Multistage Adaptive Test Procedures Based on Fisher's Product Criterion

Tables for multistage test procedures that are based on a repeated significance test version of Fisher's combination test are presented. Bauer and Köhne (1994) and Bauer and Röhmel (1995) stressed the adaptive nature of the proposed procedure. The formers incorporated the stopping of the trial with the acceptance of H0 and presented tables for the corresponding two-stage and three-stage test procedures. These methods are generalized for the application in clinical trials with more than two interim analyses. Different designing strategies are considered where stopping in favor of H0 is possible. The proposed test procedures are compared with the classical non-adaptive group sequential test designs.