Skip Navigation



American Journal of Epidemiology Advance Access published online on October 14, 2008
American Journal of Epidemiology, doi:10.1093/aje/kwn252
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplemental Material
Right arrow All Versions of this Article:
168/11/1301    most recent
kwn252v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Bell, M. L.
Right arrow Articles by Dominici, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bell, M. L.
Right arrow Articles by Dominici, F.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology © The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.

Original Contribution

Seasonal and Regional Short-term Effects of Fine Particles on Hospital Admissions in 202 US Counties, 1999–2005

Michelle L. Bell, Keita Ebisu, Roger D. Peng, Jemma Walker, Jonathan M. Samet, Scott L. Zeger and Francesca Dominici

Correspondence to Dr. Michelle L. Bell, 205 Prospect Street, Yale University, New Haven, CT 06511 (e-mail: michelle.bell{at}yale.edu).

Received for publication March 3, 2008. Accepted for publication July 23, 2008.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
The authors investigated whether short-term effects of fine particulate matter with an aerodynamic diameter ≤2.5 µm (PM2.5) on risk of cardiovascular and respiratory hospitalizations among the elderly varied by region and season in 202 US counties for 1999–2005. They fit 3 types of time-series models to provide evidence for 1) consistent particulate matter effects across the year, 2) different particulate matter effects by season, and 3) smoothly varying particulate matter effects throughout the year. The authors found statistically significant evidence of seasonal and regional variation in estimates of particulate matter effect. Respiratory disease effect estimates were highest in winter, with a 1.05% (95% posterior interval: 0.29, 1.82) increase in hospitalizations per 10-µg/m3 increase in same-day PM2.5. Cardiovascular diseases estimates were also highest in winter, with a 1.49% (95% confidence interval: 1.09, 1.89) increase in hospitalizations per 10-µg/m3 increase in same-day PM2.5, with associations also observed in other seasons. The strongest evidence of a relation between PM2.5 and hospitalizations was in the Northeast for both respiratory and cardiovascular diseases. Heterogeneity of PM2.5 effects on hospitalizations may reflect seasonal and regional differences in emissions and in particles’ chemical constituents. Results can help guide development of hypotheses and further epidemiologic studies on potential heterogeneity in the toxicity of constituents of the particulate matter mixture.

air pollution; hospitalization; Medicare; particulate matter; seasons

Abbreviations: ICD-9, International Classification of Diseases, Ninth Revision; PM2.5, particulate matter with an aerodynamic diameter ≤2.5 µm; PM10, particulate matter with an aerodynamic diameter ≤10 µm


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Numerous studies have demonstrated increased risk of cardiovascular and respiratory hospitalizations in relation to airborne particles, including particulate matter with an aerodynamic diameter ≤10 µm or ≤2.5 µm (PM10 or PM2.5) (1). Previous research identified associations between PM2.5 and chronic obstructive pulmonary disease hospital admissions (2) and between coarse particulate matter (PM10–2.5) and respiratory hospitalizations (3). PM10 has been associated with admissions for adult asthma (4), cardiopulmonary causes (5), and cardiovascular disease (6) and with emergency admissions for childhood asthma (7) and cardiovascular causes (8).

Recent studies suggest that particulate matter effects vary by region and season. A study of cause-specific cardiovascular and respiratory hospital admissions and daily PM2.5 levels among Medicare enrollees found strong regional patterns of effect across 204 US counties (9). Effect estimates for some cardiovascular causes, including ischemic heart and peripheral vascular diseases, were statistically significant in the eastern but not the western United States. For chronic obstructive pulmonary disease and respiratory tract infection, effects were observed in both eastern and western regions, but they were larger in the latter. Short-term effects of PM10 on mortality were larger in the Northeast and in summer, whereas evidence for seasonal variation was not found in the southern United States (10). In the Air Pollution and Health: A European Approach (APHEA) study, effect estimates for particles and mortality were lower for central-eastern Europe than western Europe (11), although more recent work suggests these differences are explained in part by statistical modeling choices (12). Other work shows spatial differences in PM10–mortality associations in the United States (13) and seasonal variation in coarse particulate matter effects on lung inflammation (14).

In addition to seasonal variation, time trends of effect have been examined. Methods developed in the National Morbidity, Mortality and Air Pollution Study were applied to evaluate change in short-term PM10 effects over a period of increasingly stringent regulation that might have altered particulate matter composition and toxicity (15–17). There was weak evidence that the effects declined over the period 1987–2000, primarily in the eastern United States.

