Aspirin and Stroke

n this issue of Stroke, Kronmal and colleagues report the intriguing finding that in a cohort of elderly people, self-selection for aspirin use was associated with increased risks of ischemic stroke in women and hemorrhagic stroke in men and women.The authors are commendably cautious in their interpretation of the findings and raise the possibility of confounding by reasons for aspirin use as an explanation for the observed association.However, it is useful to view these findings in the context of the totality of available evidence.1,2 Despite 100 years of widespread use, the potential for aspirin to affect risks of stroke has only been recognized relatively recently, due to the Nobel prize-winning work of Sir John Vane, who demonstrated that aspirin permanently inhibits cyclooxygenase-dependent platelet aggregation.3 Since that time, case-control and cohort studies conducted largely among middle-aged populations have suggested that individuals who self-select for aspirin therapy tend to have small (i.e., 20% to 30%) decreased risks of total and ischemic stroke.4 If true, such benefits would be clinically very worthwhile and would have an important public health impact.Unfortunately, however, the amount of uncontrolled and uncontrollable confounding in such studies, no matter how well designed and conducted, is likely to be as large as the effects being sought.In such circumstances, the most reliable design strategy to answer the question definitively is a large-scale randomized trial.5 In a collaborative worldwide overview of all randomized trials of antiplatelet agents, aspirin was shown to reduce the risk of nonfatal stroke by 25%.6 This reduction was a weighted average of a 31% (Ϯ5%) highly statistically significant reduction among high-risk patients with a history of occlusive disease and a possible but statistically nonsignificant 21% (Ϯ13%) increase among all low-risk patients (ie, primary prevention).The latter uninformative null result derived primarily from inadequate numbers of stroke end points in the two completed primary prevention trials, the US Physicians' Health Study and the British Doctors' Trial, both of which included only men. 1 With respect to cerebral hemorrhage, the data from all these randomized trials taken together are compatible with about a 1 per 1000 excess due to