Randomized clinical trials in oncology. Principles and obstacles.

Cancer clinical trials have two broad objectives: To identify promising treatments and to determine whether such treatments really do reduce morbidity and mortality. These two objectives generally require different types of clinical trials. The guiding principle of clinical trials is to ask an important question and get a reliable answer. Asking an important question generally means asking a question that has the potential for influencing medical practice, using a control treatment that is widely accepted, using an experimental treatment that is widely applicable, using an endpoint that is a direct measure of patient benefit, and studying a group of patients who are broadly representative. Getting a reliable answer also has many components, including randomized treatment assignment, adequate sample size, intent-to-treat analysis, and testing of prestated hypotheses on mature data. These components are discussed. Several recent developments in clinical trials methodology: interim monitoring boundaries, large simple trials, and metaanalysis, are highlighted here. Major obstacles to conducting randomized clinical trials are also described.