Regional and temporal differences in effect estimates may relate to heterogeneity in the particulate matter mixture. In the United States, we found substantial spatial and temporal variability in PM2.5 chemical composition (18). However, differences in effect estimates across locations could also reflect differences in exposure patterns, such as indoor versus outdoor activity patterns, and community characteristics, including the presence of susceptible subpopulations. Heterogeneity in effect estimates across seasons could reflect seasonal variation in particulate matter toxicity or confounding by a seasonally varying factor, such as ozone pollution. In addition, evidence that health effect estimates have different seasonal and regional patterns by cause would be indicative of multiple mechanisms of toxicity.

This study quantified evidence of spatial and temporal heterogeneity in the health effects of short-term exposure to particles. We applied 3 statistical approaches to investigate the short-term effects of PM2.5 on cardiovascular or respiratory hospitalizations among Medicare enrollees by season and geographic region of the United States. We also identified PM2.5 chemical components with higher levels for regions and seasons with higher effect estimates compared with regions and seasons with lower effect estimates.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
We used a national database of hospital admissions for 1999–2005 based on Medicare enrollees aged ≥65 years for 202 US counties with populations ≥200,000. Each Medicare claim includes age and place of residence. The number of hospitalizations for a given cause on a given day and for a specific community was calculated as the sum of all claims for that cause based on primary diagnosis. The number of individuals at risk was defined as the number of Medicare enrollees on a given day for that community.

We considered urgent hospitalizations for cardiovascular and respiratory causes, excluding scheduled visits that by definition are not pollution related. Cardiovascular admissions were calculated as the sum of hospitalizations for heart failure (International Classification of Diseases, Ninth Revision (ICD-9) code 428), heart rhythm disturbances (ICD-9 codes 426–427), cerebrovascular events (ICD-9 codes 430–438), ischemic heart disease (ICD-9 codes 410–414 and 429), and peripheral vascular disease (ICD-9 codes 440–449); and we determined respiratory admissions as the sum of admissions for chronic obstructive pulmonary disease (ICD-9 codes 490–492) and respiratory tract infections (ICD-9 codes 464–466, 480–487). These ICD codes were used in earlier work, which enhances comparability across studies including research on hospitalizations and PM2.5 (9) and PM10–2.5 (19).

PM2.5 data were obtained from the US Environmental Protection Agency. While some communities measured PM2.5 daily, most measured every 3 days. We used a 10% trimmed mean to average across monitors after correction for yearly monitor averages, protecting against outlier values and as applied in earlier studies (9). Daily temperature and dew point temperatures were obtained from the National Climatic Data Center.

We applied a 2-stage Bayesian hierarchical model to estimate the association between the daily PM2.5 and hospitalization rates on average across the 202 US counties. The first stage estimated this association within a single county, accounting for several time-varying confounders; the second stage combined county-specific estimates, accounting for their statistical uncertainty to generate an overall effect. At the first stage, we fitted a "main" model that assumes that the short-term PM2.5 effect on hospitalizations is constant throughout the year and a "seasonal" model allowing the effect to vary by season. As a sensitivity analysis, we applied a third approach of using a "harmonic" model allowing the PM2.5 effect to vary smoothly throughout the year (10).

The main-effect model can be defined as

Formula (1)
where

hFormula = hospitalization rate in county c, day t
βc = regression coefficient relating PM2.5 to hospitalization rates in county c
xFormula = PM2.5 level in county c, day t, at lag of l days (e.g., l = 0 is the same day)
{alpha}c = regression coefficient relating day of the week to hospitalization rates in county c
DOWt = day of the week on day t
ns(TFormula, dfT) = natural cubic spline of temperature in county c, day t with dfT (six) degrees of freedom
ns(DFormula, dfFormula) = natural cubic spline of dew point temperature in county c, day t with dfD (three) degrees of freedom
ns(TaFormula, dfTa) = natural cubic spline with dfTa (six) degrees of freedom for the average of the 3 previous days’ temperature in county c, day t, adjusted for current day temperature and dew point temperature
ns(DaFormula, dfDa) = natural cubic spline with dfDa (three) degrees of freedom for the average of the 3 previous days’ dew point temperature in county c, day t, adjusted for current day temperature and dew point temperature
ns(t, dft) = natural cubic spline of time with dft (eight/year) degrees of freedom
ns(t, dfAt) = natural cubic spline of time with dfAt (one/year) degrees of freedom
At = indicator for persons aged ≥75 years
NFormula = size of the population at risk in county c, day t.

The above model was fitted separately for each county and hospitalization cause (cardiovascular or respiratory). The nonlinear relation between health and weather was modeled by using natural splines of temperature and dew point temperature, including variables for previous days’ conditions. We accounted for temporal trends through a nonlinear function of time. Differential temporal trends by age category were modeled through an interaction term between age and the nonlinear temporal function. A version of this model was used previously to investigate PM2.5 and cause-specific hospitalizations (9).

