|
Abstract:
|
In stroke clinical trials, the NIH stroke scale (NIHSS) is commonly used as a primary efficacy endpoint. However, some subjects die before the time of evaluation of the NIHSS. Furthermore, the Data Safety Monitoring Board monitors the study for safety, measured primarily by mortality. We propose a composite endpoint that ranks early deaths together with observed NIHSS outcomes by taking death as the worst outcome, and taking earlier deaths to be worse than later deaths. We evaluate the effects of the interim analyses for mortality on the power of the final endpoint through simulations patterned after an ongoing trial in acute ischemic stroke. We consider different designs for early stopping, corresponding to different hypotheses relating to mortality and NIHSS. We derive the analytic correlation among the interim analysis and final test statistics for designing trials of this type.
|
