Acta Orthopaedica, Vol. 92, Issue 1, 2021 by Acta

1/21

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

www.heraeus-medical.com

Vol. 92, No. 1, 2021 (pp. 1–12 )

Bone cement with gentamicin ttamicin i i and clindamycin

ACTA ORTHOPAEDICA

69 %

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

Volume 92, Number 1, February 202

TF-IORT210007.indd 1

23-02-2021 11:43:00

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

E DITORIAL O F FICE

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

EDITOR

THE FOUNDATION BOARD OF

Anders Rydholm Lund, Sweden

THE NORDIC O RTHOPAEDIC F EDERATION AND A CTA O RTHOPAEDICA

DEPUTY EDITOR

Peter A Frandsen Odense, Denmark CO-EDITORS

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

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

WEB EDITOR

Magnus Tägil Lund, Sweden S TATISTICAL EDITOR

Jonas Ranstam Lund, Sweden P RODUCTION MANAGER

Kaj Knutson Lund, Sweden

Vol. 92, No. 1, 2021

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

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

Annual Institutional Subscription, Volume 91, 2020

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

$1,291

£798

€1,035

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

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

Copyright © 2021 The Author(s). Published by Taylor & Francis on behalf of the Nordic Orthopedic Federation. This is an Open Access article distributed under the terms of the Creative Commons Attribution-Non-Commercial License (https://creativecommons.org/licenses/by-nc/3.0 . Informa UK Limited, trading as Taylor & Francis Group makes every effort to ensure the accuracy of all the information (the “Content”) contained in its publications. However, Informa UK Limited, trading as Taylor & Francis Group, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Informa UK Limited, trading as Taylor & Francis Group. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Informa UK Limited, trading as Taylor & Francis Group shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. Terms & Conditions of access and use can be found at http://www.tandfonline. com/page/terms-and-conditions Indexed/abstracted in: Allied and Complementary Medicine Library (Amed); ASCA (Automatic Subject Citation Alert); Biological Abstracts; Chemical Abstracts; Cumulative Index to Nursing and Allied Health Literature(CINAHL); Current Advances in Ecological and Environmental Sciences; Current Contents/Clinical Medicine; Current Contents/Life Sciences; Developmental Medicine and Child Neurology; Energy Research Abstracts; EMBASE/ Excerpta Medica; Faxon Finder; Focus On: Sports Science & Medicine; Health Planning and Administration; Index Medicus/MEDLINE; Index to Dental Literature; Index Veterinarius; INIS Atomindex; Medical Documentation Service; Nuclear Science Abstracts (Ceased); Periodicals Scanned and Abstracted. Life Sciences Collection; Research Alert; Science Citation Index; SciSearch; SportSearch; Uncover Veterinary Bulletin. Printed in England by Henry Ling

Acta Orthopaedica

ISSN 1745-3674

Vol. 92, No. 1, February 2021 Annotation Antibiotics should not be used for back/leg pain

P Fritzell, T Bergström, B Jönsson, S G E Andersson, M Skorpil, P M Udby, M Andersen, and O Hägg

Back Back pain is also improved by lumbar disc herniation surgery

N Hareni, F Strömqvist, B Strömqvist, F G Sigmundsson, B E Rosengren, and M K Karlsson Y Wang, C Li, L Liu, and L Qi

Halo-pelvic traction for extreme lumbar kyphosis: 3 rare cases with a completely folded lumbar spine Hip 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 Custom-made 3D-printed cup-cage implants for complex acetabular revisions: evaluation of pre-planned versus achieved positioning and 1-year migration data in 10 patients The impact of socioeconomic status on the utilization of total hip arthroplasty during 1995–2017: 104,055 THA cases and 520,275 population controls from national databases in Denmark Low revision rate of dual mobility cups after arthroplasty for acute hip fractures: report of 11,857 hip fractures in the Dutch Arthroplasty Register (2007–2019) 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 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 Awareness of performance on outcomes after total hip and knee arthroplasty among Dutch orthopedic surgeons: how to improve feedback from arthroplasty registries Are functional outcomes and early pain affected by discharge on the day of surgery following total hip and knee arthroplasty? Unexpected varus deformity and concomitant metal ion release and MRI findings of modular-neck hip stems: descriptive RSA study in 75 hips with 8 years’ follow-up Knee Dutch Guideline on Knee Arthroscopy Part 1, the meniscus: a multidisciplinary review by the Dutch Orthopaedic Association Dutch Guideline on Knee Arthroscopy Part 2: non-meniscus intraarticular knee injury: a multidisciplinary review by the Dutch Orthopaedic Association Which Oxford Knee Score level represents a satisfactory symptom state after undergoing a total knee replacement? 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 Bariatric surgery prior to total knee arthroplasty is not associated with lower risk of revision: a register-based study of 441 patients Foot Ankle fracture classification using deep learning: automating detailed AO Foundation/Orthopedic Trauma Association (AO/OTA) 2018 malleolar fracture identification reaches a high degree of correct classification Patient-reported outcomes of joint-preserving surgery for moderate hallux rigidus: a 1-year follow-up of 296 patients from Swefoot

15 22

L N Van Steenbergen, K T Mäkelä, J Kärrholm, O Rolfson, S Overgaard, O Furnes, A B Pedersen, A Eskelinen, G Hallan, B W Schreurs, and R G H H Nelissen V Zampelis and G Flivik

N M Edwards, C Varnum, S Overgaard, and A B Pedersen

E M Bloemheuvel, L N Van Steenbergen, and B A Swierstra

O Wolf, S Mukka, J Ekelund, M Möller, and N P Hailer

A B Pedersen, A Mailhac, A Garland, S Overgaard, O Furnes, S A Lie, A M Fenstad, C Rogmark, J Kärrholm, O Rolfson, J Haapakoski, A Eskelinen, K T Mäkelä, and N P Hailer

P van Schie, L van Bodegom-Vos, T M Zijdeman, R G H H Nelissen, and P J Marang-van de Mheen

C E Husted, H Husted, L H Ingelsrud, C S Nielsen, A Troelsen, and K Gromov S Kiernan, B Kaptein, C Flivik, M Sundberg, and G Flivik

73 80 84 90 96

101

E R A van Arkel, S Koëter, P C Rijk, T G van Tienen, P W J Vincken, M J M Segers, B van Essen, N van Melick, and B H Stegeman S Koëter, T G van Tienen, P C Rijk, P W J Vincken, M J M Segers, B van Essen, N van Melick, B H Stegeman, and E R A van Arkel L H Ingelsrud, B Terluin, K Gromov, A Price, D Beard, and A Troelsen T Irmola, V Ponkilainen, K T Mäkelä, O Robertsson, A W-Dahl, O Furnes, A M Fenstad, A B Pedersen, H M Schrøder, A Eskelinen, and M J Niemeläinen P Ighani Arani, P Wretenberg, J Ottosson, O Robertsson, and A W-Dahl J Olczak, F Emilson, A Razavian, T Antonsson, A Stark, and M Gordon

108 M E Cöster, F Montgomery, and M C Cöster

Children Comparison of outcome between nonoperative and operative treatment of medial epicondyle fractures Systematic review of complications with externally controlled motorized intramedullary bone lengthening nails (FITBONE and PRECICE) in 983 segments Information to authors (see http://www.actaorthop.org/)

113

P Grahn, T Hämäläinen, Y Nietosvaara, and M Ahonen

119

M W Frost, O Rahbek, J Traerup, A A Ceccotti, and S Kold

Thank you, reviewers! Acta Orthopaedica thanks all reviewers for their work with manuscripts during year 2020. We depend on your work to keep and increase the quality of Acta. Your effort is much appreciated!

Adolfsson, Lars Evert Allington, Nanni Andersen, Thomas Andreou, Dimosthenis Audige, Laurent Aunan, Eirik Backteman, Torsten Bak, Klaus Baldini, Andrea Barker, Karen Berg, Hans E Bergh, Kåre Bernhoff, Karin Bieger, Ralf Bisseling, Pepijn Björnsson Hallgren, Hanna Cecilia Bobak, Peter Bodén, Henrik Bolder, Stefan Borris, Lars Bos, Pieter K Breusch, Steffen Brismar, Harald Brix, Michael Broström, Eva W Brouwer, Reinoud W Brüggemann, Anders Bråten, Martinus Buchholz, Ines Buddhdev, Pranai Burger, Bart Busch, André Buttaro, Martin Bülow, Erik Canavese, Federico Castagnetti, Marco Clarius, Michael Clauss, Martin Conway, Janet D Cordero-Ampuero, José Court-Brown, Charles Crawford, Scott Creutzfeldt, Johan Cöster, Maria C Dale, Håvard de Steiger, Richard N

Devane, Peter Dijkstra, Sander Dolatowski, Filip C Dreher, Thomas Drogset, Jon Olav Eastwood, Deborah M Egund, Niels Ekholm, Carl Ekström, Wilhelmina H G Emery, Roger John Engell, Vilhelm Englund, Martin Ericson, Mats Eriksson, Mikael Ernstbrunner, Lukas Escher, Cecilia Eskelinen, Antti Esteban, Jaime Farnebo, Simon Fernandes, Beatriz Luci Fjalestad, Tore Flivik, Gunnar Frihagen, Frede Fritz, Blaine Funk, Julia Furnes, Ove Försth, Peter Garland, Anne Gerdhem, Paul Gjertsen, Jan-Erik Gomez-Barrena, Enrique Gonzalez, Carlos Goosen, Jon Gordon, Max Gosselin, Richard A Gottliebsen, Martin Graham, Simon Grammatopoulos, George Graves, Stephen Ellis Grimm, Bermd Gromov, Kirill Guerra, Andrea Gupte, Chinmay Gustafson, Pelle Günther, Klaus-Peter Hallan, Geir

Hamilton, David F Hansen, Torben B Harris, Ian Hasan, Shaho Hays, Ron D Heckmann, Nathanael Hedin, Hanne Hedman, Leif Rune Hedström, Margareta Hellum, Christian Henricson, Anders Henriksson, Marketta Herzenberg, John Hiemstra, Laurie Anne Hindsø, Klaus Hing, Caroline Hjelholt, Thomas Johannesson Hogervorst, Tom Holen, Ketil Holla, Micha Hooper, Gary Horn, Joachim Houlihan-Burne, David Hulsbaek, Signe Husted, Henrik Hägglund, Gunnar Ilchmann, Thomas Inacio, Maria Carolina Ingelsrud, Lina Holm Itayem, R Jakobsen, Rune Jakobsen, Thomas Janssen, Dennis Jansson, Karl-Åke Johansson, Torsten Johnsen, Lars Gunnar Jozwiak, Marek Jämsen, Esa Jørgensen, Peter Holmberg Kalman, Sigridur Kamrad, Ilka Kaptein, Bart L Karlsson, Magnus K Kehlet, Henrik Keller, Johnny Kemp, Joanne L

Kjeldgaard Pedersen, Line Kjærsgaard-Andersen, Per Knutson, Kaj Koivu, Helka Kold, Søren Koster, Lennard Kralj-Iglic, Veronika Kristensen, Torbjørn Berge Kvederas, Giedrius Kvernmo, Hebe Désirée Kärrholm, Johan Laitinen, Minna K Lange, Jeppe Langvatn, Håkon Larsen, Morten Schultz Lauge-Pedersen, Henrik Launonen, Antti P Lavand–homme, Patricia Li, Yan Liavaag, Sigurd Liddle, Alexander David Lindström, Maria C Lohmander, Stefan Louwerens, Jan Willem Lundberg, Arne Maasalu, Katre Madanat, Rami Madsen, Jan Erik Magyar, Göran Marsell, Richard Mathijssen, Nina Mayr, Johannes Meermans, Geert Meurling, Lisbet Michno, Piotr Miettinen, Hannu Mikkelsen, Kim Lyngby Millis, Michael Brian Moeller-Madsen, Bjarne Mohaddes, Maziar Mukka, Sebastian Mulders, Marjolein A M Mäkelä, Keijo T Mäkinen, Tatu Märtson, Aare Möller, Michael Naylor, Justine M Nehrer, Stefan Nelissen, Rob Nemes, Szilard Nemeth, Banne Nerhus, Tor Kjetil Nestorson, Jens Nietosvaara, Yrjänä Nieuwenhuijse, Marc J

Niinimäki, Tuukka Nijhof, Marc Nilsson, Kjell G Nordsletten, Lars Olesen, Ulrik Kähler Olgun, Z Deniz Olsen, Bo Sanderhoff Ostendorf, Marieke Otten, Volker Thomas Christian Ovaska, Mikko Overgaard, Søren Parsch, Klaus Dieter Parvizi, Javad Pedersen, Alma B Pedersen, Niels Wisbech Pijls, Bart G Pitkänen, Mikko Ponzer, Sari Poolman, Rudolf W Pronk, Yvette Rahme, Hans Ranstam, Jonas Rasmussen, Jeppe Vejlgaard Rasmussen, Sten Reito, Aleksi Rice, Henry E Riddez, Louis Rogmark, Cecilia Rolf, Christer Rolfson, Ola Romijn, Marc Gerard Rosengren, Björn E D Rossvoll, Ivar Ruggieri, Pietro Rutz, Erich Ryd, Leif Ryge, Camilla Röhrl, Stephan Maximilian Rölfing, Jan Duedal Salomonsson, Björn Sayed-Noor, Arkan S Sayers, Adrian Scharff-Baauw, Marieke Schilcher, Jörg Schipper, Inger Schreurs, B. Willem Shrader, Wade M Sinikumpu, Jaakko Skogman, Roger Sköldenberg, Olof Smedby, Örjan Smolders, José M H Sousa, Ricardo Speranza, Domenico Stark, André

Stefánsdóttir, Anna Stillling, Maiken Stockley, Ian Sussman, Michael Svensson, Jonas Tarasevicius, Sarunas Terjesen, Terje Thorkildsen, Joachim Tiderius, Carl Johan Tjernström, Björn Harald Tootsi, Kaspar Trebse, Rihard Troelsen, Anders Tsagozis, Panagiotis Tsai, Jon A Tsikandylakis, Georgios Tsiridis, Eleftherios Tyson, Yosef Tönnesen, Hanne van de Bunt, Fabian van de Sande, M A J Van de Velde, Samuel Van der Heide, Huub J van Enst, Annefloor van Oldenrijk, Jakob van Raaij, Tom Marco van Susante, J L C Vehmeijer, Stephan Vendittoli, Pascal-André Verhaar, Jan A N Versteeg, Anne Virolainen, Petri Vogt, Björn von Schewelov, Thord Vos, Fidel W-Dahl, Annette Wadsten, Mats Wagner, Christof Wagner, Philippe Waldén, Markus Wartolowska, Karolina Welling, Maiju Wensaas, Anders Whitehouse, Michael Richard Whitehouse, Sarah Wik, Tina Strømdal Wilkinson, J. Mark Willigenburg, Nienke Wolf, Olof Wolterbeek, Nienke Wright, Timothy Wyat, Michael Charles Yu, Tao Zaltz, Ira Zijlstra, Wierd P

Acta Orthopaedica 2021; 92 (1): 1–3

Annotation

Antibiotics should not be used for back/leg pain Peter FRITZELL 1, Tomas BERGSTRÖM 2, Bodil JÖNSSON 3, Siv G E ANDERSSON 4, Mikael SKORPIL 5, Peter Muhareb UDBY 6, Mikkel ANDERSEN 7, and Olle HÄGG 8 1 RKC

Centre for Spine Surgery in Stockholm, Sweden/Futurum, Academy for Health and Care, Region Jönköping County, Sweden; 2 Department of Infectious Diseases, Institute of Biomedicine, University of Gothenburg, Sweden; 3 Sahlgrenska University Hospital, Göteborg, Sweden; 4 Department of Cell and Molecular Biology, Biomedical Center, Uppsala University, Sweden; 5 Karolinska University Hospital, Stockholm, Sweden; 6 Spine Unit, Ortopædkirurgisk Afdeling, Sjællands Universitetshospital, Køge, Denmark; 7 Spine Center of Southern Denmark, Lillebaelt Hospital, Middelfart, Denmark; 8 Spine Center Göteborg, Västra Frölunda, Sweden Correspondence: peterfritzell@gmail.com Submitted 2020-11-07. Accepted 2020-11-16

Antibiotics have been suggested as treatment for selected patients with chronic back pain, with or without leg pain, and in association with Modic changes type 1 (MC1) on MRI (Modic et al. 1988, Albert et al. 2013a). The hypothesis is that various bacteria, but above all the common skin bacterium Cutibacterium acnes (formerly Propionibacterium acnes), would spread hematogenously to degenerated discs, where a “low-grade subclinical infection” would trigger an inflammatory reaction and cause MC1 (Albert et al. 2013a), irritate nociceptive nerve endings, and induce pain. Other studies have not found this connection, or have been cautious in their conclusions (Birkenmaier 2013, Urquhart et al. 2015) but have not had the same public impact. Since back/leg pain is common, and Modic changes occur in people with or without back/leg pain (Wang et al. 2012), and as antibiotic resistance is a major health threat (Carlsson et al. 2019), the suggestion of treating chronic back/leg pain with antibiotics must be thoroughly investigated. Research groups in Sweden, Denmark, and Norway have independently conducted studies from 3 different perspectives, but with a focus on the same basic questions: is there a causative link between Modic changes, back pain, and bacteria. The studies, all published during 2019, conclude that antibiotics should not be used for back/leg pain, unless there is a clinically relevant infection in the disc/vertebra (discitis/ spondylitis). Bacteria and discs, Fritzell et al. (2019) In this Swedish multicenter study, the presence of bacteria in discs/vertebrae was evaluated in 2 patient groups from 7 hospitals (Figure 1). Samples from degenerated discs in 40 patients operated on for lumbar disc herniation (median age

LDH (n = 40)

Scoliosis (n = 20)

Bacteria in surrounding soft tissues and in disc/vertebra Bacteria only in surrounding soft tissues No bacteria in any samples Bacteria only in disc/vertebra 0

100

Distribution (%)

Figure 1. Data from Fritzell et al. 2019. LDH = lumbar disc herniation.

43 years) were compared with samples from non-degenerated discs in young patients operated on for scoliosis, who did not have back pain or disc herniation (median age 17 years). The samples were analyzed using culture and DNA technology at 2 independent university laboratories in Gothenburg and Uppsala. The study found no statistically significant difference in bacterial presence between the 2 groups. No association was seen between Modic changes and findings of bacteria or between back/leg pain and findings of bacteria or Modic changes. Conclusion Findings of bacteria, or traces of bacteria, as described in previous studies, are most likely a result of contamination during the surgical procedure. Modic changes and back pain, Udby et al. (2019) In Denmark, the clinical relevance of Modic changes for back pain and function was evaluated in a cohort study with a 13-year follow-up period (Figure 2). In all, 204 patients were

© 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group, on behalf of the Nordic Orthopedic Federation. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. DOI 10.1080/17453674.2020.1855561

Acta Orthopaedica 2021; 92 (1): 1–3

RMDQ score

With Modic changes

Without Modic changes p = 0.02

Treatment

Placebo Amoxicillin

20 16 12

8 5

Baseline 13-year follow-up

Weeks of follow-up

Figure 2. Data from Udby et al. 2019.

Figure 3. Data from Bråten et al. 2019.

stratified to a group with or without Modic changes. The relationship between Modic changes at study start and impaired physical function, measured with the RMDQ (Roland Morris Disability Questionnaire, 0 = best, 23 = worst possible function) (Roland et al. 1983), back pain, and sick leave at followup was evaluated. The study found no statistically significant differences in demographic data, BMI, smoking, back and leg pain at the start of the study, or at follow-up. Physical function measured with RMDQ was initially equal between the 2 groups, but statistically significantly better (without clinical relevance) at the follow-up in patients with Modic changes at the start of the study, 7.4 versus 9.6 (p = 0.02). Patients with Modic changes at the start of the study also had fewer sick days due to back pain during the study period, 9 versus 23.

Summary Based on current scientific evidence, where the 3 Nordic studies supply complementary findings, antibiotic treatment for back/leg pain with no signs of serious infection (discitis/spondylitis) cannot be recommended. It is important to counteract the development of antibiotic resistance in society due to antibiotic use without scientific evidence.

Conclusion Modic changes on MRI in patients with back pain at study start were not negatively associated with back pain or functional level after 13 years. Antibiotics and back pain, Bråten et al. (2019) In Norway, the Danish study by Albert et al. (2013b) was largely reproduced to evaluate the effect of antibiotic treatment in patients with chronic back pain (Figure 3). The research group conducted a randomized placebo-controlled doubleblind multicenter study at 6 hospitals (the Albert et al. study was a single-center study). The Norwegian study included 180 patients with back pain and Modic change type 1 or type 2. The participants were randomized to 3 months of treatment with either amoxicillin 3 × 750 mg or placebo, and follow-up was done according to “intention to treat” after 1 year. The outcome, measured with RMDQ, was similar in the 2 groups. Conclusion 3 months of treatment with amoxicillin showed no relevant clinical effect after 1 year in patients with chronic back pain and Modic changes.

This article has also been published in the Swedish Läkartidningen; 2020; 117: 20067. No author has any conflict of interest.

Albert H B, Lambert P, Rollason J, Solgaard Sorensen J, Worthington T, Bach Pedersen M, Schack Nørgaard H, Vernallis A, Busch F, Manniche C, Elliott T. Does nuclear tissue infected with bacteria following disc herniations lead to Modic changes in the adjacent vertebrae? Eur Spine J 2013a; 22(4): 690-6. Albert H B, Sorensen J S, Schiott Christensen B, Manniche C. Antibiotic treatment in patients with chronic low back pain and vertebral bone oedema (Modic type 1 changes): a double-blind randomized clinical controlled trial of efficacy. Eur Spine J 2013b; 22(4): 697-707. Birkenmaier C. Should we start treating chronic low back pain with antibiotics rather than with pain medications? Korean J Pain 2013; 26(4): 327-35. Bråten L C H, Rolfsen M P, Espeland A, Wigemyr M, Aßmus J, Froholdt A, Haugen A J, Marchand G H, Kristoffersen P M, Lutro O, Randen S, Wilhelmsen M, Winsvold B S, Kadar T I, Holmgard T E, Vigeland M D, Vetti N, Nygaard Ø P, Lie B A, Hellum C, Anke A, Grotle M, Schistad E I, Skouen J S, Grøvle L, Brox J I, Zwart J A, Storheim K; AIM study group. Efficacy of antibiotic treatment in patients with chronic low back pain and Modic changes (the AIM study): double blind, randomised, placebo controlled, multicentre trial. BMJ 2019; 367: 15654. Carlsson F, Jacobsson G, Jagers S C, Lampi E, Robertson F, Rönnerstrand B. Who is willing to stay sick for the collective? Individual characteristics, experience, and trust. Göteborg: Göteborgs universitet, institutionen för samhällsekonomi med statistik; 2019. SSM Popul Health 2019; 9: 100499. Fritzell P, Welinder-Olsson C, Jönsson B, Melhus Å, Andersson S G E, Bergström T, Tropp H, Gerdhem P, Hägg O, Laestande H, Knutsson B, Lundin A, Ekman P, Rydman E, Skorpil M. Bacteria: back pain, leg pain and Modic sign—a surgical multicentre comparative study. Eur Spine J 2019; 28(12): 2981-9.

Acta Orthopaedica 2021; 92 (1): 1–3

Modic M T, Steinberg P M, Ross J S, Masaryk T J, Carter J R. Degenerative disk disease: assessment of changes in vertebral body marrow with MR imaging. Radiology 1988; 166(1, Pt 1): 193-9. Roland M O, Morris R W. A study of the natural history of back pain. Part 1: Development of a reliable and sensitive measure of disability in low back pain. Spine (Phila Pa 1976) 1983; 8(2): 141-4. Udby P M, Bendix T, Ohrt-Nissen S, Lassen M R, Sørensen J S, Brorson S, Carreon L Y, Andersen M Ø. Modic changes are not associated with

3 long-term pain and disability: a cohort study with 13-year follow-up. Spine (Phila Pa 1976) 2019; 44(17): 1186-92. Urquhart D M, Zheng Y, Cheng A C, Rosenfeld J V, Chan P, Liew S, Monira Hussain S, Cicuttini F M. Could low grade bacterial infection contribute to low back pain? A systematic review. BMC Med 2015; 13: 13. Wang Y, Videman T, Battié M C. Modic changes: prevalence, distribution patterns and association with age in white men. Spine J 2012; 12(5): 411-16.

Acta Orthopaedica 2021; 92 (1): 4–8

Back pain is also improved by lumbar disc herniation surgery Niyaz HARENI 1,2, Fredrik STRÖMQVIST 2, Björn STRÖMQVIST 2, Freyr Gauti SIGMUNDSSON 3, Björn E ROSENGREN 2, and Magnus K KARLSSON 2 1 Department of Orthopaedics, Varberg Hospital, Varberg; 2 Departments of Clinical Sciences and Orthopedics, Lund University, Skåne University Hospital, Malmö, Sweden; 3 Department of Orthopedics, School of Medical Sciences, Faculty of Medicine and Health, Örebro University, Örebro, Sweden Correspondence: Niyaz.Hareni@regionhalland.se Submitted 2020-02-16. Accepted 2020-07-26.

Background and purpose — Indication for lumbar disc herniation (LDH) surgery is usually to relieve sciatica. We evaluated whether back pain also decreases after LDH surgery. Patients and methods — In the Swedish register for spinal surgery (SweSpine) we identified 14,097 patients aged 20–64 years, with pre- and postoperative data, who in 2000– 2016 had LDH surgery. We calculated 1-year improvement on numeric rating scale (rating 0–10) in back pain (Nback) and leg pain (Nleg) and by negative binomial regression relative risk (RR) for gaining improvement exceeding minimum clinically important difference (MCID). Results — Nleg was preoperatively (mean [SD]) 6.7 (2.5) and Nback was 4.7 (2.9) (p < 0.001). Surgery reduced Nleg by mean 4.5 (95% CI 4.5–4.6) and Nback by 2.2 (CI 2.1–2.2). Mean reduction in Nleg) was 67% and in Nback 47% (p < 0.001). Among patients with preoperative pain ≥ MCID (that is, patients with significant baseline pain and with a theoretical possibility to improve above MCID), the proportion who reached improvement ≥ MCID was 79% in Nleg and 60% in Nback. RR for gaining improvement ≥ MCID in smokers compared with non-smokers was for Nleg 0.9 (CI 0.8–0.9) and Nback 0.9 (CI 0.8–0.9), and in patients with preoperative duration of back pain 0–3 months compared with > 24 months for Nleg 1.3 (CI 1.2–1.5) and for Nback 1.4 (CI 1.2– 1.5). Interpretation — LDH surgery improves leg pain more than back pain; nevertheless, 60% of the patients with significant back pain improved ≥ MCID. Smoking and long duration of pain is associated with inferior recovery in both Nleg and Nback.

The most common indication for lumbar disc herniation (LDH) surgery is persistent sciatica that does not respond to nonoperative treatment (Blamoutier 2013). However, most patients who undergo LDH surgery also suffer from back pain (Hakkinen et al. 2003, Stromqvist et al. 2017), on a national level reported in 93% of patients having LDH surgery (Stromqvist et al. 2017). Decades ago, Mixter (1937) therefore argued that LDH extirpation should be accompanied by fusion to minimize postoperative back pain. Recent studies have opposed this view, showing that LDH surgery is not followed by increased back pain when only removing the hernia (Pearson et al. 2008, Owens et al. 2018), and in many cases even improvement of back pain seems sustainable over time. Most studies that evaluate the outcome of LDH surgery focus on the relief from sciatica and improvement in patientreported outcome measures (PROMs) (Weber 1983, Atlas et al. 2005, Peul et al. 2007, Weinstein et al. 2008, Lurie et al. 2014). A few studies have focused on back pain or included back pain in the evaluation (Kotilainen et al. 1993, Hakkinen et al. 2003, Toyone et al. 2004, Atlas et al. 2005, Pearson et al. 2008, Owens et al. 2018). While some of these infer that back pain is improved by the LDH surgery (Hakkinen et al. 2003, Toyone et al. 2004, Pearson et al. 2008, Owens et al. 2018) others report inconclusive results (Kotilainen et al. 1993, Atlas et al. 2005). There is a lack of consensus on the expected level of back pain reduction with LDH surgery. It would also be of clinical interest to identify preoperative factors that are associated with favorable reduction of back pain following LDH surgery such as age, sex, smoking, preoperative health, and duration of pain (Nygaard et al. 2000, Jansson et al. 2005, Stromqvist et al. 2016, Wilson et al. 2016, Hareni et al. 2019).

Acta Orthopaedica 2021; 92 (1): 4–8

We (i) evaluated whether back pain is reduced after LDH surgery and if so, to what extent compared with the reduction in leg pain and (ii) what proportion of patients gain improvement in back and leg pain exceeding minimum clinically important difference (MCID). The secondary aim was to identify factors associated with improvement in back pain exceeding MCID.

Patients and methods Patient data was collected from the Swedish Spine Register (SweSpine), which is a patient register with prospectively collected data. The register covers 98% of all clinics performing lower back surgery in Sweden and has a completeness of 75% (www.swespine.se). In the register the patient reports preoperative anthropometric, lifestyle, and disease-related data such as age, gender, smoking habits (yes/no), numeric rating scale (NRS) for pain in the back (Nback) and leg (Nleg), duration of pain symptoms (categorized as no pain, pain 0–3 months, 3–12 months, 12–24 months, and > 24 months) and the PROM Short Form Health Survey 36 (SF-36) (rating from 0 to 100). The outcome after the operation is evaluated after 1 year by a similar questionnaire including current NRS pain level in back and leg and SF-36. The surgeon reports data concerning diagnosis, procedure, level of surgery, side of operation, and periand postoperative complications. SweSpine has previously been described in detail, including validation, with adequate results (Stromqvist et al. 2009). We identified in SweSpine 19,815 patients aged 20–64 years during 2000–2016 with the diagnosis LDH and with baseline NRS back pain data. This age-span was chosen to include the typical LDH patient, albeit minimizing the risk of wrongful selection (that is, elderly patients with LDH diagnosis but also variable degree of spinal stenosis). The included patients had undergone open discectomy with or without microscope (87%), decompression with or without microscope (6%), various other types of surgeries (6%), or with type of surgery not reported (0.4%). 5,718/19,815 patients had not responded with postoperative NRS back pain data and were therefore excluded (Table 1). Patients included in this report had complete pre- and 1-year postoperative data for age, sex, and NRS back pain. All other included variables had above 96% response rates. Statistics IBM SPSS Statistics version 26 (IBM Corp, Armonk, NY, USA) was used for statistical analysis. Descriptive data are presented as numbers and means with standard deviations (SD) and inferential statistics as proportions (%) or means with 95% confidence intervals (CI). For group comparisons we used a paired Student’s t-test between means for continuous data and a chi-square test for categorical data. MCID was defined as an improvement by at least 2.5 units in NRS back

Table 1. Preoperative data in patients with both pre- and 1-year postoperative numeric rating scale (NRS) back pain data (n = 14,097) and in those with missing 1-year NRS back pain data (n = 5,718). Data are presented as numbers (n), proportions (%), or mean (standard deviation) Patients with pre- and Patients with postoperative incomplete NRS back NRS back Factor n pain data pain data Mean age (SD) 14,097 43 (11) Male sex (%) 14,097 55 Smokers (%) 13,975 19 Short Form-36 (SD) 13,930 47 (28) NRS back pain (SD) 14,097 4.7 (2.9) NRS leg pain (SD) 14,064 6.7 (2.5) Duration of back pain (%) 13,191 0–3 months 13 3–12 months 50 12–24 months 17 > 24 months 20 Duration of leg pain (%) 13,877 0–3 months 17 3–12 months 54 12–24 months 15 > 24 months 14

41 (10) 61 25 45 (27) 5.0 (2.9) 6.7 (2.5) 11 49 17 23 15 53 16 15

pain and 3.5 units in NRS leg pain (Solberg et al. 2013). We used negative binomial regression to determine adjusted relative risk (RR) for preoperative factors that are associated with pain reduction ≥ MCID. We selected factors that in a previous publication have been found to be associated with general outcome in LDH surgery (Wilson et al. 2016). These variables included age, sex, smoking habit, quality of life (Short Form36), and preoperative duration of pain. As binary dependent variable we dichotomized improvement in pain with improvement ≥ MCID being regarded as a successful and < MCID as an unsuccessful outcome. Furthermore, so as to be included in the binomial regression analyses, the patients had to have pain exceeding MCID at baseline (that is, having a hypothetical possibility to improve ≥ MCID). We regarded a p-value below 0.05 to indicate a statistically significant difference. Ethics, data sharing plan, funding, and potential conflicts of interest The study was approved by the Lund regional ethical review board (Dnr 2017/158). Data sharing plan: the data is available from SweSpine upon request and approval by the registry board. No specific funding has been received for this study. No conflict of interest was declared.

Results Nleg was before surgery (mean [SD]) 6.7 (2.5) and Nback 4.7 (2.9) (p < 0.001), and 1 year after surgery 2.1 (2.7) and 2.5 (2.7) (p < 0.001) (Table 2). This corresponds to a reduction in

Acta Orthopaedica 2021; 92 (1): 4–8

Table 2. Numeric rating scale 0–10 for back pain (Nback) and leg pain (Nleg) at baseline, 1 year after surgery, changes by surgery, and the proportion of patients with improvement ≥ MCID in pain (≥ 2.5 for Nback and ≥ 3.5 for Nleg) Proportion with Mean NRS (SD) Changes in NRS (CI) improvement Factor n preoperatively at 1 year absolute relative (%) ≥ MCID (%) p-value All patients Nback 14,097 4.7 (2.9) 2.5 (2.7) –2.2 (–2.1 to –2.2) –47 (–45 to –47) 43 < 0.001 Nleg 14,023 6.7 (2.5) 2.1 (2.7) –4.5 (–4.5 to –4.6) –67 (–67 to –69) 71 Patients with preoperative back and/or leg pain ≥ MCID Nback 9,975 6.2 (2.0) 3.0 (2.8) –3.2 (–3.1 to –3.3) –52 (–50 to –53) 60 < 0.001 Nleg 12,268 7.4 (1.7) 2.2 (2.8) –5.1 (–5.1 to –5.2) –69 (–69 to –70) 79 n — the number of patients who had answered this specific question; SD —standard deviation; CI — 95% confidence intervals.

Table 3. Negative binomial regression model presenting adjusted relative risk (RR) for different preoperative factors in respect of reaching improvement ≥ MCID in back pain in 9,674 patients with complete data

Table 4. Negative binomial regression model presenting adjusted relative risk (RR) for different preoperative factors in respect of reaching improvement ≥ MCID in leg pain in 11,957 patients with complete data

Factor

Mean RR (95% CI)

Factor

Mean RR (95% CI)

Age Male (reference: female) Smoker (reference: non-smoker) Short Form-36 Duration of back pain (reference: > 24 months) 0–3 months 3–12 months 12–24 months

1.0 (0.99–1.0) 0.97 (0.91–1.0) 0.9 (0.8–0.9) a 1.0 (0.99–1.0)

Age Male (reference: female) Smoker (reference: non-smoker) Short Form-36 Duration of back pain (reference: > 24 months) 0–3 months 3–12 months 12–24 months

1.0 (1.0–1.0) a 0.95 (0.90–1.0) 0.9 (0.8–0.95) a 1.0 (1.0–1.0)

1.4 (1.3–1.6) a 1.3 (1.2–1.4) a 1.1 (1.0–1.3) a

To be included in this model, patients had to have Nback ≥ 2.5 at baseline. For continuous variables results are presented as RR per unit of change in the independent variable and for categorical variables compared with a reference category. a Statistically significant.

Nleg by 4.5 (CI 4.5–4.6) and Nback by 2.2 (2.1–2.2). The relative Nleg reduction was 67% and relative Nback reduction 47% (Table 2). The proportion of patients who reached improvement ≥ MCID was 71% for leg pain and 43% for back pain (p < 0.001). When only including patients with pain ≥ MCID (3.5 for Nleg and 2.5 for Nback), that is, patients with significant baseline pain and with hypothetical possibility to improve ≥ MCID, 79% of the patients improved ≥ MCID in leg pain and 60% in back pain (p < 0.001) (Table 2). Smokers had, compared with non-smokers, RR of gaining improvement ≥ MCID in Nleg of 0.9 (CI 0.8–0.9) and in Nback of 0.9 (0.8–0.95). Older age (per year increment) had RR of gaining improvement ≥ MCID in Nleg of 0.995 (0.993– 0.998) and in Nback of 1.00 (0.99–1.00). Duration of symptom 0–3 months had, compared to > 24 months, RR of gaining improvement ≥ MCID in Nleg 1.3 (1.2–1.5) and for Nback 1.4 (1.2–1.5), and 3–12 months compared with > 24 months for Nleg 1.2 (1.1–1.3) and for Nback 1.3 (1.2–1.4). Sex and quality of life by SF-36 was not associated with either outcome (Tables 3 and 4).

1.3 (1.2–1.5) a 1.2 (1.1–1.3) a 1.1 (0.99–1.2)

To be included in this model, patients had to have Nleg ≥ 3.5 at baseline. For continuous variables results are presented as RR per unit of change in the independent variable and for categorical variables compared with a reference category. a Statistically significant.

Discussion We found that both leg and back pain was reduced by LDH surgery, leg pain more than back pain, and that as many as 60% of the patients with back pain ≥ MCID reached back pain reduction equal to or above this level. Furthermore, non-smokers and patients with shorter duration of preoperative pain had a greater probability to reach leg and back pain reduction ≥ MCID. Younger age was associated only with leg pain reduction ≥ MCID. We also found that 79% of the patients with preoperative leg pain ≥ MCID had improvement defined as a clinically successful outcome, a success rate comparable to data in the literature (Solberg et al. 2013). By the same logic, the proportion of patients who achieved clinically successful back pain reduction was lower than the proportion with clinically successful leg pain reduction, but this was still considerable as back pain is not generally regarded as an indication for LDH surgery. The explanatory mechanism behind this back-pain reduction after LDH surgery is not clear (Peng et al. 2005, Yang et al. 2015).

Acta Orthopaedica 2021; 92 (1): 4–8

Even though this study cannot conclude optimal timing for surgery, there was an association between shorter duration of symptoms (0–3 and 3–12 months compared with > 24months) and successful outcome. Peul et al. (2007) have found a similar 1-year outcome for those operated on early (6–12 weeks’ duration) and those who waited another 6 months. We highlight that our study design could draw inferences only as regards associations. Our results do not motivate LDH surgery on the basis of back pain. We can only use these data to inform patients scheduled for LDH surgery that they have a probability of 60% of having a reduction of clinical significance if they have back pain ≥ MCID. Our view is supported by another registry study that included 2,262 patients, which reported that patients with baseline back pain NRS ≥ 5 (out of maximum 10), also had significant improvement in the back pain by discectomy (Owens et al. 2018). However, that study did not evaluate the proportion of patients who reached the MCID level of pain reduction. Further studies should examine whether it is possible to identify sub-groups of patients that will specifically benefit from back-pain reduction by discectomy. Strengths in our study include a large study population, prospectively collected data, and reports of outcome on a national level that identify the results in the general health care system rather than in highly specialized spinal units. Another strength is the absence of exclusion criteria, rendering the actual outcome within general health care, in which patients with comorbidities and relative contraindications for surgery are included. This should be compared with studies that show what it is possible to achieve in selected defined patient cohorts in specialized units with highly trained surgeons (Kotilainen et al. 1993, Staartjes et al. 2019). Limitations include those unavoidable in registry-based studies, such as incomplete pre- and postoperative data collection. These limitations have not, however, biased the outcome effects (Solberg et al. 2011). Another weakness is the inability to adjust for all possible confounders, such as radiological evaluations to assess findings that have been reported in the literature to be associated with back pain (Yang et al. 2015) and data on more possible confounders to include in our model. In conclusion, both leg and back pain improve after LDH surgery, leg pain more than back pain. In patients with preoperative pain ≥ MCID, leg pain level was reduced at this level or above in 79% of the patients and back pain in 60%. The preoperative factors that in our model were associated with back pain reduction ≥ MCID by LDH surgery were virtually the same as those associated with leg pain reduction ≥ MCID. Our results improve the ability to provide accurate preoperative information as regards the probability of reaching clinically significant reduction in back and leg pain by LDH surgery. MK, FS and NH proposed the study. NH wrote the manuscript and did the statistical calculations under guidance from FS, BS, BR and MK. All authors revised the manuscript and contributed to the final manuscript.

Acta thanks J L C van Susante for help with peer review of this study.

Atlas S J, Keller R B, Wu Y A, Deyo R A, Singer D E. Long-term outcomes of surgical and nonsurgical management of sciatica secondary to a lumbar disc herniation: 10 year results from the Maine Lumbar Spine Study. Spine 2005; 30(8): 927-35. Blamoutier A. Surgical discectomy for lumbar disc herniation: surgical techniques. Orthop Traumatol Surg Res 2013; 99(1 Suppl.): S187-96. doi: 10.1016/j.otsr.2012.11.005. Hakkinen A, Ylinen J, Kautiainen H, Airaksinen O, Herno A, Tarvainen U, Kiviranta I. Pain, trunk muscle strength, spine mobility and disability following lumbar disc surgery. J Rehabil Med 2003; 35(5): 236-40. Hareni N, Stromqvist F, Stromqvist B, Rosengren B E, Karlsson M K. Predictors of satisfaction after lumbar disc herniation surgery in elderly. BMC Musculoskelet Disord 2019; 20(1): 594. doi: 10.1186/s12891-0192975-4. Jansson K A, Nemeth G, Granath F, Jonsson B, Blomqvist P. Health-related quality of life in patients before and after surgery for a herniated lumbar disc. J Bone Joint Surg Br 2005; 87(7): 959-64. doi: 10.1302/0301620X.87B7.16240. 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. Lurie J D, Tosteson T D, Tosteson A N, Zhao W, Morgan T S, Abdu W A, Herkowitz H, Weinstein J N. Surgical versus nonoperative treatment for lumbar disc herniation: eight-year results for the Spine Patient Outcomes Research Trial. Spine 2014; 39(1): 3-16. doi: 10.1097/ BRS.0000000000000088. Mixter W J. Rupture of the lumbar intervertebral disk: an etiologic factor for so-called “sciatic” pain. Ann Surg 1937; 106(4): 777-87. 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; 92(2 Suppl.): 131-4. Owens R K 2nd, Carreon L Y, Bisson E F, Bydon M, Potts E A, Glassman S D. Back pain improves significantly following discectomy for lumbar disc herniation. Spine J 2018; 18(9): 1632-6. doi: 10.1016/j. spinee.2018.02.014. Pearson A M, Blood E A, Frymoyer J W, Herkowitz H, Abdu W A, Woodward R, Longley M, Emery S E, Lurie J D, Tosteson T D, Weinstein J N. SPORT lumbar intervertebral disk herniation and back pain: does treatment, location, or morphology matter? Spine 2008; 33(4): 428-35. doi: 10.1097/ BRS.0b013e31816469de. Peng B, Wu W, Hou S, Li P, Zhang C, Yang Y. The pathogenesis of discogenic low back pain. J Bone Joint Surg Br 2005; 87(1): 62-7. Peul W C, van Houwelingen H C, van den Hout W B, Brand R, Eekhof J A, Tans J T, Thomeer R T, Koes BW, Leiden—The Hague Spine Intervention Prognostic Study G. Surgery versus prolonged conservative treatment for sciatica. N Engl J Med 2007; 356(22): 2245-56. doi: 10.1056/ NEJMoa064039. Solberg T K, Sorlie A, Sjaavik K, Nygaard O P, Ingebrigtsen T. Would loss to follow-up bias the outcome evaluation of patients operated for degenerative disorders of the lumbar spine? Acta Orthop 2011; 82(1): 56-63. doi: 10.3109/17453674.2010.548024. Solberg T, Johnsen L G, Nygaard O P, Grotle M. Can we define success criteria for lumbar disc surgery?: estimates for a substantial amount of improvement in core outcome measures. Acta Orthop 2013; 84(2): 196-201. doi: 10.3109/17453674.2013.786634. Staartjes V E, de Wispelaere M P, Vandertop W P, Schroder M L. Deep learning-based preoperative predictive analytics for patient-reported outcomes following lumbar discectomy: feasibility of center-specific modeling. Spine J 2019; 19(5): 853-61. doi: 10.1016/j.spinee.2018.11.009. Stromqvist B, Fritzell P, Hagg O, Jonsson B, Swedish Society of Spinal S. The Swedish Spine Register: development, design and utility. Eur Spine J 2009; 18(Suppl. 3): 294-304. doi: 10.1007/s00586-009-1043-4.

Stromqvist F, Stromqvist B, Jonsson B, Karlsson M K. The outcome of lumbar disc herniation surgery is worse in old adults than in young adults. Acta Orthop 2016; 87(5): 516-21. doi: 10.1080/17453674.2016. 1205173. Stromqvist F, Stromqvist B, Jonsson B, Karlsson M K. Surgical treatment of lumbar disc herniation in different ages: evaluation of 11,237 patients. Spine J 2017; 17(11): 1577-85. doi: 10.1016/j.spinee.2017.03.013. Toyone T, Tanaka T, Kato D, Kaneyama R. Low-back pain following surgery for lumbar disc herniation: a prospective study. J Bone Joint Surg Am 2004; 86-A(5): 893-6. Weber H. Lumbar disc herniation. A controlled, prospective study with ten years of observation. Spine (Phila Pa 1976) 1983; 8(2): 131-40.

Acta Orthopaedica 2021; 92 (1): 4–8

Weinstein J N, Lurie J D, Tosteson T D, Tosteson A N, Blood E A, Abdu W A, Herkowitz H, Hilibrand A, Albert T, Fischgrund J. Surgical versus nonoperative treatment for lumbar disc herniation: four-year results for the Spine Patient Outcomes Research Trial (SPORT). Spine 2008; 33(25): 2789-800. doi: 10.1097/BRS.0b013e31818ed8f4. Wilson C A, Roffey D M, Chow D, Alkherayf F, Wai E K. A systematic review of preoperative predictors for postoperative clinical outcomes following lumbar discectomy. Spine J 2016; 16(11): 1413-22. doi: 10.1016/j. spinee.2016.08.003. Yang H, Liu H, Li Z, Zhang K, Wang J, Wang H, Zheng Z. Low back pain associated with lumbar disc herniation: role of moderately degenerative disc and annulus fibrous tears. Int J Clin Exp Med 2015; 8(2): 1634-44.

Acta Orthopaedica 2021; 92 (1): 9–14

Halo-pelvic traction for extreme lumbar kyphosis: 3 rare cases with a completely folded lumbar spine Yu WANG, Chunde LI, Long LIU, and Longtao QI Department of Orthopaedics, Peking University First Hospital, Xicheng District, Beijing 100034, China Correspondence: Longtao Qi, 06340@pkufh.com Submitted 2020-05-15. Accepted 2020-08-26.

We present 3 patients who had extreme lumbar kyphosis and were treated with halo-pelvic traction (HPT) and anteroposterior fixation. In addition, we also describe a modified halopelvic apparatus that helps to relieve the discomfort and inconvenience of HPT. Preoperative and postoperative data are summarized in Table 1, and radiographic measurements are given in Table 2. Case 1 A 30-year-old female presented with a 10-year history of leg numbness and urinary incontinence (Figure 1). Physical examination showed a short trunk, and a bump in her lower back. CT revealed a congenital malformation in the lumbar region, and the lumbar spine appeared as if it was completely folded up. Considering the severity and complexity of the deformity, we decided to treat her in stages. The first-stage operation consisted of posterior release and fixation of the halo-pelvic apparatus. Posterior release included spinous process resection and laminectomy. For the first week of HPT, the halo-pelvic frame was distracted at a rate of 1 cm per day. Beginning in the second week, the distraction rate was decreased to 0.5 cm per day. Notably, the patient described a sudden feeling of a “crack” on day 7 of traction. A lateral radiograph revealed evidence that her lumbar spine was “unfolding.” The second-stage operation was performed after 10 weeks of HPT. The procedure consisted of a vertebral osteotomy and pedicle screw fixation via a posterior approach, and mesh insertion via an anterior approach. She was 8 cm taller after surgery than before traction. By 6 months after the operation her leg numbness and urinary incontinence were relieved completely. No neurological or implant-related complications were noted over a 2-year follow-up period.

Case 2 A 40-year-old female presented with numbness and weakness of both lower extremities. She also complained of chronic low back pain, which she stated significantly impaired her quality of life. She had been diagnosed with spinal tuberculosis when she was 9 years old. CT revealed angular kyphosis in the lumbar region, almost complete absence of the vertebral bodies of L2 to L4, and complete fusion of the laminas of L1 to L5 (Figure 2). Her management consisted of 2 stages. The first-stage operation consisted of posterior release and fixation of the halo-pelvic apparatus. HPT was applied for 6 weeks, and the second-stage operation was then performed. This procedure consisted of anterior instrumentation and posterior fusion (Figure 3). She was 6 cm taller after surgery than before traction. During the 1st year after the second procedure the weakness and numbness of her lower extremities gradually diminished, and her low back pain was relieved. No neu-

Table 1. Intraoperative and postoperative results Estimated Height Case Age/ Levels Days of blood Operative increase no. sex fused traction loss (L) time (h) (cm)

Complications

1 2 3

Dural tear None None

30/F T10–S2 70 40/F T8–S1 42 23/F T10–S2 46

1.6 1.1 0.4

8.1 6.6 7.4

8 6 6

Table 2. Radiographic follow-up Case no.

3 Mean

Global kyphosis (°) Before traction 180 133 112 143 After traction 137 82 57 92 After lumbar surgery 88 18 56 54 2 years after surgery 84 20 55 50 2-year correction (%) 54 85 53 64 C7-sagittal vertical axis (mm) Before traction 74 148 65 96 After traction –35 48 –18 –2 After lumbar surgery 36 39 –15 20 2 years after surgery 36 41 –12 21 Pelvic tilt (°) Before traction 48 47 35 43 After traction 11 21 26 20 After lumbar surgery 21 10 33 21 2 years after surgery 20 15 24 29 Sacral slope (°) Before traction –37 –23 –23 –27 After traction –14 –2 –18 –11 After lumbar surgery –25 15 –20 –10 2 years after surgery –22 16 –10 –5 Pelvic incidence (°) Before traction 7 23 13 14 After traction 8 20 9 12 After lumbar surgery 7 25 13 15 2 years after surgery 7 23 12 14

Preoperative

Acta Orthopaedica 2021; 92 (1): 9–14

Traction

Postoperative

Preoperative

Traction

Postoperative

Figure 1. Case 1: Computed tomography (CT) revealed a congenital malformation in the lumbar region, and the lumbar spine appeared as if it was “completely folded up.” Treatment included HPT and anteroposterior fixation.

rological or implant-related complications were noted over a 2-year follow-up period. Case 3 A 23-year-old female presented with chronic low back pain. She had been diagnosed with spinal tuberculosis when she was 2 years old. Physical examination showed a short trunk, and a bump in her lower back. CT revealed angular lumbar kyphosis, vertebral subluxation, absent vertebral bodies from L2 to L4 (presumably due to tuberculosis), and fusion of the L1–L5 laminas (Figure 4). Stage treatment was performed, similar to that used in Case 1 and Case 2. She began mobilization on day 5 after surgery, and was discharged to home on day 14 after surgery. She was 6 cm taller after surgery than before traction. She returned to work 6 months after surgery, and at that time her low back pain was relieved.

Discussion Spinal kyphosis is commonly due to ankylosing spondylitis,

fractures, spinal tuberculosis, Scheuermann’s disease, degenerative spine disorders, and congenital deformities. Kyphotic deformities that require surgical treatment primarily occur in the thoracic, thoracolumbar, or lumbosacral region. In our experience, severe lumbar kyphosis is rare. HPT was developed in the late 1960s by O’Brien et al. (1971). The powerful distraction forces generated by HPT can effectively correct various spinal deformities. The method was used extensively in the 1970s. However, its use began to decline after the 1980s because of its many drawbacks, including a long period of hospitalization, a multi-stage process, pain, discomfort, unflattering appearance, and various complications (Ransford and Manning 1975). In addition, advances in surgical techniques and methods of internal fixation rendered HPT obsolete. Currently, most spinal deformities are treated by osteotomy and internal fixation (Suk et al. 2005, Lenke et al. 2010). However, treatment of extreme spinal deformities is challenging because a complex osteotomy and large-scale correction is required, and this significantly increases the incidence of neurological, vascular, and pulmonary complications. In addition,

Acta Orthopaedica 2021; 92 (1): 9–14

Preoperative

Traction

Preoperative

Traction Postoperative

Figure 2. Case 2: CT revealed angular kyphosis in the lumbar region, almost complete absence of the vertebral bodies of L2 to L4, and complete fusion of the laminas of L1 to L5. Management consisted of HPT and anterior instrumentation and posterior fusion.

internal spinal stabilization can be difficult under some situations, such as severe rigid scoliosis, poor bone quality, revision surgery, respiratory dysfunction, and previously existing neurological deficits (Miladi 2013). HPT has been shown to increase the safety and correction rate for the surgical management of severe spinal deformities (Sponseller et al. 2008, Kim et al. 2009). Herein, we report the treatment of 3 patients from 2017 with severe lumbar kyphosis. The patients were treated with HPT and anteroposterior fusion with good results, and follow-up periods of 2 years. Why was HPT necessary for the 3 patients? Currently, spinal kyphosis is primarily treated with osteotomy and posterior fusion (Wang et al. 2015, Zhou et al. 2018, Hua et al. 2019, Zhang et al. 2019). However, performing a vertebral osteotomy via a posterior approach is difficult when a kyphosis is “folded” or the lumbar spine is collapsed. The procedure is difficult because all the nerve roots are compressed in a very small region of the posterior lumbar spine. The nerve roots block access for performing the vertebral osteotomy and mesh

insertion; indeed, it is difficult to perform a vertebral osteotomy and mesh insertion through the limited space between any 2 lumbar nerve roots. Furthermore, the lumbar nerve roots are not as dispensable as some thoracic spine nerve roots. This problem can be solved by HPT, application of which spreads the nerve roots apart making it easier to perform a posterior osteotomy (Yu et al. 2016). Another reason why HPT was necessary in these 3 patients is because there were few anchor points in the distal region of the implant construct. The lack of anchor points limits the corrective force of instrumentation. In all 3 patients, only 3 segments (L5 to S2) were available for screw insertion on the caudal side. As such, correction of the kyphosis with HPT helped to achieve better overall correction and prevent screw pull-out. The deformities in the 3 patients presented were severe and complex, and direct correction would have been associated with significant risks. On the other hand, the gradual application of HPT for 6–10 weeks allowed reduction of the kyphosis, which decreased the risk of surgical complications. Furthermore, lengthening of the HPT frame was done with

Acta Orthopaedica 2021; 92 (1): 9–14

Anterior approach

Posterior approach

Figure 3. Case 2: Osteotomy and pedicle screw fixation were performed via a posterior approach and a mesh was inserted via an anterior approach during a single procedure.

the patients conscious; therefore, any neurological deficits occurring as a result of the HPT would have been identified immediately. Lastly, the modified halo-pelvic device allowed the patients to move and lie more comfortably than if a traditional device had been used.

traction osteotomy (PSO) with a minimum 2-year follow-up was a PT < 24°. The PTs of our 3 patients presented here at 2-year follow-up were 20°, 15°, and 24°, respectively, which can be considered satisfactory outcomes according to the aforementioned criteria.

How effective is presurgical HPT? In our 3 patients, the average global kyphosis decreased from 143° before HPT to 92° after HPT. Obviously, a kyphosis of around 90° is much safer for surgery than one that is > 140°. Furthermore, the average global kyphosis further decreased to 54° after the final surgery. In recent years, more attention has been given to sagittal spinal alignment during the correction of spinal deformities. Schwab et al. (2012) proposed that sagittal vertical axis (SVA), pelvic tilt (PT), and pelvic incidence–lumbar lordosis (PI–LL) mismatch are the 3 most important spinopelvic parameters, which should be carefully considered in the preoperative planning for the treatment of adult spinal deformities. To achieve good clinical outcome, realignment objectives have been reported to be an SVA < 50 mm, PT < 25°, and PI–LL equal to ±9°. More recently, Huang et al. (2020) reported that PT was the major radiographic contributor to Oswestry Disability Index (ODI) score in patients treated for ankylosing spondylitis (AS). The optimal sagittal alignment of AS patients who had undergone a single-level pedicle sub-

A modified halo-pelvic apparatus We used a modified halo-pelvic apparatus (WEGO) that helps to relieve the discomfort and inconvenience of HPT. Unlike the whole-ring pelvic frame in the traditional HPT, the modified pelvic frame is half-ring shaped, and thus there are no pins around the posterior pelvis. The modification increases patient comfort while they are lying on their back. The pelvic ring and skull halo device are applied with the patient under general anesthesia. With the patient in the supine position, 3 pins (4.5 mm diameter) are inserted between the inner and outer table of the ilium on each side. The skull halo device is then placed. The HPT frame is not constructed at the time the pelvic and skull devices are placed. The frame is constructed 3–5 days later so that the patient has been able to accustom him/herself to the frame. After the construction of the frame, distraction can be performed. For the first week, the frame is distracted at a rate of 1 cm per day. Beginning in the second week, the distraction rate is decreased to 0.5 cm per day. With these rates, the height of the patient can be increased by 8–10 cm in the first 2 weeks.

Acta Orthopaedica 2021; 92 (1): 9–14

Preoperative

Traction

Preoperative

Traction Postoperative

Figure 4. Case 3: CT revealed angular lumbar kyphosis, vertebral subluxation, absent vertebral bodies from L2 to L4 (presumably due to tuberculosis), and fusion of the L1–L5 laminas.

Beginning in the third week, the distraction rate is decreased to 0.3–0.5 cm every 2–3 days depending on the patient’s condition and tolerance of the procedure. Traction is performed for 6–10 weeks in total, and the final surgery is then performed. Halo-pelvic traction versus halo-gravity traction (HGT) Similar to HPT, HGT is another form of traction used to obtain correction of spinal deformity prior to operative treatment. It can also be applied to early-onset scoliosis as a delaying tactic. Compared with HPT, HGT is less cumbersome, because the pelvis and the legs remain unrestrained, encouraging patient mobility. Predictably HGT produces 30–35% correction of both coronal and sagittal plane deformity, which seems less effective compared with HPT. Meanwhile, HGT shows a lower complication rate (1–1.5% incidence of neurologic complication), which is safer than HPT. In conclusion, HPT is useful in the management of extreme lumbar kyphosis. 6 weeks of HPT can greatly reduce the global kyphosis, and thus improve the conditions for corrective surgery.

Conception and design of this study: YW and CL. Acquisition of the datasets: LL and LQ. Analysis and interpretation of the results and data: YW and LQ. Drafting the manuscript: YW. Critically reviewing the manuscript: LQ and CL. Acta thanks Paul Gerdhem for help with peer review of this study.

Hua W, Zhang Y, Wu X, Gao Y, Li S, Wang K. Transpedicular wedge resection osteotomy of the apical vertebrae for the treatment of severe and rigid thoracic kyphoscoliosis: a retrospective study of 26 cases. Spine Deform 2019; 7(2): 338-45. Huang J C, Qian B P, Qiu Y, Wang B, Yu Y, Qiao M. What is the optimal postoperative sagittal alignment in ankylosing spondylitis patients with thoracolumbar kyphosis following one-level pedicle subtraction osteotomy? Spine J. 2020; 20(5): 765-75. Kim N H, Kim H J, Moon S H, Lee H M. 20-year-follow up of treatment using spine osteotomy and halo-pelvic traction for tuberculous kyphosis: a case report. Asian Spine J 2009; 3(1): 27-31. Lenke L G, Sides B A, Koester L A, Hensley M, Blanke K M. Vertebral column resection for the treatment of severe spinal deformity. Clin Orthop Relat Res 2010; 468(3): 687-99.

Miladi L. Round and angular kyphosis in paediatric patients. Orthop Traumatol Surg Res 2013; 99(1 Suppl.): S140-9. O’Brien J P, Yau A C, Smith T K, Hodgson A R. Halo pelvic traction: a preliminary report on a method of external skeletal fixation for correcting deformities and maintaining fixation of the spine. J Bone Joint Surg Br 1971; 53(2): 217-29. Ransford A O, Manning C W. Complications of halo-pelvic distraction for scoliosis. J Bone Joint Surg Br 1975; 57(2): 131-7. Schwab F J, Patel A, Shaffrey C I, Smith J S, Farcy J P, Boachie-Adjei O. Sagittal realignment failures following pedicle subtraction osteotomy surgery: are we doing enough? J Neurosurg Spine 2012; 16(6): 539-46. Sponseller P D, Takenaga R K, Newton P, Boachie O, Flynn J, Letko L. The use of traction in the treatment of severe spinal deformity. Spine 2008; 33(21): 2305-9. Suk S I, Chung E R, Kim J H, Kim S S, Lee J S, Choi W K. Posterior vertebral column resection for severe rigid scoliosis. Spine 2005; 30(14): 1682-7.

Acta Orthopaedica 2021; 92 (1): 9–14

Wang Y, Xie J, Zhao Z, Zhang Y, Li T, Bi N. Perioperative major non-neurological complications in 105 patients undergoing posterior vertebral column resection procedures for severe rigid deformities. Spine 2015; 40(16): 1289-96. Yu B, Zhu K, Zhao D, Wang F, Liang Y. Treatment of extreme tuberculous kyphosis using spinal osteotomy and halo-pelvic traction: a case report. Spine 2016; 41(4): E237-41. Zhang Y, Tao L, Hai Y, Yang J, Zhou L, Yin P. One-stage posterior multiple-level asymmetrical Ponte osteotomies versus single-level posterior vertebral column resection for severe and rigid adult idiopathic scoliosis: a minimum 2-year follow-up comparative study. Spine 2019; 44(20): E1196-e205. Zhou C, Liu L, Song Y, Feng G, Yang X, Wang L. Comparison of anterior and posterior vertebral column resection versus anterior and posterior spinal fusion for severe and rigid scoliosis. Spine J 2018; 18(6): 94853.

Acta Orthopaedica 2021; 92 (1): 15–21

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 Liza N VAN STEENBERGEN 1, Keijo T MÄKELÄ 2,3, Johan KÄRRHOLM 4–6, Ola ROLFSON 4–6, Søren OVERGAARD 7–9, Ove FURNES 10,11, Alma B PEDERSEN 9,12, Antti ESKELINEN 3,13, Geir HALLAN 11,12, Berend W SCHREURS 1,14, and Rob G H H NELISSEN 1,15 1 Dutch Arthroplasty Register (LROI), ‘s- Hertogenbosch, the Netherlands; 2 Department of Orthopaedics and Traumatology, Turku University Hospital, Turku, Finland; 3 The Finnish Arthroplasty Register, Helsinki, Finland; 4 Department of Orthopaedics, Sahlgrenska University Hospital, Gothenburg, Sweden; 5 Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; 6 The Swedish Hip Arthroplasty Register, Gothenburg, Sweden; 7 Department of Orthopaedic Surgery and Traumatology, Odense University Hospital, Odense, Denmark; 8 Department of Clinical Research, University of Southern Denmark, Odense, Denmark; 9 The Danish Hip Arthroplasty Register, Aarhus, Denmark; 10 The Norwegian Arthroplasty Register, Department of Orthopaedic Surgery, Haukeland University Hospital, Bergen, Norway; 11 Department of Clinical Medicine, Faculty of Medicine, University of Bergen, Bergen, Norway; 12 Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark; 13 Coxa Hospital for Joint Replacement, and Faculty of Medicine and Health Technologies, University of Tampere, Tampere, Finland; 14 Department of Orthopaedics, Radboudumc, Nijmegen, the Netherlands; 15 Department of Orthopaedics, Leiden University Medical Centre, Leiden, the Netherlands Correspondence: lvansteenbergen@orthopeden.org Submitted 2020-06-26. Accepted 2020-10-14.

Background and purpose — Collaborations between arthroplasty registries are important in order to create the possibility of detecting inferior implants early and improve our understanding of differences between nations in terms of indications and outcomes. In this registry study we compared patient and procedure characteristics, and revision rates in the Nordic Arthroplasty Register Association (NARA) database and the Dutch Arthroplasty Register (LROI). Patients and methods — All total hip arthroplasties (THAs) performed in 2010–2016 were included from the LROI (n = 184,862) and the NARA database (n = 290,823), which contains data from Denmark, Norway, Sweden, and Finland. Descriptive statistics and Kaplan–Meier survival analyses based on all reasons for revision and stratified by fixation were performed and compared between countries. Results — In the Netherlands, the proportion of patients aged < 55 years (9%) and male patients (34%) was lower than in Nordic countries (< 55 years 11–13%; males 35–43%); the proportion of osteoarthritis (OA) (87%) was higher compared with Sweden (81%), Norway (77%), and Denmark (81%) but comparable to Finland (86%). Uncemented fixation was used in 62% of patients in the Netherlands, in 70% of patients in Denmark and Finland, and in 28% and 19% in Norway and Sweden, respectively. The 5-year revision rate for THAs for OA was lower in Sweden (2.3%, 95% CI 2.1–2.5) than in the Netherlands (3.0%, CI 2.9–3.1), Norway (3.8%, CI 3.6–4.0), Denmark (4.6%, CI 4.4–4.8), and Finland (4.4%, CI 4.3–4.5). Revision rates in Denmark, Norway, and Finland were higher for all fixation groups.

Interpretation — Patient and THA procedure characteristics as well as revision rates evinced some differences between the Netherlands and the Nordic countries. The Netherlands compared best with Denmark in terms of patient and procedure characteristics, but resembled Sweden more in terms of short-term revision risk. Combining data from registries like LROI and the NARA collaboration is feasible and might possibly enable tracking of potential outlier implants.

Arthroplasty registries are used to evaluate patient, procedure, prosthesis, and hospital characteristics associated with revision surgery as well as to improve quality of care (Herberts and Malchau 2000, Graves 2010). Comparison of national arthroplasty registries is important to improve our understanding of national differences and similarities. Furthermore, combining data from arthroplasty registries from several countries is needed in order to increase numbers to create the possibility of detecting inferior implants as early as possible. The Nordic Arthroplasty Register Association (NARA) was established in 2007 by representatives from arthroplasty registries in Sweden, Norway, and Denmark to improve the quality of total hip and total knee arthroplasty through a registries-based research collaboration. Finland joined the association in 2010. To date, NARA is the most developed multinational arthroplasty database worldwide (Mäkelä et al. 2019). The comparison of national demographics and results was one of the main initial aims of the NARA collaboration.

Acta Orthopaedica 2021; 92 (1): 15–21

Table 1. Harmonization of LROI variables to the NARA minimal dataset for diagnosis NARA diagnosis

LROI diagnosis

Primary osteoarthritis Osteoarthritis Inflammatory arthritis Inflammatory arthritis Rheumatoid arthritis Rheumatoid arthritis Other inflammatory Ankylosing spondylitis Hip fracture Fracture (acute) Late posttraumatic Pediatric hip disease Post-Perthes Developmental dysplasia of the hip Dysplasia Perthes disease Slipped capital femoral epiphysis Combination of slipped capital femoral epiphysis and Perthes Idiopathic femoral head necrosis Osteonecrosis Other Other

Therefore, the NARA database contains only parameters that are included in all the individual registries (Havelin et al. 2011). The first NARA publication in 2009 described the results of 280,201 total hip arthroplasties (THAs) performed in 1995–2006 in Sweden, Norway, and Denmark (Havelin et al. 2009). This research was updated in 2014 by Mäkelä et al. using NARA data from 1995–2011 including Finland (Mäkelä et al. 2014). Substantial differences were found in the patient populations receiving a THA in the Nordic countries and in the procedure characteristics such as surgical approach and fixation (Mäkelä et al. 2014). Furthermore, substantial differences in 10-year survival rates were found between the Nordic countries (Havelin et al. 2009). The Dutch Arthroplasty Register (LROI) contains data on THAs since 2007. We initiated this study to compare patient and treatment characteristics as well as survival rates between THAs in the Netherlands and Nordic countries.

Patients and methods The NARA database consists of pooled data from the national hip arthroplasty registries of Denmark, Norway, Finland, and Sweden. Each register has validation routines based on national patient registries. A minimal NARA dataset was created that contains data that all registries could deliver, where personal identification numbers are deleted. The data were treated with full confidentiality, and identification of individual patients was not possible as a result of the anonymization of the NARA dataset (Havelin et al. 2009, 2011). The degree of coverage and completeness of the Nordic registries is documented to be higher than 95% (DHAR, FAR, NAR, SHAR n.d.). Selection and transformation of the respective datasets and de-identification of the patients, including deletion of the national personal identity numbers, was per-

Table 2. Harmonization of LROI variables to the NARA minimal dataset for cause of revision NARA cause of revision

LROI cause of revision

Deep infection Infection Aseptic loosening Loosening of acetabular component, no infection Loosening of femoral component, no infection Periprosthetic fracture Periprosthetic fracture Dislocation Dislocation Pain only (not available) Others Other Girdlestone situation Peri-articular ossification Inlay wear Symptomatic metal-on-metal bearing Unknown No cause of revision registered for revision procedure

formed within each national registry. Anonymous data was then merged into a common research database. Ethical approval of the study was obtained through each national registry. The LROI is the nationwide population-based register that includes information on arthroplasties in the Netherlands since 2007. The LROI has coverage of 100% of Dutch hospitals and completeness of reporting of over 95% for primary THAs and 88% for revision arthroplasties (LROI n.d., van Steenbergen et al. 2015). The LROI database contains information on patient, procedure, and prosthesis characteristics recorded by registrars from each hospital. The LROI uses the opt-out system to require informed consent of patients. For the present study we included all primary THAs in the period 2010–2016 registered in the NARA (n = 290,823) and the LROI (n = 184,862). Bilateral procedures were included in the study, since they do not introduce significant dependency problems in register studies (Ranstam et al. 2011). Resurfacing hip arthroplasties were excluded, though metal-on-metal total hip arthroplasties were included. Records with missing data were included. Age was categorized into 4 groups (< 55, 55–64, 65–74, ≥ 75). The categories for diagnosis differed slightly between NARA and the LROI. We harmonized LROI variables to the NARA minimal dataset of variables, with the main harmonization for childhood diseases being developmental dysplasia of the hip, Perthes disease, and slipped capital femoral epiphysis and a combination of the latter two diagnoses in the NARA dataset compared with post-Perthes and dysplasia in the LROI (Table 1). Furthermore, surgical approach was divided into posterior and non-posterior. Cause of revision was harmonized between the NARA and the LROI. In the LROI more than one cause of revision could be registered, which is not in line with the NARA where the main reason for revision is registered. Therefore, a hierarchical order of reasons for revision (similar to the order used on the Norwegian data: top to bottom; infection, aseptic loosen-

Acta Orthopaedica 2021; 92 (1): 15–21

Table 3. Primary total hip arthroplasty volume per country The Nether Denmark Norway Sweden Finland lands Annual primary THA volume 2010 8,478 7,254 15,730 7,153 23,206 2011 8566 7,318 15,784 7,538 23,859 2012 8,536 7,822 15,958 7,895 25,299 2013 8,774 8,087 16,278 8,172 26,038 2014 9,174 8,112 16,528 8,348 28,073 2015 9,516 8,437 16,630 9,015 28,795 2016 9,913 8,873 17,261 9,673 29,592 Total 62,957 55,903 114,169 57,794 184,862 % of total 13 12 24 12 39 Population 2016 (million) and THA incidence 2016 (per 105) Population 5.7 5.2 10.0 5.5 17.0 Incidence 174 170 173 176 150

ing, periprosthetic fracture, dislocation, other, missing) was used for the LROI data. Loosening of the acetabular and femoral component without infection as available in the LROI were combined as aseptic loosening (Table 2). In the LROI for 211 records (11%) more than one reason for revision was registered within 1 year of follow-up and 296 (8%) within the entire study period. The study is reported according to the STROBE guidelines. Statistics Descriptive statistics of patient and procedure characteristics as well as THA incidence per year based on the population per country was calculated, according to country. Survival time was calculated as the time from primary procedure to first allcause, any component revision, death of the patient, or end of the follow-up (January 1, 2017). Median follow-up was 3 years (interquartile range: 1.4–4.8 years). Kaplan–Meier survival analyses were performed to evaluate time to all-cause any component revision including 95% confidence intervals (CI) according to countries for OA THA patients. Stratified analyses were performed according to fixation and sex. Reasons for revision within 1 year and within 5 years’ follow-up time were described per country. For the 95% CIs, we assumed that the number of observed cases followed a Poisson distribution. Ethics, registration, funding and potential conflicts of interests The dataset was processed in compliance with the regulations of the LROI and NARA governing research on registry data. No external funding was received. No competing interests were declared.

