Secondary Logo

Ambient Gas Concentrations and Personal Particulate Matter Exposures

Implications for Studying the Health Effects of Particles

Sarnat, Jeremy A.*; Brown, Kathleen W.; Schwartz, Joel; Coull, Brent A.§; Koutrakis, Petros

doi: 10.1097/01.ede.0000155505.04775.33
Original Article
Free
SDC

Background: Data from a previous study conducted in Baltimore, MD, showed that ambient fine particulate matter less than 2.5 μm in diameter (PM2.5) concentrations were strongly correlated with corresponding personal PM2.5 exposures, whereas ambient O3, NO2, and SO2 concentrations were weakly correlated with their personal exposures to these gases. In contrast, many of the ambient gas concentrations were reasonable surrogates of personal PM2.5 exposures.

Methods: Personal multipollutant exposures and corresponding ambient air pollution concentrations were measured for 43 subjects living in Boston, MA. The cohort consisted of 20 healthy senior citizens and 23 schoolchildren. Simultaneous 24-hour integrated PM2.5, O3, NO2, and SO2 personal exposures and ambient concentrations were measured. All PM2.5 samples were also analyzed for SO42− (sulfate). We analyzed personal exposure and ambient concentration data using correlation and mixed model regression analyses to examine relationships among (1) ambient PM2.5 concentrations and corresponding ambient gas concentrations; (2) ambient PM2.5 and gas concentrations and their respective personal exposures; (3) ambient gas concentrations and corresponding personal PM2.5 exposures; and (4) personal PM2.5 exposures and corresponding personal gas exposures.

Results: We found substantial correlations between ambient PM2.5 concentrations and corresponding personal exposures over the course of time. Additionally, our results support the earlier finding that summertime gaseous pollutant concentrations may be better surrogates of personal PM2.5 exposures (especially personal exposures to PM2.5 of ambient origin) than they are surrogates of personal exposures to the gases themselves.

Conclusions: Particle health effects studies that include both ambient PM2.5 and gaseous concentrations as independent variables must be analyzed carefully and interpreted cautiously, since both parameters may be serving as surrogates for PM2.5 exposures.

From the *Department of Environmental and Occupational Health, Rollins School of Public Health of Emory University, Atlanta, GA; and the †Department of Environmental Health; the ‡Department of Epidemiology; and the §Department of Biostatistics, Harvard School of Public Health, Boston, Massachusetts.

Submitted 22 January 2004; final version accepted 2 December 2004.

Supported by the Health Effects Institute (Agreement #98–7) and the Harvard-EPA Center on Particle Health Effects (STAR grant #: R827353-01-0).

Correspondence: Jeremy A. Sarnat, Department of Environmental & Occupational Health, Rollins School of Public Health of Emory University, 1518 Clifton Road, N.E.—Room 214, Atlanta, GA 30322 E-mail: jsarnat@sph.emory.edu.

Numerous epidemiologic studies have found associations between ambient fine particulate matter less than 2.5 μm in diameter (PM2.5) concentrations and adverse health outcomes.1 Determining the relative contribution of PM2.5 exposure to a specific morbidity or mortality outcome is challenging because ambient concentrations of PM2.5 and its gaseous copollutants frequently are correlated with each other. This reasoning has led to suggestions that estimates of health risk associated with PM2.5 may be, in fact, confounded by gaseous copollutants.2 Correlation among ambient concentrations is a precondition for confounding because a gaseous pollutant can be a confounder of ambient particles only if it is correlated with both the exposure of interest (ie, PM2.5) and the health outcome of interest. Investigation of the potential confounding effect of gaseous copollutants on PM health effects has become one of the research priorities of the National Research Council.3

