Background: A patient who dies during the followup period of a study about total hip arthroplasty (THA) cannot subsequently undergo a revision. The presence of competing events (such as deaths, in a study on implant durability) violates an assumption of the commonly used Kaplan-Meier (KM) survivorship approach. In that setting, KM-based estimates of revision frequencies will be high relative to alternative approaches that account for competing events such as cumulative incidence methods. However, the degree to which this difference is clinically relevant, and the degree to which it affects different ages of patient cohorts, has been poorly characterized in orthopaedic clinical research. Questions/purposes: The purpose of this study was to compare KM with cumulative incidence survivorship estimators to evaluate the degree to which the competing event of death influences the reporting of implant survivorship at long-term followup after THA in patients both younger than and older than 50 years of age. Methods: We retrospectively reviewed 758 cemented THAs from a prospectively maintained single-surgeon registry, who were followed for a minimum of 20 years or until death. Revision rates were compared between those younger than or older than age 50 years using both KM and cumulative incidence methods. Patient survivorship was calculated using KM methods. A total of 21% (23 of 109) of the cohort who were younger than 50 years at the time of THA died during the 20-year followup period compared with 72% (467 of 649) who were older than 50 years at the time of surgery (p < 0.001). Results: In the cumulative incidence analysis, 19% of the younger than age 50 years cohort underwent a revision for aseptic causes within 20 years as compared with 5% in the older than age 50 years cohort (p < 0.001). The KM method overestimated the risk of revision (23% versus 8.3%, p < 0.001), which represents a 21% and 66% relative increase for the younger than/older than age 50 years groups, respectively. Conclusions: The KM method overestimated the risk of revision compared with the cumulative incidence method, and the difference was particularly notable in the elderly cohort. Future long-term followup studies on elderly cohorts should report results using survivorship curves that take into account the competing risk of patient death. We observed a high attrition rate as a result of patient deaths, and this emphasizes a need for future studies to enroll younger patients to ensure adequate study numbers at final followup. Level of Evidence: Level III, therapeutic study.
Bibliographical noteFunding Information:
One of the authors (JJC) is a consultant for and received royalties from DePuy (Warsaw, Inc, USA; for intellectual property transfer for hip and knee implant designs) and from Lippincott Williams & Wilkins (Baltimore, MD, USA; for books edited). This project was supported by the Bierbaum resident research fund.
© 2015, The Association of Bone and Joint Surgeons®.