Importance: An association between sleep duration and the trajectory of cognitive decline has not been conclusively demonstrated.
Objective: To investigate the association between sleep duration and cognitive decline by a pooled analysis of 2 nationally representative aging cohorts.
Design, setting, and participants: A pooled cohort study using data from waves 4 to 8 (2008-2009 to 2016-2017) in the English Longitudinal Study of Ageing and waves 1 to 3 (2011 to 2015) in the China Health and Retirement Longitudinal Study in a population-based setting. Participants were 2 randomly enrolled cohorts comprising 28 756 individuals living in England who were 50 years or older and those living in China who were 45 years or older.
Exposure: Self-reported sleep duration per night according to face-to-face interviews.
Main outcomes and measures: Global cognitive z scores were calculated according to immediate and delayed recall test, an animal fluency test, the serial sevens test, an intersecting pentagon copying test, and a date orientation test.
Results: Data were analyzed from 20 065 participants, including 9254 from the English Longitudinal Study of Ageing (mean [SD] age, 64.6 [9.8] years; 55.9% [5174 of 9254] women; median follow-up duration, 8 [interquartile range, 6-8] years) and 10 811 from the China Health and Retirement Longitudinal Study (mean [SD] age, 57.8 [9.0] years; 50.2% [5425 of 10 811] men; median follow-up duration, 4 [interquartile range, 4-4] years). During 100 000 person-years of follow-up, global cognitive z scores in individuals with 4 hours or less (pooled β = -0.022; 95% CI, -0.035 to -0.009 SD per year; P = .001) and 10 hours or more (pooled β = -0.033; 95% CI, -0.054 to -0.011 SD per year; P = .003) of sleep per night declined faster than in the reference group (7 hours per night) after adjusting for a number of covariates. An inverted U-shaped association between sleep duration and global cognitive decline was also observed.
Conclusions and relevance: In this pooled cohort study, an inverted U-shaped association between sleep duration and global cognitive decline was found, indicating that cognitive function should be monitored in individuals with insufficient (≤4 hours per night) or excessive (≥10 hours per night) sleep duration. Future studies are needed to examine the mechanisms of the association between sleep duration and cognitive decline.