Results 475,685 THAs were included in the NARA and LROI databases in the study period, with 39% of procedures from the

Table 4. Characteristics of the total hip arthroplasty patients and procedures registered in the NARA and LROI database, 2010–2016 (n = 475,685). Values are percentages unless otherwise specified The Den- Nor- Fin- NetherFactor mark way Sweden land lands No. of THAs 62,957 55,903 114,169 57,794 184,862 Male sex 43 35 42 43 34 Age < 55 12 11 11 13 9.0 55–64 19 23 20 23 22 65–74 38 37 38 36 38 ≥ 75 31 29 31 28 32 Diagnosis Primary osteoarthritis 81 77 81 86 87 Inflammatory arthritis 1.1 2.1 1.1 2.5 0.9 Hip fracture 11 8.2 11 3.9 6.2 Pediatric hip diseases 3.3 8.8 1.9 0.9 2.3 Idiopathic head necrosis 2.3 2.2 2.2 1.8 2.9 Others 0.9 1.3 3.0 4.8 0.7

Netherlands, 24% from Sweden, and around 12% each from Denmark, Norway, and Finland (Table 3). The volume of primary THAs increased each year in all 5 countries during the study period. The 2016 incidence rates of new THAs were around 170–175 per 100,000 inhabitants in NARA countries, and 150 in the Netherlands (Table 3). Patient characteristics In the Netherlands, the proportion of patients aged < 55 years was 9%, which was lower than that in the Nordic countries; in Finland 13% of THA patients were aged < 55 years. The proportion of patients aged ≥ 75 years was comparable between the Netherlands, Denmark, and Sweden with 31–32% each, while this proportion was lower in Finland and Norway (28–29%). The proportion of male patients was lowest in the Netherlands at 34% compared with any of the Nordic countries (35–43%). The proportion of THAs due to primary osteoarthritis (OA) was higher in the Netherlands (87%) than in Sweden, Norway, and Denmark, but comparable to Finland. The proportion of hip fractures was lower in the Netherlands compared with Denmark, Sweden, and to a lesser extent Norway. The proportion of inflammatory arthritis differed between the countries, from 0.9% in the Netherlands up to 2.5% in Finland. In Denmark and Sweden hip fracture as diagnosis for THA was almost twice as frequent as in the Netherlands and almost 3 times more frequent than in Finland (Table 4). Procedure characteristics Uncemented fixation of both components was used in 62% of patients in the Netherlands, in 70% of patients in Denmark and Finland, and in 28% and 19% in Norway and Sweden, respectively. In the Netherlands, a posterior hip approach was used in 61% of the THA procedures, which was in between Sweden

Acta Orthopaedica 2021; 92 (1): 15–21

Table 5. Procedure characteristics of the total hip arthroplasty patients and procedures registered in the NARA and LROI database, 2010–2016 (n = 475,685). Values are percentages unless otherwise specified

Revision proportion (%) – THA for OA 6 Denmark Finland Norway Sweden The Netherlands

The Den- Nor- Fin- NetherFactor mark way Sweden land lands

No. of THAs 62,957 55,903 114,169 57,794 184,862 Fixation Cemented 12 31 66 11 28 Uncemented 70 28 19 71 62 Hybrid 16 3.0 2.8 15 4.6 Reverse hybrid 0.9 37 13 1.7 4.7 Unknown, uncertain 0.6 1.4 0.0 1.8 0.6 a Posterior approach 96 39 52 67 61

Years after hip arthroplasty

For Finland approach is available since 2014, proportion shown is for 2014–2016.

Figure 1. Cumulative revision proportion (%) of THAs for OA according to country, all fixation methods.

(52%) and Finland (67%), but lower than in Denmark (96%) and higher compared with Norway (40%) (Table 5). Revision In THAs for primary OA the 1-, 3-, and 5-year overall revision rates were lowest in Sweden. The Netherlands had a higher revision rate (5-year revision rate: 3.0%, CI 2.9–3.1) compared with Sweden (2.3%, CI 2.1–2.5), but a lower revision rate compared with Denmark (4.6%, CI 4.4–4.8), Norway (3.8%, CI 3.6–4.0), and Finland (4.4%, CI 4.3–4.5) (Table 6, Figure 1). Stratified analyses showed lower revision rates for cemented THAs in Sweden and the Netherlands compared with Denmark, Norway, and Finland. Uncemented THAs had lower 3- and 5-year revision rates in the Netherlands and Sweden than in Norway, Denmark, and Finland. Hybrid (femur cemented) and reverse hybrid (cup cemented) THAs had lowest short-term revision rates in Sweden (Table 6, Figure 2). However, the number of cases with reverse hybrid THAs was small in Finland and too small in Denmark for meaningful analyses. Therefore the reverse hybrid cases from Denmark were excluded. Stratified analyses by sex showed comparable results for short-term overall revision rates examining differences between countries. Short-term overall revision rates were

Revision proportion (%) – cemented THA for OA

Revision proportion (%) – uncemented THA for OA

6 Denmark Finland Norway Sweden The Netherlands

Denmark Finland Norway Sweden The Netherlands

Years after hip arthroplasty

Revision proportion (%) – hybrid THA for OA

Revision proportion (%) – inversed hybrid THA for OA

6 Denmark Finland Norway Sweden The Netherlands

Finland Norway Sweden The Netherlands

Years after hip arthroplasty

Figure 2. Cumulative revision proportion (%) of cemented THAs (a), uncemented THAs (b), hybrid THAs (c), and inverse hybrid THAs (d) for OA according to country.

Acta Orthopaedica 2021; 92 (1): 15–21

Table 6. Kaplan–Meier revision rates (%) with 95% confidence intervals at 1, 3, and 5 years after primary total hip arthroplasties for osteoarthritis Factor Country

THAs Revisions 1–year (n) (n) revision rate

3–year revision rate

5–year revision rate

All THAs Denmark 50,777 1,910 2.6 (2.4–2.8) 3.8 (3.6–4.0) 4.6 (4.4–4.8) Norway 42,941 1,364 2.3 (2.1–2.5) 3.2 (3.0–3.4) 3.8 (3.6–4.0) Sweden 92,069 1,722 1.2 (1.2–1.2) 1.8 (1.8–1.8) 2.3 (2.1–2.5) Finland 48,597 1,672 2.2 (2.0–2.4) 3.3 (3.1–3.5) 4.4 (4.3–4.5) The Netherlands 159,386 3,791 1.4 (1.4–1.5) 2.4 (2.3–2.5) 3.0 (2.9–3.1) Cemented THA Denmark 5,716 202 2.2 (2.0–2.4) 3.5 (3.2–3.8) 4.2 (3.9–4.5) Norway 13,491 423 2.1 (2.0–2.2) 3.1 (2.9–3.3) 3.8 (3.6–4.0) Sweden 59,568 945 1.0 (1.0–1.0) 1.5 (1.4–1.6) 2.0 (1.9–2.1) Finland 5,029 145 1.6 (1.4–1.8) 2.5 (2.3–2.7) 3.4 (3.1–3.7) The Netherlands 43,095 765 1.0 (0.9–1.1) 1.8 (1.6–1.9) 2.2 (2.1–2.4) Uncemented THA Denmark 37,330 1,478 2.8 (2.7–2.9) 4.0 (3.9–4.1) 4.9 (4.8–5.0) Norway 11,610 415 2.7 (2.5–2.9) 3.8 (3.6–4.0) 4.3 (4.1–4.5) Sweden 17,558 470 1.8 (1.7–1.9) 2.8 (2.7–2.9) 3.5 (3.3–3.7) Finland 34,976 1,323 2.3 (2.2–2.4) 3.6 (3.5–3.7) 4.7 (4.6–4.8) The Netherlands 101,249 2,608 1.5 (1.4–1.6) 2.6 (2.5–2.7) 3.3 (3.2–3.4) Hybrid THA (femur cemented) Denmark 7,288 201 1.8 (1.5–2.1) 2.8 (2.4–3.3) 3.5 (2.9–4.0) Norway 1148 28 2.2 (1.3–3.1) 2.9 (1.7–4.1) 3.8 (1.6–6.0) Sweden 2395 34 1.1 (0.7–1.6) 1.5 (1.0–2.1) 1.9 (1.1–2.7) Finland 7198 165 1.9 (1.6–2.3) 2.5 (2.1–2.9) 3.1 (2.5–3.7) The Netherlands 7,242 171 1.4 (1.2–1.7) 2.5 (2.1–2.9) 3.2 (2.7–3.7) Reverse hybrid THA (cup cemented) Denmark 285 21 n.a. n.a. n.a. Norway 16,329 485 2.1 (1.9–2.3) 3.0 (2.7–3.3) 3.5 (3.2–3.9) Sweden 12,546 273 1.3 (1.1–1.5) 2.1 (1.9–2.4) 2.6 (2.2–2.9) Finland 771 22 1.7 (0.8–2.7) 2.5 (1.4–3.7) 3.1 (1.8–4.5) The Netherlands 6,992 225 1.9 (1.6–2.3) 3.4 (3.0–3.9) 4.0 (3.5–4.6) n.a. = not available, numbers are too small.

somewhat higher for males compared with females in all countries (data not shown). Analyses of the whole study population including cases with non-OA diagnoses gave the same results (data not shown).

Reason for revision The most frequent reason for revision in the first year after primary THA differed between countries; dislocation was the most frequently registered reason for revision within the first year for Denmark and the Netherlands. For Norway, Sweden, and Finland deep infection was the most frequent reason for revision (Table 7). After up to 5 years of follow-up, dislocation and infection were the most common reasons for revision in the Nordic countries, while aseptic loosening was the most frequently registered reason for revision within 5 years of follow-up in the Netherlands.

Discussion This study provides the first comparison between the NARA countries (Denmark, Finland, Norway, and Sweden) and the Netherlands for primary THA procedures. Combining data from different national arthroplasty registries like NARA and LROI is possible for patient and procedure characteristics as well as for revision rates and reasons for revision. When combining datasets some requirements are necessary such as harmonizing categories of diagnosis and reasons for revision.

Patient and procedure characteristics Our findings concerning patient and procedure characteristics of the NARA countries are in line with previously reported results. However, the proportion of male patients is somewhat

Table 7. Reasons for revision (%) within 1 year and up to 5-year follow-up of total hip arthroplasties for osteoarthritis registered in the NARA and LROI database The Den- Nor- Fin- NetherFactor mark way Sweden land lands

The Den- Nor- Fin- NetherFactor mark way Sweden land lands

Revisions within 1 year (n = 401,878) a THAs 42,628 36,149 78,073 40,552 133,985 Revisions 1,420 1,087 1,338 1,145 1,935 Cause of revision Dislocation 0.77 0.36 0.29 0.48 0.40 Deep infection 0.71 1.08 0.77 0.54 0.27 Aseptic loosening 0.25 0.26 0.10 0.19 0.34 Periprosthetic fracture 0.61 0.21 0.17 0.41 0.18 Pain only 0.06 0.03 0.01 0.01 n.a. Others 0.26 0.21 0.04 0.51 0.18 Unknown 0.02 0.15 0.00 0.21 0.09

Revisions up to the 5-year follow-up (n = 475,685) THAs (n) 50,777 42,941 92,069 48,597 159,386 Revisions 1,910 1,364 1,722 1,672 3,854 Cause of revision Dislocation 0.98 0.55 0.36 0.64 0.60 Deep infection 0.86 1.18 0.87 0.64 0.44 Aseptic loosening 0.61 0.56 0.32 0.41 0.66 Periprosthetic fracture 0.73 0.26 0.21 0.50 0.23 Pain only 0.18 0.15 0.03 0.06 n.a. Others 0.36 0.29 0.08 0.95 0.40 Unknown 0.03 0.18 0.00 0.25 0.10

n.a. = not available.

In 2010–2015 to assure at least 1-year follow-up for all records.

higher in all NARA countries, especially in Norway compared with earlier NARA results (Mäkelä et al. 2014). Dutch patients who received a THA were generally somewhat older than THA patients in NARA countries, with Finland having a known younger THA population (Mäkelä et al. 2014). Furthermore, the proportion of OA as the diagnosis for primary THA was higher in the Netherlands compared with the Nordic countries, whereas the proportion of hip fractures was lower in the Netherlands compared with Denmark, Norway, and Sweden. This might be the result of the varying indications for use of THA for fracture patients in the Nordic countries and the Netherlands. The incidence of THA is in line with reported international THA incidence rates (Merx et al. 2003, Paxton et al. 2019), but seems to be somewhat lower in the Netherlands compared with the Nordic countries. This may reflect differences in demography, prevalence of various hip diseases, indications for surgical treatment, and/or healthcare policy between countries. Apart from the differences in patient population, procedure characteristics also differed between the Netherlands and the Nordic countries. The Netherlands was most comparable to Denmark and Finland concerning THA fixation technique. These countries had a high proportion of uncemented fixation, while Norway and Sweden had a low proportion of uncemented fixation. Good results of cemented THA and inferior results of some uncemented THAs in the Swedish and Norwegian registries have encouraged the continued use of cemented THA in these countries (Havelin et al. 2009). However, uncemented fixation as well as reverse hybrid fixation increased in both Norway and Sweden during the study period. Revision There were differences in short-term revision rates between NARA countries and the Netherlands for all evaluated hip stem and acetabular fixation group combinations. Although differences were small, revision risk at short-term follow-up was lowest in Sweden for cemented THAs. The Swedish Hip Arthroplasty Register (SHAR) has provided feedback and advice to the orthopedic surgeons for over 40 years (SHAR n.d.). However, Dutch results showed comparable low shortterm revision rates as well. The higher short-term revision rates for THAs in Denmark and Finland might be explained by more dislocations due to the posterior approach (Zijlstra et al. 2017), which is frequently used in these countries. However, the posterior approach has been shown to give better patientreported outcomes (Amlie et al. 2014, Rosenlund et al. 2017). Results of large femoral head metal-on-metal THAs have been poor (Seppanen et al. 2018), which may be an important factor in the explanation for the higher revision rates in uncemented THAs in Denmark and Finland where the proportion of these THAs was high. However, in the Netherlands these large femoral head metal-on metal THAs were also used in substantial numbers (LROI n.d.). Unfortunately we were not able to exclude these big head metal-on-metal cases from our analyses because we did not include femoral head size or articulating materials in this data set. Furthermore, patient

Acta Orthopaedica 2021; 92 (1): 15–21

characteristics are likely to play a role in the revision rates of THAs, with Finland having a younger patient population that might partly explain their higher short-term revision rate (FAR, LROI n.d.). Reverse hybrid and hybrid THAs had the lowest short-term revision rates in Sweden, being lower than in the Netherlands. It has been stated previously based on NARA data, however, that reverse hybrid THA has a higher revision rate than cemented THA, mainly due to periprosthetic fractures (Wangen et al. 2017). The number of reverse THAs in the Nordic countries has decreased in recent years, after the period of the current study (2010–2016). Cause of revision differed between countries, with dislocation being the most frequent reason for revision in the Netherlands and Denmark. This is probably related to a higher proportion of the THAs done with a posterior approach. Infection represents a major share of the short-term revisions, which might substantially influence the short-term revision rate. However, infection as a reason for revision should be considered carefully, since the capture rate in arthroplasty registries is known to be suboptimal, with up to 40% of revision procedures with proven infections missing (Gundtoft et al. 2015). This might be different between countries, which most likely biases the results. Validity of revision for infection largely depends on the possibility to validate registry data with other datasets from, e.g., microbiology. Currently, a study is being performed in the Netherlands to examine the proportion of revisions for infections not registered in the LROI, which is expected to be comparable to the study by Gundtoft et al. The proportion of unknown reasons for revision differed substantially between countries, which also might bias the short-term results. Strengths and weaknesses The strength of this study is the large population-based registry dataset from 5 countries with high-quality data and minimal loss to follow-up from each national register. Completeness of data in the included registries is documented to be higher than 95% (DHAR, FAR, LROI, NAR, SHAR n.d.). Furthermore, registries provide real-world data with high generalizability and external validity. This gives us the opportunity to describe trends and differences between countries. On the other hand, our registry data are observational and can be subject to bias that we cannot account for, i.e., selection bias and confounding by indication by surgeon. Therefore, causality cannot be inferred. Data used for this project is limited by being common to all 5 registries, and hence we lack information on some factors that affect the outcomes. In this first study comparing LROI and NARA data we did not merge the datasets, hampering the possibility to perform adjusted analyses. In this study we examined and compared NARA and LROI data from a relatively recent period, resulting in a limited follow-up of THA procedures with a maximum of 7 years. Furthermore, we did not include femoral head size or type of articulation, which are known factors influencing revision rates. Moreover, we were not able to exclude metal-on-metal THAs,

Acta Orthopaedica 2021; 92 (1): 15–21

which have a known high revision rate (Smith et al. 2012). In addition, we evaluated overall revision, which contains a variety of revision procedures ranging from small revision procedures like femoral head exchange to major revision procedures such as complete exchange or extraction of all components. The type of revision procedure performed might differ between countries, resulting in differences in revision rates. Moreover, the threshold for performing revision procedures may vary between countries, which influences revision rates. Demographics of the Nordic countries and the Netherlands are comparable, although there are substantial differences in travel distance to the nearest (revision) hospital between the Netherlands and, e.g., the rural areas of the Nordic countries. Furthermore, the Nordic countries as well as the Netherlands have a solid healthcare system. However, while the health expenditure of the Nordic countries is mainly paid by the government, the health expenditure of the Netherlands is mainly covered by compulsory health insurance (OECD n.d., OECD 2019). Conclusion Patient and THA procedure characteristics as well as revision rates evinced some differences between the Netherlands and the Nordic countries, although the magnitude of differences was small. The Netherlands seemed to compare best to Denmark in terms of patient and procedure characteristics, but was most comparable to Sweden in terms of short-term revision risk for both cemented and uncemented THAs. The observed absolute differences in revision risk were small and might not be clinically relevant, since they most likely are a reflection of variations in demographics, patient selection, procedure characteristics, or indications for revision. However, most of these factors cannot be studied based on the common data available. Combining data from registries like LROI and those included in the NARA collaboration is feasible. This might possibly enable tracking of potential outlier implants, specific patient groups, or events with a relatively low occurrence. Further research is warranted to merge datasets and look at more details concerning THA, creating the opportunity to perform patient-, procedure-, and prosthesis-adjusted analyses. All authors made a substantial contribution to the study. RN, BS, and LS contributed to the conception of the study. JK created the dataset with the NARA study group. LS conducted the statistical analyses and prepared the manuscript. KM, JK, OR, SO, OV, AP, AE, GH, BS, and RN participated in the data collection and revision of the manuscript. The authors would like to thank the technical support team at the SHAR for their help. Amlie E, Havelin L I, Furnes O, Baste V, Nordsletten L, Hovik O, Dimmen S. Worse patient-reported outcome after lateral approach than after anterior and posterolateral approach in primary hip arthroplasty: a cross-sectional questionnaire study of 1,476 patients 1–3 years after surgery. Acta Orthop 2014; 85(5): 463-9. DHAR. www.dhr.dk.

FAR. www.thl.fi/far. Graves S E. The value of arthroplasty registry data. Acta Orthop 2010; 81(1): 8-9. Gundtoft P H, Overgaard S, Schonheyder H C, Moller J K, KjaersgaardAndersen P, Pedersen A B. The “true” incidence of surgically treated deep prosthetic joint infection after 32,896 primary total hip arthroplasties: a prospective cohort study. Acta Orthop 2015; 86(3): 326-34. 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. Havelin L I, Robertsson O, Fenstad A M, Overgaard S, Garellick G, Furnes O. A Scandinavian experience of register collaboration: the Nordic Arthroplasty Register Association (NARA). J Bone Joint Surg Am 2011; 93(Suppl. 3): 13-19. Herberts P, Malchau H. Long-term registration has improved the quality of hip replacement: a review of the Swedish THR Register comparing 160,000 cases. Acta Orthop Scand 2000; 71(2): 111-21. LROI. www.lroi-report.nl. 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 2014; 85(2): 107-16. Mäkelä K T, Furnes O, Hallan G, Fenstad A M, Rolfson O, Kärrholm J, Rogmark C, Pedersen A B, Robertsson O, A W D, Eskelinen A, Schroder H M, Aarimaa V, Rasmussen J V, Salomonsson B, Hole R, Overgaard S. The benefits of collaboration: the Nordic Arthroplasty Register Association. EFORT Open Rev 2019; 4(6): 391-400. Merx H, Dreinhofer K, Schrader P, Sturmer T, Puhl W, Gunther K P, Brenner H. International variation in hip replacement rates. Ann Rheum Dis 2003; 62(3): 222-6. NAR. http://nrlweb.ihelse.net/eng/Rapporter/Report2019_english.pdf. OECD. Health at a glance 2019, Paris: OECD; 2019. OECD. https://ec.europa.eu/health/sites/health/files/state/docs/chp_sv_english.pdf. Paxton E W, Cafri G, Nemes S, Lorimer M, Kärrholm J, Malchau H, Graves S E, Namba R S, Rolfson O. An international comparison of THA patients, implants, techniques, and survivorship in Sweden, Australia, and the United States. Acta Orthop 2019; 90(2): 148-52. Ranstam J, Kärrholm J, Pulkkinen P, Makela K, Espehaug B, Pedersen A B, Mehnert F, Furnes O, group Ns. Statistical analysis of arthroplasty data, I: Introduction and background. Acta Orthop 2011; 82(3): 253-7. Rosenlund S, Broeng L, Holsgaard-Larsen A, Jensen C, Overgaard S. Patientreported outcome after total hip arthroplasty: comparison between lateral and posterior approach. Acta Orthop 2017; 88(3): 239-47. Seppanen M, Laaksonen I, Pulkkinen P, Eskelinen A, Puhto A P, Kettunen J, Leskinen J, Manninen M, Mäkelä K. High revision rate for large-head metal-on-metal THA at a mean of 7.1 years: a registry study. Clin Orthop Relat Res 2018; 476(6): 1223-30. SHAR. https://registercentrum.blob.core.windows.net/shpr/r/Arsrapport_2018 _Hoftprotes_ENG_26mars_Final-rJepCXNsLI.pdf. Smith A J, Dieppe P, Vernon K, Porter M, Blom A W, National Joint Registry of England and Wales. Failure rates of stemmed metal-on-metal hip replacements: analysis of data from the National Joint Registry of England and Wales. Lancet 2012; 379(9822): 1199-204. van Steenbergen L N, Denissen G A, Spooren A, van Rooden S M, van Oosterhout F J, Morrenhof J W, Nelissen R G. More than 95% completeness of reported procedures in the population-based Dutch Arthroplasty Register. Acta Orthop 2015; 86(4): 498-505. Wangen H, Havelin L I, Fenstad A M, Hallan G, Furnes O, Pedersen A B, Overgaard S, Kärrholm J, Garellick G, Mäkelä K, Eskelinen A, Nordsletten L. Reverse hybrid total hip arthroplasty. Acta Orthop 2017; 88(3): 248-54. Zijlstra W P, De Hartog B, Van Steenbergen L N, Scheurs B W, Nelissen R. Effect of femoral head size and surgical approach on risk of revision for dislocation after total hip arthroplasty. Acta Orthop 2017; 88(4): 395-401.

Acta Orthopaedica 2021; 92 (1): 22–27

Custom-made 3D-printed cup-cage implants for complex acetabular revisions: evaluation of pre-planned versus achieved positioning and 1-year migration data in 10 patients Vasileios ZAMPELIS and Gunnar FLIVIK

Department of Orthopedics, Skane University Hospital, Clinical Sciences, Lund University, Lund, Sweden Correspondence: vasileios.zampelis@med.lu.se Submitted 2020-06-05. Accepted 2020-08-22

Background and purpose — The use of custom-made 3D-printed titanium implants for the reconstruction of large acetabular defects has been successively introduced in the last decade. In an observational cohort study we evaluated the agreement between preoperatively planned and actually achieved cup-cage position as well as 1-year follow-up migration of the cup-cage component. Patients and methods — 10 patients with Paprosky III defects underwent revision surgery using a custom-made 3D-printed cup-cage. The position of the implant on postoperative CT scan was compared with the preoperative plan and the postoperative CT scan was compared with the 1-year follow-up CT scan. Results — There was a median deviation in postoperative position versus planned in inclination of 3.6° (IQR 1.0–5.4), in anteversion of –2.8° (IQR –7.5 to 1.2), and in rotation of –1.2° (IQR –3.3 to 0.0). The median deviation in position of the center of rotation (COR) was –0.5 mm (IQR 2.9 to 0.7) in the anteroposterior (AP) plane, –0.6 mm (IQR –1.8 to –0.1) in the mediolateral (ML) plane, and 1.1 mm (IQR –1.6 to 2.8) in the superoinferior (SI) plane. The migration between postoperative and 1-year follow-up caused a mean change in inclination of 0.04° (IQR –0.06 to 0.09), in anteversion of –0.13° (IQR –0.23 to –0.06), and in rotation of 0.05° (IQR –0.46 to 1.4). The migration of COR was –0.08 mm (IQR –0.18 to –0.04) in the AP plane, 0.14 mm (IQR –0.08 to 0.22) in the ML plane, and 0.06 mm (IQR –0.02 to 0.35) in the SI plane. There was no re-revision. Interpretation — The early results show good agreement between planned and achieved cup-cage position and small measured migration values of the cup-cage component at the 1-year follow-up.

Revision of the acetabular component following primary total hip arthroplasty (THA) is especially challenging in patients with large bony defects. The Paprosky classification (Paprosky et al. 1994) is a widely used system for classifying acetabular bone loss in revision THA. The Paprosky III defects are the most complex patterns and therefore the most difficult to reconstruct. The aim of acetabular revision is to restore, when possible, the bone defects present and the center of rotation (COR) of the hip joint, thus providing a stable and durable reconstruction. Various reconstructive surgical techniques exist depending on whether the bone defects are small/contained or larger/ non-contained. In our department, for the treatment of such large acetabular bone defects, we have started to use the aMace acetabular revision system (aMace Cage, Materialise, Leuven, Belgium), a custom-made titanium cup-cage implant. The implant is designed from a computed tomography (CT) analysis of the bone-deficient acetabulum, providing a 3D-printed titanium cup-cage into which to cement a cup. The implant used was designed without an integrated augment but to fully match the individual patient’s outer pelvic surface anatomy, thus resulting in a stable reconstruction and bridging of larger bone defects. There are locations for pre-planned multiple screws, to fix the implant, aiming for the areas with best bone quality. The personalized design is intended to maximize bone preservation, allowing adaptation of the 1-piece implant to the specific patient in which a cemented cup can be placed in the desired position, thus restoring the hip COR. We evaluated the agreement between planned and achieved cup-cage position and measured possible migration of the cup-cage component, by using a novel CT-measuring technique, in patients with Paprosky III acetabular defects at 1-year follow-up.

Acta Orthopaedica 2021; 92 (1): 22–27

Table 1. Patient demographics. Values are number unless otherwise specified Variable No. of patients Mean age (range) Sex (M:F) Mean BMI (range) Paprosky III-A Right:left First-time vs. repeat revisions

10 64 (36–87) 5:5 26 (19–34) 10 7:3 3:7

Figure 1. Bone quality assessed by the software; red color reveals a poor and green shows excellent bone quality. (a): Assessment of bone quality as described by Gelaude. (b): Defect analysis quantifies in percentages and colors bone loss in the different regions of the acetabulum on which the Paprosky classification is based. Red color reveals an inferior and green shows excellent bone quality while yellow is acceptable. The output data consists of a ratio and a graph, which allow the direct comparison between specimens. The amount of original acetabular bone that is missing is defined as a ratio. The remaining bone stock in the radial direction is presented by the graph (c).

Patients and methods Study design This is a single-center observational cohort study. The inclusion criteria were patients, regardless of age, with aseptic prosthetic loosening following either a primary total hip arthroplasty (THA) or 1 or multiple earlier revisions, with a Paprosky type III-A acetabular bone defect (Table 1). The patients were scheduled for acetabular revision surgery with a custom-made 3D-printed cup-cage (aMace Cage, Materialise, Leuven, Belgium) revision system and surgery was performed between March 2014 and May 2017. Due to large bone defects present in these patients, few other surgical options were available. Cup-cage design Each patient had a routine preoperative CT scan to determine the acetabular defects in terms of missing bone and thickness, based on which the Paprosky type was determined (Figure 1). Based on the preoperative CT scan, the implant was designed to achieve the ideal center of rotation (COR), optimal implant inclination (INCL), and anteversion (AV). Additionally, a unique bone quality map was provided to determine the total radial acetabular bone loss (Gelaude et al. 2011) (Figure 1) and the ideal screw positioning. Thus, the CT scan data was used for the production of a 3D-printed bone model of the patient’s hemipelvis, an implant trial model, drill guides, and

finally a patient-specific monobloc cup-cage titanium implant (Figure 2). The implants we used had no integrated augments but a porous rough trabecular back surface designed to improve secondary fixation and central holes in the dome to be able to impact bone graft. Participants 10 patients (mean age 64 years [36–87], 5 men) with Paprosky III-A acetabular defects scheduled for acetabular cup revision with the use of a custom-made 3D-printed cup-cage (aMace Cage, Materialise, Leuven, Belgium) were included (Table 1). All patients remained available for the follow-up CT scan 1 year after surgery. Revision involved both cup and stem components. Surgery All revisions were carried out at the Orthopedic Department of Skåne University Hospital (Lund University, Sweden) and performed through a posterolateral approach by the same orthopedic surgeon (GF). After removal of the failed implant, all interface membranes in the acetabulum and areas of bone defect were removed, exposing the underlying acetabular morphologic features as modelled during prototyping. In all cases, prior to cage placement, morselized allograft bone was used to fill the gap between the cage and the host bone. Information regarding the quantity of allograft bone used in the impaction bone grafting was not recorded in all cases. In those cases the quantity was recorded; 1 large or 2 medium femoral heads were used.

Acta Orthopaedica 2021; 92 (1): 22–27

Figure 2. Custom made impalnt, planned screw positions and implant position based on CT scan. CT was taken prior to surgery. The radiograph shows the postoperative result. The position of the implant after surgery was compared with the preoperative planning using the pre- and postoperative CT scans.

The custom-made implant was fixated with screws, the number, positions, and length of which were according to the preoperative plan based on the CT scan. An Exeter X3 RimFit cup (Stryker International, London, UK) was cemented with the same AV and INCL as the cup-cage implant itself. After surgery, while the patient was still in hospital, a postoperative CT was performed. Postoperative full weight-bearing was allowed, as tolerated. 3 doses of systemic cloxacillin were given as perioperative prophylaxis. Low molecular weight heparin was given postoperatively for 10 days but 4 weeks if predisposed with any extra thromboembolic risk factors. CT analysis and evaluation For each patient, 3 CT scans were performed: 1 prior to surgery, 1 during the 1st postoperative week (CT1), and the final at 1-year follow-up (CT2). 30 CT scan sets were thus included in this analysis. For every patient, each postoperative CT examination was analyzed twice (M1 and M2), at least 1 month apart. The rater (at Materialise) was blinded to the 1st set of results by renaming the case IDs in the filename of the image sets to avoid potential bias. Repeatability, assuming no bias (M1, M2), was calculated as described by Ranstam et al. (2000) (Table 2). For deviation analysis the achieved position of the implant on postoperative CT scan (CT1) was measured and compared with the preoperatively planned position, using the average values of the 2 repeated measures (M1, M2). Further, the position on the 1-year follow-up CT scan was also compared with postoperative CT scan. For the migration analysis, the difference between these 2 measurements was calculated. All CT scans were analyzed by the same software from Materialise (Mimics Innovation Suite, Materialise NV, Leuven, Belgium). As the same orthogonal coordinate system is used for the measurements of both left and right hips, to make them comparable, the signed values were shifted for the left hips regarding inclination, anteversion and mediolateral translation. Inclination, anteversion and the center of rotation (COR) of the implant were then compared using signed values, indicating

the actual direction of the changes in positioning and migration. The position of the COR was decomposed into 3 different orthogonal components: anteroposterior (AP), mediolateral (ML), and superoinferior (SI). Values were positive when deviating anteriorly, medially, or superiorly, respectively. The evaluation was done in the same fashion as previously described by Baauw et al. (2015). Rotation was determined, clockwise values being positive and anticlockwise values being negative. In order to make rotation for left and right hips comparable, signed values were shifted for the left hips. Before measuring rotation, the difference between the planned and the postoperative anteversion and inclination was neutralized by translating the postoperative COR to the planned COR and by shifting the postoperative acetabular plane to the planned acetabular plane. Statistics Due to the small number of patients descriptive statistics only was used as number of occurrences, median, and interquartile range (IQR) (Tables 2 and 3). Ethics, funding, and potential conflicts of interest The surgical option and follow-up routine for this patient category were according to the departmental norm. Thus, for this observational study, no Ethical Board Review Committee approval was necessary. However, all patients gave informed written consent to participate in the study and the follow-up examinations. Reduced implant cost was provided for the study by the manufacturing company Materialise NV, Leuven, Belgium. GF has, in the past 3 years, given paid presentations for the same company.

Results There was a median deviation in postoperative position versus planned in inclination of 3.6° (IQR 1.0–5.4), in anteversion of –2.8° (IQR –7.5 to 1.2), and in rotation of –1.2° (IQR –3.3 to

Acta Orthopaedica 2021; 92 (1): 22–27

Table 2. Median difference between planned versus postoperative plus migration values from direct postoperative to 1-year follow-up Factor

Deviation from planned cup position median IQR range

Inclination (°) Anteversion (°) Rotation (°) Translation of COR (mm) AP plane (anterior +) ML plane (medial +) SI plane (superior +)

First year migration median IQR

range

Repeatability limit a Postop. 1 year

3.6 –2.8 –1.2

1.0 to 5.4 –7.5 to 1.2 –3.3 to 0.0

–11 to 11 –12 to 5.7 –7.4 to 14

0.04 –0.13 0.05

–0.06 to 0.09 –0.22 to 0.98 –0.23 to –0.06 –0.41 to 0.05 –0.07 to 0.36 –0.46 to 1.4

0.48 2.1 0.97

0.51 1.0 0.99

–0.5 –0.6 1.1

–2.9 to 0.7 –7.0 to 3.4 –1.8 to –0.1 –3.6 to 2.2 –1.6 to 2.8 –9.7 to 8.9

–0.08 0.14 0.06

–0.18 to –0.04 –0.48 to 0.43 –0.08 to 0.22 –0.35 to 0.51 –0.02 to 0.35 –0.15 to 0.61

0.28 0.63 0.44

0.25 0.34 0.26

Values are presented as signed, thus indicating the direction of the migration except for rotation where signed values are positive for clockwise rotation after signed values for left hips are shifted. a Repeatability limit: the value less than or equal to which the absolute difference between two tests results obtained under repeatability conditions may be expected to have a probability of 95%. IQR: interquartile range.

Table 3. The presented values are the average of the 2 repeated measurements (M1, M2) and show the difference between planned and postop position as well as the change in position postoperatively to 1 year follow-up. At the bottom rows, the median signed values show the direction of migration, whereas the unsigned median values show the amplitude of the migration Inclination (°) Anteversion (°) ∆ planned ∆ postop. ∆ planned ∆ postop. Case Planned Postop. 1 year to postop. to 1 year Planned Postop. 1 year to postop. to 1 year 1 40 43.66 43.67 3.66 0.02 20 21.39 21.29 1.39 –0.11 2 40 43.60 43.66 3.60 0.06 20 25.69 25.74 5.69 0.05 3 40 51.20 51.16 11.20 –0.04 20 12.34 11.99 –7.66 –0.35 4 40 45.50 45.76 5.50 0.25 20 20.61 20.56 0.61 –0.05 5 40 45.27 45.36 5.27 0.09 20 11.98 11.84 –8.03 –0.14 6 40 41.04 40.97 1.04 –0.06 20 16.89 16.82 –3.12 –0.07 7 40 39.76 40.74 –0.24 0.98 20 7.90 7.88 –12.10 –0.02 8 40 40.99 40.92 0.98 –0.07 20 17.48 17.07 –2.53 –0.41 9 40 47.32 47.10 7.32 –0.23 20 12.95 12.70 –7.05 –0.25 10 40 29.32 29.38 –10.69 0.07 20 22.68 22.52 2.68 –0.16 Signed median 3.63 0.04 –2.82 –0.12 Unsigned median 4.46 0.07 4.40 0.12 In the bottom rows, the median signed values show the direction of deviation/migration, whereas the unsigned median values show the amplitude of deviation/migration. ∆: difference.

0.0). The median deviation in position of COR was –0.5 mm (IQR –2.9 to 0.7) in the AP plane, –0.6 mm (IQR –1.8 to –0.1) in the ML plane, and 1.1 mm (IQR –1.6 to 2.8) in the SI plane. The migration between postoperative and 1-year follow-up caused a mean change in inclination of 0.04° (IQR –0.06 to 0.09), in anteversion of –0.13° (IQR –0.23 to –0.06), and in rotation of 0.05° (IQR –0.07 to 0.36). The migration of COR was –0.08 mm (IQR –0.18 to –0.04) in the AP plane, 0.14 mm (IQR –0.08 to 0.22) in the ML plane, and 0.06 mm (IQR –0.02 to 0.35) in the SI plane (Tables 2 and 3). Precision analysis under repeatability conditions, assuming no bias (M1, M2), was calculated as presented in Table 2. Thus, all postoperative migration values are below the repeatability limit.

No re-revision, dislocation, infection, or fracture occurred within the 1-year follow-up. To date, none of the patients included in the study has been reoperated. 1 patient has died 2 years after surgery due to reasons not related to the hip surgery.

Discussion Our results are in accordance with a recently published systematic review (Chiarlone et al. 2020), indicating good shortterm results and suggesting a reliable treatment option with a promising future for the treatment of severe acetabular defects. The use of this novel CT-based methodology makes it possible to create a custom-made implant that seems to

provide primary implant stability, a prerequisite for satisfactory results after revision hip arthroplasty. Not only does the patient-specific implant match the acetabular defect, but its flanges outline better the ilium, ischium, and pubic bone as well. There is, however, no single surgical technique to solve the problem of cup fixation, as this is challenged by the severity of different acetabular defects. The use of impaction bone grafting (IBG) with direct cemented fixation, sometimes with metal augments or with a reinforcement cage, is considered by many to be a solid, biological fixation option in the reconstruction of large acetabular defects (Slooff et al. 1993, Sheth et al. 2013, Abolghasemian et al. 2014, Gilbody et al. 2014). Nonetheless, for these large acetabular defects, the use of IBG with metal mesh has been questioned (Buttaro et al. 2008). Other alternative fixation options include uncemented, often screw-stabilized, cups with or without bone graft or metal augments (Templeton et al. 2001, Weeden and Schmidt 2007, Macheras et al. 2009, Del Gaizo et al. 2012, Whitehouse et al. 2015). Mid- and long-term results in contained defects treated with IBG are favorable (van Egmond et al. 2011), while in segmental defects of the roof or pelvic discontinuity failures have been reported (Bonnomet et al. 2001, van Haaren et al. 2007). Desirable prerequisites for successful and durable revision include viable host bone, adequate surgical technique, and a stable and durable implant. Reconstruction of large acetabular defects with trabecular augments or reconstruction cages can be done with wellknown methods (Lopez et al. 2018, Theil et al. 2019). However, the difficulty in treating Paprosky type III acetabular defects lies in the presence of extensive bone loss, jeopardizing proper placement of acetabular components due to loss of normal bony landmarks, affecting both primary stable fixation and the restoration of the hip center. The literature shows the difficulty of accurate acetabular implant positioning (Choi et al. 2013, Citak et al. 2018) as well as a high complication rate (DeBoer et al. 2007, Taunton et al. 2012, Wind et al. 2013, Myncke et al. 2017). We found good agreement between planned and achieved implant position regarding both rotational and translational values. During the last decades, attempts have been made to identify whether, and to what extent, early prosthetic micromotion results in later aseptic loosening (Karrholm et al. 1994). Based on the initial radiostereometric analysis, stem and socket migration exceeding 1.2 mm for the stem and 1.29 mm for the socket during the first 2 years increases the probability of revision (Karrholm et al. 1994, Nieuwenhuijse et al. 2012, Pijls et al. 2012). Although a 1-to-1 relationship between initial migration and long-term survivorship can only be assessed by long-term studies, 1-year migration values of our study are small, indicating a stable construct. The preoperative CT scanning results in a patient-specific implant bridging the acetabular defect and offers a better possibility to plan for more exact anatomical restoration. The anatomical fit of the implant, together with the pre-defined screws aiming for

Acta Orthopaedica 2021; 92 (1): 22–27

areas of good bone quality, minimizes possible surgical application difficulties. Altogether, the well-planned and precise process of implant positioning appears to result in a stable construct 1 year after complex acetabular revision. At the CT1 follow-up when measuring repeatability between M1 and M2, a rather high measuring error for anteversion on the postoperative CT was explained by 1 outlier of 1.6°. Otherwise we consider the measurements to be reliable, as indicated in Table 2. Results are comparable to previously published studies with a similar type of concept and a very good survival rate (Colen et al. 2013, Baauw et al. 2017, Myncke et al. 2017, Goriainov et al. 2018). In a study by Citak et al. (2018), however, 1 of 9 patients suffered an implantassociated complication, after 13 months, which required revision and other complications occurred in 5 patients. A possible explanation for the complications might be that their series included even more severe cases, the revised hip had bilateral pelvic discontinuity. A study involving 16 patients who underwent revision surgery with an associated Paprosky type III defect and a similar custom-made implant, comparing planned versus postoperative implant position with CT, indicated promising results (Baauw et al. 2015). In their study, the custom-made implant used was either a monobloc with integrated augment or in two parts as a modular construct (14 monoblocs; 2 modular constructs). Compared with our study, a higher degree of deviation from planned to postoperative position is reported, possibly related to several outliers, which may be a result of more severe bone defects. Our study is limited by the small sample size and the short follow-up time. However, all patients were included in the 1-year follow-up and no severe per- or postoperative complications were registered. To our knowledge, no previous migration study has been conducted with CT comparison between the postoperative and 1-year implant position for this type of implants. In conclusion, our study shows good agreement between pre-planned and achieved implant position as well as a very stable construct at 1-year follow-up for these complex acetabular revision cases. This encourages us to continue the use of 3D-printed custom-made acetabular implants, including longterm follow-ups. VZ: conduct of study, data analysis, writing of the manuscript. GF: study design and conduct, performing the surgery, data analysis, critical revision of the manuscript. Thanks are offered to Wouter Houthoofd (Materialise NV) for technical help in performing the CT analysis and also for review of the manuscript for accuracy of the technical details, but without any influence on the interpretation and conclusion of the results. Additionally, Helene Jacobsson at Clinical Studies Sweden—Forum South, Skane University Hospital, Lund is thanked for statistical guidance. Acta thanks Bart L Kaptein and Marieke Scharff-Baauw for help with peer review of this study.

Acta Orthopaedica 2021; 92 (1): 22–27

Abolghasemian M, Sadeghi Naini M, Tangsataporn S, Lee P, Backstein D, Safir O, Kuzyk P, Gross A E. Reconstruction of massive uncontained acetabular defects using allograft with cage or ring reinforcement: an assessment of the graft’s ability to restore bone stock and its impact on the outcome of re-revision. Bone Joint J 2014; 96-B(3): 319-24. Baauw M, van Hellemondt G G, van Hooff M L, Spruit M. The accuracy of positioning of a custom-made implant within a large acetabular defect at revision arthroplasty of the hip. Bone Joint J 2015; 97-B(6): 780-5. Baauw M, van Hellemondt G G, Spruit M. A custom-made acetabular implant for Paprosky Type 3 defects. Orthopedics 2017; 40(1): e195-e8. Bonnomet F, Clavert P, Gicquel P, Lefebvre Y, Kempf J F. [Reconstruction by graft and reinforcement device in severe aseptic acetabular loosening: 10 years survivorship analysis]. Rev Chir Orthop Reparatrice Appar Mot 2001; 87(2): 135-46. Buttaro M A, Comba F, Pusso R, Piccaluga F. Acetabular revision with metal mesh, impaction bone grafting, and a cemented cup. Clin Orthop Relat Res 2008; 466(10): 2482-90. Chiarlone F, Zanirato A, Cavagnaro L, Alessio-Mazzola M, Felli L, Burastero G. Acetabular custom-made implants for severe acetabular bone defect in revision total hip arthroplasty: a systematic review of the literature. Arch Orthop Trauma Surg 2020; 140(3): 415-24. Choi H R, Anderson D, Foster S, Beal M, Lee J A, Barr C, Malchau H, McCarthy J, Kwon Y M. Acetabular cup positioning in revision total hip arthroplasty with Paprosky type III acetabular defects: Martell radiographic analysis. Int Orthop 2013; 37(10): 1905-10. Citak M, Kochsiek L, Gehrke T, Haasper C, Suero E M, Mau H. Preliminary results of a 3D-printed acetabular component in the management of extensive defects. Hip Int 2018; 28(3): 266-71. Colen S, Harake R, De Haan J, Mulier M. A modified custom-made triflanged acetabular reconstruction ring (MCTARR) for revision hip arthroplasty with severe acetabular defects. Acta Orthop Belg 2013; 79(1): 71-5. DeBoer D K, Christie M J, Brinson M F, Morrison J C. Revision total hip arthroplasty for pelvic discontinuity. J Bone Joint Surg Am 2007; 89(4): 835-40. Del Gaizo D J, Kancherla V, Sporer S M, Paprosky W G. Tantalum augments for Paprosky IIIA defects remain stable at midterm followup. Clin Orthop Relat Res 2012; 470(2): 395-401. Gelaude F, Clijmans T, Delport H. Quantitative computerized assessment of the degree of acetabular bone deficiency: total radial acetabular bone loss (TrABL). Adv Orthop 2011; 2011: 494382. Gilbody J, Taylor C, Bartlett G E, Whitehouse S L, Hubble M J, Timperley A J, Howell J R, Wilson M J. Clinical and radiographic outcomes of acetabular impaction grafting without cage reinforcement for revision hip replacement: a minimum ten-year follow-up study. Bone Joint J 2014; 96-B(2): 188-94. Goriainov V, McEwan J K, Oreffo R O, Dunlop D G. Application of 3D-printed patient-specific skeletal implants augmented with autologous skeletal stem cells. Regen Med 2018; 13(3): 283-94. Karrholm J, Borssen B, Lowenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 4-7-year stereoradiographic follow-up of 84 cemented prostheses. J Bone Joint Surg (Br) 1994; 76(6): 912-7. Lopez T II, Sanz-Ruiz P, Sanchez-Perez C, Andrade-Albarracin R, Vaquero J. Clinical and radiological outcomes of trabecular metal systems and anti-

protrusion cages in acetabular revision surgery with severe defects: a comparative study. Int Orthop 2018; 42(8): 1811-18. Macheras G, Kateros K, Kostakos A, Koutsostathis S, Danomaras D, Papagelopoulos P J. Eight- to ten-year clinical and radiographic outcome of a porous tantalum monoblock acetabular component. J Arthroplasty 2009; 24(5): 705-9. Myncke I, van Schaik D, Scheerlinck T. Custom-made triflanged acetabular components in the treatment of major acetabular defects: short-term results and clinical experience. Acta Orthop Belg 2017; 83(3): 341-50. Nieuwenhuijse M J, Valstar E R, Kaptein B L, Nelissen R G. The Exeter femoral stem continues to migrate during its first decade after implantation: 10-12 years of follow-up with radiostereometric analysis (RSA). Acta Orthop 2012; 83(2): 129-34. Paprosky W G, Perona P G, Lawrence J M. Acetabular defect classification and surgical reconstruction in revision arthroplasty: a 6-year follow-up evaluation. J Arthroplasty 1994; 9(1): 33-44. 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. Ranstam J, Ryd L, Onsten I. Accurate accuracy assessment: review of basic principles. Acta Orthop Scand 2000; 71(1): 106-8. Sheth N P, Nelson C L, Springer B D, Fehring T K, Paprosky W G. Acetabular bone loss in revision total hip arthroplasty: evaluation and management. J Am Acad Orthop Surg 2013; 21(3): 128-39. Slooff T J, Schimmel J W, Buma P. Cemented fixation with bone grafts. Orthop Clin North Am 1993; 24(4): 667-77. Taunton M J, Fehring T K, Edwards P, Bernasek T, Holt G E, Christie M J. Pelvic discontinuity treated with custom triflange component: a reliable option. Clin Orthop Relat Res 2012; 470(2): 428-34. Templeton J E, Callaghan J J, Goetz D D, Sullivan P M, Johnston R C. Revision of a cemented acetabular component to a cementless acetabular component: a ten to fourteen-year follow-up study. J Bone Joint Surg Am 2001; 83(11): 1706-11. Theil C, Schmidt-Braekling T, Gosheger G, Moellenbeck B, Schwarze J, Dieckmann R. A single centre study of 41 cases on the use of porous tantalum metal implants in acetabular revision surgery. BMC Musculoskelet Disord 2019; 20(1): 238. van Egmond N, De Kam D C, Gardeniers J W, Schreurs B W. Revisions of extensive acetabular defects with impaction grafting and a cement cup. Clin Orthop Relat Res 2011; 469(2): 562-73. van Haaren E H, Heyligers I C, Alexander F G, Wuisman P I. High rate of failure of impaction grafting in large acetabular defects. J Bone Joint Surg Br 2007; 89(3): 296-300. Weeden S H, Schmidt R H. The use of tantalum porous metal implants for Paprosky 3A and 3B defects. J Arthroplasty 2007; 22(6 Suppl. 2): 151-5. Whitehouse M R, Masri B A, Duncan C P, Garbuz D S. Continued good results with modular trabecular metal augments for acetabular defects in hip arthroplasty at 7 to 11 years. Clin Orthop Relat Res 2015; 473(2): 521-7. Wind M A Jr, Swank M L, Sorger J I. Short-term results of a custom triflange acetabular component for massive acetabular bone loss in revision THA. Orthopedics 2013; 36(3): e260-5.

Acta Orthopaedica 2021; 92 (1): 28–34

The impact of socioeconomic status on the utilization of total hip arthroplasty during 1995–2017: 104,055 THA cases and 520,275 population controls from national databases in Denmark Nina M EDWARDS 1, Claus VARNUM 2–6, Søren OVERGAARD 4–6, and Alma B PEDERSEN 1 1 Department 3 Department

of Clinical Epidemiology, Aarhus University Hospital, Aarhus N; 2 Department of Orthopaedic Surgery, Lillebaelt Hospital, Vejle; of Regional Health Research, University of Southern Denmark; 4 Danish Hip Arthroplasty Register; 5 Department of Orthopaedic Surgery and Traumatology, Odense University Hospital, Denmark; 6 Department of Clinical Research, University of Southern Denmark, Denmark Correspondence: nme@clin.au.dk Submitted 2020-06-10. Accepted 2020-09-25.

Background and purpose — In Denmark, all citizens are guaranteed free access to medical care, which should minimize socioeconomic status (SES) inequalities. We examined the association between SES and the utilization of total hip arthroplasty (THA) by age and over time. Patients and methods — Data on education, income, liquid assets, and occupation on 104,055 THA cases and 520,275 population controls were obtained from Danish health registers. We used logistic regression to estimate adjusted odds ratios (aOR) for THA with 95% confidence intervals (CI). Results — Risk (CI) of THA was higher for 45–55-yearolds with lowest vs. highest education (aOR 1.4 [1.3–1.5]), and for those with lowest vs. highest income (aOR 1.1 [1.0– 1.2]). The association between education and income and higher risk of THA decreased with increasing age. The risk of THA was lower for persons with lowest vs. highest liquid assets in all age groups and time periods. The risk of THA was higher for persons with lowest education in 1995–2000 (aOR 1.2 [1.1-1.3]), but diminished in 2013–2017 (aOR 1.0 [1.0–1.0]). For those on lowest income there was a higher risk of THA in 1995–2000 (aOR 1.2 [1.1–1.3]), changing to lower risk in 2013–2017 (aOR 0.8 [0.8–0.9]). Interpretation — In a society where all citizens are guaranteed free access to medical care, we observed a social inequality in regard to the risk of THA with a development over time and in relation to age in most of our SES markers, showing a need for more patient involvement by implementing more focused interventions targeted to the most vulnerable patient groups identified as currently living alone, on low income, and with a low level of liquid assets.

In Denmark, the healthcare system provides tax-supported healthcare for all citizens, guaranteeing free access to medical care for emergency and hospital admission. In spite of this, inequality in healthcare has been found in Denmark in respect of socioeconomic status (SES) in stroke care, in chronic obstructive pulmonary disease, and in outcome among hip fracture patients (Tottenborg et al. 2016, Kristensen et al. 2017, Hyldgard et al. 2019). Low SES is associated with a higher risk of developing hip osteoarthritis (OA) and a lower risk of seeking medical care even in countries with tax-based healthcare systems (Agerholm et al. 2013, Peltola and Järvelin 2014, Wetterholm et al. 2016). Thus, an increased need for total hip arthroplasty (THA) among individuals with low SES may be expected. There are few studies regarding the association between SES and THA, and their results are contradictory. Some have found pro-rich-area inequality and lower rates of THA in patients of lower SES (Agabiti et al. 2007, Cookson et al. 2015, Wetterholm et al. 2016). Other studies have shown similar rates of THA across SES quintiles (Brennan et al. 2012). However, all these studies except Wetterholm et al. were based on grouplevel measures of SES or survey data. No previous study has examined the socioeconomic gradient in THA in Denmark, using individual-level register-based data and whether potential disparities are age- or time-dependent. We conducted a population-based case-control study to examine the association between SES and the utilization of THA across different age groups and over time. We hypothesized that there is a socioeconomic inequality in THA utilization in Denmark.

Acta Orthopaedica 2021; 92 (1): 28–34

Patients and methods Setting All Danish citizens are assigned a unique civil registration (CRP) number at birth, which is included in all Danish registers, allowing for unambiguous individual-level record linkage between the registers and enabling virtually complete longterm follow-up of all Danish inhabitants (Schmidt et al. 2014). We used prospectively collected data from health registers, which encompassed the entire Danish population. Data sources The Danish Civil Registration System (DCRS) contains information on the CPR number, vital and migrant status, cohabiting status, and municipality of residence (Schmidt et al. 2014). The Danish Hip Arthroplasty Register (DHR) is a national clinical quality database established in 1995 (Gundtoft et al. 2016). The main variables are type of operation (primary or revision), operation side, primary diagnosis, and operation date (Gundtoft et al. 2016). Completeness of the DHR has been high from the beginning, being between 91% and 98% for primary THA (Gundtoft et al. 2016, Danish Hip Arthroplasty Register: Annual Reports 2019). In addition, the quality of registered data is high (Pedersen et al. 2004). The Danish National Patient Registry (DNPR) is an administrative register covering all somatic admissions to all Danish hospitals since 1977 and outpatient and emergency room visits since 1995. Information reported to the DNPR includes admissions, diagnoses, treatments, and examinations. Diagnoses are reported to the DNPR according to the ICD-8 (from 1977 to 1993) or ICD-10 (starting in 1993) (Schmidt et al. 2015). Statistics Denmark is a collection of register data that contains detailed individual-level information on socioeconomic characteristics on all Danish citizens. The Income Statistics Register includes information regarding family annual household income and liquid assets and the data are primarily supplied by tax authorities. The Population Education Register obtains information on the highest completed level of education and consists of data generated from administrative records of educational institutions and from surveys. The Registerbased Labour Force Statistics (RAS) obtains a description of the affiliation with the labour market. The registers are administered by the Danish government and are updated yearly. Study population We used the DHR to identify all patients over the age of 45 undergoing primary THA in Denmark from January 1, 1995 to December 31, 2017 and diagnosed with primary hip osteoarthritis (OA) (cases), using THA as a surrogate of the most severe hip OA cases. Only the first THA during 1995–2017 was included in the study cohort. The date of THA surgery was considered as the index date and the same date was used to identify the population controls over the age of 45 by

matching for sex and region of residence on the index date. We used the DCRS to randomly select 5 population controls for each THA case. Socioeconomic status For each THA case and population control we retrieved information on marital status, cohabitation, highest obtained education, family income, and occupation. In addition, SES was measured with family liquid assets on the index date. Highest obtained education was classified into 3 categories: low, defined as none or elementary school; medium, defined as more than elementary school, but less than university completed; and high, defined as university degree completed. Since a large proportion of the THA patients are senior citizens (> 65 years of age) with a state pension, family liquid assets was used to describe SES in patients > 65 years of age, whereas family income was used to assess SES in patients < 65 years of age (Robert and House 1996). This provides a more accurate estimate of overall socioeconomic stratification than using income and liquid assets through all age groups separately (Robert and House 1996). We obtained information on family income and liquid assets for the 5 years prior to surgery. To account for yearly variations in income and liquid assets, we calculated the average yearly total income and liquid assets in the 5 years prior to surgery for the patient and cohabiting partner. The family mean income and liquid assets were categorized into tertiles of increasing amount: low, medium, and high. Occupation was divided into the following categories: Director/chief executive, employer/self-employed, skilled worker, unskilled worker, unemployed, early retirement/pension, benefits and others (Table 1, Supplementary data). Because of low numbers, we regrouped occupation to include only retirement, working, and others in the age groups 76–85 and > 85. Covariates Information on age and sex for both cases and controls was collected from the DCRS. Comorbidities were obtained from the DNPR. We summarized the 10-year pre-surgery or pre-index-date hospital comorbidity history for each case and population control (Table 2, Supplementary data). We measured the comorbidity status by Charlson Comorbidity Index (CCI) score based on the 19 disease categories and defined 3 levels of the CCI: a score of 0 (low), given to patients with no known comorbidities included in the CCI; a score of 1–2 (medium); and a score of 3 or more (high) (Schmolders et al. 2015). Statistics In order to describe the SES characteristics of cases and controls we calculated proportions of cases and controls with each specific marker, overall, and by age categories and year of surgery. Odds ratios (OR) with 95% confidence interval (CI) for THA according to each marker of SES were calculated using conditional logistic regression. We calculated crude odds

Acta Orthopaedica 2021; 92 (1): 28–34

Primary THAs in 1995–2017 n = 132,994

Matched controls (1:5) n = 664,970

Excluded Other diagnosis than primary hip OA n = 28,939 Study cases n = 104,055

Excluded Controls matched to excluded cases n = 144,695 Controls n = 520,275

Figure 1. Flowchart of cases and controls.

ratios and odds ratios adjusted for age, SES markers independently, calendar year, and CCI. Sensitivity analyses were done when stratifying for age and calendar year. The statistical analyses were performed in Stata version 15 (StataCorp, College Station TX, USA) and R version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria). Ethics, funding, and potential conflicts of interest The study was approved by the Danish Data Protection Agency’s journal number 2015-57-0002 and Aarhus University’s journal number 2016-051-000001, record number 880. This study was reported following the STROBE and RECORD guidelines. We would like to acknowledge support from the Orthopaedic Research Fund, AP Møller Fund, and Aase and Ejnar Danielsens Fund. The funders had no role in the study design, data collection, and analysis, or in the preparation of the manuscript. The authors report no conflict of interest.

Results Description of the study population (Tables 3–6 in Supplementary data) From 1995 to 2017, we identified 132,994 patients who received a THA and 664,970 population controls. We excluded 28,939 THA cases due to a diagnosis other than primary hip OA, as well as 144,695 corresponding matched population controls. The final study population included 104,055 THA cases and 520,275 controls (Figure 1). 56% of the study population were females. The mean age was 70 years (45–99) in the case group and 62 years (45–107) in the population control group. Further, cases had a lower education and a lower mean income, and there was a higher percentage of cases on early retirement or old-age pension (Table 7). Similar distribution is seen in the age- and year-stratified data. Overall associations (Table 8) Persons who never married had a lower risk of receiving a THA than patients who married (aOR 0.8 [CI 0.7–0.8]). The same association was seen in patients who were divorced (aOR 1.0 [CI 0.9–1.0]) or were widow/widower (aOR 0.9 [CI 0.9–0.9]). Persons who lived alone also had a lower risk of receiving a THA than cohabiting persons (aOR 0.9 [CI 0.9–1.0]).

Table 7. Demographic characteristics of total hip arthroplasty cases and population controls from the background population. Values are count (%) unless otherwise specified Factor

Cases Controls n = 104,055 n = 520,275

Female sex Mean age (range) Marital status Never married Married Divorced Widow/widower Cohabiting status Alone Cohabitant Other a Education Low Medium High Missing Income, tertiles Low Medium High Liquid assets, tertiles Low Medium High Missing Occupation Director/chief executive Employer/self-employed Skilled worker Unskilled worker Unemployed Early retirement/pension Benefits/public support Other Charlson comorbidity score Low Medium High

58,422 (56) 69,7 (45–99)

292,100 (56) 62.3 (56–107)

6,541 (6) 63,081 (61) 12,085 (12) 22,348 (21)

55,931 (11) 322,316 (62) 72,684 (14) 69,344 (13)

33,143 (32) 63,771 (61) 7,087 (7)

146,802 (28) 327,206 (63) 45,738 (9)

43,247 (42) 36,964 (36) 15,788 (15) 8,036 (8)

175,473 (34) 212,530 (41) 98,134 (19) 34,138 (7)

46,310 (43) 38,526 (35) 24,086 (22)

159,376 (31) 172,510 (33) 188,183 (36)

31,674 (30) 33,394 (32) 35,645 (34) 3,342 (3)

163,356 (31) 166,125 (32) 168,505 (33) 22,289 (4)

6,577 (6) 8,357 (8) 7,835 (8) 2,286 (2) 1,125 (1) 72,310 (69) 686 (1) 4,777 (5)

29,437 (6) 95,612 (18) 85,295 (16) 22,793 (4) 15,408 (3) 230,607 (44) 6,141 (1) 34,940 (7)

73,288 (70) 14,861 (14) 5,506 (5)

397,610 (77) 593,061 (11) 23,521 (4)

SD: Standard deviation. a “Other” accounts for people coded as not having children living at home and people coded as households with multiple families.

Individuals with the lowest vs. highest level of education had a higher risk of receiving a THA (aOR 1.1 [CI 1.1–1.1]). The same was seen in those with the lowest vs. highest income (aOR 1.2 [CI 1.2–1.2]). The lowest vs. highest liquid asset was associated with lower risk of receiving a THA (aOR 0.7 [CI 0.7–0.7]). There was a higher risk of receiving a THA when the persons’ occupations were employer/self-employed, skilled worker, unskilled worker, early retirement/pension, and when receiving benefits/public support than being a director or chief executive. Age trends (Figure 2, Tables 9–11 in Supplementary data) The reduced risk of receiving a THA among persons who were never married was seen in all age groups. Widows/widowers

Acta Orthopaedica 2021; 92 (1): 28–34

Table 8. Overall crude and adjusted odds ratios (OR) with 95% confidence intervals (CI) for THA during 1995–2017

Unadjusted OR (CI) Adjusted OR a (CI)

Marital status Never married Married Divorced Widow/widower Cohabiting status Alone Cohabitant Education, tertiles Low Medium High Income, tertiles Low Medium High Liquid assets, tertiles Low Medium High Occupation Director/chief executive Employer/self-employed Skilled worker Unskilled worker Unemployed Early retirement/pension Benefits/public support Other a Adjusted

Adjusted OR – marital status

Adjusted OR – cohabiting status 1.1

1.2

1.0

0.60 (0.58–0.61) 1 (reference) 0.85 (0.83–0.87) 1.65 (1.62–1.68)

0.76 (0.74–0.79) 1 (reference) 0.96 (0.93–0.99) 0.89 (0.87–0.92)

1.16 (1.14–1.18) 1 (reference)

0.93 (0.91–0.96) 1 (reference)

1.53 (1.50–1.56) 1.08 (1.06–1.10) 1 (reference)

1.08 (1.05–1.10) 1.00 (0.98–1.02) 1 (reference)

2.19 (2.14–2.22) 1.72 (1.69–1.75) 1 (reference)

1.18 (1.15–1.21) 1.19 (1.16–1.21) 1 (reference)

0.9

0.8 0.8 0.6

Divorced Never married Widow/widower Married (reference)

45–55

56–65

0.7

66–75

76–85

Age group

>85

Adjusted OR – level of education

Living alone Cohabiting (reference)

45–55

56–65

66–75

76–85

Age group

>85

Adjusted OR – level of income 1.4

1.4 1.2 1.2

0.92 (0.90–0.93) 0.95 (0.93–0.97) 1 (reference)

0.73 (0.71–0.74) 0.87 (0.85–0.88) 1 (reference)

1 (reference) 2.56 (2.47–2.65) 1.05 (1.02–1.09) 1.20 (1.14–1.26) 0.84 (0.78–0.89) 3.59 (3.50–3.67) 1.28 (1.18–1.39) 1.56 (1.51–1.62)

1 (reference) 1.74 (1.68–1.81) 1.07 (1.03–1.11) 1.17 (1.12–1.23) 0.94 (0.88 – 1.00) 1.74 (1.68–1.79) 1.36 (1.25–1.48) 1.31 (1.26–1.36)

1.0 1.0

0.8

Low Medium High (reference)

45–55

56–65

66–75

76–85

Age group

>85

Low Medium High (reference)

45–55

56–65

66–75

76–85

Age group

>85

Adjusted OR – level of income (age ≤ 65) and liquid assets (age ≥ 66)

Adjusted OR – level of liquid assets

1.2

1.0

for age, SES markers independently, calendar year, and CCI. 0.9

had a higher risk of receiving a THA than married persons in the age groups 56–65 and 66–75. The lower risk of receiving a THA when living alone was not consistent throughout the age groups. Persons with the lowest level of education had a higher risk of receiving a THA in the age group 45–55 and 56–65; this was not the case in the age group > 85. There was an association between lowest income and a higher risk of receiving a THA in the age group 45–55, but not in the age group 56–65+. The same decreasing age trend followed in the age group 66–75 with the lowest liquid assets, further decreasing the risk in the age group > 85. In the age group 45–55, being an employer/self-employed was associated with a higher risk of receiving a THA than being a director/chief executive. This association was sustained throughout the age groups. Early retirement was associated with a higher risk of receiving a THA in all age groups under the age of 75, but not above 75 years. Time trends (Figure 3, Table 12 in Supplementary data) Marital status showed a time-dependent association, moving from estimates in near proximity of each other in the years 1995–2000 to spreading and thereby more inequality in the years 2013–2017. There was a slightly reduced risk of receiving a THA in the years 1995–2000 when persons were divorced,

1.0

0.8 0.8 0.7

0.6

Low Medium High (reference)

45–55

56–65

0.6 66–75

76–85

Age group

>85

Adjusted OR – occupation 4.0

3.0

Low Medium High (reference)

45–55

56–65

66–75

76–85

Age group

>85

Adjusted OR – occupation

Employer/self-employed Skilled worker Unskilled worker Unemployed Early retirement/pension Benefits/public support Other Director/chief executive (reference)

Retirement Other Working (reference)

2.0 10 1.0 1

0 45–55

56–65

Age group

66–75

76–85

Age group

>85

Figure 2. Dot plot for the adjusted OR with 95% confidence intervals, age stratified and adjusted for age, SES markers independently, calendar year, and CCI. Striped line: reference group.

but this association was time-dependent, changing to having a higher risk of receiving a THA in the years 2013–2017. The opposite time trend was seen for persons living alone.

Acta Orthopaedica 2021; 92 (1): 28–34

Adjusted OR – marital status

Adjusted OR – cohabiting status

Discussion

1.1

1.0

0.9

0.8

Divorced Never married Widow/widower Married (reference)

0.9

1995–2000 2001–2006 2007–2012 2013–2017

Year

Adjusted OR – level of education

Living alone Cohabiting (reference)

Year

Adjusted OR – level of income

Low Medium High (reference)

1.2

Low Medium High (reference)

1.2

1.1 1.0 1.0

0.8 1995–2000 2001–2006 2007–2012 2013–2017

1995–2000 2001–2006 2007–2012 2013–2017

Year

Adjusted OR – level of liquid assets

Year

Adjusted OR – occupation

1.0

2.2

0.9

Employer/self-employed Skilled worker Unskilled worker Unemployed Early retirement/pension Benefits/public support Other Director (reference)

1.8 0.8

1.4 Low Medium High (reference)

0.7

1995–2000 2001–2006 2007–2012 2013–2017

Year

1.0 1995–2000 2001–2006 2007–2012 2013–2017

Year

Figure 3. Dot plot for adjusted OR with 95% confidence intervals, stratified for calendar year and adjusted for age, SES markers independently, calendar year, and CCI. Striped line: reference group.

