PM2.5 and Mortality: Modification of the Association by Personal and Area Level Indicators of Socioeconomic Status

Abstract

BackgroundIndividual and contextual sociodemographic characteristics are important determinants of health. We investigated the modification of the relationship between PM2.5 (particulate matter<2.5 microns) exposure and all-cause natural mortality risk, by individual sociodemographic characteristics, neighborhood census sociodemographic indicators and measures of racial and economic segregation.MethodsWe obtained geocoded 2001-2011 mortality records from the Massachusetts Department of Public Health (n=527,208), and matched them to highly spatially and temporally resolved PM2.5 predictions from satellite-based models. We examined the association between PM2.5 and natural mortality using case-crossover analysis. We defined three groups of potential modifiers: (1) individual characteristics, (2) neighborhood sociodemographic characteristics from the US Census, and (3) neighborhood measures of racial and economic segregation and dissimilarity which incorporate the individual race and the uniformity of races and economic groups within the neighborhood.ResultsEach 10µg/m3 increase in two-day average exposure to PM2.5 was associated with a 1.32% [0.94%; 1.69%] increase in mortality risk. The risk was higher among decedents without a college education and older persons (>85). The risk was modified by economic segregation and dissimilarity among younger aged decedents (<65) and effects were stronger in neighborhoods with higher percentiles of racial dissimilarity among ages ≥85 (5.93% increase, [3.03%;8.91%]). Census sociodemographic indicators did not modify the association in either of the age groups.ConclusionPM2.5 exposure is associated with increased daily deaths, and individual characteristics and sociodemographic indicators modify those risks. Neighborhood measures of segregation and dissimilarity, which incorporate individual and contextual sociodemographic characteristics, were found to be stronger modifiers than neighborhood census sociodemographic indicators.