Associations of acute and chronic adverse health effects and ambient particulate matter mass concentration have been widely reported.1 In particular, fine particles (particles ≤2.5 μm in aerodynamic diameter, PM2.5) have been linked extensively with mortality and morbidity.2 PM2.5 is a heterogeneous mixture of solid and liquid particles emitted from a variety of sources, and its composition varies spatially and temporally. Different chemical constituents within PM2.5 will likely have varying effects on specific health end points. While epidemiologic studies provide consistent evidence of adverse health effects from PM2.5 mass acute and chronic exposure, there is little information for specific chemical species and sources. Characterizing the toxicity of PM2.5 chemical species derived from various sources may be an effective tool to protect public health through more targeted emission control strategies and regulations.
Several epidemiologic studies have investigated the association of specific PM2.5 chemical constituents with mortality and morbidity,3–9 as well as the association of source-apportioned PM2.5 mass and health outcomes.10–12 However, there are no studies investigating the risk impact of PM2.5 mass sources related to health outcomes in Asia—where PM sources, chemistry, size distribution, and temporal patterns of exposure may differ from those elsewhere.
Increases in Asian urban populations due to rapid economic growth may lead to larger potential health problem related to PM air pollutants. Mitigating the effects of PM air pollutants on human health is a major challenge in this region. Seoul, the capital of South Korea, had a population of 9.8 million at the end of 2007 and 3.3 million households, representing 21% of the population of South Korea. The city’s land area is only 0.6% of South Korea’s total area, indicating high population density. PM2.5 levels remain high in Seoul relative to similar cities in developed countries, and secondary sulfate and nitrate sources are major contributors to the PM2.5.13 In addition, extreme PM2.5 exposure occurs in the region during yellow sand and smog events.14 Efforts to elucidate the effects of specific sources of PM2.5 on human health could offer new strategies to manage emission sources more effectively. We applied a time series analysis to identify the chemical species and sources of PM2.5 most strongly associated with mortality outcomes in Seoul.
Seoul daily mortality data were obtained from the Korea National Statistical Office, which coincided with the PM2.5 sampling period of March 2003 through November 2007. Using the International Classification of Disease, 10th Revision (ICD-10; World Health Organization 1993), the mortality data were classified as all nonaccidental causes (codes A00-R99), cardiovascular disease (codes I00-I99), respiratory disease (codes J00-J98), and injury (S00-T98).
Pollutant and Meteorological Data
To estimate the effects of PM2.5 chemical species and sources on mortality, ambient air samples were collected over a 24-hour period at 3-day intervals from March 2003 to November 2007. The measurement site was on the roof of the former School of Public Health (~17 m above ground, 37.5°N, 127.00°E) at the Seoul National University. Ambient air sampling methods for this study are described by Heo et al13 and Kim et al.14 Filter samples were simultaneously collected using a four-channel system. Two channels were equipped with an Annular Denuder System and the remaining two with filter packs. PM2.5 samples were analyzed for gravimetric concentration, water-soluble ionic species, carbonaceous species (ie, organic carbon and elemental carbon), and trace elements. Ionic species were analyzed by an ion chromatograph (Dionex DX-120; Dionex, Thermo Fisher Scientific, Inc., Cambridge, UK), and trace elements were analyzed using proton-induced X-ray emission. Carbonaceous species were determined by thermal and optical transmittance (Sunset Laboratories, Tigard, OR). To adjust for weather effects on mortality, daily average temperature and humidity data were collected at a meteorological station in Seoul (Korea Meteorological Admission: www.kma.go.kr).
PM2.5 mass and major chemical species have been found to be homogeneous in Seoul, and thus, we assume that the determined pollutants for this study are representative of all the included population (eTable 1 and eFigures 1–2, http://links.lww.com/EDE/A755).
