Short sleep duration and childhood obesity: Cross-sectional analysis in Peru and patterns in four developing countries

Background:We aimed to describe the patterns of nutritional status and sleep duration in children from Ethiopia, India, Peru and Vietnam; to assess the association between short sleep duration and overweight and obesity, and if this was similar among boys and girls in Peru.

Methods and Findings: Analysis of the Young Lives Study, younger cohort, third round. In Ethiopia there were 1,999 observations, 2,011, 2,052 and 2,000 in India, Peru and Vietnam, respectively. Analyses included participants with complete data for sleep duration, BMI, sex and age; missing data: 5.9% (Ethiopia), 4.1% (India), 6.0% (Peru) and 4.5% (Vietnam). Exposure was sleep duration per day: short (,10 hours) versus regular (10-11 hours). Outcome was overweight and obesity. Multivariable analyses were conducted using a hierarchical approach to assess the effect of variables at different levels. Overweight/obesity prevalence was 0.5%/0.2% (Ethiopia), 1.3%/0.3% (India), 6.1%/2.8% (Vietnam), and 15.8%/5.4% (Peru). Only Peruvian data was considered to explore the association between short sleep duration and overweight and obesity, with 1,929 children, aged 7.960.3 years, 50.3% boys. Short and regular sleep duration was 41.6% and 55.6%, respectively. Multivariable models showed that obesity was 64% more prevalent among children with short sleep duration, an estimate that lost significance after controlling for individual- and family-related variables (PR: 1.15; 95%CI: 0.81-1.64). Gender was an effect modifier of the association between short sleep duration and overweight (p=0.030) but not obesity (p = 0.533): the prevalence ratio was greater than one across all the models for boys, yet it was less than one for girls.

Conclusions:Childhood overweight and obesity have different profiles across developing settings. In a sample of children living in resource-limited settings in Peru there is no association between short sleep duration and obesity; the crude association was slightly attenuated by children-related variables but strongly diminished by family-related variables.