Sleep electroencephalography-derived brain health and long-term cognitive decline in community-dwelling adults

Abstract

Abstract Study Objectives Sleep electroencephalography (EEG) is linked with cognitive function and dementia risk. Integrated EEG-derived scores, such as the Brain Health Score (BHS), trained to jointly reflect cognitive function and disease risk, and the Brain Age Index (BAI), which estimates brain aging, also correlate with neuropsychological (NP) performance. Yet, it remains unclear whether these sleep scores are associated with the rate of future cognitive decline. Methods This study included 639 Framingham Heart Study Generation 2 and Omni 1 participants with sleep EEG in mid-to-late life and ≥ 2 NP tests over an average of 14.3 years. Linear mixed effects models estimated associations between BHS and annual changes in memory, executive function, or language. We also evaluated associations between BAI and cognitive decline among 509 Generation 2 participants with BAI values. Results Participants were, on average, 58.4 years at sleep assessment, 54 per cent female, and 86 per cent non-Hispanic white. Each 1-SD higher BHS was associated with 0.015 SD/year slower memory decline (95% CI: 0.009–0.021), a 0.009 SD/year slower executive function decline (95% CI: 0.005–0.014), and a 0.011 SD/year slower language decline (95% CI: 0.005–0.017). In contrast, BAI was not associated with changes in any cognitive domain. Conclusions Higher sleep EEG–derived brain health in mid-to-late life was associated with a slower decline in multiple cognitive domains over more than a decade of follow-up, whereas an EEG-based measure of brain aging was not. These findings suggest that sleep EEG-derived, data-driven brain health metrics may have value as novel biomarkers for the risk of cognitive decline. Statement of Significance This study demonstrates associations between a sleep-derived brain health score (BHS) in mid-to-late life and subsequent cognitive decline in memory, executive function, and language. Previous work has linked this sleep score with both traditional and digital neuropsychological performance. The current study extends this evidence to suggest that higher BHS is also related to slower rates of prospective cognitive decline. Taken together, these findings provide additional support for the potential utility of using this data-driven sleep score as a digital biomarker for the risk of cognitive decline. Future research is needed to identify whether the BHS predicts dementia risk and to replicate this study in more diverse samples to test the generalizability of these findings.