Abstract:
Risk factors for type 2 diabetes (T2D) include genetics, age, adiposity, diet, physical activity (PA) levels and environmental conditions.1–3 Putative environmental risk factors include exposure to air pollution-- specifically traffic-related air pollutants--and neighborhood conditions.4–7 A meta-analysis by Yang et al.5 of 11 studies reported an association between fine particulate matter (PM2.5) and incident T2D (HR= 1.10, 95% CI: 1.04-1.17 per 10 μg/m3 increment). Studies on NO2 have been limited and associations more mixed for incident T2D.5 There have previously been a limited number of studies on neighborhood environment, specifically neighborhood physical disinvestment (formerly referred to as disorder), and T2D. Disinvestment in the built environment is a result of systemic social and economic processes that deprive areas of resources needed to maintain and develop physical environments. Disinvestment is reflected in visual indicators of deterioration and neglect, and is a concern for health as areas of high disinvestment may reduce resident’s ability and willingness to engage in PA.8 Increasing neighborhood disorder has also been found to interact with genetic risk of T2D, as those with higher polygenic score for T2D (a summary of genome-wide genetic risk) living in neighborhoods of higher levels of disorder had a greater risk of T2D.9 Data on air pollution, neighborhood environment, and T2D risk is limited and even more so in Hispanic/Latino populations and subgroups at high risk of diabetes. This dissertation addresses those data gaps in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), a longitudinal community-based cohort study of 16,415 self-identified Hispanic/Latinos recruited between 2008 and 2011 from the Bronx, Chicago, Miami, and San Diego. Yearly follow-up calls were conducted and a second visit occurred between 2014-2017. Laboratory analysis from both exams includes fasting plasma glucose (FPG) and Glycosylated hemoglobin (HbA1c). Diabetes was defined in two way: the primary definition for diabetes required a FPG ≥126 mg/dL; HbA1c level ≥6.5%; post-OGTT glucose ≥ 200 mg/dL; and/or self-reported use of antihyperglycemic drugs,10 and the secondary definition included all those who fit the primary definition of T2D and additionally includes those who self-reported diabetes or high sugar in their blood. Additionally, different progression of diabetes were investigated and pre-diabetes was defined as having a FPG ≥100 mg/dL and <126 mg/dL; and HbA1c level ≥5.7% and <6.5%.10 The different progression of diabetes were: 1) progression from pre-diabetes to diabetes, where the sample only included those with pre-diabetes at baseline, 2) Progression of those free of diabetes and pre-diabetes to diabetes and 3) progression of those free of diabetes and pre-diabetes at baseline to pre-diabetes at visit 2.
The first aim of this dissertation focuses on estimating long-term exposure to ambient PM2.5 and NO2 in HCHS/SOL cohort members and the association of these pollutants with incident diabetes, and if these associations are modified by physical activity levels and/or adiposity. A spatiotemporal model was used to estimate air pollution exposure at residential addresses. Individual survey-weighted Poisson regression models were used to assess the association between each air pollutant’s average concentration over the follow-up period and incident diabetes at visit 2. Of the 8248 participants free of diabetes (primary definition) at visit 1, there were 950 incident diabetes cases at visit 2 with an age-centered adjusted incidence rate of 16.3 per 1000 person-years. For the secondary diabetes definition, 8129 participants were free of diabetes at visit 1 and there were 1354 incident cases by visit 2 with an age-centered adjusted incidence rate of 24.5 per 1000 person-years. When utilizing the primary definition of diabetes, a 2.48 μg/m3 increase in PM2.5 was associated with a 27% increased risk of incident diabetes when adjusting for age and sex (IRR = 1.27, 95% CI: 1.09 – 1.49). An 8.69 ppb increase in NO2 was associated with a 22% increased risk of incident diabetes when adjusting for age and sex (IRR = 1.22, 95% CI: 1.08 – 1.37). Associations with both PM2.5 and NO2 were attenuated and no longer statistically significant when adjusting for years living in the US, income, educational attainment, family history of diabetes, neighborhood SES, waist circumference, and study center/ethnic heritage. There was no evidence of effect modification by physical activity or waist circumference. When adjusting for our main set of covariates at each study center, higher exposure to PM2.5 and NO2 were only associated with increased risk of incident diabetes at the Bronx study center. Future studies should continue to investigate the association between PM2.5 and NO2 and incident diabetes in a less spatially clustered cohort with a longer follow-up period. Future studies should also continue to investigate the potential modifying effects of physical activity and adiposity.
The second aim of this dissertation focused on generating neighborhood disinvestment scores for each participants’ residential address in the HCHS/SOL and investigated the association between baseline neighborhood disinvestment and incident diabetes at visit 2. A virtual street audit of Google Street View Imagery was conducted and an item response theory model was fit to indicators (litter, graffiti, under-maintained buildings, bars on windows, and abandoned buildings) to form a scale measuring a latent level of disinvestment. Ordinary kriging was used to estimate levels for each residential address within the HCHS/SOL census tracts via spatial interpolation. Covariate-adjusted and survey-weighted Poisson regression models were used to investigate the association between neighborhood disinvestment and incident diabetes at visit 2 among 9120 participants free of diabetes at visit 1. A sensitivity analysis among those who did not move during the follow-up period was conducted as well. A one-standard deviation increase in neighborhood disinvestment score was associated with a 13% (95% CI: 1%-23%) lower risk of incident diabetes when adjusting for age, sex, education, income, study center/heritage, years in the US, family history, and a neighborhood socioeconomic index. Our sensitivity analysis yielded qualitatively similar results with lower precision. Overall, our analysis does not support the hypothesis that neighborhood physical disinvestment is associated with incident type 2 diabetes in this Hispanic/Latino population. Future studies should continue to evaluate the role of food environment, greenspace, and physical activity when considering neighborhood scale infrastructure impacts on diabetes risk.