|
Abstract:
|
There is very limited research in adaptive design for trials with recurrent events. Such endpoints usually take a long time to collect. It is often infeasible to select doses based on recurrent events data in phase 2. Thus, it is critical to perform dose selection using a surrogate endpoint, which takes shorter time to mature and is better powered to detect the difference among the studied treatments or doses, in the phase 2 part of a phase 2/3 adaptive design trial. In this talk, we considered using continuous endpoints as surrogate endpoint for dose selection. We tackled the challenging design issues including: (1) how to perform the final analysis to ensure type I error control, (2) how the correlation between the surrogate endpoints and the recurrent event outcome impacts the design performance, (3) whether subjects can switch to the selected treatment after interim analysis for ethical and enrolment considerations, (4) how to perform sample size re-estimation after dose selection to ensure the study is well powered. We compare adaptive designs with Bonferroni adjustments and fixed sample designs without interim analysis using simulations. The optimal design is recommended.
|
