Sample Size Considerations When Using Two Time-to-Event Outcomes for Comparing Two Interventions in Clinical Trials

• Activity Number: 244
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Toshimitsu Hamasaki*+ and Tomoyuki Sugimoto and Takashi Sozu and Scott Evans
• Companies: Osaka University Graduate School of Medicine and Hirosaki University Graduate School of Science and Technology and Kyoto University School of Public Health and Haravard University School of Public Health
• Keywords: Bivariate event times ; Conjunctive power ; Depending censoring ; Disjunctive power ; Multiplcity ; Semicompeting risks

Abstract

The use of two co-primary time-to-event endpoints has become common in clinical trials evaluating interventions in many disease areas such as infectious disease, oncology, or cardiovascular disease. We previously developed methods for sizing clinical trials with two co-primary time-to-event outcomes, where both events are non-fatal and each event-time is not censored by other event. In this presentation, we describe novel methods that extend these results to accommodate two additional situations, i.e., (i) when one event is fatal, and (ii) when both are fatal. For (i), the non-fatal event time may be censored by the fatal event, but for (ii), each event-time may be censored by other event. We outline the calculation of the variance-covariance matrix of the test statistics for two outcomes and describe a sample size formula under a correlation structure of several bivariate exponential distributions. We evaluate the performance of the methods and investigate the behavior of the required sample sizes via simulation.