PM2.5 source apportionment was determined using the positive matrix factorization receptor model. This is an advanced factor analysis technique based on a weighted least-squares fit and error estimates of the observed data.15 This model has been widely used in air pollution source apportionment studies.10,11,13,16 Previously published work describes identified source profiles and contributions of PM2.5 in Seoul, based on 2003–2006 data.13 For this study, extended PM2.5 speciation data sets including the 2007 calendar year were used in the positive matrix factorization model, using the procedure described by Heo et al.13
Time Series Method
Time series analyses using a generalized linear model with natural splines were used to examine the associations of PM2.5 chemical constituents and their sources with daily counts of cause-specific deaths. In the basic regression model, long-term time trends of temperature and humidity were controlled using a smoothing function with 6 degrees of freedom (df)—3df and 3df, respectively. Dummy variables for the day of the week, holiday effects, and influenza epidemics were also included. Six degrees of freedom were selected for smoothing the time trends based on the fitted mortality series and the extent of autocorrelation of the residuals. To examine the impact of alternative degrees of freedom for the smooth function of time, a sensitivity analysis using 3 to 9 degrees of freedom was applied to the time trends versus the 6 degrees of freedom in the basic model. To examine the effect of model specification of lag structure on weather covariates, we used 1-day lag values for temperature and humidity, as opposed to the same day values used in the basic model. For the chemical species concentration exposures and source apportionments, multiday exposure averages could not be constructed because the PM2.5 speciation data were collected at 3-day intervals; therefore, single-day lags of 0–3 days were examined separately.
In single-pollutant models, each PM2.5 chemical constituent was included in the regression model individually, without controlling for other chemical species. In multipollutant models, multiple PM2.5 chemical species were simultaneously included to compute the regression coefficients for each constituent while controlling the levels of other components. Because ammonium is commonly observed as condensed compounds such as ammonium sulfate ((NH4)2SO4), ammonium bisulfate (NH4HSO4), and ammonium nitrate (NH4NO3), PM2.5 ammonium particles are strongly correlated with sulfate and nitrate particles. Thus, ammonium was excluded from multipollutant models when sulfate and nitrate were included, and sulfate and nitrate were excluded in a separate multipollutant model when the ammonium component was included. Aluminum, silica, calcium, iron, magnesium, and titanium in PM2.5 are classified as crustal minerals, which are resuspended in the air from natural soil, paved and unpaved road dust; they are strongly intercorrelated. For this reason, each crustal metal was separately incorporated into the multipollutant models. To estimate the PM2.5 mass risk, the daily portions or concentrations of other chemical components were not adjusted because the source-apportioned PM2.5 mass was considered in the regression models.
All analyses were performed using the R software.17 The results are presented as the percentage of excess risk (ER = (Exp(βˆ × IQR)−1) × 100) in daily mortality associated with an interquartile range (IQR) increase in each PM2.5 chemical species and source. In the eAppendix (http://links.lww.com/EDE/A755), the full set of results, including the excess risk (95% confidence intervals) in mortality per IQR for each chemical species and source, are reported.
In total, 157,119 deaths were registered in the study population from March 2003 through November 2007 (1706 days). The daily mean estimate of all-cause deaths was 92.1, including 25.3 from cardiovascular diseases and 5.1 from respiratory diseases (Table 1).
The annual average PM2.5 concentration was 43.4 μg/m3 during the study period, which is much higher than the United State National Ambient Air Quality Standard annual PM2.5 standard and reported PM2.5 levels in other cities in the United States18 and Europe,19 which varied from 14.0 to 22.8 μg/m3. It is much lower than the PM2.5 levels in major cities in China and Taiwan,14,20 which ranged from 55.0 to 182.2 μg/m3, but higher than the 23.0 μg/m3 measurement reported from Tokyo, Japan.21 The PM2.5 mass fractions of organic carbon, sulfate, nitrate, ammonium, elemental carbon, crustal elements, and noncrustal elements accounted for 23%, 20%, 17%, 13%, 8%, 4%, and 3% of the total mass, respectively (Table 1). The concentration levels of the major chemical species of PM2.5 in Seoul were one-half to one-quarter of those found in Xi’an, Shanghai, Kaohsiung, and Beijing, China,14,20 but about two times higher than those in Tokyo.21 There was a strong correlation between PM2.5 mass and its major components, such as organic carbon, sulfate, nitrate, and ammonium, as well as some trace elements, including potassium, manganese, iron, copper, zinc, and lead (eTable 1, http://links.lww.com/EDE/A755). Among individual metals, strong intercorrelations were observed in the crustal metals group (magnesium, aluminum, silica, calcium, iron, and titanium) and industrial metals group (manganese, copper, zinc, and lead).
