The presence of Superfund sites as a determinant of life expectancy in the United States

Abstract

Superfund sites could affect life expectancy (LE) via increasing the likelihood of exposure to toxic chemicals. Here, we assess to what extent such presence could alter the LE independently and in the context of sociodemographic determinants. A nationwide geocoded statistical modeling at the census tract level was undertaken to estimate the magnitude of impact. Results showed a significant difference in LE among census tracts with at least one Superfund site and their neighboring tracts with no sites. The presence of a Superfund site could cause a decrease of -0.186 ± 0.027 years in LE. This adverse effect could be as high as -1.22 years in tracts with Superfund sites and high sociodemographic disadvantage. Specific characteristics of Superfund sites such as being prone to flooding and the absence of a cleanup strategy could amplify the adverse effect. Furthermore, the presence of Superfund sites amplifies the negative influence of sociodemographic factors at lower LEs.

Fig. 1. Differences between modeled and estimated Life expectancies (LEs) across the US.

a LE in the US estimated by the National Center for Health Statistics (NCHS). b Difference between LE predicted by the ordinary least squares (OLS) regression model and NCHS data. c Difference between LE predicted by the Random Forests model and NCHS data. Out of the 72,268 census tracts within the contiguous U.S., LE data were available for 65,226 tracts.

Fig. 2. Effect modification.

The effect of Superfund presence on life expectancy (LE) in tracts with at least one Superfund site with sociodemographic variables below and above the national median values. The effect modification was done by conducting separate ordinary least squares (OLS) regression analysis with a dummy variable. The effect modification analysis revealed that out of 12,717 census tracts with at least one Superfund site.

Fig. 3. Life expectancy (LE) and sociodemographic variables.

The effect of select sociodemographic variables and the presence of Superfund sites on the different percentile of LE. Only tracts with at least one Superfund site (N = 12,717) were chosen for the quantile regression analysis.

Fig. 4. Effect of Superfund sites characteristics on life expectancy (LE).

Zoomed-in box plot of LE among tracts with different Superfund site characteristics. Number of tracts were 52,509, 539, 1071, 2289, 3501, 1301, 1759 for the groups shown in the figure from left to right, respectively. Box plots indicate median (middle line), 25th, 75th percentile (box) as well as the minimum and maximum excluding outliers (whiskers). Only tracts with at least one Superfund site (N = 12,717) were chosen to explore the effect of Superfund characteristics.

Fig. 5. Superfund sites in the US.

Location of identified Superfund sites (11,989 uniquely identified sites in the contiguous U.S.) and their status with regard to the National Priorities List (NPL). The map also shows the areas prone to flooding identified by Federal Emergency Management Agency (FEMA).