The U-shaped association of self-reported sleep duration with all-cause mortality is generally accepted. Findings on cardiovascular (CVD) mortality and morbidity are inconsistent. We aimed to further clarify the associations of the self-reported sleep duration with CVD mortality and morbidity.
In two population based surveys in 1972 and 1977 the levels of coronary risk factors in Finland and habitual sleep duration were measured; 25,025 individuals were followed-up until 2006 by the national register data. The outcome variables were death (for any reason), CVD death, and non-fatal CVD event (non-fatal myocardial infarction or stroke). Participants with former non-fatal CVD event at baseline were excluded from CVD analyses, and socio-demographic and health-related confounders were considered in the final Cox proportional hazard models for both genders.
The U-shaped association of total mortality with self-reported sleep duration was confirmed in both genders. The association of CVD mortality with self-reported sleep duration was independent of pertinent cardiovascular risk factors in women. The highest CVD mortality risk was found in both extreme ends of sleep duration distribution (?5 and ?10h sleepers).
Sleep duration is an independent risk factor for CVD mortality and morbidity in women but not in men. The highest CVD mortality risk is associated with the extreme ends of sleep duration distribution. Thus, in epidemiological studies, combining adjacent (6 and 9h) sleep duration groups with the extreme groups may partly mask the mortality risks, especially in the long run.