Re: Ambient Air Pollution and Early Manifestation of Type 1 Diabetes.

Abstract

To the Editor Animal and epidemiologic studies suggest adverse effects of ambient air pollutants on type 1 diabetes.1–4 Increased pre- or postnatal exposure to air pollutants (ozone, particulate matter, sulfate, and nitrogen dioxide) was shown to be associated with an increased type 1 diabetes risk, and particulate matter to type 1 diabetes onset before 5 years old.2–4 However, findings regarding single air pollutants have been inconsistent. A recent study indicated that traffic-related air pollution may accelerate the manifestation of type 1 diabetes in very young children.5 We analyzed the association between air pollutants and age at onset of type 1 diabetes using data from the type 1 diabetes register for North Rhine-Westphalia (West-Germany), which is part of the EURODIAB (Europe and Diabetes) initiative.6 The completeness of registration in the age group 0–19 years was estimated to be 98% in 2002–2014. We obtained data on exposure to particulate matter with an aerodynamic diameter smaller than 10 μm (PM10), nitrogen dioxides (NO2), and accumulated ozone (O3-AOT40: excess ozone accumulated over a threshold of 40 ppb) for 2001–2005 from the German Federal Environment Agency (Umweltbundesamt II). Smoothed exposure data are estimated based on an 8 × 8 square km grid (see eAppendix; https://links.lww.com/EDE/B50). Average concentrations were calculated for each five-digit postcode area by intersection of the 8 × 8 square km grid with the German postcode map. We determined exposure of incident cases to air pollutants by linking exposure data to cases’ residential addresses at type 1 diabetes onset according to five-digit postcodes. The ethical review board of Düsseldorf University approved the study. We included 6,807 incident type 1 diabetes patients aged 0 to 19 years at diagnosis (mean age at diagnosis [SD]: 9.7 [4.5] years) in 2006–2014 in the principal analyses. To be consistent with a previous study,5 we applied simple linear and quantile regression with age at diagnosis and 10th, 30th, 50th, 70th, and 90th percentile as dependent and air pollutants as independent variables. Models were adjusted for sex, German versus non-German nationality, the German Index of Multiple Deprivation,7 family history of type 1 diabetes, level of urbanization of residence (Eurostat definition8), and additionally for patients’ body mass index at diagnosis (available for 5,625 cases). For sensitivity analyses, we performed quantile regressions at a finer grid of percentiles (5th to 95th percentile by 5%) and refitted all models including all incident cases in 2002–2014 from the North Rhine-Westphalia register (N = 9,722, mean age at diagnosis [SD]: 9.7 [4.5] years) and exposure data for 2001–2009. Results indicate that exposure to ambient air pollutants or ozone was not associated either with mean age at diagnosis or with any percentile of age at diagnosis, as evident from 95% confidence interval estimates (Figure). All sensitivity analyses confirmed these results (eFigures 1–3; https://links.lww.com/EDE/B50).FIGURE: Point estimates and 95% CIs for differences in mean and percentiles (10th, 30th, 50th [median], 70th, 90th) of the age at diagnosis of type 1 diabetes per 2 SD increases in PM10, NO2, and O3-AOT40 (accumulated ozone) adjusted for sex, German versus non-German nationality, German Index of Multiple Deprivation, family history of type 1 diabetes, level of urbanization of residence (upper panel), and additionally for body mass index at onset (lower panel). The dots represent specific quantile regression estimates and are connected by dashes to visualize trends by age quantiles, the whiskers represent 95% CI. The horizontal black lines represent the linear regression coefficients and their respective confidence intervals. The horizontal gray line depicts the age difference zero as reference. (Incident cases of 2006 to 2014, exposure data of 2001 to 2005.) CI indicates confidence interval.Our findings indicate that high exposure to traffic-related air pollutants or atmospheric ozone on scale of several kilometers are not related to the age at onset of type 1 diabetes. Thus, our study could not provide further evidence for the recently raised hypothesis that traffic-related air pollution may accelerate the manifestation of type 1 diabetes in very young children.5 Compared with the previous study,5 shortcomings of our study that may have biased our findings are the limited spatial resolution of exposure measures (8 × 8 vs. 1 × 1 square km), the coarser linkage of exposure data to patients (five-digit postcode area of residence vs. exact residential address), and the lack of individual socioeconomic data. We, however, adjusted for community-based German index of multiple deprivation as proxy for individual socioeconomic status. Strengths of our study are the high completeness of registered type 1 diabetes cases and the large study cohort. Given the limited and contradictory evidence, further studies from other regions, ideally prospective studies with improved measures of individual exposure to ambient air pollutants, are warranted to clarify the association of ambient air pollutants with accelerated manifestation and risk of type 1 diabetes. Joachim Rosenbauer Teresa Tamayo Christina Bächle Anna Stahl-Pehe Institute for Biometrics and Epidemiology German Diabetes Center Leibniz Center for Diabetes Research at Heinrich Heine University Düsseldorf Düsseldorf, Germany German Center for Diabetes Research (DZD) Partner Düsseldorf Düsseldorf, Germany [email protected] Sandra Landwehr Institute for Biometrics and Epidemiology German Diabetes Center Leibniz Center for Diabetes Research at Heinrich Heine University Düsseldorf Düsseldorf, Germany Institute of Medical Statistics Heinrich Heine University Düsseldorf Düsseldorf, Germany Dorothee Sugiri Ursula Krämer IUF-Leibniz Research Institute for Environmental Medicine Düsseldorf, Germany Werner Maier Helmholtz Zentrum München Institute of Health Economics and Health Care Management Neuherberg, Germany German Center for Diabetes Research (DZD) Partner Neuherberg Neuherberg, Germany Julia M. Hermann Reinhard W. Holl Institute for Epidemiology and Medical Biometry ZIBMT, University of Ulm Ulm, Germany German Center for Diabetes Research (DZD) Associated Partner Ulm Ulm, Germany Wolfgang Rathmann Institute for Biometrics and Epidemiology German Diabetes Center Leibniz Center for Diabetes Research at Heinrich Heine University Düsseldorf Düsseldorf, Germany German Center for Diabetes Research (DZD) Partner Düsseldorf Düsseldorf, Germany