124 – Special Topics in Clinical Trial Design and Analysis
Simulation-Based Evaluation of P-Value Quality in Phase 3 Clinical Trials
Jihao Zhou
Allergan Pharmaceuticals
Brandon Wales
University of California
Ray Zhu
Allergan Pharmaceuticals
A p-value is a most widely used measure of evidence against a null hypothesis in statistical testing of hypothesis. In Phase 3 clinical trials, a threshold of 0.05 alpha level is usually used to judge against a calculated p-value to conclude whether there is appropriate statistical evidence to support drug regulatory approval decision. The p-value, being derived from a statistical sample via test statistic, has inherent variability, which is generally ignored or not assessed and thus leads to a lack of understanding of its quality when evaluating study outcome from phase 3 clinical trials. In this paper, we use simulation-based approach to assess the p-value variability from published Phase 3 clinical trials.