We thank Drs. Hart and Solomon for their comments. First, we clearly stated in our article that our conclusions may not apply to patients with active coronary disease who require the use of aspirin, because these patients were excluded from our study. Consequently, the number of nonfatal myocardial infarctions seen in our study was too small to draw any firm conclusions on the effectiveness of aspirin in preventing this outcome.

We were interested in the effect of aspirin on the risk for a composite of clinical end points. It is clear from our data that the annual rate of nonfatal myocardial infarction in such patients is about 1%. Even under the extreme assumption that aspirin could reduce this risk by 50%, about 200 patients would need to be treated annually to prevent one nonfatal myocardial infarction. The clinical relevance of this hypothetical effect of aspirin would be doubtful because this degree of effectiveness would be substantially lower than that of other interventions in the vascular area. It would also be limited to a single nondisabling and nonfatal outcome and would ignore some of the morbidity associated with long-term aspirin use in older persons.

We did not cite the Mayo Clinic trial [1] because 1) this surgical trial was not designed to specifically assess the effect of aspirin for the prevention of myocardial infarction in neurologically asymptomatic persons; 2) three of the eight myocardial infarctions reported in the surgical group of the randomized portion of that study occurred in the period immediately after surgery, suggesting a potential confounding effect of surgery; and 3) occurrence of myocardial infarction after endarterectomy for asymptomatic carotid disease has been well documented even in patients taking aspirin [2]. Data from the results of an interim safety analysis, not originally intended to answer a specific clinical question, do not constitute strong evidence for efficacy but rather an observation open to interpretation.

The sex distribution in our study is identical to the one reported by Norris and coworkers [3] in another large observational study in which patients taking aspirin for cardiac reasons were not excluded. As a result, the rate of cardiac events in these patients was higher, but the incidence of neurologic events was similar to that in our study.

We agree with Dr. Browner that the expression "with 95% confidence" is not the most accurate one. A more precise, although rather lengthy, formulation would be the following: "The 95% CI indicates that our results are not consistent with the hypothesis that aspirin reduces the risk for the composite outcome by 40% or more and therefore are rather definitive in ruling out this hypothesis" [4].

Finally, we thank Drs. Creutzig and Ranke for their useful comments and agree that a higher aspirin dose may have a beneficial effect.