While evidence links light at night (LAN) to metabolic disturbances, large-scale, population-based studies of objectively-measured LAN exposure remain limited. This study investigates the association between LAN and the risk of diabetes mellitus (DM) among US adults using data from the National Health and Nutrition Examination Survey. Cross-sectional data from the 2011 to 2014 National Health and Nutrition Examination Survey cycles were analyzed, focusing on participants with complete data on both LAN and DM. Participants were stratified by LAN level. Multivariable logistic regression analyses were employed to examine the association between LAN and DM. Multivariable linear regression analyses assessed relationships between LAN and diabetes risk markers, including hemoglobin A1c (HbA1c), fasting blood glucose, fasting serum insulin, and Homeostasis Model Assessment of Insulin Resistance (HOMA-IR). Subgroup and sensitivity analyses with multiple imputation tested robustness. A total of 4498 participants were enrolled. The prevalence of diabetes was higher in the high-LAN exposure group compared to the no-LAN group (21.01% vs 14.48%). Consistent with this, in the fully adjusted model, high-LAN exposure was independently associated with a significantly increased risk of DM (odds ratio [OR] = 1.39, 95% confidence interval [CI]: 1.00-1.92). High-LAN exposure was linked to elevated HbA1c and fasting blood glucose in initial models. However, these associations weakened and became nonsignificant after full adjustment for covariates. No significant associations were observed for fasting serum insulin or HOMA-IR. Subgroup analyses showed the association between LAN and DM was more pronounced in middle-aged adults (40-64 years), individuals with obesity, and those with dyslipidemia or hypertension. However, no significant effect modification was found across subgroups (all P for interaction > .05). Sensitivity analyses confirmed the robustness of these findings. This study uses objective actigraphy-based LAN measurement and shows its independent association with DM, especially in metabolically susceptible subgroups. These findings identify LAN as a modifiable environmental risk factor and support its relevance to DM prevention and public health strategy.
