-operative treatment following hip fracture compared to surgical intervention. Injury 2009; 40(4): 418-21. doi: 10.1016/j.injury.2008.10.001. Johnell O, Kanis J A. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int 2006; 17(12): 172633. doi: 10.1007/s00198-006-0172-4. Johnston C B, Holleran A, Ong T, McVeigh U, Ames E. Hip fracture in the setting of limited life expectancy: the importance of considering goals of care and prognosis. J Palliat Med 2018; 21(8): 1069-73. doi: 10.1089/ jpm.2018.0029. Kanis J A. Diagnosis of osteoporosis and assessment of fracture risk. Lancet 2002; 359(9321): 1929-36. doi: 10.1016/s0140-6736(02)08761-5. Kanters T A, van de Ree C L P, de Jongh M A C, Gosens T, Hakkaart-van Roijen L. Burden of illness of hip fractures in elderly Dutch patients. Arch Osteoporos 2020; 15(1): 11. doi: 10.1007/s11657-019-0678-y.

Acta Orthopaedica 2021; 92 (6): 728–732

Loggers S A I, Van Lieshout E M M, Joosse P, Verhofstad M H J, Willems H C. Prognosis of nonoperative treatment in elderly patients with a hip fracture: a systematic review and meta-analysis. Injury 2020; 51(11): 2407-13. doi: 10.1016/j.injury.2020.08.027. NICE. Hip fracture: management. Clinical guideline. National Institute for Health and Care Excellence 2011; https://www.nice.org.uk/guidance/cg124 Rietjens J, Korfage I, Taubert M. Advance care planning: the future. BMJ Support Palliat Care 2021; 11(1): 89-91. doi: 10.1136/bmjspcare-2020-002304. Stavenuiter R, Hermanussen H, van de Ree M, Steens J. Oudere breekt heup: opereren of niet? Medisch Contact 2018; 9: 32-3. Teno J M, Nelson H L, Lynn J. Advance care planning: priorities for ethical and empirical research. Hastings Cent Rep 1994; 24(6): S32-6. van de Ree C L P, De Jongh M A C, Peeters C M M, de Munter L, Roukema J A, Gosens T. Hip fractures in elderly people: surgery or no surgery? A systematic review and meta-analysis. Geriatr Orthop Surg Rehabil 2017; 8(3): 173-80. doi: 10.1177/2151458517713821. van de Ree C L P, Landers M J F, Kruithof N, de Munter L, Slaets J P J, Gosens T, de Jongh M A C. Effect of frailty on quality of life in elderly patients after hip fracture: a longitudinal study. BMJ Open 2019; 9(7): e025941. doi: 10.1136/bmjopen-2018-025941. van der Zwaard B C, Stein C E, Bootsma J E M, van Geffen H, Douw C M, Keijsers C. Fewer patients undergo surgery when adding a comprehensive geriatric assessment in older patients with a hip fracture. Arch Orthop Trauma Surg 2020; 140(4): 487-92. doi: 10.1007/s00402-019-03294-5.

Acta Orthopaedica 2021; 92 (6): 733–738 733

Trochanteric stabilizing plate in the treatment of trochanteric fractures: a scoping review Carl Erik ALM 1,2, Jan-Erik GJERTSEN 3, Trude BASSO 4, Kjell MATRE 3, Stephan RÖRHL 1, Jan Erik MADSEN 1,2, and Frede FRIHAGEN 2,5 1 Division of Orthopaedic Surgery, Oslo University Hospital, Oslo; 2 Institute of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo; 3 Department of Clinical Medicine, University of Bergen, Bergen; 4 Department of Orthopaedic Surgery, St Olav’s University Hospital, Trondheim; 5 Department of Orthopaedic Surgery, Østfold Hospital Trust, Grålum, Norway.

Correspondence: alm.carl.erik@gmail.com Submitted 2021-03-02. Accepted 2021-06-16.

Background and purpose — The trochanteric stabilizing plate (TSP) may be used as an adjunct to a sliding hip screw (SHS) in the treatment of trochanteric fractures to increase construct stability. We performed a scoping review of the literature to clarify when and how the TSP may be useful. Methods — A systematic search was performed in 5 databases and followed by a backwards-and-forwards citation search of the identified papers. 24 studies were included. Results — 6 biomechanical studies and 18 clinical studies were included in the review. The studies presented mainly low-level evidence. All studies were on unstable trochanteric fractures or fracture models. Due to the heterogeneity of methods and reporting, we were not able to perform a metaanalysis. In the biomechanical trials, the TSP appeared to increase stability compared with SHS alone, up to a level comparable with intramedullary nails (IMNs). We identified 1,091 clinical cases in the literature where a TSP had been used. There were 82 (8%) reoperations. The rate of complications and reoperations for SHS plus TSP was similar to previous reports on SHS alone and IMN. It was not possible to conclude whether the TSP gave better clinical results, when compared with either SHS alone or with IMN. Interpretation — The heterogeneity of methods and reporting precluded any clear recommendations on when to use the TSP, or if it should be used at all.

Internal fixation of trochanteric femoral fractures is usually performed with a plate or nail with a lag screw allowing axial compression to enhance fracture healing. The agreement between surgeons on implant choice is fair (Mellema et al. 2021). While a sliding hip screw (SHS) seems sufficient in stable trochanteric fractures (Parker and Handoll 2010), several guidelines recommend the use of intramedullary nails (IMNs) in more unstable fracture patterns (NICE 2011, Roberts and Brox 2015). Fixation without a lag screw is not recommended (Parker and Handoll 2010, Parker et al. 2018). Fractures involving the lateral wall, or with posteromedial comminution, are considered as unstable. This might cause excessive medialization of the femoral shaft, malunion, poor functional results, and even fixation failure (Parker 1996, Bretherton and Parker 2016). In addition, a thin lateral wall or a concomitant fracture through the greater trochanter increases the risk for an intra- or postoperative lateral wall fracture (Palm et al. 2007, Hsu et al. 2013). Under these circumstances, with a compromised lateral buttress, implant-preventing secondary displacement is required. The trochanteric stabilizing plate (TSP) was introduced in the early 1990s as an adjunct to the sliding hip screw. The plate acts by buttressing the lateral trochanteric wall and is intended to prevent medialization of the femoral shaft (Figure 1). Despite being sparsely discussed in the literature, an SHS with an additional TSP has been widely used in some countries and regions for decades (Lunsjö et al. 2001, Bong et al. 2004, Gupta et al. 2010, Knobe et al. 2013, Hsu et al. 2015, Alm et al. 2021). We reviewed the literature on TSP to clarify existing evidence and aid in the decision-making on when to use a TSP.

734

Acta Orthopaedica 2021; 92 (6): 733–738

Articles identified through database search from 1985 to June 2020 (n = 168): – Pubmed, 24 – Scopus, 37 – Epistemonikos, 55 – Embase, 29 – Web of Science, 23

Articles identified through citation searches n=5

Articles after removal of duplicates n = 93 Articles excluded after screening title and abstract n = 60 Full-text articles assessed for eligiblity n = 33 Full-text articles excluded (n = 14): – not reporting result of TSP use, 8 – less than 3 cases with TSP, 1 – TSP used on subtrachanteric fractures, 1 – TSP used for revision surgeries, 1 – full text not available, 2 – not possible to extract results, 1

Figure 1. Pre- and postoperative images of a AO/OTA 31-A2 fracture operated on with a sliding hip screw with trochanteric stabilizing plate (TSP). The TSP should prevent excessive medialization of the femoral shaft by buttressing the lateral trochanteric wall. In this case, a loss of medial buttress with a large lesser trochanter fragment and a thin lateral wall would strengthen the traditional indication for a TSP.

Method We applied the recommendations from the Cochrane collaboration (Higgins et al. 2020) and the methodological framework for scoping reviews as proposed by Arksey and O’Malley (2005). Research questions 1. What are the mechanical properties of the SHS plus TSP compared with SHS alone or intramedullary implants? 2. Does the TSP lead to an improved clinical outcome compared with SHS alone or IMN? 3. How does the TSP function in terms of non-union, mechanical failure, and reoperations? 4. Is it possible to establish guidelines for TSP use based on the existing evidence? Eligibility criteria All papers, both clinical and biomechanical, reporting outcomes related to TSP use in trochanteric fracture treatment were included in the review. We excluded studies reporting 3 cases or less, or where the TSP was used for indications other than acute trochanteric fractures. Information sources A systematic search through PubMed, Scopus, Web of Science, Embase, and Epistemonikos was performed and last updated on June 25, 2020 by the 1st author (CEA). The com-

Studies included in the review n = 24

Figure 2. Flow chart of papers in the review. The papers identified were from Norway, Sweden, Switzerland, Germany, Taiwan, India, United States, Canada, Northern Ireland, South Korea, and Egypt. Search strategy: Languages: All. Search terms/-strings: Title/Abstract (“trochanteric stabilising plate” OR “trochanteric stabilizing plate” OR “trochanteric stabilisation plate” OR “trochanteric stabilization plate” OR “lateral support plate” OR “trochanter stabilising plate” OR “trochanter stabilizing plate” OR “trochanter stabilization plate” OR “trochanter stabilisation plate”).

plete search strategy is shown in the legend to Figure 2. In addition, we did a backwards search of all references of the papers identified and a forward search of papers citing the identified publications. We also manually searched the reference lists of review papers, meta-analyses, and guidelines until no new papers turned up. Study selection Irrelevant studies were excluded based on title and abstract screening. Full text versions of the remaining studies were screened for eligibility by CEA and FF. Funding and potential conflicts of interest No funding was received. The authors declare no conflicts of interest.

Results Study selection After removal of duplicates, 93 unique papers were identified for further analysis (Figure 2). Based on screening of title and/ or abstract, 60 records were excluded, leaving 33 studies for full text evaluation. Of these 33 studies, 14 were excluded for

Acta Orthopaedica 2021; 92 (6): 733–738 735

various reasons (Figure 2). The citation searches identified 5 additional papers (Babst et al. 1993, David et al. 1996, Friedl and Clausen 2001, Klinger et al. 2005, Bonnaire et al. 2007). Summary of the literature 6 biomechanical studies (Table 1, see Supplementary data) and 18 clinical studies (Tables 2 and 3, see Supplementary data) were identified. The clinical studies all reported surgical outcomes, and all but 2 reported at least 1 relevant clinical outcome. All papers studied unstable trochanteric fractures or fracture models. 16 studies compared SHS plus TSP (SHS/TSP) with other implants. In 3 studies the comparator was SHS without TSP (Su et al. 2003, Hsu et al. 2015, Haddon et al. 2019). In 8 studies SHS/TSP was compared with IMN (Götze et al. 1998, Friedl and Clausen 2001, Nuber et al. 2003, Bong et al. 2004, Klinger et al. 2005, Bonnaire et al. 2007, Walmsley et al. 2016, Fu et al. 2020), and in 3 studies SHS/TSP were compared with both SHS and IMN (Madsen et al. 1998, Tucker et al. 2018, Müller et al. 2020). In addition, SHS/TSP was compared with other extramedullary implants in 2 papers (Lunsjö et al 2001, Selim et al. 2020). The lack of standardized methods and reporting of outcomes made a meta-analysis infeasible. We identified 1,091 clinical cases in the literature where a TSP had been used as an adjunct to the SHS. Overall, 46 cases (4%) of mechanical failures and non-unions were reported. The number of reoperations was 82 (8%), including 19 routine removals of implants. The 10 prospective trials reported 15 reoperations (4%), while 67 (10%) reoperations were reported in the retrospective trials. Biomechanical studies SHS/TSP versus SHS (Table 1, see Supplementary data) Su et al. (2003) studied unstable trochanteric fracture models in 10 matched pairs of embalmed femora instrumented with an SHS with or without a TSP. The addition of the TSP to the SHS led to decreased displacement of the head fragment after cyclic loading at 750N. SHS/TSP versus IMN (Table 1, see Supplementary data) 2 studies compared SHS/TSP with an IMN in both cadaveric and synthetic femora using various osteotomies. Friedl and Clausen (2001) concluded that the IMN was more resistant to deformation on cyclic loading than SHS with TSP. Götze et al. (1998) reported a higher load to failure with IMN than SHS with TSP. To compare the biomechanical properties of the SHS plus a TSP with an IMN, unstable, 4-part trochanteric fractures were created in 6 pairs of cadaveric human femora, matched by bone mineral density (BMD), by Bong et al. (2004). In their study the SHS plus TSP provided equal stability and similar ability to resist femoral shaft medialization as the IMN at 250–750 N loading. Walmsley et al. (2016) created unstable intertrochanteric fractures in 24 artificial femora showing similar stiffness but lower axial compression strength

when SHS/TSP was compared with an IMN. Bonnaire et al. (2007) studied the influence of BMD on the risk of lag screw cut out in a trochanteric fracture model. They compared fixation with SHS/TSP with 2 types of IMN using cyclic loading at 2000N and found that if BMD was above 0.6 g/cm3 all implants provided sufficient stability to avoid fixation failure. Clinical studies Studies reporting SHS/TSP without comparator (Tables 2 and 3, see Supplementary data) We identified 8 patient series without comparator including from 17 to 46 patients, 234 in total. The TSP was mainly used in unstable fractures (Tables 3 and 4) (Babst et al. 1993, Hoffmann et al. 1994, David et al. 1996, Babst et al. 1998, Gupta et al. 2010, Cho et al. 2011, Prabhakar and Singh 2016, Shetty et al. 2016). The reporting of surgical and clinical results varied. All papers reported number of reoperations, and at least 1 radiographic and functional outcome. 4 papers reported results of a functional outcome score. All concluded that SHS plus TSP was a viable treatment option. Studies comparing SHS plus TSP with SHS alone or with other extramedullary implants (Table 3, see Supplementary data) SHS plus TSP was compared with SHS alone in 2 studies and with other extramedullary implants in 2 studies. In the only randomized trial in this review, 100 patients with unstable trochanteric fractures were randomized to SHS with or without a TSP. No clinically relevant differences between the groups were found, either in complications, secondary fracture displacement, or functional results (Haddon et al. 2019). Hsu et al. (2015) reported on 252 patients with AO/OTA 31 A2 trochanteric fractures. 205 patients were operated on with an SHS alone and 47 with SHS plus TSP. They performed a risk analysis for postoperative lateral wall fracture (LWF) and found that a lateral wall thickness (LWT) of less than 22 mm strongly predicted a postoperative fracture of the lateral wall. Further, they compared SHS alone (n = 125) and SHS plus TSP (n = 46) as treatment of fractures with a LWT < 22 mm and found that the TSP decreased lag screw sliding and reoperation rate. Lunsjö et al. (2001) performed a secondary analysis of a randomized trial with 569 patients with unstable trochanteric fractures. At the surgeon’s discretion 49 patients were operated on with an SHS and a TSP. No important difference was found between patients operated on with SHS and a TSP compared with patients operated with a Medoff plate or SHS without TSP. Selim et al. (2020) compared SHS plus TSP with a proximal femoral locking plate. The authors found better functional outcome, shorter time to union, and a lower failure rate in the SHS group. Studies comparing SHS plus TSP with IMN (Table 3, see Supplementary data) Nuber et al. (2003) compared SHS plus TSP with IMN in unstable trochanteric and subtrochanteric fractures and

736

reported slightly better functional results and less pain in patients treated with an IMN. Complication rates and patient satisfaction were similar between the groups. Klinger et al. (2005) compared 51 patients treated with SHS/TSP with 122 patients treated with IMN. They found shorter operating time and hospital stay and fewer complications in the IMN group, but no differences in functional results. In the largest study (n = 234) included in the review, Fu et al. (2020) found no difference in functional scores, fracture healing, failure rate, or rate of reoperation when comparing SHS plus TSP with IMN in both AO type A2 and A3 fractures Studies comparing SHS plus TSP with both SHS alone and with IMN (Table 3, see Supplementary data) Madsen et al. (1998) compared a consecutive series of 85 patients with unstable trochanteric fractures treated with SHS plus TSP with 170 patients randomized to either an IMN or an SHS. They found a trend towards better functional results and less lag screw sliding in the TSP group, but an even distribution of complications. In a register-based study by Tucker et al. (2018) reporting on more than 3,000 fractures with IMN (598), SHS (2,474), and SHS plus TSP (158), a tendency towards fewer reoperations and better clinical results with IMN was found. Another retrospective cohort compared SHS with or without TSP and IMN (AO/OTA A2 fractures only) and reported a non-significant tendency toward fewer reoperations after IMN (Müller et al. 2020).

Discussion The identified studies presented mainly low-level evidence with only 1 prospective comparison and 1 relatively small randomized controlled trial. All studies reported on unstable fractures or fracture models. A meta-analysis was not possible due to the heterogeneity of the studies. Research question 1. What are the mechanical properties of the SHS plus TSP compared with SHS alone or with IMN? The testing circumstances in the trials varied. 3 trials (Götze et al. 1998, Friedl and Clausen 2001, Walmsley et al. 2016) used supraphysiological loads, while in 2 trials (Su et al. 2003, Bong et al. 2004) the load applied was below normal loading associated with gait (Duda et al. 1997). In addition, the fracture models were simple, and thus not comparable to the comminution frequently seen in clinical practice. This complicates the interpretation of the results and limits the clinical value. In 3 trials (Götze et al. 1998, Friedl and Clausen 2001, Walmsley et al. 2016) composite bones were used alone, or in combination with cadaver specimens. Synthetic bone is probably not adequate when testing a typically osteoporotic fracture model and the results may be of limited value, as the model does not mimic the bone loss predominant in hip fracture patients (Basso et al. 2014).

Acta Orthopaedica 2021; 92 (6): 733–738

1 trial using cadaveric specimens found that SHS plus TSP provided sufficient stability within a clinical bone density range (Bonnaire et al. 2007). The biomechanical studies using IMN as a comparator all showed that the TSP provided comparable stability to intramedullary nails (Table 2). The only biomechanical study comparing SHS with and without TSP (Su et al. 2003) used a highly unstable fracture model (AO/OTA A3) and found less displacement with an additional TSP. In comparison, the SHS alone has been reported to have less ability to withstand deformation after cyclic loading than IMN (Kaiser et al. 1997, Sommers et al. 2004). Thus, the TSP appears to add stability to the osteosynthesis up to a level comparable with IMN. Research question 2. Does the TSP lead to an improved clinical outcome compared with SHS alone or with IMN? The only randomized trial comparing SHS plus TSP with SHS alone was powered to detect a difference in lag screw sliding of 4 millimeters between SHS with and without TSP (Haddon et al. 2019). At 1-year follow-up a difference in lag screw sliding of less than 1 mm was found between the groups. With a broken lateral wall (n = 44) the difference was 3 mm in favor of the TSP group (not statistically significant). In the main clinical outcome measure in the trial, the Merle d’AubignePostel score, a statistically non-significant 0.7 difference in favor of the group treated with SHS alone was found. The trial was not powered for subgroup analyses, but even with a larger number of patients included it is improbable that a meaningful clinical difference would have occurred. A thin or fractured lateral wall may, however, be a predictor of mechanical failure (Palm et al. 2007, Hsu et al. 2015) and the TSP may have a beneficial effect under these circumstances as reported by Hsu et al. (2015). A few studies (Madsen et al. 1998, Nuber et al. 2003, Klinger et al. 2005, Tucker et al. 2018, Müller et al. 2020) included in this review compared SHS plus TSP indirectly with IMN or SHS without TSP. From these studies it may be argued that the TSP protected against secondary fracture displacement. Madsen et al. observed a trend towards better functional results in the TSP group while the other publications failed to show any functional benefit of the TSP compared with IMN. The findings above may be seen in light of 2 RCTs comparing SHS without TSP with IMN. Parker et al. (2017) included both stable and unstable fractures in a large trial. The authors reported slightly better regain of mobility in patients operated on with an IMN. Hardy et al. (1998) reported similar results in a randomized study of 100 patients. They explained their findings, at least in part, by the significantly larger lag screw sliding distance and subsequent limb shortening in the SHS group. Based on the existing evidence it is not possible to conclude whether the TSP offers better clinical results than SHS alone, or when SHS plus TSP was compared with IMN for unstable trochanteric fractures.

