Abstract
STUDY OBJECTIVES: The primary aim of the study was to estimate the effect of sleep duration on prospective type 2 diabetes (T2D) risk across demographic characteristics and follow-up periods, and test body mass index (BMI) as a mediator and moderator. METHODS: Data included adults (M(age) = 39.0 ± 12.7 years) born in the United States or Mexico recruited from 2001 to 2012 in a Mexican American cohort study conducted in Houston, TX (n = 15,779). Participants completed self-reported questionnaires at baseline related to health, health behaviors (sleep duration, physical activity, smoking, drinking), and sociocultural factors and were followed up annually. RESULTS: Cox proportional hazard models estimated hazard ratios (HR) for the effect of sleep duration on T2D diagnosis at follow-up. Of the participants, 10.3% were diagnosed with T2D. Self-reported ≤5 hours of sleep, compared to 7-8 hours, at baseline predicted greater risk for T2D (HR = 1.32, P = .001), yet was no longer significant after adjusting for sociodemographic characteristics and BMI. Notably, those with BMI <25 kg/m(2) reporting ≤5 hours of sleep were at significant risk for T2D at 3 (HR = 4.13, P = .024) and 5-year follow-up (HR = 3.73, P = .008) compared to 7-8 hours. Obesity status accounted for 31.6% and 27.3% of the variance in the association between ≤5 and 6 hours of sleep and increased T2D risk, respectively. CONCLUSIONS: Results highlighted the mediating and moderating role of BMI, and its effect on T2D risk at earlier follow-up among those without obesity. T2D prevention and control for Mexican American adults should consider the role of chronic sleep loss.