Nine sources of PM2.5 were identified in this study region (Table 1). These were secondary nitrate, secondary sulfate, gasoline emission, diesel emission, biomass burning, soil, roadway emission, aged sea salt, and industry. Each resolved source had the same source profile characteristics and source contribution as seen in the study by Heo et al.13 The average source contributions to the total PM2.5 mass were estimated to be 19% by secondary nitrate, 19% by secondary sulfate, 14% by gasoline emission, 12% by diesel emission, 10% by biomass burning, 8% by soil, 5% by roadway emission, 2% by aged sea salt, and 12% by industry. Secondary nitrate was identified by high loadings of nitrate and ammonium concentrations, with distinctive seasonal patterns of high spring and winter peaks, whereas secondary sulfate was characterized by high concentrations of sulfate and ammonium with high summer peaks (Figure 1). The biomass burning source was represented by potassium and organic carbon. Two mobile emission sources (ie, gasoline and diesel) were dominated by a high loading of elemental carbon and organic carbon, with higher elemental carbon from diesel emissions versus higher organic carbon from gasoline emissions. The soil source was dominated by aluminum, silica, iron, calcium, and titanium and displayed event peaks in spring, coinciding with the yellow sand events in the region. The roadway emission was characterized by high concentrations of chloride, as well as zinc, bromine, and lead, and was most apparent during winter seasons. Aged sea salt was identified by its high mass fraction of sodium, magnesium, potassium, and organic carbon. Finally, industry source was characterized by anthropogenic elements, including nickel, copper, zinc, manganese, and bromine. A more detailed description of source characteristics can be found in the report by Heo et al.13Table 2 shows the Pearson correlation coefficients among the estimated source contributions. There were modest correlations between sources, with soil and aged sea salt being the strongest (r = 0.50).
Of the single-pollutant models, all-cause deaths were weakly associated with Na. Cardiovascular mortality was strongly associated with PM2.5 mass, ammonium, copper, and zinc and moderately associated with nickel and lead. Respiratory mortality was strongly associated with sodium and bromine and had weaker associations with elemental carbon, aluminum, silica, potassium, iron, and lead. From the multipollutant models, we found that the risk impacts of organic carbon on all-cause and cardiovascular mortality, elemental carbon on respiratory mortality, magnesium on cardiovascular mortality, and lead on cardiovascular and respiratory mortality were robust to adjustments by other chemical species.
Figures 2 and 3 represent the percentage of excess risk in daily death counts for all-cause, cardiovascular, and respiratory mortality per IQR increase in selected chemical species obtained from single-pollutant and multipollutant models during the study period. In single-pollutant models, IQR increases in PM2.5 mass, copper, and zinc at lag 0 (same day exposure) were associated with increases in cardiovascular mortality of 2.8% (0.2 to 5.4), 2.3% (0.01 to 4.6), and 2.0% (0.0 to 4.0), respectively. For a 2-day lag, respiratory mortality increased by 4.0% (0.0 to 8.1), with an IQR increase in bromine. Most risk estimates for individual chemical components in multipollutant models were lower than in the single-pollutant models, but risk estimates for organic carbon, elemental carbon, and lead remained elevated. An IQR increase in organic carbon at a 1-day lag was associated with 2.6% (0.1 to 5.1) and 6.9% (2.1 to 11.9) increases in all-cause and cardiovascular mortality, respectively. An IQR increase in elemental carbon at a 2-day lag was associated with a 9.5% (0.8 to 19.0) increase in respiratory mortality. An IQR increase in lead was associated with 9.4% (4.7 to 14.3) increase in the risk of cardiovascular mortality on the 2-day lag and a 10.1% (0.1 to 21.1) increase in respiratory mortality on the 3-day lag.
Figure 4 shows the quantitative risk effects of resolved PM2.5 sources on all-cause, cardiovascular, and respiratory mortality using single-day lags. There were no strong associations between specific sources and all-cause mortality. Cardiovascular mortality was strongly associated with the biomass burning source and had a moderate association with roadway emission and industry sources. Respiratory mortality was substantially associated with gasoline and diesel emission sources and moderately associated with the soil source. Regarding the percentage excess risk effects per IQR increases in the resolved sources, cardiovascular mortality increased by 1.9% (0.0 to 3.7) per IQR increase in the biomass burning and respiratory mortality increased by 5.5% (0.5 to 10.7) and 6.7% (0.2 to 13.7) per IQR increase in gasoline and diesel emissions, respectively.