Obtaining the lowest level of education was associated with a higher risk of receiving a THA in the years 1995–2000 than in the years 2013–2017. Likewise, the lowest income group was associated with a higher risk of receiving a THA than the highest income group in 1995–2000, but the difference vanished in the period 2013–2017. The association between the lowest liquid assets and lower risk of receiving a THA was stable throughout the entire study period. Patients who were an employer/self-employed were more likely to receive a THA in the years 1995–2000. This association decreased throughout the period in 2013–2017. Retiring early and receiving an old-age pension were associated with an increased risk of receiving a THA irrespective of the calendar year.

In this large nationwide case-control study of 104,055 THA patients, we observed a substantial socioeconomic inequality in THA utilization. Married and cohabiting persons were at increased risk of receiving a THA. The association between low education, low income, and higher risk of THA was observed among patients 45–55 years of age, but decreased with increasing age. The inequality in the risk of THA by education decreased over calendar time, whereas the inequality by income and liquid assets was persistent. Cohabitation/social support Our finding of the association between living alone or being divorced/never married/widow/widower and lower risk of THA could be due to those in need of THA often worrying about becoming reliant on family and friends for their daily activities (Mota et al. 2012), or just a lack of social and family support to assist in medical decision-making (Youm et al. 2015). There may also be a difference in willingness in regard to social support, where 1 study found OA patients to be less willing to undergo surgery if they were unmarried or beyond the age of retirement, which support our results (Mota et al. 2012). Education and income Low education and lower income are associated with an increased OA severity and increased potential need for THA, as also found by Cleveland et al. (2013). It has been suggested that individuals with higher education are better able to process information regarding healthy behaviours and therefore may be able to postpone the need for a THA, equalizing the risk in the higher age groups (Brennan et al. 2012, Cleveland et al. 2013). However, was there supposed to be a greater disparity between level of education and income, and in the risk of THA, revealing a possibly greater unmet need for THA. (Cleveland et al. 2013, Wetterholm et al. 2016). Occupation and liquid assets There is a clear dose–response relationship between occupational workload and increased risk of hip OA and therefore a greater risk of THA (Sun et al. 2019). This may explain some of our findings regarding occupation and higher risk of THA, when the patient’s occupation was employer/selfemployed, skilled worker or unskilled worker in comparison with a potentially lower manual workload such as a position as a director or chief executive. Wetterholm et al. (2016) have shown similar results. This could further fit with our finding that lower liquid assets in persons above 65 years are a predictor of THA risk. Previous studies found that patients receiving benefits were twice as likely to require THA, but were also less likely to have had surgery (Ackerman and Busija 2012). Individuals who retire early or are on a pension have a higher

Acta Orthopaedica 2021; 92 (1): 28–34

risk of THA, but also show greater disease severity before THA (Dieppe et al. 2009). Time trends Evaluation of time trends is important in order to ensure that the effort made by the Danish government to ensure equal access to healthcare irrespective of social position is working. Looking at the SES markers marital status and income, we found more inequality in the later years. In contradiction with this, Cookson et al. (2015) showed a pro-rich-area inequality in Europe, though the inequality was not significant in Denmark. However, their study has several limitations including missing data on the individual-level measures of SES, and their data on income was measured using subnational administrative areas, which could cause misclassification of SES. In relation to education, we found a time trend from a wide spread of estimates in the years 1995– 2000 to a narrowing of estimates in 2013–2017, indicating less social inequality in the latter years. Our study confirm that SES is a complex combination of an individual’s education, income, and occupation. These various SES markers may operate through different mechanisms to affect the risk of THA, including the potential for SES to influence lifestyle behaviours, preventive healthcare, and health management (Cleveland et al. 2013). Methodological considerations Strengths of our study include the prospective data collection, where information on SES markers was collected from registers on an individual basis. The few missing SES data were distributed equally between cases and controls, and at random, which is why we believe that this at most could give an overestimation of our results. Unlike previous studies, we were able to include liquid assets as a SES marker providing a more accurate estimate of SES and prosperity for the individuals older than 65. This, however, introduces an apparent paradox where we see contradicting estimates between income and liquid assets. This limitation is considered when age stratifying the markers and joining them into Figure 2: level of income and liquid assets, thereby not pooling all ages in a variable where age has a major influence. A limitation of our study is that we have no information regarding potential confounders such as lifestyle factors like smoking, alcohol, BMI, physical activity (Mota et al. 2012, Weiss et al. 2019). Another limitation is that the threshold for receiving a THA has changed with time and age possibly due to a change in surgeon-related factors as well as a change in the demands from society. This could influence our THA utilization in the age- and time-trend analysis. However, this does not have an impact on the distribution of cases in regard to the SES markers, leaving our results robust as regards this factor. In conclusion, we found an age-dependent association between living alone, lower level of education, and lower

level of income, and a high risk of THA. We also found that the inequality seen in the risk of THA by education decreased over time, suggesting a time trend towards less social inequality, while the inequality seen in the risk of THA by marital status and income increased over time. There is a development over time and in relation to age in most of our SES markers, showing a need for more patient involvement by implementing more focused interventions targeted to the most vulnerable patient groups identified as currently living alone, with low income and low level of liquid assets. Supplementary data Tables1–6 and 9–12 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2020.1840111 NME drafted the manuscript. NME, CV, SO, and AO conceived and designed the study, interpreted the results, and revised the manuscript. Acta thanks Ian Harris and Martin Englunds for help with peer review of this study.

Ackerman I N, Busija L. Access to self-management education, conservative treatment and surgery for arthritis according to socioeconomic status. Best Pract Res Clin Rheumatol 2012; 26: 561-83. Agabiti N, Picciotto S, Ceraroni G, Bisanti L, Forastiere F, Onorati R, Pacelli B, Pandolfi P, Russo A, Spadea T, Perucci C A. The influence of socioeconomic status on utilization and outcomes of elective total hip replacement: a multicity population-based longitudinal study. Int J Qual Health Care 2007; 19: 37-44. Agerholm J, Bruce D, Ponce de Leon A, Burstrom B. Socioeconomic differences in healthcare utilization, with and without adjustment for need: an example from Stockholm, Sweden. Scand J Public Health 2013; 41: 318-25. Brennan S L, Stanford T, Wluka A, Henry M J, Page R S, Graves S E, Kotowicz M A, Nicholson G C, Pasco J A. Cross-sectional analysis of association between socioeconomic status and utilization of primary total hip joint replacements 2006–7: Australian Orthopaedic Association National Joint Replacement Registry. BMC Musculoskelet Disord 2012; 13: Art. No. 63. Cleveland R J, Schwartz T A, Prizer L P, Randolph R, Schoster B, Renner J B Jordan, J M, Callahan L F. Associations of educational attainment, occupation, community poverty with hip osteoarthritis. Arthritis Care Res (Hoboken) 2013; 65: 954-61. Cookson R, Gutacker N, Garcia-Armesto S, Angulo-Pueyo E, Christiansen T, Bloor K, Bernal-Delgado E. Socioeconomic inequality in hip replacement in four European countries from 2002 to 2009: area-level analysis of hospital data. Eur J Public Health 2015; 25(Suppl 1): 21-7. Danish Hip Arthroplasty Register: Annual Reports; 2019. Available from http://danskhoftealloplastikregisterdk/en/publications/annual-reports/ Dieppe P, Judge A, Williams S, Ikwueke I, Guenther K P, Floeren M, Huber J, Ingvarsson T, Learmonth I, Lohmander L S, Nilsdotter A, Puhl W, Rowley D, Thieler R, Dreinhoefer K, Eurohip Study Group. Variations in the preoperative status of patients coming to primary hip replacement for osteoarthritis in European orthopaedic centres. BMC Musculoskelet Disord 2009; 10: 19. Gundtoft P H, Varnum C, A Pedersen B, Overgaard S. The Danish Hip Arthroplasty Register. Clin Epidemiol 2016; 8: 509-14. Hyldgard V B, Johnsen S P, Stovring H, Sogaard R. socioeconomic status and acute stroke care: has the inequality gap been closed? Clin Epidemiol 2019; 11: 933-41.

Kristensen P K, Thillemann T M, Pedersen A B, Søballe K, Johnsen S P. Socioeconomic inequality in clinical outcome among hip fracture patients: a nationwide cohort study. Osteoporos Int 2017; 28: 1233-43. Mota R E, Tarricone R, Ciani O, Bridges J F, Drummond M. Determinants of demand for total hip and knee arthroplasty: a systematic literature review. BMC Health Serv Res 2012; 12: 225. Pedersen A, Johnsen S, Overgaard S, Soballe K, Sorensen H T, Lucht U. Registration in the Danish Hip Arthroplasty Registry: completeness of total hip arthroplasties and positive predictive value of registered diagnosis and postoperative complications. Acta Orthop Scand 2004; 75: 434-41. Peltola M, Järvelin J. Association between household income and the outcome of arthroplasty: a register-based study of total hip and knee replacements. Arch Orthop Traums Surg 2014; 134: 1767-74. Robert S, House J S. SES differentials in health by age and alternative indicators of SES. J Aging Health 1996; 8: 359-88. Schmidt M, Pedersen L, Sorensen H T. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 2014; 29: 541-9. Schmidt M, Schmidt S A, Sandegaard J L, Ehrenstein V, Pedersen L, Sorensen H T. The Danish National Patient Registry: a review of content, data quality, research potential. Clin Epidemiol 2015; 7: 449-90.

Acta Orthopaedica 2021; 92 (1): 28–34

Schmolders J, Friedrich M J, Michel R, Strauss A C, Wimmer M D, Randau T M, Pennekamp P H, Wirtz D C, Gravius S. Validation of the Charlson comorbidity index in patients undergoing revision total hip arthroplasty. Int Orthop 2015; 39: 1771-7. Sun Y, Nold A, Glitsch U, Bochmann F. Exposure–response relationship and doubling risk doses: a systematic review of occupational workload and osteoarthritis of the hip. Int J Environ Res Public Health 2019; 16(19): 3681. Tottenborg S S, Lange P, Johnsen S P, Nielsen H, Ingebrigtsen T S, Thomsen R W. Socioeconomic inequalities in adherence to inhaled maintenance medications and clinical prognosis of COPD. Respir Med 2016; 119: 160-67. Weiss R J, Karrholm 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: 264-69. Wetterholm M, Turkiewicz A, Stigmar K, Hubertsson J, Englund M. The rate of joint replacement in osteoarthritis depends on the patient’s socioeconomic status. Acta Orthop 2016; 87: 245-51. Youm J, Chan V, Belkora J, Bozic K J. Impact of socioeconomic factors on informed decision making and treatment choice in patients with hip and knee OA. J Arthroplasty 2015; 30: 171-5.

Acta Orthopaedica 2021; 92 (1): 35–38

Low revision rate of dual mobility cups after arthroplasty for acute hip fractures: report of 11,857 hip fractures in the Dutch Arthroplasty Register (2007–2019) Esther M BLOEMHEUVEL 1, Liza N VAN STEENBERGEN 2, and Bart A SWIERSTRA 2 1 Department of Orthopedic Surgery, Sint Maartenskliniek, Nijmegen; 2 Dutch Arthroplasty Register (LROI), ’s Hertogenbosch, the Netherlands Correspondence: esther.bloemheuvel@gmail.com Submitted 2020-07-15. Accepted 2020-09-26.

Background and purpose — Dislocation is one of the most frequent reasons for cup revision after total hip arthroplasty (THA) for an acute fracture. A dual mobility cup (DMC) might reduce this risk. We determined the cup revision rate after THA for an acute fracture according to type of cup. Patients and methods — All THAs for an acute fracture registered in the Dutch Arthroplasty Register (LROI) during 2007–2019 were included (n = 11,857). Type of cup was divided into DMC and unipolar cup (UC). Competing risk analyses were performed with cup revision for any reason as endpoint. Multivariable Cox regression analyses with outcome cup revision were performed adjusted for sex, age, ASA class, and surgical approach, stratified for UC THA with femoral head size of 32 mm and 22–28 mm. Results — A DMC was used in 1,122 (9%) hips. The overall 5-year cup revision rate for any reason after THA for acute fracture was 1.9% (95% CI 1.6–2.2). Cup revision for dislocation within 5 years was performed in 1 of 6 DMC THAs versus 108 of 185 (58%) UC THAs. Univariable Cox regression analyses showed no statistically significant difference in cup revision rate between DMC and UC (HR = 0.8; CI 0.4–1.5). Multivariable Cox regression analyses showed lower risk of cup revision in DMC THA (n = 1,122) compared with UC THA with 22–28 mm femoral head size (n = 2,727) (HR = 0.4; CI 0.2–0.8). Interpretation — The 5-year cup cumulative incidence of revision after THA for acute fracture was comparable for DMC and UC THA. However, DMC THA had a lower risk of cup revision than UC THA with 22–28 mm femoral head.

The risk for revision in case of total hip arthroplasty (THA) after an acute fracture is higher than after hemiarthroplasty (Parker et al. 2010). Dislocation is one of the most frequent reasons for cup revision after an acute fracture (Gjertsen et al. 2007). We have shown low cup revision rates for dislocation using dual mobility cup (DMC) THA in patients with osteoarthritis (Bloemheuvel et al. 2019). The use of DMC in THA after an acute fracture might therefore be beneficial to prevent this complication. At the same time also femoral head size (in unipolar cups [UC]) and surgical approach influence the risk of revision for dislocation (Byström et al. 2003, Hailer et al. 2012, Kostensalo et al. 2013, Zijlstra et al. 2017). We hypothesized that the cup revision rate for dislocation in THA for acute fracture is lower with DMC than UC but that this can be affected by femoral head size (in UC) and surgical approach. We therefore determined the cup revision rate because of dislocation after THA for an acute fracture according to type of cup and head size.

Patients and methods The Dutch Arthroplasty Register (LROI) started in 2007 and has a completeness of 98% for primary and revision hip arthroplasty (www.lroi-report.nl). The LROI database contains patient, procedure, and prosthesis characteristics. For each component a product number is registered to identify the characteristics of the prosthesis, such as dual mobility or unipolar cup. The vital status of all patients is obtained actively on a regular basis from Vektis, the national insurance database on health care in the Netherlands, which records all deaths of Dutch citizens (www.vektis.nl).

Acta Orthopaedica 2021; 92 (1): 35–38

THA for an acute fracture 2007–2019 n = 13,107 Excluded (n = 1,250): – missing data, 1,061 – MoM articulation,189 DMC n = 1,122

Yearly number of DMC 300

200

UC n = 10,735

Figure 1. Patient flow.

For this study we included all primary THAs in the period 2007–2019 with a diagnosis of acute fracture. A cup revision was defined as a procedure where at least the cup or the cup and liner were exchanged or removed. Closed reduction after a dislocation or incision and drainage for infection without component exchange are not included in the LROI. Records with a missing cup product number (n = 1,061) and metal-on-metal hip arthroplasties were excluded (n = 189). 11,857 primary THAs were included and divided into 2 groups: DMC THA and UC THA (Figure 1). Statistics UC THA and DMC THA were described separately concerning patient and procedure characteristics. Survival time was calculated as the time from primary THA to cup revision for any reason, death of the patient, or end of follow-up (December 31, 2019). Cumulative crude incidence of cup revision was calculated using competing risk analysis, where death was considered to be a competing risk (Lacny et al. 2015, Wongworawat et al. 2015). Multivariable Cox regression analyses were performed to compare DMC and UC THA. Adjustments were made for sex, age, ASA class, and surgical approach and stratified by UC femoral head size (22–28 mm and 32 mm). BMI and smoking status were not included as covariates, since these have only been available in the LROI database since 2014. For all covariates added to the model, the proportional hazards assumption was checked by inspecting log-minus-log curves and met. Reasons for cup revision were described and compared using a chi-square test. P-values below 0.05 were considered statistically significant. For the 95% confidence intervals (CI), we assumed that the number of observed cases followed a Poisson distribution. Ethics, data sharing, funding, and potential conflicts of interests The LROI uses the opt-out system to require the informed consent of patients. The dataset was processed in compliance with the regulations of the LROI governing research on registry data. Data are available from the LROI but restrictions apply to the availability of these data, which were used

100

0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019

Year of primary THA

Figure 2. Use of DMC THA in case of an acute fracture in the period 2009–2019 in the Netherlands.

under license for the current study. No external funding was received. No competing interests were declared.

Results 11,857 THAs for acute fracture were included. In 9% a DMC THA and in 91% a UC THA was used. The median follow-up was 3.4 years (0–13), with 35% of records having a follow-up period of 5 years or longer. Of all included acute fracture THA patients, 26% (CI 22–31) in the DMC THA group died and 16% (CI 15–17) in the UC THA groups died within 5 years of the primary procedure. The use of a DMC THA in acute fracture patients increased from 15 in 2009 (3% of all THAs) to 299 (18% of all THAs) in 2019 (Figure 2). The mean age was 70 years in both groups. The proportion ASA class III–IV was higher in the DMC THA group (40%) compared with the UC DMC group (24%). In 70% the DMC THA was cemented compared with 32% in the UC THA group. The most frequent approach was posterolateral in both groups (Table 1). In the UC THA group, most often a 32 mm head was used (51%). There were 2,727 (26%) small-sized heads used (22–28 mm) and 23% had a 36 mm head size. The overall 5-year cumulative incidence of cup revision rate for any reason after THA for acute fracture was 1.9% (CI 1.6–2.2) with 6 of 1,122 cup revisions for DMC THA and 185 of 10,735 cup revisions for UC THA. The 5-year cumulative incidence of cup revision rate for DMC THA was 1.0% (CI 0.4–3.0) and 2.0% (CI 1.7–2.3) for UC THA (Figure 3). In UC THA with 36 mm heads the 5-year cumulative incidence of cup revision rate was 1.4% (CI 0.9–2.0) and for UC THA with 32 mm heads this was 1.7% (CI 1.3–2.1), while for UC THA with 22–28 mm heads this was 2.7% (CI 2.2–3.4). Univariable as well as multivariable Cox regression analyses showed a statistically significant lower risk for cup revi-

Acta Orthopaedica 2021; 92 (1): 35–38

Cumulative cup revision percentage

Table 2. Multivariable Cox regression analysis for DMC THA and UC THA compared with UC THA 32 mm femoral head and 22–28 mm femoral head in acute fracture patients

2.5 Unipolar cup THA Dual mobility cup THA 2.0

1.5

1.0

Type of hip prosthesis

crude HR

adjusted HR a

DMC THA UC THA 32 mm head

1,122 5,380

0.7 (0.4–1.4) 1.0 (ref.)

0.6 (0.3–1.2) 1.0 (ref.)

DMC THA 1,122 UC THA 22–28 mm head 2,727

0.5 (0.2–0.9) 1.0 (ref.)

0.4 (0.2–0.8) 1.0 (ref.)

Adjusted for age, sex, ASA classification, and surgical approach.

0.5

Table 3. Reason for cup revision within 5 years according to type of cup 0

Years after primary THA

Figure 3. Crude cumulative overall cup revision rate of THAs for acute fracture according to type of cup. Table 1. Patient characteristics of THAs for acute fracture according to type of cup (n = 11,857). Values are count (%) unless otherwise specified Factor

DMC THA n = 1,122

Sex Male 382 (34) Female 738 (66) 70 (52–86) Age, median (p5–p95) a Previous operation on affected hip Yes 108 (10) No 1,002 (90) ASA score I 75 (7) II 595 (53) III–IV 448 (40) Fixation Cemented 776 (70) Hybrid (acetabulum cemented) 135 (12) Hybrid (femur cemented) 58 (5) Cementless 137 (13) Approach Anterior 61 (5) Anterolateral 10 (1) Direct lateral 84 (8) Posterolateral 955 (85) Other 8 (1) Head diameter, mm 22–28 1,094 (100) 32 3 (0) 36 ≥ 38

UC THA n = 10,735 3,324 (31) 7,395 (69) 70 (54–84) 650 (6) 9,749 (94) 1,702 (16) 6,153 (59) 2,531 (25) 3,344 (32) 327 (3) 997 (9) 5,923 (56) 1,142 (11) 874 (8) 2,345 (22) 6,240 (59) 34 (0) 2,727 (26) 5,380 (51) 2,382 (23) 60 (0)

a 5th

percentile to 95th percentile DMC: dual mobility cup; UC: unipolar cup. Numbers do not add up to total due to missing data.

sion in the DMC THA group compared with UC THA with a 22–28 mm femoral head (HRadjusted 0.4 [0.2–0.8]), but no statistically significant difference in cup revision rate between DMC and UC THA with a 32 mm femoral head (HRadjusted 0.6 [CI 0.3–1.2]) (Table 2).

Factor ! All cup revisions within 5 years Reason for revision a Dislocation Infection Wear Periprosthetic fracture Loosening femoral component Loosening acetabular component Peri-articular ossification Other

DMC THA n = 1,122

UC THA n = 10,735

185

1 3 0 0 1 4 1 2

108 23 5 13 18 29 2 32

DMC: Dual mobility cup; UC: Unipolar cup. a The sum is higher than the total amount since more than 1 reason for revision can be registered.

1 of 6 DMC THAs were revised for dislocation versus 108 of 185 (58%) UC THAs. (Suspicion of) infection (3/6) and cup loosening (4/6) were other registered reasons for cup revision in the DMC group, compared with 23/185 (12%) and 29/185 (16%) in the UC group (Table 3).

Discussion We found that DMC is increasingly used in THA for acute fractures. The clinicians’ expectation to reduce the risk for dislocation is the most probable reason to use this more expensive cup. We found 6 cup revisions within 5 years when a DMC THA was used, and only 1 of these 6 was revised for dislocation. Our focus on short-term revision rates is justified as the majority of dislocations occur early after the index operation (Enocson et al. 2009). In the Nordic Arthroplasty Register Association (NARA) a reduced revision risk for DMC in THA for acute femoral neck fracture has been shown by Jobory et al. (2019). They matched 4,520 hip fractures treated with a DMC THA to 4,520 hip fractures with UC THA and found a lower risk for cup revision for dislocation for DMC, with a hazard ratio of 0.32 adjusted for approach. However, they only included

head size 32 and 36 mm in contrast to our study in which head sizes 22–28 mm were included as well. Tabori-Jensen et al. (2019) found low dislocation rates of DMC THA after acute femoral neck fracture in a cohort study of 966 hips. After mean 5.4 years follow-up, 8 cups were revised, 3 due to repeated dislocations. Their findings are comparable to our results. We found a statistically significant lower risk for cup revision in the DMC THA group compared with UC THA with a 22–28 mm femoral head. This is in accordance with our hypothesis and with the findings of Kostensalo et al. (2013), based on data from the Finnish Arthroplasty Register, who found a reduced dislocation revision rate in head sizes > 28 mm. Comparable results were found in register studies from Norway (Byström et al. 2003), Sweden (Hailer et al. 2012) and the Netherlands (Zijlstra et al. 2017). Our hypothesis that surgical approach might influence the (cup) revision rate could not be confirmed. This influence has been shown in another recent LROI study by Moerman et al. (2018), who found that posterolateral approach was a risk factor compared with other approaches (HR 1.0 versus 0.7) for revision in case of THA or hemiarthroplasty for hip fracture (74% of their study population underwent a hemiarthroplasty). Also based on LROI data, Zijlstra et al. (2017) showed that the posterolateral approach resulted in higher revision rates due to dislocation compared with all other surgical approaches (HR = 1.0 vs. 0.5–0.6) in the case of THA for primary osteoarthritis. A strength of our study is the focus on cup revisions only, since type of revision (cup, stem, insert, and/or femoral head exchange) is specified in the LROI. A limitation of register studies is the risk for selection bias. First, there is a possibility that DMC was used exclusively in a few clinics and/or by single surgeons because of preference. Second, it is possible that different cup designs were used for different types of patients for other reasons such as patient comorbidity. We tried to make an estimation of frailty and comorbidity using patient characteristics available in the LROI and found no statistically significant differences between the 2 groups based on age and ASA classification. We plan further analyses with a more extensive set of patient variables including smoking status, Charnley score and BMI. Another limitation of this study is the fact that an acute hip fracture was not further specified in the LROI database. Most often an acute femoral neck fracture will have been the indication for a THA, but some trochanteric fractures cannot be ruled out. Closed reductions for dislocations are not registered in the LROI. Reductions for UC THA can often be performed without surgery, but closed reductions are often impossible in DMC THA needing surgery with component exchange and hence registration in the LROI. This means that the dislocation revisions in DMC reflect the number of postoperative dislocations better than the dislocation revisions in UC.

Acta Orthopaedica 2021; 92 (1): 35–38

In conclusion, the 5-year cumulative incidence of cup revision rate after THA for acute fracture was 1.9% (CI 1.6–2.2) being comparable for DMC and UC THA with a 32 mm femoral head. However, DMC THA had a lower risk of cup revision than UC THA with a 22–28 mm femoral head. All authors contributed to the conception of the study, data analysis, and preparation of the manuscript. Acta thanks Torben Hansen and Johan Kärrholm for help with peer review of this study.

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 3,308 cases in the Dutch Arthroplasty Register (2007-2016). Acta Orthop 2019; 90(1): 11-14. Byström S, Espehaug B, Furnes O, Havelin LI, Norwegian Arthroplasty Register. Femoral head size is a risk factor for total hip luxation: a study of 42,987 primary hip arthroplasties from the Norwegian Arthroplasty Register. Acta Orthop Scand 2003; 74(5): 514-24. Enocson A, Hedbeck C J, Tidermark J, Pettersson H, Ponzer S, Lapidus L J. Dislocation of total hip replacement in patients with fractures of the femoral neck. Acta Orthop 2009; 80(2): 184-9. Gjertsen J E, Lie S A, Fevang J M, Havelin L I, Engesaeter L B, Vinje T, Furnes O. Total hip replacement after femoral neck fractures in elderly patients: results of 8,577 fractures reported to the Norwegian Arthroplasty Register. Acta Orthop 2007; 78(4): 491-7. 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. Jobory A, Kärrholm J, Overgaard S, Pedersen A B, Halan G, Gjertsen J, Maela K, Rogmark C. Reduced revision risk for dual-mobility cup in total hip replacement due to hip fracture. J Bone Joint Surg Am 2019; 101: 1278-85. Kostensalo I, Junnila A. Virolainen P. Effect of femoral head size on risk of revision for dislocation after total hip arthroplasty. Acta Orthop 2013; 84(4): 342-7. Lacny S, Wilson T, Clement F, Roberts D J, Faris P D, Ghali W A, Marshall D A. Kaplan–Meier survival analysis overestimates the risk of revision arthroplasty: a meta-analysis. Clin Orthop Relat Res 2015; 473(11): 3431-42. LROI. Annual report 2018. Available from www.lroi-report.nl. Moerman S, Mathijsen N M C, Tuinebrijer W E, Vochteloo A J H, Nelissen R G H H. Hemiarthroplasty and total hip arthroplasty in 30,380 patients with hip fractures: data from the Dutch Arthroplasty Register on revision and risk factors for revision. Acta Orthop 2018; 89(5): 509-14. Parker M J, Gurusamy K S, Azegami S. Arthroplasties (with and without bone cement) for proximal femoral fractures in adults. Cochrane Database Syst Rev 2010; (6): CD001706. Tabori-Jensen S, Hansen T B, Stilling M. Low dislocation rate of Saturne®/ Avantage® dual-mobility THA after displaced femoral neck fracture: a cohort study of 966 hips with a minimum 1.6-year follow-up. Arch Orthop Trauma Surg 2019; 139(5): 605-12. Vektis (www.vektis.nl) Wongworawat M D, Dobbs M B, Gebhardt M C, Gioe T J, Leopold S S, Manner P A, Rimnac C M, Porcher R. Editorial: Estimating survivorship in the face of competing risk. Clin Orthop Relat Res 2015; 473: 1173-6. Zijlstra W P, De Hartog B, van Steenbergen L N, Scheurs B M, Nelissen R G H H. Effect of femoral head size and surgical approach on risk of revision for dislocation after total hip arthroplasty. Acta Orthop 2017; 88(4): 395-401.

Acta Orthopaedica 2021; 92 (1): 39–45

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 Olof WOLF 1, Sebastian MUKKA 2, Jan EKELUND 3, Michael MÖLLER 4, and Nils P HAILER 1 1 Department of Surgical Sciences, Orthopaedics, Uppsala University, Uppsala; 2 Department of Surgical and Perioperative Sciences at Umeå University, Umeå; 3 Centre of Registers Västra Götaland, Gothenburg; 4 Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden Correspondence: olof.wolf@surgsci.uu.se Submitted 2020-05-07. Accepted 2020-09-08.

Background and purpose — Unlike hip fractures, diaphyseal and distal femoral fractures in elderly patients have not been widely studied. We investigated the demographics, comorbidities and mortality of patients with femoral fractures at any anatomical level with a focus on early mortality. Patients and methods — We analyzed 11,799 patients ≥ 65 years with a femoral fracture registered in the Swedish Fracture Register from 2011 to 2014. The cohort was matched with the National Patient Register to obtain data on comorbidities classified according to the Charlson Comorbidity Index (CCI). Generalized linear models were fitted to estimate the adjusted relative risk of mortality. Results — Mean age of the cohort was 83 years and 69% were women. Patients with distal femoral fractures had the lowest degree of comorbidity, with 9% having a CCI of ≥ 3 compared with 14% among those with proximal and 16% among those with diaphyseal fractures. Unadjusted 90-day mortalities were 13% (95% CI 9.4–16) after fractures in the distal, 13% (CI 10–16) in the diaphyseal, and 15% (CI 14–15) in the proximal segment. The adjusted relative risk for 90-day mortality was 1.1 (CI 0.86–1.4) for patients with distal and 0.97 (CI 0.76–1.2) for patients with diaphyseal femoral fractures when compared with patients with hip fractures. Interpretation — Elderly patients with femoral fractures distal to the hip may have similar adjusted early mortality risks to those with hip fractures. There is a need for larger, preferably prospective, studies investigating the effect of rapid pathways and geriatric co-management for patients with diaphyseal and distal femoral fractures.

1-year mortality after proximal femur fractures is up to 30% and is higher in men (Do et al. 2016, Mattisson et al. 2018). Proximal femoral fractures have been thoroughly investigated for outcomes after different treatment modalities (Gdalevich et al. 2004, Sircar et al. 2007, Al-Ani et al. 2008, Khan et al. 2009, Hansson et al. 2017, Bartels et al. 2018, Dolatowski et al. 2019). In contrast, little research has been conducted on femoral fractures distal to the proximal segment in the elderly population. However, there is some evidence to suggest that patients with diaphyseal and distal femur fractures have similar mortality and mobility risks to those with proximal femur fractures (Konda et al. 2015, Myers et al. 2018, Larsen et al. 2020). Elderly patients with fractures of the femur distal to the hip also seem to be similar to hip fracture patients in age and sex distribution. However, there is little information on their degree of comorbidities (Smith et al. 2015) and, to our knowledge, no comparative study has been performed on mortality after femoral fractures distal to the hip. Consequently, guidelines are lacking on the treatment of elderly patients with diaphyseal and distal femoral fractures. The Swedish Fracture Register (SFR) was launched in 2011 to prospectively monitor fracture treatment performed in Sweden and collect information on all fractures, including data on injury mechanisms, fracture characteristics, and treatments (Wennergren et al. 2015). Data from this register has been matched with the National Patient Register (NPR) to obtain information on comorbidities and causes of death (Ludvigsson et al. 2009). We thus designed an observational study on a cohort of elderly patients with femoral fractures with the primary aim to evaluate the association of fracture localization with mortality, adjusting for pre-existing comorbidities.

Patients and methods Study design and variables We designed an observational cohort study of patients with femoral fractures. The treating orthopedic surgeon registers all Swedish patients treated for any fracture at departments affiliated to the SFR. The SFR started with registration in Gothenburg only in 2011. In 2014, the SFR had a coverage of 40% of the orthopedic departments, which gave us data from 24 hospitals, and by 2020 all orthopedic departments in Sweden will be active in the SFR. Details are collected on injury mechanisms, trauma energy content, fracture type according to the AO/OTA classification, time and date of radiography, and details on treatment, including the type of osteosynthesis or joint replacement and the surgeon’s level of expertise (Wennergren et al. 2015). Information on mortality is obtained by real-time linkage to the Swedish National Death Register. This cohort was matched with the NPR to collect data on comorbidities and causes of death. The Charlson Comorbidity Index (CCI), modified according to Quan et al. (2011), was calculated from ICD codes 12 months before the date of injury and categorized into 3 groups: CCI = 0, CCI = 1–2, and CCI = ≥ 3, representing no, moderate, and high comorbidity, respectively. We followed the STROBE guidelines for the reporting of this observational study. Patient selection We retrieved information on all patients aged ≥ 65 years registered with a femoral fracture (International Classification of Diseases [ICD] S72.0–S72.4) between January 1, 2011 and December 31, 2014. To avoid dependency issues, we included only data concerning the first fracture in patients with a subsequent femoral fracture during the study period. For the same reason, patients with simultaneous bilateral or simultaneous fractures at several anatomical levels were excluded (Figure 1). Patients with periprosthetic (after hip or knee arthroplasty) and implant-related (after previous plate, nail, or screw) fractures were included in the main analysis but were also analyzed separately. Outcome measures We analyzed (1) the adjusted relative risk (RR) of 90-day mortality dependent on fracture location (proximal, diaphyseal, or distal part of the femur), (2) the adjusted RR of 30- and 365day mortality dependent on fracture location as above, (3) the association between age, sex, and pre-existing comorbidities and mortality, and (4) the distribution of fracture location and mortality of patients with subsequent fractures, periprosthetic, or implant-related fractures. Statistics Baseline variables are presented as means (SDs) and proportions, cross-tabulated by femoral segment. Differences

Acta Orthopaedica 2021; 92 (1): 39–45

Patients aged ≥18 years with a registered femur fracture in 2011–2014 n = 13,080 Excluded (n = 1,281): – age < 65 years, 1,226 – simultaneous (bilateral or 2 ipsilateral) femur fractures, 55

Patients aged ≥ 65 years with a registered femur fracture in 2011–2014 n = 11,799

Figure 1. Flowchart of patients included in the study.

between observed counts were analyzed using the chi-square test. Unadjusted cumulative mortality was estimated using the Kaplan–Meier method. Generalized linear models with binomial distribution and a log-link were fitted to estimate the RR of 30-, 90-, and 365-day mortality by fracture location, adjusted for age, sex, and CCI with 95% confidence intervals (CI). Follow-up mortality data was retrieved for 1 year for all patients. Statistical analyses were performed using IBM SPSS Statistics, version 26 for Mac (IBM Corp, Armonk, NY, USA) and SAS, version 9.4 (SAS Institute, Cary, IN, USA). Survival curves with CIs were plotted using the R software package (R Development Core Team 2017). Ethics, funding, and potential conflicts of interest The study was conducted following the ethical principles of the Helsinki Declaration and was approved by the Regional Ethical Committee in Uppsala (Dnr 2015/510; date of approval January 20, 2016). In accordance with Swedish law, individual consent was not required. This study was supported by Stiftelsen Skobranschens Utvecklingsfond and by the Swedish Research Coucil (VR 2018–02612). The authors declare no competing interests.

Results Characteristics of the study population The final study cohort comprised 11,799 patients with a femoral fracture (Figure 1, Table 1). The mean age of the patients was 83 years (SD 8) and 69% were women. Of the fractures, 3% occurred in the distal, 4% in the diaphyseal, and 93% in the proximal femur. A same-level fall was the cause of injury in 93% of all patients, 3.6% fell from another level, and 2% of the injuries were traffic-related. Most fractures (97%) were from a low-energy injury, and 0.3% (n = 31) were open fractures. The proportion of patients with high comorbidity (CCI ≥ 3) was lower among patients with distal fractures (9%) when compared with patients with diaphyseal (16%) or proximal fractures (14%, Table 1). When stratified by sex, men had similar comorbidity patterns independent of the femoral fracture

Acta Orthopaedica 2021; 92 (1): 39–45

Table 1. Baseline demographic characteristics of the study population of 11,799 patients with femoral fractures. Values are distribution of the Charlson Comorbidity Index category and subcategories (%) depending on fracture segment

Description

Proximal

Diaphyseal

Distal

Distribution, n (%) 10,964 (93) 469 (4) 366 (3) Mean age (SD) 83 (8) 82 (9) 82 (9) Sex, n (%) Male 3,441 (31) 141 (30) 68 (19) Female 7,523 (69) 328 (70) 298 (81) CCI a 0 54 54 63 1–2 33 29 28 3 or more 14 16 9 Charlson subcategories a Myocardial infarction 9.4 8.3 6.6 Congestive heart failure 13 11 13 Peripheral vascular disease 3.3 1.5 1.9 Cerebrovascular disease 11 8.7 9.8 Dementia 17 11 7.4 Chronic pulmonary disease 10 11 6.6 Rheumatic disease 3.8 6.4 5.7 Peptic ulcer disease 1.0 0.6 0.5 Mild liver disease 0.7 0.0 1.4 Diabetes without chronic complications 13 10 9.3 Hemiplegia or paraplegia 1.6 1.3 1.6 Renal disease 4.6 3.4 5.7 Diabetes with chronic complications 2.3 2.3 3.0 Malignancy 8.9 12 4.6 Moderate or severe liver disease 0.2 0.2 0.3 Metastatic solid tumor 2.3 7.5 1.1 Aids/HIV 0 0 0 a Proportion

(%) within segment

segment. Women with distal femoral fractures had a similar proportion of high comorbidity (8.7%) compared with those with proximal fractures (11%), whereas women with diaphyseal fractures had a higher proportion of high comorbidity (17%) compared with proximal fractures. For men, 19% had high comorbidity compared with 11% of the women. Finally, 48% of the men and 57% of the women had no comorbidity (Table 1). 90-day mortality The unadjusted 90-day mortality rates were 13% (CI 9.4–16) for patients with distal, 13% (CI 10–16) for those with diaphyseal, and 15% (CI 14–15) for those with proximal segment fractures (Figure 2). We found no statistically significant difference in the adjusted RR of 90-day mortality when dividing patients by fracture location although confidence intervals were wide due to the size of the sample: RR = 1.1 (CI 0.9–1.4) for patients with distal and RR = 1.0 (CI 0.8–1.2) for patients with diaphyseal fractures when compared with patients with proximal femoral fractures. No statistically significant differences in the adjusted mortality risk among

Figure 2. Unadjusted cumulative mortality up to 1 year after index fracture per femoral segment with 95% confidence intervals.

patients with the 3 different fracture locations were observed when stratifying the analyses by sex or CCI group (Table 2). 30-day and 1-year mortality After 30 days, the unadjusted cumulative mortality was 6.3% for patients with distal, 8.3% for those with diaphyseal, and 7.5% for those with proximal femoral fractures. The unadjusted cumulative mortalities after 1 year were 21% for patients with distal, 21% for those with diaphyseal, and 26% for those proximal femoral fractures (Table 3). We found no statistically significant differences in the adjusted risk of 30-day or 1-year mortality among patients with these fracture locations. Subsequent fractures During the study period, 3% (n = 350) of the patients had a subsequent femoral fracture of any kind. Of these, no patients with an initial distal or diaphyseal and 2 patients with an initial proximal femoral fracture died within 30 days of the first fracture. Comparing patients with a subsequent femoral fracture of any kind by initial fracture locations, the unadjusted 90-day mortality rates were thus 0% for patients with distal, 7.7% (CI 0–22; n = 1 of 13) for those with diaphyseal, and 5.8% (CI 3–8; n = 19 of 329) for those with proximal fractures compared with13%, 13%, and 15%, respectively, for patients without subsequent fractures. Likewise, the 1-year cumulative mortality of patients sustaining a subsequent femoral fracture was 13% for those with a distal, 15% for those with a diaphyseal, and 19% for those with a proximal femoral fracture. Periprosthetic and implant-related fractures 3.2% (n = 383) of all fractures were periprosthetic. Of all periprosthetic fractures, 29% were in the distal, 47% in the diaphyseal, and 24% in the proximal segment. 30% of all distal frac-

Acta Orthopaedica 2021; 92 (1): 39–45

Table 2. Unadjusted 90-day mortality dependent on fracture location and adjusted for age, sex, and the Charlson Comorbidity Index (CCI), including stratified analyses for sex and CCI groups. Values are percentage and relative risks (RR) with 95% confidence intervals (CI) Fracture location

Mortality (%)

RR (CI)

Unadjusted 90-day mortality Proximal 15 (CI 14–15) Diaphyseal 13 (CI 10–16) Distal 13 (CI 9.4–16) Proximal 1 ref Diaphyseal 1.0 (0.8–1.2) Distal 1.1 (0.9–1.4) 90-day mortality when stratified by sex Men Proximal 1 ref Diaphyseal 1.2 (0.8–1.6) Distal 1.1 (0.6–1.7) Women Proximal 1 ref Diaphyseal 0.8 (0.6–1.1) Distal 1.2 (0.9–1.5) 90-day mortality when stratified by CCI No (CCI = 0) Proximal 1 ref Diaphyseal 0.8 (0.5–1.2) Distal 0.9 (0.5–1.3) Moderate (CCI = 1–2) Proximal 1 ref Diaphyseal 0.8 (0.5–1.2) Distal 1.4 (1.0–2.0) High (CCI = 3 or more) Proximal 1 ref Diaphyseal 1.2 (0.8–1.6) Distal 1.1 (0.6–1.7)

tures, 38% of all diaphyseal, and 1% of all proximal fractures were periprosthetic. The 1-year cumulative mortality was 19% for patients with distal, 18% for patients with diaphyseal, and 19% for those with proximal femoral periprosthetic fractures. 1.2% (n = 137) of the cohort had implant-related femoral fractures. Of all implant-related fractures, 20% were in the distal, 45% in the diaphyseal, and 36% in the proximal femur. 7% of all patients with distal fractures, 13% of those with diaphyseal, and 0.4% of those with proximal fractures had implant-related fractures. The 1-year cumulative mortality was 19% for patients with distal, 12% for those with diaphyseal, and 18% for those with implant-related femoral fractures.

Discussion In this observational cohort study we find that elderly patients with a femoral fracture at the distal or diaphyseal level have a similar adjusted 90-day mortality risk to hip fracture patients. Patients with a distal femoral fracture are less comorbid than those with a proximal fracture, and there is a higher proportion of women among these than in patients with diaphyseal or proximal fractures.

Table 3. Unadjusted 30- and 365-day mortality dependent on fracture location. Values are mean percentage (95% CI) Fracture location Proximal Diaphyseal Distal

30-day 7.5 (7.0–8.0) 8.3 (5.8–11) 6.3 (3.8–8.8)

365-day 26 (25–27) 21 (18–25) 21 (17–25)

We are not aware of any other comparative study on femoral fractures examining the effect of fracture location on mortality. Some studies have investigated the mortality of patients with distal femoral fractures and compared those with previously reported figures on hip fracture mortality (Nyholm et al. 2017, Larsen et al. 2020). The 1-year mortality was 35% for patients > 60 years compared with 3% in those who were < 60 years in a study on distal femoral fractures (Larsen et al. 2020). Our figures on mortality after distal femoral fractures correspond well with the mortality rates of 7% at 30 days and 18% at 1 year found in a retrospective study from the UK in 105 patients > 50 years (Smith et al. 2015). Another study from the Danish Fracture Database on 392 patients > 50 years with a closed low-energy distal femoral fracture reported mortality rates of 7.1% at 30 days and 13% at 90 days (Nyholm et al. 2017). Patients with a periprosthetic distal femoral fracture and patients with comorbidities exhibit higher mortality rates (Streubel et al. 2011, Kammerlander et al. 2012). In addition, an increasing ASA score and male sex are associated with higher mortality in both proximal and distal femoral fractures (Nyholm et al. 2015, 2017). Half of the periprosthetic fractures occurred in the femoral diaphysis and the remaining half were evenly distributed between the proximal and distal femoral segments. Approximately a third of the distal and diaphyseal fractures were periprosthetic and these are often amenable to fracture fixation as opposed to the proximal periprosthetic fractures. A loose total hip arthroplasty requires extensive revision surgery and such surgery has been associated with high mortality and an overall complication rate of 18% (Lindahl et al 2005, Marsland and Mears 2012). In contrast, we found no excessive mortality rates in the groups of patients with periprosthetic or implantrelated femoral fractures. A subsequent femoral fracture would at first thought be expected to enhance mortality. Our opposite finding of lower mortality rates in patients with subsequent fractures may be due to immortal time bias introduced by the fact that patients with second fractures must have survived their first fracture. Dementia is common in hip fracture patients (Friedman et al. 2010) and has been associated with higher mortality, reduced walking ability, and lower recovery to full ADL function after a hip fracture (Larsson et al. 2019, Delgado et al. 2020). We found a higher proportion of dementia cases among patients with a proximal femoral fracture compared with patients

Acta Orthopaedica 2021; 92 (1): 39–45

with fractures further distal in the femur. Although there was a difference in the distribution of dementia between fracture locations, there was no association with mortality. Patients with pre-fracture dementia have a higher risk of developing delirium, which must be accounted for in the care of hip fracture patients (Krogseth et al. 2016). Our rates of dementia were substantially lower than the 50% of hip fracture patients included in an RCT evaluating orthogeriatric care on cognitive function following hip fracture (Watne et al. 2014), but this may be due to more stringent analysis of this specific comorbidity in the cited RCT. With over 90% of the femoral fractures in the elderly being proximal femoral fractures, mortality rates have been thoroughly investigated. Actions such as multi-professional or orthogeriatric care and hip fracture pathways have been effective in reducing mortality rates and improving functional outcomes (Pedersen et al. 2008, Adunsky et al. 2011, Leung et al. 2011, Prestmo et al. 2015, Mukherjee et al. 2020). Pathways without geriatric involvement have failed to show beneficiary results (Haugan et al. 2017, Svenoy et al. 2020). Moreover, surgery on hip fractures within 24 hours has been advocated to decrease complications and lower mortality rates (Nyholm et al. 2015). By contrast, a delay of up to 48 hours post-admission does not affect mortality according to a cohort study of over 70,000 patients from the Norwegian Hip Fracture Register (Leer-Salvesen et al. 2019). Elderly patients with hip fractures have higher mortality risk following surgical delay, a risk that is enhanced in men and patients with multiple comorbidities (Beaupre et al. 2019). Greve et al. (2020) reported higher mortality in patients with an ASA score of 3–4, but surprisingly also in women, when surgical waiting time was > 24 hours from admission. All this has led to an effort in many countries to operate hip fractures within 24 hours—or as the NICE guidelines in the UK say, “perform surgery on the day of, or the day after, admission” (National Institute for Health and Care Excellence 2011). Conflicting results from 2 studies on surgical delay and the effect on mortality after proximal (Nyholm et al. 2015) and distal femoral fractures (Nyholm et al. 2017) have been reported from the Danish Fracture Database. In these stydies delay had no effect on mortality after distal femoral fractures but a significant effect already after a 12-hour delay for proximal femoral fractures. This difference may be attributed to the fact that 80% with distal femoral fractures were women compared with 70% with proximal fractures (and diaphyseal in our findings; Konda et al. 2015, Nyholm et al. 2015, 2017). Ultra-fast surgery after hip fracture indicated no measurable effect on mortality or major complications in the HIP ATTACK trial (HIP ATTACK Investigators 2020). Of note, a higher level of surgeon expertise was associated with decreased mortality for proximal but not for distal femoral fractures (Nyholm et al. 2015, 2017). Displaced femoral neck fractures are routinely operated on with arthroplasty, which is uncommon for complex and comminuted fractures in the distal part of the femur. However, most distal femoral

fractures are extraarticular or have a simple articular component and should be amenable to routine periarticular plating or nailing without delay (Nyholm et al. 2017). The few patients needing complex reconstruction by a specialized orthopedic trauma surgeon should not dictate the routine for the majority. This study has several limitations, most of which are due to the register-based observational design. Time to surgery was added to the SFR in late 2014 and is a crucial variable in evaluating mortality in elderly patients with femoral fractures. Further research with newer data could provide information on the impact of preoperative waiting time on mortality by fractured femoral segment. The majority (93%) of patients in our cohort had proximal femoral fractures. However, with a study population of almost 12,000 patients, the distal femoral fracture group is larger than in some previous studies (Kammerlander et al. 2012, Smith et al. 2015, Nyholm et al. 2017, Myers et al. 2018, Larsen et al. 2020). We also analyzed elderly patients with diaphyseal fractures, which allowed us to compare all femoral fractures in the elderly patient. We lack information on function level and the ASA score. Such information would strengthen our comparisons and conclusions; however, the CCI retrieved from the NPR gives us a good indication of the comorbidity of the patients. Given the nature of register-based research, we have retrieved a large cohort of femoral fractures registered and classified by the treating surgeons. Such an approach increases the generalizability of our results compared with a single-center retrospective study although the data in this study was retrieved in an early phase of the SFR with 40% of the orthopedic departments taking active part in the registration. We assessed a cohort of > 11,000 femoral fractures and retrieved data for comorbidity for an adjusted analysis. Data on fracture localization affecting mortality was adjusted for age, sex, and comorbidity as assessed by the CCI. The NPR has been validated previously (Ludvigsson et al. 2011) and there is recurrent work to audit the completeness of data in the SFR. For hip fractures, most of the hospitals have completeness of registration > 90% compared with the NPR. To conclude, after adjustment for sex, age, and comorbidities femoral fractures in the elderly seem to be equally deadly independent of the anatomical segment that is injured. Our interpretation may be hampered by the limited precision of our risk estimates that leaves room for type II errors. There is a need for larger, preferably prospective, studies investigating the effect of rapid pathways, early surgery, and geriatric co-management also for patients with diaphyseal and distal femoral fractures. OW and NH planned the study. JE and OW performed the statistical analyses, and JE fitted the GLM. OW, NH, and SM wrote the initial draft. All authors contributed to the interpretation of the data and revision of the manuscript.

Acta thanks Frede Frihagen, Jan-Erik Gjertsen, and Lars Johnsen for help with peer review of this study.

Adunsky A, Lerner-Geva L, Blumstein T, Boyko V, Mizrahi E, Arad M. Improved survival of hip fracture patients treated within a comprehensive geriatric hip fracture unit, compared with standard of care treatment. J Am Med Dir Assoc 2011; 12(6): 439-44. Al-Ani A N, Samuelsson B, Tidermark J, Norling A, Ekstrom W, Cederholm T, Hedstrom M. Early operation on patients with a hip fracture improved the ability to return to independent living: a prospective study of 850 patients. J Bone Joint Surg Am 2008; 90(7): 1436-42. Bartels S, Gjertsen J E, Frihagen F, Rogmark C, Utvag S E. High failure rate after internal fixation and beneficial outcome after arthroplasty in treatment of displaced femoral neck fractures in patients between 55 and 70 years. Acta Orthop 2018; 89(1): 53-8. Beaupre L A, Khong H, Smith C, Kang S, Evens L, Jaiswal P K, Powell J N. The impact of time to surgery after hip fracture on mortality at 30- and 90-days: does a single benchmark apply to all? Injury 2019; 50(4): 950-5. Delgado A, Cordero G G E, Marcos S, Cordero-Ampuero J. Influence of cognitive impairment on mortality, complications and functional outcome after hip fracture: dementia as a risk factor for sepsis and urinary infection. Injury 2020; 51 Suppl 1: S19-S24 Do L N, Kruke T M, Foss O A, Basso T. Reoperations and mortality in 383 patients operated with parallel screws for Garden I–II femoral neck fractures with up to ten years follow-up. Injury 2016; 47(12): 2739-42. Dolatowski F C, Frihagen F, Bartels S, Opland V, Saltyte Benth J, Talsnes O, Hoelsbrekken S E, Utvag S E. Screw fixation versus hemiarthroplasty for nondisplaced femoral neck fractures in elderly patients: a multicenter randomized controlled trial. J Bone Joint Surg Am 2019; 101(2): 136-44. Friedman S M, Menzies I B, Bukata S V, Mendelson D A, Kates S L. Dementia and hip fractures: development of a pathogenic framework for understanding and studying risk. Geriatr Orthop Surg Rehabil 2010; 1(2): 52-62. Gdalevich M, Cohen D, Yosef D, Tauber C. Morbidity and mortality after hip fracture: the impact of operative delay. Arch Orthop Trauma Surg 2004; 124(5): 334-40. Greve K, Modig K, Talback M, Bartha E, Hedstrom M. No association between waiting time to surgery and mortality for healthier patients with hip fracture: a nationwide Swedish cohort of 59,675 patients. Acta Orthop 2020; 91(4): 396-400. Hansson S, Nemes S, Karrholm J, Rogmark C. Reduced risk of reoperation after treatment of femoral neck fractures with total hip arthroplasty. Acta Orthop 2017; 88(5): 500-4. Haugan K, Johnsen L G, Basso T, Foss O A. Mortality and readmission following hip fracture surgery: a retrospective study comparing conventional and fast-track care. BMJ Open 2017; 7(8): e015574. HIP ATTACK Investigators. Accelerated surgery versus standard care in hip fracture (HIP ATTACK): an international, randomised, controlled trial. Lancet 2020; 395(10225): 698-708. Kammerlander C, Riedmuller P, Gosch M, Zegg M, Kammerlander-Knauer U, Schmid R, Roth T. Functional outcome and mortality in geriatric distal femoral fractures. Injury 2012; 43(7): 1096-101. Khan S K, Kalra S, Khanna A, Thiruvengada M M, Parker M J. Timing of surgery for hip fractures: a systematic review of 52 published studies involving 291,413 patients. Injury 2009; 40(7): 692-7. Konda S R, Pean C A, Goch A M, Fields A C, Egol K A. Comparison of shortterm outcomes of geriatric distal femur and femoral neck fractures: results from the NSQIP database. Geriatr Orthop Surg Rehabil 2015; 6(4): 311-15. Krogseth M, Watne L O, Juliebo V, Skovlund E, Engedal K, Frihagen F, Wyller T B. Delirium is a risk factor for further cognitive decline in cognitively impaired hip fracture patients. Arch Gerontol Geriatr 2016; 64(38-44. 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.

Acta Orthopaedica 2021; 92 (1): 39–45

Larsson G, Stromberg U, Rogmark C, Nilsdotter A. Cognitive status following a hip fracture and its association with postoperative mortality and activities of daily living: a prospective comparative study of two prehospital emergency care procedures. Int J Orthop Trauma Nurs 2019; 35: 100705. Leer-Salvesen S, Engesaeter L B, Dybvik E, Furnes O, Kristensen T B, Gjertsen J E. Does time from fracture to surgery affect mortality and intraoperative medical complications for hip fracture patients? An observational study of 73 557 patients reported to the Norwegian Hip Fracture Register. Bone Joint J 2019; 101-B(9): 1129-37. Leung A H, Lam T P, Cheung W H, Chan T, Sze P C, Lau T, Leung K S. An orthogeriatric collaborative intervention program for fragility fractures: a retrospective cohort study. J Trauma 2011; 71(5): 1390-4. Lindahl H, Malchau H, Herberts P, Garellick G. Periprosthetic femoral fractures: classification and demographics of 1049 periprosthetic femoral fractures from the Swedish National Hip Arthroplasty Register. J Arthroplasty 2005; 20(7): 857-65. Ludvigsson J F, Otterblad-Olausson P, Pettersson B U, Ekbom A. The Swedish personal identity number: possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol 2009; 24(11): 659-67. 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. Marsland D, Mears S C. A review of periprosthetic femoral fractures associated with total hip arthroplasty. Geriatr Orthop Surg Rehabil 2012; 3(3): 107-20. Mattisson L, Bojan A, Enocson A. Epidemiology, treatment and mortality of trochanteric and subtrochanteric hip fractures: data from the Swedish fracture register. BMC Musculoskelet Disord 2018; 19(1): 369. Mukherjee K, Brooks S E, Barraco R D, Como J J, Hwang F, Robinson B R H, Crandall M L. Elderly adults with isolated hip fractures—orthogeriatric care versus standard care: a practice management guideline from the Eastern Association for the Surgery of Trauma. J Trauma Acute Care Surg 2020; 88(2): 266-78. Myers P, Laboe P, Johnson K J, Fredericks P D, Crichlow R J, Maar D C, Weber T G. Patient mortality in geriatric distal femur fractures. J Orthop Trauma 2018; 32(3): 111-15. National Institute for Health and Care Excellence (2011, May 2017). Hip fracture: management (Clinical guideline [CG124]). Available from https:// www.nice.org.uk/guidance/cg124 (retrieved April 1, 2020). Nyholm A M, Gromov K, Palm H, Brix M, Kallemose T, Troelsen A, Danish Fracture Database C. Time to surgery is associated with thirtyday and ninety-day mortality after proximal femoral fracture: a retrospective observational study on prospectively collected data from the Danish Fracture Database Collaborators. J Bone Joint Surg Am 2015; 97(16): 1333-9. Nyholm A M, Palm H, Kallemose T, Troelsen A, Gromov K, collaborators D. No association between surgical delay and mortality following distal femoral fractures: a study from the Danish Fracture Database Collaborators. Injury 2017; 48(12): 2833-7. Pedersen S J, Borgbjerg F M, Schousboe B, Pedersen B D, Jorgensen H L, Duus B R, Lauritzen J B. A comprehensive hip fracture program reduces complication rates and mortality. J Am Geriatr Soc 2008; 56(10): 1831-8. Prestmo A, Hagen G, Sletvold O, Helbostad J L, Thingstad P, Taraldsen K, Lydersen S, Halsteinli V, Saltnes T, Lamb S E, Johnsen L G, Saltvedt I. Comprehensive geriatric care for patients with hip fractures: a prospective, randomised, controlled trial. Lancet 2015; 385(9978): 1623-33. Quan H, Li B, Couris C M, Fushimi K, Graham P, Hider P, Januel J M, Sundararajan V. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol 2011; 173(6): 676-82. R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2017. Sircar P, Godkar D, Mahgerefteh S, Chambers K, Niranjan S, Cucco R. Morbidity and mortality among patients with hip fractures surgically repaired within and after 48 hours. Am J Ther 2007; 14(6): 508-13.

Acta Orthopaedica 2021; 92 (1): 39–45

Smith J R, Halliday R, Aquilina A L, Morrison R J, Yip G C, McArthur J, Hull P, Gray A, Kelly M B, Collaborative—Orthopaedic Trauma S. Distal femoral fractures: the need to review the standard of care. Injury 2015; 46(6): 1084-8. Streubel P N, Ricci W M, Wong A, Gardner M J. Mortality after distal femur fractures in elderly patients. Clin Orthop Relat Res 2011; 469(4): 1188-96. Svenoy S, Watne L O, Hestnes I, Westberg M, Madsen J E, Frihagen F. Results after introduction of a hip fracture care pathway: comparison with usual care. Acta Orthop 2020; 91(2): 139-45.

Watne L O, Torbergsen A C, Conroy S, Engedal K, Frihagen F, Hjorthaug G A, Juliebo V, Raeder J, Saltvedt I, Skovlund E, Wyller T B. The effect of a pre- and postoperative orthogeriatric service on cognitive function in patients with hip fracture: randomized controlled trial (Oslo Orthogeriatric Trial). BMC Med 2014; 12: 63. Wennergren D, Ekholm C, Sandelin A, Moller M. The Swedish Fracture Register: 103,000 fractures registered. BMC Musculoskelet Disord 2015; 16; 338.

Acta Orthopaedica 2021; 92 (1): 46–52

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 Alma B PEDERSEN 1, Aurélie MAILHAC 1, Anne GARLAND 2,8, Søren OVERGAARD 3, Ove FURNES 4,5, Stein Atle LIE 5,6, Anne Marie FENSTAD 4, Cecilia ROGMARK 7,8, Johan KÄRRHOLM 8,9, Ola ROLFSON 8,9, Jaason HAAPAKOSKI 10, Antti ESKELINEN 10,11, Keijo T MÄKELÄ 10,12, and Nils P HAILER 2,8 1 Department

of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark; 2 Department of Surgical Sciences/Orthopedics, Uppsala University, Uppsala, Sweden; 3 Department of Orthopedic Surgery and Traumatology, Odense University Hospital, Department of Clinical Research, University of Southern Denmark, Odense, Denmark, and the Danish Hip Arthroplasty Register; 4 The Norwegian Arthroplasty Register, Department of Orthopedic Surgery, Haukeland University Hospital, Bergen, Norway; 5 Department of Clinical Medicine, University of Bergen, Norway; 6 Department of Clinical Dentistry, University of Bergen, Bergen, Norway; 7 Department of Orthopedics, Lund University, Skåne University Hospital, Malmö, Sweden; 8 The Swedish Hip Arthroplasty Registry, Registercentrum Västra Götaland, Gothenburg, Sweden; 9 Department of Orthopaedics, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Sweden; 10 Finnish Arthroplasty Register, National Institute for Health and Welfare, Helsinki, Finland; 11 Coxa Hospital for Joint Replacement, and Faculty of Medicine and Health Technologies, University of Tampere, Tampere, Finland; 12 Department of Orthopedics and Traumatology, Turku University Hospital, Turku, Finland Correspondence: abp@clin.au.dk Submitted 2020-05-26. Accepted 2020-10-07.

Background and purpose — Current literature indicates no difference in 90-day mortality after cemented compared with cementless total hip arthroplasty (THA). However, previous studies are hampered by potential selection bias and suboptimal adjustment for comorbidity confounding. Therefore, we examined the comorbidity-adjusted mortality up to 90 days after cemented compared with cementless THA performed due to osteoarthritis. Patients and methods — Using the Nordic Arthroplasty Register Association database, 2005–2013, we included 108,572 cemented and 80,034 cementless THA due to osteoarthritis. We calculated the Charlson comorbidity index of each patient based on data from national patient registers. The Kaplan–Meier method was used to estimate unadjusted all-cause mortality. Cox regression was used to estimate hazard ratios (HR) with 95% confidence intervals (CI) for 14 , 30-, and 90-day mortality comparing cemented with cementless THA, adjusting for age, sex, comorbidity, nation, and year of surgery. Results — Cumulative all-cause mortality within 90 days was 0.41% (CI 0.37–0.46) after cemented and 0.26% (CI 0.22–0.30) after cementless THA. The adjusted HR for cemented vs. cementless fixation was 0.97 (CI 0.79–1.2), and similar risk estimates were obtained for mortality within 14 (adjusted HR 0.91 [CI 0.64–1.3]) and 30 days (adjusted HR 0.94 [CI 0.71–1.3]). We found no clinically relevant differences in mortality between cemented and cementless THA in analyses stratified by age, sex, Charlson comorbidity index, or year of surgery.

Interpretation — After adjustment for comorbidity as an important confounder, we observed similar early mortality between the 2 fixation techniques.

The question of whether cemented fixation of total hip arthroplasty (THA) increases early postoperative mortality, potentially by inducing bone cement implantation syndrome (BCIS) or thromboembolism, is fiercely debated (Olsen et al. 2014). The 90-day mortality after THA is reported to be 0.7% in a systematic review based on 32 observational studies including 1,129,330 patients (Berstock et al. 2014). The 3 most frequent causes of death within 90 days after THA are myocardial infarction, thromboembolism, and pneumonia (Pedersen et al. 2011). Since death is a rare complication after THA, only large register studies will have the statistical power to address the question of whether cemented THA fixation indeed confers an increased risk of early mortality. Analyses based on Norwegian, Finnish, or Swedish register data indicate similar 90-day mortality after cemented compared with cementless THA, but only a few studies were designed to adequately address comorbidity confounding using different methods (Garland et al. 2017, Dale et al. 2019, Ekman et al. 2019). The Norwegian study on 79,557 patients had access to ASA class, based on physicians’ preoperative assessment of overall health (Dale et al. 2019), whereas the Finnish study on 62,221 patients had information on comorbid conditions based on a hospital discharge register (Ekman et al. 2019). A Swed-

Acta Orthopaedica 2021; 92 (1): 46–52

ish study on 178,784 patients compared Table 1. Flow diagram a mortality after cemented and cementless THA adjusting for the Charlson comorbid Finland Denmark Norway Sweden All OA, first- All OA All OA All OA ity index (CCI), but the cohort exposed to time surgery 2005–2013 2005–2013 2005–2013 cementless fixation was only about a 10th the size of the cohort exposed to cemented No. received from each country 76,066 58,087 49,208 100,383 Keep 2005–2013 51,469 THA (Garland et al. 2017). That study is Keep only cemented/cementless also hampered by selection bias since the implants (hybrids are excluded) 42,452 48,015 34,721 86,193 practice in Sweden is to select younger and Keep non-simultaneous 40,618 47,361 34,639 85,215 Keep first surgery 41,474 30,907 75,608 fitter patients for cementless fixation. The Remove negative time to death 40,617 Nordic Arthroplasty Register Association Final study population 40,617 41,474 30,907 75,608 (NARA) was established to fuse 4 wella Some of the exclusion criteria were already applied on the national level before transferestablished Nordic arthroplasty databases ring data to NARA, whereas other exclusion criteria were applied in the NARA dataset. into a larger framework (Havelin et al. 2009). The choice of fixation techniques varies considerably among the participating countries, with a predominance of cemented fixation in hybrid THA were excluded from the study population. Thus, Sweden and Norway and a much larger proportion of cement- the final study population included 188,606 THA patients less fixation in Denmark and Finland (Mäkelä et al. 2014b). (Table 1). The index date was defined as the date of THA. Thus, the motivation for performing our study was the possibility to cope with the selection bias by analysing patients from Exposure all Nordic countries together and to improve the adjustment The exposure was THA fixation technique, including cemented for comorbidity confounding compared with previous studies. and cementless THA. Hybrid THAs were not included in In addition, we examine whether age, sex, and calendar year our study since the aim was not to examine the association are effect modifiers in the association between fixation and between exposure to hybrids and mortality. mortality, which was not examined in previous studies. We thus designed an international register-based cohort Outcome study aimed at comparing mortality within 0–14, 0–30, and The outcome was time to death within up to 90 days of the 0–90 days after cemented and cementless THA performed due index date. Information on death was available from the cento osteoarthritis with adjustment for comorbidities at the time tral statistics facilities in Denmark (Schmidt et al. 2014), Finof surgery. land, Norway, and Sweden and was linked to each patient in the national registers before transferring the data to NARA database.

Patients and methods Study design and sources of data The study was designed as a population-based cohort study on primary THA procedures performed in Denmark, Finland, Norway, and Sweden during 2005–2013, and the cohort was obtained from the combined NARA database. The database includes a common set of variables collected in all participating countries. Due to personal identification numbers used in all Nordic countries it is possible to further link national arthroplasty register data to data from national patient registries in each country, and then to anonymize and transfer data to a common NARA database. Study population Using the NARA database, we included all primary cemented and cementless THAs performed due to osteoarthritis during the period 2005 to 2013. Only 1st unilateral THAs were included. Some of the exclusion criteria were already applied on the national level before transferring data to NARA, whereas other exclusion criteria were applied in the NARA dataset. The

Potential confounders The following variables were included from the NARA database as potential confounders: age (≤ 59, 60–69, 70–79, and ≥ 80 years), sex, nation (Denmark, Norway, Finland, and Sweden), and year of THA (2005–2013). The level of comorbidity in each individual participant was assessed during the 2 years preceding the index date. The information on comorbidity was collected from the National Discharge Registries of Finland (Sund 2012), Denmark (Schmidt et al. 2015), Sweden (Ludvigsson et al. 2011), and Norway (Bakken et al. 2020) before transferring to the NARA database in anonymized form. We calculated the CCI in the adaptation suggested by Quan et al. (2011) (Charlson et al. 1987) and categorized comorbidities into 4 CCI scores: low (score 0: no known comorbidities), mild (score 1), moderate (score 2), and severe (score 3 and more). The look-back period for CCI calculation was 2 years prior to the index date and during the hospitalization for the index THA for the 19 comorbidity groups included in CCI and comorbidities that are not part of the CCI were not considered.

Statistics We tabulated characteristics of the study population by fixation technique. Numbers of patients and proportions in each category are presented. Means were used to describe age. Categorical data were summarized in cross-tables by fixation technique. We used the Kaplan–Meier method to estimate unadjusted cumulative survival probability by fixation technique. We fitted Cox regression models to estimate hazard ratios (HR) with 95% confidence intervals (CI). Overall HRs were adjusted for age (as continuous variable), sex, CCI score, nation, and year of THA surgery, and the inclusion of these covariates was based on the assessment of relevance and noninterference using directed acyclic graphs. Furthermore, we studied the association between fixation and mortality stratifying on age, sex, CCI score, and year of index THA surgery. We tested for interaction between age and fixation, as well as sex, nation and year of surgery and fixation. P-values less than 0.05 were considered statistically significant. We used the non-parametric Aalen additive regression model as a sensitivity analysis in order to assess graphically whether there were time-dependent effects of THA fixation on an additive scale, as opposed to the semi-parametric multiplicative Cox regression model. Follow-up started on the index date and ended on the day of death, or December 31, 2013, whichever came first. We used SAS version 9.4 (SAS Institute, Cary, NC, USA) to perform data management and analyses. The study follows the RECORD guidelines (Benchimol et al. 2015). Ethics, funding, and potential conflicts of interest Ethical approval was granted by the appointed authority in each participating country: the Ethics Boards of Lund (LU20-02) and Gothenburg (360/13 and T453/14) in Sweden, the Finnish National Institute of Health and Welfare (Dnro THL/1743/5.05.00/2014), the Norwegian Data Inspectorate (ref 24.1.2017: 16/01622-3/CDG and Ethical approval 2015/880/REK Vest), and the Danish Data protection agency (1-16-02-54-17). Data sharing is not possible. The study was supported by a grant from Aarhus University Research Foundation and by the Swedish Research Council (VR 201802612). The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. All authors have completed the ICMJE uniform disclosure form at http://www.icmje.org/conflicts-of-interest/ (available on request from the corresponding author) and declare that (1) no authors have received support from any company for the submitted work. The Department of Clinical Epidemiology at Aarhus University Hospital is, however, involved in studies with funding from various companies as research grants to (and administered by) Aarhus University. None of these studies have any relation to the present study; (2) no authors

Acta Orthopaedica 2021; 92 (1): 46–52

have personal financial relationships with any company that might have an interest in the submitted work in the previous 3 years; (3) no authors have non-financial interests that may be related to the submitted work. However, AE reports grants from DePuy Synthes, grants from Zimmer Biomet, personal fees from Zimmer Biomet, outside the submitted work. SO reports grants from Zimmer Biomet, outside the submitted work. NH reports grants and lecturer’s fees by Waldemar Link GmbH & Co. KG, Hamburg, and Heraeus GmbH, Wehrheim, Germany, outside the submitted work.

