Article Review

1238 volume 120 | number 9 | September 2012 • Environmental Health Perspectives

Review

In a recent published commentary (Mauderly et al. 2010), a group of scientists represent- ing academia, government, and industry groups posed the question, “Is the air pollu- tion health research community prepared to support a multi pollutant air quality manage- ment framework?” In agreement with several other contemporary reviews, editorials, and opinion papers on the subject (e.g., Dominici et al. 2010; Greenbaum and Shaikh 2010; Vedal and Kaufman 2011), the authors con- cluded that, although significant data gaps limit our current understanding of health effects resulting from exposure to air pol- lutant mixtures, much can be gained in the near future through an increased emphasis on multi pollutant issues across the spectrum from basic scientific research through imple- mentation of air quality control strategies. Although single-pollutant approaches to air pollution research, health assessments, and setting standards for ambient air quality have been successful in reducing air pollution over the past few decades, there is a clear need for parallel efforts within both the scientific and the regulatory and policy communities to advance methods for evaluating and manag- ing the effects of air pollution using a multi- pollutant approach.

This review derives from a multipollutant science and risk analysis public workshop that was held on 22–24 February 2011 in Chapel Hill, North Carolina, with the purpose of providing a brief overview of the state of the science and identifying data gaps related to addressing the health consequences of air pollu tion in a multipollutant context, includ- ing realistic steps and targets to advancing scientific and policy decisions. Co-organized and sponsored by the U.S. Environmental Protection Agency (EPA) and the Health Effects Institute (HEI), this workshop was designed to facilitate open discussions among expert scientists; these discussions are now playing a key role in multi pollutant research planning within the U.S. EPA’s Office of Research and Development and are also helping to guide the development of the U.S. EPA’s framework for conducting multipollutant science and risk assessments.

Evaluating the health impacts of multi- pollutant exposures has been identified as a priority research area in the U.S. EPA’s integrated, cross-disciplinary research planning (U.S. EPA 2012), including the establishment of four university-based Clean Air Research Centers (CLARCs) to study exposures to air pollution mixtures and their associated

health effects. As additional evidence of its commitment to this new thinking, scientists within the U.S. EPA’s National Center for Environmental Assessment (NCEA), which is responsible for evaluating and synthesizing the scientific information related to the effects of exposure to criteria air pollutants as a part of the National Ambient Air Quality Standards (NAAQS) review process, are currently developing plans for conducting a formal multi pollutant science assessment (MSA) of the health effects of exposure to air pollutant mixtures. As an initial step in the development of this proposed human health MSA, the U.S. EPA is preparing a framework describing the purpose and scope of the MSA, along with plans for conducting multipollutant analyses using existing data and information that will provide scientific support to the development of the MSA. The MSA is intended to serve as a companion document to single-pollutant Integrated Science Assessments (ISAs) of the criteria air pollutants (i.e., particulate matter, ozone, nitrogen oxides, sulfur oxides, lead, and carbon monoxide), and allow for a more effective evaluation of both the health effects of air pollutant mixtures, as well as the effects of single pollutants in a multi- pollutant context. This approach is consistent with the recommendations from the 2004 National Research Council (NRC) Report,

Address correspondence to L.W. Stanek, National Center for Environmental Assessment, U.S. Environmental Protection Agency, 109 T.W. Alexander Dr., Mail Code B243-01, Research Triangle Park, NC 27711 USA. Telephone: (919) 541-7792. Fax: (919) 541-2985. E-mail: stanek. [email protected]

We gratefully acknowledge M. Patel and A. Vette for their thoughtful reviews of this manuscript. The work reported here was performed by the U.S. Environmental Protection Agency (EPA) and the Health Effects Institute (HEI), Boston, Massachusetts. The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA, HEI, or HEI’s sponsors.

K.W. and D.G. are employed by HEI. HEI receives about half of its core funds from the U.S. EPA and half from the worldwide motor-vehicle industry, although other public and private organizations peri- odically support special projects or certain research programs. The other authors declare they have no actual or potential competing financial interests.

Received 9 January 2012; accepted 29 May 2012.

