In a case-cohort drawn from 54,546 Danes aged 50-65, 729 hemorrhagic strokes accrued over a median 21 years. Adults in the top quintile of adipose-tissue eicosapentaenoic acid (EPA) had a 36% lower risk than the bottom quintile (HR 0.64, 95% CI 0.50-0.82), with similar patterns by sex and stroke subtype, though subarachnoid hemorrhage fell short of significance. DHA, DPA and combined EPA+DHA showed no significant associations, while top-quintile alpha-linolenic acid (ALA) carried a 32% higher risk (HR 1.32, 1.01-1.72).
Adipose tissue is a strong biomarker because it reflects long-term fatty acid exposure rather than recent meals, which sidesteps the recall error of food questionnaires. Still, this is observational work: EPA content tracks fish intake, and fish eaters tend to differ in other healthy ways, so residual confounding remains plausible. The EPA-specific signal is intriguing, since EPA's antiplatelet effects have long raised worry about bleeding, and these data point the opposite way, consistent with mixed trial findings on omega-3 and intracerebral bleeding. The ALA result is borderline, sits alongside inconsistent literature, and should not drive any change in eating habits. Because DHA showed nothing, mechanism claims are premature. Treat this as reassuring and hypothesis-generating: it adds to the case that regular oily-fish intake is not a stroke-bleeding hazard, but it cannot justify EPA supplementation, which needs randomized evidence. Confirmatory and incremental, not paradigm-shifting.