Results Characteristics of the study population Among the 188,606 included THA procedures, 58% were performed using cemented and 42% using cementless fixation (Table 2). The proportion of females in the entire study population was 58%, mean age at index surgery was 69 years (SD 9.7), and the majority of patients (83%) presented without comorbidities classified according to the modified CCI. The two fixation cohorts differed with respect to some baseline characteristics (Table 2). The proportion of females was higher in the cohort receiving cemented fixation, the proportion of patients younger than 60 years was considerably higher in the cohort operated with cementless fixation, and the proportion of patients with varying degrees of comorbidity was slightly higher among those receiving cemented fixation. The majority of all cemented procedures were performed in Sweden (59%), whereas most cementless procedures were registered in Finland and Denmark (36% and 39%). During the period 2005–2008, cemented fixation was more frequently used than cementless fixation in all 4 countries, whereas in the period 2009–2013 we observed the opposite pattern. Risk of 90-day mortality During the 1st 90 days of follow-up, the cumulative mortality after cemented and cementless THA procedures was estimated at 0.41% (CI 0.37–0.46) and 0.26% (CI 0.22–0.30), respectively (with 449 deaths after cemented and 208 deaths after cementless THA procedures). This was equivalent to an unadjusted HR of 1.6 (CI 1.4–1.9) for cemented compared with cementless fixation, but after adjustment for age, sex, CCI score, nation, and year of index THA, an HR of 0.97 (CI 0.79–1.2) was attained for cemented compared with cementless fixation. We then analyzed the adjusted risk of 90-day mortality in 4 separate age groups and found similar 90-day mortality between cemented and cementless fixation in the age groups 60–69, 70–79, and > 80 years of age. However, cemented fixation conferred an increased risk of death among patients below the age of 59, with an adjusted HR of 3.5 (CI 1.4–8.9; Table 3), but the absolute risk difference was 0.09 (CI –0.04 to 0.21). We tested for interaction between age, sex, nation,

Acta Orthopaedica 2021; 92 (1): 46–52

Table 2. Baseline characteristics of patients with cemented and cementless total hip arthroplasty (THA) in the Nordic Arthroplasty Registry Association database by country, 2005–2013. Values are count (%) Fixation:

All patients Denmark Norway Sweden Finland Cemented Cementless Cemented Cementless Cemented Cementless Cemented Cementless Cemented Cementless n = 108,572 n = 80,034 n = 9,954 n = 31,520 n = 22,772 n = 8,135 n = 64,481 n = 11,127 n = 11,365 n = 29,252

Sex Male 39,655 (37) 38,789 (49) 3,446 (35) 15,452 (49) 6,957 (31) 3,091 (38) 25,601 (40) 6,279 (56) 3,651 (32) 13,967 (48) Female 68,917 (63) 41,245 (51) 6,508 (65) 16,068 (51) 15,815 (69) 5,044 (62) 38,880 (60) 4,848 (44) 7,714 (68) 15,285 (52) Age at primary THA ≤ 59 5,720 (5.3) 22,525 (28) 215 (2.2) 7,017 (22) 1,132 (5.0) 2,188 (27) 4,186 (6.5) 6,147 (55) 187 (1.6) 7,173 (25) 60–69 29,760 (27) 33,165 (41) 1,305 (13) 13,981 (44) 5,843 (26) 3,096 (38) 21,050 (33) 4,048 (36) 1,562 (14) 12,040 (41) 70–79 49,377 (46) 20,033 (25) 5,418 (54) 8,659 (28) 10,326 (45) 2,076 (26) 27,250 (42) 798 (7.2) 6,383 (56) 8,500 (29) ≥ 80 23,715 (22) 4,311 (5.4) 3,016 (30) 1,863 (5.9) 5,471 (24) 775 (9.5) 11,995 (19) 134 (1.2) 3,233 (28) 1,539 (5.3) Charlson comorbidity index 0, none 89,105 (82) 66,939 (84) 7,119 (72) 25,153 (80) 19,646 (86) 7,028 (86) 53,210 (82) 10,085 (91) 9,130 (80) 24,673 (84) 1, mild 10,268 (9.5) 6,915 (8.6) 1,370 (14) 3,216 (10) 1,828 (8.0) 698 (8.6) 5,905 (9.2) 621 (5.6) 1,165 (10) 2,380 (8.1) 2, moderate 6,499 (6.0) 4,388 (5.5) 961 (9.7) 2,182 (6.9) 852 (3.7) 296 (3.6) 3,892 (6.0) 316 (2.8) 794 (7.0) 1,594 (5.4) 3, severe 2,700 (2.5) 1,792 (2.2) 504 (5.1) 969 (3.1) 446 (2.0) 113 (1.4) 1,474 (2.3) 105 (0.9) 276 (2.4) 605 (2.1) Year of primary THA 2005 17,385 (16) 5,446 (6.8) 1,795 (18) 2,186 (6.9) 3,663 (16) 595 (7.3) 9,415 (15) 767 (6.9) 2,512 (22) 1,898 (6.5) 2006 15,654 (14) 6,278 (7.8) 1,610 (16) 2,456 (7.8) 3,218 (14) 578 (7.1) 8,719 (14) 997 (9.0) 2,107 (19) 2,247 (7.7) 2007 14,396 (13) 7,145 (8.9) 1,438 (14) 2,901 (9.2) 3,351 (15) 461 (5.7) 8,031 (13) 1,250 (11) 1,576 (14) 2,533 (8.7) 2008 13,078 (12) 8,080 (10) 1,041 (10) 3,036 (9.6) 3,021 (13) 663 (8.1) 7,631 (12) 1,368 (12) 1,385 (12) 3,013 (10) 2009 12,537 (12) 9,904 (12) 866 (8.7) 4,401 (14) 2,626 (12) 860 (11) 8,004 (12) 1,502 (14) 1,041 (9.2) 3,141 (11) 2010 11,587 (11) 10,681 (13) 938 (9.4) 4,074 (13) 2,001 (8.8) 1,169 (14) 7,843 (12) 1,671 (15) 805 (7.1) 3,767 (13) 2011 10,831 (10) 11,226 (14) 820 (8.2) 4,120 (13) 1,803 (7.9) 1,196 (15) 7,476 (12) 1,777 (16) 732 (6.4) 4,133 (14) 2012 10,221 (9.4) 11,576 (15) 727 (7.3) 4,134 (13) 1,522 (6.7) 1,303 (16) 7,362 (11) 1,795 (16) 610 (5.4) 4,344 (15) 2013 2,883 (2.7) 9,698 (12) 719 (7.2) 4,212 (13) 1,567 (6.9) 1,310 (16) 597 (5.3) 4,176 (14)

and year of surgery and fixation within 90 days and all p-values were > 0.05. The risk of revision within 90 days was 0.7% among cemented and 1.4% among cementless THA. Risk of 14- and 30-day mortality Mortality within 14 days was also analyzed, and during this time frame 158 patients (0.14%, CI 0.10–0.19) with cemented THA and 72 (0.08%, CI 0.05–0.13) who had received cementless THA died, translating into an unadjusted HR of 1.6 (CI 1.2–2.1) when comparing cemented with cementless THA (Table 4). When we fitted a Cox regression model adjusting for age, sex, CCI score, nation, and year of index THA, an adjusted HR of 0.91 (CI 0.64–1.3) for cemented compared with cementless THA was attained (Table 4). Mortality within 30 days was 0.21% (CI 0.17–0.26) for cemented THA and 0.12% (CI 0.09–0.17) for cementless fixation. The adjusted HR was 0.94 (CI 0.71–1.3) for cemented compared with cementless THA within 30 days of THA. There was no statistically significant interaction between age and fixation, nor between sex, nation, and year of surgery and fixation within either 14 or 30 days (all p values were > 0.05). The Aalen additive risk model confirmed a minor cumulated increased excess risk of death after cemented compared with cementless THA in the initial period after THA (up to 10 days), whereas no difference in the risk was observed afterwards (Figure). Overall, the Aalen model suggested no effect of cementation on the mortality up to 90 days. Since the inten-

sity is low (i.e., a low risk of dying) the curve can be interpreted as (almost) the cumulative additional number of dead patients for cemented (compared with cementless) over time.

Discussion Main findings In this hitherto largest register study on the risk of death after THA accounting for comorbidity as an important confounder, we find no clinically relevant difference in the absolute or relative risk of death up to 90 days for cemented compared with cementless fixation. Cementing is the gold standard for implant fixation in elderly patients (Mäkelä et al. 2014a). Even though cemented fixation is considered superior in terms of implant survival within the 1st decade, the surgeons fear BCIS, with at worst cardiac arrest and death of the patient (Donaldson et al. 2009). Due to its scarcity, risk factors behind BCIS are difficult to investigate; however, anecdotal evidence and a previous investigation into early mortality after THA including classical and reverse hybrids indicates that cementation of the femoral but not the acetabular component is the most dangerous moment (Garland et al. 2017). The trend towards using cementless THA may be driven by forceful marketing and production-oriented streamlining of operative procedures, as well as shorter operating times and focus on risk of BCIS (Troelsen et al. 2013, Dale et al. 2019).

Acta Orthopaedica 2021; 92 (1): 46–52

Table 3. Crude and adjusted hazard ratios (HR) with 95% confidence interval (CI) for 90-day mortality comparing cemented with cementless total hip arthroplasties (THA) a Fixation

Number Number Crude of deaths at risk HR (95% CI)

Adjusted HR (95% CI)

Among all Cementless 208 80,034 Reference Reference Cemented 449 108,572 1.6 (1.4–1.9) 0.97 (0.79–1.2) Stratified by age ≤ 59 Cementless 24 22,525 Reference Reference Cemented 11 5,720 1.8 (0.88–3.7) 3.5 (1.4–8.9) 60–69 Cementless 51 33,165 Reference Reference Cemented 37 29,760 0.81 (0.53–1.2) 1.6 (0.63–2.1) 70–79 Cementless 78 20,033 Reference Reference Cemented 164 49,377 0.85 (0.65–1.1) 1.1 (0.76–1.5) ≥ 80 Cementless 55 4,311 Reference Reference Cemented 237 23,715 0.78 (0.58–1.1) 1.1 (0.77–1.5) Stratified by gender Male Cementless 136 38,789 Reference Reference Cemented 212 39,655 1.5 (1.2–1.9) 0.81 (0.61–1.1) Female Cementless 72 41,245 Reference Reference Cemented 237 68,917 2.0 (1.5–2.6) 1.2 (0.88–1.7) Stratified by Charlson comorbidity index 0, no comorbidity Cementless 128 66,939 Reference Reference Cemented 270 89,105 1.6 (1.3–2.0) 0.83 (0.63–1.1) 1, mild comorbidity Cementless 41 6,915 Reference Reference Cemented 69 10,268 1.1 (0.77–1.7) 1.1 (0.67–1.8) 2, moderate comorbidity Cementless 16 4,388 Reference Reference Cemented 46 6,499 1.9 (1.1–3.4) 1.2 (0.60–2.6) 3, severe comorbidity Cementless 23 1,792 Reference Reference Cemented 64 2,700 1.9 (1.2–3.0) 1.3 (0.72–2.3) Stratified by year of primary operation 2005–2008 Cementless 75 26,949 Reference Reference Cemented 255 60,513 1.5 (1.2–2.0) 0.83 (0.61–1.1) 2009–2013 Cementless 133 53,085 Reference Reference Cemented 194 48,059 1.6 (1.3–2.0) 1.2 (0.92–1.6) a We

adjusted for age, gender, comorbidity, nation, and year of THA surgery in overall analyses. In stratified analyses, we adjusted for all variables expect the stratification variable. For each stratification category a new Cox model was fitted.

The true incidence of death secondary to BCIS is unknown in patients undergoing THA due to osteoarthritis, but BCIS seems to be more frequent in elderly patients receiving hemiarthroplasty due to femoral neck fracture compared with elective patients with osteoarthritis receiving THA (Olsen et al. 2014). Previous register studies indicate that the risk of BCIS and even delayed thromboembolic events may have been overestimated, at least in patients receiving THA for reasons other than femoral neck fracture. Comparisons of cemented with cementless fixation support the notion that cementation

Table 4. Crude and adjusted hazard ratios (HR) with 95% confidence intervals (CI) for 0–14- and 0–30-day mortality comparing cemented with cementless total hip arthroplasties Fixation

Number Number Crude of deaths at risk HR (95% CI)

Adjusted HR (95% CI)

0–14-day mortality Cementless 72 80,034 Reference Reference Cemented 158 108,572 1.6 (1.2–2.1) 0.91 (0.64–1.3) 0–30-day mortality Cementless 104 80,034 Reference Reference Cemented 234 108,572 1.7 (1.3–2.1) 0.94 (0.71–1.3) HR adjusted for age, gender, year of surgery, and comorbidity.

Aalen additive plot with cumulative coefficients (y-axis) with 95% confidence interval comparing cemented with cementless total hip arthroplasties within 90 days of surgery.

is associated with a very small mortality risk increase, if at all, both in absolute and in relative terms (Dale et al. 2019, Ekman et al. 2019). As an exception to this, an analysis based on data from the National Joint Registry of England & Wales dataset, led the authors to conclude that there is a statistically significant 11% increased risk of death associated with cemented compared with cementless THA (McMinn et al. 2012), but this interpretation has been contested (Whitehouse et al. 2014). Likewise, we showed no difference overall and in the most representative age groups, namely 60–69, 70–79, and +80 years of age. Our observation of an increased relative risk of mortality for cemented versus cementless THA only in the youngest age group is unexpected, but it is most likely the result of a combination of residual confounding, selection bias, and chance. Strengths and weaknesses of this study The important strength of our study is the unique collaboration of 4 national registers used to establish the NARA data-

Acta Orthopaedica 2021; 92 (1): 46–52

base, comprising a high number of patients with information on comorbidity. This is particularly important when studying rare outcomes such as early death after THA surgery. The quality of the participating national hip registries including both completeness and accuracy of data is high (Pedersen and Fernstad 2016). Adjustment for comorbidity in studies on mortality after THA surgery is important, but only a few studies on samples of similar size to ours have access to complex health administrative datasets containing information on inpatient diagnoses prior to surgery, and the use of arthroplasty registry data alone to model postoperative mortality has been criticized (Whitehouse et al. 2014). Thus, in some studies, the ASA classification is used as a proxy of comorbidity, but internal variability within this classification is high. The considerably more complex CCI as a measure of comorbidity is a well-established tool to adjust for comorbidity in studies aimed at investigating short-term mortality after THA (Pedersen et al. 2011, Garland et al. 2017). Any misclassification of CCI is most probably not related to the mode of THA fixation since the underlying healthcare data is prospectively and routinely collected. Our study has many limitations. Our choice of 90-day mortality as outcome can be questioned although the incidence of adverse events within 90 days after surgery is regularly reported as a measure of complications by many established arthroplasty registers. The longer time has elapsed after major surgery the less likely it is that the fixation mode of the index THA is related to the event of death (Pedersen et al. 2011). THA patients are elderly and can die due to numerous other reasons such as comorbid conditions or complications unrelated to the THA fixation technique. Thus, some studies investigating mortality after THA surgery performed due to fracture or osteoarthritis focus on earlier time points, and some address each day after surgery by a separate logistic regression analysis (Dale et al. 2019, Ekman et al. 2019). We therefore additionally explored 14- and 30-day mortality as measures of death occurring earlier after index surgery, with findings supporting our main conclusions. In order to account for risk modifications early during the observation period we also undertook Aalen additive risk modelling and found that there was a small risk increase associated with cemented fixation during the first 14 days after surgery but not thereafter. Socioeconomic factors are associated with morbidity, mortality, and patient-reported outcome after THA, but we had no access to such information. However, a recent investigation into the added benefit of including socioeconomic background variables in analyses of early mortality after THA indicates that the level of comorbidity may, to a certain extent, serve as a proxy measure of socioeconomic background (Weiss et al. 2019). In addition, the Scandinavian healthcare systems provide public, tax-financed care, making socioeconomic factors less pronounced. Further, we did not have information on lifestyle factors. It is possible that younger patients with cementless THA are more physically active and fitter, influencing their risk of suffering

from both intraoperative and early postoperative complications. The NARA dataset contains no information on causes of deaths, which would have enabled us to explore whether the youngest age group dies for other reasons than elderly patients early after THA surgery. Heterogeneity in data from disparate sources and between-country variation can be problematic, which would affect our study design with data from 4 Nordic countries, but this seems primarily to be an issue when pooled data are used for individual survival-probability assessments, an approach not utilized in our study (BartzJohannessen et al. 2019). We also investigated revision rates early after index THA and found 0.7% differences to the disadvantage of cementless fixation within 90 days, mostly due to a pronounced increased risk of periprosthetic fractures when compared with cemented fixation. If one used cementless fixation in order to avoid cementation-related mortality one would have to take the increased risk of early revision and the mortality associated with such secondary procedures into account. Finally, we had no information on the use of mechanical or chemical thromboprophylaxis, which are associated with 90-day mortality after THA (Hunt et al. 2013, Pedersen et al. 2019). If use of thromboprophylaxis is related to fixation method we would be confronted with residual confounding in our study. We did not examine the association between hybrid THA and mortality. Clinical implications Our study is based on a large population of THA patients from 4 Nordic countries and examines the association between THA fixation and mortality, granting generalizability of the study results. We interpret the findings from this and other studies as supportive of the continued use of cemented fixation of THA performed due to osteoarthritis although an observational study such as ours cannot entirely establish causal relationships between mode of fixation and mortality. Conclusion After adjustment for comorbidity as an important confounder, we observed similar early mortality between the 2 fixation techniques, cemented and cementless. This paper is the result of international teamwork. All authors contributed substantially to the design of the study, interpretation of data, critical revision of the manuscript, and final approval of the submitted version. ABP and AM contributed to the statistical analyses. NPH and ABP drafted the manuscript together, and accept full responsibility for the work, the conduct of the study, and access to the data. Acta thanks Adrian Sayers and Sarah Whitehouse for help with peer review of this study. Bakken I J, Ariansen A M S, Knudsen G P, Johansen K I, Vollset S E. The Norwegian Patient Registry and the Norwegian Registry for Primary Health Care: research potential of two nationwide health-care registries. Scand J Public Health 2020; 48(1): 49-55.

Bartz-Johannessen C, Furnes O, Fenstad A M, Lie S A, Pedersen A B, Overgaard S, Kärrholm J, Malchau H, Mäkelä K, Eskelinen A, Wilkinson J M. Homogeneity in prediction of survival probabilities for subcategories of hipprosthesis data: the Nordic Arthroplasty Register Association, 2000– 2013. Clin Epidemiol 2019; 11: 519-24. Benchimol E I, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, Sorensen H T, von Elm E, Langan S M, Record Working Committee. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med 2015; 1 (10): e1001885. Berstock J R, Beswick A D, Lenguerrand E, Whitehouse M R, Blom A W. Mortality after total hip replacement surgery: a systematic review. Bone Joint Res 2014; 3(6): 175-82. Charlson M E, Pompei P, Ales K L, MacKenzie C R. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation, J Chronic Dis 1987; 40(5): 373-83. Dale H, Borsheim S, Kristensen T B, Fenstad A M, Gjertsen J E, Hallan G, Lie S A, Furnes O. Perioperative, short-, and long-term mortality related to fixation in primary total hip arthroplasty: a study on 79,557 patients in the Norwegian Arthroplasty Register. Acta Orthop 2019: 1-7. Donaldson A J, Thomson H E, Harper N J, Kenny N W. Bone cement implantation syndrome. Br J Anaesth 2009; 102(1): 12-22. Ekman E, Palomaki A, Laaksonen I, Peltola M, Hakkinen U, Mäkelä K. Early postoperative mortality similar between cemented and uncemented hip arthroplasty: a register study based on Finnish national data. Acta Orthop 2019; 90(1): 6-10. Garland A, Gordon M, Garellick G, Kärrholm J, Skoldenberg O, Hailer N P. Risk of early mortality after cemented compared with cementless total hip arthroplasty: a nationwide matched cohort study. Bone Joint J 2017; 99-b(1): 37-43. 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. Hunt L P, Ben-Shlomo Y, Clark E M, Dieppe P, Judge A, MacGregor A J, Tobias J H, Vernon K, Blom A W. 90-day mortality after 409,096 total hip replacements for osteoarthritis, from the National Joint Registry for England and Wales: a retrospective analysis. Lancet 2013; 382(9898): 1097-104. 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. 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. Failure rate of cemented and uncemented total hip replacements: register study of combined Nordic database of four nations. BMJ 2014a; 348: 7592.

Acta Orthopaedica 2021; 92 (1): 46–52

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. McMinn D J, Snell K I, Daniel J, Treacy R B, Pynsent P B, Riley R D. Mortality and implant revision rates of hip arthroplasty in patients with osteoarthritis: registry based cohort study. BMJ 2012; 344: e3319. Olsen F, Kotyra M, Houltz E, Ricksten S E. Bone cement implantation syndrome in cemented hemiarthroplasty for femoral neck fracture: incidence, risk factors, and effect on outcome. Br J Anaesth 2014; 113(5): 800-6. Pedersen A B, Fernstad A M. Nordic Arthroplasty Register Association (NARA) report 2016; http://nrlweb.ihelse.net/NARA_2015_ORIG_ny.pdf. Accessed April 6, 2020. Pedersen A B, Baron J A, Overgaard S, Johnsen S P. Short- and long-term mortality following primary total hip replacement for osteoarthritis: a Danish nationwide epidemiological study. J Bone Joint Surg Br 2011; 93(2): 172-7. Pedersen A B, Andersen I T, Overgaard S, Fenstad A M, Lie S A, Gjertsen J A, Furnes O. Optimal duration of anticoagulant thromboprophylaxis in total hip arthroplasty: new evidence in 55,540 patients with osteoarthritis from the Nordic Arthroplasty Register Association (NARA) group. Acta Orthop 2019; 90(4): 298-305. Quan H, Li B, Couris C M, Fushimi K, Graham P, Hider P, Januel J-M, Sundararajan V. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol 2011; 173(6): 676-82. Schmidt M, Pedersen L, Sorensen H T. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 2014; 29(8): 541-9. Schmidt M, Schmidt S A, Sandegaard J L, Ehrenstein V, Pedersen L, Sorensen H S. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7: 449-90. Sund R. Quality of the Finnish Hospital Discharge Register: a systematic review. Scand J Public Health 2012; 40(6): 505-15. Troelsen A, Malchau E, Sillesen N, Malchau H. A review of current fixation use and registry outcomes in total hip arthroplasty: the uncemented paradox. Clin Orthop Relat Res 2013; 471(7): 2052-9. 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. Whitehouse S L, Bolland B J, Howell J R, Crawford R W, Timperley A J. Mortality following hip arthroplasty: inappropriate use of National Joint Registry (NJR) data. J Arthroplasty 2014; 29(9): 1827-34.

Acta Orthopaedica 2021; 92 (1): 53–60

Awareness of performance on outcomes after total hip and knee arthroplasty among Dutch orthopedic surgeons: how to improve feedback from arthroplasty registries Peter VAN SCHIE 1,2, Leti VAN BODEGOM-VOS 2, Tristan M ZIJDEMAN 2, Rob G H H NELISSEN 1, and Perla J MARANG-VAN DE MHEEN 2 1 Department of Orthopedics, Leiden University Medical Centre, Leiden; 2 Department of Biomedical Data Sciences, Medical Decision Making, Leiden University Medical Centre, Leiden, the Netherlands Correspondence: p.van_schie@lumc.nl Submitted 2020-03-12. Accepted 2020-07-01.

Background and purpose — The Netherlands Registry of Orthopedic Implants (LROI) uses audit and feedback (A&F) as the strategy to improve performance outcomes after total hip and knee arthroplasty (THA/TKA). Effectiveness of A&F depends on awareness of below-average performance to initiate improvement activities. We explored the awareness of Dutch orthopedic surgeons regarding their performance on outcomes after THA/TKA and factors associated with this awareness. Methods — An anonymous questionnaire was sent to all 445 eligible Dutch orthopedic surgeons performing THA/ TKA. To assess awareness on own surgeon-group performance, they were asked whether their 1-year THA/TKA revision rates over the past 2 years were below average (negative outlier), average (non-outlier), above average (positive outlier) in the funnel plot on the LROI dashboard, or did not know. Associations were determined with (1) dashboard login at least once a year (yes/no); (2) correct funnel-plot interpretation (yes/no) and; (3) recall of their 1-year THA/ TKA revision rate (yes/no). Results — 44% of respondents started the questionnaire, 158 THA and 156 TKA surgeons. 55% of THA surgeons and 55% of TKA surgeons were aware of their performance. Surgeons aware of their performance more often logged in on the LROI dashboard, more often interpreted funnel plots correctly, and more often recalled their revision rate. 38% of THA and 26% of TKA surgeons scored “good” on all 3 outcomes. Interpretation — Only half of the orthopedic surgeons were aware of their performance status regarding outcomes after THA/TKA. This suggests that to increase awareness, orthopedic surgeons need to be actively motivated to look at the dashboard more frequently and educated on interpretation of funnel plots for audit and feedback to be effective.

Several studies have shown large between-hospital variation in performance outcomes after total hip and knee arthroplasty (THA/TKA) including revision rates, suggesting opportunities to improve care (Siciliani et al. 2013, Bozic et al. 2014, Menendez et al. 2016, Weeks et al. 2016, Fry et al. 2017, van Schie et al. 2020). Audit and feedback (A&F) is a frequently used approach to reduce between-hospital variation, and defined as provision of clinical performance summaries to healthcare providers or organizations intended to initiate activities to improve performance (Brehaut and Eva 2012, Ivers et al. 2014). Worldwide, A&F from arthroplasty registries is provided in different ways. In the Netherlands, performance indicators such as revision rates, patient-reported outcome measures (PROMs), and patient characteristics are shown on surgeon-group level in a real-time password-protected webbased dashboard and the extent of variation is shown in an anonymized version in annual reports. Following a Cochrane review of 140 studies from multiple fields, A&F is effective with a median absolute improvement of 4% of the desired outcome, but with the effect size varying from a 9% decrease to a 70% increase (Ivers et al. 2012). Part of the reason for this large variation in effectiveness may be the varying degree to which A&F leads to an increased awareness of own performance. For example, A&F is not received, information including graphs (e.g., funnel plots) and/or tables is not interpreted correctly, or the reported performance outcomes are not considered interesting (Gude et al. 2018). Sufficient awareness of own performance relative to others in combination with motivation to improve is more likely to result in targeted quality improvement initiatives (Davis et al. 2006, van der Veer et al. 2010, de Vos Maartje et al. 2013). Due to a lack of awareness of own performance, it is often overestimated (Gude et al. 2018). This can limit quality

improvement initiatives, because it is assumed that performance is good even though there may be room for improvement. Furthermore, it is important that performance indicators give sufficient direction on where to improve care, so that professionals are able to select focused interventions to improve care. A recent study showed that for most surgeon groups with significantly higher revision rates, the direction of improvement could be pointed out by looking at the reason for revision (e.g., infection, prosthesis loosening, dislocation etc.) (van Schie et al. 2020). By looking at a more specific outcome, professionals can figure out in which part of the care process improvements are possible, e.g., timing of antibiotic prophylaxis (infection), cementation techniques (prothesis loosening), or femoral head size (dislocation). We explored the awareness of orthopedic surgeons regarding their performance on outcomes after THA/TKA and factors associated with this awareness, to gain insight into the ways to increase the effectiveness of A&F provided by the LROI.

Methods An anonymous internet-based questionnaire study was performed in December 2018 to explore the awareness of orthopedic surgeons on outcomes after THA/TKA provided by the LROI and associated factors. Netherlands Registry of Orthopedic Implants (LROI) The LROI was established in 2007 and in 2012 all Dutch surgeon groups participated. In 2015, the LROI dashboard was developed to allow surgeons to better monitor their performance showing information on the number of procedures performed, revision rates, PROMs and patient characteristics on surgeongroup level compared with other surgeon groups, which can be viewed at any time. The completeness for primary THA and TKA procedures is checked against Electronic Health Records and is currently above 98% for primary procedures and 96% for revisions (van Steenbergen et al. 2015). 97 surgeon groups performed THA and 98 performed TKA in the study period. Study population The questionnaire was sent to all 445 Dutch orthopedic surgeons performing primary THA/TKA who were members of the hip and knee working groups from the Dutch Orthopedic Association. Reminders were sent by email 4 and 8 weeks after the first invitation. The survey was compiled using NetQ software (version 2014.Q3; https://cumulusnetworks.com/ products/netq/). Survey The information collected with the survey regarding the feedback provided on the LROI dashboard is divided into 4 parts (Appendix, see Supplementary data).

Acta Orthopaedica 2021; 92 (1): 53–60

Eligible orthopedic surgeons invited to the questionnaire n = 445 Responded n = 194 (44%) Answered all questions n = 169 (87%a)

Figure 1. Respondent flowchart. respondents.

Did not respond n = 251 (56%)

Did not answer all questions n = 25 (13%a) a

Percentage of total number of

In the first part, awareness regarding possible deviating performance (outlier status) of their own surgeon group over the last 2 years was assessed by asking whether their 1-year revision rate was below average (negative outlier), average (non-outlier), above average (positive outlier), in the funnel plot on the LROI dashboard, or they did not know. Second, we searched for 3 potential underlying factors that might be related to the level of awareness. It was assessed whether respondents (1) logged in at least once a year on their LROI dashboard; (2) were able to interpret funnel plots correctly; (3) could recall the 1-year revision rate of their surgeon group. Respondents answering that they did not know were counted as giving a non-positive answer. By combining these 3 questions, a composite outcome was created. A respondent only scored “good” when all 3 individual measures were positive, i.e., he/she logged in at least once a year, correctly interpreted the funnel plots, and could recall his/her 1-year revision rate. We also asked about hospital work setting (university/teaching/general hospital or private clinic) and number of arthroplasties performed annually (< 50, 50–100, > 100). Third, respondents were asked about quality improvement initiatives following possible below-average performance (negative outlier) in the past 2 years, and whether the effects of these initiatives were checked using the available feedback information on the LROI dashboard. Finally, there were questions about perceived need for changes in the current feedback, which current performance indicators were considered important, which indicators should be added to improve healthcare and the preferred frequency (every 1, 3, 6, or 12 months) and way of receiving feedback (tailored for their surgeon group or ability to make selections and explore the data oneself). Statistics Analyses were performed separately for THA and TKA surgeons. First, the proportion of respondents who were aware of deviating performance for their own surgeon group in the past 2 years was assessed. To examine the associations between awareness of deviating performance and the predefined potentially underlying factors (login to the dashboard, correct interpretation of funnel plots, recall of their own revision rate), univariate logistic regression analyses were performed. All questions answered by respondents regardless of whether they completed the full survey were included in the analyses. If sur-

Acta Orthopaedica 2021; 92 (1): 53–60

Table 1. Characteristics of the respondents (n = 194). Values are frequency (%) Factor Male sexle Age < 40 40–50 51–60 > 60 Hospital setting University medical center Teaching hospital General hospital Private clinic Type of surgeon a Performing THA Performing TKA Performing THA and TKA No. of THAs/surgeon/year b < 50 50–100 > 100 No response No. of TKAs/surgeon/year c < 50 50–100 > 100 No response

n (%)

Table 2. Associations between awareness of surgeon-group performance and logging in to dashboard, correct funnel-plot interpretation, and knowledge of 1-year revision rate

177 (91)

All respondents performing

32 (16) 101 (52) 42 (22) 19 (10)

THA (n = 158) 105 (67) Aware of surgeon-group performance Yes (n = 77) 69 (90) 7.6 (3.2–18) No (n = 64) 34 (53) Reference

96 (61)

TKA (n = 156) 103 (66) Aware of surgeon-group performance Yes (n = 78) 65 (83) 4.1 (1.9–9.0) No (n = 64) 35 (55) Reference

95 (61)

20 (10) 72 (37) 78 (40) 24 (13) 158 (81) 156 (80) 120 (62)

Logging on to LROI dashboard a Yes, n (%) OR (CI) d

Correct funnel-plot interpretation b Yes, n (%) OR (CI) d

59 (77) 2.4 (1.2–4.9) 37 (58) Reference

56 (72) 1.6 (0.8–3.3) 39 (61) Reference

a Logging in at least once every year. b Correctly interpreted both funnel plots. c Know the 1-year revision rate of their healthcare d OR (CI) = odds ratio (95% confidence interval).

34 (21) 75 (48) 46 (29) 3 (2) 37 (24) 78 (50) 32 (20) 9 (6)

Does the respondent perform only THA, only TKA, or both THA and TKA? b There were 158 THA surgeons. c There were 156 TKA surgeons..

geons stopped the survey but answered the previous question, we assumed there was a reason for stopping at that specific question (e.g., because it would be not acceptable to say not logging in) and coded this question as “don’t know,” meaning these were included as non-positive answers. In addition, we examined whether the composite outcome differed across hospital settings and number of THAs/TKAs performed annually. Data were analyzed with the statistical software SPSS version 25 (IBM Corp, Armonk, NY, USA). P-values < 0.05 were considered statistically significant in all analyses. Ethics, funding, and potential conflicts of interest The LUMC Medical Ethical Committee waived the need for ethical approval under Dutch law (CME, G18.140). Author PvS received a grant from the Van Rens Foundation (VRF2018-001) to perform this study. The authors declare that there are no conflicts of interest.

Results From 445 invited orthopedic surgeons, 194 (44%) started the survey; 158 surgeons performed THA and 156 TKA. 78 answered the questions within 4 weeks, 56 after the 1st, and 60 after the 2nd reminder. 169 (87%) respondents completed

Knowledge of 1-year revision rate c Yes, n (%) OR (CI) d 105 (67) 66 (86) 4.4 (2.0–9.8) 37 (58) Reference 103 (66) 66 (85) 4.9 (2.2–11) 34 (53) Reference

center of the past 2 years.

the survey (Figure 1). Median time to complete the survey was 6.4 minutes (interquartile range 5.3–8.5). 91% of respondents were male and 52% were between 40 and 50 years old. Most respondents (40%) were employed in a general hospital and evenly distributed across volume groups for THA and TKA (Table 1). Awareness of performance and underlying factors (Tables 2 and 3) 158 THA surgeons answered the questions on logging in, funnel-plot interpretation, and recalling their revision rate. Only 141 THA surgeons answered the questions on awareness of their surgeon-group performance, with 77 (55%) THA surgeons indicating awareness of any deviating performance in their surgeon group over the past 2 years. Among the 158 THA surgeons, 105 (67%) logged in on the LROI dashboard at least once a year, 96 (61%) interpreted the funnel plot correctly, and 105 (67%) recalled their 1-year revision rate. THA surgeons who were aware of any deviating performance were 8 times more likely to log in, twice as likely to correctly interpret the funnel plot, and 4 times more likely to recall their 1-year revision rate. Overall, 66 (38%) respondents scored “good” on all these individual items and thus on the composite outcome. THA surgeons who are aware of deviating performance were 5 times more likely to score “good” on the composite outcome. 156 TKA surgeons answered the questions on logging in, funnel-plot interpretation, and recalling their revision rate. Only 142 TKA surgeons answered the questions on awareness of own surgeon-group performance, with 78 (55%) TKA surgeons indicating awareness of any deviating performance in their surgeon group over the past 2 years. Among the 156 TKA surgeons, 103 (66%) logged in to the LROI dashboard at least once a year, 95 (61%) interpreted the funnel plot correctly, and 103 (66%) recalled their 1-year revision rate. TKA surgeons who were aware of any deviating performance were 4 times more likely to log in, twice as likely to correctly inter-

Acta Orthopaedica 2021; 92 (1): 53–60

Table 3. Composite outcome, stratified by hospital setting and number of arthroplasties performed annually

Distribution (%) 60 Most interesting Interesting

Composite outcome Odds good, n ratio THA surgeons All THA-performing respondents (n = 158) Aware of surgeon-group performance Yes (n = 77) No (n = 64) Hospital setting (n = 158) University medical center Teaching hospital General hospital Private clinic No. of THAs performed per year (n = 155) a < 50 50–100 > 100 TKA surgeons All TKA-performing respondents (n = 156) Aware of surgeon-group performance: Yes (n = 78) No (n = 64) Hospital setting (n = 156) University medical center Teaching hospital General hospital Private clinic No. of TKAs performed per year (n = 147) a < 50 50–100 > 100 a Number

40 30 20

60 46 14

5.3 Ref.

4 19 31 6

0.4 0.6 Ref. 0.7

7 35 18

0.4 1.4 Ref.

41 31 10

Uninteresting Most uninteresting

10 0

Number of procedures performed

1-year revision rate

PROMs

Patient characteristics

Figure 2. Currently available performance indicators on the secure LROI dashboard ranked from most to least interesting by respondents. LROI = Netherlands Registry of Orthopedic Implants; PROMs = patient-reported outcome measures. Yes, interesting Not interesting

Prosthesis survival

3.6 Ref.

2 15 23 1

0.3 0.6 Ref. 0.1

7 28 6

2.4 1.0 Ref.

of arthroplasties performed per year by the respondent.

pret the funnel plot, and 5 times more likely to recall their 1-year revision rate. Overall, 41 (26%) respondents scored “good” on the composite outcome and TKA surgeons who are aware of deviating performance were 4 times more likely to score “good” on the composite outcome. The proportion of surgeons who met the criteria of the composite outcome did not differ in the number of arthroplasties performed annually or across hospital settings, except for a lower proportion for TKA surgeons in private clinics. Quality improvement initiatives 20 respondents indicated that they were employed in a healthcare center that had a significantly higher 1-year revision rate (negative outlier) in the past 2 years. 9 of them did not see this deviating performance coming, because they had never checked the LROI dashboard for performance indicators. 17 indicated that quality improvement initiatives had been introduced and all of them used performance indicators from the LROI dashboard to monitor the effect. A positive effect of these initiatives on the revision rate was reported by 9 respondents and a negative effect by 3 respondents when checking progress in the LROI dashboard. 5 respondents were currently following the effect.

Complications Change in PROMs Readmission Length of stay 0

100

Distribution (%)

Figure 3. Percentage of orthopedic surgeons interested in additional performance indicators. Change = difference between pre- and postoperative PROMs. For abbreviations, see Figure 2

Future feedback From the current available performance indicators, the number of procedures performed was mostly considered as the most interesting information on the LROI dashboard, followed by 1-year revision rates, PROMs, and patient characteristics respectively (Figure 2). Prosthesis survival and complications are currently not available on the LROI dashboard, but 138 (82%) THA surgeons and 129 (76%) TKA surgeons indicated this information to constitute relevant indicators (Figure 3). 106 (62%) respondents would prefer to receive feedback every 6 months, and a minority every month (n = 6, 4%), or every quarter (n = 40, 23%), with some respondents having no preference (n = 18, 11%). 139 (82%) respondents prefer feedback that is tailored to their surgeon group without making any selections and 30 respondents (18%) indicated preferring to make their own selections of LROI indicators.

Discussion Although Dutch orthopedic surgeons performing THA/TKA can view their surgeon-group performance on a web-based

Acta Orthopaedica 2021; 92 (1): 53–60

A&F dashboard, only half of them are actually aware of their performance over the past 2 years. This lack of awareness of own performance and the associations found in our study suggests that orthopedic surgeons need to be actively motivated to log in more often, need to be educated on how to interpret funnel plots correctly, and must be able to reproduce their revision rate for the A&F to be effective in improving care. To act on the feedback information all underlying factors must be met, but this was the case in only one-third of THA surgeons and one-fourth of TKA surgeons and was fairly similar across different types of hospitals and annual volume. Yet, it seems important to increase the effectiveness of feedback, given that 9 out of 20 respondents in the negative outlier surgeon groups indicated that they did not see their worsening performance coming. Without effective feedback, surgeon groups would continue to provide care without modification, while 17 of these 20 respondents indicated that they conducted quality improvement initiatives once identified as showing poor performance. Differences and similarities between national arthroplasty registries in providing A&F The way in which A&F is offered varies, from publicly available annual reports including only nationwide averages with sometimes additional surgeon-group-specific performance, whereas others publish their indicators on surgeon-group level and surgeon level only in password-protected online dashboards (Li et al. 1999, Itonaga et al. 2000, Tabak et al. 2002, Bonutti et al. 2017, Kurcz et al. 2018, Talmo et al. 2018, Assi et al. 2019, Pelt et al. 2019, Porter et al. 2019, Yoon et al. 2019). The LROI, National Joint Registry in the United Kingdom (NJR), and Swedish Hip Arthroplasty Registries (SHAR) use a web-based password-protected A&F dashboard to provide surgeons with peer-comparison indicators in visual graphs on surgeon-group level and in the United Kingdom also on surgeon level (Toomey et al. 2001, Tabak et al. 2002, Assi et al. 2019, Porter et al. 2019, Yoon et al. 2019). In contrast, the Swedish Knee Arthroplasty Registries (SKAR) and the Danish Hip Arthroplasty Registries (DHAR) make no use of online dashboards, where the SKAR publishes only some indicators (e.g., patient demographics and PROMs) on its publicly accessible website once a year. Some arthroplasty registers may inform participating hospitals once a year about their performance, e.g., by emailing performance indicators without this being listed on their website. The feedback generated by the NJR is updated every 6 months, which was also indicated as the preferred frequency to receive feedback by two-thirds of respondents in our study (Porter et al. 2019). The Finnish Arthroplasty Register (FAR) even uses a daily updated publicly accessible website, which includes patient demographics and revision rates at surgeon-group level (Tabak et al. 2002). What all these different methods of feedback have in common is that it is passive education not requiring any action, which may be one of the explanations for orthopedic surgeons being

unaware of their performance. Public availability of performance indicators may increase the likelihood of action being taken, given that both patients and other stakeholders such as insurance companies can review the data and may use them in their decision-making. Comparison with literature Besides the Cochrane Review, there are more studies that found wide variation in the effect of A&F (Ivers et al. 2012). A review evaluating interactive computer feedback found a highly variable effect of improvement in quality of care in 3 out of 7 studies (Tuti et al. 2017). Another more recent study found a significant improvement for 4 out of 6 performance indicators, 2.5 years after implementation of online A&F interventions in maternal-newborn hospitals (Weiss et al. 2018). Given the varying effect of A&F, the results of our study can make a relevant contribution to further improve current feedback as provided by arthroplasty registries. We have gained insight into whether A&F reached the target group (i.e., how often surgeons log in), the ability to interpret the funnel plot, and recall of revision rates. In addition, we investigated which performance indicators currently provided by the LROI are considered important by the target group and which indicators should be added. Furthermore, it would be useful to provide feedback on the reasons for revisions, given that this has been shown able to direct quality improvement initiatives although we did not specifically ask whether orthopedic surgeons would be interested in this information (van Schie et al. 2020). 2 meta-analyses have shown that a single A&F strategy is one of the less effective interventions showing little to no improvement when examined (Shojania et al. 2006, Tricco et al. 2012). On the other hand, it seems obvious that accessible A&F that is interpreted correctly will ultimately improve the quality of care, as 17 out of 20 orthopedic surgeons indicated that they would conduct quality improvement initiatives as soon as they become aware of poorer performance. It seems likely that more active elements need to be added both to motivate orthopedic surgeons to log in and to ensure correct interpretation of the funnel plot, which is needed to be aware of outlier status regarding their performance. Trust in A&F data quality is often identified as a barrier to change clinical behavior. This is unlikely to play a major role in the current LROI feedback given the 98% completeness for primary procedures and 96% for revisions, which is similar for the data in the above-mentioned arthroplasty registries (van der Veer et al. 2010, de Vos Maartje et al. 2013, van Steenbergen et al. 2015, Catelas et al. 2018, Gude et al. 2018). Another barrier may be that physicians do not consider some indicators as an essential part of quality or deem benchmarks unrealistic (van der Veer et al. 2011, Eva et al. 2012, Gude et al. 2016, Gude et al. 2017b, Gude et al. 2018). In this study, for instance, it was found that one-third of both THA and TKA surgeons do not know their 1-year revision rate, which may suggest that some surgeons do not recognize the importance of

this outcome. This is striking because this outcome is already widely used by arthroplasty registries and considered an indicator to reflect the quality of care (Li et al. 1999, Itonaga et al. 2000, Tabak et al. 2002, Bonutti et al. 2017, Talmo et al. 2018). Moreover, A&F does not use absolute benchmarks, but performance indicators are compared with national surgeongroup averages, thereby making it likely that other similar surgeon groups are able to achieve that level of performance. Strengths and limitations A possible limitation of this study is response bias if awareness of performance differs between responders and nonresponders and the association with underlying factors were to be different. Given that survey responses were collected anonymously, we were unable to compare whether the characteristics of the non-respondents differed from the respondents to assess whether bias may have occurred. However, considering the overall response rate of 44%, and the fact that nonrespondents in general are not as involved as respondents and are thus more likely not to be aware of their performance, the associations are likely underestimated. A second limitation is that some self-reported outcomes (e.g., frequency of logging in or recall of revision rate) were analyzed. It is therefore possible that there were socially desirable answers to certain questions, e.g., knowledge about certain indicators. If this affected the results, even fewer orthopedic surgeons may be aware of their performance. However, because this was an anonymous survey, it seems more likely that respondents are surgeons dedicated to good performance and making feedback information more useful rather than giving socially desirable answers, so that reported rates are likely to reflect actual practice. An exception on the self-reported outcomes was the funnel-plot interpretation, where answers given by respondents were compared with the correct answer so that social desirability was not an issue. A third limitation may be the generalization of our results to other countries. Increasingly, information becomes publicly available on differences between hospitals in patient outcomes, as we have previously shown for revision rates in the Netherlands and Bozic et al. (2014) have shown for complication rates after total hip and knee arthroplasty in the United States (Bozic et al. 2014, van Schie et al. 2020). The magnitude of the between-hospital variation in risk-adjusted rates in these studies is surprisingly similar, with both studies showing about 3–4-fold differences between hospitals. Furthermore, although not looking at awareness in performance specifically, a previous international survey study showed only minor differences between orthopedic surgeons operating on different continents, taking into account their demographics (e.g., sex, age), surgical experience (e.g., number of years in practice, number of arthroplasties performed per year), use of additional diagnostics (e.g., plain radiographs, CT, MRI), and final treatment chosen (e.g., surgical versus non-surgical) (Li et al. 2014). So there is no evidence to suggest that there would be smaller differences between surgeons regarding their per-

Acta Orthopaedica 2021; 92 (1): 53–60

formance in other countries, and a difference in awareness has to our knowledge not been described before. Yet, such difference in awareness may be crucial in explaining why hospital differences in performance continue to exist, rather than that public reporting of hospital differences will by itself result in improvement. Implementation and further research As alluded to earlier, more active elements need to be added to improve A&F design to make it more attractive to log in, resulting in more awareness of one’s own performance. This could be encouraged by emphasizing the importance of already available indicators (e.g., revision rates) and adding new indicators to the A&F dashboard that are considered relevant and of interest as reported in this study (prosthesis survival, complications, readmissions, and length of hospital stay). As a result, more surgeons may be actually reached by the feedback, because the number of orthopedic surgeons who log in as well as the frequency of logging in will then increase. In addition, teaching material must be available on how to interpret funnel plots and be actively promoted by the orthopedic association during meetings, which will also increase awareness and possibly increase the reach of feedback, when more surgeons can interpret the performance indicators. Ultimately, an increased awareness of one’s own performance will likely lead to more quality improvement initiatives. The question arises as to whether voluntary quality control by providing only passive A&F on performance is sufficient in a modern orthopedic society. A&F could be more effective when offered in a more active and multifaceted way instead of as a single element (which in this study was only the LROI dashboard) (Ivers et al. 2012, Soong and Shojania 2020). A possible addition to the feedback would be that indicators are also verbally explained by an independent person, with clear targets discussed and action plans created, for instance based on a toolbox (Bradley et al. 2004, 2006, de Vos et al. 2009, van der Veer et al. 2010, Ivers et al. 2012, 2014, Brehaut et al. 2016, Gude et al. 2017a, Brown et al. 2019, Roos-Blom et al. 2019). In addition, setting up committees that will actively approach poorly performing hospitals to create action plans to improve quality of care may increase interest in one’s own performance as orthopedic surgeons want to avoid being under supervision. The Dutch Orthopedic Association initiated a quality committee in 2017 with the aim to detect negative outlier hospitals using LROI data and discuss activities to improve care (Commision Quality). This new procedure may stimulate logging in to check on performance and in this way increase awareness of own performance in the coming years. After all, orthopedic surgeons have no valid reason not to be interested in their own performance, given that they want the best care for their patients and continuously improving the quality of care is thus inherently linked to this. This survey is part of the “Improving Quality based on the Joint registries project” (IQ Joint study). Within this study,

Acta Orthopaedica 2021; 92 (1): 53–60

what will be tested includes whether more active intervention including monthly feedback on THA/TKA performance indicators, active education on how to use indicators for quality improvement, asking for improvement activities, and linking hospitals with better performing hospitals to exchange information and find areas for improvement will result in better outcomes, fewer complications, and more quality improvement initiatives compared with the LROI dashboard alone. During this randomized trial, A&F on surgeon-group level will be provided according to the preferences of the orthopedic surgeons as has been evaluated in this study. Conclusion Orthopedic surgeons performing THA/TKA have limited awareness of the performance of their surgeon group. Awareness could be increased by encouraging them to log in more often on their A&F dashboard, teaching them how to interpret funnel plots, and emphasizing the importance of performance indicators. Improvement of the effectiveness of feedback is important, because the majority of orthopedic surgeons indicated that quality improvement initiatives were introduced once they learned that their performance was worsening. To provide orthopedic surgeons with better feedback in the future, the feedback information should be extended with the indicators prosthesis survival and complications compared with peers at a national level, tailored to their specific surgeon group rather than making any selections themselves, with 6-month frequency. Supplementary data TheAppendix is available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.2020. 1827523 Concept and design: PvS, RN, PM. Collecting data: PvS, TZ, PM. Interpretation of data: PvS, LvB, RN, PM. Writing of the manuscript: PvS. Critical revision of the manuscript: LvB, TZ, RN, PM. Statistical analysis: PvS, PM. Supervision: RN, PM. Acta thanks Klaus-Peter Günther and Petri Virolainen for help with peer review of this study.

Assi C C, Barakat H B, Caton J H, Najjar E N, Samaha C T, Yammine K F. Mortality rate and mid-term outcomes of total hip arthroplasty using dual mobility cups for the treatment of femoral neck fractures in a Middle Eastern population. J Arthroplasty 2019; 34(2): 333-7. Bonutti P M, Khlopas A, Chughtai M, Cole C, Gwam C U, Harwin S F, Whited B, Omiyi D E, Drumm J E. Unusually high rate of early failure of tibial component in ATTUNE total knee arthroplasty system at implant– cement interface. J Knee Surg 2017; 30(5): 435-9. Bozic K J, Grosso L M, Lin Z, Parzynski C S, Suter L G, Krumholz H M, Lieberman J R, Berry D J, Bucholz R, Han L, Rapp M T, Bernheim S, Drye E E. Variation in hospital-level risk-standardized complication rates following elective primary total hip and knee arthroplasty. J Bone Joint Surg Am 2014; 96(8): 640-7.

Bradley E H, Holmboe E S, Mattera J A, Roumanis S A, Radford M J, Krumholz H M. Data feedback efforts in quality improvement: lessons learned from US hospitals. Qual Saf Health Care 2004; 13(1): 26-31. Bradley E H, Curry L A, Webster T R, Mattera J A, Roumanis S A, Radford M J, McNamara R L, Barton B A, Berg D N, Krumholz H M. Achieving rapid door-to-balloon times: how top hospitals improve complex clinical systems. Circulation 2006; 113(8): 1079-85. Brehaut J C, Eva K W. Building theories of knowledge translation interventions: use the entire menu of constructs. Implement Sci 2012; 7: 114. Brehaut J C, Colquhoun H L, Eva K W, Carroll K, Sales A, Michie S, Ivers N, Grimshaw J M. Practice feedback interventions: 15 suggestions for optimizing effectiveness. Ann Intern Med 2016; 164(6): 435-41. Brown B, Gude W T, Blakeman T, van der Veer S N, Ivers N, Francis J J, Lorencatto F, Presseau J, Peek N, Daker-White G. Clinical Performance Feedback Intervention Theory (CP-FIT): a new theory for designing, implementing, and evaluating feedback in health care based on a systematic review and meta-synthesis of qualitative research. Implement Sci 2019; 14(1): 40. Catelas I, Lehoux E A, Ning Z, Figeys D, Baskey S J, Beaulé P E. Differential proteomic analysis of synovial fluid from hip arthroplasty patients with a pseudotumor vs. periprosthetic osteolysis. J Orthop Res 2018; 36(7): 1849-59. Commision Quality. Commision Quality (Collaboration between Dutch Orthopaedic Association and Dutch Arthroplasty Registry). Protocol: Quality procedure, 2017. Available at www.orthopeden.org. Davis D A, Mazmanian P E, Fordis M, Van Harrison R, Thorpe K E, Perrier L. Accuracy of physician self-assessment compared with observed measures of competence: a systematic review. JAMA 2006; 296(9): 1094-102. de Vos M, Graafmans W, Kooistra M, Meijboom B, Van Der Voort P, Westert G. Using quality indicators to improve hospital care: a review of the literature. Int J Qual Health Care 2009; 21(2): 119-29. de Vos Maartje L G, van der Veer S N, Graafmans W C, de Keizer N F, Jager K J, Westert G P, van der Voort P H. Process evaluation of a tailored multifaceted feedback program to improve the quality of intensive care by using quality indicators. BMJ Qual Saf 2013; 22(3): 233-41. Eva K W, Armson H, Holmboe E, Lockyer J, Loney E, Mann K, Sargeant J. Factors influencing responsiveness to feedback: on the interplay between fear, confidence, and reasoning processes. Adv Health Sci Educ Theory Pract 2012; 17(1): 15-26. Fry D E, Pine M, Nedza S M, Locke D G, Reband A M, Pine G. Risk-adjusted hospital outcomes in Medicare total joint replacement surgical procedures. J Bone Joint Surg Am 2017; 99(1): 10-8. Gude W T, van der Veer S N, de Keizer N F, Coiera E, Peek N. Optimizing digital health informatics interventions through unobtrusive quantitative process evaluations. Stud Health Technol Inform 2016; 228: 594-8. Gude W T, Roos-Blom M J, van der Veer S N, de Jonge E, Peek N, Dongelmans D A, de Keizer N F. Electronic audit and feedback intervention with action implementation toolbox to improve pain management in intensive care: protocol for a laboratory experiment and cluster randomised trial. Implement Sci 2017a; 12(1): 68. Gude W T, van Engen-Verheul M M, van der Veer S N, de Keizer N F, Peek N. How does audit and feedback influence intentions of health professionals to improve practice? A laboratory experiment and field study in cardiac rehabilitation. BMJ Qual Saf 2017b; 26(4): 279-87. Gude W T, Roos-Blom M J, van der Veer S N, Dongelmans D A, de Jonge E, Francis J J, Peek N, de Keizer N F. Health professionals’ perceptions about their clinical performance and the influence of audit and feedback on their intentions to improve practice: a theory-based study in Dutch intensive care units. Implement Sci 2018; 13(1): 33. Itonaga I, Sabokbar A, Murray D W, Athanasou N A. Effect of osteoprotegerin and osteoprotegerin ligand on osteoclast formation by arthroplasty membrane derived macrophages. Ann Rheum Dis 2000; 59(1): 26-31. Ivers N, Jamtvedt G, Flottorp S, Young J M, Odgaard-Jensen J, French S D, O’Brien M A, Johansen M, Grimshaw J, Oxman A D. Audit and feedback: effects on professional practice and healthcare outcomes. Cochrane Database Syst Rev 2012(6): Cd000259.

Ivers N M, Sales A, Colquhoun H, Michie S, Foy R, Francis J J, Grimshaw J M. No more ‘business as usual’ with audit and feedback interventions: towards an agenda for a reinvigorated intervention. Implement Sci 2014; 9:14. Kurcz B, Lyons J, Sayeed Z, Anoushiravani A A, Iorio R. Osteolysis as it pertains to total hip arthroplasty. Orthop Clin North Am 2018; 49(4): 419-35. Li T F, Santavirta S, Virtanen I, Könönen M, Takagi M, Konttinen Y T. Increased expression of EMMPRIN in the tissue around loosened hip prostheses. Acta Orthop Scand 1999; 70(5): 446-51. Li C S, Karlsson J, Winemaker M, Sancheti P, Bhandari M. Orthopedic surgeons feel that there is a treatment gap in management of early OA: international survey. Knee Surg Sports Traumatol Arthrosc 2014; 22(2): 363-78. Menendez M E, Lu N, Huybrechts K F, Ring D, Barnes C L, Ladha K, Bateman B T. Variation in use of blood transfusion in primary total hip and knee arthroplasties. J Arthroplasty 2016; 31(12): 2757-63.e2. Pelt C E, Stagg M L, Van Dine C, Anderson M B, Peters C L, Gililland J M. Early outcomes after revision total hip arthroplasty with a modern modular femoral revision stem in 65 consecutive cases. Arthroplasty Today 2019; 5(1): 106-12. Porter M, Armstrong R, Howard P, Porteous M, Wilkinson J M. Orthopaedic registries—the UK view (National Joint Registry): impact on practice. EFORT Open Rev 2019; 4(6): 377-90. Roos-Blom M J, Gude W T, de Jonge E, Spijkstra J J, van der Veer S N, Peek N, Dongelmans D A, de Keizer N F. Impact of audit and feedback with action implementation toolbox on improving ICU pain management: cluster-randomised controlled trial. BMJ Qual Saf 2019; 28(12): 1007-15. Shojania K G, Ranji S R, McDonald K M, Grimshaw J M, Sundaram V, Rushakoff R J, Owens D K. Effects of quality improvement strategies for type 2 diabetes on glycemic control: a meta-regression analysis. JAMA 2006; 296(4): 427-40. Siciliani L, Sivey P, Street A. Differences in length of stay for hip replacement between public hospitals, specialised treatment centres and private providers: selection or efficiency? Health Econ 2013; 22(2): 234-42. Soong C, Shojania K G. Education as a low-value improvement intervention: often necessary but rarely sufficient. BMJ Qual Saf 2020; 29(5): 353-7. Tabak A Y, Bicimoglu A, Celebi L, Muratli H H, Yagmurlu M F, Aktekin C N. Early results of revision total hip replacements that used self locking revision stem (Wagner stem). [Turkish]. Artroplasti Artroskopik Cerrahi 2002; 13(2): 65-72. Talmo C T, Anderson M C, Jia E S, Robbins C E, Rand J D, McKeon B P. high rate of early revision after custom-made unicondylar knee arthroplasty. J Arthroplasty 2018; 33(7s): S100-s4.

Acta Orthopaedica 2021; 92 (1): 53–60

Toomey S D, Hopper R H Jr, McAuley J P, Engh C A. Modular component exchange for treatment of recurrent dislocation of a total hip replacement in selected patients. J Bone Joint Surg Am 2001; 83(10): 1529-33. Tricco A C, Ivers N M, Grimshaw JM, Moher D, Turner L, Galipeau J, Halperin I, Vachon B, Ramsay T, Manns B, Tonelli M, Shojania K. Effectiveness of quality improvement strategies on the management of diabetes: a systematic review and meta-analysis. Lancet 2012; 379(9833): 2252-61. Tuti T, Nzinga J, Njoroge M, Brown B, Peek N, English M, Paton C, van der Veer S N. A systematic review of electronic audit and feedback: intervention effectiveness and use of behaviour change theory. Implement Sci 2017; 12(1): 61. van der Veer S N, de Keizer N F, Ravelli A C, Tenkink S, Jager K J. Improving quality of care: a systematic review on how medical registries provide information feedback to health care providers. Int J Med Inform 2010; 79(5): 305-23. van der Veer S N, de Vos M L, Jager K J, van der Voort P H, Peek N, Westert G P, Graafmans W C, de Keizer N F. Evaluating the effectiveness of a tailored multifaceted performance feedback intervention to improve the quality of care: protocol for a cluster randomized trial in intensive care. Implement Sci 2011; 6: 119. van Schie P, van Steenbergen L N, van Bodegom-Vos L, Nelissen R, Marangvan de Mheen P J. Between-hospital variation in revision rates after total hip and knee arthroplasty in the Netherlands: directing quality-improvement initiatives. J Bone Joint Surg Am 2020; 102(4): 315-24. doi: 10.2106/ JBJS.19.00312. 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. Weeks W B, Kotzbauer G R, Weinstein J N. Using publicly available data to construct a transparent measure of health care value: a method and initial results. Milbank Q 2016; 94(2): 314-33. Weiss D, Dunn S I, Sprague A E, Fell D B, Grimshaw J M, Darling E, Graham I D, Harrold J, Smith G N, Peterson W E, Reszel J, Lanes A, Walker M C, Taljaard M. Effect of a population-level performance dashboard intervention on maternal-newborn outcomes: an interrupted time series study. BMJ Qual Saf 2018; 27(6): 425-36. Yoon C, Chang M J, Chang C B, Choi J H, Lee S A, Kang S B. Does unicompartmental knee arthroplasty have worse outcomes in spontaneous osteonecrosis of the knee than in medial compartment osteoarthritis? A systematic review and meta-analysis. Arch Orthop Trauma Surg 2019; 139(3): 393403.

Acta Orthopaedica 2021; 92 (1): 61–65

Are functional outcomes and early pain affected by discharge on the day of surgery following total hip and knee arthroplasty? Christian E HUSTED, Henrik HUSTED, Lina Holm INGELSRUD, Christian Skovgaard NIELSEN, Anders TROELSEN, and Kirill GROMOV

Department of Orthopedic Surgery, Copenhagen University Hospital, Hvidovre, Denmark Correspondence: christianhusted@live.dk Submitted 2020-04-25. Accepted 2020-09-14.

Background and purpose — Outpatient total knee and total hip arthroplasty (TKA and THA) has been shown to be feasible and safe in selected patients. However, little data is available on functional outcome and early pain in patients discharged on the day of surgery (DOS). We investigated patient-reported outcomes at 1 year and early pain in outpatient TKA and THA patients discharged on the day of surgery (DOS) (DDOS) compared with patients scheduled for outpatient surgery but not discharged on the DOS (nDDOS). Patients and methods — Prospective data on 261 consecutive patients scheduled for outpatient TKA (n = 126) and THA (n = 135) were collected. 37% of TKA patients and 33% of THA patients were discharged on the DOS. Pain scores at rest and activity and use of morphine were registered on postoperative days 1–7. Oxford Knee Score (OKS) and Oxford Hip Score (OHS) were collected preoperatively and at 3 and 12 months’ follow-up. Results — DDOS and nDDOS patients were similar in respect to age, sex, procedure type (TKA vs. THA), or preoperative OKS or OHS. Neither OKS nor OHS differed between groups at 3 and 12 months’ follow-up. Pain at rest and activity and use of morphine did not differ between the 2 groups on days 1–7. Interpretation — In patients scheduled for outpatient TKA and THA, we found similar patient-reported outcomes both early and at 1 year in those discharged on the DOS and those who had at least 1 overnight stay.

Fast-track total knee arthroplasty (TKA) and total hip arthroplasty (THA) have gained significant popularity over several years due to the many favorable aspects of optimizing perioperative care in these procedures. These aspects include reduced perioperative mortality and morbidity, reduced length of stay (LOS), and reduced financial expenses (Andreasen et al. 2017, Burn et al. 2018). Outpatient surgery has become increasingly popular in recent years in selected patients, and all the appealing aspects of standardized outpatient arthroplasty have been reported to come with no increased risk to patients’ safety (Pollock et al. 2016, Goyal et al. 2017, Vehmeijer et al. 2018, Gromov et al. 2019, Xu et al. 2019) and at even lower cost (Husted et al. 2018). One of the potential concerns regarding outpatient THA and TKA is the rehabilitation that patients receive during hospital stay which can be as short as a few hours, which potentially can affect postoperative pain and functional outcome. While safety aspects following outpatient THA and TKA in selected patients have been thoroughly investigated, little information exists regarding the implications that discharge on the DOS has on postoperative pain and functional outcome. Therefore, we investigated the degree of pain in the first postoperative week, as well as pain and functional outcomes at 1 year following outpatient TKA and THA among patients discharged on the DOS (DDOS) and compared them with TKA and THA patients scheduled for outpatient surgery but not discharged on the DOS (nDDOS).

Patients and methods Patients undergoing primary unilateral THA and TKA between January 2016 and June 2017 at our high-volume center and scheduled for same-day discharge were included in this study. Patient selection and eligibility for a part of this cohort was previously described (Gromov et al. 2017). In brief, patients © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group, on behalf of the Nordic Orthopedic Federation. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. DOI 10.1080/17453674.2020.1836322

were considered eligible for inclusion in this study if they did not suffer from sleep apnea and had ASA scores of < 3. Furthermore, only patients who were operated on as 1st or 2nd in the surgical theater were included. All operations were performed in a standardized fast-track setup (Husted 2012) by surgeons specialized in THA and TKA surgery. The standard surgical protocol for both THA and TKA included intended spinal anesthesia, standardized fluid management, preoperative single-shot high-dose methyl prednisolone (Lunn et al. 2011, 2013), use of preoperative tranexamic acid (TXA)—THA patients received 2 doses IV, and TKA patients received an additional intra-articular dose— and absence of drains. All THAs were performed using a standard posterolateral approach. All TKAs were performed with a standard medial parapatellar approach without the use of a tourniquet and using local infiltration analgesia (LIA) (Andersen and Kehlet 2014). The patients were transferred from the postoperative recovery unit to the patient ward after a few hours, where mobilization was attempted as soon as possible, allowing full weight-bearing. Rivaroxaban was used as oral thromboprophylaxis, starting 6–8 hours postoperatively and continuing daily until discharge. No extended thromboprophylaxis was used. Pain medication consisting of celecoxib 200 mg/12 hours and paracetamol 1 g/6 hours were given until postoperative day 7 (POD 7). No opioids were systematically given to any patients and oral morphine 10 mg pro necessitate was used as rescue analgesic only. Physiotherapy was started on the DOS and continued until discharge. All patients were referred to public outpatient physiotherapy, which they attended as long as seemed fit by the treating therapist. Patients were discharged if discharge criteria were fulfilled before 8 p.m. on the DOS. These criteria included patients not exceeding 500 mL of intraoperative blood loss and pain scores < 3 (VAS) while resting and < 5 during weight-bearing mobilization. Furthermore, urination had to be spontaneous and mobilization had to be achieved with a physiotherapist. Also, an adult had to be present with the patient for the first 24 hours after discharge. Patients not fulfilling discharge criteria on the DOS stayed overnight until discharge criteria were fulfilled. Patient-reported outcomes were measured using Oxford Knee Score (OKS) and Oxford Hip Score (OHS)—a 12-item questionnaire regarding knee/hip pain and function that are summed to a total score of 0–48 (worst–best) (Murray et al. 2007). OKS and OHS were registered preoperatively as well as 3 months and 1 year after THA or TKA. Following surgery, the patients were asked to fill out a questionnaire regarding daily pain at rest as well as during activity. This was measured on the VAS 0–10, 10 being worst imaginable pain, and daily use of morphine (yes/no) was recorded on postoperative days 1–7. The questionnaire was given to the patients upon discharge and collected at 3 months’ follow-up.

Acta Orthopaedica 2021; 92 (1): 61–65

Later, overall satisfaction with treatment was recorded using a numeric rating scale (NRS) 0–10, 10 being very satisfied, and willingness to undergo the same treatment again (yes/no/I don’t know) was answered at 3 months’ follow-up. Statistics Normality assumption for all continuous variables was evaluated by Q–Q plots as well as skewness and kurtosis measurements. Mean values (SD) are presented for normally distributed variables, while median values and interquartile ranges (IQR) are presented for non-normally distributed variables. The Mann–Whitney U-test was used to compare continuous non-parametric variables, Student’s t-test to compare continuous parametric variables, and a chi-square test to compare categorical variables. All data were processed in R 3.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Ethics, funding, and potential conflicts of interest No approval from the National Ethics Committee was necessary as this was a non-interventional observational study. The study was approved by the Danish Data Protection Agency (entry no. 20047-58-0015). This work was sponsored by grants from the Lundbeck Foundation and ZimmerBiomet, which had no influence on any part of the study or on the content of the paper. The authors declare no conflicts of interest.

Results 275 patients were scheduled to undergo outpatient TKA/THA between December 2015 and June 2017. 261 of those patients had complete data and were included in the analysis: TKA (n = 126) and THA (n = 135). 45 (33%) of THA patients were discharged on the DOS, while 47 (37%) of TKA patients were discharged on the DOS. All patients who were not discharged on the DOS were discharged the following day. Patients in the 2 groups were similar in respect of age, sex, or preoperative OKS/OHS (Tables 1–3). OKS at 3 months/1year follow-up was 32/39 and 31/38 for DDOS and nDDOS patients, respectively (p = 0.6/p = 0.5). OHS at 3 months/1year follow-up was 39/43 and 37/43 for DDOS and nDDOS patients, respectively (p = 0.1/p = 0.9) (Tables 2 and 3). Furthermore, the 2 groups of patients had similar increases in OHS/OKS at both 3 and 12 months’ follow-up (Tables 2 and 3). Among THA patients, no statistically significant difference was found regarding pain scores between DDOS and nDDOS patients on POD 1–7 (Figure 1). Among TKA patients, no statistically significant difference was found regarding pain scores with 1 exception: on POD 2 DDOS patients reported a mean VAS score of 5.2 during rest compared with 3.6 among nDDOS patients (p = 0.002) (Figure 2). DDOS TKA patients

Acta Orthopaedica 2021; 92 (1): 61–65

Table 1. Demographics and satisfaction. Values are count (%) unless otherwise specified Factor

DDOS

nDDOS

Participants, n 92 169 Mean age (SD) 60 (11) 62 (11) Male 53 (58) 80 (47) Female 39 (42) 89 (53) THA 45 (33) 90 (67) TKA 47 (37) 79 (63) Would undergo the same treatment again yes 78 (93) 129 (87) no 4 (5) 4 (3) unsure 2 (2) 15 (10) Satisfaction with treatment (0–10), median (range) 9 (7–10) 9 (6–10)

p-value 0.2 0.2

VAS pain during activity — THA

VAS pain during rest — THA

10 DDOS nDDOS

DDOS nDDOS

●

● ●

0.6 0.05

0.4

DDOS = discharged on the day of surgery nDDOS = not discharged on the day of surgery

● ● ●

Postoperative day

Figure 1. THA VAS scores (mean and 95% CI) for pain during activity (left panel) and during rest (right panel). VAS pain during activity — TKA

VAS pain during rest — TKA

10 DDOS nDDOS

DDOS nDDOS

Table 2. Mean Oxford Knee Scores (OKS) in TKA patients 6

OKS

DDOS nDDOS Difference n = 47 n = 79 mean (95% CI) p-value

6 ● ● ●

4 ●

●

Before At 3 months At 1 year Difference 3 months—before 1 year—before

22 32 39 10 16

22 31 38 9 16

0.2 (–2.0 to 2.4) 0.8 (–2.3 to 4.1) 1.0 (–1.9 to 3.9) 0.7 (–2.6 to 4.0) 0.8 (–2.2 to 3.9)

0.9 0.6 0.5 0.7 0.6

Postoperative day

Figure 2. TKA VAS scores (mean and 95% CI) for pain during activity (left panel) and during rest (right panel).

