Sleep duration and falls: a systemic review and meta-analysis of observational studies.

Several epidemiological studies have linked sleep duration with falls; however, the findings yielded inconsistent results. No quantitative analysis has specifically assessed the influence of sleep duration on falls. PubMed and Embase were screened for observational studies from inception to 13 September 2016. A generic inverse-variance method was used to pool the outcome data for sleep duration categories of the lowest category versus reference, and the highest category versus reference with a random-effects model. Dose-response analysis was performed to evaluate the potential relationship of sleep duration with falls. Finally, seven eligible observational studies involving a total of 212 829 participants were included in the present meta-analysis. Compared with the reference category, both short and long sleep duration were significantly associated with falls, and the pooled odds ratios (95% confidence intervals) were 1.32 (1.21, 1.46) and 1.35 (1.17, 1.56), respectively, both with evidence of significant heterogeneity. An approximately 'U-shaped' curve was observed, and the lowest risk of falls was shown at 7-8 h day-1 of sleep duration. Further subgroup analysis found that the association of long sleep duration and falls was more pronounced among Caucasians. The present study is limited to its small number of included studies, considerable heterogeneity, observational study design and the large contribution of a single article. Further researches are still needed to confirm the causal association between sleep duration and falls in populations with different gender, age and ethnicity.