Particulate matter components and health: a literature review on exposure assessment

Introduction

Ambient particulate matter (PM) is a mix of heterogeneous components varies greatly by season and by region (1,2). It consists of various components, which includes organic carbon (OC), elemental carbon (EC), nitrate, sulfate and trace elements (e.g., iron, vanadium, nickel). These chemical components are derived from various anthropogenic and natural sources, such as soil or road dust, vehicle exhaust, biomass combustion, sea salt, forest fires, and may have great spatiotemporal variations associated with particular regional and local sources (3).

Since PM is a complex mixture, the toxicology of each chemical component may vary by time and locations. Multicity studies of PM show that associations between PM and morbidity and mortality vary across areas (4-6), with this variation attributed in part to chemical composition differences. Thus, it is still uncertain that whether certain PM components cause greater public health concern than others. Both National Academy of Sciences (NAS) and U.S. Environmental Protection Agency (EPA) have given prominence to the significance of investigating chemical components and characteristics of PM which contribute to their toxicity (7,8). Identifying most harmful PM components to human health could target control regulations more effectively so that it may reduce the burden of disease due to PM at a lower cost.

PM chemical components are usually measured by fixed monitoring stations which are collected and managed by government agencies on the base of regulatory purposes. For instance, in US, mainly two monitoring networks measure PM chemical components, which are the U.S. EPA Chemical Speciation Network (CSN) and the Interagency Monitoring of Protected Visual Environment (IMPROVE) sponsored by the U.S. EPA and other agencies (9-11). There are around 250 monitoring stations measuring PM components, for each single urban area less than 3 monitors on average, in contrast to 2,000 monitors for PM_2.5 mass across the whole country.

In Hong Kong, PM chemical speciation network was established by the Hong Kong Environmental Protection Department to measure twenty-six PM chemical components. Six general monitoring stations and one road side monitoring station were distributed across the whole area. The frequency of collecting samples was every six days. Thus, it is difficult to estimate the spatial and temporal variability based on the limited numbers of monitors.

PM components usually include OC, EC, ammonium ion (NH₄⁺), nitrate (NO₃⁻), sulfate (SO₄²⁻), bromide ion (Br^-), chloride ion (Cl⁻), sodium ion (Na⁺), potassium ion (K⁺), aluminium (Al), arsenic (As), beryllium (Be), barium (Ba), calcium (Ca), cadmium (Cd), chromium (Cr), copper (Cu), iron (Fe), mercury (Hg), magnesium (Mg), manganese (Mn), nickel (Ni), lead (Pb), selenium (Se), vanadium (V) and zinc (Zn). The common speciation methods and procedures for PM chemical components are summarized in Table 1. Due to the complexity of speciation methods and procedures, PM chemical components are commonly measured every 6 days for each monitoring stations (12).

The exposure assessment methods for PM components in epidemiological studies haven’t been reviewed. In this review, we summarize exposure assessment methods which have been applied in estimating the associations between PM components and adverse health effects in epidemiological studies.

Exposure assessment methods for short term exposure to PM components

Evidence have been accumulated on the association between PM components and adverse health effects on short term studies, like time series studies and case crossover studies (13,14). These studies rely on the day-to-day variations of the concentrations of PM components. Usually, associations with health effects were estimated using the exposure of the same day on which the outcome occurs or several days before the outcome occurs. Since the components were measured in every 3–6 days in general, it is very likely that during certain days there is no corresponding data for the PM components. Temporal imputation should be utilized to impute the missing data, while it is still challenging since the missing data might be substantially different from the observed data.

