QUANTIFYING THE TEMPORAL RELATIONSHIP BETWEEN SELF-REPORT SLEEP QUALITY AND COGNITION IN OLDER ADULTS

Abstract Introduction Poor sleep is a promising modifiable risk factor for impaired cognition in older adults. However, the relationship between sleep and cognition is likely bi-directional, and few studies have examined these temporal associations. We seek to investigate the temporal relationships between self-report sleep quality and global cognition. Methods Our analytic sample includes 1,610 participants from the Memory and Aging Project and Minority Aging Research Study without cognitive impairment at the initial visit (41% black, 77% female, mean[min,max] age = 77[54,100] years). Participants have cognition and sleep quality measured at an initial visit and up to 14 years of annual follow-up (median 6 years). Sleep quality was measured using a modified 10-item Pittsburgh Sleep Quality Index score (higher scores indicating worse quality) and standardized; global cognition was a composite z-score computed from an average of 19 cognitive tests. We used linear mixed effects models to quantify the concurrent and prospective (1-year) relationships of sleep quality and global cognition. Quadratic terms were also tested to allow for a potentially U-shaped relationship. Results When examining same-year associations with cognition as the outcome, sleep quality and cognition exhibit a negative quadratic association (linear term BL[p] = 0.01[0.021]; quadratic term BQ[p] = -0.01[0.051]), indicating that both better- and worse-than-average sleep quality are associated with lower cognition. Regarding 1-year associations, both better- and worse-than average sleep quality predict worse next-year global cognition (BL[p] = 0.01[0.008], BQ[p] = -0.01[0.033]). In contrast, better-than-average cognition predicts worse next-year sleep quality (B L[p] = 0.05[p=0.005]; BQ[p] = -0.01[0.650]) with a stronger association in this direction. Conclusion Understanding the temporal association between sleep and cognition has important implications for screening and development of novel treatments and interventions. The finding that both better and worse sleep quality are associated with worse cognition may reflect an underreporting of poor sleep symptoms in older adults with worsened cognition. Future work will examine these associations considering specific domains of self-report sleep (e.g., timing, efficiency, duration) and cognitive function (memory and perception), consider mechanisms relating sleep and cognition, and use objective measures of sleep (e.g., actigraphy). Support (If Any) RF1AG056331 (PI: Wallace), R01AG17917, R01AG22018