Practical Advancement of Multipollutant Scientific and Risk Assessment Approaches for Ambient Air Pollution Douglas O. Johns,1 Lindsay Wichers Stanek,1 Katherine Walker,2 Souad Benromdhane,3 Bryan Hubbell,3 Mary Ross,1 Robert B. Devlin,4 Daniel L. Costa,5 and Daniel S. Greenbaum2

1National Center for Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA; 2Health Effects Institute, Boston, Massachusetts, USA; 3Office of Air Quality Planning and Standards, 4National Health and Environmental Effects Research Laboratory, Office of Research and Development, and 5Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA

Objectives: The U.S. Environmental Protection Agency is working toward gaining a better understanding of the human health impacts of exposure to complex air pollutant mixtures and the key features that drive the toxicity of these mixtures, which can then be used for future scientific and risk assessments.

Data sOurces: A public workshop was held in Chapel Hill, North Carolina, 22–24 February 2011, to discuss scientific issues and data gaps related to adopting multipollutant science and risk assess- ment approaches, with a particular focus on the criteria air pollutants. Expert panelists in the fields of epidemiology, toxicology, and atmospheric and exposure sciences led open discussions to encour- age workshop participants to think broadly about available and emerging scientific evidence related to multipollutant approaches to evaluating the health effects of air pollution.

synthesis: Although there is clearly a need for novel research and analytical approaches to better characterize the health effects of multipollutant exposures, much progress can be made by using existing scientific information and statistical methods to evaluate the effects of single pollutants in a multipollutant context. This work will have a direct impact on the development of a multipollutant science assessment and a conceptual framework for conducting multipollutant risk assessments.

cOnclusiOns: Transitioning to a multi pollutant paradigm can be aided through the adoption of a framework for multi pollutant science and risk assessment that encompasses well-studied and ubi- quitous air pollutants. Successfully advancing methods for conducting these assessments will require collaborative and parallel efforts between the scientific and environmental regulatory and policy communities.

Key wOrDs: air pollution, exposure, human health, multipollutant, risk assessment. Environ Health Perspect 120:1238–1242 (2012). http://dx.doi.org/10.1289/ehp.1204939 [Online 29 May 2012]

Multipollutant science and risk assessment

Environmental Health Perspectives • volume 120 | number 9 | September 2012 1239

Air Quality Management in the United States (NRC 2004):

Although the committee does not believe that the science has evolved to a sufficient extent to permit development of multipollutant NAAQS [National Ambient Air Quality Standards], it would be scien tifically prudent to begin to review and develop NAAQS for related pollutants in parallel and simultaneously.

It is anticipated that the development of the MSA may eventually inform multi- pollutant risk assessment; however, conducting such a risk assessment has been one of the main challenges in multipollutant science and policy (Brook et al. 2009). Technologies for exposure assessment, knowledge of exposure– response relationships, as well as the ability to communicate risk to different stake holders are key challenges for multipollutant risk assessment. A conceptual framework that will address risk assessment in a multi pollutant context is currently under consideration by staff within the U.S. EPA’s Office of Air and Radiation (OAR). However, the introduction of this conceptual framework is not an indicator that the use of multi pollutant risk assessment is imminent. Instead, this conceptual scheme would probe the current technology, modeling tools, and the data available to help design building blocks for the implementation of a multi pollutant risk assessment. Figure 1 outlines the essential elements in air pollution risk assessment, beginning with emission sources, pollutant transport and transformation, and control. In addition to source-specific emissions, global emissions (including transboundary transport and transformation) impact regional air quality and thus contribute to overall risk. The middle section of this figure identifies traditional pillars of risk assessment that are heavily reliant on research and data collection through monitoring, modeling, and experimental evaluation of health impacts in support of assessing risk. At the base of Figure 1, the different approaches that are used to characterize adverse biological responses are provided: concentration–response curves that are obtained through observational data such as epidemiologic studies; exposure–response functions based on modeling or personal exposure studies; and dose–response curves derived from controlled human exposure studies or observational studies using bio- markers of exposure. The weight of evidence for interpreting risk results requires recognition of these different approaches that depend on the properties of the pollutant of interest and the data available.