Acta Orthopaedica 2021; 92 (6): 733–738 737

Research question 3. How does the TSP function in terms of non-union, mechanical failure, and reoperations? A total of 1,091 SHS plus TSP were reported with 46 (4%) cases of healing problems and 82 (8%) reoperations for any cause. The 2010 Cochrane review (Parker and Handoll 2010), also including a high number of stable fractures, reported a (4%) reoperation rate and 3–4% healing complications and failures after SHS. In the same review the authors found an increased relative risk of cutout with the IMN, but a reduced risk of non-union (both statistically non-significant). A study from the Norwegian Hip Fracture Register (Matre et al. 2013a) reported 10% reoperations at 3 years after AO/ OTA A3 fractures and subtrochanteric fractures treated with SHS with or without TSP compared with 7% in the IMN group. This contrasts the findings in a randomized trial (Matre et al. 2013b) comparing IMN to SHS with or without TSP where a similar reoperation rate of 8% was reported after 12 months. 10 of the 19 included clinical trials were retrospective cohorts and chart reviews vulnerable to an under-reporting of serious complications and reoperations, as the patients may have sought advice elsewhere, or not at all (Table 2). 1 trial was from a register, equally prone to reporting minimum numbers of revision surgeries (Tucker et al. 2018). There was, however, no tendency to more reoperations in the prospective trials compared with the retrospective trials in our material. The rate of complications and reoperations for SHS plus TSP was comparable to previous reports on trochanteric fractures treated with SHS alone or IMN. Research question 4. Is it possible to work out guidelines for TSP use based on the existing evidence? All reports included in the review were on unstable fracture models or fractures. This implies that no authors believe that the TSP has a role in stable fractures (AO/OTA 31 A1 and Evans Jensen I–II). The results of the randomized trial comparing SHS with or without TSP suggest that the TSP has at best a limited role (Haddon et al. 2019). Some papers report, however, that the TSP increases stability compared with SHS alone (Su et al. 2003, Hsu et al. 2015) and with a similar stability to IMN (Madsen et al. 1998, Bong et al. 2004). The limited literature identified, and the heterogeneity of methods and results, precludes any clear recommendations on when to use the TSP, or if it should be used at all. However, it might be argued that in practices where IMN is not available the TSP might be beneficial when treating trochanteric fractures with a thin or compromised lateral wall. Strengths and limitations We believe that our literature search is exhaustive, and we have included both biomechanical and clinical trials. Some papers were not included due to insufficient reporting or failure to obtain a translation. A synthesis of functional results was not possible.

Conclusion This review did not identify literature clearly advising when to use a TSP. The findings indicated, however, that the TSP may provide a more stable construct, reducing lag screw sliding and medialization of the femoral shaft, than the SHS alone in unstable trochanteric fractures. Whether this translates into improved clinical outcomes compared with SHS alone or with IMN remains unclear. There is a need for high-quality, well-powered clinical trials with relevant outcome measures to clarify any role of the TSP in the treatment of trochanteric fractures. Supplementary data Tables 1–3 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674. 2021.1954305

All authors contributed to study planning. CEA and FF selected the studies and extracted data. CEA wrote the 1st draft and all authors revised the manuscript. Acta thanks Micha Holla and Olof Wolf for help with peer review of this study.

Alm C E, Frihagen F, Dybvik E, Matre K, Madsen J E, Gjertsen J-E. Implants for trochanteric fractures in Norway: the role of the trochanteric stabilizing plate—a study on 20,902 fractures from the Norwegian hip fracture register 2011–2017. J Orthop Surg Res 2021; 16(1): 26. Arksey H, O’Malley L. Scoping studies: towards a methodological framework. Int J Soc Res Methodol 2005; 8(1): 19-32. Babst R, Martinet O, Renner N, Rosso R, Bodoky A, Heberer M, Regazzoni P. The dynamic hip screw support plate for management of unstable proximal femoral fractures. Helv Chir Acta 1993; 59(4): 521-5. Babst R, Renner N, Biedermann M, Rosso R, Heberer M, Harder F, Regazzoni P. Clinical results using the trochanter stabilizing plate (TSP): the modular extension of the dynamic hip screw (DHS) for internal fixation of selected unstable intertrochanteric fractures. J Orthop Trauma 1998; 12(6): 392-9. Basso T, Klaksvik J, Syversen U, Foss O A. A biomechanical comparison of composite femurs and cadaver femurs used in experiments on operated hip fractures. J Biomech 2014; 47(16): 3898-902. Bong M R, Patel V, Iesaka K, Egol K A, Kummer F J, Koval K J. Comparison of a sliding hip screw with a trochanteric lateral support plate to an intramedullary hip screw for fixation of unstable intertrochanteric hip fractures: a cadaver study. J Trauma 2004; 56(4): 791-4. Bonnaire F, Weber A, Bösl O, Eckhardt C, Schwieger K, Linke B. [“Cutting out” in pertrochanteric fractures—problem of osteoporosis?]. Unfallchirurg 2007; 110(5): 425-32. Bretherton C P, Parker M J. Femoral medialization, fixation failures, and functional outcome in trochanteric hip fractures treated with either a sliding hip screw or an intramedullary nail from within a randomized trial. J Orthop Trauma 2016; 30(12): 642-6. Cho S H, Lee S H, Cho H L, Ku J H, Choi J H, Lee A J. Additional fixations for sliding hip screws in treating unstable pertrochanteric femoral fractures (AO Type 31-A2): short-term clinical results. Clin Orthop Surg 2011; 3(2): 107-13. David A, Hufner T, Lewandrowski K U, Pape D, Muhr G. [The dynamic hip screw with support plate—a reliable osteosynthesis for highly unstable “reverse” trochanteric fractures?]. Chirurg 1996; 67(11): 1166-73.

738

Duda G N, Schneider E, Chao E Y. Internal forces and moments in the femur during walking. J Biomech 1997; 30(9): 933-41. Friedl W, Clausen J. Experimental examination for optimized stabilisation of trochanteric femur fractures: intra- or extramedullary implant localisation and influence of femur neck component profile on cut-out risk. Chirurg 2001; 72(11): 1344-52. Fu C W, Chen J Y, Liu Y C, Liao K W, Lu Y C. Dynamic hip screw with trochanter-stabilizing plate compared with proximal femoral nail antirotation as a treatment for unstable AO/OTA 31-A2 and 31-A3 intertrochanteric fractures. Biomed Res Int 2020; 2020: 1896935. Götze B, Bonnaire F, Weise K, Friedl H P. Loadability of osteosynthesis of unstable per- and subtrochanteric fractures: an experimental study testing the proximal femoral nail (PFN), the Gamma-nail, the DHS/trochanteric stabilization plate, the 95°angled blade plate and the UFN/spiral blade. Aktuelle Traumatol 1998; 28(5): 197-204. Gupta R K, Sangwan K, Kamboj P, Punia S S, Walecha P. Unstable trochanteric fractures: the role of lateral wall reconstruction. Int Orthop 2010; 34(1): 125-9. Haddon J, Buciuto R, Johnsen L G. A prospective randomized trial of 100 patients using trochanteric support plates; worth their mettle? Injury 2019; 50(3): 733-7. Hardy D C, Descamps P Y, Krallis P, Fabeck L, Smets P, Bertens C L, Delince P E. Use of an intramedullary hip-screw compared with a compression hip-screw with a plate for intertrochanteric femoral fractures: a prospective, randomized study of one hundred patients. J Bone Joint Surg Am 1998; 80(5): 618-30. Higgins J, Thomas J, Chandler J, Cumpston M, Li T, Page M J, Welch V A. Cochrane handbook for systematic reviews of interventions version 6.1. Cochrane 2020. Available from: https://www.cochranelibrary.com/. Hoffmann T F, Berteloot F, Renneker D. The DHS-trochanter-stabilisingplate: first results after implantation in 20 patients. Aktuelle Traumatol 1994; 24(8): 295-300. Hsu C E, Shih C M, Wang C C, Huang K C. Lateral femoral wall thickness: a reliable predictor of post-operative lateral wall fracture in intertrochanteric fractures. Bone Joint J 2013; 95-b(8): 1134-8. Hsu C E, Chiu Y C, Tsai S H, Lin T C, Lee M H, Huang K C. Trochanter stabilising plate improves treatment outcomes in AO/OTA 31-A2 intertrochanteric fractures with critical thin femoral lateral walls. Injury 2015; 46(6): 1047-53. Kaiser W, Burmester J, Hausmann H, Gulielmos V, Hätzel M, Merker H J. [Comparative stability evaluation of dynamic hip screw and gamma-nail osteosyntheses in unstable pertrochanteric femoral osteotomies]. Langenbecks Arch Chir 1997; 382(2): 100-6. Klinger H M, Baums M H, Eckert M, Neugebauer R. Eine vergleichende Untersuchung der Versorgung instabiler per- und intertrochantärer Femurfrakturen mittels DHS-Osteosynthese unter Verwendung der Trochanterabstützplatte und dem Proximalen Femurnagel (PFN) [A comparative study of unstable per- and intertrochanteric femoral fractures treated with dynamic hip screw (DHS) and trochanteric butt-press plate vs. proximal femoral nail (PFN)]. Zentralbl Chir 2005; 130(4): 301-6. Knobe M, Gradl G, Ladenburger A, Tarkin I S, Pape H C. Unstable intertrochanteric femur fractures: is there a consensus on definition and treatment in Germany? Clin Orthop Relat Res 2013; 471(9): 2831-40. Lunsjö K, Ceder L, Thorngren K G, Skytting B, Tidermark J, Berntson P O, Allvin I, Norberg S, Hjalmars K, Larsson S, Knebel R, Hauggaard A, Stigsson L. Extramedullary fixation of 569 unstable intertrochanteric fractures: a randomized multicenter trial of the Medoff sliding plate versus three other screw-plate systems. Acta Orthop 2001; 72(2): 133-40. Madsen J E, Naess L, Aune A K, Alho A, Ekeland A, Stromsoe K. Dynamic hip screw with trochanteric stabilizing plate in the treatment of unstable proximal femoral fractures: a comparative study with the Gamma nail and compression hip screw. J Orthop Trauma 1998; 12(4): 241-8.

Acta Orthopaedica 2021; 92 (6): 733–738

Matre K, Havelin L I, Gjertsen J E, Vinje T, Espehaug B, Fevang J M. Sliding hip screw versus IM nail in reverse oblique trochanteric and subtrochanteric fractures: a study of 2716 patients in the Norwegian Hip Fracture Register. Injury 2013a; 44(6): 735-42. Matre K, Vinje T, Havelin L I, Gjertsen J E, Furnes O, Espehaug B, Kjellevold S H, Fevang J M. TRIGEN INTERTAN intramedullary nail versus sliding hip screw: a prospective, randomized multicenter study on pain, function, and complications in 684 patients with an intertrochanteric or subtrochanteric fracture and one year of follow-up. J Bone Joint Surg Am 2013b; 95(3): 200-8. Mellema J J, Janssen S, Schouten T, Haverkamp D, van den Bekerom M P J, Ring D, Doornberg J N. Intramedullary nailing versus sliding hip screw for A1 and A2 trochanteric hip fractures. Bone Joint J 2021; 103b(4): 775-81. Müller F, Doblinger M, Kottmann T, Füchtmeier B. PFNA and DHS for AO/OTA 31-A2 fractures: radiographic measurements, morbidity and mortality. Eur J Trauma Emerg Surg 2020; 46(5): 947-953. NICE. The management of hip fracture in adults. Available from: https:// www.nice.org.uk/guidance/cg124. Retrieved May 2020. London: National Clinical Guideline Centre; 2011. Nuber S, Schonweiss T, Ruter A. [Stabilisation of unstable trochanteric femoral fractures. Dynamic hip screw (DHS) with trochanteric stabilisation plate vs. proximal femur nail (PFN)]. Der Unfallchirurg 2003; 106(1): 39-47. Palm H, Jacobsen S, Sonne-Holm S, Gebuhr P. Integrity of the lateral femoral wall in intertrochanteric hip fractures: an important predictor of a reoperation. J Bone Joint Surg Am 2007; 89(3): 470-5. Parker M J. Trochanteric hip fractures: fixation failure commoner with femoral medialization, a comparison of 101 cases. Acta Orthop 1996; 67(4): 329-32. Parker M J. Sliding hip screw versus intramedullary nail for trochanteric hip fractures: a randomised trial of 1000 patients with presentation of results related to fracture stability. Injury 2017; 48(12): 2762-7. Parker M J, Handoll H H. Gamma and other cephalocondylic intramedullary nails versus extramedullary implants for extracapsular hip fractures in adults. Cochrane Database Syst Rev 2010(9): CD000093. Parker M, Raval P, Gjertsen J E. Nail or plate fixation for A3 trochanteric hip fractures: a systematic review of randomised controlled trials. Injury 2018; 49(7): 1319-23. Prabhakar S, Singh R P. Does modular extension of DHS fix the unstable trochanteric fractures. J Evol Med Dent Sci 2016; 5(54): 3683-92. Roberts K C, Brox W T. AAOS clinical practice guideline: management of hip fractures in the elderly. J Am Acad Orthop Surg 2015; 23(2): 138-40. Selim A A H A, Beder F K, Algeaidy I T, Farhat A S, Diab N M, Barakat A S. Management of unstable pertrochanteric fractures, evaluation of forgotten treatment options. SICOT-J 2020; 6: 21. Shetty A, Ballal A, Sadasivan A K, Hegde A. Dynamic hip screw with trochanteric stablization plate fixation of unstable inter-trochanteric fractures: a prospective study of functional and radiological outcomes. J Clin Diagn Res 2016; 10(9): RC06-RC8. Sommers M B, Roth C, Hall H, Kam B C, Ehmke L W, Krieg J C, Madey S M, Bottlang M. A laboratory model to evaluate cutout resistance of implants for pertrochanteric fracture fixation. J Orthop Trauma 2004; 18(6): 361-8. Su E T, DeWal H, Kummer F J, Koval K J. The effect of an attachable lateral support plate on the stability of intertrochanteric fracture fixation with a sliding hip screw. J Trauma 2003; 55(3): 504-8. Tucker A, Donnelly K J, Rowan C, McDonald S, Foster A P. Is the best plate a nail? A review of 3230 unstable intertrochanteric fractures of the proximal femur. J Orthop Trauma 2018; 32(2): 53-60. Walmsley D, Nicayenzi B, Kuzyk P R, Machin A, Bougherara H, Schemitsch E H, Zdero R. Biomechanical analysis of the cephalomedullary nail versus the trochanteric stabilizing plate for unstable intertrochanteric femur fractures. Proc Inst Mech Eng H 2016; 230(12): 1133-40.

Acta Orthopaedica 2021; 92 (6): 739–745

739

30-day and 1-year mortality after skeletal fractures: a register study of 295,713 fractures at different locations Camilla BERGH 1,2, Michael MÖLLER 1,2, Jan EKELUND 3, and Helena BRISBY 1,2 1 Institute

of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg; 2 Department of Orthopaedics, Sahlgrenska University Hospital, Gothenburg; 3 Centre of Registers Västra Götaland, Gothenburg, Sweden Correspondence: camilla.m.bergh@vgregion.se Submitted 2021-03-27. Accepted 2021-07-12.

Background and purpose — Few studies have reported the mortality rate after skeletal fractures involving different locations, within the same population. We analyzed the 30-day and 1-year mortality rates following different fractures. Patients and methods — We included 295,713 fractures encountered in patients 16–108 years of age, registered in the Swedish Fracture Register (SFR) from 2012 to 2018. Mortality rates were obtained by linkage of the SFR to the Swedish Tax Agency population register. The standardized mortality ratios (SMR) at 30 days and 1 year were calculated for fractures in any location and for each of 27 fracture locations, using age- and sex-life tables from Statistics Sweden (www.scb.se). Results — The overall SMR at 30 days was 6.8 (95% CI 6.7–7.0) and at 1 year 2.2 (CI 2.2–2.2). The SMR was > 2 for 19/27 and 13/27 of the fracture locations at 30 days and 1 year, respectively. Humerus, femur, and tibial diaphysis fractures were all associated with high SMR, at both 30 days and 1 year. Interpretation — Patients sustaining a fracture had approximately a 7-fold increased mortality at 30 days and over 2-fold increased mortality at 1 year as compared with what would be expected in the general population. High mortality rates were seen for patients with axial skeletal and proximal extremity fractures, indicating frailty in these patient groups.

Compared with other medical conditions, the mortality rate after fractures has been considered to be low, and has not been frequently reported, with the exception of extensive literature on hip femur fractures; for review see Huette et al. (2020). For hip fractures, the importance of organizing care to decrease complications and mortality has been reported (von Friesendorff et al. 2016). Longer waiting time for surgery has reportedly been associated with increased mortality rates in some studies (Schnell et al. 2010, Pincus et al. 2017). A relationship between fractures in different locations and mortality rates can provide information on whether fractures in also other locations should be prioritized for treatment (Vestergaard et al. 2007a, Klop et al. 2017). Fracture types reported to be associated with increased mortality rates are vertebral fractures, distal radius fractures, diaphyseal, and distal femur fractures (Kado et al. 2003, Oyen et al. 2014, Larsen et al. 2020). There are, however, few reports comparing mortality rates for more than a few different fracture locations, within the same population. Hence, comparisons between mortality rates for different fracture locations are difficult. To describe the change in mortality rate associated with a specific condition, the standardized mortality ratio (SMR) is commonly used (Vandenbroucke 1982). We investigated the 30-day and 1-year SMR for patients with fractures in various locations by using data from the Swedish Fracture Register (SFR).