Exposure assessment methods for long term exposure to PM components

Table 2 summarized the exposure assessment methods for long term exposure to PM components. Several studies estimated the individual exposure level according to the monitoring concentrations close to their geocoded addresses. One study attempted to assess the association between PM_2.5 components and cardiopulmonary mortality in California (15,16), by assigning an concentration based on the nearest monitor according to the geocoded address. In this study, PM_2.5, Al, ammonium, Br, Ca, Cl, Cu, EC, Fe, Pb, Mn, Ni, OC, K, Si, Na, sulfate, Ti, nitrate, V and Zn were included. Particularly, different buffer distances were examined in this study. In addition, a study in California (17) explored the association between low birth weights and PM_2.5 constituents including EC and 35 metals, which applied the same exposure assessment method. In Florida, one study also tried to assess the risk of preeclampsia during pregnancy from exposure to PM components using the nearest monitoring station concentration as the surrogate for each individual. Another study attempted to explore the association between the source of fine particular matter and risk of preterm birth in Connecticut (18). BC, Zn, Pb, Cu, Br, Si, Fe, Ba, Ti, Mn, Al, Ca, K, V, Ni, Na, Cl, S were also included in the study. Exposure estimates were assigned to each woman based on the closest monitor of the women’s residence at time of birth. Source contributions to PM_2.5 mass was estimated using Positive Matrix Factorization (33). There are several advantages for Proximity method. First, this method is time-efficient for long term exposure assessment. In addition, models are derived from the monitoring station data, which have less uncertainty. Moreover, it is cost-effective to conduct compared to detailed exposure assessment which are often quite expensive. The distance based method is useful in preliminary studies when prior evidence is limited. However, this method also has limitations. First, it is very likely to lead to spatial misclassifications since only crude exposure is provided for each individual. This could lead to serious problem, when it comes to a large population and the monitoring stations are scarcely distributed. Second, the distance based method only takes the nearest fixed site monitors into account, fail to account the traffic related air pollution while commuting or indoor air pollution at schools or homes. Third, exposure misclassification tends to be smaller when restricted to small buffer radius, however, thus may lead to reduced statistical power. Fewer incidences may occur within smaller buffer which may cause imprecise estimation to some degree. Another potential issue is that urban areas tend to have more monitors than suburban or rural countries. Ambient monitoring stations are needed for more locations.

A number of studies use unweighted or distance-weighted average concentrations of several close monitors as the exposure concentrations for each individual. One study attempted to explore whether low birth weight was affected by airborne PM_2.5 chemical components (Al, NH₄, As, Cd, Ca, Cl, EC, Pb, Hg, Ni, nitrite, OCM, Si, Na, sulfate, Ti, V and Zn) in the northeastern and mid-Atlantic regions of United States (19). If one county has several monitors, the average concentration was calculated. However, different air pollutants were measured in different frequency. PM_2.5 components were measured every 3-6 days, while gaseous pollutants were measured daily. In order to avoid this problem, daily measured air pollutants and apparent temperature were incorporated to estimate weekly exposure. A study was conducted on women in Los Angeles, which tried to examine the risk of preterm birth when the women were exposed to high level of traffic-related particles, including PM₁₀ OC, PM₁₀ EC, PM_2.5 OC, PM_2.5 EC, Ammonium nitrate PM_2.5, Ammonium sulfate PM_2.5, Biomass burning PM_2.5, Diesel PM_2.5, Gasoline PM_2.5, Geological PM_2.5, Meat cooking PM_2.5, Residual Oil PM_2.5 and Sea salt PM_2.5 (20-22). For women who lived near 2 or 3 stations within 5 kilometers, the exposure values were calculated by distance-weighted method. The aforementioned model is a bit more accurate than the distance based model which incorporate weighted and unweighted concepts in the model. It is quite useful where the monitoring network is dense. Nevertheless, when the network is sparse, it may cause bias. Another limitation is that geographical patterns or topography have not been taken into account. In addition, outliers in monitoring data can affect the prediction. The third limitation is that some studies did not consider the addresses alterations during the study period. For birth cohort study, some studies only take into account the birth addresses not the actual or prior addresses for the mothers. The forth limitation is that this method does not deal with the spatial heterogeneity within a large single county, since they use a single monitoring station or average concentration as a surrogate for the whole county. In such circumstance, misclassification may occur for participants who reside far away from the monitoring stations. In addition, there is a challenge that the levels of some constituents are below the minimum detection limit which may cause exposure misclassification.

The Multi-Ethnic Study of Atherosclerosis, also known as the MESA study, utilized another approach to measure exposure to PM. They used partial least squares (PLS) and universal Kriging were used to build the national spatial exposure models (9). The aim of this study was to explore the relationship between low birth weight and exposure to particles by chemical composition (EC, OC, Si, S), size fraction and source. PLS, a kind of dimension deduction methods, was used as covariates in universal Kriging models. In addition, cross-validation was used to select the number of PLS components and assess the model’s prediction accuracy. Similarly, another study (23) conducted in California used empirical Bayesian Kriging methods to predict the concentrations of chemical composition (EC, OC, K, Cr, Fe, Ti, As, Mn, Cu, Ni, Pb), which could automatically calculate the parameters through a process of simulations comparing to other Kriging methods. The Kriging method was more accurate than the two aforementioned methods. The model will perform well with good accessibility of monitoring data, often in areas with dense monitoring stations compared to intricate spatial variability. On the contrary, when air pollution monitoring network is sparse, it might easily lead to the spatial misclassification, because large errors are exhibited which cannot represent the actual level of air pollutants. In addition, the ordinary Kriging methods assume that there is no global trend in the air pollution monitoring data, i.e., air pollutants spatial variability is only relied on the distance between monitoring stations.

