Temporal changes in short-term associations between cardiorespiratory emergency department visits and PM2.5 in Greater Los Angeles, 2005 to 2016

Abstract

Background: Emissions control programs targeting certain air pollution sources may alter PM2.5 composition, as well as the rate/risk of health events associated with PM2.5.Objectives: We examined whether there were temporal changes in associations between short-term exposure to PM2.5 and daily emergency department (ED) visits for cardiovascular disease (CVD) and asthma in the Greater Los Angeles, California.Methods: Quasi-Poisson time-series models with unconstrained distributed lags were used to estimate associations between PM2.5 and ED visits. Bi-pollutant models were considered by including individual PM2.5 components and the rest of the PM2.5 mass. The models were run separately for three predefined time periods, selected based on the implementation of multiple emissions control programs (early: 2005-2008; middle: 2009-2012; late: 2013-2016). Analyses stratified by age (<19, 19-64, 65+) and temperature (<20°C, 20-30°C, >30°C) were also conducted.Results: The relative rate of CVD per 10 μg/m3 increase in PM2.5 concentration in the previous 4 days (lag 0-3) increased from the early period (RR = 1.003; 95% CI = [0.996, 1.010]) to the late period (1.020, [1.010, 1.030]). For asthma, estimated effects were strongest in the early period (1.018, [1.006, 1.029]) and were weaker and insignificant in the following periods. Similar temporal differences in associations for CVD and asthma were observed among different age and temperature groups. No single component was identified as an obvious contributor to the changing PM2.5 effects, but some (e.g., organic carbon, sulfate, and potassium) exhibited different temporal patterns from PM2.5.Conclusions: Significant temporal changes in the health effects of PM2.5 were observed. These changes could be due to changes in PM2.5 composition (e.g., increasing organic carbon and decreasing sulfate), although the evidence at the single-component level was not clear. Other factors, such as improvements in healthcare facilities and urban environments, exposure misclassification, and residual confounding, may also contribute to the changes.