TRCLC 17-9

Assessing the Impact of Air Pollution on Public Health Along Transit Routes

PIs: Stephen P. Mattingly and Kate Hyun, University of Texas at Arlington

Summary:

Populations living, working, or going to school near major roads may be subjected to an increased risk for several adverse health effects such as respiratory diseases and conditions (e.g. asthma). Asthma affects about 25 million people in the United States including 7 million children. Due to the spatial nature of exposure, policy makers and elected officials need tools to ensure that policies and plans do not disproportionately impact particular communities and populations.

Problem:

The impact of air pollution on public health has represented a common concern for over fifty years; however, as population and economic activity continue to increase, air pollution often worsens. Exposure to air pollutants varies significantly based on a household’s location within an urban area. As a result, the local concentrations of pollutants may disproportionately impact particular communities and contribute to higher rates of morbidity and mortality in these communities. Careful strategic planning and management of transportation systems must occur to support growth and prevent worsening of health outcomes in all communities.

Research Results:

The study investigates two large metropolitan statistical areas (MSAs) in the United States (Dallas-Fort Worth (DFW) and Los Angeles). This study applies Principal Component Analysis (PCA), Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR) to investigate the impacts of the selected demographic, land use and transportation factors on the occurrence risk of respiratory diseases in two of the biggest MSAs in the US, Dallas-Fort Worth and Los Angeles, by considering respiratory hazard quotient (RHQ) as the dependent variable. As many of these variables may cause a multicollinearity problem within the model, the study applies PCA to eliminate multicollinearity and group the initial variables into fewer components, which could be used as OLS and GWR inputs.

The results of the PCA explain about 73 percent of the variation in the dependent variable in both the DFW and Los Angeles MSA using nine components. The OLS model results indicate one of the components appears insignificant for each MSA (older adults in DFW and employment density in Los Angeles), and spatial autocorrelations appear significant. As this study seeks to estimate the impacts of selected indicators locally and evaluate their effects in different locations of a MSA, a GWR addresses the spatial autocorrelations observed in the OLS. The results of GWR in both MSAs show a good fit between the final independent variables and risk of respiratory diseases, while demographic and transit access to job represent the most significant variables. The GWR results show an overall positive effect of all variables on the independent variable with a median R² value of 0.83, compared to 0.48 from OLS in DFW and 0.79 (GWR) and 0.48 (OLS) in Los Angeles.

For both MSAs, demographic characteristics represent the first principal component (PC) and transit access to jobs represents the second PC. For DFW, automobile access is the third PC while it is the fifth PC for LA. Both MSAs also have components representing older adults and miles driven. Transportation plays an important role in multiple principal components.

Results:

While demographic characteristics appear the most important determinant of aggregate respiratory disease risk in both MSAs, transit access to jobs represents the second most important component. This indicates that after controlling for demographic effects, higher transit access to jobs clearly indicates a greater risk of respiratory disease, which directly confirms the research question and hypothesis. Those living along transit corridors and likely in transit oriented developments face a greater risk of respiratory disease. While other components experience greater spatial variations in both MSAs, the transit access to jobs displays a clear pattern and significance.  While the specific variables in the components vary slightly between the DFW and Los Angeles MSAs, the components largely measure the same effects as can be noted in their descriptions. The importance of similar effects in both MSAs indicates that large MSAs may experience similar impacts related to transit access to jobs, automobile access, and vehicle miles traveled.

Presentation

Final Report