Long-term trajectories of sleep duration are associated with incident diabetes in middle-to-older aged Black and White Americans

Background
Both short and long sleep have been linked to higher diabetes risk. However, sleep duration can vary over time, and there has been limited research focusing on individual sleep trajectories and diabetes risk. There are substantial racial disparities in both sleep health and diabetes risk in the US. Thus, it is important to understand the role of suboptimal sleep patterns in diabetes risk in different racial groups.
Methods
We assessed long-term trajectories of sleep duration and incident diabetes in 22,285 Black (mean age (standard deviation), 51.1 (8.2); 64.8% women) and 13,737 White adults (54.4 (9.0), 63.8%) enrolled in the Southern Community Cohort Study. Nine sleep trajectories were derived based on self-reported sleep duration at baseline and after an average of 5 years of follow up: normal-normal (reference), short-normal, normal-short, short-short, long-normal, normal-long, long-long, long-short, short-long. Diabetes was reported using a validated questionnaire. Multivariable-adjusted logistic regression determined relationships between sleep trajectories and incident diabetes.
Results
When compared to the normal-normal trajectory, suboptimal sleep trajectories were associated with higher likelihood of developing diabetes (odds ratio, (95% confidence interval), short-normal, 1.19 (1.09, 1.31), normal-short, 1.14 (1.02, 1.27), short-short, 1.17 (1.07, 1.28), long-normal, 1.13 (0.98, 1.30), normal-long, 1.16 (1.00, 1.34), long-long, 1.23 (1.02, 1.48), long-short, 1.45 (1.19, 1.77), short-long, 1.51 (1.28, 1.77)). Stratified analyses by race and socioeconomic status (i.e., education and household income) showed that most suboptimal sleep trajectories were consistently associated with incident diabetes in all sociodemographic subgroups. We also noted potential interaction with race and education for several sleep trajectories (i.e., short-long and normal-short with race, long-long and short-short with education).
Conclusions
Adults with suboptimal sleep duration trajectories are more likely to develop incident diabetes. Future research is needed to study how sociodemographic factors modulates this relationship.