Value-Added Decisions in Oncology

Abstract

Registration of new anticancer drugs is decided too often not by their clinical value but by a p value. Approval is granted if the difference in an acceptable time-to-event outcome measure differs between experimental and control arms of a randomized controlled trial, such that the null hypothesis can be rejected based on a statistical test that meets the arbitrary criterion of p < .05. However, as stated by the American Statistical Association, a p value does not measure the size of an effect or the importance of a result; it does not provide a good measure of evidence related to a hypothesis, and policy decisions should not be made on the basis of whether a p value passes a specific threshold. Unfortunately, this statement is ignored by most journals, which emphasize p values in reporting results of clinical trials, and by regulatory agencies, such as the U.S. Food and Drug Administration and the European Medicines Agency; a significant p value is often a necessary and sufficient criterion for granting marketing approval. As a result, pharmaceutical companies often design large trials to increase the probability that a small difference in the primary outcome measure will be "significant." Moreover, the market price set for such drugs bears no relationship to the level of their benefit; drugs with small effects on outcome are sold at roughly the same price as "good drugs" that convey substantial benefit.