Competing Risks Methods in the Analysis of Clinical Trials for Early-stage Cancer

• Activity Number: 283
• Type: Invited
• Date/Time: Wednesday, August 11, 2004 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): James J. Dignam*+
• Companies: University of Chicago
• Keywords: Gray's test ; cause-specific hazard ; cumulative incidence function ; subdistribution hazard ; logrank test

Abstract

In cancer clinical trials, competing risks are frequently encountered. For example, women treated surgically for breast cancer may experience any of the following as a first failure event: recurrence near the removed tumor, recurrence at other anatomic sites, occurrence of a tumor in the opposite breast, occurrence of other cancer types, or death from noncancer causes prior to these events. Cause-specific hazards and cumulative incidence functions are typically used to summarize results by event type. Tests for event-specific differences by treatment group may be based on comparison of cause-specific hazards or quantities associated with the cumulative incidence function, such as the subdistribution hazard or the distance between cumulative incidence function curves. Results of these tests may differ depending on the relative magnitude of different event hazards, and it may be unclear as to which test to apply. Depending on the questions of principal interest, tests on either or both metrics may be appropriate. We present practical examples from cancer clinical trials and simulation study results to illustrate data summary and inferential tools for competing risk observations.