Association between ambient air pollution and diabetes mellitus in Europe and North America: systematic review and meta-analysis

Abstract

Background: Air pollution is hypothesized to be a risk factor for diabetes. Epidemiological evidence is inconsistent and has not been systematically evaluated.

Objectives: We systematically reviewed epidemiological evidence on the association between air pollution and diabetes, and synthesized results of studies on type 2 diabetes mellitus (T2DM).

Methods: We systematically searched electronic literature databases (last search, 29 April 2014) for studies reporting the association between air pollution (particle concentration or traffic exposure) and diabetes (type 1, type 2, or gestational). We systematically evaluated risk of bias and role of potential confounders in all studies. We synthesized reported associations with T2DM in meta-analyses using random-effects models and conducted various sensitivity analyses.

Results: We included 13 studies (8 on T2DM, 2 on type 1, 3 on gestational diabetes), all conducted in Europe or North America. Five studies were longitudinal, 5 cross-sectional, 2 case-control, and 1 ecologic. Risk of bias, air pollution assessment, and confounder control varied across studies. Dose-response effects were not reported. Meta-analyses of 3 studies on PM2.5 (particulate matter ≤ 2.5 μm in diameter) and 4 studies on NO2 (nitrogen dioxide) showed increased risk of T2DM by 8-10% per 10-μg/m3 increase in exposure [PM2.5: 1.10 (95% CI: 1.02, 1.18); NO2: 1.08 (95% CI: 1.00, 1.17)]. Associations were stronger in females. Sensitivity analyses showed similar results.

Conclusion: Existing evidence indicates a positive association of air pollution and T2DM risk, albeit there is high risk of bias. High-quality studies assessing dose-response effects are needed. Research should be expanded to developing countries where outdoor and indoor air pollution are high.

Figure 1

Results of systematic literature search.

Figure 2

PM_2.5 and risk of T2DM. Where I² is the variation in effect estimates attributable to heterogeneity, D + L (DerSimonian and Laird) overall is the pooled random effect estimate of all studies. I-V (inverse variance) overall is the pooled fixed effects estimate of all studies. Weights are from random-effects analysis. %Weight (D + L) is the weight assigned to each study, based on the inverse of the within- and between-study variance. The size of the blue boxes around the point estimates reflects the weight assigned to each study. The summarized studies were adjusted for age, sex, BMI, smoking, alcohol consumption, and socioeconomic status.

Figure 3

NO₂ and risk of T2DM. Where I² is the variation in effect estimates attributable to heterogeneity, D + L (DerSimonian and Laird) overall is the pooled random-effects estimate of all studies. I-V (inverse variance) overall is the pooled fixed-effects estimate of all studies. Weights are from random-effects analysis. %Weight (D + L) is the weight assigned to each study, based on the inverse of the within- and between-study variance. The size of the blue boxes around the point estimates reflects the weight assigned to each study. The summarized studies were adjusted for age, sex, BMI, smoking, and socioeconomic status.