Patients and methods Data collection in the Swedish Fracture Register (SFR) Data collection in the SFR began in 2011 and the data collection procedure has been described in detail by Wennergren et al. (2015). The number of hospitals attached to the SFR has gradually increased and at the end of 2020, 100% coverage © 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 10.1080/17453674.2021.1959003

740

was achieved with participation of all 54 departments treating fractures in Sweden. The completeness of fracture registrations in the SFR compared with the National Patient Register in 2018 was 70–95% for most participating departments. All fractures, regardless of treatment (surgically or nonsurgically), are prospectively registered in the SFR and classified according to the Müller AO/OTA classification system (Müller et al. 1990). Independent validation studies for different fracture locations regarding fracture classification have been performed (Juto et al. 2016, Wennergren et al. 2016, 2017, Knutsson et al. 2019, Morgonsköld et al. 2019). In addition to the classification of the fracture, the physicians/ surgeons responsible for the fracture registration label each fracture according to the trauma mechanism as high, low, or undefined/unknown regarding energy type. The patient’s personal identification number allows for the monitoring of patients over time and enables accurate linkage to other national databases. Fracture classification and calculation of standardized mortality ratios We included all fractures in patients aged 16 years and older, registered in the SFR between January 1, 2012 and December 31, 2018. Based on AO/OTA classification, fractures were divided into 27 anatomical regions (locations). Multiple fractures occurring at the same time, in the same patient, and in the same anatomical region, including bilateral fractures, were counted as 1 fracture for this analysis. Fractures occurring at the same time in different anatomical regions were counted once in each region. Subsequent fractures in a patient, regardless of anatomical region, were included as different entities in the analysis. Data on mortality for patients registered in the SFR were obtained by linkage of the SFR to the Swedish Tax Agency population register. The SMR was calculated for all fracture locations together as well as for each of the 27 fracture locations by using the number of deaths among the fracture patients, divided by the expected number of deaths, based on the age- and sex-specific rates in the overall population and the size of the fracture population. Statistics Mortality rates at 30 days and at 1 year were calculated as the number of patients who died divided by the total number of patients (for each fracture localization, and total, as appropriate) and expressed as a percentage. The SMR was calculated using mortality among patients registered in the SFR and the corresponding life tables for 2012–2018 retrieved from Statistics Sweden (www.scb.se). The life tables used report the 1-year mortality rates for each year of age and sex separately, for each of the relevant years. When calculating SMR for the 30-day period, this was done under the assumption that the expected number of deaths during 30 days would be 1/12th of that expected at 1 year based on the 1-year life tables.

Acta Orthopaedica 2021; 92 (6): 739–745

Table 1. Overall descriptive data Sex Men Women All

Number of fractures 120,596 175,117 295,713

mean (SD) 52 (23) 66 (20) 60 (22)

Age

median (range) 51 (16–107) 69 (16–108) 64 (16–108)

The SMR was calculated as the ratio between observed and expected mortality with 95% confidence interval (CI), according to the method by Vandenbroucke (1982). All calculations for tables and figures were performed using SAS (v9.4; SAS Institute, Cary, NC, USA). Ethics, funding, data sharing, and potential conflicts of interest This study was approved by the Central Ethical Review Board, Gothenburg (ID 792-17). This research was supported by grants from the Swedish Research Council, Government Funding of Clinical Research within the National Health Service (ALF), from Västra Götaland ALFGBG722931, the Felix Neubergh Foundation, and the Gothenburg Medical Association, all in Sweden. The data that supports the findings of this study is available from the corresponding author on reasonable request. The authors declare that they have no competing interests.

Results Baseline characteristics During the study period 295,713 fractures were registered in the SFR (based on the anatomic regions definition, i.e., we analyzed 1 fracture from the same location for multiple fractures within the same location, at the same time of injury). These were sustained during 284,625 injury occasions, where 274,934 injury occasions involved a single location (97% of injuries), and 9,691 injury occasions involved more than 1 location (3.4%). For patients sustaining 2 or more fractures, the average number of fractures was 2.14. The cohort included 262,598 patients (59% women) (Tables 1 and 2). High-energy trauma accounted for 8% of the fractures and was more commonly registered in men (15%) than in women (3.8%). For injuries involving more than 1 location the 30-day mortality was 2.3% (SMR 7.3, CI 6.4–8.3) and for single location injuries 2.0% (SMR 6.8, CI 6.6–7.0) for single location injuries. The corresponding numbers at 1 year were 7.9% (SMR 2.1, CI 2.0–2.3) and 7.8% (SMR 2.2, CI 2.2–2.2). Overall mortality rates The overall mortality for fracture patients (any location) in the study population was markedly higher compared with expected mortality in the general population. The SMR was

Acta Orthopaedica 2021; 92 (6): 739–745

741

Table 2. Fractures in all locations, fractures in another location at the same time, and a new fracture within a year. Values are count (%) Fracture location

Number of Fracture in fractures another location

New fracture within a year

30-day mortality rate (%)

30-day SMR (CI)

Male Female Total Ref. male Ref. female Ref. total

Male Female Total

All fractures Spine Pelvis Acetabulum Femur proximal Femoral diaphysis Femur distal Humerus proximal Humeral diaphysis Humerus distal Distal radius Tibia proximal Tibia diaphysis Tibia distal

295,713 7,658 8,793 1,718 1,355 2,786 2,476 23,572 3,267 2,379 50,610 6,450 3,233 2,283

20,779 (7.0) 768 (10.0) 1,207 (13.7) 374 (21.8) 1,874 (3.6) 300 (10.8) 309 (12.5) 2,009 (8.5) 368 (11.3) 388 (16.3) 2,927 (5.8) 629 (9.8) 358 (11.1) 274 (12.0)

10,502 (3.6) 291 (3.8) 562 (6.4) 71 (4.1) 2,620 (5.1) 130 (4.7) 123 (5.0) 998 (4.2) 174 (5.3) 120 (5.0) 1,679 (3.3) 204 (3.2) 87 (2.7) 56 (2.5)

30 10 20 5

0 0

100

Age (years)

100

Age (years)

Figure 2. 30-day mortality rate of all fractures for different ages (from 16 years), and sorted by sex in percentages (left). Normal population reference values included. 30-day SMR with 95% CI for all fractures for different ages (from 16 years) and sorted by sex (right). There were no observed deaths within 30 days for individuals below 24 years.

1-year mortality rate (%) 60

1-year SMR (CI) 9

Male Female Total Ref. male Ref. female Ref. total

Male Female Total

8 7 6 5

30 4 20

3 2

10 1 0

0 0

100

Age (years)

100

Age (years)

Figure 3. 1-year mortality rate for all fractures for different ages (from 16 years) and sorted by sex in percentages (left). Normal population reference values included. 1-year SMR with 95% CI for all fractures for different ages (from 16 years) and sorted by sex (right).

Figure 1. 30-day (left) and 1-year (right) SMR in different body locations illustrated with color intensity, based on SMR figures. Cut-offs and SMR color codes are different for 30-day and 1-year SMR.

6.8 at the 30-day time period and 2.2 for the 1-year time period (Table 3). The 30-day SMR in patients with fractures caused by high-energy trauma was 7.5 and 6.8 for low-energy trauma. Corresponding numbers at 1 year were 1.8 and 2.1

respectively (Table 3). An SMR > 2 was observed for 19/27 and 13/27 fracture locations at the 30-day and 1-year time points, respectively (Tables 4, 5, and 6). The SMR at 30 days and 1 year for the different fracture locations are illustrated in Figure 1. Overall, mortality rates as well as SMRs were higher for men aged ≥ 60 years, compared with women, at both the 30-day and at 1-year time points (Figures 2 and 3). The large CI due to small absolute numbers of fractures seen for patients of younger age makes comparisons between sexes difficult in these age groups.

Table 3. 30-day and 1-year mortality rate in patients sustaining a fracture at any location Proportion dead (%) Follow-up Number of Mean age Proportion Expected energy trauma time fractures years (SD) dead (%) dead (%) SMR (CI) high low 30-days 1-year

295,713 295,713

60 (22) 60 (22)

2.0 7.8

0.3 3.5

6.8 (6.6–7.0) 2.2 (2.2–2.2)

0.4 1.2

2.2 8.4

SMR (CI) energy trauma high low

7.5 (6.1–9.0) 1.8 (1.6–2.0)

6.8 (6.6–6.9) 2.1 (2.1–2.2)

742

Acta Orthopaedica 2021; 92 (6): 739–745

Table 4. Mortality rates at 30 days and 1 year for patients sustaining shoulder and upper extremity fractures 30 days 1 year HighFracture Number of Mean age energy Proportion Expected Proportion Expected location fractures years (SD) (%) dead (%) dead (%) SMR (CI) dead (%) dead (%) SMR (CI) Scapula Clavicle Humerus proximal Humerus diaphysis Humerus distal Forearm proximal Forearm diaphysis Forearm distal (wrist) Carpal Metacarpal Phalanx

2,957 10,319 23,572 3,267 2,379 11,701 2,204 50,610 4,778 16,821 17,267

58 (19) 49 (22) 69 (16) 63 (21) 66 (21) 51 (20) 50 (23) 61 (19) 42 (20) 41 (21) 46 (20)

23.3 22.1 3.9 9.6 8.0 7.5 23.0 5.3 11.0 8.3 12.2

0.6 0.6 1.7 3.2 2.4 0.4 0.4 0.4 < 0.1 0.1 0.1

0.2 0.2 0.3 0.3 0.3 0.1 0.2 0.2 0.1 0.1 0.1

3.7 (2.2–5.6) 4.1 (3.1–5.1) 5.3 (4.8–5.9) 11.0 (9.0–13.0) 7.0 (5.3–8.9) 3.2 (2.4–4.3) 2.9 (1.4–5.0) 1.9 (1.6–2.1) 0.7 (0.1–2.0) 1.0 (0.6–1.6) 1.0 (0.5–1.6)

4.3 4.0 7.4 12.4 9.7 2.3 3.5 2.8 0.8 1.4 1.2

2.1 1.8 3.8 3.6 4.2 1.4 1.9 2.5 0.7 1.0 1.0

2.0 (1.7–2.4) 2.2 (2.0–2.4) 2.0 (1.9–2.0) 3.5 (3.1–3.8) 2.3 (2.0–2.6) 1.6 (1.4–1.8) 1.8 (1.4–2.3) 1.1 (1.1–1.2) 1.0 (0.7–1.4) 1.4 (1.2–1.6) 1.2 (1.0–1.3)

Table 5. Mortality rates at 30 days and 1 year for patients sustaining fractures in the lower extremities 30 days 1 year HighFracture Number of Mean age energy Proportion Expected Proportion Expected location fractures years (SD) (%) dead (%) dead (%) SMR (CI) dead (%) dead (%) SMR (CI) Acetabulum Femur proximal Femur diaphysis Femur distal Patella Tibia proximal Tibia diaphysis Tibia distal Ankle Talus Calcaneus Midfoot Metatarsal Toe phalanx

1,718 51,355 2,786 2,476 3,700 6,450 3,233 2,283 32,975 960 1,919 2,212 12,475 8,845

71 (19) 81 (11) 71 (22) 73 (19) 62 (20) 56 (20) 50 (21) 51(21) 55 (19) 39 (17) 48 (18) 43 (18) 48 (20) 46 (18)

25.2 1.4 15.6 7.3 6.1 17.4 23.5 23.8 5.5 38.8 35.0 20.9 5.8 6.5

4.3 7.5 6.2 4.8 0.4 0.7 1.1 0.4 0.3 0.1 0.2 0.1 0.1 0.1

0.5 0.8 0.6 0.5 0.2 0.2 0.1 0.1 0.1 < 0.1 0.1 < 0.1 0.1 0.1

8.2 (6.4–10.1) 10.0 (9.7–10.3) 11.0 (9.4–12.8) 8.9 (7.4–10.6) 1.7 (0.9–2.6) 4.2 (3.0–5.5) 8.2 (5.7–11.1) 3.0 (1.4–5.2) 1.9 (1.6–2.4) 3.2 (0.0–12.4) 2.9 (0.8–6.5) 3.2 (0.6–7.7) 0.8 (0.4–1.5) 1.6 (0.7–2.9)

15.5 24.6 18.3 17.7 3.0 3.8 4.8 4.2 2.2 0.4 1.3 9.0 1.4 0.8

6.4 9.2 6.8 6.6 2.8 2.0 1.7 1.8 1.6 0.4 0.9 0.5 1.1 0.7

2.4 (2.1–2.7) 2.7 (2.6–2.7) 2.7 (2.5–2.9) 2.7 (2.4–2.9) 1.1 (0.9–1.3) 1.9 (1.7–2.1) 2.9 (2.5–3.4) 2.4 (1.9–2.9) 1.4 (1.3–1.5) 1.0 (0.3–2.3) 1.5 (1.0–2.2) 1.7 (1.0–2.6) 1.3 (1.1–1.5) 1.1 (0.9–1.4)

Table 6. Mortality rates at 30 days and 1 year for patients sustaining spinal and pelvic fractures 30 days 1 year HighFracture Number of Mean age energy Proportion Expected Proportion Expected location fractures years (SD) (%) dead (%) dead (%) SMR (CI) dead (%) dead (%) Spine Pelvis

7,658 8,793

64 (22) 75 (19)

25.0 10.9

2.4 3.8

Mortality rates at 30-day and 1-year for upper extremity fractures (including shoulder/scapula) For the 145,875 fractures that the patients sustained in the upper extremities, the mortality rate was 0.4% (n = 519) at 30 days and 3.9% (n = 5,704) at 1 year. High mortality rates at 30 days were seen for humeral diaphysis (12%) and

0.3 0.6

6.9 (5.9–8.0) 6.1 (5.4–6.8)

10.1 16.7

4.2 7.7

SMR (CI) 2.4 (2.3–2.6) 2.2 (2.1–2.3)

distal humerus fractures (7.9%), with SMR of about 11 and 7, respectively. At the 1-year time point, SMR for humeral diaphysis and distal humerus fractures was 3.5 and 2.3, respectively. Fractures in the shoulder region and of the proximal part of the upper extremity also had SMRs ≥ 2 at both time points (Table 4, Figure 2).

Acta Orthopaedica 2021; 92 (6): 739–745

Mortality rates at 30 days and 1 year for lower extremity fractures (including acetabulum) For the 133,387 fractures in the lower extremities, the mortality rate was 3.6% (n = 4,826) at 30 days and 15% (n = 20,139) at 1 year. Almost all lower extremity fracture types revealed an SMR of 3 or above at 30 days. Fractures of the femur (all 3 locations) had SMR at 30 days of ≥ 8. At the 1-year time point, all femur fracture locations and tibia diaphysis fractures were associated with an SMR ≥ 2. (Table 5, Figure 1). Mortality rates at 30 days and 1 year for spine and pelvis fractures Fractures of the spine and pelvis were associated with an SMR of roughly 6 at the 30-day time point and 2 at the 1-year time point (Table 6, Figure 1).

Discussion In this national register study, we found that fractures of most locations had an increased mortality rate. For all locations a 7-fold higher mortality at 30 days and 2-fold higher mortality at 1 year were seen, when compared with expected mortality in the general population with the same age and sex distribution. The mortality rates varied largely across different fracture locations. Proximal extremity fractures of both the upper and lower extremities, as well as vertebrae and pelvic fractures, were all associated with high SMR. The mortality rate is often reported for other acute medical conditions, e.g., myocardial infarction and stroke. The mortality rate within the first month was reported to be 25% after stroke and approximately 24% after myocardial infarction in Sweden in 2019 (National Board of Health and Welfare, Sweden 2020). The corresponding 30-day mortality rates in our study for proximal and diaphyseal femur fractures were 8% and 6%, respectively. The 1-year mortality rate after a myocardial infarction in 2019 in Sweden was approximately 33% (National Board of Health and Welfare, Sweden 2020), which is comparable to 25% for hip fractures and 18% for femoral diaphysis fractures. Most fracture locations in the proximal and diaphyseal parts of the long bones, vertebrae, and pelvis were associated with an SMR > 2 at 1 year. An increase in mortality after lower extremity fractures has been previously reported (Somersalo et al. 2016, Huette et al. 2020), confirmed in our study by high SMR for patients with femur fractures, acetabulum fractures, and tibial shaft fractures. Hip fractures are among the most studied and common fractures. Proximal femur fractures had the highest mortality rate among all fracture locations; 25% at 1 year with a corresponding SMR of 2.7, which is in accordance with previous studies (Vestergaard et al. 2007b, Gundel et al. 2020). For comparison, distal femur fractures had a 30-day mortality rate of 4.8%, which is similar to the 5% mortality reported by Larsen et al. (2020) and 6.3% by Wolf et al. (2021). The 18%

743

1-year mortality for distal femur fracture patients in our study was lower than the 25% for proximal femur fractures, with a similar SMR of 2.7. In the upper extremities, the diaphyseal and distal humerus fractures were associated with an SMR of 11 and 7.0, respectively, at 30 days, and of 3.5 and 2.3, respectively, at 1 year. Proximal humerus fractures were associated with an SMR of 5.3 at 30 days and 2.0 at 1 year. These results are in accordance with previous studies on proximal humerus fractures (Bercik et al. 2013, Wilson et al. 2014). A higher mortality rate was also seen after scapular and clavicular fractures at 1 year, both locations demonstrating an SMR of 2 or higher. Distal fractures in the upper and lower extremities were associated with only minor increases in mortality rates. The second most common fracture, after hip fracture, in our study was wrist fracture. A similar or slightly lower mortality rate of 2.5% at 1 year was seen for this location compared with the previously reported 3–3.6% (Endres et al. 2006, Oyen et al. 2014) and associated with a low SMR (1.1). The mortality rate at 1 year after non-traumatic vertebral fractures was approximately 10%, which is lower than the previously reported 12–46% (Lau et al. 2008, Harris et al. 2010, Waterloo et al. 2012). Vertebral fractures in the elderly resulting from low-energy injuries are not commonly diagnosed at accident and emergency departments but are more likely to be diagnosed and treated in primary health care. This might have influenced the mortality seen in our study, which is based on fracture register data obtained from emergency departments. The 30-day mortality rate is more likely to be influenced by factors directly linked to the fractures sustained than the 1-year mortality which may, to a larger extent, reflect the influence of comorbidities. We observed that the mortality rate from a fracture, compared with what would be expected, was higher at 30 days than at 1 year. It was beyond our scope to analyze the influence of comorbidities and the death cause. The mortality rate was, as expected, observed to be related to age, in accordance with previous reports of specific fracture locations such as the humerus (Ravindrarajah et al. 2018, Bergdahl et al. 2020) and the femur (Ravindrarajah et al. 2018, Wolf et al. 2021). The 30-day and 1-year mortality in men over the age of 60 years was higher than in women. This is in agreement with previous studies where men with comorbidities or with low bone mineral density have been reported to have an increased mortality risk when sustaining frailty fractures (Bliuc et al. 2009, Cook et al. 2017). For a frail person, the fracture is an event, often in a multifactorial chain that increases the risk of death (Johnell et al. 2004, Huette et al. 2020). The independent role of the fracture versus other factors such as comorbidities has been reported to be uncertain in previous studies (Vestergaard et al. 2007b, Teng et al. 2008). High mortality in osteoporotic fractures has, though, been relatively well described (Bliuc et al. 2009, Alarkawi et al. 2020) and there is an ongoing discussion on possible mortality reduction with

744

osteoporosis treatment (Bliuc et al. 2019). Further, in a paper from Borgen et al. (2019) patients with fractures of the axial skeletal and proximal extremities had a lower bone mineral density. This is the same pattern observed and visualized in our Figure 1. Multiple fractures on the same occasion have been suggested to signal an increased risk of mortality (Sujic et al. 2020). In our study the mortality rate as well as SMR at 30 days was slightly higher for multiple fractures than for single location fractures and no differences were seen at 1 year. The strength of our study is the use of structured data, collected within a national register and directly linked to the Swedish Tax Agency population register, providing the possibility to compare different fracture locations within a population. The major limitations of our study are that data on comorbidities and cause of death was not available. It should be noted that direct comparison of SMRs between fracture locations (or between populations) is difficult due to potential differences in the distribution of standardizing variables. Future studies, preferably with linkage to other national health databases, may provide an opportunity to directly compare the mortality risk between fracture locations while controlling for standardizing variables and other potential confounders. Another limitation is that only lifetables for mortality per year were available, which is why seasonal variations of mortality and fracture incidence may have caused an over- or underestimation of the 30-day SMR figures. Another source of potential bias in the SMR calculations is the low completeness of registrations from some hospitals, which may influence the registrations of some fracture types more than others In conclusion, patients who sustain a fracture had a marked increase in mortality rate at both 30 days and 1 year, about 7-fold and over 2-fold, respectively, when compared with the expected mortality rate of the Swedish population assuming the same age and sex distribution as in the study population. High SMR was seen for proximal fractures of both upper and lower extremities, but also for vertebrae, pelvic, and acetabular fractures. It is, however, important to emphasize that the size of SMR following specific fractures may be due to many factors: the fracture location, the treatment, the distribution of standardizing variables in the patient population, and other confounding factors.

