Effects of Short-term PM Mass and Component Exposures on Cardiopulmonary Mortality
Time-series studies of daily mortality and hospital admission counts with PM2.5 mass, chemical constituents, pollutant gases, and source-related mixtures have been conducted in numerous individual cities, but the most compelling evidence regarding these associations has been provided in the United States by recent nationwide multicity analyses, which are the focus of discussion here. In particular, the Health Effects Institute (HEI) sponsored National PArticle Component Toxicity (NPACT) studies at New York University (NYU) and at the University of Washington (UW) and the Lovelace Respiratory Research Institute (LRRI). The results of both of the NPACT studies, and their integration into our current understanding of the effects of PM2.5 inhalation exposures on cardiopulmonary diseases, were summarized and reviewed by Lippmann, and are included in the following sections.
NMMAPS US PM10–Mortality Time-series Studies
Dominici and colleagues analyzed daily PM10 data from 90 U.S. cities assembled for the National Morbidity, Mortality, and Air Pollution Study (NMMAPS). The data were analyzed with a generalized additive model (GAM) using the function in S-Plus and a generalized linear model (GLM) with natural cubic splines. The estimated effect of PM10 on total mortality from nonexternal causes was a 0.21% increase per 10 μg/m increase in PM10. As shown in Fig. 3, the largest and most significant PM10–mortality association was found in the northeast of the United States. Lippmann and colleagues further investigated these results, comparing the size of the PM10 mortality effect as a function of elemental composition, finding that, in the 60 NMMAPS cities with available PM2.5 elemental speciation data, the size of the PM10 mortality association was most significantly associated with average nickel (Ni) and vanadium (V), consistent with the increased risk in the northeastern United States, where more residual oil has been burned for heating and electricity than in other parts of the United States.
(Enlarge Image)
Figure 3.
Maximum likelihood estimate and 95% confidence interval (CI) of the percentage increase in total mortality from nonexternal causes per 10 μg/m increase in PM10, by U.S. region. Dominici et al (2005).
Another recent study assessed the multicity time-series association of daily PM2.5 and mortality across the United States. Using city–season specific Poisson regression, Dai and colleagues estimated PM2.5 effects on ~4.5 million deaths for all causes, cardiovascular disease (CVD), myocardial infarction (MI), stroke, and respiratory diseases in 75 U.S. cities for 2000 to 2006. They estimated a 1.18% (95% CI: 0.93, 1.44%) increase in all-cause mortality, a 1.03% (95% CI: 0.65, 1.41%) increase in CVD, a 1.22% (95% CI: 0.62, 1.82%) increase in MI, a 1.76% (95% CI: 1.01, 2.52%) increase in stroke, and a 1.71% (95% CI: 1.06, 2.35%) increase in respiratory deaths in association with a 10-μg/m increase in 2-day averaged PM2.5 concentration, confirming prior associations with daily PM10, but with larger effect sizes per 10 μg/m increase, suggesting the major effects are with the PM2.5 fraction of PM10.
NPACT Multicity Daily Mortality Analyses
The NPACT study, in addition to examining nationwide PM2.5 mass and its associations with daily mortality, also considered associations with PM2.5 chemical constituents, gaseous criteria pollutants, and source-oriented exposure indices derived via factor analyses of the PM2.5 constituents and gaseous pollutants in the 64 U.S. cities with available measurements. In the first stage of this analysis, Poisson time-series analysis regression models were first fitted for each of the 150 U.S. cities with PM2.5 mass concentrations and for the 64 of them having data on PM2.5, its constituents, and gaseous criteria pollutants, for the development of factor analysis-derived source indices. These analyses were then used to estimate percent excess risk for each pollutant at lag 0, 1, 2, and 3 days, adjusting for seasonal cycle, temporal trends, day-of-week, and immediate and delayed temperature effects. The risk estimates for each city at each lag day were then combined using a random effects model. The city-to-city variation in PM2.5 mass risk estimates across cities was modeled, in a second-stage analysis as a function of city-specific characteristics, including the average levels of PM2.5 chemical constituents, gaseous pollutants, land use, traffic density, and port berth volume (i.e., an indicator of emissions from marine vessels). The following lists describe a qualitative summary of the findings; however, refer to the full HEI report for detailed method descriptions and summary tables of results that allow an assessment of consistency of associations with the health outcomes across individual constituents, season, and lag days.
In the combined risk estimates from multiple cities, many more constituents were significantly associated with all-cause daily mortality.
Among the six source-related factor scores examined, the traffic, soil, and coal combustion factors showed significant associations with all-cause mortality in an all-year analysis.
Taking land use regressions into account, sulfate, V, the seaport berth volume within 60 miles, and the sum of road lengths were also significant predictors in explaining the spatial variation of the PM2.5 association with all-cause daily mortality.
Multicity Analyses of PM2.5 and Daily Hospitalizations
Bell and colleagues have evaluated the association between short-term exposure to PM2.5 and the risk of both cardiovascular and respiratory hospital admissions among Medicare enrollees (>64 years of age). The association varied by season and geographic region in 202 U.S. counties having populations greater than 200,000 from 1999 to 2005. Time-series models were applied to estimate consistent PM effects across the year, different PM effects by season, and smoothly varying PM effects throughout the year. A two-stage Bayesian hierarchical model was applied to access the association between PM2.5 and hospitalization rates, with the first stage estimating the association within a single county and the second stage combining county-specific estimates. Respiratory hospitalizations were highest in winter, with a 1.05% increase (95% PI: 0.29–1.82) in hospitalizations per 10 μg/m increase in same-day PM2.5. A 1.49% increase (95% CI:1.09–1.89) in CVD hospital admissions was also found for the winter season, and associations were observed in other seasons as well. The strongest association for both respiratory and cardiovascular admissions was found in the northeastern U.S. region.