The traditional paradigm of determining efficacy in Phase 2 and 3 trials is very limited: there is a single primary endpoint, a single population, and a single primary method of analysis. The interpretation of results is highly inadequate when any of these "singles" is deficient. This is a serious problem given the complexity and soaring cost of trials.
We introduce a more reliable way to frame questions and test hypotheses so that the results are more robust. Specifically, we propose pre-specifying multiple test statistics and relying on the minimum of p-values for testing the null. The critical value for hypothesis testing comes from permutation distributions. The null hypothesis is rejected when the smallest p-value is less than the critical value. We show that the inference with the minimum p-value is more powerful than the p-value derived from a single test.
This approach has applications in a variety of settings, including subgroups identification, biomarker analyses, and dose ranging trials. Examples are given for each application. This presentation will show that the proposed method is a better way to plan Phase 2 and 3 trials than the current, conventional approach.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.