Data collection and analyses CB, JE, HB; interpretation, writing, and editing CB, MM, JE, HB. Acta thanks Charles Court-Brown and Frede Frihagen for help with peer review of this study.

Alarkawi D, Bliuc D, Tran T, Ahmed L A, Emaus N, Bjørnerem A, Jørgensen L, Christoffersen T, Eisman J A, Center J R. Impact of osteoporotic fracture type and subsequent fracture on mortality: the Tromso Study. Osteoporos Int 2020; 31(1): 119-30.

Acta Orthopaedica 2021; 92 (6): 739–745

Bercik M J, Tjoumakaris F P, Pepe M, Tucker B, Axelrad A, Ong A, Austin L. Humerus fractures at a regional trauma center: an epidemiologic study. Orthopedics 2013; 36(7): e891-897. Bergdahl C, Wennergren D, Ekelund J, Möller M. Mortality after a proximal humeral fracture. Bone Joint J 2020; 102-B(11): 1484-90. Bliuc D, Nguyen N D, Milch V E, Nguyen V, Eisman J A, Center J R. Mortality risk associated with low-trauma osteoporotic fracture and subsequent fracture in men and women. JAMA 2009; 301(5): 513-21. Bliuc D, Tran T, van Geel T, Adachi J D, Berger C, van den Bergh J, Eisman J A, Geusens P, Goltzman D, Hanley D A, Josse R G, Kaiser S, Kovacs C S, Langsetmo L, Prior J C, Nguyen T V, Center J R; CaMOS Research Group. Mortality risk reduction differs according to bisphosphonate class: a 15-year observational study. Osteoporos Int 2019; 30(4): 817-28. Borgen T T, Bjørnerem Å, Solberg L B, Andreasen C, Brunborg C, Stenbro M B, Hübschle L M, Froholdt A, Figved W, Apalset E M, Gjertsen J E, Basso T, Lund I, Hansen A K, Stutzer J M, Omsland T K, Nordsletten L, Frihagen F, Eriksen E F. Post-fracture risk assessment: target the centrally sited fractures first! S substudy of NoFRACT. J Bone Miner Res 2019; 34(11): 2036-44. Cook M J, Oldroyd A, Pye S R, Ward K A, Gielen E, Ravindrarajah R, Adams J E , Lee D M, Bartfai G, Boonen S, Casanueva F, Forti G, Giwercman A, Han T S, Huhtaniemi I T, Kula K, Lean M E, Pendleton N, Punab M, Vanderschueren D, Wu F C, O’Neill T W; EMAS Study Group. Frailty and bone health in European men. Age Ageing 2017; 46(4): 635-41. Endres H G, Dasch B, Lungenhausen M, Maier C, Smektala R, Trampisch H J, Pientka L. Patients with femoral or distal forearm fracture in Germany: a prospective observational study on health care situation and outcome. BMC Public Health 2006; 6: 87. Gundel O, Thygesen L C, Gögenur I, Ekeloef S. Postoperative mortality after a hip fracture over a 15-year period in Denmark: a national register study. Acta Orthop 2020; 91(1): 58-62. Harris M B, Reichmann W M, Bono C M, Bouchard K, Corbett K L, Warholic N, Simon J B, Schoenfeld A J, Maciolek L, Corsello P, Losina E, Katz J N. Mortality in elderly patients after cervical spine fractures. J Bone Joint Surg Am 2010; 92(3): 567-74. Huette P, Abou-Arab O, Djebara A E, Terrasi B, Beyls C, Guinot P G, Havet E, Dupont H, Lorne E, Ntouba A, Mahjoub Y. Risk factors and mortality of patients undergoing hip fracture surgery: a one-year follow-up study. Sci Rep 2020; 10(1): 9607. Johnell O, Kanis J A, Oden A, Sernbo I, Redlund-Johnell I, Petterson C, De Laet C, Jonsson B. Mortality after osteoporotic fractures. Osteoporos Int 2004; 15(1): 38-42. Juto H, Möller M, Wennergren D, Edin K, Apelqvist I, Morberg P. Substantial accuracy of fracture classification in the Swedish Fracture Register: evaluation of AO/OTA-classification in 152 ankle fractures. Injury 2016; 47(11): 2579-83. Kado D M, Duong T, Stone K L, Ensrud K E, Nevitt M C, Greendale G A, Cummings S R. Incident vertebral fractures and mortality in older women: a prospective study. Osteoporos Int 2003; 14(7): 589-94. Klop C, van Staa T P, Cooper C, Harvey N C, de Vries F. The epidemiology of mortality after fracture in England: variation by age, sex, time, geographic location, ethnicity. Osteoporos Int 2017; 28(1): 161-8. Knutsson S B, Wennergren D, Bojan A, Ekelund J, Möller M. Femoral fracture classification in the Swedish Fracture Register: a validity study. BMC Musculoskelet Disord 2019; 20(1): 197. Larsen P, Ceccotti A A, Elsoe R. High mortality following distal femur fractures: a cohort study including three hundred and two distal femur fractures. Int Orthop 2020; 44(1): 173-7. Lau E, Ong K, Kurtz S, Schmier J, Edidin A. Mortality following the diagnosis of a vertebral compression fracture in the Medicare population. J Bone Joint Surg Am 2008; 90(7): 1479-86. Morgonsköld D, Warkander V, Savvides P, Wihlborg A, Bouzereau M, Möller H, Gerdhem P. Inter- and intra-rater reliability of vertebral fracture classifications in the Swedish fracture register. World J Orthop 2019; 10(1): 14-22.

Acta Orthopaedica 2021; 92 (6): 739–745

Müller M E, Koch P, Nazarian S, Schatzker J. The comprehensive classification of fractures of long bones/AO classification of fractures. SpringerVerlag Berlin Heidelberg 1990. National Board of Health and Welfare www.socialstyrelsen.se/globalassets/sharepoint-dokument/artikelkatalog/statistik/2020-11-7048.pd; www. socialstyrelsen.se/globalassets/sharepoint-dokument/artikelkatalog/statistik/2020-12-7062.pdf; 2020. Oyen J, Diamantopoulos A P, Haugeberg G. Mortality after distal radius fracture in men and women aged 50 years and older in southern Norway. PLoS One 2014; 9(11): e112098. Pincus D, Ravi B, Wasserstein D, Huang A, Paterson J M, Nathens A B, Kreder H J, Jenkinson R J, Wodchis W P. Association between wait time and 30-day mortality in adults undergoing hip fracture surgery. JAMA 2017; 318(20): 1994-2003. Ravindrarajah R, Hazra N C, Charlton J, Jackson S H D, Dregan A, Gulliford M C. Incidence and mortality of fractures by frailty level over 80 years of age: cohort study using UK electronic health records. BMJ Open 2018; 8(1): e018836. Schnell S, Friedman S M, Mendelson D A, Bingham K W, Kates S L. The 1-year mortality of patients treated in a hip fracture program for elders. Geriatr Orthop Surg Rehabil 2010; 1(1): 6-14. Somersalo A, Paloneva J, Kautiainen H, Lonroos E, Kiviranta I. Increased mortality after lower extremity fractures in patients < 65 years of age. Acta Orthop 2016; 87(6): 622-5. Sujic R, Luo J, Beaton D E, Bogoch E R, Sale J E M, Jaglal S, Jain R, Mamdani M. Multiple simultaneous fractures are associated with higher all-cause mortality: results from a province-wide fracture liaison service. Osteoporosis Int 2020; 31(2): 291-6. Teng G G, Curtis J R, Saag K G. Mortality and osteoporotic fractures: is the link causal, is it modifiable? Clin Exp Rheumatol 2008; 26(5 Suppl. 51): S125-37.

745

Vandenbroucke J P. A shortcut method for calculating the 95 percent confidence interval of the standardized mortality ratio. Am J Epidemiology 1982; 115: 303-34. von Friesendorff M, McGuigan F E, Wizert A, Rogmark C, Holmberg A H, Woolf A D, Akesson K. Hip fracture, mortality risk, cause of death over two decades. Osteoporos Int 2016; 27(10): 2945-53. Waterloo S, Ahmed L A, Center J R, Eisman J A, Morseth B, Nguyen N D, Nguyen T, Sogaard AJ, Emaus N. Prevalence of vertebral fractures in women and men in the population-based Tromso Study. BMC Musculoskelet Disord 2012; 13: 3. Wennergren D, Ekholm C, Sandelin A, Möller M. The Swedish fracture register: 103,000 fractures registered. BMC Musculoskelet Disord 2015; 16: 338. Wennergren D, Ekholm C, Sundfeldt M, Karlsson J, Bhandari M, Möller M. High reliability in classification of tibia fractures in the Swedish Fracture Register. Injury 2016; 47(2): 478-82. Wennergren D, Stjernstrom S, Möller M, Sundfeldt M, Ekholm C. Validity of humerus fracture classification in the Swedish fracture register. BMC Musculoskelet Disord 2017; 18(1): 251. Vestergaard P, Rejnmark L, Mosekilde L. Has mortality after a hip fracture increased? J Am Geriatr Soc 2007a; 55(11): 1720-6. Vestergaard P, Rejnmark L, Mosekilde L. Increased mortality in patients with a hip fracture-effect of pre-morbid conditions and post-fracture complications. Osteoporos Int 2007b; 18(12): 1583-93. Wilson L A, Gooding B W, Manning P A, Wallace W A, Geoghegan J M. Risk factors and predictors of mortality for proximal humeral fractures. Shoulder Elbow 2014; 6(2): 95-9. Wolf O, Mukka S, Ekelund J, Möller M, Hailer N P. How deadly is a fracture distal to the hip in the elderly? An observational cohort study of 11,799 femoral fractures in the Swedish Fracture Register. Acta Orthop 2021; 92(1): 40-6.

746

Acta Orthopaedica 2021; 92 (6): 746–752

Effects of tourniquet inflation on peri- and postoperative cefuroxime concentrations in bone and tissue Pelle HANBERG 1–3, Mats BUE 2–4, Jesper KABEL 1, Andrea René JØRGENSEN 2, Christian JESSEN 3,5, Kjeld SØBALLE 2–4, and Maiken STILLING 2–4 1 Department

of Orthopaedic Surgery, Horsens Regional Hospital, Horsens; 2 Aarhus Microdialysis Research Group, Orthopaedic Research Unit, Aarhus University Hospital, Aarhus N; 3 Department of Clinical Medicine, Aarhus University, Aarhus N; 4 Department of Orthopaedic Surgery, Aarhus University Hospital, Aarhus N; 5 Department of Anesthesiology, Horsens Regional Hospital, Horsens, Denmark Correspondence: pellehanberg@clin.au.dk Submitted 2021-01-25. Accepted 2021-05-24.

Background and purpose — Tourniquet is widely used in orthopedic surgery to reduce intraoperative bleeding and improve visualization. We evaluated the effect of tourniquet application on peri- and postoperative cefuroxime concentrations in subcutaneous tissue, skeletal muscle, calcaneal cancellous bone, and plasma. The primary endpoint was the time for which the free cefuroxime concentration was maintained above the clinical breakpoint minimal inhibitory concentration (T > MIC) for Staphylococcus aureus (4 µg/mL). Patients and methods — 10 patients scheduled for hallux valgus or hallux rigidus surgery were included. Microdialysis catheters were placed for sampling of cefuroxime concentrations bilaterally in subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone. A tourniquet was applied on the thigh of the leg scheduled for surgery (tourniquet duration time [range]: 65 minutes [58–77]). Cefuroxime (1.5 g) was administered intravenously 15 minutes prior to tourniquet inflation, followed by a second dose 6 hours later. Dialysates and venous blood samples were collected for 12 hours. Results — A cefuroxime concentration of 4 µg/mL was reached within 23 minutes in all compartments and patients. For cefuroxime the T > MIC (4 µg/mL) ranged between 4.8 and 5.4 hours across compartments, with similar results for the tourniquet and non-tourniquet leg. Comparable T > MIC and penetration ratios were found for the first and second dosing intervals. Interpretation — Administration of cefuroxime (1.5 g) 15 minutes prior to tourniquet inflation is safe in order to achieve tissue concentrations above 4 µg/mL throughout surgery. A tourniquet application time of approximately 1 hour did not affect the cefuroxime tissue penetration in the following dosing interval.

Tourniquet (TQ) is widely used in orthopedic surgery due to its ability to reduce intraoperative bleeding and improve visualization (Rama et al. 2007). However, as the blood supply to the operating field is occluded during surgery, correct timing of antimicrobial prophylaxis administration and TQ inflation is essential in order to ensure therapeutic tissue concentrations at the site of surgery. Only a few studies have investigated the ideal time interval from perioperative antimicrobial prophylaxis administration to TQ inflation, resulting in ambiguous guidelines (Johnson 1987, Deacon et al. 1996, Prokuski 2008, Ochsner et al. 2016). With regard to cefuroxime in particular, a recent randomized controlled microdialysis study in a pig model suggested that a window of 15–45 minutes between cefuroxime administration and TQ inflation results in sufficient perioperative tissue concentrations throughout a 90-minute TQ application (Hanberg et al. 2020b). TQ induces peri- and postoperative ischemia (Ejaz et al. 2015), which may result in decreased postoperative tissue perfusion and antimicrobial tissue concentration (Smith and Hing 2010). A recent study on a rat model demonstrated a reduced distribution of antimicrobials to TQ-affected tissues for up to 72 hours after TQ release (Mangum et al. 2019). Decreased postoperative antimicrobial tissue concentrations may ultimately increase the risk of surgical site infection. Therefore, we dynamically evaluated effects of TQ application on both peri- and postoperative in situ cefuroxime concentrations in subcutaneous tissue, skeletal muscle, calcaneal cancellous bone, and plasma. Cefuroxime (1.5 g) was administered intravenously prior to TQ inflation and followed by a subsequent dose 6 hours later. The primary aim was to assess the time for which the free drug concentration of cefuroxime was maintained above the clinical breakpoint minimal inhibitory concentration (T > MIC) for Staphylococcus aureus (4

Acta Orthopaedica 2021; 92 (6): 746–752

747

Table 1. Inclusion and exclusion criteria 7

2/5 1/4

Inclusion criteria Written informed consent Age ≥ 18 years Normal distal blood pressure bilaterally Normal creatinine levels Use of contraception for women of childbearing age Exclusion criteria Diabetes Unsuccessful spinal anesthesia Allergy Previous arterial surgery in either of the legs Previous surgery on either of the calcaneal bones Previous fracture or bone infection in either of the calcaneal bones

3/6

Figure 1. Illustration of the inserted microdialysis catheters. Cefuroxime concentrations were obtained by means of microdialysis catheters placed in non-tourniquet subcutaneous tissue (1), non-tourniquet skeletal muscle (2), non-tourniquet calcaneal cancellous bone (3), tourniquet subcutaneous tissue (4), tourniquet skeletal muscle (5), and tourniquet calcaneal cancellous bone (6). A tourniquet cuff (7) was placed on the leg scheduled for surgery.

µg/mL) (EUCAST 2021), which we hypothesized was maintained throughout surgery in the TQ-exposed tissues when administering cefuroxime 15 minutes prior to tourniquet inflation.

Patients and methods This study was conducted at the Department of Orthopedic Surgery, Horsens Regional Hospital, Denmark. Chemical analyses were performed at the Department of Clinical Biochemistry, Aarhus University Hospital, Denmark. This study was performed in the same setting as another study, which investigated tissue ischemic metabolites (Hanberg et al. 2021). Study procedure Microdialysis The microdialysis catheter consists of a semipermeable membrane at the tip of the catheter, which allows for sampling of water-soluble molecules such as antimicrobials (Hanberg et al. 2016, Kho et al. 2017, Bue et al. 2018, Hanberg et al. 2019b). However, as the semipermeable membrane is continuously perfused, equilibrium across the semipermeable membrane cannot be attained. Consequently, the dialysates represent only a fraction of the actual tissue concentration. This fraction is referred to as the relative recovery that can be determined by different calibration methods (Kho et al. 2017). For this study, meropenem was used as an internal calibrator for cefuroxime (Hanberg et al. 2019a). An in-depth description of the micro-

dialysis technique and the equation for calculating the relative recovery can be found elsewhere (Kho et al. 2017). We used microdialysis equipment from M Dialysis AB (Stockholm, Sweden). The microdialysis catheters consisted of CMA 63 membranes and CMA 107 precision pumps (flow rate: 2 µL/min). Study design and patients 10 patients were included in this prospective observational cohort study. The effects of TQ application on both peri- and postoperative cefuroxime concentrations were evaluated in subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone in a simultaneous paired comparison of the TQ and non-TQ leg during 12 hours of continuous microdialysis sampling (Figure 1). Patients scheduled for hallux valgus or hallux rigidus surgery were offered enrolment in the study. A single surgeon recruited 10 patients who attended the outpatient clinic. Inclusion and exclusion criteria are presented in Table 1. All patients invited for enrolment were included in the study and all completed the study. After placement of the 6 microdialysis catheters, cefuroxime (1.5 g) (FreseniusKabi AB, Sweden) was administered intravenously over 10 minutes, marking time 0. 15 minutes after initiation of the cefuroxime administration, the TQ cuff was inflated (pressure 260 mmHg) on the thigh of the leg scheduled for surgery. Prior to TQ inflation, the leg was elevated for 1 minute. The planned surgical procedure was performed after TQ inflation. When the surgical procedure was completed, the TQ cuff was released (mean TQ inflation time [range]: 65 minutes [58–77]). A second dose of 1.5 g cefuroxime was administered at 6 hours. Surgery Before the surgical procedure, microdialysis catheters were placed similarly in both legs: in the subcutaneous tissue (membrane length 30 mm), at the posterior site of the mid-lower leg, in the gastrocnemius muscle of the medial head (membrane length 30 mm), and in the calcaneal cancellous bone (mem-