Sensitivity analysis indicated that the results were insensitive to alternative degrees of freedom (ranging from 3 to 9 per year) used for the smooth function of time and to different lags for the weather covariates in the model specifications. An additional analysis was conducted for an internal check by using injury-related mortality. As expected, there were no substantial associations between exposures to PM2.5 chemical constituents and sources and deaths due to injury (eFigures 3 and 4, http://links.lww.com/EDE/A755).
We carried out a time series analysis to examine the mortality risks of exposure to PM2.5 mass and its chemical species as well as its sources. Strong and consistent associations were found between PM2.5 mass and several chemical constituents, including organic carbon, ammonium, sodium, copper, zinc, bromine, and lead, and mortality outcomes. Ambient levels of organic carbon, elemental carbon, and lead were associated with a higher risk of mortality in the multipollutant models. Moreover, there were strong associations between the source-apportioned PM2.5 mass, derived from mobile emissions (ie, gasoline and diesel emissions) and biomass burning, and cardiovascular and respiratory mortality. This suggests that the most important contributors of PM2.5 leading to adverse health effects are from local combustion sources. These results may assist in the development of effective PM2.5 management strategies to improve control programs intended to reduce the public health burden from ambient PM2.5 in the region.
The ambient levels of PM2.5 mass in Seoul were associated most strongly with cardiovascular mortality but had modest negative associations with other mortality outcomes. The magnitude of the excess risk of PM2.5 mass in this study is different from the estimates provided by Cho et al,22 which showed elevated risks of 0.9% (0.8 to 0.9) for all-cause mortality, 1.9% (1.8 to 2.1) for cardiovascular mortality, and 6.6% (6.4 to 6.8) for respiratory mortality per IQR (29.4 μg/m3) for same day PM2.5 exposures using daily PM2.5 concentrations for a year-long measurement. The difference between these two study results may be due to different sampling durations. In a meta-analysis of the association between PM2.5 species data and daily mortality outcomes from 2000 to 2003 in six California counties,7 PM2.5 mass obtained in 3- or 6-day intervals was associated only with daily cardiovascular deaths, whereas daily PM2.5 mass was strongly associated with most causes of mortality and provided enhanced statistical power. Although daily PM2.5 measurement data were not available during this study period, the relationships between PM2.5 mass and mortality outcomes may be enhanced in future studies, by increasing sample size based on daily observations.
Several studies have investigated the toxicity of PM2.5 chemical species and sources on local and regional scales in developed countries, but the empirical findings on relationships between specific constituents and sources are mixed.3,5–9,12,23,24 These differences may be due to the diversity of the study locations, health outcomes, and quantitative analysis methods.
In this study, organic carbon was strongly associated with all-cause and cardiovascular mortality in multipollutant models. Ostro et al7 looked at the association between PM2.5 chemical components and daily mortality using time series analysis in California and found that for an IQR increase (4.6 μg/m3) for organic carbon, with a 3-day lag, cardiovascular mortality increased by 1.6% (–0.1 to 3.2). Peng et al9 observed that an IQR (3.18 μg/m3) increase in organic carbon was associated with 0.6% (0.01 to 1.2) increase in cardiovascular admissions for a 1-day lag and with a 1.1% (0.1 to 2.0) increase in respiratory admissions for a 2-day lag, in 119 US counties. Mar et al25 and Metzger et al26 identified associations of organic carbon with cardiovascular mortality in Phoenix, Arizona, and with cardiovascular emergency department visits in Atlanta, Georgia, respectively.
