An Adaptive Group-Sequential Test Without Predetermination of Total and Group Sample Sizes

• Activity Number: 605
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Kosuke Kashiwabara*+ and Yutaka Matsuyama
• Companies: University of Tokyo and University of Tokyo
• Keywords: adaptive design ; clinical trials ; interim analysis ; rare disease ; sample size re-estimation ; sequential probability ratio test

Abstract

In designing confirmatory clinical trials for rare diseases, it is not uncommon that the sample size has to be determined with considerable uncertainty due to the lack of enough historical evidences for it. Accurately predicting the accrual rate of the patients would also be difficult. In such cases, adaptive designs allowing the unblinded sample size re-estimation are particularly attractive. However, if the sample size is largely increased or decreased from the initial one, efficiency of the test would considerably decrease. We propose a novel adaptive group sequential test with exact type I error rate base on the sequential probability ratio test. The total sample size and even the number of interim analyses and the sample sizes between stages can be freely determined based on the unblinded interim results with negligible loss of efficiency irrespective of the total sample size finally chosen. The critical value of the final test is determined by Müller-Schäfer principle (Müller HH, Schäfer H (2001), Biometrics 57, 886-891). We evaluate the efficiency of our proposed test through simulations in terms of the conditional and unconditional powers under several scenarios.