Sleep duration is known to significantly affect health in adults and children, but little is understood about long-term associations. This prospective cohort study is the first to examine whether childhood sleep duration is associated with lifelong mortality risk.
Data from childhood were refined and mortality data collected for 1,145 participants from the Terman Life Cycle Study. Participants were born between 1904 and 1915, lived to at least 1940, and had complete age, bedtime, and waketime data at initial data collection (1917-1926). Homogeneity of the cohort sample (intelligent, mostly White) limits generality but provides natural control of common confounds. Through 2009, 1,039 participants had confirmed deaths. Sleep duration was calculated as the difference between each child's bed and wake times. Age-adjusted sleep (deviation from that predicted by age) was computed. Cox proportional hazards survival models evaluated childhood sleep duration as a predictor of mortality separately by sex, controlling for baseline age.
For males, a quadratic relation emerged: Male children who underslept or overslept compared with peers were at increased risk of lifelong all-cause mortality (HR = 1.15, CIs [1.05, 1.27]). Effect sizes were smaller and nonsignificant in females (HR = 1.02, CIs [0.91, 1.14]).
Male children with shorter or longer sleep durations than expected for their age were at increased risk of death at any given age in adulthood. The findings suggest that sleep may be a core biobehavioral trait, with implications for new models of sleep and health throughout the entire life span.