SEQUENTIAL MONITORING OF CLINICAL-TRIALS WITH MULTIPLE SURVIVAL END-POINTS

Many clinical trials follow patients for several different types of su rvival endpoints, such as mortality, disease progression, and time unt il dose-limiting toxicity. Conduct of such trials often requires that the accumulating data be reviewed periodically to protect the safety o f participating patients and possibly identify early treatment differe nces. This paper proposes a group sequential method for assessing mult iple survival endpoints using repeated confidence intervals. Counting processes for each survival endpoint are used to estimate both the cor relation between outcomes and between times of interim analysis. The m ethods are illustrated using a clinical trial comparing two treatments for PCP prevention in AIDS patients. The operating characteristics of three strategies for constructing confidence intervals are assessed a nd compared in a simulation study.