Cross-sectional and Longitudinal Relationships Between Multi-Dimensional Sleep Composites and Cognition in...

Sleep is critical for maintaining overall brain health and may influence the trajectory of cognitive impairment. Using a multidimensional conceptualization of sleep, we hypothesized that data-derived multidimensional sleep composites would be associated with cognitive performance.

Data from four examinations (2010-2024) in the Multi-Ethnic Study of Atherosclerosis were used to examine cross-sectional and longitudinal associations between multimodal sleep and cognition. Multidimensional sleep composites were derived from 39 objective and subjective sleep variables using principal components analysis (PCA). Cognition was measured using a global cognition composite (GCC), a standardization of total scores across Digit Symbol Coding (DSC), Digit Span (DS), and Cognitive Abilities Screening Instrument (CASI). Linear regression analyzed cross-sectional associations between sleep composites and cognition, and linear mixed-effects models examined associations with within-person cognitive change over time.

In cross-sectional analyses of 1,628 participants (mean age 68.0; 54.9% female), 13 sleep composites explained about 71% of the sample's variance. After adjusting for demographic and lifestyle factors, five sleep composites were associated with GCC cross-sectionally: (1) EEG power spectral density (ß (per SD) = 0.065; 95% CI 0.03-0.10), (2) actigraphy based sleep continuity (ß = 0.049; 0.01-0.08), (3) spindle density (ß = 0.052; 0.02-0.09), (4) spindle-slow oscillation coupling (ß = 0.050; 0.01-0.09), and (5) REM sleep (ß = 0.047; 95% CI 0.01-0.08) In the longitudinal analysis, actigraphy based sleep continuity and variability were associated with cognitive decline, although significance decreased after multiple comparison adjustment.

EEG-based sleep microstructure and actigraphy-based sleep continuity were associated with cognitive performance cross-sectionally, while actigraphy-based sleep continuity was associated with cognition longitudinally. These findings support a multidimensional framework for understanding sleep-related risk factors for cognitive decline.