Many researchers have examined the potential for confounding using models that include ambient PM2.5 as well as one or more ambient concentrations of ozone (O3), nitrogen dioxide (NO2), and sulfur dioxide (SO2).4,5 These models are an appropriate methodology if the multiple pollutant measures in the models are actual personal exposures. However, because the measures used typically are ambient concentrations these models rely on 2 assumptions: (1) that the ambient pollutant concentrations in the model are good surrogates (ie, are well correlated) with their respective personal exposures and poorly correlated with personal exposures to other pollutants; and (2) that the inclusion of gaseous pollutants along with PM2.5 in the models will capture both the independent PM2.5 and gas effects while effectively controlling for any residual confounding. These previous investigations of copollutant confounding used ambient pollutant concentrations as the model inputs because information on personal exposures to the multiple air pollutants was not available. The recent development of a multiple-pollutant sampler,6 however, has enabled simultaneous measurements of PM2.5 as well as O3, NO2 and SO2, providing researchers with an opportunity to examine confounding using personal exposures. In addition, because indoor sulfate (SO42−) sources are generally limited, SO42− measurements from the PM2.5 sampler filters provide a means of assessing exposures to particles from ambient origins.

In a previous work,7 we examined associations among multiple pollutant exposures and ambient concentrations in Baltimore, MD. The Baltimore cohort consisted of 20 healthy senior adults, 21 schoolchildren, and 15 individuals with physician-diagnosed chronic obstructive pulmonary disease (COPD). Twenty-four hour ambient concentrations and personal exposure data were collected for PM2.5, O3, NO2, and SO2 during 12 consecutive days for each subject. Ambient PM2.5 and gaseous concentrations were strongly related. However, ambient concentrations of O3, NO2, and SO2 were poor surrogates of their respective personal exposures. Associations between personal PM2.5 exposures and corresponding personal gaseous exposures also were weak. Together, these results suggested that confounding of PM2.5-related health effects by these gaseous pollutants may not be likely for this location.

The results from the Baltimore study may not be generalizable to individuals living in different locations, residing in homes with varying building characteristics, or those exposed to different mixtures and levels of air pollutants. In this article, we further evaluate the role of ambient O3, NO2, SO2, and carbon monoxide (CO) as surrogates of personal PM2.5 exposures and personal exposures to the fraction of PM2.5 that is of ambient origin, using data from an additional exposure assessment study conducted in Boston, MA.

Back to Top | Article Outline

METHODS

Personal multipollutant exposures and corresponding ambient concentrations were measured for 43 subjects living in Boston, MA, during the summer of 1999 and the winter of 2000. The Boston cohort consisted of 20 healthy senior citizens and 23 schoolchildren. All subjects included in this analysis were nonsmokers and lived in residences with nonsmokers (ie, either single-family houses or apartments). Subject selection was not random and was not intended to be representative of sensitive populations in general. Cohort-specific analyses were not conducted because of the relatively small sample size for each cohort.

All subjects were measured for 12 consecutive 24-hour periods (“sampling session”) in each of the 1 or 2 seasons. Subjects were measured in sampling sessions consisting of 10 subjects per group. There were 3 sessions during each season. A total of 714 person-days of exposure data was collected for the following pollutants: PM2.5, O3, NO2, and SO2. All PM2.5 filters were analyzed for SO42− concentration by extracting the PM2.5 filters and analyzing the aqueous extract by ion chromatography. Personal SO42− exposures were used as indicators of personal exposure to PM2.5 of ambient origin.8

Personal exposure samples were collected using Multipollutant Personal Environmental Monitors to collect PM2.5, O3, NO2, and SO2 samples.6 Monitors were affixed to the shoulder strap of a backpack to correspond to the breathing zone of each subject. Subjects were instructed to remove the pack during activities when the sampler could be damaged or during prolonged periods of inactivity. During periods when the sampler was removed from the subject's body, subjects were instructed to keep the sampling inlets as close as possible to their breathing zone. Particles were collected using Personal Environmental Monitors on 37-mm Teflon filters (37-mm Teflo [Gelman Sciences, Ann Arbor, MI]). O3, NO2, and SO2 concentrations were measured using passive badge samplers. (No accurate personal CO monitors exist.)

