The Association between Multiple Environmental Exposures and Symptoms of Generalized Anxiety Disorder in a Prospective, US-Based Cohort Study

Abstract

Background: Emerging evidence has implicated built and natural environmental exposures in the etiology and exacerbation of anxiety symptoms.

Objectives: Our objective was to assess individual and joint associations between neighborhood socioeconomic status (nSES), walkability, greenness, fine particulate matter (${\mathrm{PM}}_{2.5}$), nitrogen dioxide (${\mathrm{NO}}_2$), and temperature and generalized anxiety disorder (GAD) symptoms.

Methods: We included 81,897 participants of the Nurses' Health Study II. GAD symptoms were measured in 2013 and 2017 using the seven-item Generalized Anxiety Disorder scale (GAD-7). Walkability and nSES z-scores were calculated at the census tract level. Summer and annual average greenness were measured using Normalized Difference Vegetation Index (NDVI) at 270 m and 1,230 m buffer sizes. 1-, 3-, and 12-month averages of particulate matter with an aerodynamic diameter of $<2.5\text{micrograms}$ (${\mathrm{PM}}_{2.5}$), nitrogen dioxide (${\mathrm{NO}}_2$), and temperature were predicted using spatiotemporal models. Exposures were measured at the residential level. Longitudinal logistic and generalized estimating equations (GEE) models estimated odds ratios (OR) and 95% confidence intervals (CIs) for clinically relevant GAD symptoms (GAD-7 $\ge 5$ points). We assessed effect modification by nSES, summer NDVI ($\mathrm{1,230}m$), and select covariates. We employed quantile g-computation to assess the association between environmental exposure mixtures and GAD symptoms more directly.

Results: In single-exposure models, summer NDVI (e.g., for $\mathrm{1,230}m$, $\text{OR}=0.96$; 95% CI: 0.94, 0.98), ${\mathrm{NO}}_2$ (e.g., for 12-month average ${\mathrm{NO}}_2$, $\text{OR}=1.03$; 95% CI: 1.01, 1.04), 12-month average temperature ($\text{OR}=1.05$; 95% CI: 1.03, 1.07), and walkability ($\text{OR}=1.02$; 95% CI: 1.01, 1.03) were associated with GAD symptoms. In the joint-exposure model, nSES ($\text{OR}=0.96$; 95% CI: 0.95, 0.98) and summer NDVI were associated ($\text{OR}=0.92$; 95% CI: 0.88, 0.96); all other associations were attenuated. Associations with summer NDVI ($\mathrm{1,230}m$), temperature, and ${\mathrm{NO}}_2$ were modified by nSES, with stronger associations in socioeconomically deprived areas.

Discussion: In this population, higher levels of summer greenness and nSES may be protectively associated with GAD symptoms, whereas medium- and long-term ${\mathrm{NO}}_2$ exposure, long-term temperature, and walkability are adversely associated with GAD symptoms. Socioeconomically deprived areas may bear additional risks. The results of the quantile g-computation analysis suggest that environmental exposures may act antagonistically with one another in relation to GAD symptoms. https://doi.org/10.1289/EHP14458.

Figure 1.

ORs and 95% CIs for clinically relevant symptoms of at least mild GAD (GAD-7 $\ge 5$; modeled GAD-7 score dichotomized at a score of 5, indicating clinically relevant symptoms of mild GAD associated with each IQR increase in ${\mathrm{PM}}_{2.5}$ exposure, ${\mathrm{NO}}_2$ exposure, annual and summer NDVI, and walkability stratified by neighborhood nSES and summer $\mathrm{1,230}\text{-m}$ NDVI using single-exposure GEE models for 81,897 participants with complete exposure data and outcome data in 2013 and/or 2017 in Nurses’ Health Study II. (Left) Stratification by quintiles of neighborhood socioeconomic status. (Right) Stratification by quintiles of NDVI at $\mathrm{1,230}\text{-m}$ buffer size. Effect estimates are presented in Tables S3 and S4. Note: CI, confidence interval; GAD, generalized anxiety disorder; GAD-7, seven-item Generalized Anxiety Disorder scale; GEE, generalized estimating equations; IQR, interquartile range; nSES, neighborhood socioeconomic status; NDVI, Normalized Difference Vegetation Index; OR, odds ratio; ${\mathrm{PM}}_{2.5}$, particulate matter with aerodynamic diameter $\le 2.5\mathrm{\mu }\mathrm{m}$.

Figure 2.

Scaled weights from quantile g-computation analysis of a mixture of 15 environmental exposures and their associations with clinically relevant symptoms of at least mild GAD (GAD-7 $\ge 5$; modeled GAD-7 score dichotomized at a score of 5, indicating clinically relevant symptoms of mild GAD) across 81,897 participants with complete exposure data and outcome data in 2013 and/or 2017 in Nurses’ Health Study II. Quantile g-computation analysis was adjusted for age at baseline, time, AHEI (diet) score (including alcohol consumption), smoking status (past/current/never), marital status (married/unmarried), living arrangement (with others/alone), race (White/Non-White), ethnicity (Hispanic/non-Hispanic), husband’s education status (high school or below/past high school), and nSES z-score. Numerical data available in Table S7. Note: AHEI, Alternative Healthy Eating Index; GAD, generalized anxiety disorder; GAD-7, seven-item Generalized Anxiety Disorder scale; NDVI, Normalized Difference Vegetation Index; ${\mathrm{PM}}_{2.5}$, particulate matter with aerodynamic diameter $\le 2.5\mathrm{\mu }\mathrm{m}$; nSES, neighborhood socioeconomic status.