Sleep disorders and mortality: A prospective study in the Canadian longitudinal study on aging

Abstract

Background and objectivesSleep disorders are commonly linked to various health conditions, although it remains unclear to what degree they are linked with overall mortality. We compared mortality in different self-reported sleep disorders in a large population-based prospective study. MethodsIn this case-control study within the CLSA cohort, participants completed a questionnaire at baseline (2011–2015) measuring overall sleep satisfaction, daily sleep duration, sleep-onset and sleep-maintenance insomnia, daytime somnolence, REM sleep behavior disorder (RBD), restless leg syndrome (RLS), and obstructive sleep apnea (OSA). The vital status of participants was assessed in July 2019. Baseline sleep problems of participants who died (cases) were compared to those who survived (controls). For each case, five age/sex-matched controls were selected. Binary logistic regression was used to estimate the association between sleep symptoms and mortality, adjusting for age, sex, marital status, province, education, alcohol consumption, smoking, caffeine, and body mass index. In a complementary model, anxiety and depression were also added. ResultsAmong 30,097 participants at baseline, 974 deaths were reported in 2019 (60.7 % male, age = 72.3 ± 9.4 years). In the initial analysis, mortality cases reported more baseline sleep-maintenance insomnia (12.1 % vs. 8.0 %, Adjusted OR[95%CI] = 1.62[1.15,2.29]), daytime somnolence (2.4 % vs. 1.1 %, AOR = 2.70[1.34,5.44]), and higher possible RLS (16.4 % vs. 12.4 %, AOR = 1.50[1.09,2.05]). They were also more likely to screen positive for possible OSA (33.8 % vs. 24.2 %, AOR = 1.32[1.07,1.64]); however, this effect was not related to core apnea symptoms. Sleep durations exceeding 10 h/day were also associated with increased mortality (3.4 % vs. 1.9 %, AOR = 1.83[1.04,3.24]). Other sleep symptoms/disorders, such as sleep-onset insomnia (7.3 % vs. 4.3 %, AOR = 1.54 [1.00,2.37]), possible RBD (5.3 % vs. 5.1 %, AOR = 1.02[0.62,1.69]), and overall sleep dissatisfaction (26.5 % vs. 22.6 %, AOR = 1.14[0.93,1.41]) were not different among these groups. After adding anxiety and depression to the adjustment model, all differences attenuated to become statistically non-significant, except for daytime somnolence disorder. When stratified by sex, the association between sleep disorders and mortality was only observed in women, with men showing no association. DiscussionWe confirm a relationship between numerous sleep disorders and mortality. This effect is most evident in women, and appears to be strongly related to co-existing anxiety and depression.