Twenty-four hour integrated ambient PM2.5 concentrations were measured using Harvard Impactors at the Harvard School of Public Health. For 4 days when integrated data were unavailable, continuous PM2.5 data were used. Continuous ambient PM2.5 mass concentrations were obtained from PM2.5 TEOMs (Model 1400A [Rupprecht & Patashnick, Co., Inc., East Greenbush, NY]) operated at the Harvard School of Public Health. Ambient O3, NO2, SO2, and CO data were obtained from local stationary ambient monitoring sites operated by the Massachusetts Department of Environmental Protection using UV photometric analyzers, chemiluminescence monitors, pulsed fluorescent monitors, and nondispersive infrared monitors, respectively. In cases in which pollutant concentrations were measured at multiple sites, concentrations were averaged across the sites.

Standard Quality Assurance/Quality Control procedures were followed.9 We assessed data for bias, precision, and completeness as detailed in Table 1.

TABLE 1

TABLE 1

Back to Top | Article Outline

Data Analysis

Concentrations for PM2.5 and SO42− are reported in μg/m3. Concentrations for O3, NO2, and SO2, are reported in ppb. Negative pollutant concentration values as well as values below the level of detection were included in the data analyses as measured to reduce the potential for bias in estimating relations among measurements.10

The following 4 models were used to assess the relationship between PM2.5 and its copollutants:

  • Model 1: Associations between ambient PM2.5 concentrations and ambient gaseous concentrations (“ambient-ambient associations”);
  • Model 2: Associations between ambient pollutant (PM2.5, SO42−, and gas) concentrations and their respective personal exposures (“personal-ambient associations”);
  • Model 3: Associations between ambient gaseous concentrations and personal PM2.5 exposures, including SO42− as a surrogate of personal exposure to PM2.5 of ambient origin (“cross-pollutant associations”); and
  • Model 4: Associations between personal PM2.5 exposures, including personal exposure to PM2.5 of ambient origin, and personal gaseous exposures (“personal-personal associations”).

We examined ambient-ambient (Model 1) associations using time-series regression analysis assuming a first-order autoregressive structure for the error. We conducted mixed model regression analysis for models involving personal exposures (Models 2–4) to (1) model covariance among the pollutants and (2) pool information across individuals while accounting for repeated measures on the same individual. A generalized form of the mixed model used for these analyses can be expressed as:

where Yijl represents an observed personal particle or gas exposure for subject i on day j in sampling session l; α represents the regression intercept; Xijl represents either an observed personal gas exposure or ambient particle or gas concentration for subject i on day j in sampling session l; β represents the fixed effect of X on Y; bil represents the random subject effect ∼ N (0,σ2bl), and εijl represents the random error ∼ N (0,σ2el). The l index in the variance components denotes that fact that the models allow for different between and within-subject error in the different sampling sessions. Since gas stoves constitute a potential indoor source of both NO2 and PM2.5, mixed models including a gas stove interaction term were also used to assess whether associations differed between subjects with gas and with electric stoves.

All of the mixed models were fitted using a compound symmetry (ie, random intercepts) covariance matrix, with a nested panel variable. This covariance structure yielded the lowest Akaike Information Criteria diagnostic values compared with other covariance matrices examined. The strength of association was assessed by the size and significance of the estimated slope of the mixed models.

In addition to the time-series and mixed model regression analyses, subject-specific Spearman's correlation coefficient (rs) values are reported as a secondary indicator of the strength of correlation in the observed relationships by individual. Subjects having fewer than 7 valid person-days of observations (of 12) were excluded from the analysis. The distribution of the rs values is presented.

Time-activity data indicated that several subjects were exposed to heavy or prolonged exposure to environmental tobacco smoke during sampling. A total of 37 person-days (approximately 5% of total) were therefore excluded from analyses (from 1 senior subject during the summer and winter and 1 child during the winter).