The first-stage seasonal interaction model allows effect estimates to differ by season, replacing the pollution term in equation 1, βcxFormulac, with

Formula (2)
where

Iw, ISp, ISu, IA = 0/1 indicator variables representing winter, spring, summer, and autumn, respectively
βw, βSp, βSu, βA = regression coefficients regarding the relation between PM2.5 and hospitalization rates for a given season.

We also allowed temporal trend to differ by season, replacing the temporal trend term in equation 1, ns(TFormula, dfT), with

Formula (3)

The other terms of equation 1 were maintained in the seasonal interaction model. For the seasonal interaction model, seasons were defined as 3-month periods (e.g., summer as June–August). This model provides seasonal estimates of the relation between PM2.5 and cause-specific hospitalizations by county, and it assumes no temporal variation in PM2.5 effects by season. The harmonic model relaxes this assumption by allowing PM2.5 effects to vary smoothly throughout the year and was initially developed to investigate seasonality in PM10 mortality effects (10). The harmonic model is analogous to equation 1, replacing the pollution term with

Formula (4)
All nonpollution terms from equation 1 were maintained in the harmonic model.

These 3 model structures incorporate different assumptions about the relation between PM2.5 and hospitalizations. The main-effect model assumes constant effects throughout the year. The seasonal interaction model improves upon the main-effect model by allowing the association to differ by season. The harmonic model is not subject to the specifications of seasons or the condition that the effect be constant throughout the year or in a given season. Main-effect and seasonal interaction models were applied for cardiovascular and respiratory admissions and PM2.5 at lags 0, 1, and 2 days. The harmonic model was applied for cardiovascular and respiratory admissions at the lag with the strongest effect, as reported by the main-effect and seasonal interaction models.

The main-effect, first-stage model provides an estimate of the relation between PM2.5 and hospitalizations for a given county Formula and its estimated variance, whereas the seasonal interaction, first-stage model provides an estimate of the relation between PM2.5 and hospitalizations for a given county (4-dimensional vector Formula ) and its estimated covariance matrix (VFormula) for each season. The second-stage model assumes that the true relation in a given county (βFormula) has a multivariate normal distribution with mean (µ) and the between-community variance ({Sigma}). We applied this model by using 2-level normal independent sampling estimation (TLNise) with uniform priors (20) as

Formula (5)
A similar version of the second-stage modeling structure described in equation 5 was applied to the first-stage, county-specific estimates of the main and harmonic models (21–23). Sensitivity of the risk estimates to the smooth function of time and the models for temperature and dew point have been explored previously, indicating that national average estimates obtained from hierarchical models are robust to a wide variety of modeling approaches (24–26).

We fitted all 3 models (main, seasonal, and harmonic) separately within geographic regions. Noncontinental counties in the national analysis were excluded from regional analyses (Honolulu, Hawaii; Anchorage, Alaska). We divided the remaining 200 counties into 4 regions: Northeast (n = 108), Southeast (n = 58), Northwest (n = 9), and Southwest (n = 25). Regions were based on spatial divisions applied previously in the National Morbidity, Mortality and Air Pollution Study and studies investigating hospital admissions (9, 10, 27).

We tested for evidence of seasonal and regional heterogeneity in the short-term effects of PM2.5 on hospitalizations for the lags with the strongest effects for each hospitalization cause. Specifically, we used the Wald test statistic to assess whether there is evidence of heterogeneity in 1) national average effects across seasons, and 2) regional average effects across regions for both cardiovascular and respiratory admissions. The Wald test statistic was compared with a chi-square distribution with appropriate degrees of freedom to obtain corresponding significance levels. A significance level of P < 0.05 was used; we did not include multiple testing correction for the 4 simultaneous tests.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Respiratory admissions rates showed a seasonal pattern with higher admissions in winter, whereas cardiovascular admission rates were more similar across seasons. Figure S1 provides box plots of hospitalization rates by region and season, and Figure S2 shows analogous plots of PM2.5 and weather variables (these 2 supplementary figures and other supplementary information are posted on the Journal’s website (http://aje.oupjournals.org/)).

Table 1 summarizes national average estimates of the association between PM2.5 and hospitalizations by lag for the yearly estimates (main-effect model) and by season (seasonal interaction model). Results are presented as the percentage increase in hospital admission rates per 10-µg/m3 increase in PM2.5. The value of 10 µg/m3 is close to the interquartile range of overall PM2.5 levels (8.7 µg/m3).


