Ten-year results of the randomized control trials of coronary artery bypass graft surgery: tabular data compiled by the collaborative effort of the original trial investigators. Part 2 of 2.

This metaanalysis has some important general implications for conducting studies to evaluate other treatments as well as for patient management. To avoid missing clinically important differences (eg, 15-20% differences in mortality at 5 years) between the methods of revascularization, study populations should be several times larger than those in our metaanalysis, should include a high proportion of the types of patients for whom surgery is known to be superior to medical therapy, and should be complemented by a systematic overview. We recommend that researchers running large randomized trials consider prospective collaboration with researchers from other trials with a common protocol.