25-Hydroxyvitamin D and Symptomatic Ischemic Stroke: An Original Study and Meta-Analysis

Objective: We tested the hypothesis that low plasma concentrations of 25-hydroxyvitamin D are associated with increased risk of symptomatic ischemic stroke in the general population. Methods: We measured plasma 25-hydroxyvitamin D in 10,170 individuals from the general population, the Copenhagen City Heart Study. During 21 years of follow-up, 1,256 and 164 persons developed ischemic and hemorrhagic stroke, respectively. In a meta-analysis of ischemic stroke, we included 10 studies, 58,384 participants, and 2,644 events. Results: Stepwise decreasing plasma 25-hydroxyvitamin D concentrations were associated with stepwise increasing risk of ischemic stroke both as a function of seasonally adjusted percentile categories and as a function of clinical categories of 25-hydroxyvitamin D (p for trend <= 2 x 10(-3)). In a Cox regression model comparing individuals with plasma 25-hydroxyvitamin D concentrations between the 1st and 4th percentiles to individuals with 25-hydroxyvitamin D concentrations between the 50th and 100th percentiles, multivariate adjusted hazard ratio of ischemic stroke was 1.82 (95% confidence interval, 1.41-2.34). Comparing individuals with clinical categories of severe vitamin D deficiency (<25.0nmol/l [<10.0ng/ml]) to individuals with optimal vitamin D status (>= 75.0nmol/l [>= 30.0ng/ml]), the multivariate adjusted hazard ratio of ischemic stroke was 1.36 (1.09-1.70). 25-Hydroxyvitamin D concentrations were not associated with risk of hemorrhagic stroke. In a meta-analysis comparing lowest versus highest quartile of 25-hydroxyvitamin D concentrations, the multivariate adjusted odds ratio of ischemic stroke was 1.54 (1.43-1.65) with a corresponding hazard ratio of 1.46 (1.35-1.58) in prospective general population studies. Interpretation: In this large population-based prospective study, we observed stepwise increasing risk of symptomatic ischemic stroke with decreasing plasma 25-hydroxyvitamin D concentrations. This finding was substantiated in a meta-analysis. ANN NEUROL 2013;73:38-47