Relation of dietary glycemic load with ischemic and hemorrhagic stroke: a cohort study in Greece and a meta-analysis

High glycemic load (GL) has been associated with excess stroke risk. Data suggest a different role of diet in the etiology of ischemic and hemorrhagic stroke. We analyzed data from 19,824 participants of the Greek cohort of the population-based European Prospective Investigation into Cancer and nutrition (EPIC), who were free of cardiovascular diseases, cancer, and diabetes at baseline and had not developed diabetes. Diet was assessed at enrollment through a validated, interviewer-administered semi-quantitative food frequency questionnaire. The average daily GL was derived using standard tables. We also conducted a meta-analysis on GL and stroke (overall, ischemic and hemorrhagic), using random-effects models. In the Greek EPIC cohort, 304 incident stroke cases were identified (67 ischemic, 49 hemorrhagic). Using Cox proportional hazards regression models adjusted for potential confounders, the hazard ratios for the highest versus the lowest GL tertiles were 1.07 [95 % confidence interval (CI) 0.74-1.54] for overall stroke, 1.55 (95 % CI 0.72-3.36) for ischemic and 0.48 (95 % CI 0.18-1.25) for hemorrhagic stroke (p-heterogeneity < 0.01). The meta-analysis, including a total of 3,088 incident cases and 247 deaths from stroke (1,469 cases and 126 deaths ischemic; 576 cases and 94 deaths hemorrhagic), estimated pooled relative risks for the highest versus the lowest GL levels of 1.23 (95 % CI 1.07-1.41) for overall, 1.35 (95 % CI 1.06-1.72) for ischemic, and 1.09 (95 % CI 0.81-1.47) for hemorrhagic stroke (p-heterogeneity = 0.275). This study indicates that GL is an important determinant of the more common ischemic-though not of the hemorrhagic-stroke.