The complexities involved in the develop- ment of multi pollutant science and risk assessments are considerable, and will require scientists from various disciplines and orga- nizations to work more collaboratively going forward. For the purpose of this review,

“multipollutant” is generally defined as the criteria air pollutants along with some prior- ity air toxics. Although this approach does not include all ambient air pollutants to which populations are exposed, it does repre- sent a relatively manageable set of pollutants. Further, the health impacts of exposures to the criteria air pollutants have been extensively studied, thus making them an ideal start- ing place when considering a multi pollutant approach to evaluating the health effects of exposure to air pollution.

Evaluating the Human Health Effects of Multipollutant Exposures Human exposures. Characterizing exposure to multi pollutant mixtures requires an advanced understanding of the sources of air pollutants, the chemical transformations and interactions between multiple pollutants, and informa- tion on the correlations in space and time between their individual concentrations. The most relevant consideration for risk assess- ment is how different mixtures contribute to the overall exposures of populations, which are governed by the time spent in contact with each mixture and its constituent com- ponent pollutants. Assessing multi pollutant exposure will be aided by the ability to distin- guish which mixtures or parts of mixtures are most closely associated with particular health outcomes, rather than assessing exposure to all possible mixtures. Once the most relevant mixtures are identified, it is also possible to

study the origins of those mixtures including transport and transformation of emissions from those sources to exposure environments where populations receive significant expo- sures. Although the complexity of the mixture can be considerable, recognizing and appreci- ating this complexity may, in the end, allow assessments to be conducted with information going beyond a finite set of what seem to be the so-called “most relevant” mixtures.

Several innovative, albeit ambitious, ideas have been proposed to better characterize expo- sures to air pollutant mixtures. Approaches involving the use of existing data may cen- ter around grouping air pollutants based on sources, micro environments, or chemical and physical properties (e.g., Suh et al. 2011). In developing these groupings, it is essential that atmospheric and exposure scientists work col- laboratively with health scientists to consider approaches informed by pollutant-specific toxicological pathways. Source apportionment techniques may be useful in characterizing the contributions to personal exposure from particular emission sources. However, specific source categories can be difficult to identify given multiple sources of pollutants, their dif- fering spatial distribution relative to monitors, and the formation of secondary air pollutants (Brinkman et al. 2009). In addition, the com- position and concentrations of some source emissions change over time due to technologi- cal and regulatory factors (e.g., changes in die- sel engines), as well as to economic influences (e.g., changes in the prices of fuels). Although

Figure 1. Conceptual scheme for multipollutant risk assessment. Abbreviations: CO, carbon monoxide; NOx, nitrogen oxides; Pb, lead; PM, particulate matter; O3, ozone; SOx, sulfur oxides.

Source emissions Mixtures

Global emissions Transboundary

transport Management actions Multipollutant context

Air quality (Monitored, modeled)

Exposure assessment (Modeling, measurement)

Health impacts (Experimental data)

Concentration–response

PM, O3, NOx O3, SOx CO,Pb

Transformation and control

Research and data collection

Multipollutant risk characterization

(Weight of evidence)

Interpretation of results

Exposure–response Dose–response

Johns et al.

1240 volume 120 | number 9 | September 2012 • Environmental Health Perspectives

many sources and exposure environments may be important, existing and supplementary monitoring evidence suggests that an initial subset of key source categories, including ports, roadways, industrial centers, and biomass burn- ing can form the basis for studies in the nearer term, while others may be identified in future source apportionment studies.

It is clear that neither modeling nor addi- tional monitoring alone is likely to solve the problem of characterizing multi pollutant exposures. For example, while grid modeling techniques have the potential to address uncer- tainties related to spatial variability of air pol- lutant mixtures, in many cases these models do not provide adequate temporal resolution, require detailed source and meteorological data inputs, and are difficult to evaluate (Marmur et al. 2006). Similarly, efforts to reduce com- plexity by use of a measured chemical marker species to represent a given source or micro- environment have clear advantages, but are likely overly simplistic due to differences in monitor location and frequency of data collec- tion that vary by pollutant (termed spatial and temporal misalignment, respectively) (Gryparis et al. 2009). However, one potentially promis- ing approach to reduce uncertainties associated with spatial and temporal misalignment is the integration of monitoring data, as well as satel- lite imagery, with modeling techniques that provide interpolation of atmospheric concen- trations, potentially resulting in a more com- plete characterization of spatial and temporal variations of individual pollutant and com- bined pollutant exposures (Liu et al. 2005; Villeneuve et al. 2011).

