Correlates of sleep variability in a mobile EEG-based volunteer study

Abstract

Variable sleep patterns are a risk factor for disease, but the reasons some people express greater within-individual variability of sleep characteristics remains poorly understood. In our study, we leverage BSETS, a novel mobile EEG-based dataset in which 1901 nights in total were recorded from 267 extensively phenotyped participants to identify factors related to demographics, mental health, personality, chronotype and sleep characteristics which predict variability in sleep, including detailed sleep macrostructure metrics. Young age, late chronotype, and napping emerged as robust correlates of increased sleep variability. Correlations with other characteristics (such as student status, personality, mental health and co-sleeping) generally disappeared after controlling for age. We critically examine the utility of controlling the correlates of sleep variability for the means of sleep variables. Our research shows that age and sleep habits affecting the amount of sleep pressure at night are the most important factors underlying sleep variability, with a smaller role of other psychosocial variables. The avoidance of daytime naps emerges as the most promising modifiable behavior associated with increased sleep regularity.