Organic carbon is typically emitted from both anthropogenic and biogenic primary sources, as well as by atmospheric chemical reactions of primary organic compounds.16 Key sources of organic carbon were identified from the positive matrix factorization–resolved source profiles in Seoul as combustion sources, including gasoline, diesel, roadway emission, and biomass burning, which were strongly associated with mortality outcomes. Sarnat et al12 sought to relate specific sources of PM2.5 to cardiorespiratory morbidity using chemically speciation in Atlanta, Georgia, and found that emergency department visits for cardiovascular disease were strongly associated with organic carbon-driven sources (ie, gasoline, diesel, and wood smoke). Furthermore, in a long-term exposure study from the California Teachers Study, Ostro et al23 found associations of organic carbon with all-cause, cardiopulmonary, ischemic heart disease, and pulmonary mortality. The potential biological mechanisms linking organic carbon to mortality may be oxidative stress and effects on blood pressure.27
Elemental carbon exposure was found to be substantially associated with increased risk of respiratory mortality in this study. This finding is somewhat different from several other single-city or regional studies in which elemental carbon exposure was associated with cardiovascular disease outcomes. For example, Ostro et al7 observed that cardiovascular mortality increased by 2.1% (0.3 to 3.9) for elemental carbon for an IQR (0.8 μg/m3) increase on a 3-day lag, but they did not find associations with respiratory mortality. Peng et al9 reported that an IQR (0.4 μg/m3) increase in elemental carbon at lag 0 was associated with a 0.8% (0.34 to 1.27) increase in the risk of cardiovascular admissions in a regional study of 119 US counties. Mar et al25 also found positive associations between elemental carbon and cardiovascular mortality. These findings are in agreement with source-related analyses in which exposure to elemental carbon-derived sources, most notably diesel exhaust, was associated with increased risk of cardiovascular morbidity12 and mortality.25 However, the present study’s association between elemental carbon and respiratory mortality is similar to a study at Santiago, Chile,28 which found that elevated elemental carbon levels derived from traffic sources led to increased risk of respiratory deaths. Other studies have also reported considerable evidence linking elemental carbon to risks of respiratory disease outcomes.7,29 Furthermore, traffic-related sources characterized by high elemental carbon concentrations were associated with increased risks of respiratory admissions for children and adults with asthma.30,31
Elemental carbon is emitted primarily from combustion processes and not by atmospheric chemical reactions. In this study, elemental carbon is associated with diesel and gasoline emissions. Possible mechanisms by which ambient PM may affect cardiovascular and respiratory health have been proposed,32,33 but biological mechanisms explaining how particles with high elemental carbon and organic carbon levels could cause health risks have not yet been identified. There has been growing evidence that diesel emission particles cause pulmonary system inflammation and cell damage, including alveolar macrophage death and oxidative stress capable of influencing pulmonary physiology.31,34,35 These mechanistic results confirm our study’s findings that exposure to elemental carbon, as well as to diesel- and gasoline-related PM2.5 mass, is strongly associated with increased risks of respiratory mortality.
We saw associations of lead with cardiovascular and respiratory mortality, with higher excess risks when applying multipollutant models. PM2.5 lead in Seoul is emitted predominantly from combustion sources (ie, biomass burning and two-stroke engines). Laden et al6 found that motor vehicle sources represented by lead were associated with the strongest risk effects for mortality. Hong et al36 reported a decrease in the peak expiratory flow rate in school children with lead levels in ambient PM. Bae et al37 found a positive association of lead with the oxidative stress biomarker malondialdehyde in schoolchildren. A mechanism that might contribute to greater lead toxicity is the generation of oxidative stress.38 Saxena and Flora39 also suggested that exposure to lead alters the status of reactive oxygen species or oxidative stress, including inflammatory reactions. Moreover, Bagci et al40 found that blood lead levels in workers exposed to batteries and exhaust fumes were associated with impaired pulmonary function.
Associations of cardiovascular mortality with ammonium, copper, and zinc, as well as of respiratory mortality and sodium and bromine, were notable in single-pollutant models, although these associations had reduced statistical power when we controlled for other components. These constituents are key species for characterizing industry and biomass combustion sources, as well as roadway emission. Exposure to these sources was associated with increased risk of mortality in this study. As discussed above, oxidative stress has been reported as an important mechanism inducing toxic effects of these chemical constituents. Several human and animal studies observed associations of PM metal components with pulmonary toxicity and the generation of oxidative stress.41,42 Lagorio et al43 reported that ambient zinc in particulate matter was associated with indexes of health-related responses in a panel study of patients with chronic obstructive pulmonary disease. Burnett et al4 found associations between mortality with daily exposure to particulate matter trace metals, including zinc, nickel, and iron. Ostro et al8 also found associations of hospital admissions with several metals, including copper, iron, and zinc.
The goal of the present study was to determine which chemical constituents and sources of PM2.5 have primary responsibility for adverse health effects. However, ambient PM2.5 is produced by numerous emission sources and photochemical reactions and therefore comprises various chemical components. Furthermore, many PM2.5 chemical constituents are highly correlated with each other and are likely undetermined by current analytical methods. Thus, it has been challenging to characterize the most harmful biological mechanisms on health effects corresponding to individual chemical species of PM2.5 through the epidemiological and toxicological findings. For these reasons, we conducted this analysis with a step-wise design. First, we explored which chemical species are the most harmful to the exposure population with respect to specific death outcomes using a single-pollutant model. Second, we assessed which chemical species are consistent after controlling for the other species using a multipollutant model. Finally, we explored which sources are mainly responsible for risk effects on mortality (and specifically, which species that is a key or fingerprinting constituent characterizing the harmful sources appears to be discounted due to losing an association compared with the source apportionment results).
