Independent predictors of revision following metal-on-metal hip resurfacing: a retrospective cohort study using...

Modern metal-on-metal hip resurfacing has been widely performed in the United Kingdom for over a decade. However, the literature reports conflicting views of the benefits: excellent medium- to long-term results with some brands in specific subgroups, but high failure rates and local soft-tissue reactions in others. The National Joint Registry for England and Wales (NJR) has collected data on all hip resurfacings performed since 2003. This retrospective cohort study recorded survival time to revision from a resurfacing procedure, exploring risk factors independently associated with failure. All patients with a primary diagnosis of osteoarthritis who underwent resurfacing between 2003 and 2010 were included in the analyses. Cox's proportional hazard models were used to analyse the extent to which the risk of revision was related to patient, surgeon and implant covariates. A total of 27 971 hip resurfacings were performed during the study period, of which 1003 (3.59%) underwent revision surgery. In the final adjusted model, we found that women were at greater risk of revision than men (hazard ratio (HR) = 1.30, p = 0.007), but the risk of revision was independent of age. Of the implant-specific predictors, five brands had a significantly greater risk of revision than the Birmingham Hip Resurfacing (BHR) (ASR: HR = 2.82, p < 0.001, Conserve: HR = 2.03, p < 0.001, Cormet: HR = 1.43, p = 0.001, Durom: HR = 1.67, p < 0.001, Recap: HR = 1.58, p = 0.007). Smaller femoral head components were also significantly more likely to require revision (≤ 44 mm: HR = 2.14, p < 0.001, 45 to 47 mm: HR = 1.48, p = 0.001) than medium or large heads, as were operations performed by low-volume surgeons (HR = 1.36, p < 0.001). Once these influences had been removed, in 4873 male patients < 60 years old undergoing resurfacing with a BHR, the five-year estimated risk of revision was 1.59%. In summary, after adjustment for a range of covariates we found that there were significant differences in the rate of failure between brands and component sizes. Younger male patients had good five-year implant survival when the BHR was used.