View this table:
[in this window]
[in a new window]

  		Table 1. National Estimates of the Percentage Increase in Hospital Admission Rates per 10-µg/m3 Increase in PM2.5 for 202 US Counties for 1999–2005, by Lag and by Season

 
For the main model, PM2.5 was associated with cardiovascular admissions on the same day (lag 0) and with respiratory admissions at lag 2. These lags were also identified as those with the strongest effects in earlier work using analysis similar to that for the main-effect model (9). For cardiovascular admissions, associations at lag 0 were observed for all seasons except summer, with the strongest effect in winter at lag 0. For respiratory admissions, associations were observed in winter only, at lags 0 and 2. Again, the strongest effect is in winter at lag 0.

Table 2 provides regional average estimates by season at lag 0 for cardiovascular admissions and lags 0 and 2 for respiratory admissions for the full year (main-effect model) and by season (seasonal interaction model). For lag-0 cardiovascular admissions, the largest effects occurred in the Northeast in winter, and significant effects were also observed in all other seasons in this region. For lag-0 respiratory admissions, the largest effect also occurred in the Northeast in winter. For lag-2 respiratory admissions, the yearly estimate was largest in the Southwest.


View this table:
[in this window]
[in a new window]

  		Table 2. Estimates of the Percentage Increase in Hospital Admission Rates per 10-µg/m3 Increase in PM2.5 for 202 US Counties, by Season and Region

 
We tested for evidence of regional and seasonal heterogeneity in the short-term effects of PM2.5 on hospital admissions for lag-0 cardiovascular and respiratory admissions. We found strong evidence of variability across seasons in national average effects of PM2.5 for cardiovascular (P < 0.01) and respiratory (P < 0.01) admissions. Cardiovascular effect estimates were also heterogeneous across regions (P = 0.03), whereas respiratory effects did not exhibit statistically significant evidence of heterogeneity.

Results from the sensitivity analysis using the harmonic model support those from the seasonal interaction model, as shown in Figures 1 (cardiovascular) and 2 (respiratory). These findings indicate that the seasonal patterns identified by the seasonal interaction model are not an artifact of the choice of seasonal division (e.g., summer as June–August) because the harmonic model allows smooth variation in effect estimates throughout the year by location. Supplemental Figures S3 and S4 are color versions of Figures 1 and 2. Figures 1 and 2 apply the same scale to the health effect estimates, although some 95% posterior intervals are too large to fit on this scale. Supplemental Figures S5 and S6 are alternative versions of Figures 1 and 2, with scales allowing a full view of the 95% posterior intervals.


Figure 1
View larger version (18K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1. Percentage increase in total cardiovascular hospital admissions (1999–2005) per 10-µg/m3 increase in same-day (lag 0) particulate matter with an aerodynamic diameter ≤2.5 µm, by US region, for results from the harmonic model (curved lines) and seasonal interaction model (straight lines). Vertical lines mark the divisions among seasons as defined by the seasonal interaction model. For the harmonic model (curved lines), solid lines represent the central estimate and dashed lines the 95% posterior interval. For the seasonal interaction model (straight lines), horizontal dashed lines represent the central estimate and vertical dotted lines the 95% posterior interval. The number in parentheses (n) represents the number of US counties included in each region. Note that the y-axis scale is identical across regions. The 95% intervals for some regions are too large to fit on this scale (e.g., Northwest region for the harmonic model); a color version of this figure is available as supplemental Figure S3 (posted on the Journal's website (http://aje.oupjournals.org/)).

 

Figure 2
View larger version (16K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 2. Percentage increase in total respiratory hospital admissions (1999–2005) per 10-µg/m3 increase in same-day (lag 0) particulate matter with an aerodynamic diameter ≤2.5 µm, by US region, for results from the harmonic model (curved lines) and seasonal interaction model (straight lines). Vertical lines mark the divisions among seasons as defined by the seasonal interaction model. For the harmonic model (curved lines), solid lines represent the central estimate and dashed lines the 95% posterior interval. For the seasonal interaction model (straight lines), horizontal dashed lines represent the central estimate and vertical dotted lines the 95% posterior interval. The number in parentheses (n) represents the number of US counties included in each region. Note that the y-axis scale is identical across regions. The 95% intervals for some regions are too large to fit on this scale (e.g., Northwest region for the harmonic model); a color version of this figure is available as supplemental Figure S4 (posted on the Journal’s website (http://aje.oupjournals.org/)).

