Abstract:
|
In contemporary clinical trials, there could be a need to examine multiple endpoints, such that success on more than one (k=1, 2, …, n) endpoint is crucial for claiming drug efficacy. In this study, we proposed several methods based on the superchain procedure (Kordzakhia and Dmitrienko, 2013), and compared them with several related-existing approaches for assessing efficacy in multiple families. We use the shortcut Holm procedure to assess the case where no correlation is assumed for multiple endpoints and extend the study to multiple families of hypotheses, either exchangeable or hierarchical cases, by modifying the superchain procedure based on k-truncation. Simulation studies are conducted under different scenarios to evaluate the operating characteristics of our proposed procedures. The essence of the proof for our proposed procedures will be highlighted in the presentation.
|