Changing Frequency of Analyses with Multiple Endpoints in Group-Sequential Clinical Trials

• Activity Number: 318 - Adaptive (and Other) Clinical Trial Designs
• Type: Contributed
• Date/Time: Wednesday, August 11, 2021 : 3:30 PM to 5:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Koko Asakura* and Toshimitsu Hamasaki and Diane Uschner and Frank Bretz
• Companies: National Cerebral and Cardiovascular Center and George Washington University Biostatitics Center and George Washington University and Novartis AG
• Keywords: Interim analysis; Type I error probability; Conjunctive power

Abstract

In this presentation, we discuss methods for testing hypotheses associated with two endpoints (either multiple primary endpoints or one primary and one secondary endpoint) in group-sequential clinical trials comparing a test intervention to a control intervention. We consider the specific situation where the null hypothesis for one endpoint has been rejected at some interim and investigate the other as (1) co-primary or (2) ordered endpoints. Though maximum number of analyses is usually prefixed, a promising but nonsignificant interim result sometimes leads more frequent data review at the request of a data monitoring committee, which results in inflation of the Type I error probability. While changing the frequency of subsequent analyses based on the observed data is an attractive option, in practice it would be important to consider how the Type I error probability can be controlled, and then, whether the adaptation is efficient. We discuss the impact of these adaptations on power and sample size in relation to the Type I error probability, including methods for managing inflation of it. We illustrate the approaches using a real clinical trial.