Abstract:
|
In a confirmatory clinical trial, the familywise type-I error rate (FWER) needs to be controlled in a strong sense. When a study has multiple endpoints, a widely used multiplicity adjustment is a fixed sequence hierarchical testing. While the fixed sequence hierarchical testing controls FWER strongly for a fixed design, this may not be the case when the primary endpoint is analyzed by group sequential design (GSD) (Hung et al, 2007). Hung et al performed a simulation demonstrating the type I error inflation when only the primary endpoint follows GSD. In this presentation, we extended the simulation to a case where both primary and secondary endpoints are analyzed by GSD and also time to event endpoints. In order to control FWER in a strong sense, we propose to use the sequentially rejective graphical procedure in Maurer and Bretz (2013) which is based on the closure principle and provides a strong control of FWER for studies with both multiple endpoints and multiple analysis timings. Case studies following this approach will be presented.
|