Back to Top | Article Outline

RESULTS

During each of the sampling seasons, personal exposures to O3, NO2, and SO2 were generally low, frequently below their level of detection. During the winter, in particular, subjects were exposed to extremely low levels of O3 and SO2. Wintertime O3 and SO2 personal exposures were below levels of detection more than 95% of the time, even when corresponding ambient levels were well above the level of detection. In contrast to the personal gas exposures, none of the personal PM2.5 exposures were below detectable levels during either season. Personal PM2.5 exposures were comparable with and often exceeded corresponding ambient levels. Given the extremely low personal O3 exposures during the winter and the low personal SO2 associations during both seasons, our presentation of results from the mixed model and correlation analyses focuses on the associations involving PM2.5, SO42−, NO2 and summertime O3. Because these personal exposures were almost exclusively distributed below their levels of detection, observed results are likely driven by random noise. We provide complete results from the quantitative analyses in the tables and figures to enable comparison with previously published findings.7 We present complete summary statistics for the measured ambient concentrations and personal exposures, stratified by season, panel, and cohort in Table 2.

TABLE 2

TABLE 2

Back to Top | Article Outline

Model 1: Ambient–Ambient Associations

Ambient PM2.5 concentrations were associated with corresponding ambient concentrations of several gaseous copollutants, although the strength and direction for several of these associations differed by season (Table 3). Ambient O3 was positively associated with ambient PM2.5 during the summer (slope = 0.51; 95% confidence interval [CI] = 0.34–0.68), whereas negatively associated with ambient PM2.5 during the winter (−0.53; −0.85 to −0.22). Associations between ambient PM2.5 and NO2 were positive during both seasons and more strongly so in winter (summer: 0.44; −0.04 to 0.92; winter: 0.64; 0.47–0.82). Ambient levels of SO2 were associated with ambient PM2.5 during the winter (0.80; 0.23–1.37), with little association observed during the summer. Finally, ambient PM2.5 was positively associated with ambient CO during both seasons (summer: 33.7; 6.8–45.1; winter: 24.4; 17.9–30.9).

TABLE 3

TABLE 3

TABLE 3

TABLE 3

Back to Top | Article Outline

Model 2: Personal–Ambient Associations

Ambient PM2.5 concentrations were strongly associated with corresponding personal PM2.5 exposures (summer: 0.77; 95% CI = 0.65–0.89; winter: 0.33; 0.13–0.53; Table 3). The median subject-specific rs was 0.56 during the summer and 0.38 during the winter (Fig. 1). Stronger and less variable personal-ambient associations were found for SO42−, a component of PM2.5 with few indoor sources (winter: 0.55; 95% CI = 0.49–0.61; summer: 0.74; 0.67–0.80; Table 3).11 Subject-specific personal-ambient SO42− correlations were greater than 0.8 for 25 of 29 subjects during the summer and 15 of 28 subjects during the winter. The median rs was 0.86 during the summer and 0.82 during the winter (Fig. 1).

FIGURE 1.

FIGURE 1.

Ambient O3 and ambient NO2 during the summer were modestly associated with their corresponding personal exposures (0.27; 95% CI = 0.18–0.37 and 0.19; 0.08–0.30). Eight subjects (of 29) had personal-ambient O3 correlations greater than 0.8 during the summer. In contrast, no association was found between personal and ambient SO2 during either season (summer: 0.00; 95% CI = −0.11 to 0.10; winter SO2: −0.02; −0.04 to 0.00). A gas stove in a subject's residence did not influence the association between ambient NO2 concentrations and corresponding personal NO2 exposures in either season (data not shown).