Land use regression has been applied to predict the spatial variation of the annual mean concentration for PM₁₀, PM_2.5, NO_x, NO₂, VOCs and PAHs. Recently, ESCAPE has measured the PM constituents (Cu, Fe, K, Ni, S, Si, V, Zn) in 20 areas across Europe (34). Land use regression for PM components has been developed based on these measurements for these study areas. Predictors are extracted from Geographic Information Systems (GIS), including emission sources, traffic intensity, population density, land use etc. Several studies have been used this exposure assessment to evaluate the long term PM components and health effects (24-29). The health endpoints including all natural cause, cardiovascular mortality, lung cancer mortality, lung cancer function, respiratory health in birth cohort. In the USA, the Multi-Ethnic Study of Atherosclerosis (MESA) (35) also developed the LUR model to predict the spatial variability (PM_10-2.5 Mass, Cu, Zn, Si, P, endotoxin) for individuals within the cohorts, including fixed-site ambient monitors placed in density populated areas as well as rotating monitors. Both land use regression and universal Kriging were utilized to evaluate chemical components (36). Land use regression performs well with long term traffic-related air pollution and do not require detailed information on emission sources. However, land use regression fails to separate impacts of pollutants clearly, i.e. there may be large overlaps between the predictors for different PM components, since many of them share the same source. Moreover, both studies required additional monitoring campaigns based on the original measures. Such additional campaigns are time consuming and expensive, since it depends on a great number of monitors to develop the spatial-temporal models. In addition, land use regression does not provide much information on seasonal variability. If the topography and land use are quite different, the transferability is relatively low. One of the concerns in many land use regression is that the exposure measurement in land use regression model is developed at recent times while the cohort study might start in the past. So many studies back extrapolated the exposure concentrations to the baseline as well as the follow up period. Recent studies (37-40) have shown that nitrite dioxide (NO₂) could remain the same up to 10 years, which indicates that LUR model derived from current traffic related NO₂ could predict historical exposure well. These findings may be applied to traffic-related components, for example Zn, Fe and Cu.

The chemical transport model was developed from the air quality model which could predict source appointment to visibility reduction in California (41). The CTM model was developed to evaluate PM composition, particle size and source information, which incorporated information of wet and dry deposition, emissions and transport (30,31). Recently, a study conducted in Los Angeles assesses the PM components (O₃, NO, NO₂, CO, PM_2.5 mass EC, OC, nitrate) and low birth weight by using the CTM modelled data (23). CMAQ (Community Multiscale Air Quality) model incorporates information about emissions, meteorology and atmospheric chemistry to predict the spatial concentrations of PM_2.5 composition (30,42). A study conducted in Texas applied the Bayesian Spatial-temporal model using modelled CMAQ data to evaluate the associations between PM components and the risk of congenital anomalies (30). Four main components were included in this study nitrite, sulfate, OC and EC. The dispersion model can be applied to both short term and long term air pollution modelling. For limitations, the above-mentioned methods do not consider the effects of buildings during the exposure assessment. It often requires computing demands, which might be costly to run the model at detailed spatial scale. In addition, this method also demands detailed information of emission sources (for instance, road traffic and vehicle counts). If such information is not available, the model may not develop that well.

Recently, a Canadian team developed a method which combined satellite-derived aerosol optical depth and composition information from the global chemical transport model (32). The study evaluated secondary inorganic aerosol (SO₄²⁻, NH₄⁺ and NO₃⁻), BC and organic mass. The method which developed by Canadian team is the first assessment of long term exposure to fine PM constituents all over the world. One of the limitations is that trace metals were not evaluated, which serves as one of the important fine PM constituents that should be added when their assessment capability improves. There are still opportunities remaining to improve the modelled estimates. Finer satellite resolution and simulation would provide better intra-urban gradients.

