Active-control trials: how would a new agent compare with placebo? A method illustrated with clopidogrel, aspirin, and placebo

In an active-control trial with a new treatment and a comparator that has placebo-controlled trials, how might the effect of the new therapy versus placebo be estimated? For many diseases it is not ethically justified to use a placebo-control trial, yet in the United States regulatory efficacy is conceptually defined in comparison with placebo. By use of the logarithm of the odds ratio for the active-control trial and for the placebo-controlled trials of the active-control, we estimated the effect of the new agent versus placebo. This "putative placebo" estimate assumes that the odds ratio in the active-control versus placebo trials is the same as would have been obtained had a placebo arm been possible in the trial with the new agent. A formula is given for sample size calculation in such a setting. Clopidogrel, an adenosine diphosphate (ADP) receptor antagonist, and aspirin were compared in the Clopidogrel Versus Aspirin in Patients at Risk of Ischemic Events (CAPRIE) Study of patients with recent myocardial infarction, recent ischemic stroke, or symptomatic peripheral arterial disease. The 40 trials comparing aspirin with placebo are summarized by the Antiplatelet Trialists' Collaboration. For the outcome cluster of stroke, myocardial infarction, or vascular mortality, the estimated odds ratio for clopidogrel versus placebo is 0.70 (95% confidence interval 0.64-0.78, P <.000001). The relative risk reduction of approximately 30% represents a clinically meaningful benefit. For active-control trials use of prior trials of the active-control versus placebo may be usefully used to estimate the effect of the new therapy versus a putative placebo