748

brane length 10 mm) via drill holes (ø: 2 mm; depth 30 mm) made on the posterolateral side, aiming at the anteromedial side of the calcaneal bone (Figure 1). After placement of the microdialysis catheters, all catheters were perfused with 0.9% NaCl containing 5 µg/mL meropenem, allowing for continuous calibration with meropenem as an internal calibrator. Sampling procedures Dialysates were collected from all 6 microdialysis catheters at 15-minute intervals from time 0–30 minutes, at 30-minute intervals from time 30–180 minutes, and at 60-minute intervals from both time 180–240 minutes and time 300–360 minutes. Following administration of the second dose of 1.5 g cefuroxime at time 360 minutes, dialysates were collected at 30-minute intervals from time 360–540 minutes, and at 60-minute intervals from both time 540–600 minutes and time 660–720 minutes. 17 samples from each microdialysis catheter were collected over the 12-hour period. Venous blood samples were collected at the midpoint of the sampling intervals drawn from a peripheral catheter in the cubital vein. After the last sample was collected, all microdialysis catheters were removed. Handling of samples The venous blood samples were stored at 5°C for a maximum of 10 hours before being centrifuged at 3,000 g for 10 minutes. The plasma aliquots were then stored at -80°C until analysis. The dialysate samples were immediately stored at -80°C until analysis. Quantification of cefuroxime and meropenem concentrations The concentrations of cefuroxime and meropenem were quantified using a validated ultra-high-performance liquid chromatography assay (Hanberg et al. 2018). Inter-run imprecisions (% coefficients of variation) were 4.7% at 2.5 µg/mL for quantification of cefuroxime and 3.0% at 2.0 µg/mL for quantification of meropenem. The lower limits of quantification were 0.06 µg/mL for cefuroxime and 0.5 µg/mL for meropenem. Pharmacokinetic analysis and statistics The cefuroxime concentrations of the dialysate were attributed to the midpoint of the sampling intervals. Pharmacokinetic parameters, areas under the concentration-time curves (AUC), peak drug concentration (Cmax), time to Cmax (Tmax), and T1/2, were determined separately for each compartment for each patient by non-compartmental analysis using the pharmacokinetic series of commands in Stata (v. 15.1, StataCorp, College Station, TX, USA). AUC were calculated using the linear up-log down trapezoidal method. The maximum of all the recorded concentrations was defined as Cmax, enabling calculation of Tmax. The T1/2 was calculated as ln(2)/λeq, where λeq is the terminal elimination rate constant estimated by linear regression of the log concentration on time. The AUCtissue/ AUCplasma ratio was calculated as a measure of tissue penetration. Microsoft Excel (v. 16.16.11, Microsoft Corp, Redmond,

Acta Orthopaedica 2021; 92 (6): 746–752

WA, USA) was used to estimate the T > MIC (4 µg/mL) using linear interpolation. The pharmacokinetic parameters and T > MIC were calculated separately for both the first (time 0–6 hours) and second (time 6–12 hours) dosing intervals. The pharmacokinetic parameters and T > MIC were obtained in all 7 compartments from the same patient and a mixed model for repeated measurements had compartments as fixed effect and subject identification variable as a random effect was applied. Also, distinct residual variance was assumed within each compartment. The normality of the residuals was estimated using a quantile–quantile plot for the residuals and the homogeneity of the residual variance was checked by plotting residuals vs. best linear unbiased prediction estimates. The normality of the estimated random effects was checked using a quantile–quantile plot of the estimated random effects. The Kenward–Roger approximation method was used for degrees of freedom correction due to the small sample size (Kenward and Roger 1997). The F-test was used to determine the overall comparisons between the compartments and a t-test was used to determine pairwise comparisons. A significance level of 5% was used. Statistical analyses were performed using Stata. Sample size Sample size calculation indicated 8 patients, based on cefuroxime target tissue concentrations above 4 μg/mL throughout a predicted surgery time of 105 minutes (pre TQ time [15 minutes] + expected TQ time [90 minutes]). With a significance level of 5% and a power of 90%, a sample size calculation comparing 1 mean to the reference value of 105 minutes was performed for T > MIC (4 μg/mL) for plasma (mean [SD] 145 [28] minutes), TQ subcutaneous tissue (mean [SD] 198 [37] minutes), and TQ calcaneal cancellous bone (mean [SD] 208 [43] minutes) (Hanberg et al. 2020b). To accommodate dropout of patients/microdialysis probes, 10 patients were included. Ethics, registration, funding, and potential conflicts of interest The study was approved by the Danish Medicines Agency (EudraCT number 2018-000217-21), the Central Denmark Region Committees on Health Research Ethics (Registration number 1-10-72-47-18), and the Danish Data Protection Agency (Registration number 1-16-02-88-18). The study was registered at www.clinicaltrialsregister.eu (number 2018000217-21) and conducted in accordance with the Declaration of Helsinki and the ICH Harmonized Tripartite Guideline for Good Clinical Practice. The Good Clinical Practice Unit at Aalborg and Aarhus University Hospitals conducted the mandatory monitoring procedures. This work was supported by grants from the Health Research Foundation of Central Denmark Region, the Elisabeth og Karl Ejnar Nis-Hansens Mindelegat Foundation, the Læge Sofus Carl Emil Friis og Hustru Olga Doris Friis’ legat Foundation, the Augustinus Foundation, the A. P. Møller

Acta Orthopaedica 2021; 92 (6): 746–752

749

Table 2. Patient characteristics. Values are mean (range) unless otherwise specified Sex (female/male), n Age, years Height, cm Weight, kg BMI Plasma creatinine, µmol/L Tourniquet duration, min Ankle–brachial index tourniquet leg Ankle–brachial index non-tourniquet leg

7/3 58 (45–67) 169 (156–185) 72 (56–89) 25 (20–33) 75 (60–90) 65 (58–77) 1.11 (0.90–1.28) 1.08 (0.91–1.28)

Table 3. Mean time (95% confidence interval) with concentrations above the minimal inhibitory concentration (T > MIC) (4 µg/mL) in minutes for plasma, subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone on both the tourniquet and non-tourniquet leg from the first dosing interval

Non-tourniquet leg Tourniquet leg p-value

Plasma Subcutaneous tissue Skeletal muscle Calcaneal canc. bone

318 (297–338) 312 (292–333) 320 (300–341) 306 (285–326)

– – 322 (302–343) 0.4 316 (295–336) 0.7 289 (269–310) a 0.2

a p < 0.05 for comparison with all compartments on the tourniquet side and with plasma.

Normal range: Plasma creatinine (males), 60–106 µmol/L; plasma creatinine (females), 45–90 µmol/L; ankle–brachial index, ≥ 0.9.

Cefuroxime concentration (µg/mL) Tourniquet inflation

Tourniquet release

Plasma Non-tq subcutaneous tissue Tq subcutaneous tissue Non-tq skeletal muscle Tq skeletal muscle Non-tq calcaneal cancellous bone Tq calcaneal cancellous bone MIC 4 µg/mL

100

1 0 1st dose of cefuroxime

100

200

300

400

500

2nd dose of cefuroxime

Foundation, and the Familien Hede Nielsen Foundation. The funding sources did not have any roles in the investigation, data interpretation, or paper presentation. The authors have no conflicts of interest.

Results The patients’ characteristics are presented in Table 2. No adverse events related to the microdialysis technique or cefuroxime infusion occurred. The mean relative recovery (SD) values were 23% (9) for TQ subcutaneous tissue, 20% (7) for non-TQ subcutaneous tissue, 39% (4) for TQ skeletal muscle, 33% (12) for non-TQ skeletal muscle, 21% (8) for TQ calcaneal cancellous bone, and 19% (7) for non-TQ calcaneal cancellous bone. T > MIC Comparable results were observed for T > MIC (4 µg/mL) between the first and second dosing intervals. Therefore, the T > MIC results are presented only for the first dosing interval in

600

700

Time (min)

Figure 2. Mean concentration-time profiles of cefuroxime for plasma, subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone on both the tourniquet and non-tourniquet leg. Bars represent 95% CI. The y-axis is in log scale. The first and second dose of 1.5 g cefuroxime was administered at time 0 and 6 h, respectively. Tourniquet inflation and mean release times were 15 and 80 minutes, respectively. Abbreviations: Tq = tourniquet; MIC = minimal inhibitory concentration.

Table 3. A cefuroxime concentration of 4 µg/mL was reached within 23 minutes in all compartments and patients. The T > MIC (4 µg/mL) ranged between 4.8 and 5.4 hours across compartments, and no significant differences were found between the TQ and non-TQ exposed leg (Figure 2 and Table 3). When comparing TQ and non-TQ legs separately, lower T > MIC values were found for calcaneal cancellous bone compared with the remaining compartments in the TQ leg, including plasma (p < 0.05). No differences were found between the compartments in the non-TQ leg. Pharmacokinetic parameters Comparable pharmacokinetic results were seen between the first and second dosing intervals in all investigated compartments. Only the TQ calcaneal cancellous bone Tmax was longer in the first dosing interval (mean [range], 84 minutes [23–135]) compared with the second dosing interval (mean [range], 51 minutes [15–75]) (p < 0.01). The pharmacokinetic parameters are presented only for the first dosing interval in Table 4. The concentration time profiles are depicted for both the first and second dosing interval in Figure 2.

750

Acta Orthopaedica 2021; 92 (6): 746–752

Table 4. Pharmacokinetic parameters for plasma, subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone on both the tourniquet and non-tourniquet leg Compartment

Non-tourniquet

Tourniquet p-value

AUC0–6h, mean (95% CI), 104 min μg/mL Plasma 8.2 (6.6–9.8) – – Subcutaneous tissue 8.5 (7.0–10) 7.6 (6.0–9.1) 0.3 Skeletal muscle 7.3 (5.7–8.9) 7.8 (6.2–9.4) 0.6 a Calcaneal canc. bone 6.6 (5.1–8.2) 7.1 (5.6–8.7) 0.6 Cmax, mean (95% CI), μg/mL Plasma 97 (84–110) b – – Subcutaneous tissue 58 (45–70) 51 (38–64) 0.4 c Skeletal muscle 61 (48–73) 70 (60–83) 0.3 Calcaneal canc. bone 59 (47–72) 53 (40–66) 0.4 Tmax, mean (range), min Plasma 7.5 (7.5–7.5) b – – Subcutaneous tissue 45 (23–75) 49 (23–105) 0.7 Skeletal muscle 27 (23–45) 33 (23–105) 0.5 Calcaneal canc. bone 35 (23–75) 84 (23–135) d < 0.01 T1/2, mean (95% CI), min Plasma 74 (56–93) – – Subcutaneous tissue 94 (75–113) 99 (81–118) 0.7 Skeletal muscle 97 (78–116) 87 (68–105) 0.4 Calcaneal canc. bone 86 (67–105) 95 (77–114) 0.5 AUCtissue/AUCplasma Subcutaneous tissue 1.1 (0.86–1.3) 0.96 (0.73–1.2) 0.3 Skeletal muscle 0.92 (0.69–1.2) 0.98 (0.75–1.2) 0.6 Calcaneal canc. bone 0.84 (0.61–1.1) 0.88 (0.65–1.1) 0.8 AUC, area under the concentration-time curve from 0 to 6 hours; Cmax, peak drug concentration; Tmax, time to Cmax; T1/2, half-life; AUCtissue/AUCplasma, area under the concentration-time curve ratio of tissue/plasma. a p = 0.04 for comparison with non-tourniquet subcutaneous tissue. b p < 0.05 for comparison with all tissues. c p < 0.05 for comparison with tourniquet subcutaneous tissue and calcaneal cancellous bone. d p < 0.01 for comparison with tourniquet subcutaneous tissue and skeletal muscle.

No statically significant differences were observed for AUC, Cmax, T1/2, and tissue penetrations when comparing the TQ and non-TQ leg (Table 3). Only the calcaneal cancellous bone Tmax was longer in the TQ leg (mean [range], 84 minutes [23–135]) compared with the non-TQ leg (mean [range], 35 minutes [22–75]) (p < 0.01) in the first dosing interval. No statistically significant differences were found for the remaining compartments. When comparing the TQ and non-TQ leg separately, a lower AUC was found for the non-TQ calcaneal cancellous bone compared with non-TQ subcutaneous tissue (Table 3). Plasma Cmax was higher compared with all investigated compartments. Moreover, the TQ skeletal muscle Cmax was higher compared with both TQ calcaneal cancellous bone and TQ subcutaneous tissue. Finally, plasma Tmax was shorter compared with all tissues in both the TQ and non-TQ leg, and the TQ calcaneal cancellous bone was longer than both TQ subcutaneous tissue and TQ skeletal muscle.

Discussion This is the first clinical study to investigate the effects of TQ application on both peri- and postoperative cefuroxime concentrations in subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone in a simultaneous paired comparison of the TQ and non-TQ leg. Our main finding was a cefuroxime T > MIC (4 µg/mL) range between 4.8 and 5.4 hours across compartments. Furthermore, comparable T > MIC and penetration ratios were found for the first and second dosing intervals. TQ is widely used in orthopedic surgery, but only a few studies have investigated antimicrobial tissue concentrations during the TQ application, and no clinical studies have investigated antimicrobial tissue concentrations after TQ release. Using bone and fat tissue specimens, Johnson (1987) investigated different time intervals from administration of cefuroxime (1.5 g) to TQ inflation, and concluded that a time interval of 10 minutes was sufficient to achieve tissue concentrations above 4 µg/mL. Recently, a randomized controlled microdialysis study in a pig model suggested that a window of 15–45 minutes between cefuroxime (1.5 g) administration and TQ inflation was sufficient to achieve calcaneal cancellous bone and subcutaneous tissue concentrations above 4 µg/mL (Hanberg et al. 2020b). The present clinical study confirms these findings, suggesting that cefuroxime has fully penetrated the investigated tissues after 15 minutes. It has previously been hypothesized that perioperative ischemia reduces the postoperative antimicrobial tissue penetration (Smith and Hing 2010, Mangum et al. 2019). However, studies investigating tissue ischemia during and after TQ application found that ischemia-exposed tissue fully recovers 2.5 hours after TQ release (Ejaz et al. 2015, Hanberg et al. 2020b). Our findings do not indicate any decreased postoperative cefuroxime penetration in the TQ exposed tissues for a TQ application of approximately 1 hour. Interestingly, our study showed that TQ calcaneal cancellous bone Tmax is longer than in non-TQ calcaneal cancellous bone. Furthermore, a wider range of the Tmax values was found for both subcutaneous tissue and skeletal muscle in the TQ leg compared with the non-TQ leg. These Tmax results may be attributed to a combination of the limited elimination rate of cefuroxime during TQ time and a second peak in the cefuroxime concentration after TQ release. For 5 patients, this peak was higher than the initial peak prior to TQ inflation in TQ calcaneal cancellous bone. This may indicate a favorable hyperemic effect when the TQ is released, which was also observed in a pig model (Hanberg et al. 2020b). For antimicrobial prophylaxis it is generally recommended that the antimicrobial plasma and tissue concentrations exceed the MIC values of relevant bacteria throughout surgery (Mangram et al. 1999). In our study a TQ cuff was inflated 15 minutes after initiation of the cefuroxime admin-

Acta Orthopaedica 2021; 92 (6): 746–752

istration and a cefuroxime concentration of 4 µg/mL was reached within 23 minutes in all tissues and patients, which was maintained above this target for a minimum of 4.5 hours in all the investigated compartments. As such, these findings indicate that cefuroxime appears to be a good choice for antimicrobial prophylaxis in terms of tissue penetration and T > MIC. Only 1 clinical study has previously investigated cefuroxime bone tissue concentrations by means of microdialysis (Tottrup et al. 2019). Tottrup et al. (2019) found a shorter T > MIC in plasma, subcutaneous tissue, and tibial cancellous bone after a postoperative intravenous bolus administration of 1.5 g cefuroxime compared with our study compartments. While the plasma creatinine was comparable between the patient groups in the 2 studies, Tottrup et al. recorded a substantially higher mean BMI compared with the present study (31 vs. 25). Weight-based dosing of cefuroxime, in addition to consideration of renal function, may therefore be considered in order to achieve therapeutic tissue concentrations in heavy patients. The few clinical studies that have investigated antimicrobial concentrations during TQ application have been based on tissue specimens (Johnson 1987, Deacon et al. 1996). However, this approach suffers from important methodological limitations because sampling in clinical studies is limited to the time of surgery, free extracellular concentrations cannot be measured selectively, and drug concentrations are given by mass rather than volume (Landersdorfer et al. 2009). Microdialysis, on the other hand, allows for simultaneous and serial sampling of the free and active fraction of drugs in the interstitial space from multiple compartments, both peri- and postoperatively (Tottrup et al. 2016, Hanberg et al. 2020a). These features are desirable, as the majority of infections occur in the interstitial space. However, microdialysis remains a sampling technique that has limitations associated with calibration procedures and chemical assays (Landersdorfer et al. 2009, Kho et al. 2017). The major limitation of our study is the small sample size. Although a paired design and no statistically significant differences for the T > MIC (4 µg/mL) between TQ and non-TQ exposed tissues were demonstrated for this specific patient population, a larger study population may alter these findings. However, as all mean tissue cefuroxime concentrations were above 4 µg/mL approximately 4–5 times longer than the presented surgery time, any potential difference between the TQ and non-TQ exposed tissue may be without clinical relevance. In summary, administering cefuroxime (1.5 g) 15 minutes prior to TQ inflation seems safe in order to achieve tissue concentrations above 4 µg/mL throughout 1 hour surgery, as T > MIC (4 µg/mL) ranged between 4.8 and 5.4 hours in subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone. A TQ application time of approximately 1 hour did not affect the cefuroxime tissue penetration in the following dosing interval.

751

PH, MB, JK, KS, and MS initiated and designed the study. PH, JK, and CJ conducted the surgery and placed all the probes. PH, MB and ARJ collected the data. Statistical analysis and interpretation of data was done by PH, MB, JK, KS, and MS. All authors drafted and revised the manuscript. The authors would like to thank the funding organizations, the Department of Orthopaedic Surgery, Horsens Regional Hospital, and the Orthopaedic Research Unit, Aarhus University Hospital for supporting this study. Finally, they express their sincere gratitude to the patients who participated in this study. Acta thanks Joan Elizabeth Bechtold and Pim Langendijk for help with peer review of this study.

Bue M, Hanberg P, Koch J, Jensen L K, Lundorff M, Aalbaek B, Jensen H E, Søballe K, Tøttrup M. Single-dose bone pharmacokinetics of vancomycin in a porcine implant-associated osteomyelitis model. J Orthop Res 2018; 36(4): 1093-8. Deacon J S, Wertheimer S J, Washington J A. Antibiotic prophylaxis and tourniquet application in podiatric surgery. J Foot Ankle Surg 1996; 35(4): 344-9. Ejaz A, Laursen A C, Kappel A, Jakobsen T, Nielsen P T, Rasmussen S. Tourniquet induced ischemia and changes in metabolism during TKA: a randomized study using microdialysis. BMC Musculoskelet Disord 2015; 16: 326. EUCAST. https://eucast.org/ (Accessed January 25, 2021). Hanberg P, Bue M, Birke Sorensen H, Soballe K, Tottrup M. Pharmacokinetics of single-dose cefuroxime in porcine intervertebral disc and vertebral cancellous bone determined by microdialysis. Spine J 2016; 16(3): 432-8. Hanberg P, Obrink-Hansen K, Thorsted A, Bue M, Tottrup M, Friberg L E, Hardlei T F, Søballe K, Gjedsted J. Population pharmacokinetics of meropenem in plasma and subcutis from patients on extracorporeal membrane oxygenation treatment. Antimicrob Agents Chemother 2018; 62(5): e02390-17. doi: 10.1128/AAC.02390-17. Hanberg P, Bue M, Obrink-Hansen K, Kabel J, Thomassen M, Tottrup M, Søballe K, Stilling M. Simultaneous retrodialysis by drug for cefuroxime using meropenem as an internal standard: a microdialysis validation study. J Pharm Sci 2019a; 109(3): 1373-9. doi: 10.1016/j.xphs.2019.11.014. Hanberg P, Lund A, Soballe K, Bue M. Single-dose pharmacokinetics of meropenem in porcine cancellous bone determined by microdialysis: an animal study. Bone Joint Res 2019b; 8 (7): 313-22. Hanberg P, Bue M, Jorgensen A R, Thomassen M, Obrink-Hansen K, Soballe K, Stilling M. Pharmacokinetics of double-dose cefuroxime in porcine intervertebral disc and vertebral cancellous bone: a randomized microdialysis study. Spine J 2020a; 20(8): 1327-32. doi: 10.1016/j. spinee.2020.03.006. Hanberg P, Bue M, Öbrink-Hansen K, Thomassen M, Søballe K, Stilling M. Timing of antimicrobial prophylaxis and tourniquet inflation: a randomized controlled microdialysis study. J Bone Joint Surg Am 2020b; 102(21): 1857-64. doi: 10.2106/JBJS.20.00076. Hanberg P, Bue M, Kabel J, Jørgensen A R, Søballe K, Stilling M. Tourniquet-induced ischemia and reperfusion in subcutaneous tissue, skeletal muscle, and calcaneal cancellous bone. Apmis 2021; 129(4): 225-31. Johnson D P. Antibiotic prophylaxis with cefuroxime in arthroplasty of the knee. J Bone Joint Surg Br 1987; 69(5): 787-9. Kenward M G, Roger J H. Small sample inference for fixed effects from restricted maximum likelihood. Biometrics 1997; 53(3): 983-97. Kho C M, Enche Ab Rahim S K, Ahmad Z A, Abdullah N S. A review on microdialysis calibration methods: the theory and current related efforts. Mol Neurobiol 2017; 54(5): 3506-27.