Discussion

We reviewed different PM exposure assessment methods applied in epidemiological studies of adverse health effects of short-term and long-term exposure to PM components. For short term epidemiological study, the exposure largely relies on the day-to-day temporal variation of the pollutant concentrations. For long term epidemiological study, they include PM composition measurements by ambient fixed-site monitors (e.g., unweighted or distance-weighted methods), or incorporation of both measuring and modeling data (e.g., Kriging, land use regression, CTM). While these methods enable the assessment of adverse health associations of PM components, several challenges remain. Firstly, the large-scale routinely collected PM chemical composition data only started in the early 2000s, and it may vary by country/location. Thus, monitor-specific data on PM components remain sparse or incomplete in certain areas, resulting in lack of spatial variation and also greater exposure assessment error during land use model development and validation. Secondly and most importantly, ambient PM composition measurements at fixed-site monitors may not reflect personal exposures to ambient PM or its composition due to spatial variability of ambient PM concentrations, the house-to-house and temporal variations in indoor infiltration (i.e., attenuation) of ambient PM, and personal variation in the time spent in different outdoor and indoor places. For example, a panel study of four French metropolitan areas has showed little association between ambient monitors and personal measurements (43). As a result, the attempts to estimate long-term effects using average exposure in epidemiological studies, might lead to bias and wrong confidence intervals in health effect estimates, as well as reduce the power of such studies to build up the proper association between exposure and health effects.

Despite the challenges, there lie opportunities for future development and evaluation of appropriate exposure metrics that can estimate personal exposure well. For example, new research has emerged to explore the use of Exposure Model for Individual (EMI) to predict exposure metrics for actual individuals in the study using ambient air pollutants monitoring concentrations, questionnaire information (e.g., the characteristics of different buildings), meteorology, and information of time and location (44). Another study conducted in Saint Louis intended to explore the association between heart rate ability for elderly people during trips on public transportation and other activities (45). Portable monitors were utilized to measure the air pollution (PM_2.5 mass, BC, fine particle counts, coarse PM) exposure for each participant. So far, the personal monitoring technique has not been applied in the cohort studies, since it is relatively expensive to measure PM_2.5 components exposure for each individual. In addition, new research interests have accumulated on the estimating infiltration rates (46), since people spend most of their time indoor and it is not feasible to measure indoor exposure for each individual. Both outdoor and indoor light scattering measurements were conducted and combined with meteorological data and spatial property assessment data (46). Such ideas have already been applied to PM_2.5 mass, so it is promising to apply it to PM components. Furthermore, as more and more people are living in three dimensional landscape and exhibit a great deal of population mobility, neither traditional land use regression model nor dispersion model takes into account the vertical profile of the air pollution and the mobility of the population, especially in cities. A recent study which incorporates both three dimensional air pollution and population mobility has been developed in Hong Kong. Such research interest is quite promising in future.

To conclude, exposure assessment for PM components is complicated and need to be considered before exploring the association between PM components and adverse health effects. For example, due to the high spatial variability of PM components, the use of central site concentration alone may not reveal their real spatial variability. More concerns should be given to PM components infiltration rate from outdoor sources and personal monitoring methods.

Acknowledgments

Funding: None.