 
Variation in seasonal and regional effect estimates may result from differences in the chemical composition of particulate matter. The chemical component database currently is not sufficiently large to estimate the short-term effects of PM2.5 chemical components on hospital admissions by season and region. As a descriptive analysis, we identified the chemical components of PM2.5 that are higher for the regions and seasons with higher effect estimates than for the regions and seasons with lower effect estimates. First, we identified the regions and seasons with higher effect estimates for PM2.5 and hospital admissions based on overall trends. Then, for each chemical component, we calculated ratios between the average concentration for the region or season with the largest effect estimate divided by the average concentration for all other regions and seasons combined. In this paper, results are presented for the PM2.5 chemical components with >20% levels in the places or time periods with higher effect estimates.

County-level component averages were based on a database of 52 PM2.5 chemical components from 2000–2005, previously developed from the US Environmental Protection Agency's monitoring network (18). Chemical composition data were available for 106 counties and were used to estimate long-term levels and seasonal averages for each region and for the United States. For the main-effect model, associations between cardiovascular admissions and PM2.5 were strongest in the Northeast regarding both magnitude of effect and degree of statistical significance. Figure 3A presents components that were ≥20% higher in the Northeast than in other regions, and it provides box plots for the Northeast (gray boxes) and other regions (white boxes). National effect estimates for respiratory admissions were higher in winter than in other seasons. Levels of sulfate, selenium, ammonium, nitrate, nickel, zinc, and lead were higher in this region than elsewhere. Results from the seasonal interaction model for respiratory admissions based on all 202 counties exhibited higher effects in winter than in other seasons. For nitrate, elemental carbon, nickel, bromine, zinc, and chlorine, national averages were ≥20% higher in winter than in other seasons (Figure 3B).


Figure 3
View larger version (24K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 3. Comparison of concentrations of chemical components of particulate matter with an aerodynamic diameter ≤2.5 µm (PM2.5) (µg/m3), by US region and season, 1999–2005. A) Concentrations of specific components that were ≥20% higher in the Northeast than in other regions; B) concentrations of specific components that were ≥20% higher in winter than in other seasons. Each part uses 2 y-axis scales. Components to the left of the vertical line use the left-side y-axis scale; components to the right of the vertical line use the right-side y-axis scale. SO4=, sulfate; Se, selenium; NH4+, ammonium; NO3–, nitrate; Ni, nickel; Zn, zinc; Pb, lead; EC, elemental carbon; Br, bromine; Cl, chlorine.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
Efforts to protect human health from ambient particulate matter are limited by scientific understanding of the toxicity of various components of the particulate matter mixture and the sources that contribute injurious particles (28). The components of the particle mixture vary seasonally and regionally with source use patterns and weather (18). If the chemical composition of particles affects toxicity, we would expect to find evidence of seasonal and regional heterogeneity in the short-term risks associated with PM2.5 total mass. A lack of spatial and temporal heterogeneity would provide evidence for the hypothesis that particulate matter risk is not related to chemical composition. The presence of heterogeneity could result from variation in the toxicity of particulate matter chemical constituents, the nature of the air pollution mixture and the presence of other pollutants, or differences in populations’ susceptibility. Characterization of spatial and temporal heterogeneity in risks associated with particulate matter provides an opportunity to test hypotheses regarding the significance of particle characteristics for human health and to develop focused hypotheses based on variation in risks by time period and region.

We conducted a multisite time-series analysis to investigate whether regional and seasonal heterogeneity exists in the short-term effects of PM2.5 on cardiovascular and respiratory hospital admissions. Both cardiovascular and respiratory disease effect estimates exhibited spatial and temporal differences, although the patterns differed by disease outcome. This pattern might indicate that the particle component or set of components most strongly associated with the adverse health response may vary by health outcome. However, for both outcomes, PM2.5 had the largest effects in winter and the Northeast.

Components with higher concentrations in the seasons and regions with the largest short-term effects of PM2.5 total mass on hospitalization (Figure 3) relate to several source categories. In particular, these components correspond to several combustion sources as well as metals and sea salt (1). As additional chemical component measurements become available, further analysis is warranted, such as multisite time-series studies investigating whether the health effects of particular components or set of components also vary seasonally and spatially.

To date, findings from several epidemiologic studies indicate that certain particulate matter components may be more harmful than others with respect to mortality (29, 30). Daily mortality in 6 California counties was associated with various PM2.5 chemical components and source categories, including metals (zinc, lead, vanadium), combustion-related products (sulfate), crustal components (silicon, calcium), chlorine, and carbon (organic carbon, elemental carbon) (30). Combustion-related PM2.5 (sulfate, elemental carbon) was associated with cardiovascular mortality in Phoenix, Arizona (31). Sulfate and metals (iron, nickel, zinc) were linked to mortality in 8 Canadian cities (32). Combustion and traffic-related PM2.5 sources, but not crustal PM2.5, were associated with mortality in 6 US eastern cities (33). Higher effect estimates for PM10 on mortality were observed in communities with a higher percentage of primary PM10 from traffic sources (34). Other studies have linked individual components to health responses (29), such as sulfate and organic carbon PM2.5, and heart rate variability in cardiovascular disease patients.