With respect to monitoring data, trade- offs must inevitably be made between quan- tity and quality of the data. For example, the development of inexpensive personal air quality sensors has considerable appeal as an approach to gathering personal expo- sure data, but equal concern exists about the analytical sensitivity and accuracy of these monitors. However, if sensors are deployed widely, patterns of pollutants and variations in time and space may be invaluable in model development despite short comings in accu- racy. Recognizing the limitations of personal monitoring and uncertainties associated with the use of central-site monitors to represent personal exposure to ambient air pollutants, the implementation of an exposure supersite program may be warranted. At such a site, the spatial and temporal variability of the con- centrations of multiple pollutants would be characterized in multiple micro environments using similar sampling methods. Although each of these approaches has merit, adequately characterizing exposures to air pollutant mix- tures to support multi pollutant risk assess- ments may best be accomplished through integration of central site monitors, widely

deployed sensors, satellite measurements, and refined atmospheric chemistry and expo- sure modeling that can draw across the rela- tive strengths of each approach, to allow for focused validation and tool develop ment.

Health studies. The development of a MSA for health effects will entail extensive integration of findings from observational, experimental, and exposure studies. For observational studies, moving toward a multi- pollutant focus will require drawing upon many established epidemiologic approaches. Single-pollutant models can be used to evalu- ate one pollutant as an indicator of a group of pollutants, and under certain conditions, traditional two-pollutant models can provide insight into combined effects through sum- ming model coefficients or partial derivatives (Bateson et al. 2007). Epidemiologic studies that rely on source apportionment methods to characterize exposure are another way to gain insight into the relationship between health effects and a group of correlated pollutants.

Generally, there are three broad approaches to evaluating interactions or other types of effect modification in epidemiologic studies— traditional regression models, dimension reduction techniques, and Bayesian hierarchical methods (Billionnet et al. 2012). All have their merits, which include ease of use for the former two approaches and flexibility for the latter approach. Future approaches may borrow and build on newer techniques from the genomics and other “omics” communities, applied in a multi pollutant context, including clustering methods and random forest approaches (Breiman 2001; Siroux et al. 2011).

Monitoring networks will need to provide enhanced capabilities to evaluate temporal and spatial scales. Spatial considerations include improved resolution of exposure and a better understanding of the representativeness of central-site monitors. Spatial misalignment of exposure data is an issue when evaluating multi pollutant effects because the ability to estimate exposures in different locations for different pollutants varies. Observational studies already consider some temporal patterns of health outcomes; for example, daily concentrations (or < 24 hr) are necessary for assessing acute effects and long-term concentrations (including seasonal variation) are required for assessing more chronic effects. Moving forward, the development of statistical methods that allow for clearer separation of the impacts of temporal and spatial variation may enable the resolution of multi pollutant effects (Bell et al. 2007; Peng and Bell 2010). These types of methods may also be valuable in understanding why changes are observed in the magnitude of risks associated with individual pollutants over time. We can hypothesize that these changes may be due to changing levels and mixes of pollutants, potentially due to

new regulations on diesel engine and power plant emissions, but additional research is needed to test these hypotheses.