Footnote

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at http://dx.doi.org/10.21037/jphe.2018.03.03). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Krall JR, Anderson GB, Dominici F, et al. Short-term exposure to particulate matter constituents and mortality in a national study of U.S. urban communities. Environ Health Perspect 2013;121:1148-53. [PubMed]
2. Bell ML. Assessment of the health impacts of particulate matter characteristics. Res Rep Health Eff Inst 2012;5-38. [PubMed]
3. Mirowsky J, Hickey C, Horton L, et al. The effect of particle size, location and season on the toxicity of urban and rural particulate matter. Inhal Toxicol 2013;25:747-57. [Crossref] [PubMed]
4. Katsouyanni K, Samet JM, Anderson H, et al. Air pollution and health: a European and North American approach (APHENA). Res Rep Health Eff Inst 2009;5-90. [PubMed]
5. Janssen NA, Schwartz J, Zanobetti A, et al. Air conditioning and source-specific particles as modifiers of the effect of PM (10) on hospital admissions for heart and lung disease. Environ Health Perspect 2002;110:43-9. [Crossref] [PubMed]
6. Zanobetti A, Franklin M, Koutrakis P, et al. Fine particulate air pollution and its components in association with cause-specific emergency admissions. Environ Health 2009;8:58. [Crossref] [PubMed]
7. Wyzga RE, Rohr A. Long-term particulate matter exposure: Attributing health effects to individual PM components. J Air Waste Manag Assoc 2015;65:523-43. [Crossref] [PubMed]
8. National Research Council. Research priorities for airborne particulate matter: IV. Continuing research progress. Washington, DC: National Academies Press, 2004.
9. Bergen S, Sheppard L, Sampson PD, et al. A national prediction model for PM2. 5 component exposures and measurement error–corrected health effect inference. Environ Health Perspect 2013;121:1017-25. [PubMed]
10. 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;118:363-9. [Crossref] [PubMed]
11. Pope CA 3rd, Burnett RT, Thun MJ, et al. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA 2002;287:1132-41. [Crossref] [PubMed]
12. Sun S, Qiu H, Ho KF, et al. Chemical components of respirable particulate matter associated with emergency hospital admissions for type 2 diabetes mellitus in Hong Kong. Environ Int 2016;97:93-9. [Crossref] [PubMed]
13. Levy JI, Diez D, Dou Y, et al. A meta-analysis and multisite time-series analysis of the differential toxicity of major fine particulate matter constituents. Am J Epidemiol 2012;175:1091-9. [Crossref] [PubMed]
14. Pun VC, Yu IT, Qiu H, et al. Short-term associations of cause-specific emergency hospitalizations and particulate matter chemical components in Hong Kong. Am J Epidemiol 2014;179:1086-95. [Crossref] [PubMed]
15. 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;115:13-9. [Crossref] [PubMed]
16. Basu R, Harris M, Sie L, et al. Effects of fine particulate matter and its constituents on low birth weight among full-term infants in California. Environ Res 2014;128:42-51. [Crossref] [PubMed]
17. Ibrahimou B, Salihu HM, Aliyu MH, et al. Risk of preeclampsia from exposure to particulate matter (PM2.5) speciation chemicals during pregnancy. J Occup Environ Med 2014;56:1228-34. [Crossref] [PubMed]
18. Pereira G, Bell ML, Lee HJ, et al. Sources of fine particulate matter and risk of preterm birth in Connecticut, 2000-2006: a longitudinal study. Environ Health Perspect 2014;122:1117-22. [PubMed]
19. Ebisu K, Bell ML. Airborne PM2.5 chemical components and low birth weight in the northeastern and mid-Atlantic regions of the United States. Environ Health Perspect 2012;120:1746-52. [PubMed]
20. Wilhelm M, Ghosh JK, Su J, et al. Traffic-related air toxics and preterm birth: a population-based case-control study in Los Angeles County, California. Environ Health 2011;10:89. [Crossref] [PubMed]
21. Wilhelm M, Ghosh JK, Su J, et al. Traffic-related air toxics and term low birth weight in Los Angeles County, California. Environ Health Perspect 2012;120:132-8. [Crossref] [PubMed]
22. Lippmann M, Chen L, Gordon T, et al. National Particle Component Toxicity (NPACT) Initiative: integrated epidemiologic and toxicologic studies of the health effects of particulate matter components. Res Rep Health Eff Inst 2013;5-13. [PubMed]
23. Laurent O, Hu J, Li L, et al. Sources and contents of air pollution affecting term low birth weight in Los Angeles County, California, 2001–2008. Environ Res 2014;134:488-95. [Crossref] [PubMed]
24. Raaschou-Nielsen O, Beelen R, Wang M, et al. Particulate matter air pollution components and risk for lung cancer. Environ Int 2016;87:66-73. [Crossref] [PubMed]
