Cumulative Meta-analyses and the Problem of Multiple Drug Effects

Abstract

To the Editor. —Antman et al 1 have observed that expert recommendations often lag behind pooled results of clinical trials. Using therapies for acute myocardial infarction as a model, they demonstrated that cumulative meta-analyses can allow more rapid clarification of the effectiveness of individual interventions. This innovative and promising method should be used to enhance our understanding of clinical trials, but it is not yet clear that these analyses should rapidly lead to standard textbook recommendations. For example, a textbook chapter on therapy of myocardial infarction must consider the combinations of drugs to be used, yet there is a paucity of data in this area. Meta-analyses demonstrating the benefit of six separate therapies in acute myocardial infarction do not address the effects of combinations of these same treatments. The fact that meta-analyses showed in 1976 that aspirin reduced mortality after myocardial infarction and in 1990 that warfarin did the same