OP38 A nationally representative study exploring the relationship between sleep and obesity in the United Kingdom -conference paper abstract-
AuthorsG.T.H. Ellison, E.M. Scott, A. Al-Naji, N. Al Afif, M. Barnes and G.R. Law
Background This study explored the potential role that sleep, and sleep improvement, might play in the aetiology of obesity, using data from the first wave of the Understanding Society (USoc) cohort – a longitudinal household panel study representative of the UK population.
Methods The first wave of USoc collected data on seven self-reported sleep-related variables comprising measures of: sleep duration; sleep quality; snoring and/or coughing while asleep; falling asleep within 30 minutes; use of sleep medication; night time waking; and staying awake during the day. Linear (for log-transformed BMI) and multinomial logistic (for BMI categorised as underweight through morbidly obese) regression analyses were conducted using STATA 12.0, before and after adjustment for age, sex, pregnancy, household income per person and household crowding index (bedrooms per person) as potential confounders. Models were specified using a Directed Acyclic Graph.
Results Complete data were available on 20,956 (41.1%) of the 50,994 cohort participants. Regardless of the sleep-related variable and whether BMI was operationalised as a continuous or multinomial variable, there was a strong, consistent and statistically significant association between sleep and BMI. Shorter, worse quality and more frequently medicated, disturbed and inadequate sleep were all associated with elevated BMI or an increased risk of overweight and obesity before and after adjustment for confounding. The sleep-related variable most strongly associated with BMI was “cough or snore loudly” where participants who reported doing so “most nights” had an adjusted odds of morbid obesity of 11.94 (95% CI 8.35, 17.07). However, even the sleep-related variable with the weakest association with BMI (“medicine for sleep”) was associated with more than twice the odds of obesity and morbid obesity amongst those participants reporting “medicine for sleep” use three or more time per week. Indeed, the overall consistency of the association between sleep and BMI was evident not only across all seven of the different sleep measures but also in the trends observed for those measures with continuous or ordinal scales.
Conclusion This is the first nationally representative study to demonstrate a powerful association between sleep duration, sleep quality and obesity in the UK. However, the study is limited by: the substantial number of participants with missing data; its cross-sectional design; and potential clustering of sleep and BMI within households. Further research is therefore required to explore whether naturally occurring changes in sleep duration and quality are associated with changes in BMI to strengthen the evidence for a causal link between sleep and obesity.