Urban nighttime environments are rapidly shifting from traditional sodium lamps to high-intensity, blue-enriched LEDs (characterized by a higher proportion of short-wavelength light, typically in the 450-490 nm range), raising concerns about emerging forms of spectral artificial light at night (ALAN) and their metabolic consequences. Evidence on how specific light spectra and LED-dominated exposure relate to diabetes prevalence, however, remains limited. Leveraging multispectral nighttime imagery from the SDGSAT-1 Glimmer Imager, this cross-sectional study linked residential ALAN intensity, spectral composition (red, green, and blue bands), and predominant lighting types to data from 3,903 middle-aged and older adults in multiple cities. Logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for diabetes in relation to ALAN indicators after adjustment for sociodemographic, behavioral, and environmental covariates. Higher ALAN intensity was positively associated with diabetes; a ten-fold increase in panchromatic ALAN corresponded to an 18% increase in diabetes odds (OR = 1.18, 95% CI: 1.02-1.37). Among spectral bands, blue light exhibited the strongest association, with the highest quartile showing 61% higher odds of diabetes compared with the lowest (OR = 1.61, 95% CI: 1.12-2.31). Associations between ALAN and diabetes were more pronounced in areas dominated by white LEDs than in those dominated by high-pressure sodium lighting. These findings highlight the importance of considering spectral characteristics and lighting technologies in urban ALAN assessments and suggest that adopting warmer, low-blue-light LEDs in residential areas may support healthier and more sustainable nighttime environments.
