All Times EDT
Keywords: RAR, confirmatory study
In efforts to advance the drug development program with efficiency, speed and precision, adaptive designs such as sample size re-estimation, seamless phase II/III design, and response adaptive randomization have been gaining popularity in recent years. Specifically, designs with response adaptive randomization may provide statistical, ethical, and pragmatic advantages in which subjects are more likely to be randomized to the more promising treatment groups. However, these designs are in general more commonly endorsed and utilized in phase II studies for exploratory purposes. Confirmatory studies with such adaptive features are subject to careful evaluations by both the sponsor and the regulatory agency. In this presentation, we propose an innovative response adaptive randomization scheme with plausible characteristics to serve for confirmatory studies: (i) the family-wised error rate (FWER) can be controlled analytically at level a in a strong sense; (ii) the practical adaptation process improves the traceability from study conduct and operational perspective. Using simulations, we also show the improved power and efficiency from the proposed design, to detect the treatment effect of selected treatment groups.