752

Landersdorfer C B, Bulitta J B, Kinzig M, Holzgrabe U, Sorgel F. Penetration of antibacterials into bone: pharmacokinetic, pharmacodynamic and bioanalytical considerations. Clin Pharmacokinet 2009; 48(2): 89-124. Mangram A J, Horan T C, Pearson M L, Silver L C, Jarvis W R. Guideline for prevention of surgical site infection, 1999. Centers for Disease Control and Prevention (CDC) Hospital Infection Control Practices Advisory Committee. Am J Infect Control 1999; 27(2): 97-132; quiz 3-4; discussion 96. Mangum L C, Garcia G R, Akers K S, Wenke J C. Duration of extremity tourniquet application profoundly impacts soft-tissue antibiotic exposure in a rat model of ischemia-reperfusion injury. Injury 2019; 50(12): 2203-14. Ochsner P E, Borens O, Bodler P M, Broger I, Eich G, Hefti F, Maurer T, Nötzli H, Seiler S, Suvà D, Trampuz A, Uçkay I, Vogt M, Zimmerli W. Infections of the musculoskeletal system: basic principles, prevention, diagnosis and treatment. Grandvaux: Swiss orthopaedics in-house publisher; 2016.

Acta Orthopaedica 2021; 92 (6): 746–752

Prokuski L. Prophylactic antibiotics in orthopaedic surgery. J Am Acad Orthop Surg 2008; 16(5): 283-93. Rama K R, Apsingi S, Poovali S, Jetti A. Timing of tourniquet release in knee arthroplasty: meta-analysis of randomized, controlled trials. J Bone Joint Surg Am 2007; 89(4): 699-705. Smith T O, Hing C B. Is a tourniquet beneficial in total knee replacement surgery? A meta-analysis and systematic review. Knee 2010; 17(2): 141-7. Tottrup M, Bue M, Koch J, Jensen L K, Hanberg P, Aalbaek B, Fuursted K, Jensen H E, Søballe K. Effects of implant-associated osteomyelitis on cefuroxime bone pharmacokinetics: assessment in a porcine model. J Bone Joint Surg Am 2016; 98(5): 363-9. Tottrup M, Soballe K, Bibby B M, Hardlei T F, Hansen P, Fuursted K, Birke-Sørensen H, Bue M. Bone, subcutaneous tissue and plasma pharmacokinetics of cefuroxime in total knee replacement patients: a randomized controlled trial comparing continuous and short-term infusion. Apmis 2019; 127(12): 779-88.

Acta Orthopaedica 2021; 92 (6): 753–759 753

Cat at home? Cat scratch disease with atypical presentations and aggressive radiological findings mimicking sarcoma, a potential diagnostic pitfall Florian AMERSTORFER 1,a, Jasminka IGREC 2,a, Thomas VALENTIN 3, Andreas LEITHNER 1, Lukas LEITNER 1, Mathias GLEHR 1, Jörg FRIESENBICHLER 1, Iva BRCIC 4, and Marko BERGOVEC 1 1 Department of Orthopedics and Trauma, Medical University of Graz, Graz; 2 Division of General Radiology, Department of Radiology, Medical University of Graz, Graz; 3 Section of Infectious Diseases and Tropical Medicine, Department of Internal Medicine, Medical University of Graz, Graz; 4 Diagnostic and Research Institute of Pathology, Medical University of Graz, Neue Graz, Austria a Shared first authorship. Correspondence: iva.brcic@medunigraz.at Submitted 2020-06-12. Accepted 2021-05-26.

Background and purpose — Cat scratch disease (CSD) is a self-limiting disease caused by Bartonella (B.) henselae. It is characterized by granulomatous infection, most frequently involving lymph nodes. However, it can present with atypical symptoms including musculoskeletal manifestations, posing a diagnostic challenge. We describe the prevalence and demographics of CSD cases referred to a sarcoma center, and describe the radiological, histological, and molecular findings. Patients and methods — Our cohort comprised 10 patients, median age 27 years (12–74) with clinical and radiological findings suspicious of sarcoma. Results — 7 cases involved the upper extremities, and 1 case each involved the axilla, groin, and knee. B. henselae was found in 6 cases tested using polymerase chain reaction and serology in 5 cases. 9 cases were soft tissue lesions and 1 lesion involved the bone. 1 patient had concomitant CSD with melanoma metastasis in enlarged axillary lymph nodes. On MRI, 5 soft tissue lesions were categorized as probably inflammatory. In 3 cases, with still detectable lymph node structure and absent or initial liquefaction, the differential diagnosis included lymph node metastasis. A sarcoma diagnosis was suggested in 4 cases. The MRI imaging features of the bone lesion were suspicious of a bone tumor or osteomyelitis. Interpretation — Atypical imaging findings cause a diagnostic challenge and the differential diagnosis includes malignant neoplasms (such as sarcoma or carcinoma metastasis) and other infections. The distinction between these possibilities is crucial for treatment and prognosis.

Bartonella (B.) henselae infection with regional lymphadenopathy may mimic neoplastic processes such as soft tissue or bone tumor, metastasis, or lymphoma, leading to a delayed diagnosis and unnecessary invasive procedures resulting in overtreatment (Huang et al. 1989, Gielen et al. 2003, MazurMelewska et al. 2015). The classical differentiation of CSD from soft tissue neoplasm are enlarged lymph node with preserved hilar architecture and reactive changes of the surrounding fat and fascia, suggesting inflammation (Wang et al. 2009, Mazur-Melewska et al. 2015, Bernard et al. 2016, Chen et al. 2018). In atypical CSD cases, soft tissue mass or a solitary bone lesion may mimic a sarcoma due to the overlapping clinical and radiological findings. Previous studies, analyzing B. henselae infections and their clinical and radiological presentation, are mainly focused on imaging features of lymphadenopathy/lymphadenitis at the epitrochlear region (Gielen et al. 2003, Bernard et al. 2016, Chen et al. 2018). However, the assessment regarding the potential differential diagnosis of sarcoma is scarce, as only single cases of atypical B.henselae infection mimicking sarcoma have been published so far (Frank et al. 1952, Nimityongskul et al. 1992, Fox and Gurtler 1993, Eichhorn-Sens et al. 2008, Colman et al. 2014, Dhal et al. 2020). To our knowledge, this retrospective study represents the first large case series of atypical CSD cases who were transferred to a sarcoma center with the primary diagnosis of sarcoma. The objectives of this study were (i) to describe the prevalence and demographics of atypical CSD cases in a musculoskeletal, orthopedic sarcoma center, (ii) to describe the radiological, histological, and molecular findings, (iii) to highlight the specific MRI criteria for differentiation, and (iv) to discuss differential diagnoses with the aim of raising awareness of this rare disease.

754

Acta Orthopaedica 2021; 92 (6): 753–759

Clinical records of patients scheduled in our sarcoma center assessed for eligibility January 2014 to December 2020 n = 2,467 Excluded Patients not meeting inclusion criteria n = 2,457 Diagnosis of CSD based on serological and/or histopathological and/or molecular diagnostic test. Patients with equivocal radiological finding of soft tissue lesion suspicious of sarcoma n = 10 Missing data for radiological analysis n=2 Patients eligible for radiological analysis n=8

Figure 1. Diagram depicting the patient population and the reasons for inclusion in our cohort.

Patients and methods 2,467 cases of soft tissue tumors (2,077 benign and 390 malignant) presented at the Sarcoma Center, Comprehensive Cancer Center, Medical University of Graz from 2014 until the end of 2020 and were searched for the terms “B. henselae,” “CSD,” and “granulomatous infection” originating in soft tissue, bone and/or lymph node (Figure 1). Inclusion criteria were: (1) diagnosis of CSD by either serology and/or histopathology, as well as molecular diagnostic tests and (2) clinical and/or radiological findings suspicious of soft tissue sarcoma. The analysis included clinical data (age, sex, symptoms, diagnostic pathway, treatment, and follow-up) and MRI, histological, serological, and/or molecular findings. MRI analyses of the lesions included: site, number of lesions, size, margin, signal intensity, and contrast enhancement (CE) pattern, changes of the surrounding fat and fascia, and differential diagnosis, and was performed by the musculoskeletal radiologist (JI) without the knowledge of the diagnosis of CSD. The study’s minimal MRI protocol included T1-weighted and a fluid-sensitive, fat-saturated (FS) sequence parallel to the lesion’s long axis, an axial proton-density, or T2-weighted FS sequence. When applicable, the postcontrast sequence after intravenous application of gadolinium (Gd) was included in the analysis. Available hematoxylin and eosin (HE) slides, cut in 4 µm-thick whole tissue sections from formalin-fixed, paraffinembedded (FFPE) paraffin tissue blocks, were evaluated by a bone and soft tissue pathologist (IB). Molecular analysis (PCR amplification of Bartonella spp. genes) FFPE tissue blocks were cut in 30 x 5 μm-thick whole tissue sections with a microtome using a new sterile blade for each case. According to the manufacturer’s recommendations,

genetic DNA was extracted with the Maxwell 16 FFPE Plus LEV DNA purification kit (Promega, Mannheim, Germany). The concentration and quality of DNA were determined spectrophotometrically with a NanoDrop ND-3300 instrument and the PicoGreen assay (Thermo Fisher Scientific, Waltham, MA, USA). As previously described (Raoult et al. 2006), the Bartonella genus-specific real-time polymerase chain reaction (PCR) amplification was performed on a LightCycler 2.0 instrument using the LightCycler® TaqMan® Master (Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s recommendations with 200–750 ng input DNA. The AMPLIRUN® Bartonella DNA control (Vircell, Granada, Spain) served as a positive control. Ethics, funding, and potential conflicts of interests This retrospective study (EK 30-075 ex 17/18) was approved by the Ethical Review Board. The authors have not received any funding and have no potential conflicts of interests.

Results Clinical findings (Table 1) 10 cases of CSD with radiological findings suspicious of malignancy were identified, resulting in a low prevalence of 0.4% CSD cases in our musculoskeletal sarcoma center. 8 of the patients were primarily evaluated in an external institution and referred to our Center for further work-up. An additional 2 patients presented to our outpatient clinic with MRI findings suspicious of soft tissue tumor and atypical clinical presentation (growing lesions in the soft tissues of the knee and epitrochlear with unclear trauma without inflammation). The patients’ ages ranged from 12 to 74 years (median 27 years); 6 were female. Medical history revealed swelling for several weeks prior to first presentation, median 31 days (17–270). In 7 patients the lesion was located in the upper extremities (elbow [n = 5] and upper arm [n = 2]), followed by the groin/inguinal region (n = 1), axilla (n = 1), and knee (n = 1). Pain was present in 6 cases, and in 1 case signs of infection at the middle part of the left upper arm were noted. In all cases, soft tissue neoplasms were suspected by the external radiologist or clinician. In addition, in case number 5, the lesion additionally involved bone, and chronic osteomyelitis was considered as a differential diagnosis. Only in 5 patients were cat scratches on the lower arm or hand and/or contact with cats documented. Inflammatory serum markers were documented in 8 patients, showing elevated C-reactive protein (CRP) values in only 2 cases. In contrast, normal CRP levels were documented in 6 of the cases. MRI imaging analysis (Table 2) In 8 cases, MRI images were available for radiological analysis, and in 2 cases, MRI images were not available for reevaluation.

Acta Orthopaedica 2021; 92 (6): 753–759 755

Table 1. Summary of clinical, serological, pathological and molecular data Case Swelling CRP WBC count no. Sex Age (days) (< 5 mg/L) (< 11.3 G/L) Biopsy 1 m 26 38 0.6 4.8 2 m 16 29 21.9 11.2 3 f 22 28 0.6 6.6 4 m 36 42 2.1 5.1 5 f 12 23 1.0 7.6 6 f 26 17 1.1 6.3 7 f 40 33 n.a. n.a. 8 f 27 21 16.9 7.7 9 m 74 150 3.6 6.2 10 f 37 270 n.a. n.a.

IB n.d. n.d. EB n.d. CNB n.d. EB CNB, IB, EB EB

Serologic testing

Histology

Granulomatous inflammation with necrosis n.a. n.a. Granulomatous inflammation with necrosis n.a. Granulomatous inflammation with necrosis n.a. Granulomatous inflammation with necrosis Granulomatous inflammation with necrosis, metastasis of malignant melanoma Granulomatous inflammation with necrosis

PCR Culture

n.a. 1:256 1:64 n.a. 1:256 n.a. 1:64 1:256 n.a.

pos n.a. n.a. pos n.a. pos n.a. pos pos

neg n.a. n.a. neg n.a. neg n.a. neg neg

n.a.

pos

n.a.

Legend: CNB: core-needle biopsy; CRP: C-reactive protein; EB: excision biopsy; IB: incision biopsy; n.a.: not available; n.d.: not done; neg: negative; PCR: polymerase chain reaction; pos: positive; WBC: white blood cell.

Table 2. MRI characteristics of the lesions with differential diagnosis No.

1 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 2 Distal UA (ST) epi 28x45x28 c-s inf im h yes rim yes no Inflammation Vascular lesion with internal bleeding 3 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 4 Inguinal (ST) epi 29x15x35 s wd h h no hetero yes no Pathologic LN Soft tissue tumor (synovial sarcoma, neural tumor) 5 Distal UA (ST+B) no 29x26x11 s inf im h yes no no no Inflammation; Osteomyelitis; bone tumor bone sarcoma (Ewing sarcoma) 6 Distal UA (ST) epi 26x14x24 s inf h h yes hetero yes no Inflammation; Partial thrombosed STS venous structure; hematoma 7 Proximal UA (ST) epi 18x6x21 s inf im h yes rim yes no Inflammation; Sarcoma nodular fasciitis; STS 8 Proximal LA (ST) epi 37x16x22 c-s inf im h yes hetero yes yes Pathologic LN; Sarcoma/pathologic LN inflammation 9 Axilla (ST+LN) epi 49x75x110 s wd h h no hetero yes yes Pathologic LN; Liposarcoma STS 10. Knee (ST) epi 24x25x10 s inf im h yes homo no no Nodular fasciitis; Unclear lesion inflammation, STS A. Location (tissue type): B: bone; LA: lower arm; UA: upper arm, ST: soft tissue; LN: lymph node B. Fascial relation: epi: epifascial C. Size (mm) D. Structure: c-s: Cystic-solid; s: Solid E. Margin: inf: infiltrative; wd: well defined F. T1-weighted sequence, h: high; im: intermediate G. T2-weighted sequence, h: high H. Surrounding edema I. Enhancement hetero: heterogenous; homo: homogenous; J. Necrosis K. Satellite lesion(s) L. Differential diagnosis (internal); LN: lymph node, STS: soft tissue sarcoma M. Differential diagnosis (external): LN: lymph node, STS: soft tissue sarcoma; n.a.: not available;

The average diameter of the soft tissue lesions was 40 mm (21– 110). The dimensions of the intra- and extraosseous extension of the bone lesion in case number 5 measured 29 x 26 x 11 mm. All soft tissue lesions were located epifascially. 2 soft tissue lesions had a well-defined margin, while the rest showed an infiltrative

margin. Moreover, 2 lesions had a solid-cystic appearance and in the rest of the lesions no necrosis was observed. Additionally, in 2 lesions (cases 8 and 9) satellite lymph nodes were present. Due to MRI characteristics, 6/8 lesions were categorized as probably inflammatory (5 soft tissue lesions and 1 bone

756

Figure 2. Case no. 10. A 37-year-old female with an infrapatellar infiltrative soft tissue lesion of the knee. (A) Proton density (PD) image shows irregular heterogeneous soft tissue lesion in subcutaneous tissue just anterior to patellar ligament with surrounding edema (white arrow) and joint effusion. (B) Homogenous hypointense signal intensity on T1-weighted image (white arrow). (C) Axial PD image shows the hyperintense lesion and extension of soft tissue edema (white arrow). (D) Sagittal T1 VIBE image after application of Gd-contrast shows homogenous contrast enhancement of the lesion with an irregular margin (white arrow). On all sequences, the patellar ligament is intact. (E–F) Histology shows fatty tissue with extensive fibrosis and numerous granulomas with central necrosis. (G–H) Central necrosis (black stars) is often stellate in appearance with admixed neutrophils and surrounded by palisading histiocytes (black arrows, H).

lesion). In 3 cases, with still detectable lymph node structure and absent or initial liquefaction, the differential diagnosis included lymph node metastasis. In 2 cases, nodular fasciitis was suspected with an epifascially located mass in the infrapatellar region of the knee and upper arm (Figures 2 and 3) and sarcoma was suggested in 4 cases. In comparison with radiography, the MRI imaging features of the bone lesion (Figure 4) were suspicious of a bone tumor (such as lymphoma or Ewing sarcoma) or osteomyelitis. In case number 9, atypical MRI imaging findings included metastasis in the differential diagnosis (Figure 5).

Acta Orthopaedica 2021; 92 (6): 753–759

Figure 3. Case no. 7. A 40-year-old female with an epifascial soft tissue lesion of the upper arm. (A) On a T1-weighted image, the lesion with intermediate signal intensity and irregular margin; wide contact with the underlying fascia (arrow). (B) Homogenous high T2-weighted signal intensity without any surrounding edema (arrow). (C) Coronal T1-weighted image with fat saturation after application of Gd-contrast shows heterogeneous contrast enhancement (arrow). (D–E) DWI image (D) with corresponding (E) ADC map shows diffusion restriction due to necrotic collection (circles).

