Chemical Composition of Fine Particulate Matter and Life Expectancy: In 95 US Counties Between 2002 and 2007

Abstract

Background: In a previous study, we provided evidence that a decline in fine particulate matter (PM2.5) air pollution during the period between 2000 and 2007 was associated with increased life expectancy in 545 counties in the United States. In this article, we investigated which chemical constituents of PM2.5 were the main drivers of the observed association.

Methods: We estimated associations between temporal changes in seven major components of PM2.5 (ammonium, sulfate, nitrate, elemental carbon matter, organic carbon matter, sodium, and silicon) and temporal changes in life expectancy in 95 counties between 2002 and 2007. We included US counties that had adequate chemical components of PM2.5 mass data across all seasons. We fitted single pollutant and multiple pollutant linear models, controlling for available socioeconomic, demographic, and smoking variables and stratifying by urban and nonurban counties.

Results: In multiple pollutant models, we found that: (1) a reduction in sulfate was associated with an increase in life expectancy; and (2) reductions in ammonium and sodium ion were associated with increases in life expectancy in nonurban counties only.

Conclusions: Our findings suggest that recent reductions in long-term exposure to sulfate, ammonium, and sodium ion between 2002 and 2007 are associated with improved public health.

FIGURE 1

Changes in life expectancy in 95 US counties (2007 vs. 2002).

FIGURE 2

County-specific yearly average of PM_2.5 total mass and its seven components in 2002 versus 2007.

FIGURE 2

County-specific yearly average of PM_2.5 total mass and its seven components in 2002 versus 2007.

FIGURE 3

A, Point estimates and 95% confidence intervals of the association across counties between a decrease (between 2002 and 2007) in an IQR (μg/m³) of the PM_2.5 total mass or in each of the seven chemical components and an increase (between 2007 and 2002) in life expectancy (years). Results are reported for all the 95 counties (left panel), 43 nonurban counties (middle panel), and 52 urban counties (right panel). These estimates were obtained under an “unadjusted single pollutant model” that includes a single exposure predictor of the change in PM_2.5 total mass or the change in one of the seven components. B, Point estimates and 95% confidence intervals of the association across counties between a decrease (between 2002 and 2007) in an IQR μg/m³) of the PM_2.5 total mass or in each of the seven chemical components and an increase (between 2007 and 2002) in life expectancy (years). Results are reported for all the 95 counties (left panel), 43 nonurban counties (middle panel), and 52 urban counties (right panel). These estimates were obtained under an “adjusted single pollutant model” that includes a single exposure predictor (change in PM_2.5 total mass or the change in one of the seven components) and the additional confounders that represent changes in demographics, socioeconomic, population, lung cancer, and COPD variables (surrogate information on smoking population). C, Point estimates and 95% confidence intervals of the association across counties between a decrease (between 2002 and 2007) in an IQR μg/m³) of the PM_2.5 total mass or in each of the seven chemical components and an increase (between 2007 and 2002) in life expectancy (years). Results are reported for all the 95 counties (left panel), 43 nonurban counties (middle panel), and 52 urban counties (right panel). These estimates were obtained under an “adjusted multiple pollutant model” that includes changes in all the seven components as exposure predictors and the additional confounders that represent changes in demographics, socioeconomic, population, lung cancer, and COPD variables (surrogate information on smoking population).