Back to Top | Article Outline

Model 3: Cross-Pollutant Associations

The ambient gas concentrations were each associated with personal PM2.5 during the summer but not winter (Table 3), which was consistent with the stronger summer associations between ambient PM2.5 and the ambient gases. The direction of the associations between personal PM2.5 and the ambient gas concentrations mirrored those of the corresponding ambient associations between PM2.5 and the gases. The associations between most of the ambient gas concentrations and personal SO42− exposures was stronger than that found for personal PM2.5 (Table 3). Subject-specific correlation coefficients between the ambient O3, NO2, and SO2 concentrations and personal PM2.5 exposures showed considerable variation by subject, pollutant and season (Fig. 2).

FIGURE 2.

FIGURE 2.

Models examining the opposite cross correlations (ie, ambient PM2.5 concentrations with personal gas exposures) showed that ambient PM2.5 was associated with personal O3 during the summer and with personal NO2 exposures during the winter (PM2.5 − O3 slope = 0.27; 95% CI = 0.14–0.39; PM2.5 − NO2 slope = 0.21; 0.10–0.33; Table 3).

Back to Top | Article Outline

Model 4: Personal–Personal Associations

We found associations between personal PM2.5 and personal O3 during the summer and between personal PM2.5 and personal NO2 during both seasons (Table 3). Weaker personal-personal associations were found for models using personal SO42− instead of personal PM2.5. Most of the166 subject-specific personal–personal rs values showed little or no correlation as presented in Figure 3.

FIGURE 3.

FIGURE 3.

We compared results from the Boston study with the earlier Baltimore7 study. A schematic diagram summarizes the results from both cities for associations involving PM2.5 and SO42− (Fig. 4). Similarly, Table 3 indicates differences between Boston and Baltimore in models combining data from both cities and including a city interaction term.

FIGURE 4.

FIGURE 4.

Back to Top | Article Outline

DISCUSSION

The results show that associations between measurements of ambient particulate and gaseous pollutants with their corresponding personal exposures vary by pollutant and individual. Given the correlation patterns among ambient pollutant concentrations, the findings also showed that variability in the levels of one ambient pollutant can be associated with variability in numerous personal pollutant exposures. It is important to verify the generalizability of these findings, since the implications of these results may have key implications for interpreting air pollution epidemiologic results. Our discussion focuses on similarities and differences between the results of this study with those found in a previous analysis of data collected in Baltimore.7

Back to Top | Article Outline

Similarities Between Boston and Baltimore

Subjects from Boston and Baltimore7 generally were exposed to very low levels of the gases over an integrated 24-hour period, particularly O3 during the winter and SO2 during the summer. These personal exposures were low in both cities even when corresponding ambient levels of these pollutants were high. For epidemiologic studies using 24–hour integrated measurements, it is unlikely that a gaseous pollutant will serve as a confounder of ambient particles if actual exposure to that pollutant is negligible. For example, during the wintertime sampling in Boston, a 6.1 ppb increase in ambient O3, or one standard deviation, was associated with corresponding increases of 0.2 ppb in personal exposure to O3. Therefore, inclusion of these ambient concentrations of these pollutants in multivariate models is questionable, since changes in the observed ambient concentrations may not be associated with exposures that are of biologic significance.

Likewise, in both Baltimore and Boston, ambient PM2.5 was correlated with many of the ambient gases in the summer and winter. Correlations among pollutants are common throughout many parts of the United States and are largely attributable to the similar impact of meteorology on the transport and removal of these ambient pollutants.12,13 Identifying the relative importance of these pollutants for observed mortality and morbidity outcomes is made more difficult by the presence of collinearity among ambient pollutant concentrations.

Ambient concentrations of PM2.5 also were associated with personal exposures for subjects from both cities during both seasons, despite the presence of nonambient sources of PM2.5. These findings are consistent with results from other recent longitudinal studies of PM personal exposure.14–16 Moreover, associations between personal and ambient SO42− were substantially stronger than those found between personal and ambient PM2.5 in both cities.

