Dealing with competing risks: testing covariates and calculating sample size

Authors

  • Melania Pintilie

    Corresponding author
    1. Ontario Cancer Institute, Princess Margaret Hospital, Toronto, Canada
    • Biostatistics Department, Ontario Cancer Institute, Princess Margaret Hospital, 610 University Avenue, Toronto, ON, M5G 2M9, Canada
    Search for more papers by this author

Abstract

It is universally agreed that Kaplan–Meier estimates overestimate the probability of the event of interest in the presence of competing risks. Kalbfleisch and Prentice recommend using the cumulative incidence as an estimate of the probability of an event of interest. However, there is no consensus on how to test the effect of a covariate in the presence of competing risks. Using simulations, this paper illustrates that the Cox proportional hazards model gives valid results when employed in testing the effect of a covariate on the hazard rate and when estimating the hazard ratio. A method to calculate the sample size for testing the effect of a covariate on outcome in the presence of competing risks is also provided. Copyright © 2002 John Wiley & Sons, Ltd.

Ancillary