Experimental studies with animals or intentional environmental exposures involving humans will likely continue to provide evi- dence of biological plausibility for the impact of mixtures as they do for single pollutants. Because they are hypothesis driven, they can be used to directly identify individual causal agents and their contributions to an overall effect. Although information on effects occur- ring at ambient levels is most relevant when considering human health risks, differences in respiratory tract deposition and biological response between animals and humans may justify the use of exposure concentrations that exceed ambient concentrations in ani- mal toxicology studies (Brown et al. 2005). Various experimental approaches are cur- rently available to evaluate multi pollutant effects. One such approach involves animal and human exposures to “real world” mix- tures of air pollutants under controlled condi- tions or environ ments that simulate ambient conditions, such as in photochemical cham- bers (Lemos et al. 2011; Sexton et al. 2004) or a traffic tunnel (Kooter et al. 2006). One limitation of this type of study is its inability to precisely control or manipulate the expo- sure concentrations or pollutant mixture. The potential strength of this study design resides in the spatial and temporal contrasts for exposure and response. Another approach is laboratory-generated mixture studies where a few pre-selected individual pollutants are combined to create an exposure atmosphere, with the most informative studies being those that have an air control as well as exposures to individual pollutants (Mauderly and Samet 2009). Although the mixtures being studied do not reflect the aging and transformation that occur under ambient conditions, they do provide relevant information on the nature of the interactions among different chemicals.

A mode-of-action (MOA) framework describing a sequence of key events leading to an end point or clinical outcome represents a possible unifying theme when considering multiple pollutants and the common toxicity pathways through which they act (Ankley et al. 2010). Such a framework should make it possible to characterize and illustrate the interconnectedness of key event pathways and the upstream responses that produce those events. The focus of the framework should be on commonalities between air pollutants, such as robust toxicity pathways informed by well established measurements of phenotypic changes and biomarkers linking exposure and effect. This approach also has the capabil- ity of evaluating the nature of interactions between pollutants and identifying potentially susceptible populations.

Multipollutant science and risk assessment

Environmental Health Perspectives • volume 120 | number 9 | September 2012 1241

Thus, two possible approaches to address pollutant interactions are currently being developed: the common occurrence approach where a related group of pollutants from a particular source or resulting atmospheric mixture are associated with a health end point, and the MOA approach that groups specific chemicals in the atmosphere based on their effects with the same health end point. Limitations in toxicological approaches and epidemiologic data and statistical methods will require cross-disciplinary hypothesis development, research planning, and execution of this research (Dominici et al. 2008).

Synthesis and Discussion The historical focus on single pollutants has brought us a long way—a substantial data base of exposure monitoring data; a large and grow- ing body of evidence on different facets of the MOA for individual pollutants; a diverse toolbox of toxicological, clinical, and epide- miologic study designs; statistical and other data analysis methods; and air quality models that we can rely on in the short term. But it is also clear that we have a ways to go—whether it is toward the adaptation of existing data and tools or toward the development of new ones. In short, the transition to a deliberative multi pollutant approach will require substan- tial research effort and time. The U.S. EPA’s current efforts to develop a MSA, along with plans for designing a framework for conduct- ing multi pollutant risk assessments, provide an important structure and context for assessing the strengths and limitations of the existing science related to the health effects of expo- sures to air pollutant mixtures. In addition, the current and planned U.S. EPA and U.S. EPA- funded multi pollutant scientific research is

clearly vital in making a successful transition from single to multi pollutant approaches to air quality evaluation and management.

The U.S. EPA began using a multi- pollutant approach in a limited context in the recent evaluation of the secondary standards for nitrogen oxides and sulfur oxides. The development of a U.S. EPA policy assessment for nitrogen oxides and sulfur oxides resulted in consideration of a multi pollutant NAAQS to protect against the combined effects of nitrogen and sulfur oxides on aquatic acidi- fication (EPA 2011). OAR also has recently conducted a pilot study that demon strated how multi pollutant approaches to imple- menting primary (human health–based) air pollution standards could result in greater health benefits and more cost-effective imple- mentation, which may also help to achieve greater reductions in air toxics when com- pared with single-pollutant approaches (Fann et al. 2011). The adoption of a multi pollutant policy for the review of ambient air pollutants would potentially result in increased efficien- cies, bene fits, and cost savings throughout the process beginning with the initial evaluation of the scientific evidence. More important, increasing the emphasis on multi pollutant approaches may allow for a better under- standing of the types of air pollutant mixtures most likely to result in adverse health effects, which could, in turn, facilitate the identifica- tion of control strategies to minimize expo- sures to these mixtures.

