Background Insufficient sleep is a public health problem, particularly among adolescents, where sleep duration is often lower than recommended. Insufficient sleep has been associated with weight gain and metabolic dysregulation, with one suggested mechanism being via a reduction in diet quality. To date however, findings have been based on either laboratory studies, which do not reflect variation seen in free-living populations, or observational analyses of cross-sectional associations between usual sleep timings and habitual diet, which are unable to support inferences about the possible direction of causality. In this analysis we use daily measures of sleep timings and diet quality to investigate the associations between each night of sleep and diet quality on the subsequent day, allowing us to assess short-term unidirectional associations over time.

Methods Data are from the ROOTS study, a prospective cohort study recruited from secondary schools in Cambridgeshire and Suffolk (UK). Participants (n = 872) who took part in a sub-study at mean age 15.0y (SD 0.3y) completed a diet diary and wore a combined heart rate monitor and accelerometer over 4 consecutive days. Sleep onset and termination times were determined from review of accelerometer and heart rate traces, and sleep duration and midpoint calculated for each night of data. Daily energy density and daily energy-adjusted fruit and vegetable intakes were derived from diet diaries. Multilevel, random effects models were used to test associations between daily sleep and subsequent day diet, with daily measures nested within individuals and schools, and adjusted for sex, socioeconomic status and day of week. We additionally tested for confounding by daily physical activity energy expenditure, also measured using the combined heart rate monitor and accelerometer.

Results Adolescents slept a mean of 7.9hrs (SD 1.0hr) per night, reporting a mean energy density of 2.09 kcal/g (SD 0.44) and median energy-adjusted daily fruit and vegetable intake of 111.2g (IQR 52.1;202.0). A one hour longer sleep duration was associated with reduced energy density of diet (-0.017 kcal/g, 95%CI -0.034, -0.003), and greater daily intake of fruit and vegetables (7.14 g, 95%CI 3.07, 11.21) the following day. Sleep midpoint did not show significant associations with either energy density diet (-0.001 kcal/g, 95%CI -0.021, 0.019) or fruit and vegetable intake (-4.39 g, 95%CI -9.28, 0.50). Adjustment of models for daily physical activity did not substantially alter the findings.

Conclusion Longer sleep is associated with better diet quality the following day in this cohort of adolescents. This study supports the consideration of sleep duration when designing interventions to promote a healthy diet among adolescents.