A key finding from Baltimore7 was the association between the ambient gas concentrations and personal PM2.5 and SO42− exposures. This association was also found in Boston, particularly with SO42−. In both cites, the associations between the ambient gas concentrations and personal SO42− exposures were generally stronger than those observed for personal PM2.5. These results are expected given the lack of nonambient SO42− contributions to personal exposures that could introduce noise into the cross-pollutant relationships. Conversely, the presence of nonambient contributions to personal PM2.5 exposures likely produced weaker cross-pollutant associations involving PM2.5.

Back to Top | Article Outline

Differences Between Boston and Baltimore

Despite these similarities, there are several key differences between the results from Boston and Baltimore.7 Most important were the moderately strong personal-ambient gaseous associations in Boston, indicating that changes over time for some gaseous pollutants measured at central sites did, in fact, reflect corresponding changes in the personal exposures for our subjects. This was not the case in Baltimore, where ambient gas concentrations were not associated with their respective personal exposures. The ambient gas measurements in both cities were more strongly associated with personal exposure to PM2.5 than with their respective personal exposures, as evidenced from the t-statistics associated with the observed mixed model slopes.

For personal-ambient O3 associations, the city-specific discrepancy in the results may be attributable to differences in ventilation. Previous studies have shown that the degree of indoor ventilation can influence personal-ambient associations for several pollutants including O3 and PM2.5.9,17,18 Although no quantitative measures of ventilation were measured in either Boston or Baltimore, it is likely that average air exchange rates for the relatively older, leakier homes of Boston subjects were higher than those in Baltimore, where many of the subjects lived in apartment complexes with central air conditioning.

In Boston, personal PM2.5 was associated with personal O3 during the summer and with personal NO2 during both seasons, which is another finding not seen in Baltimore. Correlation among personal PM2.5 and NO2 exposures may be induced by common indoor sources such as gas stoves, which emit both PM2.5 and NO2.19

Back to Top | Article Outline

Implications

If gaseous pollutant exposures are responsible, in part, for the observed health effects, then ambient gas concentrations should be included along with PM2.5 in multivariate epidemiologic models to control for the potential confounding effect of these gases. However, the results from these studies indicate that the summertime associations between the ambient measurements of the gases may be more strongly associated to personal exposure to PM2.5 and SO42−, than with their respective exposures. The Boston results suggest there may be differences, by location in the strength of the personal-ambient association for the gases. These differences are likely attributable to geographical variability in housing characteristics such as ventilation.

Given these findings, one should be cautious when interpreting results from time-series epidemiologic studies that include both gaseous and particulate pollutant concentrations in the models. Numerous epidemiologic studies, for example, have shown associations between ambient SO2 concentrations, measured at central monitoring sites, and adverse health outcomes including hospital admissions and mortality.20–23 To date, there have been no exposure studies showing ambient SO2 to be a suitable surrogate of personal SO2 exposures. This issue has been addressed previously by several researchers.20,21 Katsouyanni et al21 speculate that associations of ambient SO2 with excess mortality in 12 European cities may be attributable to SO2 serving as a “surrogate of other substances.” The authors point out that SO2 is highly reactive with a short indoor half-life. These 2 factors would likely result in weak personal-ambient associations for this pollutant.

Our results from Boston showed that summertime O3 and NO2 concentrations are modestly associated with their corresponding personal exposures. These findings suggest that it is incorrect to assume that ambient gas measurements are consistent surrogates for PM exposures. Indeed, controlled exposure studies have shown that O3, NO2, and SO2 exposures can have adverse health effects.24–27 Likewise, environmental and occupational exposure assessment studies have also reported higher mean personal O3, NO2, and SO2 exposures for some individuals and cohorts as compared with the levels reported in the current study.28–30

In addition, the Boston results show that for some pollutants during specific seasons, ambient PM2.5 also was associated with personal O3 and NO2 exposures. Although the strength of these cross-pollutant associations was not as strong as between ambient PM2.5 and personal PM2.5, the findings suggest that ambient PM2.5 also can serve as a surrogate for exposures to pollutants other than their respective exposures.

