Association between long-term exposure to air pollution and urban green and incidence of type2 diabetes

Abstract

Introduction: Several studies investigated the association between environmental exposures and incidence of type2 diabetes with conflicting results. Our aim was to evaluate the long-term effect of particulate matter (PM), nitrogen oxides (NOx), and residency close to green areas on the incidence of type2 diabetes in a large metropolitan cohort. Methods: We recruited 1,191,589 subjects aged 35+ years diabetes-free at baseline (January 2008), and followed them up for 6 years. We estimated PM10, PM2.5, NO2 and NOx exposures at residence using land use regression models. We defined urban green exposure as km2 of greenness in a 1-km buffer around the address. We used Cox proportional-hazard regression models adjusted for individual and contextual characteristics, to estimate the association between environmental exposures and incidence of diabetes. Moreover, we evaluated prevalent and incident myocardial infarction, use of statin and antihypertensive drugs as potential effect modifiers. Results: We identified 66,986 incidence cases during follow-up. Average exposures were: 19.6 (standard deviation=1.9) ug/m3 for PM2.5, 36.6 (5.2) ug/m3 for PM10, 84.3 (24.4) ug/m3 for NOx, 42.5 (10.4) ug/m3 for NO2, and 0.17 (0.3) km2 for Urban Green. We found weak evidence of an association between air pollutants and incidence of diabetes, the only exception being NOx exposure, with a hazard ratio (HR) of 1.011 (95%CI: 1.005-1.018) per 20ug/m3 increases. Proximity to green spaces displayed a protective effect on the incidence of diabetes, even after adjustment for concurrent NOx exposure (HR=0.96, 95%CI: 0.94-0.98 comparing >75th percentile VS null exposure). Pre-existent medical factors did not modify the associations. Conclusions: There was weak evidence of an association between long-term exposure to air pollution and incidence of type2 diabetes. Greenness significantly reduced the risk of disease incidence and pre-existent medical factors did not alter the associations under study.