PM2.5 in the Northeast, the region with the largest effect estimates in this study, contains higher sulfate PM2.5 levels than elsewhere (Figure 3) (18). In this region, PM2.5 levels are linked to emissions of sulfur oxides and nitrogen oxides from power-generation point sources in the Midwest, and sulfate levels are particularly related to sulfur oxide emissions (35, 36). In addition to coal combustion for electricity production, local sources of residual oil combustion contribute to particulate matter in the Northeast (37).

The observed regional heterogeneity in the short-term effects of PM2.5 may also be explained by differences in population susceptibility, access to health care, and socioeconomic status. US maps of chronic obstructive pulmonary disease and heart disease mortality rates show substantial regional variation, although rates are not consistently higher in the Northeast, the area with the highest effect estimates in this study (38). Short-term effects of PM10 on mortality were greater in communities with higher population densities (34), while short-term effects of PM10 on hospital admissions were greater in communities in which a lower percentage of the population used air conditioning (39, 40). Thus, although our results indicate that the impact of chemical composition on toxicity is important, other factors may also be relevant.

Seasonal and regional variation in effect estimates could also be explained by variation in exposure patterns. Studies of persons aged >64 years in Baltimore, Maryland, and Boston, Massachusetts, identified different personal PM2.5 exposures, varied activity patterns, and differing use of gas stoves by season (41, 42). The relation between ambient and personal PM2.5 exposure for an older population can also differ by season (43). A study of 3 retirement communities found that housing factors, such as open versus closed windows and use of heating, ventilation, and air conditioning systems, influence the relation between ambient and personal PM2.5 exposure (44).

Disentangling effects of multiple pollutants can be challenging, especially because of the different measurement frequencies for various pollutants as well as common sources producing multiple pollutants in the overall air pollution mixture. For example, traffic contributes to particles and ozone precursors. Analysis directed at separating the risk from PM2.5 and from gaseous pollutants, such as nitrogen dioxide and sulfur dioxide that contribute to secondary particle formation, is subject to substantial uncertainty.

This study demonstrated regional and temporal patterns in the association between PM2.5 and cardiovascular and respiratory hospitalizations, with the strongest evidence in the Northeast and in winter for both causes. The clear finding of heterogeneity provides a rationale for further work to understand its basis. The higher effect estimates in the Northeast, previously observed for mortality (10), need further explanation and testing of hypotheses related to pollution sources for this region, such as power plants.


   ACKNOWLEDGMENTS
 
Author affiliations: School of Forestry and Environmental Studies, Yale University, New Haven, Connecticut (Michelle L. Bell, Keita Ebisu); Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland (Roger D. Peng, Scott L. Zeger, Francesca Dominici); Medical Statistics Unit, London School of Hygiene and Tropical Medicine, London, United Kingdom (Jemma Walker); and Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland (Jonathan M. Samet).

The National Institute for Environmental Health Sciences (NIEHS) Outstanding New Environmental Scientists (ONES) Award (grant R01 ES015028), Health Effects Institute Rosenblith Award (grant 4720-RFA04-2/04-16), and US Environmental Protection Agency–sponsored Johns Hopkins Particulate Matter Research Center (grant RD-83241701) funded this research.

The views expressed in this article are those of the authors and do not necessarily reflect the views or polices of the funding agencies.

