Revision rates after knee replacement. Cumulative results from worldwide clinical studies versus joint registers

Objective: To assess revision rates after knee arthroplasty by comparing the cumulative results from worldwide clinical studies and arthroplasty registers. We hypothesised that the revision rate of all clinical studies of a given implant and register data would not differ significantly. Methods: A systematic review of clinical studies in indexed peer-reviewed journals was performed followed by internal and external validation. Parameters for measurement of revision were applied (Revision for any reason, Revisions per 100 observed component years). Register data served as control group. Results: Thirty-six knee arthroplasty systems were identified to meet the inclusion criteria: 21 total knee arthroplasty (TKA) systems, 14 unicondylar knee arthroplasty (UKA) systems, one patello-femoral implant system. For 13 systems (36%), no published study was available that contained revision data. For 17 implants (47%), publications were available dealing with radiographic, surgical or technical details, but power was too weak to compare revision rates at a significant level. Six implant systems (17%) had a significant number of revisions published and were finally analysed. In general, developers report better results than independent users. Studies from developers represent an overproportional share of all observed component years. Register data report overall 10-year revision rates of TKA of 6.2% (range: 4.9-7.8%), rates for UKA are 16.5% (range: 9.7-19.6%). Conclusion: Revision rates of all clinical studies of a given implant do not differ significantly from register data. However, significant differences were found between the revision rates published by developers and register data. Therefore the different data need to be interpreted in the context of the source of the information. (C) 2012 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.