The randomized clinical trial presents us with a striking paradox. While arguably the single most important methodological advance in medical research, it remains poorly understood outside the circle of experts skilled in its practice. Even within that circle, difficult issues remain. Moreover, the most important issues encountered in clinical trials are virtually invisible to the volunteers who participate in them--and certainly these people should have a much clearer understanding of clinical trial research and exactly what is at stake.

By some estimates, more than 300,000 randomized clinical trials have been conducted since October 1948, when the British Medical Journal published the results of what is widely believed to be the first modern randomized trial, "Streptomycin Treatment of Pulmonary Tuberculosis." Clinical trials have wrought innumerable changes in health care; in breast cancer treatment alone, randomized trials have resolved controversies in the treatment of operable breast cancer, showing that for many tumors, surgical removal alone is as effective as complete removal of the affected breast. Recent trials in HIV-AIDS demonstrate that new antiretroviral therapies suppress viral burden and substantially delay the onset of AIDS. Indeed, the use of clinical trials is so abundant in medicine today that it is difficult to imagine how most chronic diseases would be treated without them. And it is rare that an issue of a major medical journal, like the New England Journal of Medicine or the Journal of the American Medical Association, does not contain a report of at least one randomized trial. Policymakers and the public almost take for granted that clinical trials are an essential part of the FDA’s approval process for new drugs.

Yet the clinical trial still presents difficult, almost insolvable issues that arise due to the difference between them and any other methodology used to study human populations. Not only is it a controlled experiment using human subjects, but a randomized clinical trial asks of the human subject something that would be unthinkable outside of the context of scientific study: to receive a treatment chosen neither by the subject, nor the caregiver, but instead by random assignment, the modern-day equivalent of the proverbial coin toss. As a result, the randomized clinical trial must stand up to special scrutiny.

Every scientist designing a randomized trial faces one fundamental question at the onset: is it ethical to enroll subjects in this trial? Ethicists argue that for the answer to be "yes" there must be a state of equipoise, or balance between the two interventions being compared in the trial. Either individual clinicians must be unsure which intervention is better, or the evidence in favor of one must be balanced by the evidence in favor of the other. Equipoise is an elusive concept, yet it is only the first in a chain of difficult issues. By relinquishing the right to choose a therapy and by allowing observations of their outcome to be used in scientific literature, people enrolled in clinical trials enter into an implicit contract with the investigators. For their part, the investigators ensure to the best of their ability that the trial will be free of bias in its evaluation, and that differences in outcome will be measured precisely enough to answer the question posed by the trial. Trial designers must also decide on the best way to monitor a trial so that if it becomes clear early in the trial what the conclusion will be, the enrollment of patients will be stopped. These questions all involve a delicate interplay between the ethical context of a trial and the statistical or mathematical basis of its experimental design. The computations done to determine the enrollment needed on a trial allow a quantitative "answer" to some of the ethical decisions that investigators face, such as, how many patients should be enrolled? However, emphasis on the quantitative aspect sometimes makes it difficult for nonstatisticians--clinical investigators and especially trial participants--to understand why and how a trial is being done. That difficulty inhibits the public debate that should surround more trials.

These debates will become more informed as the general understanding of the quantitative basis for public health studies improves. Raising the level of "numeracy" among both public health professionals and the general public may be the most important role of statisticians and epidemiologists in schools of public health.