Daily associations between time-use and sleep onset timing in children: A multilevel compositional data analysis

OBJECTIVES: This study aimed to determine how time-use behaviors in the 4 hours before a child goes to sleep relate to sleep onset time at a daily level. METHODS: A total of 3438 children (51.6% male, aged 2-4 at baseline) with 13,748 days of time-use diaries completed between 2006 and 2008 were included in this study. Sleep onset time was assessed as the time that parents reported their child was in bed asleep. Time-use in the 4 hours before this time was categorized into screen time, physical activity, non-screen-based leisure, routine behaviors, and other. Associations were estimated using multilevel Bayesian compositional data analysis. RESULTS: At the between-person level, substituting 15 minutes of habitual screen time with physical activity in the 4 hours before bed was associated with earlier sleep onset by 7 minutes in younger children (95% CI: 6-9 minutes) and 8 minutes in older children (95% CI: 7-9 minutes). At the within-person level, higher-than-usual physical activity was associated with earlier sleep among younger (b = -6.19, 95% CI = -7.08, -5.33) and older children (b = -3.34, 95% CI = -4.02, -2.44). More time engaged in routine behaviors were related to later sleep onset at the within-person level among younger children (b = 6.50, 95% CI = 5.05, 7.95) and earlier sleep onset at the between-person level among older children (b = -8.92, 95% CI = -13.33, -7.83). CONCLUSIONS: These results suggest that time-use behaviors in the afternoon and evening are associated with both habitual and day-to-day variability in sleep onset in children.