Objectives: Previous studies have indicated that improvement in sleep duration might correlate with better cognition. We aimed to examine the associations between changes in sleep duration and cognitive function.

Results: A change from short sleep duration (SSD) to moderate sleep duration (MSD) was associated with better global cognition scores (β=0.54, P <0.01). A change from SSD to long sleep duration (LSD) (β=-0.94, P <0.001) or a change from LSD to SSD (β=-1.38, P <0.01) was associated with lower global cognition. For individuals with MSD, a≥2 h increase (β=-0.89, P <0.001) or decrease (β=-0.70, P <0.001) in sleep duration was associated with lower global cognition.

Conclusions: For short sleepers, improvement in sleep duration correlated with better cognition. For long sleepers, there was no need to reduce sleep duration. Excessive changes or deviation from the moderate duration was associated with lower cognition.

Methods: A total of 10325 individuals aged 45 and older from the China Health and Retirement Longitudinal Study (CHARLS) were included. Self-reported nocturnal sleep duration and cognitive function were assessed in the three waves of CHARLS from 2011 to 2015. Cognitive function was assessed by a global cognition score, which included episodic memory, visuospatial abilities, calculation, orientation and attention.