Abstract: The problem of testing for differences between two treatments with respect to multiple competing risk time-to-event endpoints is considered. Using a nonparametric approach, a multivariate two-sample statistic is proposed for simultaneous testing for differences in cumulative incidence of the multiple competing survival endpoints.
Establishment of the statistical properties of the test requires consideration of covariation processes among martingales associated with all cause-specific counting processes. The procedure is illustrated by a simulation study and by an application to data from a recent large randomized cancer prevention clinical trial.
Key words and phrases: Cause-specific cumulative incidence, cause-specific failure probability, cause-specific hazard, clinical trial, counting process, covariation process, martingale, multiple outcomes, nonparametric estimates, simultaneous inference, survival data.