We found that mobile emissions were strongly associated with respiratory mortality, and in contrast, biomass burning and industrial sources were associated with the largest risk of cardiovascular mortality. Of the individual constituents, elemental carbon and lead were associated with respiratory mortality, whereas organic carbon, magnesium, and lead were substantially associated with cardiovascular mortality after adjustment for other constituents. In general, mobile sources are characterized by high levels of insoluble chemical species, including water-insoluble organic carbon (eg, heavy polycyclic aromatic hydrocarbons) and elemental carbon, whereas biomass burning is characterized by mostly soluble chemical species, such as water-soluble organic carbon that comprises polar organic compounds, including levoglucosan and fatty acids, and water-soluble potassium. In addition, the industrial source is dominated mainly by transition metals, including copper and zinc that are soluble species. Although it is not yet clear why risk effects of PM2.5 chemical species and sources on mortality vary by disease groups, the findings in this study would support the biological plausibility that soluble components penetrating deep into organelles induce cardiovascular responses via oxidative stress and nervous system activation, whereas insoluble components may stay longer on the surfaces of alveoli, resulting in respiratory inflammation. Furthermore, by the source apportionment technique, inconsistent risk effects of specific constituents that lose an association in multipollutant models due to the undetected levels, as well as correlation between other pollutants or other potential confounders (such as weather and season), are complemented, when categorized as the same sources.
In Asian countries, two studies have been conducted to estimate the risk effects of PM2.5 chemical constituents on mortality outcomes in an urban area.20,44 The findings in our study are consistent with a previous study performed in Xi’an, China, that showed combustion-related PM2.5 chemical species such as organic carbon, elemental carbon, ammonium, and nickel were associated with mortality, but common crustal elements in PM2.5 were not.20 However, our results are somewhat different from those of the other study at Seoul that applied a time series analysis to real-time measurements of organic carbon, elemental carbon, and water-soluble ions collected over a 1-year period and found that only magnesium and ammonium ions were substantially associated with mortality outcomes.44 This difference is likely due to measurement differences. Epidemiologic studies examining the relationship of PM2.5 sources with health outcomes have been reported in Europe and North America, to help target future pollution control strategies, but neither of the studies in East Asia considered the risk effects of various PM2.5 sources. The findings here linking PM2.5 sources and mortality outcomes complement the results of both studies and may provide insights into applications of modeled source-apportioned data to health risk assessment.
There are several limitations to this study. First, PM2.5 speciation data were collected at a single monitoring site, and thus, there are innate random measurement errors. These errors could temper risk estimates. Second, cumulative short-term exposure effects were not estimated with consecutive daily data because PM2.5 samples were collected at 3-day intervals. Previous time series analyses reported that continuous daily exposure data tend to generate larger effect estimates than that of a single-day lag.7–10,45 The risk estimates for a single-day lag in this study presumably hold true but remain underestimated.
In summary, we examined the risks of PM2.5 mass and its chemical species, as well as its sources, on mortality using time series analyses. Several chemical species and PM2.5 mass were found to be strongly associated with daily mortality in Seoul during 2003–2007. PM2.5 mass derived from local combustion sources (ie, gasoline emission, diesel emission, and biomass burning) had the largest risk effects on daily death counts for cardiovascular and respiratory mortality, suggesting that, although PM2.5 levels have been influenced by both urban activities and regional-scale transport, local combustion sources remain particularly important in the study region.
