Abstract:
|
Subgroup analysis is frequently used to account for the treatment effect heterogeneity in clinical trials. When a promising subgroup is selected from existing trial data, we must address the question of how good the selected subgroup really is. The usual statistical inference applied to the selected subgroup, assuming that the subgroup is chosen independent of the data, may lead to an overly optimistic evaluation of the selected subgroup. In this talk, we address the issue of selection bias and develop a de-biasing bootstrap inference procedure for the selected subgroup effect. The proposed inference procedure is model-free, easy to compute, and asymptotically sharp. We demonstrate the merit of our proposed method by re-analyzing the MONET1 trial and show that the risk of data snooping, an increasingly more common problem in the big data era, should be addressed appropriately in any management or regulatory decision. This talk is based on a paper awarded the Biopharmaceutical Section Student Paper Award of American Statistical Association.
|