As the science, risk assessment, and risk management communities advance toward adopting multi pollutant approaches, it is important to set realistic targets for progress amidst a very large number of possible direc- tions for improvement. We should aim to

take concrete steps in that direction in at least four major areas (Figure 2):

Exposure—seeking to increase the scien-• tific base of health studies in which multiple pollutants are measured simultaneously in ambient air Toxicology—seeking enhanced measure-• ment and analysis techniques for source and ambient mixture (e.g., concentrated ambient particles) exposure, and better understanding of common MOAs (key events) as a useful way to group and assess pollutants Epidemiology and statistics—seeking • enhanced techniques, especially statistical analysis techniques, that enable the charac- terization of associations between multiple pollutants (i.e., more than two at a time), sources, and health outcomes Modeling and risk assessment—seeking the • development and testing of multi pollutant approaches to estimating population expo- sure and risk.

Enhanced science in each of these areas can inform the three major components of air quality management—science assessment, risk assessment, and risk management (as in Figure 2). Additional scientific questions may arise leading to new scientific developments, further supporting progress toward under- standing effects of joint exposures to mul- tiple pollutants (e.g., the criteria pollutants and major air toxics) and, if the data permit, identifying individual pollutants and sources within the mixture that may be dispropor- tionately responsible for adverse effects on human health.

RefeRences

Ankley GT, Bennett RS, Erickson RJ, Hoff DJ, Hornung MW, Johnson RD, et al. 2010. Adverse outcome pathways: a conceptual framework to support ecotoxicology research and risk assessment. Environ Toxicol Chem 29:730–741.

Bateson TF, Coull BA, Hubbell B, Ito K. Jerrett M, Lumley T, et al. 2007. Panel discussion review: session three—issues involved in interpretation of epidemiologic analyses— statistical modeling. J Expo Sci Environ Epidemiol 17(suppl 2):S90–S96.

Bell ML, Dominici F, Ebisu K, Zeger SL, Samet JM. 2007. Spatial and temporal variation in PM2.5 chemical composition in the United States for health effects studies. Environ Health Perspect 115:989–995.

Billionnet C, Sherrill D, Annesi-Maesano I, GERIE study. 2012. Estimating the health effects of exposure to multipollutant mixture. Ann Epidemiol 22:126–141.

Breiman L. 2001. Random forests. Mach Learn 45:5–32. Brinkman GL, Milford JB, Schauer JJ, Shafer MM, Hannigan MP.

2009. Source identification of personal exposure to fine particulate matter using organic tracers. Atmos Environ 43:1972–1981.

Brook JR, Demerjian KL, Hidy G, Molina LT, Pennell WI, Scheffe R. 2009. New directions: results-oriented multipollutant air quality management. Atmos Environ 43:2091–2093.

Brown JS, Wilson WE, Grant LD. 2005. Dosimetric comparisons of particle deposition and retention in rats and humans. Inhal Toxicol 17:355–385.

Dominici F, Peng RD, Barr CD, Bell ML. 2010. Protecting human health from air pollution: shifting from a single-pollutant to a multipollutant approach. Epidemiology 21:187–194.

Dominici F, Wang C, Crainiceanu C, Parmigiani G. 2008. Model selection and health effect estimation in environmental epidemiology. Epidemiology 19:558–560.Figure 2. Schema showing continuous enhancement of multipollutant science.

Continuously enhanced multipollutant science:

• Exposure • Toxicology • Epidemiology • Risk models

Science assessment:

Multipollutant science assessment in context of single pollutant NAAQS

Risk assessment:

Multipollutant exposure and risk analysis

Risk management:

Multipollutant source controls and

multipollutant air quality management plans

Enhanced science input to MSA; new tools for data

synthesis and assessment

Enhanced input to source control decisions;

multipollutant priorities for AQ plans

Enhanced exposure and risk models

Johns et al.