Histological, molecular, and laboratory/serological findings (Table 1) Due to inconclusive findings and suspicion of malignancy, core-needle, incision, or excision biopsies were performed in 6 cases. Histological examination showed numerous granulomas composed of central necrotic areas admixed with neutrophils surrounded by the palisading histiocytes (Figure 2); occasionally giant cells were also found. In 4 cases, granulomas were found only in the soft tissue and in 2 patients, soft tissues and the lymph nodes were affected. In 1 of the latter cases with soft tissue swelling in the left axilla (case number 9), a core-needle and incision biopsy was performed first, and CSD was diagnosed. However, due to findings suspicious of malignancy and after multidisciplinary discussion, an excision biopsy was performed. Histology revealed granulomatous inflammation in the soft tissue and the adjacent lymph node. A concomitant malignant melanoma composed of diffusely and atypical melanocytes arranged in clusters, with high mitotic activity and necrosis, was found in the same lymph node (Figure 5). The diagnosis was confirmed immunohistochemi-

Acta Orthopaedica 2021; 92 (6): 753–759 757

Figure 4. Case no. 5. A 13-year-old female with osteomyelitis of distal humerus as a late manifestation of cat scratch disease. (A) AP radiography of the elbow with discrete permeative osteodestruction pattern, cortical irregularity, and no periosteal reaction (circle). (B–C) Bone marrow infiltration of the medial epicondyle with permeative destruction of underlying cortex on axial proton density image (arrow, B) and coronal T1-weighted (arrow, C). (D) Bone marrow edema in the area and extraosseous extension with cortex destruction. Abnormal thickening and increased T2-weighted signal intensity within the common flexor origin from the lateral epicondyle due to inflammatory infiltration (white arrow) with edema of the surrounding subcutaneous fat tissue. (E) Corresponding color Doppler sonography shows cortical discontinuity with extraosseous soft tissue extension without significant hypervascularisation of the surrounding structures (yellow star). Table 3. Summary of treatment and follow-up No. Antibiotics

Treatment Follow-up length (days) (days)

1 ciprofloxacin/doxycycline 2 clarithromycin 3 ciprofloxacin/doxycycline 4 none 5 clarithromycin 6 doxycycline 7 azithromycin 8 none 9 doxycycline + rifampicin 10 none

55 7 11 − 49 14 14 − 28 −

60 37 11 59 100 33 111 11 53 28

cally by strong positive reaction of the tumor cells to S100, SOX10, HMB-45, and Melan A. To confirm the presence of Bartonella organisms, PCR analysis was performed in 6 cases

Figure 5. Case no. 9. A 74-year-old male with lymph node melanoma metastasis and synchronous cat scratch disease. (A) Enlarged axillary lymph nodes with heterogenous contrast enhancement on coronal postcontrast T1-weighted image with fat saturation; no signs of necrosis (star). (B) Sonography shows irregular 28 x 26 mm large lymph nodes with the heterogeneous structure without necrosis and hyperechogenic surrounding subcutaneous tissue (arrow). (C) On axial short tau inversion recovery (STIR) image and (D) T1-weighted image pathologically changed lymph nodes with a heterogenous signal. Centrally, on T2-weighted image low signal intensity with hyperintense signal intensity on T1-weighted image corresponds to melanin (arrows). (E) Histologically shows granuloma with central necrosis (star) and malignant melanoma metastasis (arrow). (F) Higher-power view of the granuloma with extensive central necrosis, surrounded by palisading histiocytes and (G) atypical melanocytes. (H) Malignant melanoma is immunohistochemically positive with SOX10 (Inlet: positive reaction to Melan A).

and was positive. In addition, in 5 cases serologic testing for the presence of antibodies to B. henselae was positive. Moreover, in 5 patients, microbiological cultures were performed and all came back negative. Treatment and follow-up (Table 3) In short, in 7 cases, antibiotic therapy was given for a median duration of 14 days (7–55). In the other 3 cases, patients did

758

not receive any further therapy after the excision. Median follow-up was 45 days (11–111). In all patients, swelling and pain decreased and all were free of CSD at the last follow-up.

Discussion We report 10 patients with CSD, 4 of them presenting at atypical locations with soft tissue involvement mimicking sarcoma. Due to unspecific symptoms and the diversity of systemic manifestations, CSD recognition can be difficult and easily confused with soft tissue tumors. The patient’s history, physical examination, and diagnostic imaging play an important part in establishing the correct diagnosis. This helps in the prevention of further invasive procedures and the provision of rapid, appropriate treatment. 5 patients presented with visible cat scratches most frequently located on the upper extremities in our case series, as previously published (Carithers 1985, Gielen et al. 2003, Raoult et al. 2006). Sonography, an initial imaging method used for the assessment of superficial soft tissue tumors, is used also in typical CSD, demonstrating enlarged hypovascularized lymph node(s) adjacent to the entry point of infection with different degrees of liquefaction (Mazur-Melewska et al. 2015, Bernard et al. 2016). On MRI, in early disease, the affected enlarged lymph node has a typical appearance with a thickened cortex and timely evolution of the sharp delineated perinodal abscesses with gradual necrosis of the lymph node (Mazur-Melewska et al. 2015, Chen et al. 2018). Computed tomography (CT) is inferior to MRI because of its limited differentiation of softtissue structures and edema (Bernard et al. 2016). An atypical manifestation of CSD, such as knee soft tissue mass and osteomyelitis, as in our cohort, is rare and mainly affects children (Carithers 1985, Robson et al. 1999, Donà et al. 2018). Osteomyelitis associated with B. henselae frequently occurs along with regional lymphadenopathy due to hematogenous or lymphatic spread after the inoculation of the microorganisms (Florin et al. 2008, Erdem et al. 2018). In atypical CSD, MRI is the imaging technique of choice in characterization of inflammatory soft tissue mass with adjacent soft tissue edema and in evaluation of bone lesions (Mazur-Melewska et al. 2015, Erdem et al. 2018). For bone lesions, radiography or CT are complementary imaging methods to MRI, useful in evaluating the bone destruction pattern and periosteal reaction. A permeative destruction pattern with cortical destruction and extraosseous soft tissue extension can mimic primary malignant bone processes, as seen in one of our patients (Figure 4, case number 5). Histologically, the finding of granulomatous inflammation with central necrosis is a feature associated with various infections, such as CSD and tuberculosis caused by Mycobacterium spp. Therefore, further analysis of the acquired specimens should be performed. Due to its slow growth, B. henselae is difficult to culture. Better diagnostic methods for detecting B.

Acta Orthopaedica 2021; 92 (6): 753–759

henselae include serologic testing and molecular analysis, such as PCR amplification of Bartonella spp. genes (Sander et al. 1998, 1999, Herremans et al. 2007, Edouard et al. 2015). In our study, in all 6 patients tested prior to antibiotic therapy, PCR was positive and in 5 patients, serology for B. henselae came back positive. Of note, a positive B. henselae serology or PCR does not exclude other etiology, such as malignancy or osteomyelitis (Florin et al. 2008). In the study by Rolain et al. (2006), the histopathological analysis of 181 lymph node biopsy specimens from patients with suspected CSD showed concurrent diseases in 13 B. henselae positive patients: 10 patients had a mycobacterial infection, and 3 had a malignant disease (2 lymphomas and 1 Hodgkin disease). Biopsy is not indicated in lesions with unequivocal clinical and imaging features of inflammation. In atypical cases, when B. henselae PCR is positive and histology characteristic, concurrent malignancy (primary or metastasis) should be ruled out by performing a biopsy (Florin et al. 2008, Baranowski and Huang 2020). In our study, in 3 patients biopsy was performed, as radiological findings were suspicious of malignancy despite positive serology in 1 of the cases, and in 1 patient a synchronous metastasis of malignant melanoma was found. Treatment of CSD frequently includes surgery and antibiotic therapy. Up to one-third of cases develop suppuration of the affected lymph nodes and evacuation of the pus is indicated (Wang et al. 2009). Lymphadenectomy is not advised due to the possibility of the development of fistulas (Baranowski and Huang 2020). In our study, even though B. henselae infection was confirmed in 2 cases, lymphadenectomy was performed due to imaging findings suspicious of malignancy. 1 patient (case number 4) was free of local postoperative complications and in the second patient (case number 9) histology showed granulomatous inflammation with synchronous metastasis of the malignant melanoma. This case illustrates that only histologic analysis can rule out malignant diseases, and if the findings are inconclusive, the biopsy should be repeated. The radiological differential diagnosis of CSD includes other infections and a range of benign and malignant soft tissue tumors, such as peripheral nerve sheath tumors, synovial sarcoma, leiomyosarcoma, and distant nodal metastasis (Mazur-Melewska et al. 2015, Bernard et al. 2016, Chen et al. 2018). As seen in our study, the radiologist’s experience and knowledge of different imaging features may play an important role in diagnosing typical and atypical findings in the spectrum of CSD, resulting in a significantly lower number of equivocal findings suspicious of malignancy. In addition, for the differentiation of lymph nodes in the epitrochlear region from other soft tissue masses, key anatomical features, including location posterior to the basilic vein, superficial to the brachialis muscle and brachial fascia or intermuscular septum covering the medial head of the triceps and ulnar nerve, can be used (Bernard et al. 2016). In conclusion, this study showed a very low prevalence of 0.4% CSD cases in a musculoskeletal orthopedic sar-

Acta Orthopaedica 2021; 92 (6): 753–759 759

coma center. Radiologically, CSD is typically presented as a pathologically enlarged lymph node with different grades of liquefaction and inflammatory changes of surrounding tissues. However, in atypical cases, CSD may mimic soft tissue or bone sarcoma due to undefined soft tissue mass without discernible lymph node structure or bone involvement. These cases cause diagnostic challenges, and the differential diagnosis includes benign and malignant neoplasms (such as sarcoma or carcinoma metastasis) and infections. Besides the clinical picture, patient history, and radiologic imaging, histopathology together with PCR and serology remain the diagnostic methods of choice to establish the correct diagnosis. The distinction between these possibilities is crucial for treatment and prognosis. Furthermore, this study highlights the importance of treating patients with suspicious soft tissue lesions irrespective of size and patients with a deep-seated lesion larger than 5 cm in a specialized multidisciplinary sarcoma center. AF, IJ, BI, BM, and FJ substantial contributed to research design, AF, IJ, BI, VT, LL, and GM analyzed and interpreted the data. AF, IJ, BI, and LA drafted the paper and all authors revised it critically. All authors have read and approved the final submitted manuscript Acta thanks Rudolf W. Poolman help with peer review of this study. Baranowski K, Huang B. Cat scratch disease. In: StatPearls. Treasure Island, FL: StatPearls Publishing LLC; 2020. Bernard S A, Walker E A, Carroll J F, Klassen-Fischer M, Murphey M D. Epitrochlear cat scratch disease: unique imaging features allowing differentiation from other soft tissue masses of the medial arm. Skeletal Radiol 2016; 45(9): 1227-34. doi: 10.1007/s00256-016-2407-6. Carithers H A. Cat-scratch disease: an overview based on a study of 1,200 patients. Am J Dis Child 1985; 139(11): 1124-33. doi: 10.1001/archpedi.1985.02140130062031. Chen Y, Fu Y-B, Xu X-F, Pan Y, Lu C-Y, Zhu X-L, Li Q-H, Yu R-S. Lymphadenitis associated with cat-scratch disease simulating a neoplasm: imaging findings with histopathological associations. Oncology Letters 2018; 15(1): 195-204. doi: 10.3892/ol.2017.7311. Colman M W, Lozano-Calderon S, Raskin K A, Hornicek F J, Gebhardt M. Non-neoplastic soft tissue masses that mimic sarcoma. Orthop Clin North Am 2014; 45(2): 245-55. doi: 10.1016/j.ocl.2013.12.006. Dhal U, Tarrand J, Kontoyiannis D P. 1551. Cat scratch disease as a mimicker of malignancy: rare and elusive. Open Forum Infectious Diseases 2020; 7(Supplement_1): S777-S. doi: 10.1093/ofid/ofaa439.1731. Donà D, Nai Fovino L, Mozzo E, Cabrelle G, Bordin G, Lundin R, Giaquinto C, Zangardi T, Rampon O. Osteomyelitis in cat-scratch disease: a never-ending dilemma—a case report and literature review. Case Rep Pediatr 2018; 2018: 1679306. doi: 10.1155/2018/1679306. Edouard S, Nabet C, Lepidi H, Fournier P E, Raoult D. Bartonella, a common cause of endocarditis: a report on 106 cases and review. J Clin Microbiol 2015; 53(3): 824-9. doi: 10.1128/jcm.02827-14.

Eichhorn-Sens J, Bund T, Vogt P M. [Painful soft-tissue swelling of the upper arm]. Der Chirurg; Zeitschrift fur alle Gebiete der operativen Medizen 2008; 79(3): 249-51. doi: 10.1007/s00104-007-1317-5. Erdem G, Watson J R, Hunt W G, Young C, Tomatis Souverbielle C, Honegger J R, Cassady K A, Ilgenfritz M, Napolitano S, Koranyi K. Clinical and radiologic manifestations of bone infection in children with cat scratch disease. J Pediatr 2018; 201:274-80.e12. doi: 10.1016/j. jpeds.2018.05.033. Florin T A, Zaoutis T E, Zaoutis L B. Beyond cat scratch disease: widening spectrum of Bartonella henselae infection. Pediatrics 2008; 121(5): e141325. doi: 10.1542/peds.2007-1897. Fox B C, Gurtler R A. Cat-scratch disease mimicking rhabdomyosarcoma. Orthop Rev 1993; 22(10): 1148-9. Frank G M, Marshall C S, Theodore W. Cat scratch disease simulating sarcoma of the neck. Am J Surg 1952; 84(4): 483-5. doi: https://doi. org/10.1016/0002-9610(52)90021-4. Gielen J, Wang X L, Vanhoenacker F, De Schepper H, De Beuckeleer L, Vandevenne J, De Schepper A. Lymphadenopathy at the medial epitrochlear region in cat-scratch disease. Eur Radiol 2003; 13(6): 1363-9. doi: 10.1007/s00330-002-1673-y. Herremans M, Vermeulen M J, Van de Kassteele J, Bakker J, Schellekens J F, Koopmans M P. The use of Bartonella henselae-specific age dependent IgG and IgM in diagnostic models to discriminate diseased from non-diseased in cat scratch disease serology. J Microbiol Methods 2007; 71(2): 107-13. doi: 10.1016/j.mimet.2007.09.004. Huang C L, Kitano M, Nagasaki H, Tatsumi A, Yamanaka A, Matsui T, Yamashita N. [A case report of cat scratch disease which mimicked a metastasis of lung cancer]. Bull Chest Dis Res Inst Kyoto Univ 1989; 22(1-2): 11-18. Mazur-Melewska K, Jończyk-Potoczna K, Mania A, Kemnitz P, Szydłowski J, Służewski W, Figlerowicz M. The significance of Bartonella henselae bacterias for oncological diagnosis in children. Infect Agent Cancer 2015; 10: 30. doi: 10.1186/s13027-015-0025-x. Nimityongskul P, Anderson L D, Sri P. Cat-scratch disease: orthopaedic presentation. Orthop Rev 1992; 21(1): 55-9. Raoult D, Roblot F, Rolain J M, Besnier J M, Loulergue J, Bastides F, Choutet P. First isolation of Bartonella alsatica from a valve of a patient with endocarditis. J Clin Microbiol 2006; 44(1): 278-9. doi: 10.1128/ jcm.44.1.278-279.2006. Robson J M, Harte G J, Osborne D R, McCormack J G. Cat-scratch disease with paravertebral mass and osteomyelitis. Clin Infect Dis 1999; 28(2): 274-8. doi: 10.1086/515102. Rolain J M, Lepidi H, Zanaret M, Triglia J M, Michel G, Thomas P A, Texereau M, Stein A, Romaru A, Eb F, Raoult D. Lymph node biopsy specimens and diagnosis of cat-scratch disease. Emerg Infect Dis 2006; 12(9): 1338-44. doi: 10.3201/eid1209.060122. Sander A, Posselt M, Oberle K, Bredt W. Seroprevalence of antibodies to Bartonella henselae in patients with cat scratch disease and in healthy controls: evaluation and comparison of two commercial serological tests. Clin Diagn Lab Immunol 1998; 5(4): 486-90. Sander A, Posselt M, Böhm N, Ruess M, Altwegg M. Detection of Bartonella henselae DNA by two different PCR assays and determination of the genotypes of strains involved in histologically defined cat scratch disease. J Clin Microbiol 1999; 37(4): 993-7. Wang C W, Chang W C, Chao T K, Liu C C, Huang G S. Computed tomography and magnetic resonance imaging of cat-scratch disease: a report of two cases. Clin Imaging 2009; 33(4): 318-21. doi: 10.1016/j.clinimag.2009.01.006.

760

Acta Orthopaedica 2021; 92 (6): 760–762

Correspondence

Isometric hip strength impairments in patients with hip dysplasia are improved but not normalized 1 year after periacetabular osteotomy: a cohort study of 82 patients Jacobsen et al. Acta Orthop 2021; 92(3): 285-291. DOI 10.1080/17453674.2020.1864911

Sir,—We read with great interest the article “Isometric hip strength impairments in patients with hip dysplasia are improved but not normalized 1 year after periacetabular osteotomy: a cohort study of 82 patients” by Jacobsen et al. (2021). The authors investigated isometric hip muscle strength in patients with hip dysplasia, before and 1 year after periacetabular osteotomy, and compared this with healthy volunteers. We agree with the conclusions drawn by the authors. However, there are some issues we like to comment on. First, healthy volunteers did not undergo any imaging, and it is therefore unknown if they were radiologically healthy despite having no hip symptoms or other joint abnormalities. Furthermore, the findings of positive FADIR test (3/50) and positive FABER test (2/50) in some healthy volunteers could be related to persons not being “healthy.” Hip disorders should be diagnosed based on a combination of symptoms, clinical signs and imaging findings. Hence, it is currently unknown whether asymptomatic hips have hidden other hip disorders. Consequently, we believe that healthy people should be evaluated more comprehensively. Second, 89 patients with bilateral hip dysplasia were recruited in this study. However, inclusion of bilateral hip dysplasia patients may also lead to bias. Previous studies showed that the hip muscle strength of the contralateral hip joint in patients with unilateral femoroacetabular impingement syndrome (Malloy et al. 2019) or hip osteoarthritis (Arokoski et al. 2002, Diamond et al. 2016) will also be affected. We firmly believe that this condition can also occur in patients with hip dysplasia. However, side differences of hip muscle strength between affected and contralateral leg were not analyzed bilaterally in patients affected with hip dysplasia. Third, hand-held instead of stabilized dynamometry which may be less robust than other forms of strength testing (e.g., Biodex system) was used for strength assessment of hip flexion, extension, abduction, and adduction. Intrarater reliability

of hand-held dynamometry has been shown to be lower compared with stabilized dynamometry due to the influence of the investigator’s strength to resist the measured forces (Thorborg et al. 2009, Casartelli et al. 2010). Furthermore, the considerable hip dysplasia-related hip muscle weakness observed pre- and post-operation in this study could potentially originate from different factors: a mechanical/anatomical limit, qualitative (fatty degeneration), quantitative (atrophy) morphologic alterations or reduced muscle activation (possibly related to pain and/or fear of pain) during isometric hip muscle contraction. Although the authors found that an increased hip muscle strength was associated with higher hip functional scores, however, finding that hip muscle weakness was predictive of hip–specific outcomes does not imply causality. It is possible that poor surgical outcomes caused disuse muscle weakness that are then detected through dynamometer rather than hip muscle weakness affecting patient clinical outcomes. Further research would be necessary to clarify this point. Additionally, the cross sectional design did not clarify the development in muscle strength over time in patients with hip dysplasia and the hand-held dynamometer can’t indicate which specific muscles of the hip is weakened. We do not know which specific muscles of hip have weakened and which have not after periacetabular osteotomy. However, hip muscle size can be assessed by quantifying the cross-sectional area using axial cuts of pelvis magnetic resonance imaging (Malloy et al. 2019, Gao et al. 2021). Mingjin Zhong and Weimin Zhu Department of Sports Medicine The First Affiliated Hospital of Shenzhen University Health Science Center, Shenzhen Second People’s Hospital Shenzhen, Guangdong, China Email:sportsmedzhong@sina.com