1. U.S. EPA. Air Quality Criteria for Particulate Matter. EPA 600/P/-99/002a,bF. 2004 Research Triangle Park, NC U.S. Environmental Protection Agency
2. Pope CA 3rd, Dockery DW. Health effects of fine particulate air pollution: lines that connect. J Air Waste Manag Assoc. 2006;56:709–742
3. Bell ML, Ebisu K, Peng RD, Samet JM, Dominici F. Hospital admissions and chemical composition of fine particle air pollution. Am J Respir Crit Care Med. 2009;179:1115–1120
4. Burnett RT, Brook J, Dann T, et al. Association between particulate- and gas-phase components of urban air pollution and daily mortality in eight Canadian cities. Inhal Toxicol. 2000;12(suppl 4):15–39
5. Franklin M, Koutrakis P, Schwartz P. The role of particle composition on the association between PM2.5 and mortality. Epidemiology. 2008;19:680–689
6. Laden F, Neas LM, Dockery DW, Schwartz J. Association of fine particulate matter from different sources with daily mortality in six U.S. cities. Environ Health Perspect. 2000;108:941–947
7. Ostro B, Feng WY, Broadwin R, Green S, Lipsett M. The effects of components of fine particulate air pollution on mortality in California: results from CALFINE. Environ Health Perspect. 2007;115:13–19
8. Ostro B, Roth L, Malig B, Marty M. The effects of fine particle components on respiratory hospital admissions in children. Environ Health Perspect. 2009;117:475–480
9. Peng RD, Bell ML, Geyh AS, et al. Emergency admissions for cardiovascular and respiratory diseases and the chemical composition of fine particle air pollution. Environ Health Perspect. 2009;117:957–963
10. Lall R, Ito K, Thurston GD. Distributed lag analyses of daily hospital admissions and source-apportioned fine particle air pollution. Environ Health Perspect. 2011;119:455–460
11. Ostro B, Tobias A, Querol X, et al. The effects of particulate matter sources on daily mortality: a case-crossover study of Barcelona, Spain. Environ Health Perspect. 2011;119:1781–1787
12. Sarnat JA, Marmur A, Klein M, et al. Fine particle sources and cardiorespiratory morbidity: an application of chemical mass balance and factor analytical source-apportionment methods. Environ Health Perspect. 2008;116:459–466
13. Heo JB, Hopke PK, Yi SM. Source apportionment of PM2. 5 in Seoul, Korea. Atmos Chem Phys. 2009;9:4957–4971
14. Kim HS, Huh JB, Hopke PK, Holsen TM, Yi SM. Characteristics of the major chemical constituents of PM2. 5 and smog events in Seoul, Korea in 2003 and 2004. Atmos Environ. 2007;41:6762–6770
15. Paatero P, Tapper U. Positive matrix factorization: a non-negative factor model with optimal utilization of error estimates of data values. Environmetrics. 1994;5:111–126
16. Heo J, Dulger M, Olson MR, et al. Source apportionments of PM2.5 organic carbon using molecular marker Positive Matrix Factorization and comparison of results from different receptor models. Atmos Environ. 2013;73:51–61
17. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. 2012 Vienna, Austria Available at: http://www.R-project.org
. Accessed October 10, 2012.
18. Bell ML, Dominici F, Ebisu K, Zeger SL, Samet JM. Spatial and temporal variation in PM(2.5) chemical composition in the United States for health effects studies. Environ Health Perspect. 2007;115:989–995
19. Putaud J-P, Van Dingenen R, Alastuey A, et al. A European aerosol phenomenology—3: Physical and chemical characteristics of particulate matter from 60 rural, urban, and kerbside sites across Europe. Atmos Environ. 2010;44:1308–1320
20. Cao J, Xu H, Xu Q, Chen B, Kan H. Fine particulate matter constituents and cardiopulmonary mortality in a heavily polluted Chinese city. Environ Health Perspect. 2012;120:373–378
21. Minoura H, Takahashi K, Chow JC, Watson JG. Multi-year trend in fine and coarse particle mass, carbon, and ions in downtown Tokyo, Japan. Atmos Environ. 2006;40:2478–2487
22. Cho YS, Lee JT, Jung CH, Chun YS, Kim YS. Relationship between particulate matter measured by optical particle counter and mortality in Seoul, Korea, during 2001. J Environ Health. 2008;71:37–43
23. Ostro B, Lipsett M, Reynolds P, et al. Long-term exposure to constituents of fine particulate air pollution and mortality: results from the California Teachers Study. Environ Health Perspect. 2010;118:363–369
