Recycling of Significance Levels in Testing Multiple Hypotheses of Confirmatory Clinical Trials

Abstract

Confirmatory clinical trials for the demonstration of the effects of new treatments generally classify their hypotheses into the primary and secondary types and sometimes into other lower types. The primary hypotheses enjoy a special status; if the trial wins for one or more primary hypotheses , then one can characterize clinically relevant benefits of the study treatment. This framework of classifying hypotheses based on their importance into primary and secondary types allows using statistical test methods that maximize the power for the test of primary hypotheses . These methods also recycle the significance level of a successfully rejected hypothesis to other hypotheses (e.g., from a rejected primary or secondary hypothesis to other primary and secondary hypotheses ). Often when there is a set of primary or secondary hypotheses , then hypotheses within each set or family can be assigned with different weights by importance or power considerations. This structuring also allows using methods that allow recycling of the significance level of a successfully rejected hypothesis within a set or family to other hypotheses in the same set or family or to a hypothesis in the next family. A number of novel statistical test methods have been introduced over the last two decades that are based on such approaches for testing multiple hypotheses in clinical trials. The purpose of this chapter is to review some of these methods.