Acta Orthopaedica 2021; 92 (6): 760–762

761

Sir,— We are pleased that Dr. Zhong agrees with the conclusion of our findings which states 1) “Isometric hip muscle strength is impaired in patients with symptomatic dysplastic hips measured before PAO”, and 2) “1 year after surgery, isometric hip flexion and abduction strength had improved but muscle strength did not reach that of healthy volunteers.” Furthermore, we would also like to thank Dr. Zhong for the possibility to elaborate on key issues raised in the letter, described point by point. Point 1: The healthy volunteers were asymptomatic and excluded in case of pain, comorbidity, previous trauma or surgery. Radiology was used only in the patient population as it was considered unethical to expose the asymptomatic population to radiation. Therefore, “asymptomatic” would have been a better term than “healthy” volunteers. Regarding imaging findings, we agree that imaging alone cannot be used to determine whether (or not) participants are healthy. Instead a combination of symptoms, clinical signs and imaging should be used to assess the presence of “hip disease”, as agreed for femoroacetabular impingement syndrome (Griffin et al. 2016). This is why the findings of positive FADIR/FABER tests do not indicate whether participants are healthy. Painful FADIR tests have been documented in 12–15% of asymptomatic participants (Czuppon et al. 2017); this is most likely due to the high sensitivity and false positive rate of this test (Reiman et al. 2013). We consider the FADIR test to be positive only if it replicates known symptoms (Troelsen et al. 2009). Therefore, in the asymptomatic volunteers, it would have been less confusing if we had described whether a test was painful instead of labeling the test as positive or negative. In the case of missing “hidden” pathology in the asymptomatic volunteers, the muscle strength deficit seen in the patients in our study might have been larger, but this would not have changed our conclusion that “isometric hip muscle strength is impaired in patients with symptomatic dysplastic hips.” Point 2: In our sample 63% had bilateral symptoms (89% radiological bilateral affection). The patients had a significant strength deficit of both symptomatic and asymptomatic side

compared to asymptomatic volunteers, and apart from hip abduction, there was no statistically significant differences between the two, as expected, and as correctly pointed out by Dr. Zhong (Table, new data). Moreover, the majority of patients with hip dysplasia are bilaterally affected, and therefore we do not consider side comparison relevant as most often the contralateral side cannot be considered “normal.” Point 3: Dr. Zhong stated that the intra-rater reliability of hand-held dynamometry (HHD) was worse compared with stabilized dynamometry. This statement is not supported by the studies referred to in the letter (Thorborg et al. 2010, Casartelli et al. 2011) or in previous studies (Thorborg et al. 2010, 2013, Casartelli et al. 2011, Kemp et al. 2013, Chamorro et al. 2017). On the contrary, it has been shown that the reliability of HHD compared with isokinetic dynamometry was not inferior but comparable (Chamorro et al. 2017). On this basis, we consider HHD applicable to evaluate muscle strength in this population, where muscle strength of the tester seems to surpass the strength of the person being tested. Finally, our study was not designed to investigate whether the reported strength deficit originated from either mechanical, qualitative or quantitative causes. Instead, muscle strength was tested with a HHD in 4 directions (Thorborg et al. 2010), and therefore deficits of specific muscles were not investigated. Nevertheless, we consider knowledge of deficits in specific directions useful in clinical practice since this knowledge can support monitoring of weakness and direct exercise interventions (Kemp et al. 2019). Julie Sandell Jacobsen 1,2, Stig Storgaard Jakobsen 3, Kjeld Søballe 3,4, Per Hölmich 5, Kristian Thorborg 5,6 1 Research Centre for Health and Welfare Technology, Programme for Rehabilitation, VIA University College, Aarhus. 2 Research Unit for General Practice in Aarhus, Aarhus. 3 Department of Orthopaedic Surgery, Aarhus University Hospital, Aarhus.

Mean hip muscle strength (Nm/kg) in 50 asymptomatic volunteers and 37 patients with unilateral symptoms related to hip dysplasia divided in asymptomatic and symptomatic hips Asymptomatic Asymptomatic Symptomatic hip hip hip (volunteers) (dysplasia ptts.) (dysplasia ptts.) n Mean (SD) n Mean (SD) n Mean (SD) Flexion Extension Abduction Adduction

50 50 50 50

a Difference

1.8 (0.34) 2.5 (0.63) 1.5 (0.37) 1.5 (0.46)

37 36 36 36

1.2 (0.40) 1.9 (0.68) 1.3 (0.35) 1.1 (0.37)

37 36 37 37

1.2 (0.35) 1.8 (0.59) 1.1 (0.38) 1.0 (0.31)

Difference a (95% CI) p-value 0.60 (0.44–0.76) < 0.001 0.60 (0.32–0.88) 0.001 0.20 (0.04–0.36) 0.01 0.40 (0.22–0.58) < .001

Difference b Difference c (95% CI) p-value (95% CI) 0.60 (0.45–0.75) 0.70 (0.43–0.97) 0.40 (0.24–0.56) 0.50 (0.33–0.67)

< 0.001 < 0.001 < 0.001 < 0.001

p-value

0.00 (-0.17 to 0.17) 0.10 (-0.20 to 0.40) 0.20 (0.31 to 0.37) 0.10 (-0.06 to 0.26)

1.0 0.5 0.02 0.2

in muscle strength between hip in asymptomatic volunteers compared with asymptomatic hip in hip dysplasia patients (ptts.) with unilateral hjp symptoms. in hip muscle strength between hip in asymptomatic volunteers compared with symptomatic hip in hip dysplasia patients with unilateral hip symptoms. c Difference in hip muscle strength between hip in asymptomatic and symptomatic hip in hip dysplasia patients with unilateral hip symptoms. b Difference

762

4 Department of Clinical Medicine, Aarhus University, Aarhus. 5 Sports Orthopaedic Research Center-Copenhagen (SORCC), Department of Orthopaedic Surgery, Copenhagen University Hospital, Hvidovre. 6 Physical Medicine and Rehabilitation Research- Copenhagen (PMR-C), Department of Physical and Occupational Therapy, Copenhagen University Hospital, Hvidovre, Denmark. Correspondence: jsaj@via.dk Arokoski M H, Arokoski J P, Haara M, Kankaanpää M, Vesterinen M, Niemitukia L H, Helminen H J. Hip muscle strength and muscle cross sectional area in men with and without hip osteoarthritis. J Rheumatol 2002; 29: 2185-95. Casartelli N C, Maffiuletti N A, Item-Glatthorn J F, Staehli S, Bizzini M, Impellizzeri F M, Leunig M. Hip muscle weakness in patients with symptomatic femoroacetabular impingement. Osteoarthritis Cartilage 2011; 19: 816-21. Chamorro C, Armijo-Olivo S, De la Fuente C, Fuentes J, Javier Chirosa L. Absolute reliability and concurrent validity of hand held dynamometry and isokinetic dynamometry in the hip, knee and ankle joint: systematic review and meta-analysis. Open Med (Wars) 2017; 12: 359-75. Czuppon S, Prather H, Hunt D M, Steger-May K, Bloom N J, Clohisy J C, Larsen R, Harris-Hayes M. Gender-dependent differences in hip range of motion and impingement testing in asymptomatic college freshman athletes. PM R 2017; 9(7): 660-7. Diamond L E, Wrigley T V, Hinman R S, Hodges P W, O’Donnell J, Takla A, Bennell K L. Isometric and isokinetic hip strength and agonist/antagonist ratios in symptomatic femoroacetabular impingement. J Sci Med Sport 2016; 19: 696-701. Gao Y, Lyu X, Liu Q, Meng Y, Wang J, Pan S. Quantitative evaluation of hip muscle atrophy in patients with unilateral slipped capital femoral epiphysis based on magnetic resonance imaging. Acad Radiol 2021; 28(8): 1125-32.

Acta Orthopaedica 2021; 92 (6): 760–762

Griffin D R, Dickenson E J, O’Donnell J, Agricola R, Awan T, Beck M, Clohisy J C, Dijkstra H P, Falvey E, Gimpel M, Hinman R S, Hölmich P, Kassarjian A, Martin H D, Martin R, Mather R C, Philippon M J, Reiman M P, Takla A, Thorborg K, Walker S, Weir A, Bennell K L. The Warwick Agreement on femoroacetabular impingement syndrome (FAI syndrome): An international consensus statement. Br J Sports Med 2016; 50(19): 1169-76. Jacobsen J S, Jakobsen S S, Søballe K, Hölmich P, Thorborg K. Isometric hip strength impairments in patients with hip dysplasia are improved but not normalized 1 year after periacetabular osteotomy: a cohort study of 82 patients. Acta Orthop 2021; 92(3): 285-91. Kemp J L, Schache A G, Makdissi M, Sims K J, Crossley K M. Greater understanding of normal hip physical function may guide clinicians in providing targeted rehabilitation programmes. J Sci Med Sport 2013; 16(4): 292-6. Kemp J L, King M G, Barton C, Schache A G, Thorborg K, Roos E M, Scholes M, Grimaldi A, Semciw A I, Freke M, Risberg M A, Reiman M P, Mayes S, Pizzari T, Heerey J J, Lawrenson P R, Ingelsrud L H H, Crossley K M. Is exercise therapy for femoroacetabular impingement in or out of FASHIoN? We need to talk about current best practice for the non-surgical management of FAI syndrome. Br J Sports Med 2019; 53(19): 1204-5. Malloy P, Stone A V, Kunze K N, Neal W H, Beck E C, Nho S J. Patients with unilateral femoroacetabular impingement syndrome have asymmetrical hip muscle cross-sectional area and compensatory muscle changes associated with preoperative pain level. Arthroscopy 2019; 35: 1445-53. Reiman M P, Goode A P, Hegedus E J, Cook C E, Wright A A. Diagnostic accuracy of clinical tests of the hip: a systematic review with meta-analysis. Br J Sports Med 2013; 47(14): 893-902. Thorborg K, Petersen J, Magnusson S P, Hölmich P. Clinical assessment of hip strength using a hand-held dynamometer is reliable. Scand J Med Sci Sports 2010(3); 20: 493-501. Thorborg K, Bandholm T, Hölmich P. Hip- and knee-strength assessments using a hand-held dynamometer with external belt-fixation are inter-tester reliable. Knee Surgery, Sport Traumatol Arthrosc 2013; 21(3): 550–5. Troelsen A, Mechlenburg I, Gelineck J, Bolvig L, Jacobsen S, Søballe K. What is the role of clinical tests and ultrasound in acetabular labral tear diagnostics? Acta Orthop 2009; 80(3): 314-8.

Acta Orthopaedica 2021; 92 (6): 763–764 763

Erratum

Physical child abuse demands increased awareness during health and socioeconomic crises like COVID-19 A review and education material Polina MARTINKEVICH 1,2a, Lise Langeland LARSEN 1,2a, Troels GRÆSHOLT-KNUDSEN 3a, Gitte HESTHAVEN 4, Michel Bach HELLFRITZSCH 2,5, Karin Kastberg PETERSEN 5, Bjarne MØLLER-MADSEN 1,2,6, and Jan Duedal RÖLFING 1,2,6,7 1 Department

of Orthopaedics, Aarhus University Hospital; 2 Danish Paediatric Orthopaedic Research; 3 Research Unit for Mental Public Health, Department of Public Health, Aarhus University; 4 Department of Paediatrics, Aarhus University Hospital; 5 Department of Radiology, Aarhus University Hospital; 6 Department of Clinical Medicine, Aarhus University; 7 Corporate HR, MidtSim, Central Denmark Region, Denmark a Shared first authorship Correspondence: jan.roelfing@clin.au.dk Acta Orthop 2020; 91 (5): 527–533, doi: 10.1080/17453674.2020.1782012

The infographic about red flags that should raise suspicion of physical child abuse (Figure 1 in Supplementary data) has been updated with minor text corrections. Furthermore, a complete reference list for the infographic has been added.

1. Stith S M, Liu T, Davies L C, Boykin E L, Alder M C, Harris J M, Som A, McPherson M, Dees J E M E G. Risk factors in child maltreatment: A meta-analytic review of the literature. Aggress Violent Behav 2009; 14(1): 13-29. doi:10.1016/j.avb.2006.03.006 2. Baird E. Non-accidental injury in children in the time of COVID-19 pandemic. The Transient Journal. Published April 8, 2020. Accessed July 21, 2021. https://www.boa.ac.uk/resources/knowledge-hub/non-accidentalinjury-in-children-in-the-time-of-covid-19-pandemic.html 3. Vanderminden J A. A longitudinal analysis of the effect of disability type and emotional/behavior problems on different forms of maltreatment across childhood. Published online 2013. 4. Kelly P, Thompson J M D, Koh J, Ameratunga S, Jelleyman T, Percival T M, Elder H, Mitchell E A. Perinatal risk and protective factors for pediatric abusive head trauma: a multicenter case-control study. J Pediatr 2017; 187: 240-6.e4. doi:10.1016/j.jpeds.2017.04.058 5. Benedict M I, Zuravin S, Brandt D, Abbey H. Types and frequency of child maltreatment by family foster care providers in an urban population. Child Abuse Negl 1994; 18(7): 577-85. doi:10.1016/01452134(94)90084-1 6. Sorenson S B, Peterson J G. Traumatic child death and documented maltreatment history, Los Angeles. Am J Public Health 1994; 84(4): 623-7. doi:10.2105/AJPH.84.4.623 7. Cross D, Vance L A, Kim Y J, Ruchard A L, Fox N, Jovanovic T,

Bradley B. Trauma exposure, PTSD, and parenting in a community sample of low-income, predominantly African American mothers and children. Psychol Trauma Theory Res Pract Policy 2018; 10(3): 327-35. doi:10.1037/tra0000264 8. Kostelny K, Lamin D, Manyeh M, Ondoro K, Stark L, Lilley S, Wessells M. “Worse than the war”: An ethnographic study of the impact of the Ebola crisis on life, sex, teenage pregnancy, and a community-driven intervention in rural Sierra Leone. London: Save the Children. Published online 2016. 9. Keenan H T, Marshall S W, Nocera M A, Runyan D K. Increased incidence of inflicted traumatic brain injury in children after a natural disaster. Am J Prev Med 2004; 26(3): 189-93. doi:10.1016/j.amepre.2003.10.023 10. Douglas E, Vanderminden J. A longitudinal, multilevel analysis of homicide against children aged 0–9 years using state-level characteristics: 1979–2007. Violence Vict 2014; 29(5): 757-70. doi:10.1891/08866708.VV-D-12-00085 11. Molnar B E, Goerge R M, Gilsanz P, Hill A, Subramanian S V, Holton J K, Duncan D T, Beatriz E D, Beardslee W R. Neighborhood-level social processes and substantiated cases of child maltreatment. Child Abuse Negl 2016; 51: 41-53. doi:10.1016/j.chiabu.2015.11.007 12. Black D A, Heyman R E, Smith Slep A M. Risk factors for child physical abuse. Aggress Violent Behav 2001; 6(2): 121-88. doi:10.1016/ S1359-1789(00)00021-5

764

Acta Orthopaedica 2021; 92 (6): 763–764

Supplementary data

SUSPICION OF PHYSICAL CHILD ABUSE DO NOT HESITATE Consult with Child Protection Team Act according to your local guidelines and laws

ANAMNESTIC RED FLAGS

OBJECTIVE RED FLAGS

RADIOLOGICAL RED FLAGS

All injuries in child < 6 months of age

Respiratory deterioration

Any fracture in child unable to transport itself

Seizures

No trauma/denial of trauma No witness to injury

Retinal hemorrhage

AND/OR

Long-bone fractures: femur, humerus, tibia

SKIN

Rib fractures

Delay in seeking medical care

Infant with any bruise

Multiple fractures

History not compatible with injury

Child < 4 years: bruise in specific regions a

Multiple fractures of different ages

Repeated emergency room visits

Burns and bite marks

Classic metaphyseal lesion(s)

Regions: chest, abdomen, back, buttocks, genitourinary region, hips, ears, neck Forms

Skull fracture(s) Subdural hematoma/hygroma

RISK INDICATORS CHILD

Calls-to-Action RISK Landing INDICATORS Pages CAREGIVER

RISK INDICATORS ENVIRONMENT

Internalizing behaviors (fearfulness, social distancing) (1)

Low parental age (1)

Health crises (2,8)

Low socioeconomic status (1)

Natural disasters (9)

Violence in family (1)

High levels of violent crime (10)

Lack of social competences (1)

Maltreated in childhood (1)

Low neighborhood cohesion (11)

Appears neglected (2)

Multiple pregnancies/twins/triplets (1)

Neighborhood poverty (12)

Learning and intellectual disability (3)

Substance abuse (1)

Born prematurely (4)

Psychiatric illness (1,7)

Living in foster care (5)

For example depression, posttraumatic stress disorder, attention deficit, hyperactivity disorder, internet gaming disorder

Externalizing behaviors (violence, hyperactivity) (1)

Prior contact with social services (6)

Figure 1. Red flags and risk indicators of physical child abuse.

Relational issues between caregivers (1) Inappropriate interaction between caregiver–child (2)

Acta Orthopaedica 2021; 92 (6): 765

765

Erratum

Isometric hip strength impairments in patients with hip dysplasia are improved but not normalized 1 year after periacetabular osteotomy: a cohort study of 82 patients Julie Sandell JACOBSEN 1,2, Stig Storgaard JAKOBSEN 3, Kjeld SØBALLE 3,4, Per HÖLMICH 5, and Kristian THORBORG 5,6 1 Research

Centre for Health and Welfare Technology, Programme for Rehabilitation, VIA University College, Aarhus; 2 Research Unit for General Practice in Aarhus, Aarhus; 3 Department of Orthopaedic Surgery, Aarhus University Hospital, Aarhus; 4 Department of Clinical Medicine, Aarhus University, Aarhus; 5 Sports Orthopaedic Research Center-Copenhagen (SORC-C), Department of Orthopaedic Surgery, Copenhagen University Hospital, Hvidovre; 6 Physical Medicine and Rehabilitation Research-Copenhagen (PMR-C), Department of Physical and Occupational Therapy, Copenhagen University Hospital, Hvidovre, Denmark Correspondence: jsaj@via.dk

Acta Orthop 2021; 92(3): 285-291. DOI 10.1080/17453674.2020.1864911

In the published article Figure 2 had a typographical error in the y-axis values which went from 0 to 6 now corrected to 0 to 5.

Isometric hip muscle strength (Nm/kg) 5 Healthy volunteers Patients before PAO Patients 1 year after PAO

Flexion

Extension Abduction Adduction

Figure 2. Median isometric hip muscle strength in patients with hip dysplasia and in healthy volunteers in Nm/kg; box represents 25th and 75th percentiles and error bars represent 10th and 90th percentiles. PAO = periacetabular osteotomy.

6/21

Medical ACTA ORTHOPAEDICA

LONGER IMPLANT SURVIVAL. WITH THE RIGHT BONE CEMENT.

Real World Data show:

19%

lower revision risk* with PALACOS® R+G compared to other bone cements

COVER.indd 1

10666

www.heraeus-medical.com

Vol. 92, No. 6, 2021 (pp. 633–765)

The element of success in joint replacement

Volume 92, Number 6, December 2021

27-10-2021 21:25:06