To assess sleep duration-mortality relationships across a 27-year follow-up period in a well characterized random sample of older people, and to test the hypothesis that mortality risks associated with long sleep duration confound with, and can be explained by, low levels of functional capacity indicative of frailty.

Face-to-face interviews conducted among 1002 randomly sampled older (65+) people in 1985 provided baseline profiles of health, functional capacity, physical activity, and sleep quality and duration. Health and functional status in each of 6 sleep duration categories (≤4, 5, 6, 7, 8, ≥9 h) was examined. At censorship in 2012, 927 deaths were recorded. Relationships between sleep duration and 27-year all-cause mortality were then examined in a series of incrementally adjusted Cox regression models.

Associations between sleep duration and measures of sleep quality were predominantly linear, with longer sleep times indicating superior sleep quality. Relationships between sleep duration and functional capacity, on the other hand, were predominantly quadratic, with most approximating a U-shaped function. Adjusted for age, gender, social class, insomnia symptoms, physical health, depression, BMI and smoking status, long sleep duration and continuous hypnotic drug use at baseline were significantly and independently associated with elevated mortality risk (HR: 1.37; 95% CI: 1.05–1.78; HR: 1.24; 95% CI: 1.01–1.51). When indices of frailty were added to the model, hazard ratios for long sleep duration and hypnotic drug became non-significant, while the lowest physical activity quintile and very slow walking speed significantly increased mortality risk (HR: 1.79; 95% CI: 1.40–2.30; HR: 1.41; 95% CI: 1.15–1.73 respectively).

In analyses of sleep-related mortality outcomes long sleep durations confound with, and may be indicative of, incipient frailty among older participants.