Short sleep duration and adolescent health: does weekend catch-up sleep work and for whom?

Objectives: Despite evidence that sleep duration affects adolescent health, there remain important research gaps in the literature. Little is known about (1) whether the association between weekday sleep duration and health is confounded by unobserved individual heterogeneity and (2) the extent to which weekend catch-up sleep (WCS) duration moderates this association. This study addresses these gaps.Study design: Using six waves of longitudinal data from the 2011-2016 Korean Children and Youth Panel Survey (N = 6633), this study examined the relationship between weekday sleep duration, WCS duration, and two measures of adolescent health, body mass index (BMI) and self-rated health (SRH).Methods: We estimated fixed effects models to account for individual-level heterogeneity.Results: Fixed effects estimates suggest that part of the associations between short sleep duration and adolescent health are confounded by unobserved individual heterogeneity (62% for BMI and 30% for poor SRH), although the associations remain statistically significant. Sleeping less than 6 h increased BMI by 0.203 and the probability of reporting poor SRH by about 2 percentage points. Controlling for individual heterogeneity, however, changed the sign of the WCS duration coefficient, suggesting that a longer WCS duration is positively associated with BMI (b = 0.021). No such patterns were found for SRH.Conclusions: Short weekday sleep duration threatens adolescent health. WCS duration is protective only for those who are most sleep deprived.(c) 2022 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

Automated summary

This study addresses important research gaps by examining the relationship between sleep duration and adolescent health in South Korea. The findings suggest that the association between short sleep duration and health is partly confounded by unobserved individual heterogeneity, and weekend catch-up sleep has limited moderating effects on this association.