1242 volume 120 | number 9 | September 2012 • Environmental Health Perspectives

Fann N, Roman HA, Fulcher CM, Gentile MA, Hubbell BJ, Wesson K, et al. 2011. Maximizing health benefits and mini- mizing inequality: incorporating local-scale data in the design and evaluation of air quality policies. Risk Anal 31:908–922.

Greenbaum D, Shaikh R. 2010. First steps toward multipollutant science for air quality decisions. Epidemiology 21:195–197.

Gryparis A, Paciorek CJ, Zeka A, Schwartz J, Coull BA. 2009. Measurement error caused by spatial misalignment in environmental epidemiology. Biostatistics 10:258–274.

Kooter IM, Boere AJ, Fokkens PH, Leseman DL, Dormans JA, Cassee FR. 2006. Response of spontaneously hypertensive rats to inhalation of fine and ultrafine particles from traffic: experimental controlled study. Part Fibre Toxicol 15:3–7.

Lemos M, Diaz EA, Gupta T, Kang CM, Ruiz P, Coull BA, et al. 2011. Cardiac and pulmonary oxidative stress in rats exposed to realistic emissions of source aerosols. Inhal Toxicol 23(suppl 2):75–83.

Liu Y, Sarnat JA, Kilaru V, Jacob DJ, Koutrakis P. 2005. Estimating ground-level PM 2.5 in the eastern United States using satellite remote sensing. Environ Sci Technol 39:3269–3278.

Marmur A, Park SK, Mulholland JA, Tolbert PE, Russell AG. 2006. Source apportionment of PM2.5 in the southeastern

United States using receptor and emissions-based mod- els: conceptual differences and implications for time- series health studies. Atmos Environ 40:2533–2551.

Mauderly JL, Burnett RT, Castillejos M, Özkaynak H, Samet JM, Stieb DM, et al. 2010. Is the air pollution health research community prepared to support a multi pollutant air quality management framework? Inhal Toxicol 22(suppl 1):1–19.

Mauderly JL, Samet JM. 2009. Is there evidence for synergy among air pollutants in causing health effects? Environ Health Perspect 117:1–6.

NRC (National Research Council). 2004. Air Quality Management in the United States. Washington, DC:National Academies Press.

Peng RD, Bell ML. 2010. Spatial misalignment in time series studies of air pollution and health data. Biostatistics 11:720–740.

Sexton KG, Jeffries HE, Jang M, Kamens RM, Doyle M, Voicu I, et al. 2004. Photochemical products in urban mixtures enhance inflammatory responses in lung cells. Inhal Toxicol 16(suppl 1):107–114.

Siroux V, Basagaña X, Boudier A, Pin I, Garcia-Aymerich J, Vesin A, et al. 2011. Identifying adult asthma phenotypes using a clustering approach. Eur Respir J 38:310–317.

Suh HH, Zanobetti A, Schwartz J, Coull BA. 2011. Chemical properties of air pollutants and cause-specific hospital admissions among the elderly in Atlanta, Georgia. Environ Health Perspect 119:1421–1428.

U.S. EPA (U.S. Environmental Protection Agency). 2011. Policy Assessment for the Review of the Secondary National Ambient Air Quality Standards for Oxides of Nitrogen and Oxides sf Sulfur. EPA-452/R-11-005a. Research Triangle Park, NC:U.S. EPA.

U.S. EPA (U.S. Environmental Protection Agency). 2012. Air, Climate, and Energy Strategic Research Action Plan 2012- 2016. Available: http://www.epa.gov/research/docs/strap- ace2012.pdf [accessed 26 July 2012].

Vedal S, Kaufman JD. 2011. What does multipollutant air pollu- tion research mean? [Editorial]. Am J Respir Crit Care Med 183:4–6.

Villeneuve PJ, Goldberg MS, Burnett RT, van Donkelaar A, Chen H, Martin RV. 2011. Associations between ciga- rette smoking, obesity, sociodemographic characteristics, and remote-sensing–derived estimates of ambient PM2.5: results from a Canadian population-based survey. Occup Environ Med 68:920–927.

Copyright of Environmental Health Perspectives is the property of Superintendent of Documents and its content

may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express

written permission. However, users may print, download, or email articles for individual use.