The alpha spending function approach to interim data analyses.

Abstract

Over the past three decades, clinical trials have become one of the major standards for evaluating new therapies and interventions in medicine [1–3]. Numerous clinical trials have been conducted during this period across a wide variety of diseases, evaluating drugs, procedures, devices, and biologic materials. The fundamentals of the design, conduct, and analyses of clinical trials have been developed and refined during this period as well. One such fundamental is that clinical data should be carefully monitored during the course of the trial so that unexpected or unacceptable toxicity can be detected as soon as possible in order to minimize patient exposure; in addition, trials should not be continued longer than necessary to prove the benefits of the therapy or intervention under study, or to understand the trade-offs between the benefits and risks of the therapy. In order to accomplish this goal, the National Institutes of Health sponsored a committee in the 1960s to develop guidelines for the conduct of clinical trials. The chair of this committee was Dr. Bernard Greenberg from the University of North Carolina, and the report, which was issued in 1967, has become known as the Greenberg Report [4], although it was only recently published in the literature. This report endorses the concept of interim review of data by an independent Data and Safety Monitoring Board (DSMB), a committee that has no conflict of interest for the study. This typically means that committee members should not be investigators entering patients into the trial. The Coronary Drug Project (CDP) [5] was one of the first trials to implement the Greenberg model.