For abbreviations, see Table 1

Use of morfin, % — THA

Use of morfin, % — TKA

1.0

DDOS nDDOS

Table 3. Mean Oxford Hip Scores (OHS) in THA patients

●

0.8

●

OHS

DDOS nDDOS Difference n = 45 n = 90 mean (95% CI) p-value

0.6

●

Before At 3 months At 1 year Difference 3 months—before 1 year—before

24 39 43

23 37 43

0.6 (–1.7 to 2.9) 2.2 (–0.5 to 4.8) 0.1 (–2.3 to 2.4)

0.6 0.1 1.0

16 19

14 20

1.6 (–1.7 to 4.7) 0.5 (–4.0 to 2.9)

0.4 0.7

For abbreviations, see Table 1

had a mean VAS score of 4.6 on POD 1 during rest compared with 4.0 among nDDOS TKA patients (p = 0.3). Respectively, these scores dropped to 3.6 and 3.3 on POD 7 during rest (p = 0.7). During activity, DDOS TKA patients scored 5.4 on POD 1 whereas nDDOS TKA patients scored 5.6 (p = 0.7). On POD 7, these scores dropped to 4.6 and 4.7 (p = 0.9), respectively. Regarding use of morphine, a similar pattern was found in both DDOS and nDDOS patients, as there was no statistically significant difference in THA or TKA patients (Figure 3). Patients from both groups also reported similar satisfaction, as DDOS patients on average rated their experience 9.1/10

0.4

●

● ● ●

●

0.2

Postoperative day

Figure 3. Use of morphine (percentages with 95% CI) among TKA patients (left panel) and THA patient (right panel) .

compared with nDDOS patients who scored 9.0/10 (p = 0.4) (Table 1). 93% of DDOS patients claimed that they would would go through the process again compared with 87% of nDDOS patients (p = 0.05) (Table 1).

Discussion In this prospective cohort study, we found very similar patientreported outcomes when comparing a group of patients who were discharged on the DOS after THA/TKA with a group

of patients who were not discharged on the DOS. This was the case with regards to postoperative OHS and OKS, as well as pain, and use of morphine during the first week after surgery. Furthermore, similar levels of satisfaction were found between the two groups. Patient-reported outcomes measured using OKS and OHS were similar between the 2 groups prior to surgery and at 3and 12-months’ follow-up. Change in OKS and OHS at 3- and 12-months’ follow-up were also similar between groups suggesting that the outcome of THA/TKA is not affected negatively by discharge on the DOS. DDOS patients included in this study were able to undergo the same postoperative rehabilitation after discharge as nDDOS patients despite shorter hospital stays and therefore fewer instructions during hospital stay. The mean preoperative OHS in our study was 24 among DDOS patients and 23 among nDDOS patients, which is similar to previous studies investigating early outcome following total joint arthroplasty (Klapwijk et al. 2017, Porsius et al. 2018). Klapwijk et al. (2017) also measured OHS at 6 weeks after THA surgery and found a median OHS of 43 among a mixed group of inpatients and outpatients. The mean preoperative OKS of 22 (DDOS) and 22 (nDDOS) is also similar to a previous study by Hoeffel et al. (2019), who found OKS at 3 months and 1 year after surgery to be mean 35 and 39, respectively. The patients in our study had similar increases in OHS and OKS to the patients in the aforementioned studies, and overall our patients reached expected and satisfying scores. Pain scores in the first 7 days after surgery, where pain may be most pronounced, did not—with 1 exception on POD 2—differ statistically between groups. However, there seemed to be a trend towards higher pain levels among DDOS TKA patients compared with nDDOS patients on POD 2–4 both at rest and during activity. A possible explanation for this could be that patients discharged on DOS are more active both before and after discharge and therefore have higher pain levels compared with patients who spent 1 night at the hospital. Another potential explanation could be that compliance regarding use of pain medication other than morphine is lower among DDOS TKA patients as they spend a shorter time in hospital and therefore receive more information during a shorter period of time. Generally lower pain scores following THA may explain why no difference is seen between DDOS and nDDOS THA patients. The literature on pain in the first days after surgery is inconsistent as some find differences between inpatients and outpatients with regards to pain (Goyal et al. 2017) whereas others do not (Gauthier-Kwan et al. 2018). Goyal et al. (2017) found that THA outpatients had higher pain levels than inpatients on POD 1 and proposed that insufficient pain management may be the cause of this increased pain among outpatients, which is in line with our findings and proposed explanations—although the difference was among TKA patients and not THA patients in our study. Few studies have compared pain scores between outpatients and inpatients following THA and TKA during the first postoperative days, but other studies have done so with

Acta Orthopaedica 2021; 92 (1): 61–65

regards to the first postoperative weeks and months. These studies have found similar pain levels between inpatients and outpatients after THA and TKA (Schotanus et al. 2017, Füssenich et al. 2020). Schotanus et al. (2017) found that 6 weeks after TKA both inpatients and outpatients had mean NRS pain scores of 2.6. As expected, those scores are lower than pain scores measured in our study on POD 7 as pain is still expected to decrease after the first week. Also, the use of rescue medication in the form of opioids was similar between the 2 groups. This goes to show that discharge on the DOS does not lead to increased opioid use among selected patients. Gauthier-Kwan et al. (2018) found similar results, as outpatients and inpatients had statistically similar use of opioids during the first postoperative days. The literature on this subject is limited and further studies are necessary to determine whether or not there is a difference between the 2 groups of patients. Patients having THA or TKA are generally satisfied with their overall experiences (Husted et al. 2008, Hoeffel et al. 2019). That was also the case in this study as the 2 groups of patients on average scored 9.0/10 and 9.1/10, respectively. These findings are in accordance with an earlier study (Kelly et al. 2018). Furthermore, the willingness to do the same procedure under the same circumstances again was similar in both groups. A nearly statistically significant difference in willingness to undergo the same procedure was found between the 2 groups of patients, with slightly more DDOS patients responding that they were willing to undergo the same procedure again. Porsius et al. (2018) found similar levels of willingness to undergo fast-track THA among three different groups of patients: fast-, average-, and slow-recovery patients. A large proportion of the fast-recovery patients had outpatient surgery compared with the proportion of slowrecovery patients, which goes to show that outpatients are not less satisfied than inpatients after THA. Though this study sheds light on some aspects of patientreported outcomes and pain after THA and TKA, there are also limitations associated with this study. 1 of the limitations of our study lies in possible differences between the 2 groups of patients as this was a prospective cohort study without randomization. This allows for potential confounding between the 2 groups. Even though recorded patient demographics and preoperative PROMS were similar between groups, other nonaccounted for factors may be responsible for some patients being discharged on the DOS, while others were not. As this study is observational, there is a potential for both selection bias and residual confounding. However, we chose not to adjust the analysis for potential bias but present the PROM data for the 2 cohorts as is, to present the actual clinical reality for such patients scheduled for outpatient surgery. Patients’ mentality may play a role with some patients having a more positive attitude toward DOS discharge. Such patients might be more likely to report high levels of satisfaction, provided they were discharged on the DOS; this, however, is speculative. Further-

Acta Orthopaedica 2021; 92 (1): 61–65

more, this study would be strengthened by additional data, as the reasons for inpatients’ overnight stays were not registered. The differences in the use of physiotherapy between the 2 groups of patients following discharge may be a confounding variable in this study, though it seems unlikely. Tracking the use of physiotherapy is not possible as it takes place outside of the hospital after discharge. Additionally, a potential bias could exist in the form of a difference in pain between the two groups of patients. One could argue that increased pain after surgery could lead to a higher probability of overnight stay in hospital but this study did not aim to investigate reasons for overnight stay in hospital. Finally, external validity is a limitation, as patients scheduled for outpatient surgery in this study may differ compared with other setups, making results more difficult to apply to a different patient population. In conclusion, in this prospective cohort study, we found similar pain scores, use of rescue medication (morphine), and OHS/OKS up to 1 year reported by THA and TKA patients discharged on the day of surgery and patients who stayed at least 1 night in the hospital. These findings support continuous utilization of outpatient TKA and THA in selected patients. CEH, HH, and KG planned the study. HH, LHI, CSN, AT, and KG were responsible for the logistical setup and collected the data. CH and KG analyzed the data. CH wrote the first draft of the paper; all authors revised the paper. Acta thanks Michael Clarius and Stephan Vehmeijer for help with peer review of this study.

Andersen LØ, Kehlet H. Analgesic efficacy of local infiltration analgesia in hip and knee arthroplasty: a systematic review. Br J Anaesth 2014; 113(3): 360-74. Andreasen S E, Holm H B, Jørgensen M, Gromov K, Kjærsgaard-Andersen P, Husted H. Time-driven activity-based cost of fast-track total hip and knee arthroplasty. J Arthroplasty 2017; 32(6): 1747-55. Burn E, Edwards C J, Murray D W, Silman A, Cooper C, Arden N K, PinedoVillanueva R, Prieto-Alhambra D. Trends and determinants of length of stay and hospital reimbursement following knee and hip replacement: evidence from linked primary care and NHS hospital records from 1997 to 2014. BMJ Open 2018; 8(1): e019146. Füssenich W, Gerhardt D M, Pauly T, Lorenz F, Olieslagers M, Braun C, van Susante J L. A comparative health care inventory for primary hip arthroplasty between Germany versus the Netherlands: is there a downside effect to fast-track surgery with regard to patient satisfaction and functional outcome? Hip Int 2020; 30(4): 423-30. Gauthier-Kwan O Y, Dobransky J S, Dervin G F. Quality of recovery, postdischarge hospital utilization, and 2-year functional outcomes after an outpatient total knee arthroplasty program. J Arthroplasty 2018; 33(7): 2159-64.e1.

Goyal N, Chen A F, Padgett S E, Tan T L, Kheir M M, Hopper R H Jr, Hamilton W G, Hozack W J. Otto Aufranc Award: A multicenter, randomized study of outpatient versus inpatient total hip arthroplasty. Clin Orthop Relat Res 2017; 475(2): 364-72. Gromov K, Kjærsgaard-Andersen P, Revald P, Kehlet H, Husted H. Feasibility of outpatient total hip and knee arthroplasty in unselected patients. Acta Orthop 2017; 88(5): 516-521. Gromov K, Jørgensen C C, Petersen P B, Kjaersgaard-Andersen P, Revald P, Troelsen A, Kehlet H, Husted H. Complications and readmissions following outpatient total hip and knee arthroplasty: a prospective 2-center study with matched controls. Acta Orthop 2019; 90(3): 281-5. Hoeffel D P, Daly P J, Kelly B J, Giveans M R. Outcomes of the first 1,000 total hip and total knee arthroplasties at a same-day surgery center using a rapid-recovery protocol. J Am Acad Orthop Surg Glob Res Rev 2019; 3(3): e022. Husted H. Fast-track hip and knee arthroplasty: clinical and organizational aspects. Acta Orthop Suppl 2012; 83(346): 1-39. Husted H, Holm G, Jacobsen S. Predictors of length of stay and patient satisfaction after hip and knee replacement surgery: fast-track experience in 712 patients. Acta Orthop 2008; 79(2): 168-73. Husted H, Kristensen B B, Andreasen S E, Skovgaard Nielsen C, Troelsen A, Gromov K. Time-driven activity-based cost of outpatient total hip and knee arthroplasty in different set-ups. Acta Orthop 2018; 89(5): 515-21. Kelly M P, Calkins T E, Culvern C, Kogan M, Della Valle C J. Inpatient versus outpatient hip and knee arthroplasty: which has higher patient satisfaction? J Arthroplasty 2018; 33(11): 3402-6. Klapwijk L C, Mathijssen N M, Van Egmond J C, Verbeek B M, Vehmeijer S B. The first 6 weeks of recovery after primary total hip arthroplasty with fast track. Acta Orthop 2017; 88(2): 140-144. Lunn T H, Husted H, Solgaard S, Kristensen B B, Otte K S, Kjersgaard A G, Gaarn-Larsen L, Kehlet H. Intraoperative local infiltration analgesia for early analgesia after total hip arthroplasty: a randomized, double-blind, placebo-controlled trial. Reg Anesth Pain Med 2011; 36(5): 424-9. Lunn T H, Andersen L Ø, Kristensen B B, Husted H, Gaarn-Larsen L, Bandholm T, Ladelund S, Kehlet H. Effect of high-dose preoperative methylprednisolone on recovery after total hip arthroplasty: a randomized, double-blind, placebo-controlled trial. Br J Anaesth 2013; 110(1): 66-73. 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. Pollock M, Somerville L, Firth A, Lanting B. Outpatient total hip arthroplasty, total knee arthroplasty, and unicompartmental knee arthroplasty: a systematic review of the literature. JBJS Rev 2016 27; 4(12). pii: 01874474201612000-00004. Porsius J T, Mathijssen N M C, Klapwijk-Van Heijningen L C M, Van Egmond J C, Melles M, Vehmeijer S B W. Early recovery trajectories after fast-track primary total hip arthroplasty: the role of patient characteristics. Acta Orthop 2018; 89(6): 597-602. Schotanus M G M, Bemelmans Y F L, Grimm B, Heyligers I C, Kort N P. Physical activity after outpatient surgery and enhanced recovery for total knee arthroplasty. Knee Surg Sports Traumatol Arthrosc 2017; 25(11): 3366-71. Vehmeijer S B W, Husted H, Kehlet H. Outpatient total hip and knee arthroplasty. Acta Orthop 2018; 89(2): 141-4. Xu J, Cao J Y, Chaggar G S, Negus J J. Comparison of outpatient versus inpatient total hip and knee arthroplasty: a systematic review and meta-analysis of complications. J Orthop 2019; 17: 38-43.

Acta Orthopaedica 2021; 92 (1): 66–72

Unexpected varus deformity and concomitant metal ion release and MRI findings of modular-neck hip stems: descriptive RSA study in 75 hips with 8 years’ follow-up Sverrir KIERNAN 1, Bart KAPTEIN 2, Carl FLIVIK 1, Martin SUNDBERG 1, and Gunnar FLIVIK 1 1 Department

of Orthopedics, Skåne University Hospital, Clinical Sciences, Lund University, Lund, Sweden; 2 Department of Orthopedics, Leiden University Medical Centre, Leiden, The Netherlands Correspondence: Sverrir.Kiernan@med.lu.se Submitted 2020-07-02. Accepted 2020-10-07.

Background and purpose — Modular-neck hip stems have been identified with corrosion-related problems from the neck–stem junction. We report an ongoing varus deformity of modular-neck hip stems with simultaneous metal ion release observed during a study comparing the migration of modular vs. standard hip stems. Patients and methods — We followed 50 patients with modular and 25 with standard neck stems using radiostereometry (RSA). At 5-year follow-up, we noted a compromised integrity of the modular stem with varus deformity in the neck–stem interface. Changes in head–tip distance as well as whole-blood ion concentration and MRI findings were analyzed. The modular stems were followed further up to 8 years. Results — The head–tip distance decreased continuously by 0.15 mm per year resulting in 1.2 (95% CI 1.0–1.4) mm at 8 years for modular stems, while for the standard stems at 5 years, the decrease was 0.09 (CI 0.0–0.2) mm or 0.02 mm/year. For the modular stems, the reduction in head–tip distance correlated to the increase in whole-blood cobalt concentration at 8 years but not to the MRI grading of tissue reactions. At 5 years, cobalt levels were 4.9 µg/L for modular stems and at 8 years 4.8 µg/L, whereas for standard stems this was 1.0 µg/L. After 8 years, 9 of 72 stems had been revised for different reasons, but only 1 with obvious adverse local tissue reaction (ALTR). Interpretation — We present a surprisingly large progressive deformation at the modular neck–stem junction, but so far without a definite clinical problem. Even the femoral head seems to show slight compression onto the taper over time. A high rate of revisions for the modular type of this stem has raised general concerns, and it has been recalled from the market.

Modular hip stems with different versions, angles, and lengths of neck can adapt to different femoral geometries. These modifications are valuable theoretically for improving range of motion and soft tissue balance (Barrack 1994, Jones 2004, Archibeck et al. 2011, Srinivasan et al. 2012). Mechanically assisted crevice corrosion (MACC) became clinically relevant with the emergence of large metal-on-metal (MoM) implants early in the 21st century (McGrory and McKenny 2016). Recently, corrosion has also been reported for metal-on-polyethylene (MoP) in a variety of stem designs caused by fretting in the head–neck junctions (Gilbert et al. 1993, Morlock 2015, Patel et al. 2016, Morlock et al. 2018). The increased number of interfaces introduced by modular systems has the potential to increase the risk for adverse local tissue reaction (ALTR) caused by the release of metal ions and inflammatory mediators (Molloy et al. 2014, McGrory and McKenney 2016). Taper corrosion at the modular junctions of THA femoral stems are known to cause ALTR (Lindgren et al. 2011, Gill et al. 2012). Revisions as a consequence of ALTR associated with neck–stem taper corrosion have now been reported (Nahhas et al. 2019, Shah et al. 2019) and should be considered as a potential cause for progressive, disabling groin pain (Cooper et al. 2013). The cause of corrosion in the neck–stem junction has been debated. Some say that the shape of the neck–stem tapers may deviate from ideal dimensions causing relative motions between the neck and stem (Frisch et al. 2018). Others state that corrosion occurs regardless of design and that the primary cause is mixed-metal couples with unequal modulus of elasticity (Young’s modulus), allowing for increased metal transfer and surface damage (galvanic mode of corrosion) (Su et al. 2017). In 2010, we started a randomized controlled trial (Kiernan et al. 2020) to evaluate the potential superiority of modular

Acta Orthopaedica 2021; 92 (1): 66–72

Table 1. Demographic and treatment data of study population

Standard

Modular

Modular Standard Factor (n = 47) (n = 25) Male/female sex 30/17 16/9 Age (range) 58 (34–80) 60 (46–74) BMI (SD) 28 (3.8) 28 (4.0) Weight, kg (SD) 84 (13) 88 (17) Components used Size, mean (range) 5 (1–7) 6 (4–8) Short/long neck 23/24 Retroverted/standard/anteverted neck 16/19/12 CCD angle (125°/130°/135°) 37/6/4 Head length, (–5/standard/+5) 10/30/7 12/11/2 Total neck length, mm median (range) 58 (47–69) 58 (53–65)

Figure 1. Standard and modular stem designs used in this study.

stems in restoring hip symmetry and examine postoperative migration rates of ABG II modular vs. standard stems (ABG II system, Stryker, Exeter, UK) with RSA (Figure 1). During the follow-up and data-processing, we achieved some unexpected results regarding the modular design. At the 5-year follow-up, we noted that the complex consisting of the stem–neck–head as a rigid body used for RSA was no longer a fixed segment, and the prosthetic head seemed to have migrated with respect to the body of the stem. To investigate this phenomenon, we decided to measure the movement of the head in relation to the tip point of the stem. To find out if it might be due to corrosion and fretting in the neck–stem junction of the modular stem, we correlated the rate of deformation with whole–blood ion levels of cobalt, chromium, and titanium. We measured ALTR formation based on Metal Artefact Reduction Series MRI (MARS-MRI) and analyzed whether stem size, neck length, head length, caput-collum diaphyseal (CCD) angle, version, total neck length, and body weight influenced this deformation of the stem complex.

Patients and methods Of the 75 stems that we had in our cohort for the former ABG II study (Kiernan et al. 2020), we were able to include 47 modular stems with sufficient RSA data during the followup period and 25 non-modular standard stems (up to 5 years’ follow-up). We lost 1 patient to follow-up due to an early periprosthetic femoral fracture (PPFF); we excluded 2 because of problems in performing the postoperative RSA examination (Table 1). The 47 modular hips were studied in further detail to reveal the influence of stem size, neck length (short/long), neck angle (125°/130°/135°), head length (–5 mm/standard/+ 5mm), neck version (anteverted, standard, retroverted), and patient body weight. We also looked at the total neck length (combined neck and head length) and correlated it to the HTD reduction over time.

The ABG II prosthesis has a titanium alloy (TMZF) stem. The modular neck is a cobalt-chromium (CoCr) alloy. We used a 36 mm CoCr LFIT femoral head for all patients except 2 modular stems that had 32 mm head due to small-sized cups. We used uncemented cups with a highly crosslinked poly ethylene liner. Radiostereometric analysis We used a uniplanar technique with the patient supine (Valstar et al. 2005) with 2 fixed X-ray sources. We used a type41 calibration cage (Tilly Medical, Lund, Sweden) and the model-based RSA software (Version 4.0, RSAcore, Leiden, The Netherlands). The reference examination was performed on the 1st postoperative day before mobilization and served as the reference point for all further examinations. Follow-up examinations were carried out at 14 days, 3 months, and then at 1, 2, and 5 years and additionally for the modular stems at 8 years. We performed model-based RSA analysis using the EGS hip analysis method that includes accurate estimation of the positions of the head and distal tip of the hip stem (Kaptein et al. 2006). Our primary outcome was the change in the distance between the center of rotation of the prosthetic head and the tip of the stem measured by successive RSA using the postoperative examination as reference (Figure 2). The precision of the RSA measurements for the head–tip distance was 0.15 mm and based on 68 patients’ double examinations. The precision represents the smallest migration value that is considered significant and is based on 2 standard deviations of the error obtained representing the 95% confidence interval. Metal ion measurements We measured the levels of cobalt, chromium, and titanium at 5 years for both stem designs for comparison and additionally at 8-year follow-up for correlation with the rate of stem deformation in patients operated on with the modular stem. Measurements obtained metal ion concentrations in whole blood

Acta Orthopaedica 2021; 92 (1): 66–72

to the point of its intersection with the longitudinal axis of the stem on calibrated templates (Figure 3). Clinical evaluation For clinical evaluation, preoperatively and at 1, 2, and 5-year follow-up, we used the hip-specific Hip Osteoarthritis Outcome Score (HOOS), Swedish version LK 2.0, as well as VAS pain and VAS satisfaction.

Figure 2. Measurement of the head–tip distance.

Figure 3. Examples of measurements of total neck length. The figure shows all possible measuring points along the 3 different CCD angles. The point of intersection is defined as the intersection between the CCD 130° line and the longitudinal axis of the stem.

(SGAB Analytica, Luleå University of Technology, S-971 87 Luleå, Sweden) (Rodushkin et al. 2000). ALTR assessment on MARS-MRI We evaluated all patients who agreed to an MRI at 5 years for the occurrence of ALTR and graded the MARS-MRI findings. 3 patients with modular stems and 1 with standard refused the examination. We used the 1.5 Tesla MRI system (MAGNETOM Avanto, Siemens AG, Healthcare Sector, Erlangen, Germany) using spine matrix and body matrix coils, running a protocol consisting of coronary T1 view angle tilting (VAT) + STIR VAT, sagittal T2 VAT and axial T1 VAT. Intravenous contrast (19 mL Dotarem) was administered, and then an axial T1 VAT together with an axial subtraction image was conducted. This resulted in 6 image sequences. An experienced musculoskeletal radiologist analyzed all MRIs and graded the findings using a modified version of the Hauptfleisch grading system (Hauptfleisch et al. 2012). This system consists of type I: cystic ALTR with a wall thickness < 3 mm; Type II: Cystic ALTR with a wall thickness > 3 mm; Type III: solid ALTR. We added type 0 (no ALTR) and divided type II into subgroups IIa (cystic ALTR without solid parts) and b (with a solid part, but comprising less than 50% of the total ALTR area). Measurement of total neck length The modular design had 3 parts, i.e., a stem, neck, and head. With the standard design, the offset increases with size, while the modular design allows an improved range of motion and soft tissue balance by various choices in neck length, version, and CCD angle. We measured the total neck length of all implanted modular stem combinations for estimation of the effect of the combined lever arm gained by different component choices. We did this by measuring the length of a line from the head center

Statistics We used estimates from a general linear mixed model for the analysis of head–tip distance reduction, where the subject effect was taken into consideration, and estimates from a linear regression model for the analysis of whole-blood ion levels with regard to the rate of stem deformation. All models were specified as the main effect of each predictor and time, and the interaction between them. Both random slopes and intercepts were used and the covariance matrix was specified as an AR-1 correlation. Metal ion levels were modeled with linear regression, where each metal was modeled against movement, respectively. This was done separately for the 5and 8-year follow-up. All linear mixed models were estimated using the R package nlme (R Foundation for Statistical Computing, Vienna, Austria). The Mann–Whitney U-test was used to compare the distribution for the different grades of ALTR as well as to test for differences between groups for VAS and HOOS. All calculations were conducted in R v.3.5.2 (R Foundation for Statistical Computing, Vienna, Austria). Ethics, registration, data sharing, funding, and potential conflicts of interest The Regional Ethical Review Board at Lund University approved the original study (Dr 2009/6), and it was carried out in compliance with the Helsinki Declaration and registered in ClinicalTrials.gov Identifier: NCT01512550. Data is available on reasonable request. The Southern Healthcare Region in Sweden (Södra Sjukvårdsregionen) provided a doctoral grant for labor costs. Stryker gave financial support for part of the RSA and MRI examinations but did not influence how we conducted or interpreted the study. The authors declare no conflict of interest related to this study.

Results Revisions 8 modular stems had been revised at the 8-year follow-up, 1 due to hip pain in combination with raised metal ion levels and MRI signs of ALTR. 1 revision was due to loosening of the stem, and 2 revisions due to loosening of the cup where we also decided to revise the well-fixed stems. 3 stems were revised because of periprosthetic femoral fractures (PPFF) with adequate trauma and 1 because of late periprosthetic infec-

Acta Orthopaedica 2021; 92 (1): 66–72

Table 2. Time to and cause for revisions

Factor

Years to Cobalt Chromium Titanium ALTR revision (µg/L) (µg/L) (µg/L) grade a

Modular: Infection 1.9 n/a n/a n/a n/a Loose cup 3.6 3.3 1.0 1.7 n/a Loose cup 4.3 3.8 1.7 0.5 n/a PPFF b 5.8 3.0 0.9 2.8 4 PPFF 6.4 5.4 1.9 0.5 0 ALTR c 6.4 8.2 2.4 1.3 3 PPFF 7.1 9.3 1.8 0.5 1 Loose stem 7.8 4.8 1.1 0.5 1 Standard: Infection 1.4 n/a n/a n/a n/a ALTR: Adverse local tissue reaction . PPFF: Periprosthetic femoral fracture. a At 5 years’ follow-up. b Accompanying groin pain prior to PPFF. c ALTR type 3. Skin reaction with proved hypersensitivity to cobalt. Accompanying groin pain.

Figure 4. ABG II modular head (LFit) and neck after revision with corrosion on the neck part engaged in the stem–neck junction.

Change in head–tip distance (mm)

Head Y-position change (mm) Modular stems

Standard Modular

0.5

Head Y-position change (mm) Standard stems

0 0.0

–0.5

–1

Follow-up (months)

–1.0

12 24

–2

–1.5

100

–2

Months from index operation

Figure 5. Mean values with 95% CI of the reduction in head–tip distance in mm for different follow-up moments in months up to 5 years for the standard design and up to 8 years for the modular version.

–1

Head X-position change (mm)

–2

–1

Head X-position change (mm)

Figure 6. Change in position of the hip head with respect to the postoperative situation in X-direction (perpendicular to the hip–stem axis) and Y-direction (along the hip–stem axis), for 1, 2, 5, and 8 years’ postoperative follow-ups. The ellipse presents the 95% prediction interval of the head position change for each follow-up moment.

tion. None of the modular necks showed perioperative signs of loosening from the stem, and they had to be dismounted with force. The metal on both stem and neck junctions showed signs of corrosion with black discoloration (Figure 4). 1 of the hips in the standard group was revised before the 5-year follow-up because of periprosthetic infection (Table 2). Radiostereometric analysis For the modular group, at 5 years, the mean change in HTD was –0.75 mm (range –1.64 to 0.14), equivalent to –0.15 mm/ year. For the standard group, the change was –0.09 mm (–1.07 to 0.33) or –0.02 mm/year. We then continued to follow the modular group, and at 8 years the change in HTD was –1.21 mm (–1.94 to –0.10) or still at the same pace of –0.15 mm/year.

This HTD reduction was statistically significant over time for the modular group (p < 0.001) but not for the standard group (p = 0.3). There was a statistically significant difference in HTD between the modular and standard groups from the 2nd year follow-up onwards, and by the 5-year follow-up the difference was 0.66 mm (Figure 5). No comparison could be made at 8 years as we followed the standard group for only 5 years. The reduction in the HTD was evident in modular stems but not in standard stems (Figure 6). Metal ion measurements We found a difference between standard and modular designs for all metal ion results at 5 years, with higher levels for the modular group (Table 3).

Acta Orthopaedica 2021; 92 (1): 66–72

Table 3. Metal ion levels at 5- and 8-year follow-up for stem designs. Values are mean µg/L (95% CI) Factor Modular Standard

5 years

Cobalt Chromium Titanium Cobalt Chromium Titanium

8 years

4.9 (4.1−5.7) 1.8 (1.5−2.0) 1.3 (1.1−1.5) 1.0 (0.7−1.4) 0.9 (0.4−1.4) 0.8 (0.6−1.0)

4.8 (4.3−5.3) 1.3 (1.0−1.6) 1.2 (1.0−1.5)

Cobalt serum concentration (µg/L) 10

Table 4. ALTR grades assessed on MARS-MRI 0

ALTR grades

Modular (n = 45)

Standard (n = 22)

0 19 13 1 10 2 2 2 0 3 6 1 4 3 5 Missing 5 1 0 = no ALTR; 1 = cystic ALTR with a wall thickness of < 3 mm; 2 = cystic ALTR with a wall thickness of > 3 mm and without any solid parts; 3 = cystic ALTR with a wall thickness of > 3 mm with a solid part, but comprising less than 50% of the total ALTR area; 4 = solid ALTR.

According to estimates from our linear regression model for the modular stem, a 1mm reduction in HTD corresponds to a 1.9 µg/L (CI 1.03–2.8) increase in whole-blood cobalt concentration at 8 years (Figure 7). We found no statistically significant reduction in HTD and whole-blood concentration of chromium or titanium (p = 0.4 and 0.6). The metal ion concentrations (Co, Cr, and Ti) leveled out after the 5-year follow-up. ALTR assessment on MARS-MRI There were no differences in ALTR grade between stem designs (Table 4), nor was there any correlation between the level of any of the metal ions and grade of ALTR. Modular components used Stem size, neck length, neck angle, head length, neck version (anteverted, standard, retroverted), sex, and patient body weight did not influence HTD reduction. We used the median neck length to divide all modular stems into 2 equal groups (long vs. short) (Table 1) and found a mean of 0.3 (CI –0.0 to 0.6) mm greater reduction in HTD in longer total neck lengths. Clinical evaluation Pain and satisfaction were similar between modular and standard stems preoperatively and throughout the 5-year follow-up for pain and satisfaction. The increase in whole-blood cobalt

0.5

1.0

1.5

2.0

Reduction in head-tip distance (mm)

Figure 7. Estimates from our linear regression model showing cobalt whole-blood concentration vs. reduction in head–tip distance at the 8-year follow-up.

concentration and the reduction in HTD did not affect the HOOS. No correlation was found between either type or size of ALTR and pain or satisfaction scores at 5 years. All patients showed good HOOS score throughout the 5-year follow-up.

Discussion We studied the reduction in head–tip distance (HTD) of a modular hip system and compared it with the HTD reduction of a standard hip system. The HTD reduced for the modular group at a rate of 0.15 mm/year. At 8-year follow-up, this HTD reduction was 1.2 mm. We interpret the reduced HTD as a varus deformation in the neck–stem junction. The modular neck– stem deformation correlated with the level of cobalt concentration, and the revised stems showed signs of corrosion. We therefore suspect that the HTD reduction was caused by corrosion at the neck–stem interface, probably in combination with a deformation of the softer stem titanium alloy (TMZF) under the load of the harder neck, made of CoCr alloy (Vitallium). The increased modification possibilities of modular stems with different neck options have previously been claimed as valuable for ROM, soft tissue balance, and to minimize leg length discrepancies (Barrack 1994, Jones 2004, Archibeck et al. 2011, Srinivasan et al. 2012). The RSA stem migration data showed that the ABG II proximal fit stem, in both the modular and non-modular version, is stable in its resting position after an initial slight subsidence and retroversion within the 3 first postoperative months (Kiernan et al. 2020). There have, however, also been reports on disadvantages related to the additional neck–stem interface when using modular stems. Some have reported fractures of the modular neck (Atwood et al. 2010, Wright et al. 2010) and others have reported ALTR to the metal debris caused by corrosion related to titanium–cobalt–chromium interfaces in modular stem junctions

Acta Orthopaedica 2021; 92 (1): 66–72

(Duwelius et al. 2010, Gill et al. 2012, Walsh et al. 2015). Pivec et al. (2014) evaluated 202 ABG II modular stems and reported a 3% revision rate up to 2-year follow-up for reasons unrelated to corrosion and 30% revision rate because of corrosion-related symptoms before 2 years. Restrepo et al. (2014) reported in 195 patients 13% revisions at 2-year follow-up for the ABG II modular stem. 8 of our 50 patients with modular stems had been revised 8 years after surgery. Although only 1 of these revisions was directly related to pain in association with corrosion at the neck–stem junction, another patient already had severe ALTR with groin pain before the incidence of PPFF, which led to its revision. The revised necks have all been firmly attached to the stem body, and they did not seem to be loose. A high force was needed to separate the necks from the stem. This high rate of revisions has raised general concerns, and Stryker recalled the ABG II modular stem in June 2012 when concerns arose due to the potential for corrosion at the neck–stem junction. We are the first to report on the steady rate of HTD reduction in ABG II modular stems, and we are not aware that this phenomenon has been described for any other modular hip stem design before. We found that the modular ABG II stem releases more metal ions into the surrounding tissue, compared with the standard ABG II. This seems to be the case for all modular stems and is confirmed by other studies (Pivec et al. 2014, Kwon et al. 2016, Nawabi et al. 2016). We did not find a correlation between elevated metal ion levels and type of ALTR in MRI as reported in other studies (Cooper et al. 2013, Walsh et al. 2015, Kwon et al. 2018), nor did we find a difference in ALTR between the two study groups or a relation between ALTR and clinical results. The slow deformation caused by the ongoing corrosion process might not be so harmful in itself, but how long can it continue? We are as yet unable to draw any conclusions regarding the outcome in the long term. It is hoped we shall see a continued leveling out of whole-blood ion concentrations in future follow-ups. This might be an indicator of a steadier state at the neck–stem junction. A far worse scenario is continuing varus deformity of the stem with modular-neck fractures or dislocation of the neck–stem junction. We therefore will continue to follow these patients with RSA and metal ions in blood and, when needed, with MRI. A limitation of this study is that we lack patient-reported outcome measures for the 8-year follow-up. It was an administrative mistake that these questionnaires were not sent out for the modular stem patients at 8 years, but we will continue to do so in later follow-ups. Our study is unique in the sense that we measured the HTD with RSA, and we suggest it as a valuable tool for measuring the integrity of a modular implant. If we had been specifically looking for this phenomenon, we would have noticed the difference within the first 2 years after surgery. In conclusion, in this modular stem design, there is a corrosion-related release of cobalt ions in particular, with a cor-

related progressive varus deformation of the neck–stem connection. This does not, however, seem to correlate with ALTR, and up to 8 years we have not yet seen a definite clinical problem with neck deformation, but further follow-up is needed. SK: Planning, collection of data, and data analysis, writing of manuscript. BK: Critical comments and help with the writing of the manuscript. RSA analysis. CF: Collection of data regarding metal ions and MARS MRI and help with writing that part of the manuscript. MS and GF: Planning of study, performing surgery, critical comments, and help in the writing of the manuscript. The principal investigator, SK, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors would like to thank Anna Åkesson, medical statistician; Clinical Studies Region Skåne for statistical analysis and Håkan Leijon, research engineer at the biomechanics lab for RSA analysis. They also thank Åsa Björkqvist, research secretary, Department of Orthopedics, Skåne University Hospital, for keeping track of our patients administratively. Acta thanks Aleksi Reito and Timothy Wright for help with peer review of this study.

Archibeck M J, Cummins T, Carothers J, Junick D W, White R E Jr. A comparison of two implant systems in restoration of hip geometry in arthroplasty. Clin Orthop Relat Res 2011; 469(2): 443-6. doi: 10.1007/s11999010-1678-9 [doi]. Atwood S, Archibeck A, Patten E W, Bozic K J, Pruitt L A, Ries M D. Corrosion-induced fracture of a double-modular hip prosthesis: a case report. J Bone Joint Surg Am 2010; 92(6): 1522-5. doi: 10.2106/jbjs.I.00980. Barrack R L. Modularity of prosthetic Implants. J Am Acad Orthop Surg 1994; 2(1): 16-25. doi: 10.5435/00124635-199401000-00003 [doi]. Cooper H J, Urban R M, Wixson R L, Meneghini R M, Jacobs J J. Adverse local tissue reaction arising from corrosion at the femoral neck–body junction in a dual-taper stem with a cobalt-chromium modular neck. J Bone Joint Surg Am 2013; 95(10): 865-72. doi: 1684766 [pii] 10.2106/ JBJS.L.01042 [doi]. Duwelius P J, Hartzband M A, Burkhart R, Carnahan C, Blair S, Wu Y, Grunkemeier G L. Clinical results of a modular neck hip system: hitting the “bull’s-eye” more accurately. Am J Orthop 2010; 39(10 Suppl.): 2-6. Frisch N B, Lynch J R, Pourzal R, Banglmaier R F, Silverton C D. Dual-taper modular hip implant: investigation of 3-dimensional surface scans for component contact, shape, and fit. Arthroplast Today 2018; 4(3): 370-5. doi: 10.1016/j.artd.2018.06.001 [doi] S2352-3441(18)30063-3 [pii]. Gilbert J L, Buckley C A, Jacobs J J. In vivo corrosion of modular hip prosthesis components in mixed and similar metal combinations: the effect of crevice, stress, motion, and alloy coupling. J Biomed Mater Res 1993; 27(12): 1533-44. doi: 10.1002/jbm.820271210 [doi]. Gill I P, Webb J, Sloan K, Beaver R J. Corrosion at the neck–stem junction as a cause of metal ion release and pseudotumour formation. J Bone Joint Surg Br 2012; 94(7): 895-900. doi: 10.1302/0301-620x.94b7.29122. Hauptfleisch J, Pandit H, Grammatopoulos G, Gill H S, Murray D W, Ostlere S. A MRI classification of periprosthetic soft tissue masses (pseudotumours) associated with metal-on-metal resurfacing hip arthroplasty. Skeletal Radiol 2012; 41(2): 149-55. doi: 10.1007/s00256-011-1329-6. Jones R E. Modular revision stems in total hip arthroplasty. Clin Orthop Relat Res 2004; (420): 142-7. doi: 00003086-200403000-00020 [pii] 10.1097/00003086-200403000-00020 [doi]. Kaptein B L, Valstar E R, Spoor C W, Stoel B C, Rozing P M. Modelbased RSA of a femoral hip stem using surface and geometrical shape models. Clin Orthop Relat Res 2006; 448: 92-7. doi: 10.1097/01. blo.0000224010.04551.14. [doi] 00003086-200607000-00015 [pii].

Kiernan S, Geijer M, Sundberg M, Flivik G. Effect of symmetrical restoration for the migration of uncemented total hip arthroplasty: a randomized RSA study with 75 patients and 5-year follow-up. J Orthop Surg Res 2020; 15(1): 225. Kwon Y M. Evaluation of the painful dual taper modular neck stem total hip arthroplasty: do they all require revision? J Arthroplasty 2016; 31(7): 13859. doi: S0883-5403(16)00296-5 [pii] 10.1016/j.arth.2016.01.074 [doi]. Kwon Y M, MacAuliffe J, Arauz P G, Peng Y. Sensitivity and specificity of metal ion level in predicting adverse local tissue reactions due to head– neck taper corrosion in primary metal-on-polyethylene total hip arthroplasty. J Arthroplasty 2018; 33(9): 3025-9. doi: S0883-5403(18)30356-5 [pii] 10.1016/j.arth.2018.04.007 [doi]. Lindgren J U, Brismar B H, Wikstrom A C. Adverse reaction to metal release from a modular metal-on-polyethylene hip prosthesis. J Bone Joint Surg Br 2011; 93(10): 1427-30. doi: 93-B/10/1427 [pii] 10.1302/0301620X.93B10.27645. McGrory B J, McKenney B R. Revision for taper corrosion at the head–neck junction: pearls and pitfalls. Curr Rev Musculoskelet Med 2016; 9(1): 97-102. doi: 10.1007/s12178-016-9325-z [doi] 10.1007/s12178-0169325-z [pii]. Molloy D O, Munir S, Jack C M, Cross M B, Walter W L, Walter W K Sr. Fretting and corrosion in modular-neck total hip arthroplasty femoral stems. J Bone Joint Surg Am 2014; 96(6): 488-93. doi: 1840111 [pii] 10.2106/ JBJS.L.01625 [doi]. Morlock M M. The taper disaster: how could it happen? Hip Int 2015; 25(4): 339-46. doi: ACB92214-20C1-4377-82CF-9F58ABEFDFEF [pii] 10.5301/hipint.5000269[doi] . Morlock M M, Dickinson E C, Gunther K P, Bunte D, Polster V. Head taper corrosion causing head bottoming out and consecutive gross stem taper failure in total hip arthroplasty. J Arthroplasty 2018; 33(11): 3581-90. doi: S0883-5403(18)30660-0 [pii] 10.1016/j.arth.2018.07.017[doi] . Nahhas C R, Yi P H, Moric M, Puri R, Jacobs J J, Sporer S M. High failure at a minimum 5-year follow-up in primary total hip arthroplasty using a modular femoral trunnion. J Arthroplasty 2019; 34(7): 1395-9. doi: S08835403(19)30256-6 [pii] 10.1016/j.arth.2019.03.033 [doi]. Nawabi D H, Do H T, Ruel A, Lurie B, Elpers M E, Wright T, Potter H G, Westrich G H. Comprehensive analysis of a recalled modular total hip

Acta Orthopaedica 2021; 92 (1): 66–72

system and recommendations for management. J Bone Joint Surg Am 2016; 98(1): 40-7. doi: 10.2106/jbjs.N.01121. Patel S, Talmo C T, Nandi S. Head–neck taper corrosion following total hip arthroplasty with Stryker Meridian stem. Hip Int 2016; 26(6): e49e51. doi: 2A01EAF5-5D6F-4D7B-8029-FD399BEAFF42 [pii] 10.5301/ hipint.5000440 [doi]. Pivec R, Meneghini R M, Hozack W J, Westrich G H, Mont M A. Modular taper junction corrosion and failure: how to approach a recalled total hip arthroplasty implant. J Arthroplasty 2014; 29(1): 1-6. doi: 10.1016/j. arth.2013.08.026.

Restrepo C, Ross D, Restrepo S, Heller S, Goyal N, Moore R, Hozack W J. Adverse clinical outcomes in a primary modular neck/stem system. J Arthroplasty 2014; 29(9 Suppl.): 173-8. doi: 10.1016/j.arth.2014.01.040. Rodushkin I, Ödman F, Olofsson R, Axelsson M D. Determination of 60 elements in whole blood by sector field inductively coupled plasma mass spectrometry. J Analytical Atomic Spectrometry [10.1039/B003561K] 2000; 15(8): 937-44. doi: 10.1039/B003561K.

Shah R, Talmo C, Nandi S. Titanium neck–titanium stem taper corrosion in a modular neck stem. Arthroplast Today 2019; 5(2): 145-7. doi: 10.1016/j. artd.2019.04.009 [doi] S2352-3441(19)30042-1 [pii]. Srinivasan A, Jung E, Levine B R. Modularity of the femoral component in total hip arthroplasty. J Am Acad Orthop Surg 2012; 20(4): 214-22. doi: 10.5435/jaaos-20-04-214. Su S L, Koch C N, Nguyen T M, Burket J C, Wright T M, Westrich G H. Retrieval analysis of neck–stem coupling in modular hip prostheses. J Arthroplasty 2017; 32(7): 2301-6. doi: 10.1016/j.arth.2017.02.016.

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: V42136W7L1733G68 [pii] 10.1080/17453670510041574 [doi]. Walsh C P, Hubbard J C, Nessler J P, Markel D C. Revision of recalled Modular Neck Rejuvenate and ABG femoral implants. J Arthroplasty 2015; 30(5): 822-6. doi: 10.1016/j.arth.2014.12.002.

Wright G, Sporer S, Urban R, Jacobs J. Fracture of a modular femoral neck after total hip arthroplasty: a case report. J Bone Joint Surg Am 2010; 92(6): 1518-21.

Acta Orthopaedica 2021; 92 (1): 73–79

Dutch Guideline on Knee Arthroscopy Part 1, the meniscus: a multidisciplinary review by the Dutch Orthopaedic Association Ewoud R A VAN ARKEL 1, Sander KOËTER 2, Paul C RIJK 3, Tony G VAN TIENEN 4, Patrice W J VINCKEN 5, Michiel J M SEGERS 6, Bert VAN ESSEN 7, Nicky VAN MELICK 8, and Bernardine H STEGEMAN 9 1 FocusKliniek Orthopedie, Haaglanden Medisch Centrum, Den Haag; 2 Department of Orthopedic Surgery, Canisius Wilhelmina Ziekenhuis, Nijmegen; 3 Department of Orthopedic Surgery, Medisch Centrum Leeuwarden, Leeuwarden; 4 Department of Orthopedic Surgery, Radbout UMC, Nijmegen; 5 Department of Radiology, Alrijne Hospital, Leiderdorp; 6 Department of Surgery, Sint Antonius Ziekenhuis, Utrecht; 7 Department of Sportsmedicine, Maxima Medisch Centrum, Veldhoven; 8 Knee Expert Center, Eindhoven; 9 Kennisinstituut van de Federatie Medisch Specialisten, Utrecht, The

Netherlands Correspondence: e.van.arkel@haaglandenmc.nl Submitted 2020-06-09. Accepted 2020-10-09.

Background and purpose — A guideline committee of medical specialists and a physiotherapist was formed on the initiative of the Dutch Orthopedic Association (NOV) to update the guideline Arthroscopy of the Knee: Indications and Treatment 2010. This next guideline was developed between June 2017 and December 2019. In this Part 1 we focus on the meniscus, in Part 2 on all other aspects of knee arthroscopy. Methods — The guideline was developed in accordance with the criteria of the AGREE instrument (AGREE II: Appraisal of Guidelines for Research and Evaluation II) with support of a professional methodologist from the Dutch Knowledge Institute of Medical Specialists. The scientific literature was searched and systematically analyzed. Conclusions and recommendations were formulated according to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) method. Recommendations were developed considering the balance of benefits and harms, the type and quality of evidence, the values and preferences of the people involved, and the costs. Here in Part 1 we focus on the diagnosis of meniscal pathology, treatment of meniscal injuries, and persistent pain after meniscal repair and after treatment. In this guideline, the importance of nonoperative treatment of degenerative meniscal tears, the potential for spontaneous healing of traumatic meniscal tears, and encouragement to repair rather than resect meniscal tears when possible are highlighted. The guideline is published online, in Dutch, and is available from the Dutch Guideline Database (https://richtlijnendatabase.nl/?query=Artroscopie+van+de+knie+&special ism=)

The Dutch Orthopedic Association (NOV) has a long tradition of development of practical clinical guidelines with the use of the GRADE method. This is a systematic and transparent approach to collecting and grading of the available evidence and to weighing the evidence together with complementary arguments, so-called considerations, relevant to the clinical question—including patient values and preferences, and resource use (cost, organization of care issue). It is a dynamic tool in which a particular module can be altered if there are new insights (Besselaar et al. 2017). The guideline Knee Arthroscopy describes the indication, diagnosis, and treatment for knee arthroscopy. For better readability we have divided the guideline into 2 parts. In Part 1 we focus on the meniscus. In Part 2 we discuss all other aspects of arthroscopic knee surgery (Koëter et al. 2020). We do not address knee arthroscopy in children, with specific diagnoses such as discoid meniscus and osteochondritis dissecans. The implementation of the guideline Arthroscopy of the Knee: Indications and Treatment 2010 contributed to a decrease in incidence of meniscal procedures in all age groups in the Netherlands from 2005 to 2014, with a more pronounced decrease in the younger patients (Rongen et al. 2018). Due to “game changing” studies on the meniscus and further technical developments in the field of arthroscopic treatment of knee complaints and developments of the technique of MRI, it became necessary to revise the 2010 guideline Arthroscopy of the Knee. 5 clinical questions on the meniscus were formulated by a steering group of the Dutch Orthopedic Association (see Guideline recommendations below). This guideline aims to provide a uniform policy for the care of patients with knee injuries that could possibly be treated with an arthroscopic procedure. It is written for orthopedic surgeons, sport medicine specialists, physiotherapists, radi-

ologists, and trauma surgeons who are involved in the care of patients with (acute) knee injuries. In addition, this guideline is intended to inform healthcare providers who are also involved in the care of these patients, including pediatricians, rehabilitation doctors, general practitioners, physician assistants, and nurse practitioners. Funding and potential conflicts of interest The guideline development was financially supported by the Dutch Orthopaedic Association (NOV), using governmental funding from the Quality Foundation of the Dutch Association of Medical Specialists in the Netherlands. The authors declare that there is no relevant conflict of interest.

Method Guideline panel In November 2016, a guideline panel, tasked with revising the guideline, was formed consisting of orthopedic surgeons, a radiologist, a trauma surgeon, a physiotherapist, and a sports medicine specialist. A methodologist from the Knowledge Institute of the Federation of Medical Specialists supported the guideline panel by ensuring proper design and systematic evidence-based development of the guideline using the GRADE methodology, to meet all the criteria of the AGREE instrument. Methodology and workflow The guideline was developed in agreement with the criteria set by the advisory committee on guideline development of the Federation Medical Specialist in the Netherlands, which are based on the AGREE II instrument (Brouwers et al. 2010). The guideline was developed using an evidence-based approach endorsing the GRADE methodology, and meets all criteria of AGREE II. Grading of Recommendations Assessment, Development, and Evaluation (GRADE) is a systematic approach for synthesizing evidence and grading of recommendations offering transparency at each stage of the guideline development (Guyatt et al. 2011, Schünemann et al. 2014). The guideline development process involves a number of phases: a preparatory phase, a development phase, a commentary phase, and an authorization phase. After authorization, the guideline has to be disseminated and implemented. Furthermore, uptake and use must be evaluated. Finally, the guideline must be kept up-to date. Each phase involves a number of practical steps (Schunemann et al. 2014). As a first step in the early preparatory phase, a broad forum discussion was held and all relevant stakeholders were consulted to define and prioritize key issues, which were extensively discussed in the guideline panel. The selected, highpriority issues were translated into carefully formulated clinical questions. These questions defined patient problems, intervention, comparison, and outcomes. Furthermore, the patient out-

Acta Orthopaedica 2021; 92 (1): 73–79

comes relevant to decision-making were prioritized and minimal clinically important differences were defined. In the development phase, the literature was systematically searched using the databases MEDLINE and Embase. Selection of the relevant literature was based on predefined inclusion and exclusion criteria and was carried out by 1 member of the guideline panel (EvA) in collaboration with the methodologist (BS). For each of the clinical questions, the evidence was summarized by the guideline methodologist using the GRADE approach. A systematic review was performed for each of the relevant outcomes and the quality of evidence was assessed in 1 of 4 grades (high, moderate, low, very low) by analyzing limitations in study design or execution (risk of bias), inconsistency of results, indirectness of evidence, imprecision, and publication bias. The evidence synthesis was complemented by a guideline panel member considering any additional arguments relevant to the clinical question, including patient values, preferences, and resource use (costs, organization of care issues). Evidence synthesis, complementary arguments, and concept recommendations were extensively discussed in the guideline committee. Final recommendations were then formulated. The final recommendations are based on the balance between desirable and undesirable outcomes, the quality of the body of evidence across all relevant outcomes, values and preferences, and resource use. The strength of a recommendation reflects the extent to which the guideline panel was confident that desirable effects of the intervention would outweigh undesirable effects or vice versa, across the range of patients for whom the recommendation is intended. The strength of a recommendation is determined by weighing all relevant arguments together. This includes the weight of the body of evidence from the systematic literature analysis, and also the weight of all complementary arguments formulated, the so-called considerations. When using the GRADE approach, guideline panels must use judgment in integrating these arguments to make a strong or weak recommendation. Thus, although low quality of the body of evidence from the systematic literature analysis will generally result in a weak recommendation, it does not a priori exclude a strong recommendation, and weak recommendations may also result from high-quality evidence (Schunemann et al. 2014). After reaching consensus from the guideline panel, the concept guideline was subjected to peer review by all the relevant stakeholders: the commentary phase. Amendments were made and agreed upon by the guideline panel, and the final text was presented to the Dutch Orthopedic Association (NOV), the Dutch Society for Radiology (NVvR), the Royal Dutch Society for Physical Therapy (KNGF), the Dutch Sports Medicine Association (VSG), and the Association of Surgeons of the Netherlands (NVvH) for authorization. In this authorization phase, additional amendments were made to the guideline text based on the outcome of a general assembly of the NOV. The guideline was finally approved and officially authorized by the NOV in March 2019.

Acta Orthopaedica 2021; 92 (1): 73–79

Guideline recommendations according to 5 clinical questions, for literature reviews, see Supplementary data 1. What is the value of the different meniscus tests during physical examination?

Recommendation

• Do not perform arthroscopy based on 1 or more meniscus tests without additional information from history, physical examination, and any additional radiological examination.

• Always consider meniscal repair or follow a wait-and-see policy. Meniscal injury does not necessarily mean meniscectomy. • Leave the peripheral rim of the meniscus intact. • Always try to repair a meniscal tear in young patients if the tear is in the vascularized part of the meniscus. Here, a stable knee or an unstable knee that is stabilized within 6 weeks is indispensable. Degenerative meniscus injury • Start with nonoperative treatment in degenerative meniscus injury. • Consider treating nonoperatively for at least 3 months in the event of a meniscal tear.

2. What is the place of imaging techniques such as conventional radiographs, MRI, and ultrasound in the diagnostic process?

5. What is the added value of physiotherapy after arthroscopic meniscus surgery of the knee?

Recommendation

• Do not refer patients with an expected normal recovery to the physiotherapist after a meniscectomy. • Discuss with the patient with a delayed recovery the expected effects of physical therapy.

• Perform imaging, such as conventional radiographs and/or MRI, before performing an arthroscopy. • Perform an MRI in addition to history and physical examination in patients younger than 50 years, unless there is a high a priori chance of intra-articular injury. In that case an arthroscopy without MRI is indicated (provided conventional radiograph is done). A high a priori chance is defined as: history of a traumatic moment, hydrops, and a locked knee. • Do not routinely perform MRI in patients older than 50 years with knee complaints, but start with an AP and lateral radiograph of the knee, preferably with fixed flexion at 20°. • Be cautious when applying ultrasound in the indication for arthroscopy due to insufficient visibility of (intra)osseous and intra-articular structures of the knee. 3. What is the additional value of CT and MRI arthrography compared with conventional MRI in patients with a meniscus repair after injury?

Recommendation

• Consider (direct) MR arthrography over conventional MRI as additional diagnostics for persistent complaints after an arthroscopic treatment of a meniscal injury. 4. Which meniscus injuries should be treated, when and how?

Recommendation Acute meniscal injury • Perform arthroscopy within 2 weeks of injury when a patient has a locked knee with the most likely cause of a ruptured meniscus.

Recommendation

Discussion The guideline is aimed at providing evidence-based advice to medical specialists and physiotherapists, in order to minimize unwarranted variation in treatment of meniscus injuries and to improve the therapeutic results. The 1st question concerned the values of diagnostic meniscus tests. The execution of meniscus tests is part of the standard physical examination during consultation, in all patients with (non)traumatic knee complaints. Other findings during the physical examination, such as a locked knee, or swelling of the knee, may also support the suspicion of meniscal disease. A locked knee may be a reason to qualify a patient for arthroscopy without prior MRI, provided a conventional weightbearing knee radiograph has been taken to exclude other pathology. Analysis of the literature clearly shows that performing meniscus tests alone, separately or in combination is diagnostically insufficiently accurate (Goossens et al. 2015, Smith et al. 2015). In a combination of both a negative Thessaly and McMurray test, the presence of a meniscal injury is unlikely. Further caution with the use of meniscus tests is appropriate if there is concomitant ligamentous injury such as an anterior cruciate ligament rupture. This further decreases the reliability of the meniscus tests. Additional information from demographic factors, history, physical examination, and possibly additional examination (MRI) has to support the diagnosis before treatment decisions can be made. The 2nd question concerned the place and reliability of imaging of ultrasound compared with MRI techniques in the

Acta Orthopaedica 2021; 92 (1): 73–79

diagnostic process. Prior to arthroscopy osseous pathology (fracture, neoplasm, and infectious processes such as osteomyelitis) should be excluded by imaging. A general MRI scanning protocol, suitable for the assessment of menisci, ligaments, and cartilage, should consist of scans in different planes: sagittal, coronal, and, where relevant, axial. In the choice of 35 sequences at least a combination of short TE (T1- or PD-weighted images) and T2-weighted images with or without fat suppression should be used. Because there is a learning curve in reading of MR images, experience and training of the reader will increase the accuracy of MRI (White et al. 1997). Experience of the reader has more influence on accuracy of MRI than field strength of the system (Krampla et al. 2009). Therefore, reading of the MRI by a (musculoskeletal) radiologist, with adequate training and experience, is essential to achieve the highest accuracy. A limited number of studies, compared with those on MRI, indicate that ultrasound of the knee can be accurate in the assessment of menisci and cruciate ligaments. Transducers used in the aforementioned studies ranged from 5 to 14 MHz. No information is available on the influence of the frequency of transducers on accuracy, but higher frequencies yield higher resolution images, probably increasing accuracy. Cartilage can be evaluated only to a minor extent. Bone and bone marrow cannot be assessed with ultrasound. This seriously limits the diagnostic yield of ultrasonography when compared with MRI. The learning curve for performing musculoskeletal ultrasound is considerable, because not only the interpretation of the obtained images but also eye–hand coordination benefits greatly from training and experience. Therefore, ultrasonography is only a reliable diagnostic tool in the hands of an experienced musculoskeletal ultrasonographer. Advocates of ultrasound point to the fact that availability of ultrasound machines is higher than that of MRI systems. This may be true, but the limiting factor will probably be the availability of adequately trained and experienced ultrasonographers. The disadvantage compared with MRI is that the ultrasonographer has to perform the examination in person to make an adequate report, whereas MRI can be reported by a radiologist independent of moment and location of examination. This facilitates plan-

Acta Orthopaedica 2021; 92 (1): 73–79

ning. An MRI scan consists of multiple images of predefined thickness and interval in at least two orthogonal planes. This means every orthopedic surgeon or radiologist can identify and locate pathology in the knee based on information visible in these images. Because ultrasound is a dynamic examination and the number of obtained images, orientation, and quality of images is entirely operator-dependent; only the reporter/ultrasonographer can extract all the information from an examination. Others will have to rely on the report. This will diminish the added value of ultrasound in preoperative planning or in giving the patient insight into the pathology. When arthroscopy is warranted, MRI can be used to exclude (oncologic) osseous pathology and radiography is not required. Ultrasound cannot fulfill that role because visualization of osseous structures is insufficient. Additional imaging (radiography) is required. This means that ultrasound cannot be a stand-alone diagnostic tool before arthroscopy. Another disadvantage of ultrasound is the inability to visualize bone marrow changes at all, or cartilage in a sufficient manner. Bone bruising or focal chondral lesions can mimic meniscal pathology and the bone bruise pattern can point to specific trauma mechanisms and associated pathology. And one could also assume the use of ultrasound in obese patients is limited. Compared with MRI, ultrasound therefore lacks in completeness. In our opinion, MRI is the diagnostic imaging of choice in patients younger than 50 years and without history of locking or catching or extension deficit (history of trauma, effusion, and extension deficit) (Vincken et al. 2007). Ultrasound is not equivalent. The 3rd question involved patients with persistent complaints after repair. MR arthrography may have additional value over conventional MRI, as there is evidence that MR arthrography has a higher accuracy for detecting unhealed or re-ruptured menisci. The cost of the study is likely to be higher due to the use of intra-articular contrast. There will also be a higher risk of complications due to the (minimally) invasive nature. No literature data is available showing which strategy (conventional MRI and possibly additional MR arthrography, MR arthrography alone, or direct second-look arthroscopy) is preferable from the point of view of cost-effectiveness or clinical outcome. CT arthrography is also believed to have higher accuracy for detecting unhealed or torn menisci than with conventional MRI. However, no article was included in our literature search confirming this. De Filippo et al. (2009) examined CT arthrography in patients with MRI contraindication, but studies comparing different imaging modalities are lacking. Availability, expertise, and local customs also play a role in these considerations. The 4th question addressed the treatment of acute meniscus injury and degenerative meniscal lesions. Concerning acute meniscal injury, the working group is of the opinion that studies more than 8 years old, comparing total meniscectomy with partial meniscectomy, are no longer relevant, because standard

care has changed in recent years. It is now preferable to repair a meniscus and if not possible then to perform an arthroscopic partial meniscectomy. Xu and Zhao (2015) undertook a metaanalysis of the comparison between meniscus repair and a meniscectomy with better outcomes for meniscus repair. There is currently insufficient scientific evidence to determine when and which meniscal lesions should be repaired or removed to obtain optimal outcomes in the short and longer term. However, it seems more prudent to have low-threshold suturing in younger patients with a lateral meniscal injury than to perform a partial meniscectomy because of the long-term results after lateral meniscectomy (Hulet et al. 2015). The working group considers meniscus tears to be repairable when they are torn close to or separated from the knee joint capsule, or a longitudinal tear in the peripheral part the “so-called red-red zone” where the healing potential is best because of the vascularization, provided that the torn meniscus tissue is of good quality and the knee is stable. This usually concerns traumatic meniscal tears. Spontaneous healing of meniscal injuries has also been extensively described, both in combination with anterior cruciate ligament rupture and with isolated meniscal injuries. The working group is of the opinion that in the case of a peripheral longitudinal tear of the meniscus proven on MRI and no restriction of movement in the knee, a wait-and-see policy can be pursued. Due to the chance of spontaneous healing, overtreatment may result. Recovery after meniscus repair takes longer than partial meniscectomy. No evidence-based post-treatment protocol is available, but it is generally advised to do partial weightbearing for 4 to 6 weeks and limit flexion to 90 degrees. Return to sport level is advised after 3 to 6 months. Postoperative rehabilitation should be discussed explicitly with a top athlete so that a well-considered decision can be made with regard to repair or partial meniscectomy. We recommend a different approach to medial vs. lateral meniscus tears. The younger the patient, the more aggressive the surgeon should be in repair of the lateral meniscus. And in the case of a bucket-handle tear in combination with an anterior cruciate ligament rupture, meniscus repair should be performed in combination with an ACL reconstruction. The second part of the 4th question concerns degenerative menisci, as regards treatment of degenerative meniscal tears. Most studies concerning degenerative meniscus injury used 3 months as the “short-term follow-up.” During the first 3 months after surgery and nonoperative treatment, a reduction in complaints was measured and the difference in effect of treatment between the two groups seemed minimal. This minimal difference continued up to 24 months. In case of nonobstructive meniscus complaints, conservative treatment is therefore preferable to surgery for the first 3 months after initiations of complaints. The working group has set the age limit for degenerative meniscal injuries to > 50 years, but this is open to debate. The reason for choosing 50 years was to stay in line with the knee osteoarthritis guideline and the asso-

ciated radiographic diagnostics. Progressive insight shows that lowering the age of 45 years produces the same results and thus can also be considered. Perhaps in the future the age recommendation will further decrease to 35 years, but more research is needed. The5th question addressed physical therapy. Physical therapy is an intervention that entails hardly any risks or complications. In the Netherlands, the first 20 physiotherapy treatments after surgery are reimbursed by the patient’s additional insurance. From the 21st treatment onward, reimbursement from the basic insurance applies up to 12 months after the meniscectomy. If the patient has less than 20 treatments in his additional insurance, he will therefore have to pay for physiotherapy treatments him- or herself. Today’s society demands a return to work as soon as possible and physiotherapy may be able to contribute to this. In addition, it is often the wish of the patient to be able to quickly return to the old level, particularly when it comes to sports. In certain professional groups (top athletes, heavy physical work), counseling in postoperative recovery can therefore be useful, also to prevent secondary injuries. For example, it may be useful to monitor a top athlete more frequently in connection with a step-by-step build-up. This group of patients often ignore symptoms because they want to get back into competition quickly or because of pressure from the media, the coach, the team, or the athletes. It is not only the type and treatment of meniscal injury that determines whether the patient should be referred to a physiotherapist, but more whether there is normal or (expected) delayed recovery. In the Royal Dutch Physical Therapists (KNGF) meniscectomy guideline two patient profiles are distinguished. Patient profile 1 concerns patients after partial meniscectomy who require little or no physical therapy because of expected normal recovery. These are usually the younger patients with an acute injury of the meniscus, with a blank history. Physiotherapy is desirable in this group only if there is comorbidity (such as an ACL rupture) or fear of movement. Patient profile 2 are patients with a burdened history of previous knee surgeries, in whom the complaints have arisen after repeated (micro)trauma, causing multiple and recurrent ruptures in the meniscus. This may be a sign of nascent osteoarthritis, but the distinction between meniscus pathology and early stage degenerative knee disease may not be clear. These patients are at high risk of delayed recovery and physical therapy is indicated. Other signs of delayed recovery are insufficient increase in function (mobility, gait pattern) and insufficient increase or even decline in activities and participation. In that case, physical therapy can help improve mobility and gait recovery, and increase strength and neuromuscular control, which may also increase activity and participation. However, as discussed in this guideline, in most cases these patients are no longer even eligible for arthroscopy and therefore should be treated nonoperatively.

Acta Orthopaedica 2021; 92 (1): 73–79

This guideline was composed for arthroscopic treatment of knee injuries. Here we present our recommendations concerning the meniscus. We should keep in mind that there is a continuum from a traumatic meniscal injury at a younger age to a degenerative meniscal lesion around midlife. We have changed our thoughts from performing partial meniscectomy in the case of meniscal lesions to first considering nonoperative treatment before meniscal repair. If surgery is indicated and meniscal repair is not possible partial meniscectomy should be considered. Spontaneous decrease in pain with meniscus lesions is possible. We should be aware that a degenerative meniscus lesion is one of the first signs of osteoarthritis of the knee. Supplementary data Literature reviews are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2020.1850086 Acta thanks Jon Drogset and Samuel Van de Velde for help with peer review of this study.

Besselaar A T, Sakkers R J B, Schuppers H A, Witbreuk M M, Zeegers E V, Boekestijn R A, Marges S D, van der Steen M C, Burger K N. Guideline on the diagnosis and treatment of primary idiopathic clubfoot. Acta Orthop 2017; 88(3): 305-9. Brouwers M C, Kho M E, Browman G P, Burgers J S, Cluzeau F, Feder G, Fervers B, Graham I D, Grimshaw J, Hanna S E, Littlejohns P, Makarski J, Zitzelsberger L. AGREE II: advancing guideline development, reporting and evaluation in health care. CMAJ 2010; 182(18): E839-E842. De Filippo M, Bertellini A, Pogliacomi F, Sverzellati N, Corradi D, Garlaschi G, Zompatori M. Multidetector computed tomography arthrography of the knee: diagnostic accuracy and indications. Eur J Radiol 2009; 70(2): 34251. doi:10.1016/j.ejrad.2008.01.034. Goossens P, Keijsers E, van Geenen R J, Zijta A, van den Broek M, Verhagen A P, Scholten-Peeters G G M. Validity of the Thessaly test in evaluating meniscal tears compared with arthroscopy: a diagnostic accuracy study. J Orthop Sports Phys Ther 2015; 45(1): 18-24, B1. doi: 10.2519/ jospt.2015.5215. Guyatt G H, Oxman A D, Schunemann H J, Tugwell P, Knottnerus A. GRADE guidelines: a new series of articles in the Journal of Clinical Epidemiology. J Clin Epidemiol 2011; 64(4): 380-2. Hulet C, Menetrey J, Beaufils P, et al. Clinical and radiographic results of arthroscopic partial lateral meniscectomies in stable knees with a minimum follow up of 20 years. Knee Surg Sports Traumatol Arthrosc 2015; 23(1): 225-31. doi: 10.1007/s00167-014-3245-5. Koëter S, van Tienen T G, Rijk P C, Vincken P W J, Seger M J M, van Essen B, van Melick N, Stegeman B H, van Arkel E R A. Dutch Guideline on Knee Arthroscopy Part 2: non-meniscus intra-articular knee injury: a multidisciplinary review by the Dutch Orthopaedic Association. Acta Orthop 2020; 91(x): 1–7. Epub ahead of print. Krampla W, Roesel M, Svoboda K, Nachbagauer A, Gschwantler M, Hruby W. MRI of the knee: how do field strength and radiologist’s experience influence diagnostic accuracy and interobserver correlation in assessing chondral and meniscal lesions and the integrity of the anterior cruciate ligament? Eur Radiol 2009; 19(6): 1519-28. doi: 10.1007/s00330-009-1298-5. Rongen J J, van Tienen T G, Buma P, Hannink G J. Meniscus surgery is still widely performed in the treatment of degenerative meniscus tears in The

Acta Orthopaedica 2021; 92 (1): 73–79

Netherlands. Knee Surg Sports Traumatol Arthrosc 2018; 26: 1123-9. doi: 10.1007/s00167-017-4473-2. Schünemann H J, Wiercioch W, Etxeandia I, Falavigna M, Santesso N, Mustafa R, Ventresca M, Brignardello-Petersen R, Laisaar K T, Kowalski S, Baldeh T, Zhang Y, Raid U, Neumann I, Norris S L, Thornton J, Harbour R, Treweek S, Guyatt G, Alonso-Coello P, Reinap M, Brozek J, Oxman A, Akl E A. Guidelines 2.0: systematic development of a comprehensive checklist for a successful guideline enterprise. CMAJ 2014; 186(3): E123-E142. Smith B E, Thacker D, Crewesmith A, et al. Special tests for assessing meniscal tears within the knee: a systematic review and meta-analysis. Evid Based Med 2015; 20(3): 88-97. doi: 10.1136/ebmed-2014-110160.

