Abstract
STUDY OBJECTIVES: To examine whether there is a temporal association between sleep disturbance and multimorbidity. METHODS: We performed a cross-sectional and longitudinal observational analysis in people aged 40 years or more, recruited from the knee pain and related health in the community cohort study. The primary exposure was the Sleep Problems Index II score in tertiles measured at baseline. The primary outcome was count of chronic conditions developed in 5 years. Pain, low mood, and anxiety were measured at 2 years as mediators. Poisson regression was used to calculate adjusted relative risk and 95% confidence intervals. RESULTS: We included 4488 participants in the cross-sectional analysis at baseline and 1941 in the 5-year longitudinal analysis. At baseline, the adjusted relative risks for prevalent multimorbidity were 1 (reference) for tertile 1, 1.09 (95% confidence interval; 1.01-1.18) for tertile 2, and 1.21 (95% confidence interval; 1.11-1.32) for tertile 3 of the sleep disturbance score (p for trend <.001). Of the total association between sleep disturbance and multimorbidity, 14 per cent (95% confidence interval; 9% to 19%) were mediated by pain and 7 per cent (95% confidence interval; 2% to 13%) by low mood. In the 5 year follow-up, the adjusted relative risk for incident multimorbidity were 1 (reference) for tertile 1, 1.12 (95% confidence interval; 0.98-1.28) for tertile 2, and 1.25 (95% confidence interval; 1.06-1.47) for tertile 3 (p for trend .007). Of the total association between sleep disturbance and multimorbidity, 10 per cent (95% confidence interval; 2% to 18%) was mediated by pain. CONCLUSIONS: Sleep disturbance is associated with multimorbidity. The association is dose-dependent, temporal, and partially mediated by pain.