Résumé
To estimate trajectories of sleep lost over worry as a function of age, using longitudinal modeling, and compare these trajectories with those for insomnia symptoms. Data from two prospective, occupational cohorts (the Whitehall II and Finnish Public Sector studies) comprising 84,384 observations from four to eight repeat measurements in 1985-2010. There were 16,408 men and women age 34-79 yr. Age-related trajectories of sleep lost over worry and insomnia symptoms (sleep initiation or maintenance problems, nonrefreshing sleep) were estimated using repeated-measures log-binomial regression analysis and generalized estimating equations. These analyses were adjusted for year of birth and time of measurement to minimize confounding by cohort or period effects. The prevalence ratio for insomnia symptoms was higher in older age groups compared with participants age 34-45 yr. In contrast, the age-related trajectory of sleep lost over worry included two phases: a period of high prevalence of sleep complaints at age 34-60 yr followed by a declining trajectory at older ages. Compared with participants age 34-45 yr, prevalence ratios for sleep lost over worry were 0.63 (0.49-0.80) and 0.59 (0.41-0.84) in the Whitehall II study participants ages 61-65 and 71-79 years. Corresponding figures were 0.62 (0.52-0.75) and 0.46 (0.32-0.66) in the Finnish Public Sector study. This study shows a general age-related decrease in sleep lost over worry between late midlife and old age, a pattern strikingly different from the age-related increase in insomnia symptoms.