25. Eeftens M, Hoek G, Gruzieva O, et al. Elemental composition of particulate matter and the association with lung function. Epidemiology 2014;25:648-57. [Crossref] [PubMed]
26. Gehring U, Beelen R, Eeftens M, et al. Particulate matter composition and respiratory health: the PIAMA Birth Cohort study. Epidemiology 2015;26:300-9. [Crossref] [PubMed]
27. Hampel R, Peters A, Beelen R, et al. Long-term effects of elemental composition of particulate matter on inflammatory blood markers in European cohorts. Environ Int 2015;82:76-84. [Crossref] [PubMed]
28. Wang M, Beelen R, Stafoggia M, et al. Long-term exposure to elemental constituents of particulate matter and cardiovascular mortality in 19 European cohorts: results from the ESCAPE and TRANSPHORM projects. Environ Int 2014;66:97-106. [Crossref] [PubMed]
29. Adhikari R, D’Souza J, Solimon EZ, et al. Long-term coarse particulate matter exposure and heart rate variability in the Multi-Ethnic Study of Atherosclerosis (MESA). Epidemiology 2016;27:405-13. [Crossref] [PubMed]
30. Warren J, Fuentes M, Herring A, et al. Bayesian Spatial-temporal Model for Cardiac Congenital Anomalies and Ambient Air Pollution Risk Assessment. Environmetrics 2012;23:673-84. [Crossref] [PubMed]
31. Hu J, Zhang H, Chen S-H, et al. Predicting primary PM2. 5 and PM0. 1 trace composition for epidemiological studies in California. Environ Sci Technol 2014;48:4971-9. [Crossref] [PubMed]
32. Philip S, Martin RV, van Donkelaar A, et al. Global chemical composition of ambient fine particulate matter for exposure assessment. Environ Sci Technol 2014;48:13060-8. [Crossref] [PubMed]
33. Lee S, Liu W, Wang Y, et al. Source apportionment of PM2.5: Comparing PMF and CMB results for four ambient monitoring sites in the southeastern United States. Atmospheric Environment 2008;42:4126-37. [Crossref]
34. de Hoogh K, Wang M, Adam M, et al. Development of land use regression models for particle composition in twenty study areas in Europe. Environ Sci Technol 2013;47:5778-86. [Crossref] [PubMed]
35. Adar SD, D'Souza J, Mendelsohn-Victor K, et al. Markers of inflammation and coagulation after long-term exposure to coarse particulate matter: a cross-sectional analysis from the multi-ethnic study of atherosclerosis. Environ Health Perspect 2015;123:541-8. [PubMed]
36. Zhang K, Larson TV, Gassett A, et al. Characterizing spatial patterns of airborne coarse particulate (PM10-2.5) mass and chemical components in three cities: the multi-ethnic study of atherosclerosis. Environ Health Perspect 2014;122:823-30. [PubMed]
37. Eeftens M, Beelen R, Fischer P, et al. Stability of measured and modelled spatial contrasts in NO 2 over time. Occup Environ Med 2011;68:765-70. [Crossref] [PubMed]
38. Gulliver J, de Hoogh K, Hansell A, et al. Development and back-extrapolation of NO2 land use regression models for historic exposure assessment in Great Britain. Environ Sci Technol 2013;47:7804-11. [Crossref] [PubMed]
39. Cesaroni G, Porta D, Badaloni C, et al. Nitrogen dioxide levels estimated from land use regression models several years apart and association with mortality in a large cohort study. Environ Health 2012;11:48. [Crossref] [PubMed]
40. Wang M, Beelen R, Basagana X, et al. Evaluation of land use regression models for NO2 and particulate matter in 20 European study areas: the ESCAPE project. Environ Sci Technol 2013;47:4357-64. [Crossref] [PubMed]
41. Ying Q, Lu J, Allen P, et al. Modeling air quality during the California Regional PM 10/PM 2.5 Air Quality Study (CRPAQS) using the UCD/CIT source-oriented air quality model–Part I. Base case model results. Atmospheric Environment 2008;42:8954-66. [Crossref]
42. Robledo CA, Mendola P, Yeung E, et al. Preconception and early pregnancy air pollution exposures and risk of gestational diabetes mellitus. Environ Res 2015;137:316-22. [Crossref] [PubMed]
43. Nerriere É, Zmirou-Navier D, Blanchard O, et al. Can we use fixed ambient air monitors to estimate population long-term exposure to air pollutants? The case of spatial variability in the Genotox ER study. Environ Res 2005;97:32-42. [Crossref] [PubMed]
44. Breen MS, Long TC, Schultz BD, et al. Air pollution exposure model for individuals (EMI) in health studies: evaluation for ambient PM2. 5 in central North Carolina. Environ Sci Technol 2015;49:14184-94. [Crossref] [PubMed]
45. Adar SD, Gold DR, Coull BA, et al. Focused exposures to airborne traffic particles and heart rate variability in the elderly. Epidemiology 2007;18:95-103. [Crossref] [PubMed]
46. Hystad PU, Setton EM, Allen RW, et al. Modeling residential fine particulate matter infiltration for exposure assessment. J Expo Sci Environ Epidemiol 2009;19:570-9. [Crossref] [PubMed]