Conflict of interest: none declared.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. US EPA. Air Quality Criteria for Particulate Matter (2004) Research Triangle Park, NC: US Environmental Protection Agency, Office of Research and Development. (EPA/600/P-99/002).
  2. Ko FW, Tam W, Wong TW, et al. Temporal relationship between air pollutants and hospital admissions for chronic obstructive pulmonary disease in Hong Kong. Thorax (2007) 62(9):780–785.[Abstract/Free Full Text]
  3. Fung KY, Khan S, Krewski D, et al. Association between air pollution and multiple respiratory hospitalizations among the elderly in Vancouver, Canada. Inhal Toxicol (2006) 18(13):1005–1011.[CrossRef][Web of Science][Medline]
  4. Chen CH, Xirasagar S, Lin HC. Seasonality in adult asthma admissions, air pollutant levels, and climate: a population-based study. J Asthma (2006) 43(4):287–292.[CrossRef][Web of Science][Medline]
  5. Arena VC, Mazumdar S, Zborowski JV, et al. A retrospective investigation of PM10 in ambient air and cardiopulmonary hospital admissions in Allegheny County, Pennsylvania: 1995–2000. J Occup Environ Med (2006) 48(1):38–47.[CrossRef][Web of Science][Medline]
  6. Le Tertre A, Medina S, Samoli E, et al. Short-term effects of particulate air pollution on cardiovascular diseases in eight European cities. J Epidemiol Community Health (2002) 56(10):773–779.[Abstract/Free Full Text]
  7. Erbas B, Kelly AM, Physick B, et al. Air pollution and childhood asthma emergency hospital admissions: estimating intra-city regional variations. Int J Environ Health Res. (2005) 15(1):11–20.[CrossRef][Web of Science][Medline]
  8. Ballester F, Rodríguez P, Iñíguez C, et al. Air pollution and cardiovascular admissions association in Spain: results within the EMECAS project. J Epidemiol Community Health (2006) 60(4):328–336.[Abstract/Free Full Text]
  9. Dominici F, Peng RD, Bell ML, et al. Fine particulate air pollution and hospital admission for cardiovascular and respiratory diseases. JAMA (2006) 295(10):1127–1134.[Abstract/Free Full Text]
  10. Peng RD, Dominici F, Pastor-Barriuso R, et al. Seasonal analyses of air pollution and mortality in 100 US cities. Am J Epidemiol (2005) 161(6):585–594.[Abstract/Free Full Text]
  11. Katsouyanni K, Touloumi G, Spix C, et al. Short-term effects of ambient sulphur dioxide and particulate matter on mortality in 12 European cities: results from time series data from the APHEA project. Air Pollution and Health: a European Approach. BMJ (1997) 314(7095):1658–1663.[Abstract/Free Full Text]
  12. Samoli E, Schwartz J, Wojtyniak B, et al. Investigating regional differences in short-term effects of air pollution on daily mortality in the APHEA project: a sensitivity analysis for controlling long-term trends and seasonality. Environ Health Perspect (2001) 109(4):349–353.[Web of Science][Medline]
  13. Dominici F, McDermott A, Zeger SL, et al. National maps of the effects of particulate matter on mortality: exploring geographical variation. Environ Health Perspect (2003) 111(1):39–44.[Web of Science][Medline]
  14. Becker S, Dailey LA, Soukup JM, et al. Seasonal variations in air pollution particle-induced inflammatory mediator release and oxidative stress. Environ Health Perspect (2005) 113(8):1032–1038.[Web of Science][Medline]
  15. Dominici F, Peng RD, Zeger SL, et al. Dominici et al. respond to "Heterogeneity of particulate matter health risks" [letter]. Am J Epidemiol (2007) 166(8):892–893.[Free Full Text]
  16. Dominici F, Peng RD, Zeger SL, et al. Particulate air pollution and mortality in the United States: did the risks change from 1987 to 2000? Am J Epidemiol (2007) 166(8):880–888.[Abstract/Free Full Text]
  17. Tolbert PE. Invited commentary: heterogeneity of particulate matter health risks. Am J Epidemiol (2007) 166(8):889–891.[Abstract/Free Full Text]
  18. Bell ML, Dominici F, Ebisu K, et al. Spatial and temporal variation in PM2.5 chemical composition in the United States for health effects studies. Environ Health Perspect (2007) 115(7):989–995.[Medline]
  19. Peng RD, Chang HH, Bell ML, et al. Coarse particulate matter air pollution and hospital admissions for cardiovascular and respiratory diseases among Medicare patients. JAMA (2008) 299(18):2172–2179.[Abstract/Free Full Text]
  20. Everson P. Two-Level Normal Independent Sampling Estimation (TLNise) (2000) Swarthmore, PA: Swarthmore College.
  21. Gelman A, Carlin JB, Stern HS, et al. Bayesian Data Analysis (2003) New York, NY: Chapman & Hall.
  22. Lindley DV, Smith AFM. Bayes estimates for the linear model. J R Stat Soc (B) (1972) 34(1):1–41.
  23. DuMouchel WH, Harris JE. Bayes methods for combining the results of cancer studies in humans and other species. J Am Stat Assoc. (1983) 78(382):293–315.[CrossRef][Web of Science]
  24. Peng RD, Dominici F, Louis TA. Model choice in time series studies of air pollution and mortality. J R Stat Soc (A) (2006) 169(2):179–203.
