Surgeon, Implant, and Patient Variables May Explain Variability in Early Revision Rates Reported for Unicompartmental Arthroplasty

Background: The reported revision rates for cemented unicompartmental knee arthroplasties have varied widely. The effect of implant selection, patient characteristics, and surgeon characteristics on revision risk has not been evaluated fully. The purpose of the present study was to determine the impact of these variables on the risk for aseptic revision arthroplasty. Methods: We identified all cemented primary unicompartmental knee arthroplasties performed in our health-care system from 2002 to 2009 (median follow-up time = 2.6 years) to assess the risk for aseptic revision. A multivariate marginal Cox proportional-hazards model with robust standard errors (to adjust for the nesting of surgical cases within surgeons) was used to calculate the differential risk for revision of implants after adjusting for surgeon and hospital volume of unicompartnnental knee arthroplasties performed; surgeon experience with unicompartmental knee arthroplasties at the time of surgery; surgeon fellowship training; and patient age, sex, weight, body mass index, and American Society of Anesthesiologists (ASA) score. Results: A total of 1746 unicompartmental knee arthroplasties were identified. The overall revision rate during the study period was 4.98% (95% confidence interval [CI], 4.0% to 6.1%). In a multivariate Cox model, the hazard ratio (HR) for aseptic revision relative to a modern, fixed, metal-backed tibial bearing was significantly higher for an all-polyethylene tibial tray (HR = 3.85, 95% CI = 1.54 to 9.63, p = 0.004) but not significantly higher for a mobile-bearing implant (HR = 2.42, 95% CI = 0.55 to 10.65, p = 0.242) or an older-design, fixed, metal-backed bearing (HR = 1.89, 95% CI = 0.67 to 5.33, p = 0.23). Younger age was associated with increased risk (age less than fifty-five years compared with more than sixty-five years: HR = 4.83, 95% CI = 2.60 to 8.96, p < 0.001), and a higher ASA score compared with >= 3 points: HR = 0.54, 95% CI = 0.32 to 0.93, p = 0.027) and a greater mean yearly surgeon volume of unicompartmental knee arthroplasties (twelve or fewer compared with more than twelve: HR = 2.18, 95% CI = 1.28 to 3.74, p = 0.004) were associated with reduced risk. Conclusions: Implant selection can have a considerable effect on the risk for aseptic revision following a cemented unicompartmental knee arthroplasty, as can patient and surgeon factors. Therefore, the variation among risk estimates reported in the literature for unicompartmental knee arthroplasty revision may be explained by differences in patient characteristics and implant selection as well as the surgeons' yearly volume of unicompartmental knee arthroplasties.