Keywords: Un-blinded interim analysis, type I error rate
Clinical trials that are intended to provide primary evidence for product approval are usually designed based on smaller and limited information from earlier studies. There exist uncertainties of the planned sample sizes. Consequently, adaptive design allowing using interim results of the on-going studies to re-size the trials to ensure an adequate power to detect a treatment difference at study completion is attractive. In fact, sample size re-estimation (SSR) is a design adaptation that has been proposed and occurred most frequently in regulatory submissions just after group sequential design. There have been numerous approaches for sample size re-estimation proposed in literatures and have been applied to clinical trials. For SSR based on unblinded interim treatment effects, the major issues are the potential type I error inflation and/or operational biases. Therefore, clear analytical and statistical justifications to demonstrate a control of type I error rate and adequate strategies /plans to mitigate operational bias are expected to be laid out in regulatory applications. In this talk, I will present regulatory experiences in sample size re-estimation approaches followed by recommendations.