We read with interest the recent report by Bartolucci et al relating to the use of aspirin for primary prevention of cardiovascular events and wish to submit the following comments relating to the primary prevention of myocardial infarction (MI). The investigators’ report is a follow-up to a previous meta-analysis that was based on 6 trials. We have suggested that the original 6 trials had methodologic defects that argued against their inclusion in a single meta-analysis. The investigators contend that 3 recent trials in particular could enhance the possibility of “detecting moderate, but potentially meaningful, differences that individual trials cannot detect” with respect to the prevention of MI. However, meta-analysis “is only properly applicable if the data summarized are homogenous—that is, treatment, patients and end points must be similar or least comparable.” Pogue and Yusuf, in discussing the limitations of meta-analyses, asserted that the procedure is unlikely to provide the same quality of information about the efficacy of an intervention as a randomized clinical trial if its standards are less stringent. The new analysis appears to violate these fundamental requirements in several important aspects. For instance, in the trials included in the first meta-analysis, 1 was not a randomized clinical trial, and in others, the end points were established on the basis of self-reports, with no provision to detect silent MI (see Kappagoda and Amsterdam for details).

Primary prevention has been defined by the American College of Cardiology and the American Heart Association as the prevention of the first occurrence of cardiovascular disease. Thus, to substantiate a claim for the primary prevention of MI, studies must be performed on subjects who have not sustained previous MI and have no history of clinically overt cardiovascular disease. Of the 3 studies included in the new meta-analysis by Bartolucci et al, 1 was in patients with overt peripheral artery disease (PAD), another in subjects with diabetes, and the third in patients with both. In the study in patients with PAD, the primary end point was a composite of initial (earliest) fatal or nonfatal coronary event or stroke or revascularization. The investigators concluded that no statistically significant differences were found in individual primary end point events between the groups (their Table 2). The trial reported by Ogawa et al was undertaken in patients with type 2 diabetes mellitus, and the primary end point was again a composite of 1 of the following atherosclerotic events during the follow-up period: sudden death; death from coronary, cerebrovascular, and aortic causes; nonfatal acute MI; unstable angina pectoris; newly developed exertional angina; nonfatal ischemic and hemorrhagic stroke; transient ischemic attack; or nonfatal aortic and PAD (arteriosclerosis obliterans, aortic dissection, mesenteric arterial thrombosis). The study showed no significant differences in the incidence of fatal and nonfatal MI (their Table 1). The third study, which was undertaken in patients with type 2 diabetes mellitus and PAD, had 2 hierarchical composite primary end points of death from coronary heart disease or stroke, nonfatal MI or stroke, or amputation above the ankle for critical limb ischemia and death from coronary heart disease or stroke. The differences in primary end points were not statistically significant (their Table 3). Taken individually, none of the new studies showed a statistically significant reduction in MI after the use of aspirin.

A related critical issue is whether these studies could be included in a single meta-analysis along with the 6 trials described previously. Most patients in the latter were in the low- and intermediate-risk categories for coronary heart disease risk on the basis of Framingham risk scores, while the 3 new studies were done in patients in a high-risk category. PAD and type 2 diabetes mellitus are considered coronary artery disease equivalents by the National Cholesterol Education Program Adult Treatment Panel III, and management of these patients cannot be equated with those who are considered at low or intermediate risk by Framingham criteria.

We believe that it is not sound practice to combine the findings of the new studies in patients with coronary heart disease equivalents with those obtained from studies in low- and intermediate-risk patients (based on Framingham criteria). Thus, the new analysis does not provide “additional support” for the claim that aspirin protects against MI in all subjects.