Vincken P W, ter Braak A P, van Erkel A R, Coerkamp E G, de Rooy T P, de Lange S, Mallens W M, Coene L N, Bloem R M, van Luijt P A, van den Hout W B, van Houwelingen H C, Bloem J L. MR imaging: effectiveness and costs at triage of patients with nonacute knee symptoms. Radiology 2007; 242(1): 85-93. White L M, Schweitzer M E, Deely D M, Morrison W B. The effect of training and experience on the magnetic resonance imaging interpretation of meniscal tears. Arthroscopy 1997; 13(2): 224-8. Xu C, Zhao J. A meta-analysis comparing meniscal repair with meniscectomy in the treatment of meniscal tears: the more meniscus, the better outcome? Knee Surg Sports Traumatol Arthrosc 2015; 23(1): 164-70. doi: 10.1007/ s00167-013-2528-6.

Acta Orthopaedica 2021; 92 (1): 80–83

Dutch Guideline on Knee Arthroscopy Part 2: non-meniscus intraarticular knee injury: a multidisciplinary review by the Dutch Orthopaedic Association Sander KOËTER 1, Tony G VAN TIENEN 2, Paul C RIJK 3, Patrick W J VINCKEN 4, Michiel J M SEGERS 5, Bert VAN ESSEN 6, Nicky VAN MELICK 7, Bernadine H STEGEMAN 8, and Ewoud R A VAN ARKEL 9 1 Department

of Orthopedic Surgery Canisius Wilhelmina Ziekenhuis, Nijmegen; 2 Department of Orthopedic Surgery Via Sana, Mill; 3 Department of Orthopedic Surgery Medisch Centrum Leeuwarden, Leeuwarden; 4 Department of Radiology Alrijne Hospital, Leiderdorp; 5 Department of Surgery Sint Antonius Ziekenhuis, Utrecht; 6 Department of Sports Medicine Maxima Medisch Centrum, Veldhoven; 7 Knee Expert Center, Eindhoven; 8 Kennisinstituut van de Federatie Medisch Specialisten, Utrecht; 9 FocusKliniek Orthopedie Haaglanden Medisch Centrum, Den Haag, The Netherlands Correspondence: s.koeter@cwz.nl Submitted 2020-06-14. Accepted 2020-09-08.

Background and purpose — A guideline committee of medical specialists and a physiotherapist was formed on the initiative of the Dutch Orthopedic Association (NOV) to update the Guideline Arthroscopy of the Knee: Indications and Treatment 2010. This next Guideline was developed between June 2017 and December 2019. In part 1 we focused on the meniscus; this part 2 addresses all other aspects of knee arthroscopy. Methods — The guideline was developed in accordance with the criteria of the AGREE instrument (AGREE II: Appraisal of Guidelines for Research and Evaluation II) with support of a professional methodologist from the Dutch Knowledge Institute of Medical Specialists. The scientific literature was searched and systematically analyzed. Conclusions and recommendations were formulated according to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) method. Recommendations were developed considering the balance of benefits and harms, the type and quality of evidence, the values and preferences of the people involved, and the costs. In this part 2 we focus on anterior knee pain, patellar tendinopathy, the role of arthroscopy in the osteoarthritic knee, arthroscopy and patellar dislocation, and osteochondral fractures, and additionally ask what the role is of arthroscopy in bacterial arthritis or ligamentous injury of the knee, and whether arthroscopy is supplemental in the treatment of tibial plateau fractures. We did not address the role of arthroscopy in the discoid meniscus and osteochondritis dissecans in children. The guideline is published online, in Dutch, and is available from the Dutch Guideline Database (https://richtlijnendatabase.nl/?query=Artroscopie+van+de+knie+&specialism=)

Like most musculoskeletal injuries, knee injuries can be painful and debilitating. Most knee injuries occur during activities of daily living or while participating in sports. In the Netherlands 17% of patients with knee complaints are referred to an orthopedic surgeon (Wagemakers 2010). The most frequent indication for arthroscopy is (suspected) meniscal injury, but other causes of persistent knee complaints may also necessitate arthroscopic surgical treatment. Technical advances in both diagnostic modalities and surgical possibilities as well as shifting insights on indications warrant the necessity for the guideline to address meniscal and non-meniscal injury. 7 clinical questions on non-meniscal related intra-articular pathology of the knee were formulated by a steering group of the Dutch Orthopedic Association (see Guideline recommendations below). This guideline aims to provide a uniform policy for the care of patients with knee disorders that could possibly be treated with an arthroscopic procedure. It is written for orthopedic surgeons, sports medicine specialists, physiotherapists, radiologists, and trauma surgeons who are involved in the care of patients with (acute) knee injuries. In addition, this guideline is intended to inform healthcare providers who are also involved in the care of these patients, including pediatricians, rehabilitation doctors, general practitioners, physician assistants, and nurse practitioners. Funding and potential conflicts of interest The guideline development was financially supported by the Dutch Orthopaedic Association (NOV), using governmental funding from the Quality Foundation of the Dutch Association

Acta Orthopaedica 2021; 92 (1): 80–83

of Medical Specialists in the Netherlands. The authors declare that there is no relevant conflict of interest.

Method See Dutch Guideline on Knee Arthroscopy, Part 1 (van Arkel et al. 2020).

Guideline recommendations according to 7 clinical questions, for literature reviews, see Supplementary data 1. What is the value of arthroscopy in patients with anterior knee pain (AKP), apexitis patella (Jumper’s knee) or patellar tendon tendinopathy?

Recommendation

• Do not perform arthroscopy in patients with AKP, because there is no difference in level of pain or function in patients with AKP after arthroscopy compared with nonoperative treatment. In patients with apexitis patella or patellar tendon tendinopathy, most patients do well with nonoperative treatment, but there was a positive effect of arthroscopic shaving compared with nonoperative treatment on level of pain. 2. Is there a role for arthroscopy of the knee in patient with osteoarthritis?

Recommendation

• Do not perform an arthroscopy in patients with osteoarthritis of the knee with or without debridement or lavage except if the knee is locked due to a sizable loose fragment in the knee. 3. Is arthroscopy indicated after patellar dislocation?

Recommendation

• Do not perform an arthroscopy in patients in the acute phase after a patellar dislocation; only consider an arthroscopy in case of osteochondral fracture. 4. Is arthroscopy indicated for treatment of (osteo) chondral fracture?

Recommendation

• Do not perform a diagnostic arthroscopy in patients with a suspected chondral lesion. Consider an arthroscopy in the treatment of an osteochondral fracture.

5. Is there a role for arthroscopy in the case of septic arthritis of the knee?

Recommendation

• An arthroscopic treatment of septic arthritis combined with systemic antibiotics provides a good treatment option. 6. Is arthroscopy indicated for ligamentous injury of the knee?

Recommendation

• Do not perform a diagnostic arthroscopy in patients with suspected ligamentous injury. 7. Is arthroscopy helpful in the treatment of tibial plateau fracture?

Recommendation

• Arthroscopy can have added value in the treatment of unicondylar tibial plateau fracture.

Discussion For each question, the scientific level of evidence on which the conclusion was based was graded using the 4 levels of evidence of the GRADE approach (Schünemann et al. 2013). RCTs start with a high level of evidence but must be downgraded if risk of bias (RoB) exists. The RoB tables for RCTs are based on the recommendation made by the Cochrane Collaboration (Higgins et al. 2011). The recommendations given are influenced by many considerations apart from the scientific evidence—such as patient preferences, availability of facilities, or organizational aspects. The recommendations for each question have been based on the scientific evidence, combined with the most important considerations, such as input from the guideline committee experts and feedback from the participating medical societies. The 1st question addresses the role of arthroscopy in patients with anterior knee pain (AKP). The term “anterior knee pain” is a descriptive term that covers all the pain surrounding the patellofemoral joint. It is therefore not a diagnosis in the narrow sense, but a symptom. The working group regarded pain and self-reported knee function as the 2 critical outcome measures. The evidence for this is low grade, because of the small sample size of the included RCT (Kettunen et al. 2007, 2012), which found no difference between the effect of arthroscopy and the effect of nonoperative therapy on level of pain and self-reported function in patients with patellofemoral pain syndrome. In patients with apexitis patella (jumper’s knee) or patellar tendinopathy there was low-grade evidence that there was a positive effect of arthroscopy compared with nonopera-

tive therapy on level of pain in patients with apexitis patella (jumper’s knee) or patellar tendinopathy (Willberg et al. 2011). Due to this low grade of evidence a reticent approach with regard to advising arthroscopy for patients with apexitis is advisable. The 2nd question regards the relevance of whether arthroscopy of the knee is of value in patients with osteoarthritis, because in older patients a degenerative meniscal lesion can be diagnosed in up to 50% in men in the age range 70–90 years old (Englund et al. 2008), and it can be difficult to differentiate between symptoms caused by the degenerative meniscal injury and symptoms due to early osteoarthritis of the knee. The guideline committee considered self-reported pain scores and self-reported knee function to be critical outcome measures for decision-making. and complications to be an important outcome measure. It was concluded with moderate-grade evidence that knee arthroscopy did not result in an extra reduction in pain scores or function in the short or long term when compared with nonoperative management in patients with osteoarthritis. The level of evidence was downgraded by 1 level due to serious risk of bias (4 out of 5 trials did not blind the participants, care providers, or outcome assessors) (Brignardello-Petersen et al. 2017). With low-grade evidence it was concluded that arthroscopy may have a small risk of venous thromboembolism and a very small risk of infection compared with nonoperative management in patients with degenerative knee disease. The level of evidence for the outcomes VTE and infections was downgraded for both by 2 levels due to serious risk of bias (used data were not collected for the study) and serious inconsistency (Brignardello-Petersen et al. 2017). To diagnose (early) osteoarthritis of the knee the working group advises making a standing full weight-bearing conventional radiograph in AP, lateral, and fixed flexion in patients over 50 years old. Additional imaging, such as MRI, is necessary only in the absence of osteoarthritis. Arthroscopy in osteoarthritis can be considered when repeated or persistent locking occurs, which is based on engagement of sizable loose fragments in the knee. The 3rd question addresses the treatment of knee complaints after patellar dislocation. After a patellar dislocation chondral fractures frequently occur; osteochondral fractures are seen less often (Sillanpaa et al. 2008). Chondral and osteochondral fragments can form loose bodies in the knee and arthroscopic removal of the loose bodies is propagated by some surgeons, sometimes combined with other arthroscopic or open operative procedures. This early surgical repair is now more common, without clear evidence to support this approach. We found no recent literature in the databases Medline (via OVID) and Embase (via Embase.com) between 2009 and January 17, 2018 that met the selection criteria. The guideline committee concluded (expert opinion) that only in patients with accompanying osteochondral fractures that can possibly be reattached is arthroscopic or open surgery indicated; in all other cases conservative treatment is the best first treatment option.

Acta Orthopaedica 2021; 92 (1): 80–83

The 4th question addresses the indication for arthroscopy in the case of knee complaints caused by chondral or osteochondral fractures in the acute phase. The working group regarded the development of osteoarthritis and response to treatment as critical outcome measures. We found no recent literature that met the selection criteria, the old guideline concerned 3 case series and a dissertation on fixation techniques that are not eligible according to the current selection criteria. The guideline committee advises, based on expert opinion, that arthroscopy is not indicated in case of chondral fracture, but can be considered in the case of refixation of an osteochondral fracture or removal of sizable fragments engaging persistent or recurrent locking. Concerning the clinical questions 5 to 7 we could not find new literature in our search that we could analyze according to the GRADE criteria; therefore the old text of the guideline Knee Arthroscopy 2010 was adopted. The 5th question addresses the treatment of ligamentous injury of the knee. Because the treatment of anterior cruciate ligament (ACL) injury is described in a separate guideline (Meuffels et al. 2012), ACL injury is excluded from this guideline. In former years the use of arthroscopy to address concomitant injury in case of hemarthrosis was widespread because the incidence of additional injury is high: in more than 50% of patients 1 or more ligamentous injuries are present. Based on expert opinion, the working group concludes that there is no role for diagnostic arthroscopy. Clinical examination, conventional radiographs (to exclude fractures), and MRI are the diagnostic tools of choice. The treatment of septic arthritis, addressed in the 6th question, is controversial, and differences persist between clinical specialties (orthopedic surgeons, rheumatologist, family physicians). We found no recent literature that met the selection criteria; all evidence on the treatment of septic arthritis is based on older, retrospective studies. Based on these studies and expert opinion, the working group concludes that arthroscopic debridement of septic arthritis seems to provide favorable results when combined with systemic antibiotics (Stutz et al. 2000, Wirtz et al. 2001). In addition, arthroscopic debridement in the acute phase seems to provide better results than recurrent needle aspiration (Ayral 2005). In the case of persistent septic arthritis arthroscopic debridement can be repeated (Stutz et al. 2000). In the case of tibial plateau fractures (question 7), arthroscopy was advocated in the literature in the 1990s (Jackson 1995). More recent literature focused on selected fractures (unicondylar fractures type II [split depression], and type III [isolated depression]). In these fracture types arthroscopic-assisted techniques resulted in fewer complications and faster rehabilitation (Ohdera et al. 2003, Musahl et al. 2009). Based on the limited literature and expert opinion the working group concludes that arthroscopic treatment can be indicated in the treatment of unicondylar tibial plateau fracture. Diagnostic arthroscopy is not indicated in the treatment of tibial plateau fracture.

Acta Orthopaedica 2021; 92 (1): 80–83

In recent years evidence has accumulated that questions the effectiveness and rationalization of arthroscopy for the treatment of AKP and osteoarthritis. This observation might have induced a more critical appraisal of other indications for arthroscopy, such as after a patellar dislocation, osteochondral fracture, bacterial arthritis, ligamentous injury of the knee, or tibial plateau fracture. This guideline provides evidence-based consideration of the current indications for arthroscopy. Supplementary data Literature reviews are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2020.1850081

Acta thanks Jon Drogset and Samuel Van de Velde for help with peer review of this study.

Ayral X. Arthroscopy and joint lavage. Best Pract Res Clin Rheumatol 2005; 19: 401-15. Brignardello-Petersen R, Guyatt G H, Buchbinder R, Poolman R W, Schandelmaier S, Chang Y, Sadeghirad B, Evaniew N, Vandvik P O. Knee arthroscopy versus conservative management in patients with degenerative knee disease: a systematic review. BMJ Open 2017; 7(5): e016114. doi: 10.1136/bmjopen-2017-016114. Englund M, Guermazi A, Gale D, Hunter J D, Aliabadi P, Clancy M, Felson D T. Incidental MRI findings on knee MRI in middle aged and elderly persons. N Engl J Med 2008; 359: 1108-15. Higgins J P, Altman D G, Gøtzsche P C, Jüni P, Moher D, Oxman A D, Savovic J, Schulz K F, Weeks L, Sterne J A, Cochrane Bias Methods Group. Cochrane Statistical Methods Group. Version 2; 2011. Jackson D W. Reconstructive knee surgery. New York: Raven Press; 1995. Kettunen J A, Harilainen A, Sandelin J, Schlenzka D, Hietaniemi K, Seitsalo S, Malmivaara A, Kujala U M. Knee arthroscopy and exercise versus exercise only for chronic patellofemoral pain syndrome: a randomized con-

trolled trial. BMC Med 2007; 5: 38. doi: 10.1186/1741-7015-5-38. Kettunen J A, Harilainen A, Sandelin J, Schlenzka D, Hietaniemi K, Seitsalo S, Malmivaara A, Kujala U M.. Knee arthroscopy and exercise versus exercise only for chronic patellofemoral pain syndrome: 5-year follow-up. Br J Sports Med 2012; 46(4): 243-6. Meuffels D, Poldervaart M T, Diercks R L, Fievez A, Patt T W, van der Hart C P, Hammacher E R, van der Meer, Goedhart E A, Lenssen AF, MullerPloeger SB, Pols M A, Saris D B F. Guideline on anterior cruciate ligament injury. Acta Orthop 2012; 83(4): 379-86. Musahl V, Tarkin I, Kobbe P, Tzioupis C, Siska P A, Pape H C. New trends and techniques in open reduction and internal fixation of fractures of the tibial plateau. J Bone Joint Surg Br 2009; 91: 426-33. Ohdera T, Tokunaga M, Hiroshima S, Yoshimoto E, Tokunaga J, Kobayashi A. Arthroscopic management of tibial plateau fractures: comparison with open reduction method. Arch Orthop Trauma Surg 2003; 123: 489-93. Schünemann H, Brożek J, Guyatt G, et al. GRADE handbook for grading quality of evidence and strength of recommendations. Updated October 2013. GRADE Working Group; 2013. Sillanpaa P J, Mäenpää H M, Mattila V M, Visuri T, Pihlajamäki H. Arthroscopic surgery for primary traumatic patellar dislocation: a prospective, nonrandomized study comparing patients treated with and without acute arthroscopic stabilization with a median 7-year follow-up. Am J Sports Med 2008; 36(12): 2301-9. doi: 10.1177/0363546508322894. Stutz G, Kuster M S, Kleinstuck F, Gachter A. Arthroscopic management of septic arthritis: stages of infection and results. Knee Surg Sports Traumatol Arthrosc 2000; 8: 270-4. van Arkel E R A, Koëter S, Rijk P C, van Tienen T G, Vincken P W J, Segers M J M, B van Essen B, van Melick N, Stegeman B H. Dutch Guideline on Knee Arthroscopy Part 1, the meniscus: a multidisciplinary review by the Dutch Orthopaedic Association. Acta Orthop 2020; 91(x): 1–7. Epub ahead of print. Wagemakers H P, Luijsterburg P A, Heintjes E M, Berger M Y, Verhaar J, Koes B W, Bierma-Zeinstra S M. Outcome of knee injuries in general practice: 1-year follow-up. Br J Gen Pract 2010; 60: 56-63. Willberg L, Sunding K, Forssblad M, Fahlström M, Alfredson H. Sclerosing polidocanol injections or arthroscopic shaving to treat patellar tendinopathy/jumper’s knee? A randomised controlled study. Br J Sports Med 2011; 45(5): 411-15. Wirtz D C, Marth M, Miltner O, Schneider U, Zilkens K W. Septic arthritis of the knee in adults: treatment by arthroscopy or arthrotomy. Int Orthop 2001; 25: 239-41.

Acta Orthopaedica 2021; 92 (1): 84–89

Which Oxford Knee Score level represents a satisfactory symptom state after undergoing a total knee replacement? Lina H INGELSRUD 1, Berend TERLUIN 2, Kirill GROMOV 1, Andrew PRICE 3, David BEARD 3, and Anders TROELSEN 1 1 Department

of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark; 2 Department of General Practice and Elderly Care Medicine, VU University Medical Center, Amsterdam, Netherlands; 3 Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, UK Correspondence: lina.holm.ingelsrud@regionh.dk Submitted 2020-06-29. Accepted 2020-09-20.

Background and purpose — Meaningful interpretation of postoperative Oxford Knee Score (OKS) levels is challenging. We established Patient Acceptable Symptoms State (PASS) and Treatment Failure (TF) values for the OKS in patients undergoing primary total knee replacement (TKR) in Denmark. Patients and methods — Data from patients undergoing primary TKR between February 2015 and January 2019 was extracted from the arthroplasty registry at the Copenhagen University Hospital, Hvidovre in Denmark. Data included 3, 12, and 24 months postoperative responses to the OKS and 2 anchor questions asking whether they considered their symptom state to be satisfactory, and if not, whether they considered the treatment to have failed. PASS and TF threshold values were calculated using the adjusted predictive modeling method. Non-parametric bootstrapping was used to derive 95% confidence intervals (CI). Results — Complete 3, 12, and 24 months postoperative data was obtained for 187 of 209 (89%), 884 of 915 (97%), and 575 of 586 (98%) patients, with median ages from 68 to 70 years (59 to 64% female). 72%, 77%, and 79% considered as having satisfactory symptoms, while 6%, 11%, and 11% considered the treatment to have failed, at 3, 12, and 24 months postoperatively, respectively. OKS PASS values (CI) were 27 (26–28), 30 (29–31), and 30 (29–31) at 3, 12, and 24 months postoperatively. TF values were 27 (26–28) and 27 (26–29) at 12 and 24 months postoperatively. Interpretation — The OKS PASS values can be used to guide the interpretation of TKR outcome and support quality assessment in institutional and national registries.

The patient perspective on outcome of total knee replacement (TKR) is captured with patient-reported outcome measures (PROMs) (Price et al. 2018). The Oxford Knee Score (OKS) measures the degree of knee pain and functional status of the knee on a scale ranging from 0 to 48 (worst to best score) (Dawson et al. 1998). Registry-based data suggest that 6-months postoperative OKS results are on average 36 points (NHS 2020). However, judging whether the outcome of surgery was successful or not can be challenging, because it is not clear which symptom level patients consider to be satisfactory. The Patient Acceptable Symptom State (PASS) concept was defined by Tubach et al. (2005) as the score on a PROM above which patients consider themselves well. The contrary concept, Treatment Failure (TF), was introduced for patients undergoing ACL reconstruction, to define patients who consider their symptom levels unsatisfactory to a degree that they find the treatment has failed (Ingelsrud et al. 2015). Suggested satisfaction thresholds for the OKS range from 30 to 38 points after knee replacement (Judge et al. 2012, Keurentjes et al. 2014, Petersen et al. 2017). The time-points evaluated in these studies were either 6 months or shorter/ longer than 3 years postoperatively. A dichotomized visual analogue scale (VAS) or a numeric rating scale (NRS) was used as anchor question to measure patients’ satisfaction. However, having the patients’ explicit judgements of whether they have reached a satisfactory symptom state or not after surgery is necessary to derive credible PASS values. Moreover, interpretation characteristics of PROMs are context dependent (Tubach et al. 2007), which highlights the relevance of evaluateing the time-dependency of PASS values for the OKS after TKR. We therefore defined PASS and TF values for the OKS at 3 months, and 1 and 2 years after a TKR.

Acta Orthopaedica 2021; 92 (1): 84–89

Patients and methods Data from patients undergoing primary TKR due to primary or secondary OA between February 2015 and January 2019 were extracted from the arthroplasty registry at the Copenhagen University Hospital, Hvidovre in Denmark. Registry data were predominantly collected electronically, with patients responding to an electronic questionnaire during their pre-surgical visit to the hospital. Links to follow-up questionnaires were sent by email at 3, 12, and 24 months postoperatively. Paper versions were sent to patients without an e-mail address and to all patients not responding to an electronic reminder. Patients’ BMI was calculated using self-reported preoperative height and weight while the ASA score and Kellgren and Lawrence classification of radiographic OA was reported to the registry by the operating surgeon. Questionnaire data for this study included the OKS and 2 additional anchor questions that were responded to postoperatively. The first question asked, “Taking into account all the activities you have during your daily life, your level of pain, and also your functional impairment, do you consider that your current state is satisfactory?” (yes/no). If the patients answered that they did not have a satisfactory symptom state, they were asked the second question: “Would you consider your current state as being so unsatisfactory that you think the treatment has failed?” (yes/no). Administration of these anchor questions in the registry was initiated at different points in time for the 3 follow-up time-points, which is why the numbers of eligible patients differ across the 3 follow-up time-points. Statistics Patient characteristics are reported as median (interquartile range [IQR]) for non-normally distributed continuous variables and number (proportion) for categorical variables. Postoperative OKS distributions across anchor response categories were investigated with boxplots. The association between the postoperative OKS score and anchor responses were investigated with Spearman’s correlation. R version 3.4.1 (https:// www.r-project.org/) was used for analyses. PASS and TF threshold values for the OKS were calculated using the predictive modeling method (Terluin et al. 2015), which was originally developed to estimate minimal important change thresholds, but can also be used to estimate thresholds in cross-sectional data. The method is based on logistic regression, with the PASS and TF anchors as dependent variables and postoperative OKS as the independent variable. The threshold is the OKS score that corresponds to a likelihood ratio of 1. With a likelihood ratio of 1, the posttest odds of having a satisfactory symptom state are the same as the pretest odds of having a satisfactory symptom state. The predictive modeling method identifies thresholds that are close to optimal ROC cut-offs with greater precision than

ROC analysis (Terluin et al. 2015). However, both thresholds tend to be biased if the proportions of the dependent variable are unequally distributed, resulting in overestimation of the threshold if the proportion having a satisfactory state is greater than 50% or underestimation if the proportion is smaller than 50%. We therefore applied an adjustment to the threshold for unequal proportions of patients, with the equation proposed by Terluin et al. (2017): PASSadjusted = PASSpred – (0.090 + 0.103*Cor)*SD*log-odds(sat)

In this equation, Cor is the point biserial correlation between the postoperative OKS and the anchor, SD is the SD of the postoperative OKS, and log-odds(sat) is the natural logarithm of (proportion with satisfactory symptom state/[1 – proportion with satisfactory symptom state]). Non-parametric bootstrapping (n = 1,000) was used to derive 95% confidence intervals (CI) and is reported as 0.025–0.975 quantiles. Subgroup analyses were performed to investigate the effect of preoperative severity level on the PASS and TF values. We calculated stratified PASS and TF values for high-severity (lower OKS score) and low-severity (higher OKS score) subgroups that were split by the median preoperative OKS score. Furthermore, to generate CIs around the differences in PASS and TF values for the high- and low-severity subgroups, we median split 1,000 non-parametric bootstrap samples. Severity subgroup PASS and TF values were considered to be statistically different if the 95% CI of the differences did not include 0. For comparison with previous studies, we calculated cutoffs with the receiver operating characteristics (ROC) statistics. The cut-off was determined according to the Youden principle as the point yielding the largest sum of sensitivity and specificity (Youden 1950). We expected the Youden threshold to be close to the predictive modeling threshold and higher than the adjusted predictive modeling threshold if the proportion satisfactory state was greater than 50%. We also applied an 80% specificity rule, since other studies have suggested that thresholds determined as the point with the highest degree of sensitivity and at least 80% specificity improve comparability across studies (Aletaha et al. 2009). Ethics, funding, and potential conflicts of interest The local arthroplasty registry was approved by the Danish Data Protection Agency (Journal number HVH-2012-048). In Denmark, approval from the ethical committee is not required for register-based studies involving only questionnaire data. The study was fully funded by the Department of Orthopaedic Surgery at the hospital. The authors declare no potential conflicts of interest in relation to this study.

Results Of the eligible patients in the registry, complete data was obtained from 187 (54%), 884 (56%), and 575 (52%) patients

Acta Orthopaedica 2021; 92 (1): 84–89

3-months cohort Surgery between April 2018 and January 2019 n = 346

1-year cohort Surgery between February 2015 and January 2018 n = 1,584

Excluded 3-month form missing n = 137 Patients with completed 3-month form n = 209

Excluded 1-year form missing n = 669 Patients with completed 1-year form n = 915

Excluded 2-year form missing n = 529 Patients with completed 2-year form n = 586

Excluded (n = 31): – missing PASS anchor, 25 – missing OKS, 6

Excluded Missing anchor n = 22 Patients with complete data for primary analyses n = 187 (54%)

2-year cohort Surgery between February 2015 and January 2017 n = 1,115

Patients with complete data for primary analyses n = 884 (56%)

Excluded (n = 11): – missing PASS anchor, 8 – missing OKS, 3

Patients with complete data for primary analyses n = 575 (52%)

Figure 1. Flow chart of patient enrollment.

Table 1. Patient demographics and preoperative characteristics. Values are median (2.5–97.5% quantile range) and count (percentage) Factor

3 months n = 187

1 year n = 884

2 years n = 575

Age 70 (60–75) 69 (61–74) 68 (61–74) Female sex 110 (59) 545 (62) 365 (64) BMI 28 (25–32) 29 (26–33) 29 (26–33) missing data – 6 7 ASA 1 22 (12) 115 (13) 78 (14) 2 132 (71) 623 (70) 398 (69) 3 33 (18) 144 (16) 98 (17) 4 – 2 (0) 1 (0) Kellgren and Lawrence grade 1 1 (1) 5 (1) 4 (1) 2 12 (6) 91 (10) 61 (11) 3 70 (37) 377 (43) 251 (44) 4 104 (56) 411 (47) 259 (45) Oxford Knee Score 24 (18–29) 22 (17–27) 22 (17–27) missing data 33 251 158 EQ5D index 0.72 0.68 0.66 (0.63–0.72) (0.56–0.72) (0.56–0.72) missing data 32 254 160

at 3 months, 1 year, and 2 years’ follow-up (Figure 1). At surgery, the median age was 68–70 years and 59–64% were female (Table 1). At 3 months postoperatively, 72% considered themselves to have satisfactory symptoms, while 6% considered their symptom state as being so unsatisfactory that they considered the treatment to have failed. The proportions of patients who were satisfied with their symptom level were 77% and 79% at 1 and 2 years postoperatively, while 11% considered the treatment to have failed (Table 2).

Table 2. Proportions of patients achieving a satisfactory symptom level, considering treatment failure, or neither at 3 months, 1 year, and 2 years after surgery. Values are count (percentage) Factor

3 months n = 187

Satisfactory symptom level 135 (72) Neither satisfactory symptoms nor treatment failure 39 (21) Treatment failure 12 (6) Treatment failure anchor missing 1 (1)

1 year n = 884

2 years n = 575

684 (77)

456 (79)

99 (11) 93 (11) 8 (1)

52 (9) 63 (11) 4 (1)

Postoperative OKS were in general higher for patients considering their symptom level to be satisfactory, in comparison with those considering the treatment to have failed or neither (Figure 2). Spearman’s correlation between the postoperative OKS and the classification “satisfactory symptoms,” “neither satisfactory nor treatment failure,” or “treatment failure” was 0.52 at 3 months, 0.59 at 12 months, and 0.58 at 24 months. When adjusting the predictive modeling method PASS values for the high proportion having satisfactory symptoms, OKS PASS values were 27 at 3 months, 30 at 12 months, and 30 at 24 months after TKR. TF values were 27 points at 12 months, and 27 at 24 months postoperatively (Table 3). At 3 months, the absolute number of patients considering the treatment to have failed (n = 12) was too low to calculate TF values. Subgroup analyses showed that mean bootstrapped PASS and TF values were 4–6 points higher in the low-severity subgroups in comparison with the high-severity subgroups at 12 and 24 months, respectively (Table 4). Further, the adjusted predictive modeling method yielded smaller PASS and TF values than the ROC method and had smaller CIs for all follow-up time-points (Supplementary data).

Acta Orthopaedica 2021; 92 (1): 84–89

Table 4. Baseline dependency of Patient Acceptable Symptom State (PASS) and Treatment Failure (TF) cut-off values calculated with the adjusted predictive modeling method for the Oxford Knee Score at 12 and 24 months after total knee replacement

Oxford Knee Score 48

Mean threshold Mean threshold value a (95% CI b) value a (95% CI b) Mean difference High-severity Low-severity in threshold Factor subgroup c subgroup c value (95% CI)

PASS 12 months 29 (28–30) 33 (32–34) –4 (–5 to –2) 24 months 27 (26–29) 34 (32–35) –6 (–9 to –4) TF 12 months 24 (22–26) 31 (29–32) –6 (–9 to –3) 24 months 25 (23–27) 30 (28–32) –6 (–9 to –3)

3 months

Satisfactory symptoms

1 year

2 years

Neither satisfactory nor treatment failure

a PASS

Treatment failure

Figure 2. Oxford Knee Score (OKS) distributions at 3, 12, and 24 months postoperatively for patients with satisfactory symptoms, considering the treatment to have failed, or neither. Red bars present the median, the box the interquartile range (IQR), and the whiskers the maximum and minimum scores within 1.5 * IQR from the box. Outliers are values beyond 1.5 * IQR from the box.

Table 3. Patient Acceptable Symptom State (PASS) and treatment failure (TF) cut-off values calculated with the adjusted predictive modeling method for the Oxford Knee Score at 3, 12, and 24 months after a total knee replacement n PASS value (95% CI) a

Follow-up 3 months 12 months 24 months

187 884 575

n TF value (95% CI) a

27 (26–28) 30 (29–31) 876 30 (29–31) 571

– 27 (25–28) 27 (26–29)

a 95%

confidence intervals (CI) are the 0.025–0.975 quantiles of the 1,000 bootstrap threshold values.

Discussion We estimated OKS PASS values at 3 months, 1, and 2 years after a total knee replacement. A PASS value can be interpreted as the threshold between what the average patient would consider a satisfactory state and what they would consider a nonsatisfactory state. Our finding that PASS values increased by approximately 3 points from 3 months to 1 and 2 years postoperatively suggests that patients accept lower levels of knee functional status in the shorter term than they do in the longer term after surgery. Furthermore, we found that PASS values varied with the preoperative functional level. The PASS values we derived lie in the lower range of proposed cut-offs between 30 and 38 points for the OKS that previously have been suggested to reflect satisfaction with knee replacement (Judge et al. 2012, Keurentjes et al. 2014, Petersen et al. 2017, Hamilton et al. 2018). The variations in values from previous studies may be caused by differences

and TF values are presented as the mean of 1,000 bootstrap threshold values. b 95% confidence intervals (CI) are the 0.025–0.975 quantiles of the 1,000 bootstrap threshold values. c High- and low-severity subgroups were generated by splitting each bootstrap sample by the median preoperative OKS score.

in statistical methods, anchor questions used, and follow-up time-points. Our PASS value of 27 at 3 months postoperatively is lower than the values of 30 and 35 that were previously found at 6 months postoperatively (Judge et al. 2012, Petersen et al. 2017). Furthermore, our PASS values of 30 at 12 and 24 months are lower than any of the other previously published values using the same or longer follow-up time-periods (Keurentjes et al. 2014, Hamilton et al. 2018). Increasing PASS values with increasing follow-up time-points were previously suggested by Keurentjes et al. (2014), who found OKS PASS values of 34 for patients < 3 years and 38 for patients > = 3 years postoperatively. In contrast, we found an increase in PASS values from 3 to 12 months, but almost similar PASS values at 12 and 24 months postoperatively. Differences in statistical approaches hinder direct comparison of PASS values. We have shown in this study that different methods yield different results. The PASS values we derived with other methods were 2–5 points higher than those derived using the primary analysis method. The overestimation of PASS values calculated with the ROC method is associated with the proportions of patients with satisfactory symptom levels exceeding 50% (Terluin et al. 2020). Advantages of using the adjusted predictive modeling method include that we are able to overcome the issue of biased PASS values in the direction of the largest group (Terluin et al. 2017). Furthermore, smaller CIs reflect greater precision of this method in comparison with the more traditional ROC method. Another particular strength of our study is the dichotomous anchor question used. All previous studies have anchored the postoperative OKS on dichotomized 0–100 VAS or 0–10 NRS measuring satisfaction with the outcome. These anchors were dichotomized using thresholds of ≥ 5 on the NRS (range 0 to 10) (Keurentjes et al. 2014), and ≥ 50 or ≥ 70 on the

VAS scales (range 0 to 100) (Judge et al. 2012, Petersen et al. 2017). A shortcoming of previous studies is therefore that the dichotomization thresholds are seemingly arbitrarily chosen. Judge et al. (2012) tested the effect of varying the cut-off on the satisfaction VAS and concluded that the PASS values varied by only 3 points with the choice of cut-off on the VAS scale. However, considering that the concept PASS reflects the threshold level of symptoms that patients consider satisfactory, the experts to judge are the patients themselves. An advantage of our PASS anchor question is therefore that the classification of being in a satisfactory symptom state or not relies totally on the patients’ own judgement. The presented TF thresholds were only 3 points lower than the PASS thresholds. The small difference implies that the undecided area between considering treatment success and treatment failure is not prominent and, in turn, suggests that treatment outcome can be dichotomized into successful and not successful outcome using the PASS values. Depending on the question patients are asked and the definition of treatment success, the proportion of patients with successful outcome after knee replacement varies (Hamilton et al. 2018). A possible limitation of the PASS anchor item we used is that it does not specifically ask about the current state of the operated knee. However, we believe that patients actually do reflect on their knee status when responding to this anchor item, since this question is asked subsequently after knee-specific PROMs. A qualitative study approach would be necessary to clarify the anchor items’ degree of content validity and to further investigate the use of PASS and TF anchor questions in reflecting postoperative treatment success. We found that PASS and TF values were baseline dependent. Patients with lesser symptoms preoperatively required fewer symptoms to consider being in a satisfactory symptom state, in comparison with patients with worse preoperative symptom levels. In other words, patients with higher OKS scores preoperatively required higher OKS scores at the postoperative follow-up time points to consider their symptom levels satisfactory. The differences in 12- and 24-month PASS values were 3 and 6 points, between patients with higher and lower preoperative severity levels. Baseline dependency in satisfaction thresholds was also previously found in Judge and colleagues’ (2012) study of the OKS, and also the Oxford Hip Score in patients undergoing total hip replacement (Arden et al. 2011). That PASS values vary with preoperative severity levels underlines the importance of considering whether preoperative patient characteristics are comparable, when applying PASS values to interpret OKS results from other data sources. PASS values can be used as responder criteria. In clinical trials, such responder analyses, presenting the proportions of patients with postoperative OKS exceeding the PASS values in each trial arm, may increase the interpretation of treatment success, as was previously exemplified in patients undergoing ACL reconstruction (Roos et al. 2019). Likewise, the inter-

Acta Orthopaedica 2021; 92 (1): 84–89

pretation of registry-based PROM collection or cohort study results can be improved in a similar manner, and may contribute to quality assessment of TKR based on institutional or national registries. From a clinical perspective, PASS values serve as group-based reference values and should not be misinterpreted as representing every individual patient. These values that represent thresholds for the average patient may serve as comparability values from a reference population in the postoperative management, rather than fixed goals of treatment for every patient. Furthermore, PASS values and proportions of responders may be helpful to patients and clinicians in their shared-decision dialogue when deciding whether to undergo surgery and to leverage expectations. Importantly, the finding that PASS values are baseline dependent stresses the fact that application in other populations and cohorts must be done with careful consideration of the patient characteristics’ comparability to our study cohort. The study data, collected in a single public hospital in Denmark, possibly limits the generalizability of the PASS values. Furthermore, we had complete data from only just above 50% of the patients who underwent surgery in the study period, which may introduce selection bias. However, considering the hospitals’ uptake area, covering both rural and urban geographical areas, and patient characteristics from our sample mirroring the characteristics from the nationwide Danish Knee Arthroplasty Register with regard to age and sex distribution, this supports the representativeness of our study population in a Danish context (Odgaard et al. 2019). Whether the suggested PASS values are applicable in other countries and cultures needs to be established using data from large-scale international registries. Furthermore, we included only patients undergoing a TKR, while unicompartmental knee replacement is increasingly used in Denmark (Henkel et al. 2019). PASS values are considered context dependent, and specific to the patient population and intervention under study (Tubach et al. 2007). Whether PASS values differ with regard to surgical strategy for knee OA remains unanswered. In conclusion, in patients undergoing TKR, the symptom level patients experience as satisfactory is higher in the shorter term than in the longer term postoperatively. Furthermore, across all investigated follow-up time-points, thresholds for considering their symptom levels as satisfactory are higher for patients who have lower symptom levels preoperatively. The established PASS values can be used to guide the interpretation of TKR outcome when measured with the OKS. Future studies should investigate the external validity of the derived PASS values. Supplementary data PASS and TF values calculated with Receiver Operating Characteristics analyses are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/ 17453674.2020.1832304

Acta Orthopaedica 2021; 92 (1): 84–89

Conception and study design: LHI, BT, AT. Collection and assembly of data: LHI. Analysis: LHI. Interpretation of the data: LHI, BT, KG, AP, DB, AT. Drafting of the manuscript: LHI. Critical revision and final approval of the article: LHI, KG, BT, AP, DB, AT. The authors would like to thank the patients for responding to the questionnaires and the staff at the Department of Orthopaedic Surgery for handling the database on a daily basis. Acta thanks Tor Kjetil Nerhus and Ola Rolfson for help with peer review of this study.

Aletaha D, Funovits J, Ward M M, Smolen J S, Kvien T K. Perception of improvement in patients with rheumatoid arthritis varies with disease activity levels at baseline. Arthritis Care Res 2009; 61(3): 313-20. Arden N K, Kiran A, Judge A, Biant L C, Javaid M K, Murray D W, Carr A J, Cooper C, Field R E. What is a good patient reported outcome after total hip replacement? Osteoarthritis Cartilage 2011; 19(2): 155-62. Dawson J, Fitzpatrick R, Murray D, Carr A. Questionnaire on the perceptions of patients about total knee replacement. J Bone Joint Surg 1998; 80(1): 63-9. Hamilton D F, Loth F L, MacDonald D J, Giesinger K, Patton J T, Simpson A H, Howie C R, Giesinger J M. Treatment Success following joint arthroplasty: defining thresholds for the Oxford Hip and Knee Scores. J Arthroplasty 2018; 33(8): 2392-7. 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(4): 354-9. Ingelsrud L H, Granan L-P, Terwee C B, Engebretsen L, Roos E M. Proportion of patients reporting acceptable symptoms or treatment failure and their associated KOOS values at 6 to 24 months after anterior cruciate ligament reconstruction: a study from the Norwegian Knee Ligament Registry. Am J Sports Med 2015; 43(8): 1902-7. Judge A, Arden N K, Kiran A, Price A, Javaid M K, Beard D, Murray D, Field R E. Interpretation of patient-reported outcomes for hip and knee replacement surgery: identification of thresholds associated with satisfaction with surgery. Bone Joint J 2012; 94-B(3): 412-18.

Keurentjes J C, Tol F R Van, Fiocco M, So-Osman C, Onstenk R, KoopmanVan Gemert A W M M, Pöll R G, Nelissen R G H H. Patient acceptable symptom states after total hip or knee replacement at mid-term follow-up: thresholds of the Oxford Hip and Knee Scores. Bone Joint Res 2014; 3(1): 7-13. NHS Digital [Internet]. NHS PROMs programme [cited 2020 May 15]. Available from: https://digital.nhs.uk/data-and-information/publications/ statistical/patient-reported-outcome-measures-proms/finalised-hip--kneereplacements-april-2018---march-2019 Odgaard A, Emmeluth C, Schrøder H, Østgaard S E, Madsen F, Troelsen A, Iversen P, van Bugge L, Hjelm A. Danish Knee Arthroplasty Register Annual Report; 2019. Petersen C L, Kjærsgaard J B, Kjærgaard N, Jensen M U, Laursen M B. Thresholds for Oxford Knee Score after total knee replacement surgery: a novel approach to post-operative evaluation. J Orthop Surg Res 2017; 12(1): 89. 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. Roos E M, Boyle E, Frobell R B, Lohmander S, Ingelsrud L H. It is good to feel better, but better to feel good: Whether a patient finds treatment successful’ or not depends on the questions researchers ask. Br J Sports Med 2019;53(23): 1474-8. Terluin B, Eekhout I, Terwee C B, de Vet H C. Minimal important change (MIC) based on a predictive modeling approach was more precise than MIC based on ROC analysis. J Clin Epidemiol 2015; 68: 1388–96. Terluin B, Eekhout I, Terwee C B. The anchor-based minimal important change, based on receiver operating characteristic analysis or predictive modeling, may need to be adjusted for the proportion of improved patients. J Clin Epidemiol 2017; 83: 90-100. Terluin B, Griffiths P, van der Wouden JC, Ingelsrud L H, Terwee C B. Unlike ROC analysis, a new IRT method identified clinical thresholds unbiased by disease prevalence. J Clin Epidemiol 2020; 124:118-25. Tubach F, Ravaud P, Baron G, Falissard B, Logeart I, Bellamy N, Bombardier C, Felson D, Hochberg M, van der Heijde D, Dougados M. Evaluation of clinically relevant states in patient reported outcomes in knee and hip osteoarthritis: the patient acceptable symptom state. Ann Rheum Dis 2005; 64(1): 34-7. Tubach F, Ravaud P, Beaton D, Boers M, Bombardier C, Felson D T, Van Der Heijde D, Wells G, Dougados M. Minimal clinically important improvement and patient acceptable symptom state for subjective outcome measures in rheumatic disorders. J Rheumatol 2007; 34(5): 1188-93. Youden W J. Index for rating diagnostic tests. Cancer 1950; 3(1): 32-5.

Acta Orthopaedica 2021; 92 (1): 90–95

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 Tero IRMOLA 1, Ville PONKILAINEN 1, Keijo T MÄKELÄ 2,3, Otto ROBERTSSON 4, Annette W-DAHL 4, Ove FURNES 5,6, Anne M FENSTAD 5, Alma B PEDERSEN 7, Henrik M SCHRØDER 8, Antti ESKELINEN 1, and Mika J NIEMELÄINEN 1 1 Coxa

Hospital for Joint Replacement, and Faculty of Medicine and Health Technologies, University of Tampere, Tampere, Finland; 2 Finnish Arthroplasty Register, National Institute for Health and Welfare, Helsinki, Finland; 3 Department of Orthopaedics and Traumatology, Turku University Hospital, Turku, and University of Turku, Turku, Finland; 4 The Swedish Knee Arthroplasty Register, Department of Orthopedics, Skane University Hospital, Lund, Sweden. Department of Clinical Sciences, Orthopedics, Lund University, Sweden; 5 The Norwegian Arthroplasty Register, Department of Orthopaedic Surgery, Haukeland University Hospital, Bergen, Norway; 6 Department of Clinical Medicine, University of Bergen, Bergen, Norway; 7 Department of Clinical Epidemiology, Aarhus University Hospital. Denmark and Danish Knee Arthroplasty Registry; 8 Department of Orthopaedic Surgery, Naestved Hospital, Denmark Correspondence: tero.irmola@fimnet.fi Submitted 2020-06-16. Accepted 2020-09-23.

Background and purpose — The population of the Nordic countries is aging and the number of elderly patients undergoing total knee arthroplasty (TKA) is also expected to increase. Reliable fixation methods are essential to avoid revisions. We compared the survival of different TKA fixation concepts with cemented fixation as the gold standard. Patients and methods — We used data from the Nordic Arthroplasty Register Association (NARA) database of 265,877 unconstrained TKAs performed for patients aged ≥ 65 years with primary knee osteoarthritis between 2000 and 2016. Kaplan–Meier (KM) survival analysis with 95% confidence intervals (CI) and the Cox multiple-regression model were used to compare the revision risk of the fixation methods. Results — Cemented fixation was used in 243,166 cases, uncemented in 8,000, hybrid (uncemented femur with cemented tibia) in 14,248, and inverse hybrid (cemented femur with uncemented tibia) fixation in 463 cases. The 10-year KM survivorship (95% CI) of cemented TKAs was 96% (96−97), uncemented 94% (94−95), hybrid 96% (96−96), and inverse hybrid 96% (94−99), respectively. Uncemented TKA was associated with increased risk of revision compared with the cemented TKA; the adjusted hazard ratio was 1.3 (95% CI 1.1−1.4). Interpretation — Cemented, hybrid, and inverse hybrid TKAs showed 10-year survival rates exceeding 95%. Uncemented fixation was associated with an increased risk of revision in comparison with cemented fixation. As both hybrid and inverse hybrid fixation were used in only a limited number of TKAs, indicating possibility of selection bias in their favor, cemented TKA still remains the gold standard, as it works reliably in the hands of many.

Cemented fixation of total knee arthroplasty (TKA) has been regarded as the gold standard. Uncemented fixation potentially offers some advantages, such as shorter operative times and elimination of possible complications of using bone cement (Yayac et al. 2020), but also some obvious disadvantages such as higher implant costs and increased risk of revision as shown in register-based studies (Nugent et al. 2019, NJR 2019). In a study from the New Zealand Joint Registry (NZJR) uncemented TKAs had similar patient-reported outcomes but higher revision rates when compared with hybrid and cemented TKAs (Nugent et al. 2019). Survivorship between unconstrained cemented and hybrid TKAs did not differ either in the National Joint Registry for England, Wales and North Ireland (NJR) or in the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR), while uncemented TKAs showed a slightly higher risk of revision (AOANJRR 2019, NJR 2019). Conversely, a Norwegian register-based study reported better survivorship of hybrid than cemented TKAs at 11 years (Petursson et al. 2015). Further, a recent meta-analysis found no differences in either implant survivorship or clinical outcomes between uncemented and cemented fixation (Zhou et al. 2018). We demonstrated recently that cemented TKA should be considered the gold standard in patients younger than 65 years of age even if promising survivals were detected in hybrid TKAs (Niemeläinen et al. 2020). In the current study, we assessed whether the traditional assumption of cemented TKA as gold standard still holds true also in elderly patients. We analyzed survivorships of different fixation methods in unconstrained TKA in patients aged 65 years and older based on the Nordic Arthroplasty Register Association (NARA) database.

Acta Orthopaedica 2021; 92 (1): 90–95

Patients and methods Study design and setting We conducted this population-based cohort study using prospectively collected data available from the NARA knee dataset, which contains data from 4 Nordic countries (the Swedish Knee Arthroplasty Register [SKAR], the Danish Knee Arthroplasty Register [DKR], the Norwegian Arthroplasty Register [NAR], and the Finnish Arthroplasty Register [FAR]). The dataset includes only variables that all countries can deliver, currently 20 variables for knee arthroplasty (Mäkelä et al. 2019). All Swedish, Norwegian, Danish, and Finnish citizens are assigned a unique civil registration number, permitting unambiguous linkage between knee registries and other medical databases in each country. All registers have used individual-based registration of operations. Selection and transformation of the particular datasets and de-identification of the operations were performed within each national register. The anonymous data was then merged into a common dataset. Data were treated with full confidentiality, according to the rules of the respective countries. The quality of data in the Nordic registers is high, including both 100% coverage and following completeness for primary TKA: SKAR 97%, DKR 97%, NAR 97%, FAR 96% and for revision TKA: SKAR > 95% (estimate, OR, personal communication), DKR 94%, NAR 91%, FAR 80% (NAR 2018, DKR 2019, FAR 2019, SKAR 2019). The study follows the RECORD and STROBE guidelines.

All knee arthroplasties in NARA database n = 550,570 Exluded (n = 284,693): – age < 65 years, 184,460 – other than primary OA, 25,855 – posterior stabilized implant, 22,243 – unicompartmental arthroplasty, 17,842 – degree of constrain unknown, 10,906 – performed before 2000, 7,215 – operated in 2017, 6,693 – revisions, 4,202 – other types of implant, 3,508 – type of implant unknown, 639 – patellofemoral arthroplasty, 479 – fixation unknown, 322 – fully stabilized, 216 – patella treatment unknown, 94 – other partial knees, 6 – missing data, 13 Included primary TKAs (n = 265,877): – cemented, 243,166 – uncemented, 8,000 – inverse hybrid, 463 – hybrids, 14,248

Figure 1. Flow chart of the study.

Outcome The primary outcome measure was time to 1st revision, defined as removal, addition, or exchange of at least 1 of the components for any reason. Thus, the 1st revision of the index knee was the endpoint.

to assess implant survival probability with 95% confidence intervals (CI) of the TKA fixation. Groups with less than 40 knees at risk are not presented in the tables. We used multivariable Cox proportional hazard regression analysis to compare the survival between different fixation types adjusting for confounding variables (hazard ratios). Fixation type was used as the primary dependent variable and all analyses were adjusted for potential confounders such as sex, country, patellar resurfacing, and age. Age was included in the model as continuous variable whereas the others were categorical. Sensitivity analyses were performed for subgroups based on age (65−75 years of age and older than 75), sex, and NexGen TKA model. Because of the obvious risk for case-mix bias, an additional sensitivity analysis was conducted for patients operated on with NexGen TKAs. We examined violations of proportional hazard (PH) assumptions by evaluating the correlation of scaled Schoenfeld residuals with time. In addition, the correlation of scaled Schoenfeld residuals and log–log survival curves were inspected visually to evaluate the PH assumptions. Violations of PH assumptions were handled by constructing a time-stratified model (Zhang et al. 2018). Thus, the time-stratification was conducted for a unique set of variables based on the Schoenfeld residuals and log–log survival curves of the particular model. Correlations of Schoenfeld residuals with time were repeatedly evaluated to ensure that the non-proportionality was fixed. Statistical analyses were performed using R 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria), with packages survival, survminer, and tidyverse.

Statistics Descriptive statistics were presented as numbers (%), as mean (SD), or as median with interquartile range (IQR) based on the distribution. Kaplan–Meier (KM) survival analysis was used

Ethics, funding, and potential conflicts of interest Formal approval for the study was granted by the ethical approval process of each national register. Permission numbers from each country are: the Danish Data protection agency

Study population We included all uni- or bilateral unconstrained primary TKAs that had been implanted in patients aged 65 years or older for primary OA 2000–2016 (Figure 1). Bilaterals were both same-day bilateral and staged bilateral. Previous reports have shown that the effect of including bilateral cases in studies of hip and knee prosthesis survival is negligible (Robertsson and Ranstam 2003, Lie et al. 2004). The fixation of TKAs was divided into 4 groups: (1) cemented, (2) uncemented, (3) hybrid (uncemented femur with cemented tibia), and (4) inverse hybrid (cemented femur with uncemented tibia). The numbers of included implants and reasons for exclusions are shown in a flow chart (Figure 1).

Acta Orthopaedica 2021; 92 (1): 90–95

Table 1. Demographic data. Values are count (%) unless otherwise specified Inverse Factor Uncemented hybrid

Hybrid

Cemented

TKAs 8,000 (3.0) 463 (0.2) 14,248 (5.4) Mean age (SD) 73 (5.5) 73 (5.6) 74 (5.7) Men, % 43 39 37 TKAs per country Finland 1,204 (1.5) 340 (0.4) 307 (0.4) Norway 2,017 (6.2) 15 (<0.1) 4,793 (15) Sweden 1,955 (1.8) 22 (<0.1) 36 (<0.1) Denmark 2,824 (6.4) 86 (0.2) 9,112 (21)

243,166 (92) 74 (5.7) 36 80,195 (98) 25,900 (79) 104,863 (98) 32,208 (73)

Table 5. Unadjusted Kaplan–Meier (KM) 10- and 15-year survival rates with 95% confidence intervals (CI) for uncemented, inverse hybrid, hybrid, and cemented TKA At 10 years Type of fixation

No. of No. of No. at knees revisions risk

Uncemented 8,000 Inverse Hybrid 463 Hybrid 14,248 Cemented 243,166

321 11 423 6,767

1,271 40 2,078 50,845

Figure 2. Number of TKAs with different fixation methods.

At 15 years

KM survival rate (%)(CI)

No. at risk

94 (94–95) 96 (94–99) 96 (96–96) 96 (96–97)

201 – 254 6,594

(1-16-02-54-17), Denmark, the National Institute of Health and Welfare (Dnro THL/1743/.5.05.00/2014), Finland, the Norwegian Data Inspectorate (ref 24.1.2017: 16/01622-3/ CDG), Norway, and the Ethics Board of Lund University (LU20-02), Sweden. This work was supported by the competitive research funds of Pirkanmaa Hospital District, Tampere, Finland, representing governmental funding. The authors have no conflicts of interests to declare.

Results Cemented fixation was used in 92% of all TKAs, although there was some variation between the countries: Sweden (98%), Finland (98%), Norway (79%) to Denmark (73%). Hybrid fixation was used in 5% of all cases: Denmark (21%), Norway (15%), Finland (0.4%), and Sweden in 36 cases (0.0%) (Table 1). The total number of TKAs performed annually increased notably (102%) between the years 2000 (n = 8,733) and 2009 (n = 17,668) but remained relatively stable after that. The use of hybrid fixation increased by 104% between the years 2009 (n = 798) and 2016 (n = 1,631) (Figure 2). The TKA models varied between countries without a common trend and the most commonly used TKA models are shown in the Table 2 (see Supplementary data). NexGen, PFC, and Triathlon were the most commonly used models within the fixation concepts (Table 3, Supplementary data). The patella was resurfaced in 56,596 TKAs (22%) and uncemented patellar buttons were

KM survival rate (%)(CI) 93 (92–94) – 94 (93–95) 96 (95–96)

Figure 3. Unadjusted Kaplan–Meier cumulative risk of revision by fixation type in patients > 65 years of age (stopping rule, n = 40).

used in only 371 (0.1%) of the TKAs. There were differences between the countries when considering proportion of patellar resurfacing, from Norway (3%), Sweden (6%), and Finland (18%), to Denmark (79%) (Table 4, Supplementary data). In the subgroup of NexGen TKAs, the patella was resurfaced in 12,160 (18%) TKAs, and an uncemented patellar button was used in only 14 knees. Of the 265,877 TKAs, altogether 7,522 underwent revision after median follow-up time of 5.5 years. The median followup time was 5.8 years for cemented, 4.7 years for uncemented, 4.7 years for inverse hybrid, and 4.2 years for hybrid TKA. Between the fixation groups, there were marginal differences in the proportion of men, ranging from 36% in the cemented to 43% in the uncemented group (Table 1). KM-based 10-year survival rates were: cemented 96%, inverse hybrid 96%, hybrid 96%, and uncemented 94%. Due to low numbers, the 15-year survival rate of inverse hybrid

Acta Orthopaedica 2021; 92 (1): 90–95

Table 6. Multivariate Cox regression with adjusted hazard ratios (aHR) and 95% confidence intervals (CI)

Table 9. Multivariate Cox regression of patients in NexGen subgroup with adjusted hazard ratios (aHR) and 95% confidence intervals (CI)

Age ≥ 65 years Age 65–75 years Age > 75 years Type of fixation aHR a (CI) aHR b (CI) aHR c (CI)

Type of fixation

aHR a (CI)

Uncemented Inverse hybrid Hybrid Cemented

Uncemented Inv hybrid Hybrid Cemented

1.1 (0.8–1.5) 0.9 (0.4–1.8) 1.3 (1.1–1.6) 1.0 Reference

1.3 (1.1–1.4) 0.8 (0.4–1.4) 1.0 (0.9–1.1) 1.0 Reference

1.1 (0.9–1.5) 0.9 (0.4–1.8) 1.5 (1.1–1.7) 1.0 Reference

1.4 (1.1–1.7) 1.1 (0.4–2.9) 0.9 (0.7–1.0) 1.0 Reference

Hazard ratios adjusted by age, sex, patellar resurfacing, and nation —age, sex, and nation as time-dependent coefficients divided into time intervals of: a 0.1, 0.3, 0.5, 1.5, 3.5, and 6 years. b 0.1, 0.5, 1.0, 3.5, and 6 years. c 0.1, 1.0, 3.5, 8, and 10 years.

a Adjusted

by age, sex, patellar resurfacing, and nation age, sex, and nation as time-dependent coefficients — divided into time intervals of 0.2, 1.5, 3.5, and 6 years.

Table 8. Unadjusted Kaplan–Meier 7- and 10-year survival rates with 95% confidence intervals (CI) are presented for uncemented, inverse hybrid, hybrid, and cemented TKA in the NexGen subgroup At 7 years

At 10 years

Type of fixation

No. of No. of No. at knees revisions risk

KM survival rate (%)(CI)

No. at risk

KM survival rate (%)(CI)

Uncemented Inverse Hybrid Hybrid Cemented

1,976 379 3,887 61,376

96 (94–97) 97 (94–99) 95 (94–96) 98 (98–98)

– – – 6,982

– – – 97 (97–97)

61 8 133 1,191

164 85 275 16,858

was not reliable, yet the 15-year survival rates for other fixation methods were: cemented 96%, hybrid 94%, uncemented 93% (Table 5, Figure 3). Uncemented fixation evinced an increased risk of revision compared with the cemented TKA in the adjusted Cox regression analysis (HR 1.3) (Table 6). We found no differences in the risk of revision between the hybrid or inverse hybrid and the cemented TKAs. The additional Cox regression analyses were conducted for 2 different age groups: 65−75 years of age and older than 75 years of age (Table 6). In patients aged 65−75 years, the risk of revision with hybrid TKAs was increased in comparison with the cemented reference group (HR 1.5) (Table 6). In patients older than 75 years, there was an increased risk of revision with uncemented fixation (HR 1.4) (Table 6). Most of the TKAs in the inverse hybrid group were NexGen (82%) (Table 3, Supplementary data)). Cemented fixation was used in 91% of the NexGen TKAs (Table 7, Supplementary data). Because of the obvious risk for case-mix bias, an additional sensitivity analysis was conducted for patients operated on with NexGen TKAs. In this sensitivity analysis, 7-year survival rates of different fixations were in descending order: cemented 98%, inverse hybrid 97%, uncemented 96%, and hybrid 95%. At 10 years, survival rate was available only for cemented NexGen TKAs (98%) (Table 8). An increased risk of revision was found for hybrid NexGen TKAs as compared

with the cemented NexGen TKAs (HR 1.3). The risk of revision for uncemented and inverse hybrid TKAs was comparable to cemented TKAs (Table 9).

Discussion This multinational register-based study revealed that cemented fixation was used in the vast majority of the TKAs (91%) in the Nordic countries among elderly patients. Cemented, hybrid, and inverse hybrid TKAs all evinced acceptable 10-year survival rates exceeding 95% in patients aged 65 years and older. Uncemented fixation was associated with increased risk of revision compared with cemented fixation. The population is aging, and patients aged 65 years or more still contribute to most of the total incidence of knee arthroplasty (Niemeläinen et al. 2017), meaning that most TKAs will be performed in elderly patients. In our study the vast majority of TKAs were cemented. Cemented fixation is also the most commonly performed type of knee replacement in arthroplasty registers (AOANJRR 2019, NJR 2019, NZJR 2019). Conversely, the NZJR showed that the usage of uncemented fixation had increased during the last 3 years (NZJR 2019). In our study, the small increase in the use of hybrid fixation was mainly seen in Denmark and Norway between 2009 and 2016.

All fixation methods evinced acceptable 10-year survival rates in patients aged 65 years and older. Cemented and hybrid TKAs still showed good survivorship at 15 years. Uncemented TKAs had the lowest survivorship. These findings are in line with the majority of the previous literature. Based on the AOANJRR, the cumulative 15-year revision rate of minimally stabilized TKA was lower with cemented fixation compared with uncemented and lowest with hybrid fixation (AOANJRR 2019). The NJR (2019) and the NZJR (2019) reported the same trend among 65–74-year-old patients. Moreover, the revision rate of patients older than 75 years with cemented TKAs was lower than with uncemented and hybrid TKAs in the NJR and slightly higher with hybrid compared with cemented and uncemented TKAs in the NZJR. A higher risk of revision in cemented TKAs compared with hybrid TKAs at 11 years was noted in a Norwegian registry-based study, but after exclusion of a high-volume hospital the difference was no longer statistically significant. Of most obvious concern is that the reported result involved only 1 prosthesis brand (Petursson et al. 2015). The risk of revision with hybrid TKAs was increased in comparison with cemented TKAs in the age group 65−75 years in our study. This is contrary to annual reports from the AOANJRR (2019), NJR (2019) and NZJR (2019), where hybrid TKAs are not worse than cemented. The risk of revision was increased with uncemented fixation in patients aged above 75 years. In this age group uncemented and hybrid fixation showed slightly increased risk of revision in the NJR annual report of 2019. In New Zealand there was no difference in the revision risk between cemented and uncemented fixation, but this was slightly higher with hybrid fixation (NZJR 2019). Some studies report contradictory results on the association of fixation on TKA outcomes. In a randomized controlled trial comparing uncemented and cemented fixation in TKA (PFC), the authors reported no differences in revision rates and survival between the cemented and uncemented TKAs with mean follow-up of 9 years (Baker et al. 2007). When interpreting the results of that study, it is important to keep in mind that it was a single-surgeon series from a clinic with the same prosthesis design. Similarly, Zhou et al. (2018) found no differences between uncemented and cemented TKAs in implant survivorship and clinical outcomes in their systematic review and meta-analysis consisting of 409 uncemented and 403 cemented TKAs. There was a wide range in the average length of follow-up among the trials and population characteristics like mean age were different between the trials, which may have affected the results. As affirmed earlier, NexGen covered the majority (82%) of the TKAs in the inverse hybrid group, and 87% of these NexGen TKAs had been used with TM tibial components, which are known to have good results (Niemeläinen et al. 2014). We tried to grasp the obvious possibility of selection bias by conducting a sensitivity analysis including only NexGen TKAs (Tables 8 and 9). In that analysis we found

Acta Orthopaedica 2021; 92 (1): 90–95

similar mid-term survival rates or Cox-adjusted revision risks between inverse hybrid and cemented NexGen TKAs. Further, hybrid fixation showed an increased risk for revision in this NexGen subgroup. Similar increased risk for revision with hybrid fixation was also seen in the TKAs in the age group 65–75 years. Thus, the more expensive uncemented or hybrid/ inverse hybrid versions did not provide the older age group with any advantage over cemented fixation in the 10-year follow-up of NexGen TKAs. We acknowledge certain strengths and limitations in our study. The major strength is the unique collaboration of 4 national registers in the creation of a multinational dataset comprising a high number of non-selected TKAs, reflecting the real-world outcomes of TKA. There are also some limitations in our study. First, these results concern TKA concepts, not single components and their fixation. It must be noted that there were clearly fewer patients in the alternative fixation groups as compared with the cemented reference group. This may have caused some selection bias, and in this case it might have favored concepts other than cemented fixation. Further, especially inverse hybrid fixation, but also hybrid fixation to some extent, had another advantage over cemented fixation in our study setting. In the inverse hybrid group, 82% were NexGen TKAs and more than 80% approximately of the inverse hybrid NexGen TKAs utilized TM monoblock tibial components (an estimate from national registers’ data), which are known to have good longterm results (Kim et al. 2012, Niemeläinen et al. 2014, Robertsson et al. 2020). What is more, in Finland NexGen inverse hybrid TKAs (with TM tibial component) have been performed in only 3 hospitals, one of which is a high-volume specialized center (Niemeläinen et al. 2014). In the hybrid group, 3 TKA designs with a good track record (PFC, NexGen, Profix) comprised 75% of all TKAs. Thus, the result of inverse hybrids should be interpreted with caution. Second, revision as the only outcome of interest also has several disadvantages. Revision is a surgeon and patient-dependent outcome and decision regarding which knee symptoms require revision is mostly subjective and may vary. Of course, the decision to reoperate is shared and the patient has to agree. With the exception of prosthetic joint infection and periprosthetic fracture, other indications for knee revision are less clear and less objective. Also, differences in proportions of patellar resurfacing in each country may have some role, especially on a patient’s risk of secondary patellar resurfacing. Third, completeness of revision TKA is lower in FAR than in other Nordic arthroplasty registries. In Finland, 20% of revisions are missing from the FAR database when compared with the National Patient Discharge Register. Finally, the median follow-up time of 5.5 years is rather short when evaluating the revision rate for TKA. In conclusion, cemented TKA deserves the status of gold standard in TKA irrespective of the patients’ age. The advantages of cemented TKA are even more pronounced with increasing patient age. Even though hybrid/inverse hybrid versions of the well-performing contemporary TKA designs

Acta Orthopaedica 2021; 92 (1): 90–95

provided older patients with a good mid-term outcome, these results do not support systematic use of these more expensive components in TKA for older patients. Thus, for patients aged 65 years and older, cemented TKA is still the method of choice. Supplementary data Tables 2–4 and 7 are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453 674.2020.1837422 Design of the work: MN, AE. Drafting the work: VP, AE, MN, TI. Final approval of the version to be published: TI, VP, KM, OR, AW-D, OF, AF, AP, HS, AE, MN. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: TI, VP, KM, OR, AW-D, OF, AF, AP, HS, AE, MN. Acta thanks Peter Devane nad Michael Richard Whitehouse for help with peer review of this study.

AOANJRR. Australian Orthopaedic Association National Joint Replacement Registry, hip, knee & shoulder arthroplasty 20th annual report; 2019, https://aoanjrr.sahmri.com Baker P N, Khaw F M, Kirk L M G, Esler C N A, Gregg P J. A randomised controlled trial of cemented versus cementless press-fit condylar total knee replacement: 15-year survival analysis. J Bone Joint Surg Br 2007; 89(12): 1608. DKR. Dansk knealloplastikregister, årsrapport; 2019, https://sundhed.dk FAR. Finnins arthroplasty register; 2019, https://www.thl.fi/far (accessed April 24, 2020). Kim Y, Park J, Kim J. High-flexion total knee arthroplasty: survivorship and prevalence of osteolysis: results after a minimum of ten years of follow-up. J Bone Joint Surg Am 2012; 94(15): 1378-84. Lie S A, Engesæter L B, Havelin L I, Gjessing H K, Vollset S E. Dependency issues in survival analyses of 55 782 primary hip replacements from 47 355 patients. Stat Med 2004; 23(20): 3227-40. Mäkelä K T, Furnes O, Hallan G, Fenstad A M, Rolfson O, Kärrholm J, Rogmark C, Pedersen A B, Robertsson O, W-Dahl A, Eskelinen A, Schrøder H M, Äärimaa V, Rasmussen J V, Salomonsson B, Hole R, Overgaard S. The benefits of collaboration: the Nordic Arthroplasty Register Association. EFORT Open Rev 2019; 4(6): 391-400.

NAR. Norwegian arthroplasty register, annual report; 2018, http://Nrlweb. ihelse.net. Niemeläinen M, Skyttä E T, Remes V, Mäkelä K, Eskelinen A. Total knee arthroplasty with an uncemented trabecular metal tibial component: a registry-based analysis. J Arthroplasty 2014; 29(1): 57-60. Niemeläinen M J, Mäkelä K T, Robertsson O, W-Dahl A, Furnes O, Fenstad A M, Pedersen A B, Schrøder H M, Huhtala H, Eskelinen A. Different incidences of knee arthroplasty in the Nordic countries. Acta Orthop. 2017; 88(2): 173-8. 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; 91(2): 184-90. NJR. National Joint Registry for England, Wales, Northern Ireland and the Isle of Man, 16th annual report; 2019, https://reports.njrcentre.org.uk 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. NZJR. The New Zealand Joint Registry, annual 20 years report; 2019, https:// nzoa.org.nz Petursson G, Fenstad A M, Havelin L I, Gøthesen Ø, Lygre S H L, Röhrl S M, Furnes O. Better survival of hybrid total knee arthroplasty compared to cemented arthroplasty. Acta Orthop 2015; 86(6): 714-20. Robertsson O, Ranstam J. No bias of ignored bilaterality when analys ing the revision risk of knee prostheses: analysis of a population based sample of 44,590 patients with 55,298 knee prostheses from the national Swedish Knee Arthroplasty Register. BMC Musculoskelet Disord 2003; 4(1): 1. Robertsson O, Sundberg M, Sezgin E A, Lidgren L, W-Dahl A. Higher risk of loosening for a four-pegged TKA tibial baseplate than for a stemmed one: a register-based study. Clin Orthop Relat Res 2020; 478(1): 58-65. SKAR. Swedish Knee Arthroplasty Register, annual report; 2019, https:// myknee.se Yayac M, Harrer S, Hozack W J, Parvizi J, Courtney P M. The use of cementless components does not significantly increase procedural costs in total knee arthroplasty. J Arthroplasty 2020; 35(2): 407-12. Zhang Z, Reinikainen J, Adeleke K A, Pieterse M E, Groothuis-Oudshoorn C G M. Time-varying covariates and coefficients in cox regression models. Ann Transl Med 2018; 6(7): 121. Zhou K, Zhou Z, Yu H, Li J, Wang H, Pei F. No difference in implant survivorship and clinical outcomes between full-cementless and full-cemented fixation in primary total knee arthroplasty: a systematic review and metaanalysis. Int J Surg 2018; 53: 312-19.

