Duração do sono, insônia e função cognitiva: performance e declínio em quatro anos de seguimento na coorte ELSA-Brasil

Abstract

Objectives: The objectives of this thesis were twofold: 1) to investigate the isolated and combined association between sleep disorders (sleep duration, insomnia symptoms in the last 30 nights and daytime tiredness) and performance on cognitive tests; and 2) evaluate the longitudinal association between insomnia symptoms (difficulty falling asleep or staying asleep or waking up earlier) and cognitive performance in different domains, and verifying whether the associations found varied according to the level of education. Methods: Results presented in two original articles. In the first, a cross-sectional analysis was carried out including participants from visit 2 (2012-2014) of the Longitudinal Study of Adult Health (ELSA-Brazil). Polynomial regression with a quadratic term and multiple linear regression models were performed to evaluate isolated and combined associations between sleep disturbances and duration and performance in memory, fluency, executive function, and global cognition. The second article is a longitudinal analysis with data obtained in visits 2 (2012-2014) and 3 (2017-2019) of ELSA-Brazil, totaling around four years of follow-up. Mixed-effects linear regression with random intercept were used to estimate associations and test the impact of insomnia on longitudinal changes in standardized scores for tests of memory, fluency, and executive function performed at the two visits. The variable time between visits was estimated by the difference between the ages of participants in Visits 3 and 2. An interaction term between education and insomnia was added to the final models to verify whether the level of education modified the association found between insomnia and cognition. Results: In the first article, inverted U-shaped associations were observed between sleep duration and cognitive performance in all abilities, indicating that durations shorter or longer than seven hours were associated with worse performance, regardless of age. Reporting insomnia was associated with worse performance on the executive function test (β: -0.08; 95% CI -0.15 to -0.01), with the magnitudes of associations being greater for individuals with insomnia at two or more moments (β: -0.12; confidence interval [CI]95% 0.19 to -0.05) or, especially, for insomnia combined with short sleep (β: -0.18; CI95% -0.24 to -0.11). Insomnia in two or more periods was also associated with lower performance on the memory test and global cognition. There was no association between any sleep disorder and verbal fluency. Isolated daytime fatigue was not associated with performance in the evaluated tests. In the second article, frequent insomnia symptoms at the beginning of sleep were associated with worse performance in verbal fluency (β: -0.0998; CI95%: -0.1774 to -0.0222) and in executive function (β: -0.0899; CI95%: - 0.1655 to -0.0143), with evidence of a doseresponse gradient in the association with the frequency of insomnia in this last ability. Insomnia at the end of sleep, almost always/always, was associated with worse performance in the memory test (β: -0.0918; CI95%: -0.1774; -0.0062) and executive function (β: -0.0918; CI95%: -0.1734 to -0.0162). Mid-sleep insomnia was inconsistently associated with cognitive performance: while those who showed symptoms sometimes had worse memory performance (β: -0.0658; CI95%: -0.1263 to -0.0054), those who had difficulty sleep maintenance, almost always/always, performed better on the verbal fluency test (β: 0.1258; CI95%: 0.0440 to 0.2075). However, insomnia symptoms were not associated with changes in the speed of cognitive decline between the two visits. We also did not verify any interaction between the level of education and the report of insomnia with the performance in any cognitive abilitie assessed. Conclusion: The results suggested that extreme sleep duration is detrimental to almost all cognitive functions investigated, while insomnia appears to most strongly affect executive function. Although insomnia symptoms at different periods were shown to be detrimental to abilities such as memory, verbal fluency and executive function, these symptoms did not alter the speed of cognitive decline during the study period. Furthermore, the results do not support the hypothesis that education modifies the association between insomnia and cognition.