24. Zanobetti A, Franklin M, Koutrakis P, Schwartz J. Fine particulate air pollution and its components in association with cause-specific emergency admissions. Environ Health. 2009;8:58
25. Mar TF, Norris GA, Koenig JQ, Larson TV. Associations between air pollution and mortality in Phoenix, 1995-1997. Environ Health Perspect. 2000;108:347–353
26. Metzger KB, Tolbert PE, Klein M, et al. Ambient air pollution and cardiovascular emergency department visits. Epidemiology. 2004;15:46–56
27. Brook RD, Franklin B, Cascio W, et al.Expert Panel on Population and Prevention Science of the American Heart Association. Air pollution and cardiovascular disease: a statement for healthcare professionals from the Expert Panel on Population and Prevention Science of the American Heart Association. Circulation. 2004;109:2655–2671
28. Cakmak S, Dales RE, Vida CB. Components of particulate air pollution and mortality in Chile. Int J Occup Environ Health. 2009;15:152–158
29. Anderson HR, Bremner SA, Atkinson RW, Harrison RM, Walters S. Particulate matter and daily mortality and hospital admissions in the west midlands conurbation of the United Kingdom: associations with fine and coarse particles, black smoke and sulphate. Occup Environ Med. 2001;58:504–510
30. Gent JF, Koutrakis P, Belanger K, et al. Symptoms and medication use in children with asthma and traffic-related sources of fine particle pollution. Environ Health Perspect. 2009;117:1168–1174
31. McCreanor J, Cullinan P, Nieuwenhuijsen MJ, et al. Respiratory effects of exposure to diesel traffic in persons with asthma. N Engl J Med. 2007;357:2348–2358
32. Kendall M, Brown L, Trought K. Molecular adsorption at particle surfaces: a PM toxicity mediation mechanism. Inhal Toxicol. 2004;16(suppl 1):99–105
33. O’Neill MS, Veves A, Sarnat JA, et al. Air pollution and inflammation in type 2 diabetes: a mechanism for susceptibility. Occup Environ Med. 2007;64:373–379
34. Diaz-Sanchez D, Penichet-Garcia M, Saxon A. Diesel exhaust particles directly induce activated mast cells to degranulate and increase histamine levels and symptom severity. J Allergy Clin Immunol. 2000;106:1140–1146
35. Hiura TS, Li N, Kaplan R, Horwitz M, Seagrave JC, Nel AE. The role of a mitochondrial pathway in the induction of apoptosis by chemicals extracted from diesel exhaust particles. J Immunol. 2000;165:2703–2711
36. Hong YC, Hwang SS, Kim JH, et al. Metals in particulate pollutants affect peak expiratory flow of schoolchildren. Environ Health Perspect. 2007;115:430–434
37. Bae S, Pan XC, Kim SY, et al. Exposures to particulate matter and polycyclic aromatic hydrocarbons and oxidative stress in schoolchildren. Environ Health Perspect. 2010;118:579–583
38. Daggett DA, Oberley TD, Nelson SA, Wright LS, Kornguth SE, Siegel FL. Effects of lead on rat kidney and liver: GST expression and oxidative stress. Toxicology. 1998;128:191–206
39. Saxena G, Flora SJ. Lead-induced oxidative stress and hematological alterations and their response to combined administration of calcium disodium EDTA with a thiol chelator in rats. J Biochem Mol Toxicol. 2004;18:221–233
40. Bagci C, Bozkurt AI, Cakmak EA, Can S, Cengiz B. Blood lead levels of the battery and exhaust workers and their pulmonary function tests. Int J Clin Pract. 2004;58:568–572
41. Dye JA, Lehmann JR, McGee JK, et al. Acute pulmonary toxicity of particulate matter filter extracts in rats: coherence with epidemiologic studies in Utah Valley residents. Environ Health Perspect. 2001;109(suppl 3):395–403
42. Pritchard RJ, Ghio AJ, Lehmann JR, et al. Oxidant generation and lung injury after particulate air pollutant exposure increase with the concentrations of associated metals. Inhal Toxicol. 1996;8:457–477
43. Lagorio S, Forastiere F, Pistelli R, et al. Air pollution and lung function among susceptible adult subjects: a panel study. Environ Health. 2006;5:11
44. Son JY, Lee JT, Kim KH, Jung K, Bell ML. Characterization of fine particulate matter and associations between particulate chemical constituents and mortality in Seoul, Korea. Environ Health Perspect. 2012;120:872–878
45. Schwartz J. The distributed lag between air pollution and daily deaths. Epidemiology. 2000;11:320–326
Supplemental Digital Content
© 2014 by Lippincott Williams & Wilkins, Inc