Frontiers in Endocrinology · Oct 30, 2026

Long-term exposure to PM2.5 chemical constituents and prevalent diabetes in Northwest China: a cross-sectional analysis of a Shaanxi cohort

By Xiaohui Wang, Qian Huang, Huimeng Liu et al.

BackgroundEvidence on the constituent-specific and joint associations of PM2.5 with diabetes remains limited in Northwest China.MethodsWe analyzed baseline data from 10,854 adults in the Shaanxi RECS cohort (2018–2019). Five-year PM2.5 and constituent concentrations (BC, OM, NO3-, NH4+, and SO42-) were assigned using TAP estimates. Diabetes was defined by self-reported physician diagnosis. Generalized propensity score (GPS)-adjusted logistic regression, restricted cubic splines, and quantile g-computation (QgC) assessed single-exposure and joint associations, with sensitivity analyses for lifestyle adjustment, GPS components, weight stability, and missing covariates.ResultsDiabetes was reported by 872 participants (8.03%). Per interquartile range increase, adjusted ORs were 2.21 (95% CI, 2.01–2.42) for PM2.5 and 2.82 (95% CI, 2.57–3.10) for SO42-, the largest single-exposure estimate. A simultaneous one-quartile increase in all five constituents was associated with higher odds of diabetes (OR 2.27; 95% CI, 2.04–2.52). BC accounted for 63.9% of positive QgC weights. Associations remained positive across sensitivity analyses, but the mixture OR decreased to 1.60 (95% CI, 1.49–1.71) with three GPS components.ConclusionsPM2.5 and its constituents were positively associated with self-reported prevalent diabetes, with mixture estimates sensitive to adjustment specification. These cross-sectional findings do not establish causality or rank constituent toxicity.

doi.org/10.3389/fendo.2026.2007890Open accessPopulations

AgingResearch.News