Collective Evidence in Drug Evaluation

Abstract

Multiple doses, endpoints, and tests are used in several clinical studies to establish drug efficacy. Statistical evaluation relies heavily on multiplicity adjustments within one study to control the type I error rate. The use of multiplicity adjustment procedures (MAPs) sometimes leads to conclusions that may not seem logical. As drug efficacy evaluation involves aspects such as assessing efficacy, selecting optimal doses, and labeling claims, incorporating all the aspects under the umbrella of controlling type I error may not be an optimum strategy. Alternatively, a practical approach that uses collective evidence is proposed to evaluate multiple studies, doses, endpoints, and tests. Instead of controlling the type I error, specific types of errors are controlled, such as the error of wrongly approving an ineffective drug and the error of labeling false information. With the collective evidence approach, the need of MAPs in individual studies is debated when multiple studies are available.