Sleep duration and total cancer mortality: a meta-analysis of prospective studies.

Objective: Epidemiological evidence suggests a possible association between sleep duration and cancer-related mortality, but the reported findings are inconsistent. We conducted a meta-analysis of prospective studies to evaluate the relationships between sleep duration and cancer mortality.

Methods: Potentially relevant studies were identified by searching PubMed and Embase databases in addition to manual searches of references of retrieved full publications. The summary relative risks (RRs) with 95% confidence intervals (CIs) were computed using a random-effect model. The meta-regression analyses were performed to explore any potential effect modifier.

Results: A total of 17 reports from 11 independent prospective studies were included in this meta-analysis. When comparing with reasonable sleep duration (mostly defined as 7 or 7-8 h), the summary RR for long sleep duration (mostly defined as ≥9 or ≥10 h) and short sleep duration (mostly defined as ≤6 or ≤5 h) was 1.11 (95% CI = 1.05-1.18) and 1.05 (95% CI = 0.99-1.11), respectively, with little evidence of heterogeneity. There was evidence of publication bias for the association of long sleep duration with cancer mortality, and the summary RR was slightly attenuated to be 1.10 (95% CI = 1.02-1.18) after using a statistical method to correct for the bias.

Conclusion: This meta-analysis of prospective studies suggests that long, not short sleep duration is associated with significantly increased risk of total cancer mortality.