Background

In the presence of a competing risk of death, the Kaplan-Meier method is known to overestimate the probability of implant failure. To our knowledge, the magnitude of the competing risk of death in survivorship analyses in total knee arthroplasty has not been studied. The purpose of this study was to determine the extent of the competing risk of death in different age groups during a long-term follow-up of patients who had undergone primary total knee arthroplasty.

Methods

The study population comprised 22,864 primary total knee arthroplasties at a large medical center in the United States. We compared the overall revision outcomes estimated using the Kaplan-Meier method and the cumulative incidence function over a mean follow-up of 7.8 years (range, 0.1 to 26.3 years).

Results

The risk of death exceeded the risk of revision by a factor of 4 at twelve years following the surgical procedure. The Kaplan-Meier method overestimated the risk of revision by 3% at five years, 14% at ten years, 32% at fifteen years, and 57% at twenty years. At ten years after the surgical procedure, the risk of death exceeded the risk of revision by a factor of 15 in the all-polyethylene group and by a factor of 3.5 in the metal modular group. The Kaplan-Meier method overestimated the ten-year risk of revision by 18% in the all-polyethylene group and by 11% in the metal modular group. After accounting for the higher competing risk of death in the all-polyethylene group, the cumulative incidence of revision in the metal modular group was about 2.5 to three times higher than that in the all-polyethylene group at ten years.

Conclusions

Although the Kaplan-Meier and cumulative incidence methods yielded different implant survival estimates beyond ten years (Kaplan-Meier overestimates implant failure), the size of the bias was small at earlier time points. The extent of overestimation depends on the duration of follow-up and the magnitude of the competing risk of death.