Street: 5600 Fishers Ln
Authors: Gang Chen, Yong-Cheng Wang, and George Chi
Title: Multiplicity Issues in Sample Size Re-estimation for Active Control Non-inferiority Trials
Abstract: The sample size required for testing non-inferiority hypotheses depends on an appropriate assessment of the control effect and the assumption of the new treatment effect relative to the control. Test of the null hypothesis with a fraction retention of the control effect is conditionally equivalent to an asymmetric two confidence interval testing procedure, where the control effect is estimated by the lower limit of a 100(1-?n?×)%, 0 < ×n¬n¡znconfidence interval . The value of this ×nis not known ahead of time for most clinical benefit endpoints and the sample size of the study cannot be reasonably estimated at the study design stage. In this presentation, a randomized 2-stage design is proposed for active control non-inferiority studies. The ×ncan be evaluated at the end of first stage of the study so that the sample size can be re-estimated. A simulation has been done to exam the potential for the type 1 error inflation due to the interim sample! size adjustment and an analysis method for controlling the type I error inflation has been proposed. An example based on an approved NDA data is given.
References: . Rothmann, M, Li, N, Chen, G, Chi, G, and Tsou, HH. Non-inferiority methods for mortality trials. ASA Proceedings of the Biopharmaceutical Section 2001.