Abstract
Ethical justification for starting a clinical trial requires at the outset an accurate statement of "no difference" regarding the two or more agents to be compared. This may be expressed as "theoretical equipoise" (no data to support the superiority of one of the agents) or, preferably, as "clinical equipoise" (there are insufficient data to resolve controversy among experts as to which is superior). This presents a problem when a placebo control is proposed, particularly when the outcome measure entails irreversible loss of function; preliminary data often suggest the superiority of the "active agent." Informed consent should ordinarily include the fact that treatment assignments will be accomplished by a process of randomization and, in the case of double-blind designs, that neither the subject nor the investigator will know the subject's treatment assignment until the end of the trial. In general, clinical trials should be monitored by data and safety monitoring boards that have access to unblinded data; they should be guided by stopping rules that should be agreed upon by all concerned before the trial is begun. Women and minorities must be included unless there is strong justification for their exclusion. Care must be taken to balance the competing objectives of validity (the results will be correct), generalizability (the results will be broadly applicable), and efficiency (the study will not be unduly expensive).
