Comparison of medical and surgical management of coronary artery disease: methodologic issues.

Coronary artery bypass surgery has been recommended for patients with a wide variety of signs, symptoms, and medical histories. Investigators have used historically controlled studies, registries and randomized clinical trials to compare the efficacy of medical and surgical management of coronary artery disease in well-defined patient groups. Historically controlled studies offer a weak basis for inference because of rapid changes in medical practice and in the prognosis for patients with coronary artery disease. Randomized trials are the ideal method for comparing therapies, but only a few trials, in highly selected patients, have been completed. Thus, registry studies provide an important part of the currently available evidence. In this paper, the strengths of randomized and registry studies are compared and the need for "meta-analyses" that integrate the evidence from both types of study is discussed. The institutional variation in surgical mortality and its implication for treatment strategy are also discussed.