In view of conflicting prior reports, we prospectively studied associations between alcohol consumption and subsequent hospitalization for hemorrhagic stroke (HS) in 431 persons. Alcohol use was determined at examinations in 1978-1984 among 128,934 members of a prepaid health plan. Cox proportional hazards models, with 6 covariates, yielded the following multivariate relative risks (95% CI's) for HS: lifelong abstainers (ref) = 1.0, exdrinkers = 0.9 (0.5-1.5), persons drinking <1/month = 1.1 (0.8-1.4), >1/month; <1 drink/day = 0.7 (0.5-0.9), 1-2/day = 0.8 (0.6-1.1), 3-5/day = 1.0 (0.6-1.5), 6+/day = 1.9 (1.0-3.5). Relationships to alcohol were similar for subarachnoid (31% of HS) or intracerebral hemorrhage (69% of HS) and in men or women. Beverage choice (wine, beer, and liquor) was not independently related. We conclude that only heavy drinking is weakly related to increased HS risk and that light drinking need not be proscribed with respect to HS risk.
Copyright 2002 S. Karger AG, Basel