Acta Orthopaedica 2021; 92 (1): 96–100

Bariatric surgery prior to total knee arthroplasty is not associated with lower risk of revision: a register-based study of 441 patients Perna IGHANI ARANI 1,2, Per WRETENBERG 1,2, Johan OTTOSSON 2–4, Otto ROBERTSSON 5,6, and Annette W-DAHL 5,6 1 Department 3 Department

of Orthopedics, Orebro University Hospital, 2 Faculty of Medicine and Health, School of Medical Sciences, Örebro University, Örebro, of Surgery, Örebro University Hospital, 4 Scandinavian Obesity Surgery Registry, Örebro, 5 Lund University, Faculty of Medicine, Department of Clinical Sciences Lund, Orthopedics, Lund, 6 The Swedish Knee Arthroplasty Register, Lund, Sweden Correspondence: perna.arani@gmail.com Submitted 2020-04-29. Accepted 2020-10-06.

Background and purpose — Obesity is a considerable medical challenge in society. We investigated the risk of revision for any reasons and for infection in patients having total knee arthroplasty (TKA) for osteoarthritis (OA) within 2 years after bariatric surgery (BS) and compared them with TKAs without BS. Patients and methods — We used the Scandinavian Obesity Surgery Registry (SOReg) and the Swedish Knee Arthroplasty Register (SKAR) to identify patients operated on in 2009–2019 with BS who had had primary TKA for OA within 2 years after the BS (BS group) and compared them with TKAs without prior BS (noBS group). We determined adjusted hazard ratio (HR) for the BS group and noBS group using Cox proportional hazard regression for revision due to any reasons and for infection. Adjustments were made for sex, age groups, and BMI categories preoperatively. Results — 441 patients were included in the BS group. The risk of revision for infection was higher for the BS group with HR 2.2 (95% CI 1.1–4.7) adjusting for BMI before the TKA, while the risk of revision for any reasons was not statistically significant different for the BS group with HR 1.3 (CI 0.9–2.1). Corresponding figures when adjusting for BMI before the BS were HR 0.9 (CI 0.4–2) and HR 1.2 (CI 0.7–2). Interpretation — Our findings did not indicate that BS prior to TKA was associated with lower risk of revision.

Overweight and obesity in society is a considerable medical challenge. It has been estimated that between 19% and 31% of the population in Europe is obese (BMI ≥ 30) and the proportion continues to increase (Krzysztoszek et al. 2019). Obesity is associated with a number of medical conditions such as type 2 diabetes, obstructive sleep apnea, cardiovascular diseases, dyslipidemia, fatty liver disease, and several cancers, contributing to a decline in both quality of life and life expectancy (Ng et al. 2014). It has been shown that bariatric surgery (BS) is an effective method of achieving significant long-term weight loss for obese patients in comparison with nonsurgical interventions (Sjostrom 2013). BS results in significant improvement in multiple metabolic and cardiovascular diseases, reduction in new cancer development, and reduces risk for premature death in patients with BMI > 35 (Sjostrom et al. 2007, Adams et al. 2012, Sundbom et al. 2017). A significant improvement in health-related quality of life is also seen (Raoof et al. 2015, Kolotkin et al. 2018). With the growing number of obese patients undergoing primary TKA, revision surgery among patients with obesity has increased over the past decades (Odum et al. 2016). Obesity has been described to negatively influence complications and mortality after TKA (Kerkhoffs et al. 2012, Electricwala et al. 2017, Tohidi et al. 2018). Several studies have shown an increased overall risk of revision after TKA in obese patients without looking at the specific reason for revision (Roche et al. 2018, Tohidi et al. 2018, Boyce et al. 2019). However, a recent study showed that obesity was associated with overall risk of revision and revision for infection, but not for revision for reasons other than infection (Sezgin et al. 2020).

Acta Orthopaedica 2021; 92 (1): 96–100

SOreg 2007–2019 67,706 patients

plasty after the BS. Those with missing BMI before the TKA were excluded. In the cohort of TKAs Excluded (n = 3,404): – TKAs in SOreg, 1,992 without BS (noBS group) – missing BMI pre TKA, 1,412 we excluded those with Eligible TKAs for OA missing BMI before the n = 128,266 TKA as well as the TKAs included in the SOReg (Figure 1). Excluded Excluded in BMI before TKA group: in BMI before BS group: We used 2 separate ways Outside age and BMI interval Outside age and BMI interval n = 32,318 n = 102,559 to compare the risk of revision in the BS group and BMI before BS group BMI before TKA group the noBS group. In the 1st with TKA for OA with TKA for OA included in the analysis included in the analysis comparison we adjusted (noBS-TKA group) (noBS-TKA group) n = 25,707 n = 95,948 for the BMI prior to the BS in the BS group (BMI before BS), and in the 2nd comparison we adjusted for their BMI prior to the TKA (BMI before TKA) and compared them with the noBS TKA with comparable selections. SKAR 2009–2019 TKA for OA n = 131,670

Excluded (n = 65,794): – no valid personal ID, 108 – no match in SKAR, 65,686 Eligible patients/knee arthroplasties n = 1,912/2,603 Excluded Knee arthroplasty before BS n = 827 Remaining knee arthroplasties n = 1,776 Excluded (n = 1,335): – no TKA, 125 – TKA before 2009, 3 – bilateral surgery, second knee, 366 – TKA not within 2 years, 798 – no OA, 41 – missing BMI pre TKA, 2 TKA for OA included in the analysis (BS-TKA group) n = 441

Flow-chart of the study.

There is no national guideline for BS in Sweden. Most hospitals follow the international accepted guidelines (Hubbard and Hall 1991). BS that may reduce BMI and subsequently comorbidities prior to a TKA could then be a beneficial measure to reduce the risk of revision. Nevertheless, this remains unclear considering previous studies (Inacio et al. 2014, Martin et al. 2015, Lee et al. 2018, McLawhorn et al. 2018) as well as in a recent systemic review (Gu et al. 2019). Therefore, we evaluated the risk of revision for any reasons and for infection in OA patients having a TKA within 2 years after BS and compared them with TKAs without BS using data from the Scandinavian Obesity Surgery Registry (SOReg) and the Swedish Knee Arthroplasty Register (SKAR).

Patients and methods Patients who underwent BS in 2007–2019 were identified using the SOReg. From the SKAR we identified patients having primary TKA for OA in 2009–2019 within 2 years after the BS. From the SOReg we included all patients with primary gastric bypass and sleeve gastrectomy. The SOReg was established in 2007 and the SKAR in 1975. Both registers have high completeness and the correctness of data is validated with hospital records and found to be high (Hedenbro et al. 2015, SKAR 2019). The unique personal identification number that all Swedish citizens have was used when linking the datasets. In patients having BS before the TKA (BS group) we excluded the 2nd knee if both knees had a primary knee arthro-

TKA selections In order to adjust for BMI before the BS in the BS group we selected the noBS group within the same BMI interval (BMI 32–63). For the analysis adjusting for BMI prior to the TKA in the BS group, we selected noBSs within the same BMI interval (range 16.9–50). In both analyses, we selected noBS TKAs within the same age interval as the BS TKA group had prior to the TKA surgery (range 39–76 years) (Figure 1). BMI, age, and sex were obtained from the SKAR and SOReg. BMI was classified into 5 categories: underweight (BMI < 18.5), normal weight (BMI 18.5–24.9), overweight (BMI 25–29.9), obese (BMI 30–39.9), and morbidly obese (BMI ≥ 40). The ASA physical status classification was available from the SKAR. We decided not to adjust for the ASA classification in the primary analysis since BMI interferes with comorbidity (Owens et al. 1978) and BMI has also become a factor in deciding the ASA classification. The outcome measures were revision for any reason and revision for suspected or verified infection. Revision is defined as a new operation in a previously resurfaced knee in which 1 or more of the components are exchanged, removed, or added (including arthrodesis and amputation) (SKAR 2019). Statistics Adjusted hazard ratio (HR) was obtained for the BS groups and the noBS groups using Cox proportional hazards regressions for 2 endpoints: revision for any reasons over the whole period and revision for suspected or verified infection within 2 years after the primary TKA. Using BMI before the BS, adjustments were made for prior BS or no BS, sex, age groups (< 45, 45–54, 55–64, 65–74, and 75–84), and BMI category (30–39.9 and ≥ 40). Using BMI before the TKA adjustments

Acta Orthopaedica 2021; 92 (1): 96–100

Table 1. Patient characteristics Restricting for BMI before BS BS group noBS group Factor n = 441 n = 25,707 Women a 336 (76) 16,452 (64) Age b 55 (37–74) 65 (39–76) BMI b 43 (32–63) 35 (32–50) a Count (%) b Mean (range)

Restricting for BMI prior to TKA BS group noBS group n = 441 n = 95,948 336 (76) 57 (39–76) 35 (32–50)

52,771 (55) 66 (39–76) 29 (17–50)

were made for prior BS or no BS, sex, age groups (< 45, 45–54, 55–64, 65–74 and 75–84), and BMI category (< 18.5, 18.5–24.9, 25–29.9, 30–39.9, and ≥ 40). Sensitivity analyses were made including the ASA classification in the Cox regression analysis together with sex, age, and BMI when using BMI before the TKA with the 2 endpoints revision for any reasons and revision for suspected or verified infection. Results are reported as HRs with 95% confidence intervals (CIs). Statistical analyses were carried out using Stata version 15 (StataCorp, College Station, TX, USA). Ethics, funding, and potential conflicts of interest The study was approved by the Swedish Ethical Review authority (2017/466). A research grant for the project was received from Region Örebro län. The authors have no conflicts of interest to report.

Results 441 patients operated on with bariatric surgery prior to the TKA for OA (BS group) were included. When adjusting for BMI and age before the BS in the BS TKA group, the comparison was with 25,707 TKAs for OA not subject to BS but within the same age and BMI range as the patients in the BS group. When adjusting for the BMI and age prior to the TKA the BS group was compared with 95,948 TKAs for OA within the same age and BMI range. The mean time between BS and TKA was 1.1 year (0.1– 2). The mean BMI decreased from 43 (32–63) to 31 (17–50) between the BS and the TKA in the BS group. The mean follow-up for infection was 23 months in the BS group and 22 months in the noBS group and the follow-up for revision for any reason was 65 months in the BS group and 62 months in the noBS group. Adjusting for BMI prior to the BS In the BS group the mean age was 55 years (37–74) and 76% were women. The mean BMI was 43 (32–63) before the BS (Table 1). The noBS group had a mean age of 65 years (39–76) and 64% were women. The mean BMI was 35 (32–50).

Table 2. Adjusted hazard ratios (aHR) for the risk of revision for infection and any reasons for revision when using BMI before the BS Revision for infection Revision for any reasons Variable aHR (95% CI) aHR (95% CI) noBS group Reference Reference BS group 0.9 (0.4–2.0) 1.2 (0.8–2.0)

Table 3. Adjusted hazard ratios (aHR) for the risk of revision for infection and any reason for revision when using BMI before the TKA Variable

Revision for infection aHR (95% CI)

noBS group Reference BS group 2.2 (1.1–4.7)

Revision for any reasons aHR (95% CI) Reference 1.3 (0.9–2.1)

Using Cox proportional hazard regression, considering revision for any reasons as well as revisions for suspected or verified infection, we found similar risk, HR 1.2 (CI 0.7–2) and HR 0.9 (CI 0.4–2) respectively, for the BS TKA group as compared with the noBS TKA group without statistical significance (Table 2). Adjusting for BMI prior to the TKA The mean age of the BS group was 57 years (39–76) and 76% were women. The mean BMI was 31 (17–50) before the TKA (Table 1). In the noBS group the mean age was 65.8 years (39–76) and consisted of 55% women. The mean BMI was 29 (17–50). Using Cox proportional hazard regression when analyzing revision for suspected or verified infection, we found a higher risk for the BS TKA group with HR 2.2 (CI 1.0–4.7) (Table 2). The risk of revision for any reasons was also higher for the BS TKA group with HR 1.3 (CI 0.9–2.1,), however without statistical significance (Table 3). The sensitivity analysis including ASA classification in the Cox regression analysis changed the results marginally: for the risk of revision for infection from HR 2.2 (CI 1.1–4.7) in the BS group to HR 2.0 (CI 1.0–4.3) with the noBS group as reference. Corresponding figures for the risk of revision for any reason was HR 1.3 (CI 0.9–2.1) and 1.3 (CI 0.8–2.0).

Discussion When analyzing the risk of revision for all reasons, we found no benefit for obese patients of having BS before the TKA when compared with TKAs without BS. This was independent of whether we adjusted for their BMI before the BS or before the TKA. However, the risk of revision due to suspected or verified infection was higher in patients having a BS before the TKA when using their BMI before the TKA, while the

Acta Orthopaedica 2021; 92 (1): 96–100

higher risk was not statistically significant when using the BMI before the BS in the BS group. Our observational study based on national registers suffers from the limitations of this type of design. However, the data from SKAR and SOReg is prospectively collected, and both are national registers with a high completeness and quality (SOReg 2018, SKAR 2019). Another limitation is the relatively low number of 441 BS TKA cases. Nevertheless, this is the largest study of this type that has been performed. When adjusting for BMI, for the BS group we used their BMI prior to the BS as well as prior to the TKA. The rationale was to make the comparison as if the BMI of the patients in the noBS group had been comparable to that of the BS group both before and after their weight loss. In order to make the groups more comparable we only included noBSs within the same BMI range as the BSs had on both occasions. Additionally, we only included the same age interval as in the BS TKA group, which on average was 13 years younger than the general TKA population in Sweden (SKAR 2019). We set the cut-off time between BS and TKA as within 2 years to minimize the risk of other confounding health factors affecting outcome. In addition, time to TKA after BS has been suggested to be between 6 months and 2 years (Schwarzkopf et al. 2018). The mean age in patients having BS in Sweden is just over 40 years and almost 80% are women (SOReg 2018). This may be reflected in the BS group were the patients were younger and the proportion of women was higher than in the general knee arthroplasty population in Sweden at the time of TKA. The Swedish bariatric surgery cohort has somewhat lower BMI (5–6 BMI units) and a lower percentage of comorbidity compared with most North American cohorts. Compared with some European cohorts the Swedish cohort have similar BMI and lower percentage of comorbidity (Poelemeijer et al. 2018). We did not include information on comorbidity in the primary Cox regression analysis although it could affect the risk of revision. The SKAR includes the ASA classification while the SOReg includes the obesity surgery mortality risk score (OS-MRS). However, the patients undergoing elective surgery are assessed by both the surgeon and an anesthesiologist with optimization prior to surgery. Earlier studies have demonstrated the relationship between BS prior to TKA regarding complications and outcome (Kulkarni et al. 2011, Inacio et al. 2014, Martin et al. 2015, Werner et al. 2015, Lee et al. 2018, McLawhorn et al. 2018). The findings in previous studies investigating the risk of revision in patients having BS before the TKA as compared with patients not having BS are inconsistent. Martin et al. (2015) evaluated the BMI both before the BS and before the TKA in BS patients. They showed an increased risk of both reoperation and revision in 91 patients having a BS prior to the TKA (4 months to 11 years between the surgeries) as com-

pared with a matched cohort of patients without BS. Infection was the most common reason for reoperation and revision. Also Lee et al. (2018) found an increased risk of revision in 70 patients having a BS before TKA within 2 years compared with a cohort of patients with chronic metabolic conditions, but found no increased risk of revision due to infection. Inacio et al. (2014) found no difference in risk of revision in 62 BS patients operated on with a TKA within 2 years after the BS as compared with 2,616 TKA patients without BS. McLawhorn et al. (2018) used propensity scores to match morbidly obese patients receiving and not receiving BS prior to TKA (2,636 TKA patients within each group) and found that BS did not reduce the risk of revision surgery. However, the time between the BS and the TKA was not presented. These studies included different selections of patients, BMI (i.e., BMI at the time for BS, BMI at the time for TKA, or both), selection of control group, time between the BS and TKA, and some have matched BS patients with TKA patients without BS for various variables. This makes comparison of these studies with ours difficult. Despite the medical benefits in patients undergoing BS (Sjostrom et al. 2007, Adams et al. 2012, Sundbom et al. 2017), nutritional deficiencies may occur. Protein malnutrition is extremely uncommon with gastric bypass and sleeve gastrectomy but micronutrient deficiencies may be present (Xanthakos 2009, Bal et al. 2012). Also, bone demineralization and increased risk of fractures may occur after gastric bypass (Raoof et al. 2016, Axelsson et al. 2018). Our cohort includes gastric bypass and sleeve gastrectomy, the two most common types of BS today. Since the number of sleeve gastrectomies is low, no subgroup analysis was done. To our knowledge, this is the 1st study to include a whole nation with close to complete registration of all TKA and BS procedures. In Sweden, there are no national guidelines for knee arthroplasty surgery. However, a national project with the intention to stop prosthetic-related infection after hip and knee arthroplasty surgery (the PRISS project) resulted in recommendations (2013), among others, for optimizing patients before surgery. The recommendations include, despite the absence of evidence, a suggestion that patients with BMI > 40 should be referred for help with preoperative weight loss (LÖF n.d.). While the proportion of obese individuals has increased in the Swedish population during the last decade (SCB n.d.) the proportion of patients with BMI ≥ 35 having a knee arthroplasty has decreased slightly (SKAR 2019). This may be a result of surgeons being more restrictive in operating on obese patients and/or recommending that obese patients consider BS or other measures to decrease their preoperative weight before the TKA surgery. In conclusion, our findings indicate that having BS prior to TKA does not decrease the risk of revision. This information may be valuable when advising and scheduling obese OA patients before TKA surgery.

100

PW, OR, AWD, and JO conceived and designed the study. OR and AWD performed the analysis. PIA and AWD wrote the initial draft. All authors contributed to the interpretation of the data and to revision of the manuscript. The authors would like to thank all the contact surgeons and associated staff at the hospitals in Sweden for their dedicated registration work over the years. Acta thanks Reinoud W Brouwer and Maria Carolina Inacio for help with peer review of this study.

Adams T D, Davidson L E, Litwin S E, Kolotkin R L, LaMonte M J, Pendleton R C, et al. Health benefits of gastric bypass surgery after 6 years. JAMA 2012; 308(11): 1122-31. Axelsson K F, Werling M, Eliasson B, Szabo E, Naslund I, Wedel H, et al. Fracture risk after gastric bypass surgery: a retrospective cohort study. J Bone Miner Res 2018; 33(12): 2122-31. Bal B S, Finelli F C, Shope T R, Koch T R. Nutritional deficiencies after bariatric surgery. Nat Rev Endocrinol 2012; 8(9): 544-56. Boyce L, Prasad A, Barrett M, Dawson-Bowling S, Millington S, Hanna S A, Achan P. The outcomes of total knee arthroplasty in morbidly obese patients: a systematic review of the literature. Arch Orthop Trauma Surg 2019; 139(4): 553-60. Electricwala A J, Jethanandani R G, Narkbunnam R, Huddleston J I 3rd, Maloney W J, Goodman S B, Amanatullah D F. Elevated body mass index is associated with early total knee revision for infection. J Arthroplasty 2017; 32(1): 252-5. Gu A, Cohen J S, Malahias M A, Lee D, Sculco P K, McLawhorn A S. The effect of bariatric surgery prior to lower-extremity total joint arthroplasty: a systematic feview. HSS J 2019; 15(2): 190-200. Hedenbro J L, Naslund E, Boman L, Lundegardh G, Bylund A, Ekelund M, et al. Formation of the Scandinavian Obesity Surgery Registry, SOReg. Obes Surg 2015; 25(10): 1893-1900. Hubbard V S, Hall W H. Gastrointestinal surgery for severe obesity. Obes Surg 1991; 1(3): 257-65. Inacio M C, Paxton E W, Fisher D, Li R A, Barber T C, Singh J A. Bariatric surgery prior to total joint arthroplasty may not provide dramatic improvements in post-arthroplasty surgical outcomes. J Arthroplasty 2014; 29(7): 1359-64. Kerkhoffs G M, Servien E, Dunn W, Dahm D, Bramer J A, Haverkamp D. The influence of obesity on the complication rate and outcome of total knee arthroplasty: a meta-analysis and systematic literature review. J Bone Joint Surg Am 2012; 94(20): 1839-44. Kolotkin R L, Kim J, Davidson L E, Crosby R D, Hunt S C, Adams T D. 12-year trajectory of health-related quality of life in gastric bypass patients versus comparison groups. Surg Obes Relat Dis 2018; 14(9): 1359-65. Krzysztoszek J, Laudanska-Krzeminska I, Bronikowski M. Assessment of epidemiological obesity among adults in EU countries. Ann Agric Environ Med 2019; 26(2): 341-9. Kulkarni A, Jameson S S, James P, Woodcock S, Muller S, Reed M R. Does bariatric surgery prior to lower limb joint replacement reduce complications? Surgeon 2011; 9(1): 18-21. Lee G C, Ong K, Baykal D, Lau E, Malkani A L. Does prior bariatric surgery affect implant survivorship and complications following primary total hip arthroplasty/total knee arthroplasty? J Arthroplasty 2018; 33(7): 2070-4. LÖF. Landstingens Ömsesidiga Försäkringsbolag (Patient Insurance). Available from http://www.patientforsakring.se

Acta Orthopaedica 2021; 92 (1): 96–100

Martin J R, Watts C D, Taunton M J. Bariatric surgery does not improve outcomes in patients undergoing primary total knee arthroplasty. Bone Joint J 2015; 97-B(11): 1501-5. McLawhorn A S, Levack A E, Lee Y Y, Ge Y, Do H, Dodwell E R. Bariatric surgery improves outcomes after lower extremity arthroplasty in the morbidly obese: a propensity score-matched analysis of a New York statewide database. J Arthroplasty 2018; 33(7): 2062-9 e2064. Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono C, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2014; 384(9945): 766-81. Odum S M, Van Doren B A, Springer B D. National obesity trends in revision total knee arthroplasty. J Arthroplasty 2016; 31(9): 136-9. Owens W D, Felts J A, Spitznagel E L, Jr. ASA physical status classifications: a study of consistency of ratings. Anesthesiology 1978; 49(4): 239-43. Poelemeijer Y Q M, Liem R S L, Vage V, Mala T, Sundbom M, Ottosson J, Nienhuijs S W. Perioperative outcomes of primary bariatric surgery in North-Western Europe: a pooled multinational registry analysis. Obes Surg 2018; 28(12): 3916-22. Raoof M, Naslund I, Rask E, Karlsson J, Sundbom M, Edholm D, et al. Health-Related Quality-of-Life (HRQoL) on an average of 12 years after gastric bypass surgery. Obes Surg 2015; 25(7): 1119-27. Raoof M, Naslund I, Rask E, Szabo E. Effect of gastric bypass on bone mineral density, parathyroid hormone and vitamin D: 5 years follow-up. Obes Surg 2016; 26(5): 1141-5. Roche M, Law T Y, Kurowicki J, Rosas S, Rush A J, 3rd. Effect of obesity on total knee arthroplasty costs and revision rate. J Knee Surg 2018; 31(1): 38-42. SCB. Statistics Sweden. Available from http://www.scb.se Schwarzkopf R, Lavery J A, Hooper J, Parikh M, Gold H T. Bariatric surgery and time to total joint arthroplasty: does it affect readmission and complication rates? Obes Surg 2018; 28(5): 1395-1401. Sezgin E A, W-Dahl A, Lidgren L, Robertsson O. Weight and height separated provide better understanding than BMI on the risk of revision after total knee arthroplasty: report of 107,228 primary total knee arthroplasties from the Swedish Knee Arthroplasty Register 2009-2017. Acta Orthop 2020; 91(1): 94-7. Sjostrom L. Review of the key results from the Swedish Obese Subjects (SOS) trial: a prospective controlled intervention study of bariatric surgery. J Intern Med 2013; 273(3): 219-34. Sjostrom L, Narbro K, Sjostrom C D, Karason K, Larsson B, Wedel H, et al., Swedish Obese Subjects S. Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med 2007; 357(8): 741-752. SKAR. The Swedish Knee Arthroplasty Register: annual report 2019. Available from http://www.myknee.se/pdf/SVK_2019_1.0_Eng.pdf SOReg. The Scandinavian Obesity Surgery Registry. Annual report 2018, part 3/3. Available from https://www.ucr.uu.se/soreg/component/edocman/ arsrapport-soreg-2018-del-3/viewdocument/1426?Itemid Sundbom M, Hedberg J, Marsk R, Boman L, Bylund A, Hedenbro J, et al., Scandinavian Obesity Surgery Registry Study G. Substantial decrease in comorbidity 5 years after gastric bypass: a population-based study from the Scandinavian Obesity Surgery Registry. Ann Surg 2017; 265(6): 1166-71. Tohidi M, Brogly S B, Lajkosz K, Grant H J, VanDenKerkhof E G, Campbell A R. Ten-year mortality and revision after total knee arthroplasty in morbidly obese patients. J Arthroplasty 2018; 33(8): 2518-23. Werner B C, Kurkis G M, Gwathmey F W, Browne J A. Bariatric surgery prior to total knee arthroplasty is associated with fewer postoperative complications. J Arthroplasty 2015; 30(9 Suppl.): 81-5. Xanthakos S A. Nutritional deficiencies in obesity and after bariatric surgery. Pediatr Clin North Am 2009; 56(5): 1105-21.

Acta Orthopaedica 2021; 92 (1): 101–107

101

Ankle fracture classification using deep learning: automating detailed AO Foundation/Orthopedic Trauma Association (AO/OTA) 2018 malleolar fracture identification reaches a high degree of correct classification Jakub OLCZAK, Filip EMILSON, Ali RAZAVIAN, Tone ANTONSSON, Andreas STARK, and Max GORDON

Karolinska Institute, Institution for Clinical Sciences, Danderyd University Hospital, Stockholm, Sweden Correspondence: jakub.olczak@ki.se Submitted 2020-05-05. Accepted 2020-09-14.

Background and purpose — Classification of ankle fractures is crucial for guiding treatment but advanced classifications such as the AO Foundation/Orthopedic Trauma Association (AO/OTA) are often too complex for human observers to learn and use. We have therefore investigated whether an automated algorithm that uses deep learning can learn to classify radiographs according to the new AO/OTA 2018 standards. Method — We trained a neural network based on the ResNet architecture on 4,941 radiographic ankle examinations. All images were classified according to the AO/OTA 2018 classification. A senior orthopedic surgeon (MG) then re-evaluated all images with fractures. We evaluated the network against a test set of 400 patients reviewed by 2 expert observers (MG, AS) independently. Results — In the training dataset, about half of the examinations contained fractures. The majority of the fractures were malleolar, of which the type B injuries represented almost 60% of the cases. Average area under the area under the receiver operating characteristic curve (AUC) was 0.90 (95% CI 0.82–0.94) for correctly classifying AO/OTA class where the most common major fractures, the malleolar type B fractures, reached an AUC of 0.93 (CI 0.90–0.95). The poorest performing type was malleolar A fractures, which included avulsions of the fibular tip. Interpretation — We found that a neural network could attain the required performance to aid with a detailed ankle fracture classification. This approach could be scaled up to other body parts. As the type of fracture is an important part of orthopedic decision-making, this is an important step toward computer-assisted decision-making.

Ankle fractures are recognized among the most common fractures, with peak incidence between 15 and 29 years (67 per 100,000 person-years) and elderly women ≥ 60 years (174 per 100,000 person-years) (Westerman and Porter 2007, Thur et al. 2012). Efforts to classify ankle fractures in clinically relevant entities have a long history, ending in 3 classic systems, i.e., the Lauge-Hansen (Hansen 1942), Danis–Weber, and the AO/OTA classifications (Association Committee for Coding and Classification 1996; Budny and Young 2008), where the Danis–Weber with its A, B, and C classes is probably the most used in everyday practice. The most recent update for the AO/OTA classification system was published in 2018 (Meinberg et al. 2018). The AO/OTA system contains classifications for the entire body. The ankle is divided into (1) malleolar, (2) distal tibia, and (3) fibular fractures. For malleolar fractures, the subcategories correspond to the Danis–Weber ABC classification (Hughes et al. 1979) with the addition of a suffix of 2 digits (range 1–3), e.g., the common intra-syndesmotic B-injury without widening of the mortise corresponds to the B1.1 class. The numbers correspond roughly to the severity of each fracture. The complexity of this classification makes it difficult to learn and apply, limiting inter-observer reliability and reproducibility (Fonseca et al. 2017). This has hindered its use in an everyday clinical setting, suggesting the need for better aid during the classification. During recent years, the resurgence of neural networks, a form of artificial intelligence (AI), has proven highly successful for image classification. In some medical image classification applications neural networks attain (Olczak et al. 2017, Kim and MacKinnon 2018, Gan et al. 2019), and surpass, human expert performance (Esteva et al. 2017, Lee et al. 2017,

102

Acta Orthopaedica 2021; 92 (1): 101–107

than 15 years. 145 examinations were excluded from training and 2 examinations from testing. The final study set included 4,676 examinations in the training set and 409 examinations in the test set.

Danderyd University Hospital 2002–2016 Train Rough initial labelling 20,000 exams Study sample 5,086 exams

Train set 4,676 exams

Validation set 410 exams

Test set 409 exams

Figure 1. Data flowchart.

Chung et al. 2018, Urakawa et al. 2019). Machine learning and neural networks are also becoming more commonplace research tools in orthopedics. They hold great potential, as the diagnostic underpinning and intervention decision relies heavily on medical imaging (Cabitza et al. 2018). The strength of these learning algorithms is their ability to review a vast number of examinations and examples, and the speed and consistency with which they can review each examination and at the same time remember thousands of categories without issue. We therefore hypothesized that a neural network can learn to classify ankle fractures according to the AO/OTA 2018 classification from radiographs.

Method Study design The initial dataset consisted of deidentified orthopedic radiographic examinations of various anatomical regions taken between 2002 and 2016 at Danderyd University Hospital in Stockholm, Sweden. Through using the radiologist’s report, we identified images with a high likelihood of fracture, comminution, dislocation, and/or displacement. Based on these categories, we randomly selected a study set of 5,495 ankle examinations where the categories allowed for selecting cases with a higher likelihood of pathology. We introduced this bias to include as many sub-classifications of fractures in the dataset as possible. From the study set, we selected 400 random patients (411 examinations) to include into the test set. 75% were chosen for having reports suggesting a fracture. Similarly, we chose the training and validation sets to have approximately 50% chance of having a fracture (Figure 1). This introduced a selection bias towards pathology, as the primary task was to distinguish different types of fractures and not just the presence of a fracture. We excluded any examination within 90 days of a previously included examination, to ensure that the same fracture was not included more than once, e.g., pre/post reposition/surgery. We further excluded the few pediatric fractures (defined as open physes) as nearly all patients at the hospital are older

Labeling and outputs All examinations selected for the study set were manually reviewed and labelled according to the AO/OTA classification (Meinberg et al. 2018) down to subgroup but excluding subgroup qualifiers. We use the term class for a possible classification outcome, as a summary term for bone, segment, type, group, and subgroup outcome, and specify more clearly when necessary. This means we have 39 classes of malleolar fractures with 3 types (A–C), 3 groups per type (1–3) for each class, and 27 subgroups (3 subgroups per group). Each exam in the training set was reviewed by a minimum of 2 out of 5 reviewers (FE, AS, MG, JO, TA) using a custombuilt image-labeling platform displaying the entire full-scale examination together with the original radiologist report. Reviewer FE was a 5th-year medical student, JO and TA were medical doctors. FE, TA, and JO were specifically trained for the task of labeling radiographic ankle examinations according to the AO/OTA 2018 classification for ankle fractures and labeled between 2,000 and 4,000 examinations each. MG is a senior orthopedic surgeon specializing in orthopedic trauma and AS is a senior orthopedic surgeon. In a second step, all examinations classified as having fractures were rereviewed by MG before being added to the training set. The test set was reviewed by MG and AS. We required a minimum of at least 5 fractures per outcome in the training dataset before including that outcome. The AO classification is partially ligamentous based and as ligaments are not visible on radiographs we therefore used proxies for these classes. For infra-syndesmotic lateral malleolar fractures, if the avulsion fragment was ≤ 3 mm from the tip we classified it as A1.1, 3–10 mm from the tip as A1.2, and ≥ 10 mm as A1.3. As the B1.1 and B1.2 class differ only by syndesmotic injury, information that was not available to us, we chose to separate these by the presence of a step-off in the fracture that could suggest a rotation of the distal fragment. Another important note is that we defined B2.1 based on the presence of a widening of the ankle fork, and this can thus be falsely negative if the ankle has been well repositioned in a cast. Visible fractures of the tibia and fibula were classified as far as possible. Only the complete ankle examinations were included, but no additional examinations of the tibia, fibula, or the foot. In the AO/OTA 2018 version there is an inherent overlap between fibular fractures of the distal end segment (4F3) and fractures of the lateral malleolus (44A–C). A distal end segment fibular fracture (4F3) cannot necessarily be distinguished from ankle fractures involving the distal fibula (44A–C). If the fracture was deemed not to be associated with an ankle fracture it was coded as a fibular fracture (4F) and if it was deemed

Acta Orthopaedica 2021; 92 (1): 101–107

103

Neural network design We used a modified ResNet architecture (He et al. 2015) with a layered structure, Train (n = 4,941) Test (n = 409) which was randomly initiated at the beginFracture type Yes Maybe No Yes Maybe No ning of the experiment. The network, training setup including overfitting strategies, is Fracture 2,156 (44) 121 (2) 2,664 (54) 306 (75) 13 (3) 90 (22) Malleolar (44) 1,696 (34) 63 (1) 3,182 (64) 210 (51) 6 (1) 193 (47) presented in Table 2 (Supplementary data). Tibia distal (43) 254 (5) 6 (0) 4,681 (95) 63 (15) 2 (0) 344 (84) Each output had its own 2-layer subnetwork Fibula (4F2–3) 129 (3) 3 (0) 4,809 (97) 37 (9) 0 (0) 372 (91) and a margin loss. To merge outcomes from Tibia diaphyseal (42) 88 (2) 0 (0) 4,853 (98) 27 (7) 0 (0) 382 (93) Other bone 210 (4) 47 (1) 4,684 (95) 35 (9) 5 (1) 369 (90) various images within the same examination we used the max. function, i.e., if the net“Other bone” generally indicates a visible fracture of the foot. It was possible for an work predicted 2 or more outcomes, the one examination to have multiple fracture labels. with the highest predicted likelihood was selected, ensuring each examination had a to be part of an ankle fracture it was coded as (44A–C), as by unique outcome. The “maybe” outcome was included in the Meinberg et al. (2018). The final verdict was decided by MG. margin loss during training, but was categorized as “no fracThe 2018 AO/OTA revision has separate classifications for ture” during validation and testing. Each outcome was calepiphyseal, metaphyseal, and diaphyseal fractures, and it was culated separately so classifying a fracture as type B did not possible to have multiple labels when multiple fractures and follow from classifying a fracture as group B1, which in turn fracture systems were present. was a separate classification from subgroup B1.1. However unlikely, it is possible for the network to classify a fracture Data set as a type B fracture (between types A and C) and at the same The training data consisted of labeled examinations passed time determine that it is a C1.1 fracture for subgroup clasto the network. A subset of the initial dataset was randomly sification. selected for the test set and was never used during training or validation. We used a biased selection, 75% of fractures, to Outcome performance/statistics increase the likelihood of selecting rare fracture types. The The primary outcome was receiver-operating curve (ROC) test set was manually and independently classified and veri- area under curve (AUC) accuracy for AO/OTA malleolar fracfied by MG and AS using the same platform as in the training ture type, group, and subgroup or no fracture outcome for the set. Any cases where there was disagreement were then subse- complete examination. Secondary outcomes were fibular and quently re-reviewed for a consensus on the final classification tibial AO/OTA classes, as well as any foot fracture when presof the test set (Table 1). ent. These were secondary outcomes as we did not look at the complete examinations, e.g., proximal femur or foot examinaValidation set and active learning tions. To test the diagnostic accuracy of the neural network, we Before each round of training a new validation set of 400 also calculated the sensitivity, specificity, and Youden’s index patients was randomly selected. Based on the validation out- (Youden 1950) for each outcome. There is no consensus as to come we: what an adequate J is, but bigger J is generally more useful. • re-validated categories for training images where the net- Chung et al. (2018) found that J > 0.71 indicated performance work performed poorly to ensure the quality of training superior to an orthopedic surgeon for detecting any fracture labels; in hip radiographs. Two-way interobserver reliability Cohen’s • used targeted sampling via the network outputs combined kappa and percentage agreement was computed between all with specific searches in the radiologist’s reports to extend observers. The overall best performing model (highest AUC) the original training dataset for low-performing categories; on the validation set was used for final testing on the test set. • implemented active learning, where categories with low As there is a large number of categories we also present a performance despite having plenty of training examples weighted mean for groups. The weighting is according to the were targeted with more data and targeted review of train- number of cases as we want small categories that may perform ing labels during training. well by chance to have less influence on the weighted mean; for AUC the calculation was: Image input AUCi * ni Σ categories i=1 AUC = = WAUC The labeled radiographic images were scaled down with weighted categories n Σ = 1 i retained proportions, so that the largest side had 256 pixels. i Only outcomes with ≥ 2 cases in the test set were evaluated If the image was not square, the shorter side was extended with black pixels resulting in a 256 × 256 square proportion- during testing. Main outcomes were classes A–C, group A1– C3, subgroup A1.1–C3.3. ally scaled copy. Table 1. Base distribution of fractures according to the AO classification. Values are count (%)

104

Acta Orthopaedica 2021; 92 (1): 101–107

Table 3. Distribution of malleolar fractures by type (44A–C), specified by type, group, and subgroup. Values are count (%) for samples > 100 AO type

Train (n = 4,941)

Test (n = 409)

44A (483 train and 31 test cases) 1.1 78 (22) 6 1.2 165 (46) 7 1.3 114 (32) 9 2.1 105 (93) 5 2.2 1 (1) – 2.3 7 (6) 2 3.1 11 – 3.3 2 2 44B (1,015 train and 136 test cases) 1.1 385 (74) 39 1.2 132 (25) 26 1.3 6 (1.1) 2 2.1 99 (44) 20 2.2 105 (47) 16 2.3 19 (8.5) 2 3.1 76 (28) 12 3.2 152 (56) 13 3.3 41 (15) 6 44C (255 train and 47 test cases) C1 1.1 85 (67) 17 1.2 20 (16) 5 1.3 22 (17) 2 2.1 30 6 2.2 21 3 2.3 39 9 3.1 10 3 3.2 9 1 3.3 19 1

Ethics, funding, and potential conflicts of interest This study was approved by the Regional Ethics Committee fort Stockholm, Sweden (Dnr. 2014/453-31/3, April 9, 2014). This project was supported by grants provided by Region Stockholm (ALF project), the Swedish Society of Doctors (Svenska Läkaresällskapet) and by the Karolinska Institute. AS and MG are co-founders and shareholders in DeepMed AB. AR is a shareholder in DeepMed AB.

Table 4. Outcome measures for the most important groups and weighted average AUC for each malleolar AO type, group, and subgroup combined Cases Sensitivity Specificity AO type n = 409 (%) (%) Youden’s J AUC (95% CI) 44A Base 32 73 81 0.54 1 22 88 75 0.63 1.1 6 75 93 0.68 1.2 7 80 83 0.63 1.3 9 75 88 0.63 2 7 100 74 0.74 2.1 5 100 74 0.74 3 2 100 86 0.86 44B Base 137 89 88 0.77 1 67 90 88 0.77 1.1 39 87 84 0.71 1.2 26 92 85 0.77 2 38 82 84 0.65 2.1 20 100 72 0.72 2.2 16 88 74 0.62 2.3 2 100 98 0.98 3 32 78 90 0.68 3.1 12 83 75 0.58 3.2 13 92 82 0.74 3.3 6 100 91 0.91 44C Base 47 74 90 0.65 1 24 75 79 0.54 1.1 17 76 85 0.61 1.2 5 80 92 0.72 1.3 2 100 88 0.88 2 18 100 72 0.72 2.1 6 83 93 0.76 2.2 3 100 88 0.88 2.3 9 100 77 0.77 3 5 100 88 0.88 Malleolar 216 86 90 0.76 Weighted mean AUC A B C Malleolar

0.81 (0.72–0.88) 0.87 (0.77–0.94) 0.87 (0.70–0.98) 0.79 (0.54–0.94) 0.84 (0.70–0.95) 0.91 (0.83–0.97) 0.89 (0.80–0.97) 0.90 (0.83–0.96) 0.93 (0.90–0.95) 0.93 (0.88–0.96) 0.89 (0.85–0.93) 0.90 (0.81–0.96) 0.87 (0.80–0.92) 0.87 (0.83–0.92) 0.82 (0.68–0.91) 0.99 (0.97–1.00) 0.90 (0.85–0.94) 0.79 (0.63–0.90) 0.91 (0.84–0.96) 0.96 (0.93–0.98) 0.86 (0.79–0.92) 0.83 (0.72–0.91) 0.86 (0.74–0.94) 0.89 (0.77–0.97) 0.92 (0.86–0.97) 0.91 (0.86–0.95) 0.91 (0.79–0.98) 0.96 (0.88–1.00) 0.88 (0.84–0.92) 0.95 (0.90–0.98) 0.92 (0.89–0.95) 0.84 0.90 0.87 0.90

Criterion based on Youden’s Index (Youden 1950, Aoki et al. 1997, Shapiro 1999, Greiner et al. 2000) defined as YI(c) = maxc (Se(c) + Sp(c)–1). This is identical (from an optimization point of view) to the method that maximizes the sum of sensitivity and specificity (Albert 1987, Zweig and Campbell 1993) and to the criterion that maximizes concordance, which is a monotone function of the AUC.

Results 5,495 radiographic examinations were used in the experiment. 5,086 examinations were used for training and validation and 409 examinations (400 unique patients) were withheld in the test set, with no patient overlap. In the combined data, there were 2,462 examinations with a fracture. Malleolar fractures were by far the most prevalent fractures (1,906 out of 2,462 fractures) and the majority of them were type B injuries (1,147), followed by type A injuries (456) and type C injuries (300). The training set had 1,753 malleolar fractures for 39 possible outcomes, averaging 48 positive training cases per outcome, though some

classes had more fractures than others (Table 3 and Figures 1 and 2). Main results 32 out of 39 outcomes had 2 or more examinations in the test set. Most outcomes were possible to train and most classes that disappeared had too few test cases (Table 4). For malleolar fractures, weighted mean AUC came to 0.90 with varying 95% confidence intervals (CI) for individual classes. The network could identify malleolar fractures with

Acta Orthopaedica 2021; 92 (1): 101–107

No. of observations (train) A

500

105

No. of observations (test)

Second number 1 2 3

400

C Second number 1 2 3

300

200 20 100

First number in the AO-classification

Figure 2. Distribution of AO classes in the malleolar fracture data.

an AUC 92 (CI 0.89–0.95). For malleolar fracture overall best performance was achieved for type B (1,015 in the training and 136 in the test set) injuries with AUC 0.93 (CI 0.90– 0.95), then type C (255 cases in the training and 47 cases in the test set) with AUC 0.86 (CI 0.78–0.92), and then type A (483 in the training and 31 in the test set) with AUC 0.81 (CI 0.72–0.88). Type A injuries exhibited the poorest results with weighted average AUC 0.84. It was not possible to evaluate subgroups A2.1, A2.3, and subgroups of A3. Average AUC for type B injuries was 0.90 and all classes, except the subgroup B13, were evaluated. Weighted average AUC for type C injuries was 0.87 but it was not possible to evaluate subgroups to C3. Despite there being almost twice as many type A fractures in the data set there were fewer type A fractures in the test set, which resulted in few outcomes for type A fractures. Other anatomies The number of fractures in the other anatomies did not allow for a detailed analysis for many of the classes. In the test set the second most common fracture group was the distal tibia group with weighted average AUC 0.90. We found similar values for the isolated fibular and tibial diaphysis fractures. The foot fractures were somewhat less performant, mostly due to metatarsal fractures (see Supplementary data).

Figure 3. The network failed to identify this image as a malleolar type A fracture. Among the malleolar fractures it was predicted as a type C fracture.

Other analyses Overall Cohen’s kappa between reviewers was 0.65 (and 0.55 on the AO classification task) (see Supplementary data). When reviewing the failed images there was no obvious pattern. The presence of casts was common (Figure 3) or discrete findings (Figure 4) were common but we could not see any clear pattern that the failures followed.

Discussion This study is the first, to our knowledge, that classifies fractures according to the AO/OTA classification, and ankle fractures in particular, using machine learning. We believe that an

Figure 4. The examination should have been a malleolar type C fracture but the network predicted type B fracture.

106

average AUC 0.90 for the relatively complex AO/OTA classification task, on a small training set with many categories, is a good outcome. This study shows the potential benefits of an AI classification, where complex classifications can become commonplace to the benefit of patients and their treatment. We have shown that a neural network, using a combinatory approach with different machine learning methods and targeted labeling, can learn even rare fracture types. In the AO/OTA classification, C injuries tend to be more complex and severe than type B injuries, which in turn are worse than type A injuries. Higher group and subgroup numbers also tend to entail more severe and complex injuries. We found that malleolar type A injuries decreased in frequency with severity, and that there were fewer type A than type B injuries. One reason for this was that many minor fractures, e.g., simple avulsion fragments, are a form of distortion that, despite being commonplace, are difficult to diagnose through radiographs. Fonseca et al. (2017) found a kappa of 0.38 for the AO classification (not subgroups) whereas our study found kappa 0.55 between the human reviewers MG and AS. In a separate test, 388 of the examinations in the test set were reviewed by a resident emergency medical specialist (TA). Kappa for this subset of the training set was 0.53 or an agreement of 92%. One reason for this could be that reviewers usually had access to the radiologist’s report, probably improving kappa, while the network never did. While the report never specified AO classification and mostly helped identify discrete fractures, it also helped fill in the lack of additional patient information. In addition, the very unsymmetrical distribution of outcomes (marginal probabilities for each individual class) between AO classes (e.g., C3.3 is much less uncommon than B1.1) likely unduly penalizes kappa for the AO classification task (Delgado and Tibau 2019). Compared with Juto et al. (2018) we found a percentile agreement of ≥ 91% for all levels (fracture type, group, and subgroup) between observers. Both Fonseca et al. (2017) and Juto et al. (2018) used the previous AO/OTA classification. Detecting a fracture was easy for humans and computer alike and there was great agreement, but in line with other studies AO/OTA classification is complicated as is shown by the declining kappa. This strengthens the case for an automated classification system that can assist in making uniform classifications. Overall, the network was good at classifying ankle fractures and its subgroups though some subclasses were difficult and many had insufficient data. Limitations We relied on only radiographs and the radiologist’s report, which does not fully allow for discrimination between the AO/OTA classifications, especially where ligamentous injuries are important. However, as most ankle fractures will never undergo a CT or MRI examination, extracting additional chart

Acta Orthopaedica 2021; 92 (1): 101–107

information would have little impact on the outcome. CT and MRI are also not performed randomly on fractures and including them in our results, when available, would introduce an information bias. We also strongly believe that clinicians will always have to add their clinical exam to the interpretation even with these new technologies, as some information simply is not present in a radiographic image. This study reports the outcome of the top classification, the highest AUC. For many malleolar subgroups the difference is small and it would make sense to present additional likely outcomes, in particular outcomes where the differences are only in ligamentous injuries, alongside each other—for example B1.1 and B1.2. A repositioned or stabilized fracture can hide a previously obvious ligamentous injury, changing the classification. Our data entailed a selection bias towards pathological material and did not represent the average population. Despite this, there were insufficient cases for many subclasses and for some outcomes the statistical significance and confidence intervals were difficult to assess. Uncommon pathologies are problematic for any human observer or deep learning system. We have combated this by selecting new cases for annotation where the network has either (1) difficulties distinguishing a category, or (2) high likelihood of a rare fracture class, a form of active learning. This interactive approach to machine learning proved useful and could be repeated, and adding more data could help target rare fractures. The human observers had access to full-scale radiographs and reports whereas the network, at best, had proportionally scaled 256 × 256 representations. Despite this limitation, many of the categories were correctly identified. We believe that this is most likely due to the fact that the network reviews each image and thus is able to find even tiny changes. We chose this approach as our experience has indicated that increasing image size has little benefit. Similarly, we have tried some different permutations of the network structure with mostly similar outcomes. It is important to keep in mind that the literature surrounding deep learning is vast and there are many interesting network designs that could be tested. Regardless, we believe that the chosen structure fulfills our aim, to find a network that can help clinicians to use complex fracture classifications on an everyday basis. Despite having a large dataset and actively searching for pathology we found it hard to find an adequate number of fractures for many of the classes. While we can retrain the network to fit new categories, it is important to remember that fractures in case reports and other rare entities will be a challenge for deep learning applications and clinicians alike. Generalizability The source population was dominated by a Caucasian population. We excluded only examinations with open physes and believe that our results generalize well in a regular clinical setting, though we would expect more negative cases and simple

Acta Orthopaedica 2021; 92 (1): 101–107

fractures than in our material. The clinical performance of the algorithm may therefore differ from the sample performance. Our results also extend to the Danis–Weber classification to the extent that it is a subset of the AO classification. Interpretation A neural network can learn the AO/OTA classification from relatively few training examples. Even with this small data set we find that we can achieve high predictive accuracy for most categories. The strength of an AI model is the ability to further improve the model by adding more training cases and its potential for uniform classification. Supplementary data Table 2, inter-rater reliability (IRR) results, and data on other fracture classes are available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674. 2020.1837420 Great thanks are offered to Tor Melander, Hans Nåsell, and Olof Sköldenberg for great feedback. According to CRediT (https://casrai.org/credit/). JO: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, visualization, writing—original draft, writing—review & editing; FE: conceptualization, data curation, formal analysis, writing—review & editing; ASR: conceptualization, formal analysis, methodology, software, validation, visualization, writing—review & editing; TA: data curation, writing—review & editing; AS: data curation, funding acquisition, supervision, validation, writing—review & editing; MG: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing—review & editing. Acta thanks Hans E Berg and Jan Erik Madsen for help with peer review of this study.

Albert. Multivariate interpretation of clinical laboratory data. Boca Raton, FL: CRC Press; 1987. Aoki K, Misumi J, Kimura T, Zhao W, Xie T. Evaluation of cutoff levels for screening of gastric cancer using serum pepsinogens and distributions of levels of serum pepsinogen I, II and of PG I /PG II ratios in a gastric cancer case-control study. J Epidemiol 1997; 7(3): 143-51. Association Committee for Coding and Classification. Fracture and dislocation compendium. Orthopaedic Trauma Association Committee for Coding and Classification. J Orthop Trauma 1996; 10(Suppl. 1): v-ix, 1-154. Budny A M, Young B A. Analysis of radiographic classifications for rotational ankle fractures. Clin Podiatr Med Surg 2008; 25(2): 139-52. Cabitza F, Locoro A, Banfi G. Machine learning in orthopedics: a literature review. Front Bioeng Biotechnol [Internet] 2018 Jun 27 [cited 2019 Oct 29]; 6. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/ PMC6030383/ Chung S W, Han S S, Lee J W, Oh K-S, Kim N R, Yoon J P, et al. Automated detection and classification of the proximal humerus fracture by using deep learning algorithm. Acta Orthop 2018; 89(4): 468-73.

107

Delgado R, Tibau X-A. Why Cohen’s Kappa should be avoided as performance measure in classification. PLOS ONE 2019; 14(9): e0222916. Esteva A, Kuprel B, Novoa R A, Ko J, Swetter S M, Blau H M, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature 2017; 542(7639): 115. Fonseca L, Nunes I, Nogueira R, Martins G, Mesencio A, Kobata S. Reproducibility of the Lauge-Hansen, Danis–Weber, and AO classifications for ankle fractures. Rev Bras Ortop Engl Ed 2017; Dec 1: 53. Gan K, Xu D, Lin Y, Shen Y, Zhang T, Hu K, et al. Artificial intelligence detection of distal radius fractures: a comparison between the convolutional neural network and professional assessments. Acta Orthop 2019; 90(4): 394-400. Greiner M, Pfeiffer D, Smith R D. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med 2000; 45(1): 23-41. Guillaumin M, Verbeek J, Schmid C. Multimodal semi-supervised learning for image classification. In: 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition [Internet]. San Francisco, CA: IEEE; 2010 [cited 2020 Jan 14]. p. 902-9. Available from: http://ieeexplore.ieee.org/document/5540120/ Hansen N L. Ankelbrud i genetisk Diagnose og Reposition: experimentalchirurgiske og radiografiske Undersøgelser Repositionsforsøg i Kliniken. Copenhagen: E. Munksgaard; 1942. He K, Zhang X, Ren S, Sun J. Delving deep into rectifiers: surpassing human-level performance on ImageNet Classification. 2015 [cited 2016 Aug 19]. p. 1026-34. Available from: http://www.cv-foundation. org/openaccess/content_iccv_2015/html/He_Delving_Deep_into_ ICCV_2015_paper.html Hughes J L, Weber H, Willenegger H, Kuner E H. Evaluation of ankle fractures: non-operative and operative treatment. Clin Orthop Relat Res 1979; (138): 111. Izmailov P, Podoprikhin D, Garipov T, Vetrov D, Wilson A G. Averaging weights leads to wider optima and better generalization. ArXiv180305407 Cs Stat [Internet] 2019 Feb 25 [cited 2019 Nov 16]. Available from: http:// arxiv.org/abs/1803.05407 Juto H, Nilsson H, Morberg P. Epidemiology of adult ankle fractures: 1756 cases identified in Norrbotten County during 2009–2013 and classified according to AO/OTA. BMC Musculoskelet Disord 2018; 19(1): 441. Kim D H, MacKinnon T. Artificial intelligence in fracture detection: transfer learning from deep convolutional neural networks. Clin Radiol 2018; 73(5): 439-45. Lee J-G, Jun S, Cho Y-W, Lee H, Kim GB, Seo JB, et al. Deep Learning in Medical Imaging: General Overview. Korean J Radiol 2017; 18(4): 570-84. Meinberg E G, Agel J, Roberts C S, Karam M D, Kellam J F. Fracture and dislocation classification compendium—2018. J Orthop Trauma 2018; 32(Suppl. 1): S1-170. Olczak J, Fahlberg N, Maki A, Razavian A S, Jilert A, Stark A, et al. Artificial intelligence for analyzing orthopedic trauma radiographs. Acta Orthop 2017; 88(6): 581-6. Shapiro D E. The interpretation of diagnostic tests. Stat Methods Med Res 1999; 8(2): 113-34. Thur C K, Edgren G, Jansson K-Å, Wretenberg P. Epidemiology of adult ankle fractures in Sweden between 1987 and 2004. Acta Orthop 2012; 83(3): 276-81. Urakawa T, Tanaka Y, Goto S, Matsuzawa H, Watanabe K, Endo N. Detecting intertrochanteric hip fractures with orthopedist-level accuracy using a deep convolutional neural network. Skeletal Radiol 2019; 48(2): 239-44. Westerman R W, Porter K. Ankle fractures in adults: an overview. Trauma 2007; 9(4): 267-72. Youden W J. Index for rating diagnostic tests. Cancer 1950; 3(1): 32-5. Zweig M H, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinic medicine. Clin Chem 1993; 39(4): 561-77.

108

Acta Orthopaedica 2021; 92 (1): 108–112

Patient-reported outcomes of joint-preserving surgery for moderate hallux rigidus: a 1-year follow-up of 296 patients from Swefoot Marcus E CÖSTER 1,2, Fredrik MONTGOMERY 1, and Maria C CÖSTER 1,3 1 Department

of Clinical Sciences, Lund University; 2 Department of Orthopedics, Central Hospital in Växjö; 3 Department of Orthopedics, Skåne University Hospital in Malmö, Sweden Correspondence: maria.coster@skane.se Submitted 2020-04-07. Accepted 2020-09-01.

Background and purpose — Hallux rigidus (HR) may cause decreased range of motion, joint pain, and gait disturbances. There is a lack of evidence regarding the outcome of different surgical procedures for moderate HR. We report patient-reported outcomes after joint-preserving surgical procedures for moderate HR. Patients and methods — We included 296 patients registered in Swefoot (Swedish national registry of foot and ankle surgery) who underwent primary surgery for moderate HR 2014 through 2018. We extracted information on anthropometrics, grading of HR, chosen surgical procedure, and patient-reported data including the PROMs SEFAS (summary score 0–48) and EQ-5D-3L (index 0–1) preoperatively and 1 year postoperatively. Results — 115 patients underwent metatarsal decompression (i.e., Youngswick) osteotomy (YOT) and 181 underwent cheilectomy. The mean improvement in SEFAS score 1 year after surgery was 12 points (95% CI 10−13) for YOT and 10 points (CI 9−12) for cheilectomy. Also, EQ-5D improved in both groups. Patients who underwent YOT were more satisfied with the procedure (84% vs. 70 % for cheilectomy, p = 0.02). Interpretation — Surgically treated patients with moderate HR improved after both YOT and cheilectomy, according to patient-reported data from Swefoot. Patients who underwent a YOT were more satisfied with their procedure. One possible explanation may be that more patients in the YOT group had a concomitant hallux valgus; however, we have no information on this.

Hallux rigidus (HR), which affects approximately 2% of the population, can be graded based on the severity of the disease (Lucas et al. 2015, Lam et al. 2017, Galois et al. 2020). Beeson et al. (2008) conducted a literature review of the existing multiple classification systems where the system promoted by Coughlin and Shurnas (2003) was deemed to have the most reliable grading. Surgical treatment options can be joint-preserving and joint-sacrificing procedures. Several studies have shown positive outcomes for different surgical procedures (Coughlin and Shurnas 2003, Stevens et al. 2017, Coutts and Kilmartin 2019, Sidon et al. 2019, Slullitel et al. 2019), but there is a lack of consensus on what procedure should be chosen for a certain grade of HR (Galois et al. 2020). For moderate HR, when joint-preserving surgery is preferred, techniques such as cheilectomy, phalangeal and metatarsal osteotomies could be used (Coughlin and Shurnas 2003, Slullitel et al. 2016, Sidon et al. 2019, Galois et al. 2020). The evidence regarding which of these techniques should be used is, however, sparse. In Sweden most joint-preserving procedures are Youngswick osteotomies (YOT) and cheilectomies. YOT is a chevronshaped decompression osteotomy of the first metatarsal where a slice of bone is removed in the dorsal arm of the osteotomy to achieve plantar and proximal displacement of the head. Most surgeons use a modified osteotomy, where the slice of bone being removed is smaller than what was originally described (Youngswick 1982, Gerbert et al. 2001). This procedure gives the possibility to address the hallux valgus deformity additionally. Even though studies have shown positive outcomes for both YOT and cheilectomy (Coughlin and Shurnas 2003, Sidon et al. 2019, Slullitel et al. 2019) there are few studies

Acta Orthopaedica 2021; 92 (1): 108–112

comparing these procedures. There is some evidence suggesting that cheilectomy has a higher frequency of revision surgery (Cullen et al. 2017), but no consensus on recommendations exists. Therefore, we analyzed patient-reported outcome for YOT and cheilectomy reported to Swefoot in patients with moderate HR. We hypothesized that YOT would have better outcomes than cheilectomy.

Patients and methods Swefoot We used data from Swefoot, a Swedish national quality registry of foot and ankle surgery. Swefoot was launched in 2014 and covers approximately 50% of all units performing foot and ankle surgery in Sweden. The registry contains baseline preoperative data; for example, sex, age, anthropometrics, comorbidities, and smoking habits. Patients are asked to complete 2 patient-reported outcome measures (PROMs): SEFAS (Self-reported Foot and Ankle Score) and EQ-5D 3L (EuroQol 5-dimensional 3 level version) before surgery, and 1 and 2 years after surgery. In addition to these PROMs patients are at the same time asked 4 questions regarding appearance, shoes, strength, and forefoot pain. After 1 and 2 years, they also complete 4 specific questions regarding their satisfaction with the result of the procedure and existence of any adverse events (Appendix, see Supplementary data). At the time of surgery, the surgeon reports the diagnosis, radiographic findings, severity of disease, type of surgical procedures, type of fixation, and postoperative routine. Patient-reported outcome measures (PROMs) SEFAS is a self-reported PROM specific for disorders in the foot and ankle, which has been described and validated for both the forefoot and hindfoot in previous publications (Cöster et al. 2012, 2014, 2018). A summary score is calculated based on the answers ranging from 0 (severe disability) to 48 (normal function). SEFAS covers different constructs, which are not reported separately in subscales. The most important of these constructs are pain, functional limitations, and quality of life (QoL). EQ-5D is a generic PROM evaluating health-related quality of life (QoL) (EuroQol 1990) that is also validated and used in patients with foot and ankle disorders. We used the UK tariff to calculate an index score ranging from 0 to 1.0 where 1.0 represents full health. Patients We extracted data from Swefoot for all 623 reported patients who underwent primary surgery for moderate HR with YOT (including dorsal osteophyte resection) or cheilectomy between January 1, 2014 and December 31, 2018. To determine the severity of HR we used the grading system suggested by Coughlin and Shurnas where grade 0 represents

109

mild disease and grade 4 advanced disease. We regarded grade 2 and 3 as moderate HR and included only patients in these grades. We excluded 327 patients who had not completed the PROMs both preoperatively and 1 year postoperatively (154 pre- and 173 postoperatively). These inclusion criteria resulted in 296 patients. Due to the large proportion of excluded patients we undertook a dropout analysis, which showed that preoperative age, sex, SEFAS score, and EQ-5D index were similar in both the study group and dropouts. Patients in the dropout group had slightly higher BMI, but in a sub-analysis this difference was equal for the YOT and the cheilectomy group. Statistics Data are reported as numbers and proportions (%), mean (SD), or median (range). We considered a probability of less than 5% as statistically significant and used 95% confidence intervals (CI) to describe uncertainty. Outcome is reported in the registry as the summary score for SEFAS and index for EQ-5D before and after surgery. Delta score and index is calculated as postoperative value minus preoperative value. The delta score, i.e., the absolute difference, could be without clinical relevance. Due to this we related the absolute difference to the minimally important change (MIC) for the PROMs. MIC reflects the smallest measured change in score that patients perceive as being important and defines a threshold when a treatment should be regarded as clinically relevant. The MIC value for the SEFAS in patients with forefoot disorders is 5 score points (Cöster et al. 2017). The MIC value for the EQ-5D in patients with foot and ankle disorders it is not defined, but in patients with back pain, another musculoskeletal disorder, the value is 0.173 (Johnsen et al. 2013). Group comparisons were performed using an independent-samples t-test for parametric data and Mann–Whitney U-test or chisquare test for non-parametric data. We used IBM SPSS Statistics version 24 (IBM Corp, Armonk, NY, USA) to perform the statistical analyses. Ethics, data sharing, funding, and potential conflicts of interest The study protocol was approved by the Ethical Review Board (Etikprövningsmyndigheten) in Sweden (reference number 2019-02733). The study was conducted in accordance with the Helsinki Protocol. The registration of data and the study were performed confidentially after patient consent and according to Swedish and EU data protection rules. Data may be accessible upon application to the registry. The study was supported by grants from Herman Järnhardt foundations. The Swedish Foot and Ankle Registry (Swefoot) is financed by the Swedish Association of Local Authorities and Regions. The funders had no influence on the design of the study, the collection, analysis, and interpretation of data, on writing the manuscript, or on any other part of the study. The authors declare no conflict of interest.

110

Acta Orthopaedica 2021; 92 (1): 108–112

Table 1. Preoperative baseline data presented as numbers with percentages unless otherwise specified

Osteotomy Cheilectomy (n = 115) (n = 181)

Mean age (SD) 55 (10) 58 (11) median (range) 60 (21–78) 55 (29–82) Female sex 85 (74) 112 (62) Mean BMI (SD) 26 (4) 27 (4) Diabetes mellitus 5 (4) 8 (4) Rheumatoid arthritis 5 (4) 6 (3) Smoker 4 (4) 2 (1) Quit smoking before surgery 7 (6) 12 (7) HR grade 2 47 (41) 70 (39) 3 68 (59) 111 (61)

p-value 0.04 0.03 0.2 1.0 0.3 0.2 0.7

Table 3. Patient-reported satisfaction. Values are numbers with percentages Osteotomy Cheilectomy (n = 115) (n = 181) p-value Satisfaction with outcome of surgery Satisfied 97 (84) 127 (70) Neither satisfied nor dissatisfied 6 (5) 16 (9) Dissatisfied 12 (10) 38 (21) Satisfaction with appearance of the feet Preoperatively Satisfied 46 (40) 103 (57) Dissatisfied 30 (26) 37 (20) Postoperatively Satisfied 98 (85) 139 (77) Dissatisfied 3 (3) 15 (8)

0.006 0.02 0.005 0.3 0.08 0.04

in the YOT group compared with the cheilectomy group reported dissatisfaction with the results (Table 3). Plantar forefoot problems were reported postoperatively Osteotomy (n = 115) Cheilectomy (n = 181) Group comparison among 3% in the osteotomy group and 8% in the cheilectomy group (p = 0.1). SEFAS summary score a Before surgery a higher percentage of preoperative 26 (7) 26 (8) postoperative 38 (8) 36 (9) 1.5 (–0.6 to 3.5) patients in the YOT group compared with delta 12 (8) (10–13) 10 (9) (9.1–12) 1.2 (–0.7 to 3.2) the cheilectomy group reported that they EQ-5D index b were dissatisfied with the appearance of preoperative 0.59 (0.3) 0.61 (0.3) postoperative 0.80 (0.2) 0.77 (0.2) 0.04 (–0.01 to 0.09) their feet. Conversely, postoperatively delta 0.21 (0.3) (0.15–0.26) 0.16 (0.3) (0.11–0.20) 0.05 (–0.02 to 0.12) more patients in the YOT group reported satisfaction with the appearance of their Delta is calculated as postoperative value minus preoperative value. a p = 0.8 feet compared with the cheilectomy group b p = 0.7 (Table 3). 77% of patients in the YOT group were satisfied with the shoes they were able to wear postoperatively compared with 76% in the cheilectomy group. Also, 82% from the YOT group were satisfied with Results their postoperative foot strength compared with 79% in the Surgery was performed on 296 patients, 115 with a YOT and cheilectomy group (p = 0.5). 181 with a cheilectomy. Both methods were used equally in most of the regions in Sweden except in the central part of Sweden where cheilectomy was the preferred method. In the Discussion YOT group the patients were younger and included more women compared with the cheilectomy group (Table 1). SEFAS In this registry study we found that both YOT and cheilecsummary score and EQ-5D index increased and the increase tomy resulted in improved foot-related pain and function and was statistically significant but also clinically relevant (based health-related QoL for patients with moderate HR. Both the on MIC values) for both the YOT and cheilectomy group 1 year SEFAS score and EQ-5D index increased more than their postoperatively compared with before surgery (Table 2). The identified thresholds for clinical relevance. These results are CI of the changes by surgery all exceeded MIC, inferring that in line with previous studies that have shown positive outwith more than 95% probability there is improvement exceed- comes both short term and long term and suggest that both ing MIC in the general population also. Postoperative scores techniques are adequate to treat moderate HR surgically with positive effects evolving within 1 year after surgery (Waizy et and indices were similar in both groups (Table 2). 84% of patients in the YOT group reported that they were al. 2010, Cullen et al. 2017). When we started this study, we expected that the YOT and satisfied with their procedure compared with 70% in the cheilectomy group (p = 0.01). Also, a lower percentage of patients the cheilectomy group were identical and comparable in baseTable 2. Preoperative, postoperative and mean increase in SEFAS score and EQ-5D index from before surgery until the 1-year follow-up for respective procedure. Values are mean (SD) (95% CI)

Acta Orthopaedica 2021; 92 (1): 108–112

line features such as age, percentage of women, anthropometrics, and presence of other foot and ankle disorders. We hypothesized that the YOT group should be more satisfied than the cheilectomy group after surgery. However, we found that patients in the YOT group were younger and included more women. When comparing the 2 surgical procedures we found similar improvement in PROMs in the 2 groups. However, we also found that a higher percentage of patients in the YOT group reported that they were satisfied with their surgery and that patients in the cheilectomy group were dissatisfied at a higher percentage. A possible explanation for this is that there are other aspects than PROMs can capture which also affect satisfaction with surgery. Another possible explanation may be that the YOT group contains proportionally more patients with concomitant hallux valgus (HV). The YOT can be used to decompress the joint with osteoarthritis and in the same procedure additionally to correct a HV deformity. There was a higher percentage of women in the osteotomy group and thus perhaps also more patients with concomitant HV (HV is more common in women than men; Coughlin and Jones 2007). Also, the fact that the patients in the YOT group were less satisfied with the appearance of their feet preoperatively could indicate that they also had HV and thus they would be more satisfied with surgery if this deformity was corrected as well. This would then also explain why these patients were satisfied at a higher grade with their appearance of their feet postoperatively. In summary, all these facts indicate that HV could be overrepresented in the YOT group. Thus, the 2 groups are not completely comparable. Although HR is a disease with strict diagnostic criteria the group of patients with HR is to some extent a mixed group between HR and HV. This must be considered when analyzing the results as it makes it possible that with the YOT we treat 2 conditions in these patients and thereby receive better results. Swefoot does not specify HR patients with a concomitant HV. With the knowledge from this study we will adjust the registry to recognize these patients. We will also add a possibility for the surgeon to report additional corrective surgery. Our hypothesis was that YOT would have better results than cheilectomy in moderate HR because YOT with dorsal osteophyte resection not only has immediate effect on ROM and thereby pain and gait function (as does cheilectomy), but also changes the length and angle of the metatarsal, which should also improve symptoms. In theory, these changes could reduce the risk for secondary surgery, but our 1-year follow-up time might be too short to capture such differences. Previous studies have used a 5-year follow-up regarding reoperation (Cullen et al. 2017, Slulittel et al. 2019). Future studies should include a longer follow-up time and have reoperation as an outcome to be able to address this question. With decompression YOT there have been concerns regarding postoperative development of metatarsalgia due to the shortening of the metatarsal bone. We asked patients at the 1-year follow-up what degree of plantar forefoot problems

111

they experienced and in fact tendencies were that the cheilectomy group experienced more symptoms. At the very least there was similar symptomatology in both groups. Although plantar forefoot problems are not directly transferrable to a diagnosis of metatarsalgia this is an indication that there were no obvious differences in experienced problems and thus the concerns regarding postoperative metatarsalgia after decompression osteotomy seems to be a minor issue when deciding on the procedure for moderate HR. We did not present adverse events, revision rate, or secondary surgery because the registry is too new. There are some limitations with this study. As this is a multicenter register study data reporting could vary between centers and surgeons. This could affect for example grading of HR as it is based on findings reported from the surgeon. Some centers might prefer one of the surgical procedures and combined with different experience of foot and ankle surgery this could affect outcomes. Furthermore, we are not able to make longterm statements due to the short follow-up time of 1 year. In Sweden, phalangeal osteotomies are not performed although this is a common procedure in several other countries. Hence, we are not able to include these procedures in the analysis. The strengths of this study are the multicenter register design, the large sample size, and that we prospectively examine data collected in routine clinical practice. Another strength is that we have used a region-specific PROM, SEFAS, that is thoroughly evaluated in patients with foot and ankle disorders. In conclusion, surgically treated patients with moderate HR improved within 1 year after surgery after both YOT and cheilectomy according to patient-reported outcomes, and the improvement was clinically relevant. In HR patients with a concomitant HV the YOT might be the better surgical method. This ought to be evaluated further in future studies. Supplementary data The Appendix is available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674. 2020.1824762 MEC performed the statistical analyses and wrote the manuscript. FM assisted with the writing and statistics, as well as the planning of the study. MCC planned the study and supervised the writing. Acta thanks Jan Willem Louwerens for help with peer review of this study.

