LAUSR

OBJECTIVES Assess longitudinal associations between diary-measured sleep duration and clinically assessed body mass index (BMI). DESIGN Multilevel growth curve analyses examined how within-person changes and between-person differences in habitual sleep duration were associated with BMI trajectories. SETTING Sleep diaries across 2-6 consecutive weekday and weekend nights at each data collection point, repeatedly collected at approximate 4-year intervals, for an average of 9.2 (standard deviation [SD] = 3.6) years between 1989 and 2011. PARTICIPANTS About 784 participants (47% women) enrolled in the Wisconsin Sleep Cohort Study (mean [SD] age = 51.1 [8.0] years at baseline). MEASUREMENTS The outcome variable was BMI (kg/m2). Key predictors were habitual sleep duration (defined as average weekday nighttime sleep duration) and sleep duration differential (defined as the difference between average weekday and average weekend nighttime sleep duration) at each data collection wave. RESULTS Men with shorter habitual sleep duration on weekdays had higher BMI than men with longer habitual sleep duration on weekdays (β = -0.90 kg/m2/hour, se = 0.34, p = .008). Participants with larger differentials between weekday and weekend sleep duration experienced more rapid BMI gain over time for both men (β = 0.033 kg/m2/year per hour differential, se = 0.017, p = .044) and women (β = 0.057 kg/m2/year per hour differential, se = 0.027, p = .036). CONCLUSION This study suggests that habitual short sleep is associated with higher BMI levels in men and that a larger weekday-weekend sleep differential is associated with increasing BMI trajectories among both men and women in mid-to-late life.