A Comparison of Multiple Testing Procedures That Compare Two Interventions with Respect to Two Time-to-Event Outcomes in Clinical Trials

• Activity Number: 563
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yuki Ando*+ and Toshimitsu Hamasaki and Scott R. Evans and Koko Asakura and Yuko Ohno
• Companies: Pharmaceuticals & Medical Devices Agency and Osaka University Graduate School of Medicine and Harvard School of Public Health and Osaka University and Osaka University Graduate School of Medicine
• Keywords: Multiplicity ; Multiple primary endpoints ; Power ; Type I error rate ; Semi-competing risk ; Time-to-event outcome

Abstract

The use of two primary time-to-event endpoints has become common in clinical trials evaluating interventions in e.g., infectious disease, oncology, and cardiovascular disease. If a trial is designed to detect an effect on at least one endpoint, then control of the Type I error rate must be considered. Many methods to control the Type I error rate with/without a prespecified ordering of endpoints exist including the Bonferroni and related procedures, the adaptive alpha allocation approach, and the fallback-type procedure. We evaluate the performance of these procedures when comparing two interventions with respect to two time-to-event outcomes in clinical trials, with regard to power and the probability of rejecting at least one or both hypotheses. We consider three censoring paradigms for these endpoints: (i) when both are non-terminal, (ii) when one event is terminal, and (iii) when both are terminal. The results of this work will help practitioners select the most optimal method in various situations.