Beeson P, Phillips C, Corr S, Ribbans W. Classification systems for hallux rigidus: a review of the literature. Foot Ankle Int 2008; 29(4): 407-14. Cöster M, Karlsson M K, Nilsson J A, Carlsson A. Validity, reliability, and responsiveness of a self-reported foot and ankle score (SEFAS). Acta Orthop 2012; 83(2): 197-203. Cöster M C, Bremander A, Rosengren B E, Magnusson H, Carlsson A, Karlsson M K. Validity, reliability, and responsiveness of the Self-reported Foot and Ankle Score (SEFAS) in forefoot, hindfoot, and ankle disorders. Acta Orthop 2014; 85(2): 187-94.

112

Cöster M C, Nilsdotter A, Brudin L, Bremander A. Minimally important change, measurement error, and responsiveness for the Self-Reported Foot and Ankle Score. Acta Orthop 2017; 88(3): 300-4. Cöster M C, Rosengren B E, Karlsson M K, Carlsson A. Age- and genderspecific normative values for the Self-Reported Foot and Ankle Score (SEFAS). Foot Ankle Int 2018; 39(11): 1328-34. Coughlin M J, Jones CP. Hallux valgus: demographics, etiology, and radiographic assessment. Foot Ankle Int 2007; 28(7): 759-77. Coughlin M J, Shurnas P S. Hallux rigidus: grading and long-term results of operative treatment. J Bone Joint Surg Am 2003; 85-A(11): 2072-88. Coutts A, Kilmartin T E. Dorsiflexory phalangeal osteotomy for grade ii hallux rigidus: patient-focused outcomes at eleven-year follow-up. J Foot Ankle Surg 2019; 58(1): 17-22. Cullen B, Stern A L, Weinraub G. Rate of revision after cheilectomy versus decompression osteotomy in early-stage hallux rigidus. J Foot Ankle Surg 2017; 56(3): 586-8. EuroQol Group. EuroQol–a new facility for the measurement of healthrelated quality of life. Health Policy 1990; 16(3): 199-208. Galois L, Hemmer J, Ray V, Sirveaux F. Surgical options for hallux rigidus: state of the art and review of the literature. Eur J Orthop Surg Traumatol 2020; 30(1): 57-65. Gerbert J, Moadab A, Rupley K F. Youngswick–Austin procedure: the effect of plantar arm orientation on metatarsal head displacement. J Foot Ankle Surg 2001; 40(1): 8-14. Johnsen L G, Hellum C, Nygaard O P, Storheim K, Brox J I, Rossvol I, Leis-

Acta Orthopaedica 2021; 92 (1): 108–112

veth G, Grotle M. Comparison of the SF6D, the EQ5D, and the Oswestry Disability Index in patients with chronic low back pain and degenerative disc disease. BMC Musculoskeletal Dis 2013; 14: 148. Lam A, Chan J J, Surace M F, Vulcano E. Hallux rigidus: how do I approach it? World J Orthop 2017; 8(5): 364-71. Lucas D E, Hunt K J. Hallux rigidus: relevant anatomy and pathophysiology. Foot Ankle Clin 2015; 20(3): 381-9. Sidon E, Rogero R, Bell T, McDonald E, Shakked R J, Fuchs D, Daniel J N, Pedowitz D I, Raikin S M. Long-term follow-up of cheilectomy for treatment of hallux rigidus. Foot Ankle Int 2019; 40(10): 1114-21. Slullitel G, Lopez V, Seletti M, Calvi J P, Bartolucci P, Pinton G. Joint preserving procedure for moderate hallux rigidus: does the metatarsal index really matter? J Foot Ankle Surg 2016; 55(6) 1143-7. Slullitel G, Lopez V, Calvi J P, D’Ambrosi R, Usuelli F G. Youngswick osteotomy for treatment of moderate hallux rigidus: thirteen years without. arthrodesis. Foot Ankle Surg 2019; S1268-7731(19)30208-5. Stevens J, de Bot R, Hermus J P S, van Rhijn L W, Witlox A M. Clinical outcome following total joint replacement and arthrodesis for hallux rigidus: a systematic review. JBJS Rev 2017; 5(11): e2. Waizy H, Czardybon M A, Stukenborg-Colsman C, Wingenfeld C, Wellman M, Windhagen H, Frank D. Mid- and long-term results of the joint preserving therapy of hallux rigidus. Arch Orthop Trauma Surg 2010; 130: 165-70. Youngswick F D. Modifications of the Austin bunionectomy for treatment of metatarsus elevatus associated with hallux limitus. J Foot Surg 1982; 21: 114-16.

Acta Orthopaedica 2021; 92 (1): 113–118

113

Comparison of outcome between nonoperative and operative treatment of medial epicondyle fractures Petra GRAHN 1, Tero HÄMÄLÄINEN 2, Yrjänä NIETOSVAARA 1,3, and Matti AHONEN 1 1 Department

of Pediatric Orthopedics and Traumatology, New Children’s Hospital, HUS Helsinki University Hospital, Helsinki; 2 Department of Orthopedics and Traumatology, HUS Helsinki University Hospital, Helsinki; 3 Department of Pediatric surgery, Kuopio University Hospital, Kuopio, Finland Correspondence: petra.grahn@hus.fi Submitted 2020-06-24. Accepted 2020-09-09.

Background and purpose — Controversy exists regarding the optimal treatment for displaced medial epicondyle fractures. We compared the results of nonoperative and operative treatment and calculated the incidence of medial epicondyle fractures in the pediatric census population. Patients and methods — 112 children under 16 years old who sustained > 2 mm displaced fracture of the medial epicondyle were treated in our institution between 2014 and 2019. 80/83 patients with 81 non-incarcerated fractures were available for minimum 1-year follow-up. 41 fractures were treated with immobilization only, 40 by open reduction and internal fixation, according to the preference of the attending surgeon. Outcome was assessed at mean 2.6 years (1–6) from injury with different patient-reported outcome measures. Elbow stability, range of motion, grip strength, and distal sensation were registered in 74/80 patients. Incidence was calculated for 7- to 15-year-olds. Results — Nonoperatively treated children had less pain according to the PedsQL Pediatric Pain Questionnaire (3 vs. 15, p = 0.01) with better cosmetic outcome (VAS 95 vs. 87, p = 0.007). There was no statistically significant difference between the groups in respect of QuickDASH, PedsQL generic core scale, Mayo Elbow Performance Score, grip strength, carrying angle, elbow stability, or range of motion (p > 0.05). All 41 nonoperatively treated children returned to pre-injury sports; of the surgically treated 6/40 had to downscale their sporting activities. The incidence of displaced (> 2 mm) fractures of the medial epicondyle in children aged 7–15 years was ≥ 3:100,000. Interpretation — Displaced fractures of the medial humeral epicondyle in children heal well with 3–4 weeks’ immobilization. Open reduction and screw fixation does not improve outcome.

Fractures of the medial humeral epicondyle have been reported to account for 12–20% of all pediatric elbow fractures, but the incidence is not known. Elbow dislocation is associated with 30–50% of these fractures (Gottschalk et al. 2012), with an incarceration rate of the fracture fragment into the elbow joint of 5–18%. Ulnar nerve lesions are registered in 10–16% of cases (Louhaem et al. 2010). Nonoperative treatment is advised in minimally displaced (< 2 mm) fractures of the medial humeral epicondyle, whereas surgery is recommended for fractures incarcerated in the elbow joint as well as for fractures that are either grossly unstable or where the ulnar nerve is entrapped (Smith 1950, Blount 1955, Maylahn and Fahey 1958, Bede et al. 1975, Gottschalk et al. 2012, Tarollo et al. 2015). Significant controversy concerning the treatment of displaced (3–15 mm) fractures exists, with some surgeons advocating early mobilization, some immobilization, and some internal fixation (Lee et al. 2005, Hughes et al. 2019, Pezzutti et al. 2020). It has also been suggested that competitive athletes or fractures occurring in combination with elbow dislocation should be treated surgically with a lower threshold than in children without sporting activities (Baety and Kasser 2014). The reported outcome of nonoperative and operative treatment in displaced fractures of the medial humeral epicondyle in terms of elbow function and complications has been similar (Farsetti et al. 2001, Biggers et al. 2015, Axibal et al. 2019). We compared subjective and objective outcomes and calculated the incidence of medial humeral epicondyle fractures in children treated either with immobilization or with open reduction and internal fixation (ORIF).

114

Figure 1. Medial epicondyle avulsion with displacement measures as described by Edmonds et al. (2010).

Patients and methods We conducted a controlled treatment trial based on prospectively collected data from consecutive patients identified from our institutional fracture registry. 112 (62 female) less than 16-year-old children who had sustained a more than 2 mm displaced fracture of the medial humeral epicondyle (modified ICD-10 code: S42.45) were treated in our tertiary level teaching hospital during a 6-year-long study period between January 2014 and December 2019. The incidence of displaced fractures of the medial humeral epicondyle was calculated in 7- to 15-year-olds in the catchment area, as nearly all children (109/112) who had sustained a fracture of the medial epicondyle were older than 6 years. Mean age of patients was 12 years (4–16). 34/112 (30%) patients had a trampoline injury. Patients with the medial epicondyle incarcerated in the elbow joint (n = 9) and patients with less than 1-year follow-up (n = 20) were excluded from the outcome study. All patients with partial avulsions (n = 4) were excluded from both studies. 81/83 remaining children with 82 fractures could be contacted and were available for follow-up (FU). Treatment method was chosen by the attending surgeon’s preference and cases were not uniformly presented in a consensus conference. 41 of the patients available for FU had been treated primarily nonoperatively and 40 by ORIF. 1 primarily nonoperatively treated patient underwent reduction and screw fixation of a malunited fracture due to pain under load 5 months from injury; this outcome data is not included in the group analysis. Patients medial elbow pain continued why the fixation screw together with a hypertrophic scar were removed at 9 months postoperative. At the last FU 1 year from surgery she still had pain under load. There was little difference in sex distribution (24/41 vs. 20/40 male patients) and mechanism of injury between the nonoperatively and surgically treated children (Table 1, see Supplementary data). Fracture displacement

Acta Orthopaedica 2021; 92 (1): 113–118

was calculated by the method described by Edmonds (2010) in 50 cases from AP radiographs only and in the remaining 31 radiographs from both AP and lateral view (Figure 1). The mean fracture displacement in nonoperatively treated patients measured from AP radiographs was 8 mm (3–12) vs. 7 mm (4–13) in the surgically treated children and from the lateral view respectively (n = 17 and 16) 9 mm (6–22) vs. 9 mm (6–16). 2 of the 3 patients with less than 5 mm of fracture displacement were treated nonoperatively. Nonoperatively treated patients were somewhat younger with a mean age of 11 years (4–16) vs. 12 (7–16) for the surgically treated, and their fractures were more often associated with clinically or radiologically documented elbow dislocation (19/41 vs. 10/40). Nonoperative treatment was carried out by immobilizing the injured upper extremity (1 bilateral injury) either with an above-elbow cast (n = 38) or a collar-and-cuff sling (n = 3) for a mean 24 days (19–34). Closed reduction was not attempted in the nonoperative group. Internal fixation was performed at mean 5 days (0–19) from the injury by a cannulated screw in 33, smooth pins in 6, or with a bone anchor in 1 patient. Bone anchors were additionally used in 2 instances, once in combination with a screw and once with pins. Half of the operations were done by an attending pediatric orthopedic surgeon or a pediatric surgeon, half by registrars. Mean length of postoperative immobilization was 30 days (21–44) either in an aboveelbow splint (n = 36) or with a collar-and-cuff sling (n = 5). All wounds healed uneventfully without recorded infections. The rate and timing of hardware removal was registered. 39 of the nonoperatively treated children were examined in our outpatient clinic and 2 interviewed by phone at mean 2.8 years (1–5) from the injury against 36 of the operated children, with 4 interviewed by phone at 2.4 years (1–6). Subjective outcome was assessed in 80 patients with QuickDASH (Beaton et al. 2005), Pediatric Quality of Life InventoryTM (PedsQL) Generic Core Scale, PedsQL Pediatric Pain Questionnaire (Varni et al. 1999), Mayo Elbow Performance Score (Morrey 1993), as well as a cosmetic visual analogue scale (VAS 0–100). Patients interviewed by phone were asked to answer the PedsQL Pain Questionnaire, Cosmetic VAS, and QuickDASH main and hobby module. Patients’ pre- and postinjury participation in non-organized and organized sports was registered. In the outpatient clinic carrying angle, and active and passive range of motion (ROM) of both elbows was measured using a goniometer. Stability of the elbow was assessed by the moving valgus test (O’Driscoll et al. 2005) and the valgus stress test (Flynn et al. 2008). Grip strength of both hands was recorded as the mean of 3 efforts with a hydraulic hand-held dynamometer. Distal sensation was examined by Semmes–Weinstein monofilaments (Bell-Krotoski 1990). Prevalence of cold intolerance was assessed. Statistics Data was analyzed using the Wilcoxon rank-sum test in Python 3.8 (Python Software Foundation, Wilmington, DE,

Acta Orthopaedica 2021; 92 (1): 113–118

115

95 vs. 87, p = 0.007). For carrying angle, elbow stability, extension deficiency, flexion deficiency, active or passive ROM and grip Non-operative ORIF Score n = 41 a n = 40 strength, we discovered no statistically significant differences Quick Dash 1.8 (0–13.6) SD 4 [0.1–2.9] 4.2 (0–23) SD 6 [2.5–5.9] between uninjured and injured Quick Dash hobby module 0.5 (0–6.3) SD 2 [0.0–1.1] 6.7 (0–100) SD 21 [0.3–13] PedsQL side within the group or between total score 89 (74–100) SD 9 [86–92] 90 (74–100) SD 7 [88–92] the groups (Table 2). A separate physical functioning analysis excluding the 3 patients health summary score 93 (75–100) SD 7 [90–95] 91 (75–100) SD 7 [89–93] pain module, VAS (0–100) b 3.3 (0–44) SD 8 [0.9–5.7] 15 (0–85) SD 22 [8–22] with < 5 mm displacement yielded Mayo Elbow Performance score 99 (85–100) SD 3 [98–100] 95 (80–100) SD 8 [93–97] the same results. Cosmetic score, VAS (0–100) 95 (64–100) SD 9 [92–98] 86 (35–100) SD 17 [81–92] All examined patients had a 1 bilateral normal sensation in the pulp b PedsQL Pain module score represents the worst pain the patient has experienced in the injured of their fingers as measured by elbow during the last 7 days. Semmes–Weinstein monofilament test, but 1 operatively treated child interviewed by phone reported Table 3. Results of clinical examination at last follow-up. Values are mean (range), SD, and [95% diminished sensation in the ulnar confidence interval]. Patients interviewed by phone excluded (n = 6). fingers. Another surgically treated patient had decreased sensation in Non-operative ORIF a 5 x 6 cm area distal to the scar. Outcome n = 39 a n = 36 Cold intolerance was reported by Carrying angle difference (°) 0.9 (0–8) SD 3 [0.3–1.8] 1.1 (0–5) SD 2 [0.4–1.9] 1 nonoperatively and 2 surgically Extension deficiency (°) 1.0 (0–15) SD 4 [–0.3 to 2.3] 3.0 (0–20) SD 6 [1.0–5.0] treated patients. Pain at the medial Flexion deficiency (°) 1.6 (0–10) SD 5 [0.1–3.1] 2.1 (0–15) SD 4 [0.9–3.3] Valgus stress test 3 unstable without pain 1 unstable without pain humeral epicondyle either with 2 stable with pain 4 stable with pain direct contact or under load was 1 unstable with pain reported by 4 nonoperatively and Moving valgus test 7 pain 9 pain by 6 operatively treated children a 1 bilateral with otherwise normal sensation and elbow stability (Table 3). All non-operatively treated patients USA). Our null hypothesis was that there is no difference had returned to the same or higher level of sport as pre-injury, in outcome between nonoperatively and operatively treated whereas 6 surgically treated patients had downgraded their sporting activities (Table 1, see Supplementary data). patients. Level of significance was set at p < 0.05. Pins were removed in the outpatient clinic from 5/6 children Ethics, funding, and potential conflict of interest: who had had their fractures pin fixed. 10 of the 33 children who Hospital ethical board approval was received in 1999. Exten- had their fractures fixed with a cannulated screw had had their sion permission for the study was approved on December 17, screws removed due to local pain at mean 16 months (7–29) 2015 (approval number HUS 621/1999). None of the authors from the injury. 6 children’s fractures were fixed in malposireceived any funding for the study and none of the authors tion (4 with screws, 2 with pins), but 5 of these 6 children were pain free with normal elbow stability and function. The trainreport any conflict of interest. ing level of the operating surgeon did not affect the outcome assessed by the different PROMs used (p > 0.05, Figures 2–3). Between 2014 and 2019, 525,966 children between the ages Results of 7–15 years lived in the catchment area (national registry). All patients completed the requested follow-up forms. There During the same period 76 children from the same area in was no statistically significant difference in the QuickDASH, the same age range had a displaced (> 2mm) fracture of the QuickDASH hobby module, PedsQL Physio Social Health medial humeral epicondyle, giving a mean annual incidence of Summary Score, PedsQL Physical Functioning Health Sum- 3:100,000. The peak incidence occurred at 11 years of age. The mary Score, PedsQL Total Score, or Mayo Elbow Perfor- real incidence may be slightly higher, as 10 additional chilmance Score. However, nonoperatively treated children had dren’s medial epicondyle fractures (city of residence unknown) less pain according to the PedsQL Pediatric Pain Question- had been treated in private clinics according to a survey connaire (3 vs. 15, p = 0.01) with better cosmetic outcome (VAS ducted among our region’s pediatric orthopedic surgeons. Table 2. Results of patient-reported outcome measures at last follow-up. Values are mean (range), SD, and [95% confidence interval]

116

Acta Orthopaedica 2021; 92 (1): 113–118

ture displacement could be measured only from the AP view and the actual displacement could thus be bigger. More reliable measurements could have been made from CT scans, which were not routinely taken. Our results suggest, however, that CT is unnecessary as the degree of fracture displacement does not seem to affect the outFigure 2. a, b. 11.8-year-old boy with 11 mm displaced fracture of the medial humeral epicondyle, which was come. treated with an above-elbow splint for 3 weeks. c, d. He had returned to climbing without pain and his elbow Louahem et al. (2010) was stable with a full range of motion at 4 years from injury. argued that damage to the main medial stabilizer of the elbow, the medial collateral ligament, has far greater influence on elbow joint stability and outcome than actual fracture displacement and recommended surgery in patients with a positive valgus stress test, regardless of amount of fracture displacement. We did not routinely examine elbow stability in our patients at Figure 3. a, b. 12.6-year-old gymnast’s 8 mm displaced fracture of the medial humeral epicondyle, which was time of injury, an examinaanatomically reduced and fixed with a well-positioned 4 x 45 mm cannulated screw. c, d. At follow-up 3 years from injury, she had returned to competitive gymnastics. She reported no pain and she had no functional prob- tion that often requires sedation. At follow-up there was lems, although her valgus stress test was positive (unstable without pain). no difference between the 2 treatment groups regarding stability of the elbow under valgus load. Nearly half of Discussion the nonoperatively treated children had an elbow dislocation, Operative treatment of pediatric medial epicondyle fracture which was a clearly higher rate than one-fourth in the surgihas gained popularity, although there is little evidence in sup- cally treated children, thus one could argue that good results port of surgical treatment over nonoperative. Decision to oper- in the nonoperatively treated group could in part be due to an ate is often based on the degree of displacement and mecha- intact ulnar collateral ligament as suggested by Gottschalk et nism of injury. Surgery is often recommended in displaced al. (2012). 9 of the 42 children with a medial epicondyle fracture fractures and in fractures sustained in association with elbow dislocation (Blount 1955, Maylahn and Fahey 1958, Vecsei et who were treated nonoperatively in the series of Smith et al. 1975, Josefsson and Danielsson 1986, Farsetti et al. 2001, al. (2010), developed a symptomatic nonunion in a 1-year follow-up. Contributing factors to the cause of the pain Lee et al. 2005, Lawrence et al. 2013, Pezzutti et al. 2020). Currently there is no consensus on how fracture displace- were not found. On the other hand, none of 139 surgically ment should be measured or how to define a clinically signifi- treated patients reported pain at mean 3.9 years follow-up in cant fracture displacement. Edmonds et al. (2010) have shown the series of Louhaem et al. (2009). However, Axibal et al. that measurements from plain radiographs are unreliable, and (2018) found no difference in elbow pain according to a phone argued that computer tomography (CT) should be used. Some survey at minimum 1.5 years after a medial epicondyle fracauthors advocate surgery in fractures with as little as 3 mm ture between 28 patients treated with cast immobilization and displacement (Vescei et al. 1975, Baety and Kasser 2014, 14 operatively treated children. Our findings contradict these Hughes et al. 2019), while most surgeons would consider previous reports because our nonoperatively treated patients operative treatment if it exceeds 5 mm (Smith 1950, Pezzutti had less pain than the operatively treated children. We also et al. 2020). In most of our patients’ plain radiographs frac- found that nonoperatively treated children were more pleased

Acta Orthopaedica 2021; 92 (1): 113–118

with the appearance of their injured elbow than children who had undergone surgery. There is little information concerning return to sport in children who have sustained a fracture of the medial humeral epicondyle. According to the study by Lawrence et al. (2013) there was no difference in outcome assessed by QuickDASH and elbow range of motion at 2 years from injury in 6 nonoperatively and 14 operatively treated athletes. Axibal et al. (2018) showed similar results with no difference in the objective outcome in less than 1-year follow-up between 22 operated patients matched with 22 nonoperated patients. All nonoperatively treated children returned to their previous sports, whereas 6 of the operatively treated patients could not continue their sports at all or returned to a lower level. It thus appears that the rate of return to sports cannot be improved by open reduction and pin or screw fixation of the fractured medial humeral epicondyle. This finding should be interpreted with caution, since patients were younger in the nonoperative group than in the operatively treated group. Medial humeral epicondyle fractures are reported to represented 1/5 of elbow fractures in children (Gottschalk et al. 2012, Baety et al. 2010). To our knowledge, the incidence of humeral medial epicondyle fractures in the pediatric census population has not been reported. We chose to calculate and report the incidence for 7- to 15-year-olds (≥ 3:100,000 in the catchment area) rather than for the entire pediatric population as 99% of patients with medial epicondyle fracture in the study population were 7 years or older. The most common injury mechanism was trampoline, followed by falls from height often associated with cheerleading or different types of gymnastics (Table 1, see Supplementary data). There are few prospective and no randomized controlled treatment trials for fractures of the medial humeral epicondyle in children. Most published studies represent small retrospective hospital-based patient series. The degree of fracture displacement or the stability of the elbow before commencing treatment is seldom registered. Strengths and limitations This is a comparative study of 81 consecutive children prospectively collected who had sustained a > 2 mm displaced medial epicondyle fracture treated by surgeon’s preference either by immobilization or by ORIF with a high follow-up rate, 81/83. Treatment was not randomized, which may cause a bias. Mean age of patients in the nonoperative group was lower than in the ORIF group. We do not have an obvious explanation for this discrepancy, but in general younger children less often require operative treatment in pediatric orthopedic trauma, which may have had an effect on selecting treatment modality. CTs had not been taken routinely and the exact fracture displacement could not therefore be measured. Regardless of treatment some patients remain symptomatic under valgus stress. This raises the question as to whether our treatment decisions are based on the right parameters, e.g.,

117

displacement of the fracture fragment vs. medial collateral ligament injury. In light of the shortcomings of this study we have been granted ethical review board permission to start a randomized control trial conducted as a non-inferiority trial. In summary, non-incarcerated fractures of the medial humeral epicondyle in children and adolescents can be safely and reliably treated nonoperatively with 3–4 weeks’ cast immobilization. The degree of primary fracture displacement or elbow dislocation does not seem to affect the outcome, since normal elbow functions are restored with few exceptions. Supplementary data Table 1 is available as supplementary data in the online version of this article, http://dx.doi.org/10.1080/17453674.2020. 1832312 PG conceived and designed the study, collected the data, performed the data analysis, wrote the manuscript. TH contributed to the design and implementation of the study, to the analysis of the results, and to the writing of the manuscript. Y N contributed to the analysis and presentation of the results and writing of the manuscript. MA conceived and designed the study, collected the data, and wrote the manuscript. All authors discussed the results and commented on the manuscript. Acta thanks Gunnar Hägglund and Bjarne Moeller-Madsen for help with peer review of this study.

Axibal D, Carry P, Skelton A, Mayer S. No difference in return to sport and other outcomes between operative and nonoperative treatment of medial epicondyle fractures in pediatric upper-extremity athletes. Clin J Sport Med 2018; 00: 1-5. [Online ahead of print]. Axibal D, Ketterman B, Skelton A, Carry P, Georgopoulos G, Miller N, Mayer S W. No difference in outcomes in a matched cohort of operative versus nonoperatively treated displaced medial epicondyle fractures. J Pediatr Orthop B 2019; 28(6): 520-25. Baety J H, Kasser J R. Chapter 15. The Elbow. In: Rockwood & Wilkins’ fractures in children (Eds. Waters P M, Skaggs D L, Flynn J M). 7th ed. Philadelphia, PA: Lippicott Williams & Wilkins; 2010; pp. 553-89. Beaton D E, Wright J G, Katz J N. Development of the Quick-DASH: comparison of three item-reduction approaches, J Bone Joint Surg Am 2005; 87: 1038-46. Bede W B, Lefebvre A R, Rosman M A. Fractures of the medial humeral epicondyle in children. Can J Surg 1975; 18(2): 137-42. Bell-Krotoski J. “Pocket filaments” and specifications for the Semmes–Weinstein monofilaments. J Hand Ther 1990; 3(1): 26-31. Biggers M D, Bert T M, Moisan A, Spence D D, Warner W C Jr, Beaty J H, Sawyer J R, Kelly D M. Fracture of the medial humeral epicondyle in children: a comparison of operative and nonoperative management. J Surg Orthop Adv 2015; 24(3): 188-92. Blount W. Fractures in children. London: Williams & Wilkins; 1955. Edmonds E. How displaced are “nondisplaced” fractures of the medial humeral epicondyle in children? Results of a three-dimensional computer tomography analysis. J Bone Joint Surg Am 2010; 92: 2785-91. Farsetti P, Potenza V, Caterini R, Ippolito E. Long-term results of treatment of fractures of the medial humeral epicondyle in children. J Bone Joint Surg Am 2001 83(9): 1299-305. Flynn T, Cleland J, Whitman J. User’s guide to the musculoskeletal examination: fundamentals for the evidence-based clinician. Buckner, KY: Evidence in Motion; 2008.

118

Gottschalk H P, Eisner E, Hosalkar H S. Medial epicondyle fractures in the pediatric population. J Am Acad Orthop Surg 2012; 20: 223-32. Hughes M, Dua K, O’Hara N N, Brighton B K, Ganley T J, Hennrikus W L, Herman M J, Hyman J E, Lawrence J T, Mehlman C T, Noonan K J, Otsuka N Y, Schwend R M, Shrader M W, Smith B G, Sponseller P D, Abzug J M. Variation among pediatric orthopaedic surgeons when treating medial epicondyle fractures. J Pediatr Orthop 2019; 39(8): e592-e596. Josefsson P O, Danielsson L G. Epicondylar elbow fracture in children: 35-year follow-up of 56 unreduced cases. Acta Orthop Scand 1986; 57: 313-15. Lawrence J T, Patel N M, Macknin J, Flynn J M, Cameron D, Wolfgruber H C, Ganley T J. Return to competitive sports after medial epicondyle fractures in adolescent athletes: results of operative and nonoperative treatment. Am J Sports Med 2013; 41: 1152-7. Lee H, Shen H, Lee C, Wu S. Operative treatment of displaced medial epicondyle fractures in children and adolescents. J Shoulder Elb Surg 2005; 14(2): 178-85. Louahem D, Bourelle S, Cottalorada J. Displaced medial epicondyle fractures of the humerus: surgical treatment and results: a report of 139 cases. Arch Orthop Trauma Surg 2010; 130: 649-55.

Acta Orthopaedica 2021; 92 (1): 113–118

Maylahn D, Fahey J. Fractures of the elbow in children: review of three hundred consecutive cases. J Am Med Assoc 1958; 166: 220-8. Morrey B F. The elbow and its disorders. Philadelphia: WB Saunders; 1993. O’Driscoll S, Lawton R, Smith A. The “moving valgus stress test” for medial collateral ligament tears of the elbow. Am J Sports Med 2005; 33(2): 231-9. Pezzutti D, Lin J, Singh S, Rowan M, Samora J. Pediatric medial epicondyle fracture management: a systematic review. J Pediatr Orthop 2020; February 14,. [Online ahead of print]. Smith F M. Medial epicondyle injuries. J Am Med Assoc 1950; 142: 396-402. Smith J, McFeely E, Bae D, Peter M, Micheli L, Kocher M. Operative fixation of medial epicondyle fracture nonunion in children. J Pediatr Orthop 2010: 30(7): 644-8. Tarollo L, Mugani R, Catani F. Pediatric medial epicondyle fractures with intra-articular elbow incarceration. J Orthop Traumatol 2015; 16: 117-23. Varni J, Seid M, Rode C. The PedsQL: measurement model for the pediatric quality of life inventory. Med Care 1999; 37(2): 126-39. Vecsei V, Pernecky A, Polterauer P. Teraphy of fracture of epicondylus medialis humeri. Arch Orthop Unfallchir 1975; 83(2): 233-43.

Acta Orthopaedica 2021; 92 (1): 119–126

119

Systematic review of complications with externally controlled motorized intramedullary bone lengthening nails (FITBONE and PRECICE) in 983 segments Markus W FROST 1,2, Ole RAHBEK 1,2, Jens TRAERUP 1, Adriano A CECCOTTI 1, and Søren KOLD 1,2 1 Department

of Orthopedic Surgery, Aalborg University Hospital, Aalborg; 2 Department of Clinical Medicine, Faculty of Medicine, Aalborg University, Aalborg, Denmark Correspondence: markus.frost@rn.dk Submitted . 2020-04-23 Accepted 2020-09-19

Background and purpose — In recent years motorized intramedullary lengthening nails have become increasingly popular. Complications are heterogeneously reported in small case series and therefore we made a systematic review of complications occurring in lower limb lengthening with externally controlled motorized intramedullary bone lengthening nails. Methods — We performed a systematic search in PubMed, EMBASE, and the Cochrane Library with medical subject headings: Bone Nails, Bone Lengthening, and PRECICE and FITBONE nails. Complications were graded on severity and origin. Results — The search identified 952 articles; 116 were full text screened, and 41 were included in the final analysis. 983 segments were lengthened in 782 patients (age 8–74 years). The distribution of nails was: 214 FITBONE, 747 PRECICE, 22 either FITBONE or PRECICE. Indications for lengthening were: 208 congenital shortening, 305 acquired limb shortening, 111 short stature, 158 with unidentified etiology. We identified 332 complications (34% of segments): Type I (minimal intervention) in 11% of segments; Type II (substantial change in treatment plan) in 15% of segments; Type IIIA (failure to achieve goal) in 5% of segments; and Type IIIB (new pathology or permanent sequelae) in 3% of segments. Device and bone complications were the most frequent. Interpretation — The overall risk of complications was 1 complication for every 3 segments lengthened. In 1 of every 4 segments, complications had a major impact leading to substantial change in treatment, failure to achieve lengthening goal, introduction of a new pathology, or permanent sequelae. However, as no standardized reporting method for complications exists, the true complication rates might be different.

Distraction osteogenesis through an externally applied fixator is a well-established treatment for lower limb lengthening (De Bastiani et al. 1987, Paley 1988, Ilizarov 1990). However, complication rates of this treatment are high, amounting to 1–3.2 complications per patient (Tjernström et al. 1994, Noonan et al. 1998). The wires or pins penetrating soft tissues result in complications such as pin site infection, pain, scarring, muscle transfixation, reduced joint movement, and immobility (Paley 1990, Mazeau et al. 2012, Landge et al. 2015). When the external fixator is removed, there is a risk of further complications such as fracture or malalignment (Noonan et al. 1998, Simpson and Kenwright 2000). To reduce complications and improve patient comfort, limb lengthening by fully implantable bone lengthening nails has been introduced (Guichet 1999, Cole et al. 2001). Problems with purely mechanically driven lengthening nails were resolved by the introduction of motorized (FITBONE) or magnetically driven (PRECICE) bone lengthening nails (Baumgart et al. 1997, Kirane et al. 2014, Paley et al. 2014, Shabtai et al. 2014). A few case-control studies have compared these nails with external fixation (13–15 patients), and the largest case series on intramedullary bone lengthening reports on 92 patients (Black et al. 2015, Horn et al. 2015, Calder et al. 2019). However, the majority of reports of complications of the FITBONE and PRECICE lengthening nails are small case series (Krieg et al. 2008, Dinçyürek et al. 2012, Birkholtz and De-Lange 2016, Hammouda et al. 2017). In recent years motorized intramedullary lengthening nails have become increasingly popular, and we thus hypothesized that standardized data on complications could now be extracted from the literature. We performed a systematic literature review of complications using PRECICE and FITBONE bone lengthening nails in lower limb bone lengthening. The primary outcome was risk of complications imposing a new pathology or permanent sequelae in the patient.

120

Acta Orthopaedica 2021; 92 (1): 119–126

Table 1. Classification of severity of complications in accordance with Black et al. and Paley Modified Black et al. 2015

Complication severity grade Paley 1990

I Minimal intervention required; Problems Potential expected difficulty arising treatment goal still achieved during distraction or fixation period which is fully resolved non-operatively by end of the treatment period II Substantial change in treatment Obstacle Potential expected difficulty that arose plan; treatment goal still achieved doing distraction or fixation period that is fully resolved by end of the treatment period by operative means IIIA Failure to achieve treatment goal; Complication Complication include any local or no new pathology or permanent systemic intraoperative or perioperative sequelae. Peri- or intraoperative complication, difficulty during distraction complication without sequelae or fixation that remains unsolved at the end of treatment period, and any early or late post-treatment difficulty IIIB Failure to achieve treatment goal Complications were divided into minor and/or new pathology or and major depending on whether the permanent sequelae original treatment goal was achieved

Method Search criteria An electronic search in the databases PubMed, Embase, and Cochrane Library was performed by a health science librarian with expertise in systematic literature searching. For details of the search strategy see Supplementary data 1. There was no limit concerning study design, publishing date, or language. We searched reference lists of included studies, relevant reviews identified through the systematic search and authors’ personal files to ensure literature saturation. Inclusion and exclusion criteria We included only published full-text original studies designed as randomized controlled trials, prospective and retrospective cohort studies, case-control studies, case series, and case reports. Cross-sectional studies were excluded. Studies in both English and German were included. Studies were included if: the bone lengthening nails applied were FITBONE (Wittenstein Intens GmbH, Igersheim, Germany) and/or PRECICE (Nuvasive, San Diego, CA, USA), conducted in humans, and bone lengthening was performed on lower extremities. Descriptions of complications included origin, severity, and management of complications or a statement of no complications. Studies were excluded if: reporting only bone transport treatment, nails were used only for compression, there was no involvement of lower extremities, or reporting stump lengthening. If patients were represented in more than one study, only one of the studies was included. A

Examples of complications Pin-site infection. Temporary joint contracture

Unplanned return to surgery, such as delayed consolidation requiring additional intervention, and device problem needing revision Premature consolidation with aborted lengthening, inability to tolerate lengthening, and fracture at fixation site or regenerate bone with shortening Joint subluxation, joint dislocation, regenerate fracture with deformity, and deep infection. Thromboembolic complication such as deep vein thrombosis

single patient or a group of patients from one study could be included if patient-/group-specific data was available. Data collection and management The primary search was performed at the end of November 2019 and updated at the end of March 2020. The literature search was assembled in www.covidence.org as well as the management of article selection flow. Titles and abstracts were screened by the first author (MWF) to select articles for full text reading. Among the full text articles, MWF selected papers for possible inclusion. SK and MWF assessed articles in accordance with inclusion criteria and agreed on studies relevant for final inclusion. During the initial data collection, MWF collected the following information from each study: title, author(s), year of publication, study design, evidence level, number of patients, number of lengthening segments, sex, nail type (FITBONE or PRECICE), participant age range, and bone segments (femur or tibia). Etiology was divided into 3 groups: (1) congenital, (2) short stature, (3) acquired/developed limb length discrepancy diagnoses in accordance with the modified Stricker and Hunt classification (Stricker and Hunt 2004) (see Supplementary data 2), min./max. leg length, and perioperative soft tissue release. Complications were assessed according to the particular point in time when they occurred: intraoperative complication (Early 1:E1), postoperative complication prior to distraction start (Early 2:E2), during distraction period (Late 1:L1), after end of distraction and prior to implant removal (Late 2:L2), and after implant removal (Late 3:L3) (see Supplementary data 3). The severity of complications was classified

Acta Orthopaedica 2021; 92 (1): 119–126

according to Black et al. (2015) (Table 1). If a complication was graded according to Paley, we used Table 1 to compile the complication into the Black classification. If the treatment of a complication was not thoroughly described, we generally downgraded it, assuming that the treatment goal was achieved and no new pathology or permanent sequalae had emerged. As an example, a joint contracture with no described changes in treatment was classified as grade I. If a complication was graded by article authors as grade I, but the described treatment included additional surgery, we graded it as grade II. A deep vein thrombosis was graded as grade IIIB. The type of complication was categorized into origin representing 8 main groups (soft tissue, joint, vascular, bone, neurological, infection, device-related, others) and 33 subgroups according to Table 5 (for specific examples see Supplementary data 4). Intra-articular nail placement causing irritation and residual deformity was categorized into origin as Others/Surgical. Patient requesting to stop the lengthening procedure was categorized into origin as Others/Patient. MWF identified all complications and graded them according to severity, time of treatment, and origin. A second reviewer (SK) subsequently evaluated and graded the complication concerning severity and origin. Disagreement between reviewers was solved by consensus discussion. The Oxford Centre for Evidence-Based Medicine—Levels of Evidence 2009 grading of Harm was used to assess the level of evidence in the included studies (case reports were not included). A study was classified as a case report if reporting less than 5 bone lengthening segments. Case series with a subgroup analysis were classified as a cohort study. A study was considered prospective if data were collected prospectively; all other studies were considered retrospective. We used a methodology quality assessment score for all studies: Methodological Index for Non-Randomized Studies (MINORS) for non-randomized studies; Murad et al. for case reports (Slim et al. 2003, Murad et al. 2018). MWF and AG independently assessed the studies and solved difference through discussion. 3 specific questions concerning harms (from the McHarm scale) were used (Santaguida et al. 2011, Kronick et al. 2014) (see methodology quality assessment score, Supplementary data 5). Statistics Microsoft Excel 2019 version 16.33 (Microsoft Corp, Redmond, WA, USA) was used for data storage and descriptive analysis. Inter-rater agreement between the 2 assessors of complications was calculated as Kappa values for both severity grading of complications (4 types) and categorization of origin (8 main groups and 33 sub-groups) with Stata/MP 15.1 (StataCorp, College Station, TX, USA). For the strength of agreement, values less than 0 were rated as Poor; 0–0.20 Slight; 0.21–0.40 Fair; 0.41–0.60 Moderate; 0.61–0.8 Substantial; and 0.80–1 Almost perfect (Landis and Koch 1977).

121

Studies identified through database searching (n = 831): – Pubmed, 510 – Embase, 310 – Cochrane library, 11

Additional studies identified through other sources n = 121

Duplicates removed n = 330 Abstracts screened after duplicate removal n = 622 Studies excluded n = 506 Full-text articles assessed for eligibility n = 116 Full-text articles excluded (n = 75): – wrong publication types, 32 – other operation types, 16 – other nail types, 11 – patients included in more than 1 study, 6 – wrong outcomes, 6 – wrong indications, 3 – wrong study designs, 2 – insufficient description of complications, 1 Studies included in qualitative synthesis (n = 41): – case series, 26 – case reports, 8 – cohort studies, 6 – case-control study, 1

Flow diagram of selection of studies.

Registration, funding, and potential conflicts of interest Prior to conducting the study, we searched the PROSPERO database (http://www.crd.york.ac.uk/PROSPERO) for ongoing reviews and recently completed systematic reviews; we did not identify any results. This study was then submitted to PROSPERO on November 22, 2019. Due to waiting time at PROSPERO, this pre-study registration has unfortunately not been published before submission (April 23, 2020). During the review period, the pre-study registration was published by PROSPERO (ID number: CRD42020159272). The majority of studies included were case series and case reports and we changed the risk of bias/quality assessment tool to MINORS and added Murad et al. for case reports (Slim et al. 2003, Murad et al. 2018). Thus, reporting guidelines were also changed to Meta-analysis of Observational Studies in Epidemiology (MOOSE), and this systematic review was organized in agreement with this (Stroup et al. 2000). The change was submitted to PROSPERO. The authors’ institutions funded the study. No conflicts of interest are declared.

Results Our search identified 952 articles of which 41 were included (Table 2); for flowchart of article selection see Figure. There was 1 case-control study, 6 cohort studies, 26 case series, and 8 case reports. Of the 33 studies that were not case reports,

122

Acta Orthopaedica 2021; 92 (1): 119–126

Table 2. Included studies with the corresponding number of patients, segments, and complications used in this review Reference

Pro- Segspec- Patients ments Nail Complitive n n type a cations

Accadbled et al. 2019 Accadbled et al. 2016 Al-Sayyad 2012 Baumgart et al. 1997 Baumgart et al. 2005 Birkholtz and De-Lange 2016 Black et al. 2015 Calder et al. 2019 Cosic and Edwards 2020 Couto et al. 2018 Dinçyürek et al. 2012 Frommer et al. 2018 Haider and Wozasek 2019 Hammouda et al. 2017 Harkin et al. 2018 Havitcioglu et al. 2020 Horn et al. 2019 Iobst et al. 2018 Karakoyun et al. 2016 Karakoyun et al. 2015 Kariksiz and Karakoyun 2019 Kirane et al. 2014 Krieg et al. 2008 Krieg et al. 2011 Küçükkaya et al. 2015 Laubscher et al. 2016 Lee et al. 2017 Lenze et al. 2011 Morrison and Sontich 2016 Muratori et al. 2018 Nasto et al. 2020 Paley et al. 2015 Paley et al. 2014 Rozbruch 2017 Schiedel et al. 2014 Shabtai et al. 2014 Singh et al. 2006 Steiger et al. 2018 Tiefenboeck et al. 2016 Wiebking et al. 2016 Wu and Kuhn 2018 aF

Yes Yes Yes No No No No No No No No No No No No No No No No No No No Yes No No No No No No No Yes No No No Yes Yes No No No No No

5 23 10 11 1 9 13 92 21 1 14 54 20 17 3 8 47 27 23 22 1 24 8 32 22 15 41 11 1 4 26 51 46 2 24 18 10 5 10 9 1

7 26 14 11 3 11 15 107 21 2 15 60 20 17 3 16 50 27 27 22 1 25 8 32 25 20 80 11 1 4 26 116 62 2 26 21 24 5 10 9 1

F F F F F P F P P P F P P P P P/ F P/ F P P F/P P P F F F P P F P P P P P P P P F F P P P

1 9 1 6 0 2 20 31 9 0 12 7 8 4 0 4 16 4 10 2 0 6 4 10 5 5 36 6 1 1 10 20 31 1 9 9 14 2 13 2 1

= FITBONE; P = PRECICE.

there was 1 level 3 study and 32 level 4 studies. The mean MINORS score was 8.3 (n = 26, range 5–12, ideal score 16) for non-comparative studies and 15.1 (n = 7, range 12–18, ideal score 24) for comparative studies (for full score of studies, see Supplementary data 6). The mean score for case reports was 4.3 (n = 8, 3–6 range, ideal score 8). Concerning the McHarm questions: (1) 1 study included predefined/standardized descriptions of complications, (2) standard scale of complications was used in 15 studies, and (3) number of each type of event and total number were specified on study groups in 31 studies (see Supplementary data 6). The 41 studies included 782 patients and 983 bone lengthening segments (Table 3). We found 332 complications cor-

Table 3. Descriptive study data collected from the studies reporting group-level data Factor Numbers

Studies reporting data

Number of patients 782 41 Number of bone segments 983 41 Male / female, n 384 / 234 29 / 33 Age, min / max 8 / 74 39 Etiology, n Congenital disease 208 22 Short stature 111 14 Acquired/developmental LLD 305 29 Femur / tibia, n 813 / 170 40 / 28 Bone lengthening cm, min / max 1 / 14 35 / 35 FITBONE/ PRECICE nails 214 / 747 15 / 27 164 patients were unidentified regarding gender, and 158 patients regarding etiology. 22 nails could not be differentiated between FITBONE or PRECICE. LLD: limb length discrepancy

Table 4. Severity grading of complications divided into specific numbers and percentages of lengthened segments and patients

Factor

Severity grade of complications I II IIIA IIIB Sum

Number of complications 113 Complications per segment, % 11 Complications per patient, % 14

146 15 19

45 5 6

28 3 4

332 34 42

Grading according to severity by Black et al. (2015).

responding to 34% of segments; 14 complications were not classified with origin, only severity. We observed 28 type IIIB complications, which was our primary outcome, corresponding to 3% of segments. Type IIIA complications not achieving the lengthening goal were seen in 45 cases (5% of segments). There were 113 type I complications and 146 type II complications, corresponding to 11% and 15% complications per segment, respectively (Table 4). Device-related complications (12% of segments) were the most frequent type of complication followed by bone (8% of segments) and then joint complications (6% of segments) (Table 5). 5 studies reported a systematic approach to soft-tissue release during primary surgery (Shabtai et al. 2014, Paley et al. 2015, Laubscher et al. 2016, Rozbruch 2017, Calder et al. 2019). None of the 41 studies systematically reported the timing of the complication; in 332 complications, timing was established in 177 (53%) cases with 6 and 5 complications of E1 and E2, respectively. L1 and L2 were seen in 85 and 81 cases, respectively and no L3 complications were found. In 18 (8 of these were case reports) of the 41 studies, it was possible to connect the complication and the individual patient data. These 18 studies represent only 160 patients and we considered this number too low for subgroup analysis of complica-

Acta Orthopaedica 2021; 92 (1): 119–126

Table 5. Complications categorized into 8 main groups (soft tissue, joint, vascular, bone, neurological, infection, device-related, others) and 33 subgroups Group

Severity grade and origin of complications I II IIIA IIIB Sum

Soft tissue Skin 2 1 3 Muscles 0 Tendons 0 Pain 5 5 Others 2 1 2 (CS) 5 Sum of soft tissue 13 Soft tissue complications in % of segments 1 Joint Pain 1 1 Contracture 19 24 5 5 53 Subluxation 6 6 Dislocation 1 1 Others 0 Sum of joint 61 Joint complications % of segment: 6 Vascular Vascular damage 1 1 Deep vein thrombosis 4 4 Hemorrhage/hematoma 2 2 Others 2 1 (AV) 3 Sum of vascular 10 Vascular complications in % of segments: 1 Bone Premature consolidation 15 4 19 Delayed healing 16 27 2 1 46 Secondary malalignment 1 2 3 Fracture 6 1 1 8 Others 1 1 2 Sum of bone 78 Bone complications in % of segments: 8 Neurology Paresthesia 2 1 2 5 Paralysis 0 Others 3 3 Sum of neurology 8 Neurology complication in % of segments: 0.8 Infection Superficial soft tissue 2 1 3 Deep soft tissue 1 1 Osteomyelitis 3 1 4 Others 0 Sum of infection 8 Infection complications in % of segments: 0.8 Device-related Distraction mechanism 16 20 9 45 Mechanical strength 25 14 3 2 44 Attachment failure 8 24 1 33 Others 0 Sum of device-related 122 Device-related complications in % of segments: 12 Others Patient 6 6 Surgical 3 7 1 11 Others 1 1 Sum of others 18 Others, complications in % of segments 1.8 CS: compartment syndrome; AV: arteriovenous fistula of the posterior tibial artery decompensated during tibial lengthening and an embolization procedure had to be performed. 14 complications could not be categorized due to missing descriptions.

123

tion risks. A few possible risk factors for complications could be estimated at a study group level. We found 31% complications per segment for the PRECICE nail and 46% complications per segment for the FITBONE nail. Surgical unit experience was assessed by dividing studies into studies with less than 20 patients (49% complications per segment) and studies with more than 40 patients (30% complications per segment) (see Supplementary data 6 for studies included in sub-analysis and Supplementary data 7 for full data presentation). The inter-rater agreement between the 2 assessors of complications was 0.87 for severity grading and 0.94 for categorization of origin.

Discussion To our knowledge, this is the first systematic review on complications related to bone lengthening nails. The primary outcome was the risk of type IIIB complications resulting in a new pathology or permanent sequelae. This review found such IIIB complications in 3% of lengthened segments. Furthermore, a complication of any type was found in 34% of lengthened segments, and 5% of segments did not achieve the planned lengthening due to a complication (IIIA). In 15% of segments treated with intramedullary PRECICE and FITBONE lengthening nails, a complication (II) resulted in substantial change in treatment, such as unplanned re-surgery. 6% (11/177) of time-determined complications occurred intra- or perioperatively prior to start of distraction, and 94% of complications (166/177) occurred during or after the end of distraction. The high diversity of complications demonstrates that several means must be applied to reduce the high number of complications in intramedullary bone lengthening. Concerning the primary outcome, where the (type IIIB) complication resulted in a new pathology or permanent sequelae, the majority of complications were a result of joint-related complications such as contracture, subluxation, or dislocation. It is likely that a reduction in joint-related complications is accomplished by improved patient selection and attention to soft-tissue release as well as individualized protocols for lengthening, temporary extraarticular screw arthrodesis, splints/orthoses, or physiotherapy. The risk of joint subluxation and dislocation was 6 and 1 per 1,000 segments, respectively. Joint contracture was seen in 5% (53/983) of the segments, and primary soft-tissue release might be a key to address this complication; this was, however, only reported in 5 of the 41 studies (Shabtai et al. 2014, Paley et al. 2015, Laubscher et al. 2016, Rozbruch 2017, Calder et al. 2019). Calder et al. made a systematic division of the iliotibial band (ITB) if the planned lengthening was above 3 cm. They found that, in femoral lengthening, females lost joint movement in the hip and knee earlier than males. Moreover, it took substantially more time to regain range of motion in patients treated with retrograde compared with antegrade nails. However, we believe that higher rates of severe joint

124

complications must be anticipated in high-risk patients such as congenital femoral deficiency and fibular hemimelia. We believe there is a need for systematic reporting of primary soft-tissue release as there is a lack of knowledge of benefits and challenges concerning this issue. A device-related complication was seen in 12% (122/983) of segments with 5% (45/983) assigned to distraction mechanism-related complications, and 1% (13/983) of segments did not reach the lengthening goal due to device-related type IIIA complications. The overall complication rate per segment was 46% for studies only reporting the use of a FITBONE nail and 31% for studies only reporting the use of a PRECICE nail. However, the quality of data is not sufficient to compare complication rates between the 2 nail types; for example, did the studies that used only FITBONE nails include tibial lengthening in 27% compared with 16% in the PRECICE studies. In addition, in the FITBONE group the average number of patients per study was only 13 compared with an average number of 28 patients per study in the PRECICE group. However, the relatively high rate of device-related complications shown in this review warrants a constant focus on the technology of bone lengthening nails. Future studies should specifically report the type and generation of the applied nail to assess complication risk related to different nails and generations. Complications related to bone regeneration were mainly due to delayed healing in 5% (46/983) of segments or premature consolidation in 2% (19/983) of segments. These complications might be reduced by increasing knowledge and handling of nail stability, patient compliance, mobilization, and biological factors such as type of osteotomy, latency period, and distraction rate/force. Another solution might be providing realtime feedback on surrogate markers of bone healing to allow for individualized distraction treatment. It seems logical that a surgeon’s ability to avoid or recognize, manage, and solve complications strongly correlates with the surgeon’s experience of this highly specialized treatment. This was to some extent supported by this review as studies with fewer than 20 patients had more complications per segment compared with studies with more than 40 patients. The validity of a review depends on the quality of the included studies and on the validity of the data extraction. The level of evidence in the studies included in this review was low. Of the 41 included studies, there were 1 level 3 study, 32 level 4 studies, and 8 case reports and mean MINORS scores of about half of the ideal scores. Our study found 146 type II complications compared with 113 type I complications, and most patients thus had a more complex type of complication. This might reflect both underreporting and the lack of accurate reporting of complications in elective surgery (Martin et al. 2002). We assessed complications in relation to segment lengthening and not to each patient because, in some patients, bone lengthening occurred in more than 1 bone in the same leg,

Acta Orthopaedica 2021; 92 (1): 119–126

lengthening involved both legs (short stature patients), and some patients underwent multiple lengthening procedures of the same bone. Complications per segment were lower than complications per patient, but since most of the patients had lengthening of only one segment, segment was chosen for main reporting. 41 studies with 983 bone lengthening segments reported either complications or stated absence of complications. With the increased popularity of lengthening by PRECICE and FITBONE nails, there is a knowledge gap concerning the distribution of severity grade and origin of complications in all treated patients. We believe that the demographics and number of included patients in this review are sufficiently diverse to illuminate even rare complications. 4 different classifications for reporting severity in bone lengthening complications were used (Paley 1990, Dahl et al. 1994, Dinçyürek et al. 2012, Black et al. 2015), and 29 studies did not use a classification. We are familiar with at least 4 more classifications of complications in limb lengthening, which challenges comparison between reported complications (Caton et al. 1985, Popkov 1991, Donnan et al. 2003, Lascombes et al. 2012). In this review we have classified the reported complications to achieve consistent reporting. However, it is a limitation that data on complications could be classified only from reported complications and not from original data. In the case of uncertainty between different grades of a complication, the complication was graded with the lower severity grade. Thereby, a systematic risk of reporting too low a complication severity grade was introduced. Another limitation of our review is that the reported complication rates could not be specified on subgroup level. We would expect that the complication rates differ substantially between a patient with idiopathic lower limb lengthening undergoing 3 cm of simple antegrade femoral lengthening without deformity correction and a patient with congenital fibular hemimelia and multiple previous operations undergoing 5 cm of tibial lengthening. However, it was not possible to extract data on a single patient level from the current literature. Therefore, we could not make correlations between complication rates and individual risk factors. We encourage future studies to report complications on a single patient level where complications can be related to possible risk factors such as age, diagnosis/etiology, segment, approach, nail type nail, nail generation, and timing of complications. Conclusion This review of the literature shows an overall high rate of complications, with complications occurring in 1 of every 3 segments undergoing lower limb lengthening. In 1 of every 4 segments, complications have a major impact leading to substantial change in treatment (15%), failure to achieve lengthening goal (5%), or introduction of a new pathology or permanent sequelae (3%). As no standardized method of reporting complications exists, the true complication rate might be different.

Acta Orthopaedica 2021; 92 (1): 119–126

A standardized reporting method would substantially improve the knowledge needed to reduce the rate of complications. Supplementary data Supplementary data items 1–7 are available in the online version of this article, http://dx.doi.org/10.1080/17453674.2020. 1835321 Conceptualization: MWF, JT, SK. Data extraction: MWF, SK, AAC. Formal analysis: MWF. Supervision: SK, OR. Writing original draft: MWF. Writing, reviewing, and editing: MWF, AAC, SK, OR, JT. The authors would like to thank Arash Ghaffari, MD, for competent help with study quality assessment. Acta thanks Joachim Horn and Björn Vogt for help with peer review of this study. Accadbled F, Pailhé R, Cavaignac E, Sales de Gauzy J. Bone lengthening using the Fitbone motorized intramedullary nail: the first experience in France. Orthop Traumatol Surg Res 2016; 102(2): 217-22. Accadbled F, Thévenin Lemoine C, Poinsot E, Baron Trocellier T, Dauzere F, Sales De Gauzy J. Bone reconstruction after malignant tumour resection using a motorized lengthening intramedullary nail in adolescents: preliminary results. J Child Orthop 2019; 13(3): 324-9. Al-Sayyad M J. Lower limb lengthening and deformity correction using the Fitbone motorized nail system in the adolescent patient. J Pediatr Orthop Part B 2012; 21(2): 131-6. De Bastiani G, Aldegheri R, Renzi-Brivio L, Trivella G. Limb lengthening by callus distraction (callotasis). J Pediatr Orthop 1987. p 129-34. Baumgart R, Betz A, Schweiberer L. A fully implantable motorized intramedullary nail for limb lengthening and bone transport. Clin Orthop Relat Res 1997; (343): 35-43. Baumgart R, Bürklein D, Hinterwimmer S, Thaller P, Mutschler W. The management of leg-length discrepancy in Ollier’s disease with a fully implantable lengthening nail. J Bone Joint Surg Br 2005; 87-B(7): 1000-4. Birkholtz F, De-Lange P. Evaluation of the first experience of intramedullary nail lengthening using PRECICE in a South African limb lengthening and reconstruction unit. South African Orthop J 2016; 15(1): 67-71. Black S R, Kwon M S, Cherkashin A M, Samchukov M L, Birch J G, Jo C-H H. Lengthening in congenital femoral deficiency: a comparison of circular external fixation and a motorized intramedullary nail. J Bone Joint Surg Am 2015; 97(17): 1432-40. Calder P R, McKay J E, Timms A J, Roskrow T, Fugazzotto S, Edel P, Goodier W D. Femoral lengthening using the Precice intramedullary limblengthening system. Bone Joint J 2019; 101-B(9): 1168-76. Caton J, Dumont P, Berard J, Michel C. Intermediate results of a series of 33 cases of leg lengthening using H. Wagner’s technique. Rev Chir Orthop Reparatrice Appar Mot 1985; 71(Suppl. 2): 44-8. Cole J D, Justin D, Kasparis T, DeVlught D, Knobloch C. The intramedullary skeletal kinetic distractor (ISKD): first clinical results of a new intramedullary nail for lengthening of the femur and tibia. Injury 2001; 32(Suppl. 4): SD-129-39. Cosic F, Edwards E. PRECICE intramedullary nail in the treatment of adult leg length discrepancy. Injury 2020; Epub ahead of print. Couto A, Freitas J, Alegrete N, Coutinho J, Costa G. Two consecutive limb lengthenings with the same PRECICE nail: a technical note. Strateg Trauma Limb Reconstr 2018; 13(3): 199-204. Dahl M T, Gulli B, Berg T. Complications of limb lengthening: a learning curve. Clin Orthop Relat Res 1994; (301): 10-18.

125

Dinçyürek H, Kocaoǧlu M, Eralp I L, Bilen F E, Dikmen G, Eren I. Functional results of lower extremity lengthening by motorized intramedullary nails. Acta Orthop Traumatol Turc 2012; 46(1): 42-9. Donnan L T, Saleh M, Rigby A S, Bank W, Sheffield S. Acute correction of lower limb deformity and simultaneous lengthening with a monolateral fixator. J Bone Joint Surg 2003; 85(2): 254-60. Frommer A, Rödl R, Gosheger G, Vogt B. Application of motorized intramedullary lengthening nails in skeletally immature patients: indications and limitations. Unfallchirurg 2018; 121(11): 860-7. Guichet J M. Beinverlängerung und Deformitätenkorrektur mit dem FemurAlbizzia Nagel. Orthopade 1999; 28(12): 1066-77. Haider T, Wozasek G E. Repeated intramedullary stabilization following failed telescopic nail lengthening: an appropriate treatment strategy. Injury 2019; 50(11): 2060-4. Hammouda A I, Jauregui J J, Gesheff M G, Standard S C, Conway J D, Herzenberg J E. Treatment of post-traumatic femoral discrepancy with PRECICE magnetic-powered intramedullary lengthening nails. J Orthop Trauma 2017; 31(7): 369-74. Harkin E, Rozbruch S R, Liskutin T, Hopkinson W, Bernstein M. Total hip arthroplasty and femoral nail lengthening for hip dysplasia and limb-length discrepancy. Arthroplasty Today 2018; 4(3): 279-86. Havitcioglu H, Gursan O, Isin Y. Cosmetic bilateral leg lengthening using intramedullary nail experience of 9 cases. J Orthop 2020; 20: 232-5. Horn J, Grimsrud Ø, Dagsgard A H, Huhnstock S, Steen H. Femoral lengthening with a motorized intramedullary nail: a matched-pair comparison with external ring fixator lengthening in 30 cases. Acta Orthop 2015; 86(2): 248-56. Horn J, Hvid I, Huhnstock S, Breen A B, Steen H. Limb lengthening and deformity correction with externally controlled motorized intramedullary nails: evaluation of 50 consecutive lengthenings. Acta Orthop 2019; 90(1): 81-7. Ilizarov G A. Clinical application of the tension-stress effect for limb lengthening. Clin Orthop Relat Res 1990; (250): 8-26. Iobst C A, Rozbruch S R, Nelson S, Fragomen A. Simultaneous acute femoral deformity correction and gradual limb lengthening using a retrograde femoral nail: technique and clinical results. J Am Acad Orthop Surg 2018; 26(7): 241-50. Karakoyun Ö, Küçükkaya M, Erol M F. Does lengthening after acute correction negatively affect bone healing during distraction osteogenesis? Acta Orthop Traumatol Turc 2015; 49(4): 405-9. Karakoyun O, Sokucu S, Erol M F, Kucukkaya M, Kabukcuoglu Y S. Use of a magnetic bone nail for lengthening of the femur and tibia. J Orthop Surg 2016; 24(3): 374-8. Kariksiz M, Karakoyun O. Limb lengthening with one Precice nail over its capacity. Saudi Med J 2019; 40(10): 1058-62. Kirane Y M, Fragomen A T, Rozbruch S R. Precision of the PRECICE internal bone lengthening nail. Clin Orthop Relat Res 2014; 472(12): 3869-78. Krieg A H, Lenze U, Speth B M, Hasler C C. Intramedullary leg lengthening with a motorized nail: indications, challenges, and outcome in 32 patients. Acta Orthop 2011; 82(3): 344-50. Krieg A H, Speth B M, Foster B K. Leg lengthening with a motorized nail in adolescents: an alternative to external fixators? Clin Orthop Relat Res 2008; 466(1): 189-97. Kronick R, Slutsky J, Stephanie C. Methods guide for effectiveness and comparative effectiveness review. AHRQ Publ. No. 10(14)-EHC063-EF. Rockville, MDL Agency Healthc Res Qual; 2014. Küçükkaya M, Karakoyun Ö, Sökücü S, Soydan R. Femoral lengthening and deformity correction using the Fitbone motorized lengthening nail. J Orthop Sci 2015; 20(1): 149-54. Landge V, Shabtai L, Gesheff M, Specht S C, Herzenberg J E. patient satisfaction after limb lengthening with internal and external devices. J Surg Orthop Adv 2015; 24(3): 174-9. Landis J R, Koch G G. The measurement of observer agreement for categorical data. Biometrics 1977; 33(1): 159-74.

126

Lascombes P, Popkov D, Huber H, Haumont T, Journeau P. Classification of complications after progressive long bone lengthening: proposal for a new classification. Orthop Traumatol Surg Res 2012; 98(6): 629-37. Laubscher M, Mitchell C, Timms A, Goodier D, Calder P. Outcomes following femoral lengthening: an initial comparison of the Precice intramedullary lengthening nail and the LRS external fixator monorail system. Bone Joint J 2016; 98-B(10): 1382-8. Lee D H, Kim S, Lee J W, Park H, Kim T Y, Kim H W. A comparison of the device-related complications of intramedullary lengthening nails using a new classification system. Biomed Res Int 2017; 2017. Lenze U, Hasler C C, Krieg A H. Ausgleich posttraumatischer Beinverkürzungen mit einem motorisierten intramedullären Nagel. Unfallchirurg 2011; 114(7): 604-10. Martin R C G, Brennan M F, Jaques D P. Quality of complication reporting in the surgical literature. Ann Surg 2002; 235(6): 803-13. Mazeau P, Assi C, Louahem D, L’Kaissi M, Delpont M, Cottalorda J. Complications of Albizzia femoral lengthening nail: an analysis of 36 cases. J Pediatr Orthop Part B 2012; 21(5): 394-9. Morrison T, Sontich J. Premature consolidation with resultant implant failure using PRECICE femoral nail lengthening. JBJS Case Connect 2016; 6(2): E2-E2. Murad M H, Sultan S, Haffar S, Bazerbachi F. Methodological quality and synthesis of case series and case reports. BMJ Evid Based Med 2018; 23(2): 60-3. Muratori F, Scoccianti G, Beltrami G. Is an intramedullary nail a valid treatment for limb-length discrepancy after bone tumor resection? Surg Technol Int 2018; 11(33): 281-8. Nasto L A, Coppa V, Riganti S, Ruzzini L, Manfrini M, Campanacci L, Palmacci O, Boero S. Clinical results and complication rates of lower limb lengthening in paediatric patients using the PRECICE 2 intramedullary magnetic nail: a multicentre study. J Pediatr Orthop B 2020; Epub ahead of print. Noonan K J, Leyes M, Forriol F, Cañadell J. Distraction osteogenesis of the lower extremity with use of monolateral external fixation: a study of two hundred and sixty-one femora and tibiae. J Bone Joint Surg Ser A 1998; 80(6): 793-806. Paley D. Current techniques of limb lengthening. J Pediatr Orthop 1988; 8(1): 73-92. Paley D. Problems, obstacles, and complications of limb lengthening by the Ilizarov technique. Clin Orthop Relat Res 1990; 250: 81-104. Paley D, Harris M, Debiparshad K, Prince D, Matthew H, Debiparshad K, Prince D. Limb lengthening by implantable limb lengthening devices. Tech Orthop 2014; 29(2): 72-85. Paley D, Debiparshad K, Balci H, Windisch W, Lichtblau C. Stature lengthening using the precice intramedullary lengthening nail. Tech Orthop 2015; 30(3): 167-82.

Acta Orthopaedica 2021; 92 (1): 119–126

Popkov A. Errors and complications of operative lengthening of the lower extremities in adults by the Ilizarov method. Vestn Khir Im I I Grek 1991; 146(1): 113-16. Rozbruch S R. Adult posttraumatic reconstruction using a magnetic internal lengthening nail. J Orthop Trauma 2017; 31(6): S14-19. Santaguida P, Keshavarz H, MacQueen G, Levine M, Beyene J, Raina P. Development of the McHarm: a tool evaluating validity of the collection and reporting of harms. Abstr 19th Cochrane Colloq Madrid, Spain. Chichester: Wiley; 2011. Poster. Schiedel F M, Vogt B, Tretow H L, Schuhknecht B, Gosheger G, Horter M J, Rödl R. How precise is the PRECICE compared to the ISKD in intramedullary limb lengthening? Reliability and safety in 26 procedures. Acta Orthop 2014; 85(3): 293-8. Shabtai L, Specht S C, Standard S C, Herzenberg J E. Internal lengthening device for congenital femoral deficiency and fibular hemimelia. Clin Orthop Relat Res 2014; 472(12): 3860-8. Simpson A H R W, Kenwright J. Fracture after distraction osteogenesis. J Bone Joint Surg Ser B 2000; 82(5): 659-65. Singh S, Lahiri A, Iqbal M. The results of limb lengthening by callus distraction using an extending intramedullary nail (Fitbone) in non-traumatic disorders. J Bone Joint Surg Br 2006; 88-B(7): 938-42. Slim K, Emile N, Forestier D, Kwiatkowski F, Panis Y, Chipponi J. Methodological index for non-randomized studies (minors): development and validation of a new instrument. ANZ J Surg 2003; 73(9): 712-6. Steiger C N, Lenze U, Krieg A H. A new technique for correction of leg length discrepancies in combination with complex axis deformities of the lower limb using a lengthening nail and a locking plate. J Child Orthop 2018; 12(5): 515-25. Stricker S J, Hunt T. Evaluation of leg length discrepancy in children. Int Pediatr 2004; 19(3): 134-46. Stroup D, Berlin J, Morton S, Olkin I, Williamson G, Rennie D, Moher D, Becker B, Sipe T, Thacker S. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283(15): 2008-12. Tiefenboeck T M, Zak L, Bukaty A, Wozasek G E. Pitfalls in automatic limb lengthening: first results with an intramedullary lengthening device. Orthop Traumatol Surg Res 2016; 102(7): 851-5. Tjernström B, Olerud S, Rehnberg L. Limb lengthening by callus distraction: complications in 53 cases operated 1980–1991. Acta Orthop 1994; 65(4): 447-55. Wiebking U, Liodakis E, Kenawey M, Krettek C. Limb lengthening using the PRECICETM nail system: complications and results. Arch Trauma Res 2016; 5(4): e36273. Wu W W, Kuhn K M. Mechanical failure of a femoral lengthening nail: a case report. Patient Saf Surg 2018; 12(7): eCollection 2018.

1/21

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

www.heraeus-medical.com

Vol. 92, No. 1, 2021 (pp. 1–12 )

Bone cement with gentamicin ttamicin i i and clindamycin

ACTA ORTHOPAEDICA

69 %

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

Volume 92, Number 1, February 202

TF-IORT210007.indd 1

23-02-2021 11:43:00