The Influence of Socioeconomic Status on the Relationship of Daily Cardiovascular Mortality with Daily Ambient PM2.5 Concentrations in Phoenix, AZ

Abstract

Background/Aim. Test the suggestion that people with high socioeconomic status (SES) are less susceptible to air pollution than those with low SES. Methods. Standard regression analysis using R. Results. People in different SES populations die on different lag days. By breaking the county into three SES populations, much higher associations are found in the SES populations than in the county. For example, betas for the strongest single lag day association for the low, medium, and high SES populations are 0.063, 0.061, and 0.046 but only 0.030 for the whole county. Since strong associations occur only on 4 or 5 lag days out of 12, and there are not enough deaths on other days to give meaningful associations, an unconstrained distributed lag model is used to estimate the total effect of PM2.5 on mortality. Total increase in % risk for an interquartile increase in PM2.5 (95% confidence interval) (range) is: low, 13.0% (5.1, 21.5) (16.5); middle 7.3% (0.2, 14.8) (14.6); high 11.6% (4.4, 19.3) (7.3); entire county 7.6 % (4.0, 11.3) (7.3). Conclusions. When risks from all lag days with meaningful associations are combined, people with low and high SES have approximately the same % increase in risk for a unit increase in PM2.5. However, it takes longer for the high SES people to die. If on a given lag day, only people in one SES population die due to PM2.5 exposure, the association will be high for that SES population lag day, but much lower for that lag day for the total population, since there will be the same number of deaths due to PM2.5, but within three times the total number of deaths. Thus, by breaking the county population into three SES populations, much higher associations are found in the low and high SES populations than in the county.