Exploring changes in treatment effects across design stages in adaptive trials

The recently published committee for Medicinal Products for Human Use reflection paper on flexible designs highlights a controversial issue regarding the interpretation of adaptive trials. The guideline suggests that a test for heterogeneity should he preplanned and if treatment effect estimates differ significantly between design stages then data collected be ore and after the interim analysis might not be combined in a formal analysis. In this paper we investigate error rates for such a procedure ill the presence of calendar-time effects. Furthermore, we present all alternative testing strategy based oil change point methods. In a simulation study we demonstrate that our procedure performs well in comparison to that suggested by the guideline. Copyright (C) 2008 John Wiley & Sons, Ltd.