Composite endpoints: Sometimes more than a solely economic consideration

Endpoints are response variables, or outcomes, that are measured during the course of a clinical trial. I consider endpoints that are either events (e.g., death) or the time to an occurrence of an event (e.g., time to disease progression). A composite endpoint (CEP) is an endpoint that consists of a number of component endpoints, and is considered to have occurred as soon as any one of its components occurs. For example if CEP = death + disease progression, the CEP is said to have occurred as soon as either the disease progresses or the patient dies. It is seen that one of the results of using a CEP is to increase the event rate; and this in turn can reduce the sample size or the time required to observe a specified number of events, thereby resulting in a speedier, less costly clinical trial. Many believe that the only reason CEPs are ever employed is to this end, viz., saving money. I argue that there may be other circumstances that suggest the use of CEPs – that the choice of the primary response variable should be driven by the question the trial is being designed to answer.