Back to Top | Article Outline

Potential Limitations of This Analysis Method

Numerous limitations exist to the analyses conducted for both the current Boston dataset and the previously analyzed data from Baltimore.7 One possible explanation for the location-specific differences in the personal-ambient associations may be the fact that the personal gaseous sampler measured exposures with greater imprecision compared with the PM2.5 measurements, resulting in greater random noise in the associations involving the gases. Estimates of precision for the gas measurements during the Boston study indicated that they were as precise, and in some cases more precise, than the PM2.5 measurements. However, these estimates were derived from measurements conducted in conditions with generally higher gaseous levels (as indicated by the reference mean values in Table 1) than those observed during actual subject sampling.

Second, our results are for a small, nonrandom selection of subjects living in the eastern United States. Caution should be exercised, therefore, in generalizing the results to other locations and cohorts. The Boston results provide some indication that the relationships between personal and ambient concentrations and among the pollutants may be influenced by either differences in ventilation or by the amount of time the subjects spent outdoors. Thus, results are likely to vary for locations where average ventilation rates differ or where subjects tend to spend more time outdoors.

Our objective was to examine the potential confounding effect of gaseous pollutants within the context of commonly used 24-hour time-series analysis studies investigating associations between daily PM2.5 concentrations and mortality and morbidity outcomes. It is possible that these patterns of association may differ for exposure periods either longer or shorter than 24 hours. Of particular interest will be studies that examine pollutant associations for short peaks of exposure which may be more relevant for understanding O3 health effects.31,32

Despite these limitations, the results from Boston provide an indication from another location in the eastern United States that ambient gas concentrations may be more strongly associated with exposures to PM2.5 than with their respective personal exposures. These estimates can potentially distort estimates of health effects due to specific pollutants. Results from other regions and future improvements in personal sampler design that provide the ability to characterize exposures to pollutants at shorter time intervals will provide further insights.

Back to Top | Article Outline

ACKNOWLEDGMENTS

We thank the participants of this study.

