Partition testing in confirmatory adaptive designs with structured objectives.

In this paper, we propose a partition testing for adaptive clinical trials with structured study objectives. The proposed approach is a combination of the graphical approach with partition testing. The proposed approach enables one to handle many types of structured objectives, to tailor a multiple test procedure to given structured objectives, and to draw sensible conclusions in adaptive clinical trials. In addition, the proposed approach reduces to the Bonferroni-based graphical approaches that can allow adaptations such as treatment selection and sample size reassessment during the course of the trial. Some practical aspects of the proposed approach are investigated via a simulation study.