Relationship Between Sleep Duration and Cognitive Function in Older Adults: Analysis of NHANES and UK Biobank GWAS Data

Objectives To explore the relationship between sleep duration and cognitive functions in older adults using NHANES, a national US population study dataset, and to explore the causal association with Mendelian randomization (MR) using the UK Biobank. Methods First, an observational study was conducted with the NHANES database with participants ≥60 years. Sleep duration was measured with accelerometers for 7 consecutive days. Participants were divided into habitual short sleep (<7 h) and long sleep (>9 h) groups. Cognitive functions were measured with the CERAD Word Learning sub-set, Animal Fluency, and Digit Symbol Substitution test (DSST). Multivariate regression models were used to explore relationships between sleep duration and cognitive functions. Second, bidirectional MR was conducted with data for self-reported sleep duration, which came from a genome-wide association study (GWAS) comprising 446,118 adults from the UK Biobank, and general cognitive performance, which was obtained from a recent GWAS study ( N = 257,841). Inverse-variance weighted (IVW) was used as the primary estimation of the outcome. Results In the observational study, 2687 participants were included. Sleep duration was associated with cognitive functions in a non-linear way. Habitual long sleep (>9°h) was associated with lower scores on DSST (OR = 0.01, p = .003) in the fully-adjusted model. The association between habitual short sleep and cognitive functions was insignificant. For the MR, genetically predicted lower general cognitive performance was causally associated with a higher prevalence of habitual short sleep (OR = 0.97, p = 5.1 × 10 −7 ) and long sleep (OR = 0.97, p = 8.87 × 10 −16 ). Discussion Short and long sleep duration might be both causally associated with worse outcomes of cognitive functions in older adults, highlighting the importance of maintaining sleep health.