Back to Top | Article Outline

REFERENCES

1. Pope CA. Review: epidemiological basis for particulate air pollution health standards. J Aerosol Sci. 2000;32:4–14.
2. Green LC, Crouch EAC, Ames MR, et al. What's wrong with the National Ambient Air Quality Standard (NAAQS) for fine particulate matter (PM2. 5)? Regul Toxicol Pharm. 2002;35:327–337.
3. National Research Council. Research Priorities for Airborne Particulate Matter. I. Immediate Priorities and a Long-Range Research Portfolio. Washington, DC: National Academy Press; 1998.
4. Burnett RT, Smith-Doiron M, Stieb D, et al. Effects of particulate and gaseous air pollution on cardiorespiratory hospitalizations. Arch Environ Health. 1999;54:130–139.
5. Samet JM, Dominici F, Curriero FC, et al. Fine particulate air pollution and mortality in 20 U.S. cities, 1987–1994. N Engl J Med. 2000;343:1742–1749.
6. Demokritou P, Kavouras IG, Ferguson ST, et al. Development and laboratory performance evaluation of a multipollutant sampler for simultaneous measurements for particulate and gaseous pollutants. Aerosol Sci Tech. 2001;35:741–752.
7. Sarnat JA, Schwartz J, Catalano PJ, et al. The role of gaseous pollutants in particulate matter epidemiology: confounder or surrogate? Environ Health Perspect. 2001;109:1053–1062.
8. Sarnat JA, Long CM, Koutrakis P, et al. Using sulfur as a tracer of outdoor fine particulate matter. Environ Sci Technol. 2002;36:5305–5314.
9. 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;50:1184–1198.
10. McBean EA, Rovers FA. Statistical Procedures for Analysis of Environmental Monitoring Data & Risk Assessment. Upper Saddle River, NJ: Prentice Hall PTR; 1998.
11. Mage D, Wilson W, Hasselblad V, et al. Assessment of human exposure to ambient particulate matter. J Air Waste Manage Assoc. 1999;49:174–185.
12. Chen C, Chock DP, Winkler SL. A simulation study of confounding in generalized linear models for air pollution epidemiology. Environ Health Perspect. 1999;107:217–222.
13. Wallace L. Correlations of personal exposure to particles with outdoor air measurements: a review of recent studies. J Aerosol Sci. 2000;32:15–25.
14. Ebelt ST, Petkau AJ, Vedal S, et al. Exposure of chronic obstructive pulmonary disease patients to particulate matter: relationships between personal and ambient air concentrations. J Air Waste Manage Assoc. 2000;50:1081–1094.
15. Janssen NAH, Hoek G, Harssema H, et al. Childhood exposure to PM10: relation between personal, classroom, and outdoor concentrations. J Occup Environ Med. 1997;54:888–894.
16. Williams R, Creason J, Zweidinger R, et al. Indoor, outdoor, and personal exposure monitoring of particulate air pollution: the Baltimore elderly epidemiology-exposure pilot study. Atmos Environ. 2000;34:4193–4204.
17. Brauer M, Brook JR. Ozone personal exposures and health effects for selected groups residing in the Fraser Valley. Atmos Environ. 1997;31:2113–2121.
18. Rojas-Bracho L, Suh HH, Koutrakis P. Relationships among personal, indoor, and outdoor fine and coarse particle concentrations for individuals with COPD. J Expo Anal Environ Epidemiol. 2000;10:294–306.
19. Levy JI. Impact of residential nitrogen dioxide exposure on personal exposure: an international study. J Air Waste Manag Assoc. 1998;48:553–560.
20. Schwartz J, Dockery DW, Neas LM, et al. Acute effects of summer air pollution on respiratory symptom reporting in children. Am J Resp Crit Care. 1994;150:1234–1242.
21. 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. Br Med J. 1997;314:7095.
22. Krewski D, Burnett RT, Goldberg MS, et al. Reanalysis of the Harvard Six Cities Study and the American Cancer Society Study of Particulate Air Pollution and Mortality. Cambridge, MA: Health Effects Institute; 2000.
23. 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;109:349–353.
24. Devlin RB, Folinsbee LJ, Biscardi F, et al. Inflammation and cell damage induced by repeated exposure of humans to ozone. Inhal Toxicol. 1997;9:211–235.
25. Frampton MW, Morrow PE, Gibb FR, et al. Effects of nitrogen dioxide exposure on pulmonary function and airway reactivity in normal humans. Am Rev Respir Dis. 1991;143:522–527.
26. Gong HJ, Wong R, Sarma RJ, et al. Cardiovascular effects of ozone exposure in human volunteers. Am J Respir Crit Care. 1998;158:538–546.
27. Sheppard D, Saisho A, Nadel JA, et al. Exercise increases sulfur dioxide-induced bronchoconstriction in asthmatic subjects. Am Rev Respir Dis. 1981;123:486–491.
28. Mosqueron L, Momas I, Le Moullec Y. Personal exposure of Paris office workers to nitrogen dioxide and fine particles. Occ Envir Med. 2002;59:550–556.
29. Geyh AS, Xue JP, Ozkaynak H, et al. The Harvard Southern California chronic ozone exposure study: assessing ozone exposure of grade-school-age children in two Southern California communities. Environ Health Perspect. 2000;108:265–270.
30. Lee WJ, Teschke K, Kauppinen T, et al. Mortality from lung cancer in workers exposed to sulfur dioxide in the pulp and paper industry. Environ Health Perspect. 2002;110:991–995.
31. Lippmann M. Health-effects of ozone—a critical review. JAPCA. 1989;39:672–695.
32. Thurston GD, Lippmann M, Scott MB, et al. Summertime haze air pollution and children with asthma. Am J Respir Crit Care. 1997;155:654–660.
Copyright © 2005 Wolters Kluwer Health, Inc. All rights reserved.