PubMed and Embase databases were searched through January 2016, and multivariate-adjusted relative risks were pooled by using fixed-effects models. Semiparametric and dose–response methods were used to assess the relationship of sleep duration and risk of stroke and stroke mortality.
Eleven articles with 16 independent reports were included in our meta-analysis. An approximate J-shaped relationship was detected between sleep duration and risk of stroke and stroke mortality. No evidence of a curve linear relationship was seen between sleep duration and risk of stroke or stroke mortality. Compared with 7-h sleep duration per day, the pooled relative risks for stroke events were 1.07 (95% CI 1.02–1.12) for each 1-h shorter sleep duration among individuals who slept < 7 h per day and 1.17 (1.14–1.20) for each 1-h increase of sleep duration among individuals with longer sleep duration and the pooled RR for stroke mortality was 1.17 (95% CI 1.13–1.20) per 1-h increase of sleep duration.
Both short and long duration of sleep are predictors of stroke outcomes, and long sleep duration is significant marker of stroke mortality.