  25. Welty LJ, Zeger SL. Are the acute effects of particulate matter on mortality in the National Morbidity, Mortality, and Air Pollution Study the result of inadequate control for weather and season? A sensitivity analysis using flexible distributed lag models. Am J Epidemiol (2005) 162(1):80–88.[Abstract/Free Full Text]
  26. Touloumi G, Samoli E, Pipikou M, et al. Seasonal confounding in air pollution and health time-series studies: effect on air pollution effect estimates. Stat Med (2006) 25(24):4164–4178.[CrossRef][Web of Science][Medline]
  27. Samet JM, Zeger SL, Dominici F, et al. The National Morbidity, Mortality, and Air Pollution Study, Part II: Morbidity and Mortality From Air Pollution in the United States (2000) Cambridge, MA: Health Effects Institute.
  28. NRC Committee on Research Priorities for Airborne Particulate Matter, National Research Council, Board on Environmental Studies and Toxicology. Research Priorities for Airborne Particulate Matter IV—Continuing Research Progress (2004) Washington, DC: National Academy Press.
  29. Chuang KJ, Chan CC, Su TC, et al. Associations between particulate sulfate and organic carbon exposures and heart rate variability in patients with or at risk for cardiovascular diseases. J Occup Environ Med (2007) 49(6):610–617.[CrossRef][Web of Science][Medline]
  30. Ostro B, Feng WY, Broadwin R, et al. The effects of components of fine particulate air pollution on mortality in California: results from CALFINE. Environ Health Perspect (2007) 115(1):13–19.[Web of Science][Medline]
  31. Mar TF, Norris GA, Koenig JQ, et al. Associations between air pollution and mortality in Phoenix, 1995–1997. Environ Health Perspect (2000) 108(4):347–353.[Web of Science][Medline]
  32. 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.[Medline]
  33. Laden F, Neas LM, Dockery DW, et al. Association of fine particulate matter from different sources with daily mortality in six U.S. cities. Environ Health Perspect (2000) 108(10):941–947.[Web of Science][Medline]
  34. Zeka A, Zanobetti A, Schwartz J. Short term effects of particulate matter on cause specific mortality: effects of lags and modification by city characteristics. Occup Environ Med (2005) 62(10):718–725.[Abstract/Free Full Text]
  35. Mueller SF, Bailey EM, Kelsoe JJ. Geographic sensitivity of fine particle mass to emissions of SO2 and NOx. Environ Sci Technol (2004) 38(2):570–580.[Medline]
  36. Dutkiewicz VA, Qureshi S, Khan AR, et al. Sources of fine particulate sulfate in New York. Atmos Environ (2004) 38(20):3179–3189.
  37. Grahame T, Hidy G. Using factor analysis to attribute health impacts to particulate pollution sources. Inhal Toxicol (2004) 16(suppl 1):143–152.[CrossRef][Web of Science][Medline]
  38. Pickle LW, Mungiole M, Jones GK, et al. Atlas of United States Mortality (1999) Hyattsville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics.
  39. Medina-Ramón M, Zanobetti A, Schwartz J. The effect of ozone and PM10 on hospital admissions for pneumonia and chronic obstructive pulmonary disease: a national multicity study. Am J Epidemiol (2006) 163(6):579–588.[Abstract/Free Full Text]
  40. Janssen NAH, Schwartz J, Zanobetti A, et al. Air conditioning and source-specific particles as modifiers of the effect of PM10 on hospital admissions for heart and lung disease. Environ Health Perspect (2002) 110(1):43–49.[Web of Science][Medline]
  41. Sarnat JA, Brown KW, Schwartz J, et al. Ambient gas concentrations and personal particulate matter exposures: implications for studying the health effects of particles. Epidemiology (2005) 16(3):385–395.[CrossRef][Web of Science][Medline]
  42. Koutrakis P, Suh HH, Sarnat JA, et al. Characterization of particulate and gas exposures of sensitive subpopulations living in Baltimore and Boston. Res Rep Health Eff Inst (2005) 131:1–65. discussion 67–75.[Medline]
  43. Sarnat JA, Koutrakis P, Suh HH. Assessing the relationship between personal particulate and gaseous exposures of senior citizens living in Baltimore, MD. J Air Waste Manag Assoc. (2000) 50(7):1184–1198.[Web of Science][Medline]
  44. Rodes CE, Lawless PA, Evans GF, et al. The relationships between personal PM exposures for elderly populations and indoor and outdoor concentrations for three retirement center scenarios. J Expo Anal Environ Epidemiol (2001) 11(2):103–115.[CrossRef][Web of Science][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Am. J. Respir. Crit. Care Med.Home page
M. L. Bell, K. Ebisu, R. D. Peng, J. M. Samet, and F. Dominici
Hospital Admissions and Chemical Composition of Fine Particle Air Pollution
Am. J. Respir. Crit. Care Med., June 15, 2009; 179(12): 1115 - 1120.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplemental Material
Right arrow All Versions of this Article:
168/11/1301    most recent
kwn252v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Bell, M. L.
Right arrow Articles by Dominici, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bell, M. L.
Right arrow Articles by Dominici, F.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?