Both sleep and cognition are multidimensional constructs. Using univariate methods to examine associations between sleep and cognition may inadequately characterize the association between these arrays of variables. The current study used a multivariate approach to identify key sleep metrics and cognitive domains contributing to the maximum sleep-cognition covariance in healthy older adults. In 773 community-dwelling older adults of ages 65-80 years, sleep was assessed using the Oura Ring worn for 15-28 days. Cognition performance in seven domains was assessed using standardized tests. The overall covariance between sleep and cognition was examined by a partial least square correlation (PLSC) analysis. Sleep metrics and cognitive domains contributing to significant PLSC components were identified by bootstrapping. PLSC analysis identified a component that explained 82 % of covariance between sleep and cognition matrices (r = 0.2, p < 0.001). Bootstrapping tests further identified 11 sleep continuity and regularity metrics and 3 corresponding cognitive domains that contributed significantly to the observed covariance. Post-hoc univariate analyses showed that sleep continuity metrics correlated with speed of processing, while sleep regularity metrics correlated with verbal memory, executive functions, and speed of processing. Our results suggest that sleep continuity and regularity may be more sensitive markers of impairments across multiple cognitive domains in healthy aging compared to sleep duration and timing.
Keywords: Cognitive performance; Multivariate analysis; Sleep health; Sleep measures; Wearable sleep tracker.
Copyright © 2025 The Authors. Published by Elsevier B.V. All rights reserved.