|
Abstract:
|
Group sequential designs (GSD) are a well-established approach to maintain strong control of type 1 error while enabling a trial to stop early for efficacy and/or futility based on interim decision rules. Developing the decision rule framework becomes increasingly difficult as the complexity of the trial’s features and number of key scientific objectives increases (i.e. as the number of treatment arms, interim analyses, subpopulations, and endpoints increase). In this talk, we provide two case examples of proposed pediatric group sequential design trials that extend the work of Urach and Posch (2016) on multi-arm GSDs. One trial accounts for the testing of an overall and subpopulation, and the other trial considers multiple arms and interim analyses. Efficient decision rules and alpha-exhaustive closed testing procedures that utilize the known correlation structure among test statistics are proposed, and therefore increase the power of the design. These trials enable early stopping while maximizing the power of both the elementary and global tests, and thus, increase the likelihood of answering key objectives of the trial while enrolling only the necessary amount of participants.
|
