Study prospectively the effect of sleep-related complaints and sleep duration on all cause mortality in a general population sample.

The data were gathered from the adult population from the County of Nord-Trøndelag as part of a general health survey whichhad a participation rate of 71.2%. Data included self-reported somatic disorders, somatic symptoms,health related behaviour, impairment, public benefits, medication use, anxiety and depression as well as anthropometric measures, blood pressure and cholesterol level.

Mortality during a 4-year period following the general health survey as recorded in the Norwegian Death register.

An ordinal five point scale of sleep disturbance predicted mortality in the observation period, even in the probable over-adjusted model including all available confounders. The variables that most strongly accounted for the effects of the sleep disturbance were (in order of magnitude) somatic diagnoses, health related behaviour, anxiety and depression, subjectively reported physical impairment, educational and social differences, blood-pressure, cholesterol level, and BMI. Time in bed was strongly associated with mortality, and the association was U-shaped. Compared to the median value of 7 hours, spending either less or more time in bed predicted death.

Sleep disturbances as well as spending either short or long time in bed are predictors of mortality. Both predictors are robust for adjustment for multiple confounding factors.