Introduction

Accurate assessment of human exposures is an important part of environmental health research. Both outdoor and indoor sources of pollutants influence an individual's exposures during the course of their lives and daily activities. Understanding and quantifying these exposures are essential but they are often not adequately addressed in health effect studies for air pollution. Air pollution epidemiology, risk assessment, health tracking, and accountability assessments are examples of health effect studies that require but often lack adequate exposure information. For example, most air pollution epidemiology studies make use of potentially unreliable surrogates of personal exposures, such as information based on available central-site outdoor concentration monitoring or modeling data. However, more explicit studies reveal that personal exposures tend to be greater in magnitude and more variable in location and time than the corresponding ambient concentrations. Examples are individuals near major point or on-road-emission sources of air pollutants such as particulate matter (PM) and hazardous air pollutants. Other examples are individuals exposed during their daily activities to higher levels of outdoor pollutants such as ozone or indoor pollutants such as combustion products including nitrogen dioxide (NO₂) and semi-volatile organic compounds (SVOCs) (Spengler et al., 1994; Özkaynak et al., 2008a).

Complex patterns in the spatial variation of exposures among different population cohorts, especially in the context of cross-sectional or intra-urban analysis of air-pollution health effects, remain challenging. Recent research has shown that the variations in exposure-to-concentration ratios can be highly dependent on pollutant type and the locations and activities of the exposed population (Isakov et al., 2006; Özkaynak et al., 2008a). Thus, the assignment of outdoor concentrations as a proxy for personal exposures introduces varying degrees of exposure misclassification. The degree of misclassification depends on the nature of the epidemiologic model tested and the statistical methodology employed. Moreover, the use of simple proxies in the assignment of exposures often have important implications for both the design and the interpretation of the findings derived from community air pollution health studies. Recent advances in exposure modeling along with better information on time–activity, commuting patterns, and exposure factors data, provide us with unique opportunities to improve the assignment of exposures for both future and ongoing studies linking air pollution to health impacts.

In September 2006, scientists from the US Environmental Protection Agency (EPA) and the Centers for Disease Control and Prevention (CDC) along with scientists from the academic community and state health departments convened a symposium on air pollution exposure and health to identify, evaluate, and improve current approaches for linking air pollution exposures to disease. Scientists from these different sectors worked to develop consensus on the capabilities and limitations of the current state of the science and then made recommendations and set priorities for near-term and long-term research goals that could improve this linking process. The symposium participants divided themselves into working groups that focused on the specific topics of (i) exposure, (ii) health information, (iii) emerging issues, and (iv) linking exposure and health data. This manuscript focuses on the research efforts of the first of these working groups and presents the key issues, challenges, and recommendations identified by the exposure working group — referred from here on as the workgroup. To explore key issues and develop findings and recommendations, this workgroup used case studies of PM, ozone, and toxic air-pollutant exposure in the context of air pollution epidemiology, risk assessment, environmental public health tracking (EPHT) or surveillance studies, and accountability. The two companion papers by Özkaynak et al. (2008b); Thurston et al. (2008), which appear in this issue of the journal, summarize the overall findings from this symposium and the issues and recommendations regarding health information, respectively.

Our objectives in this manuscript are to report the issues and approaches used in the September 2006 workshop to determine short-term and long-term goals to improve the use of monitoring data and exposure model results for PM, ozone, and toxic air pollution in health surveillance and health research studies. The Approach and methods section below describes how the workshop was organized and used to select and set priorities among the issues that should be considered and the research needed to confront these issues. This section also describes how the participants in the exposure workgroup selected a set of key issues and used these issues to explore limitations, barriers, and challenges to improving health surveillance and health impact research. As part of the process of methods development, the workgroup agreed to address the following key issues:

1. Differential exposures — indoor, outdoor, and other environments
2. The impact of population location and mobility
3. Biologically relevant timescales
4. Age-specific exposure issues
5. Exposures to multiple pollutants
6. Tracking effect modifiers
7. Tracking exposures for long-latency outcomes
8. Tracking exposure outcome over short timescales

The Results section provides the workgroup's findings on each of these key issues in the context of limitations, barriers, and challenges. This is followed by a discussion of priorities for actions to understand limitations, confront challenges, and remove barriers.

Approach and methods

Our approach to this research and the process for developing our findings include three primary activities, all structured around the workgroup process. First, the symposium organizers described to the participants how the workgroup process would be used to explore key issues and set research priorities. Second, the symposium organizers developed the conceptual framework in which participants could identify and evaluate short-term and long-term efforts needed to confront the limitations, barriers, and challenges to linking disease and exposure. In describing this framework, we compare here classical epidemiology, risk assessment, health tracking, and accountability to evaluate the needs of these different health-research efforts for information relating exposure to disease. In Table 1, we define each of these terms. Finally, the organizers identified and described substances of interest that provide informative case studies for the workgroup evaluations.

Table 1 - Distinctions among epidemiology, risk assessment, health tracking, and accountability.

Full table

The Conceptual Framework for Linking Air Pollution Exposure to Disease

Epidemiology, risk assessment, health tracking, and accountability require information that link environmental exposure to disease. Each of these efforts requires identification of individuals at increased exposures and tools to determine how and whether the more exposed individuals show a greater likelihood of increased risk of disease. The inadequacy of exposure assessment in these types of health studies was an underlying premise of the workgroup discussions and findings. In many areas of epidemiological research, this may not necessarily be the case, but the workgroup found it a key issue for air pollution epidemiology. The workgroup developed Figure 1 to illustrate the conceptual link between exposure and disease along with the internal pathway from exposure to intake, dose, biological changes, early biological effect, altered structure or function, and finally disease. Historically, scientists had to rely primarily on measures of exposure and disease diagnosis to make this link. But the use of better population data, exposure models, exposure biomarkers, and early detection of disease states have improved our ability to explore exposure/disease links (Ryan et al., 2007). We use this conceptual framework both to assess current limitations for health studies applied to air pollution and to identify research opportunities. On the basis of this framework, we characterize information needs and a strategy for our evaluation.

Figure 1.

Conceptual framework indicating the links among exposure information, modeling, monitoring, tracking, and disease.

Full figure and legend (71K)

Distinctions Among Classical Epidemiology, Risk Assessment, Health Tracking, and Accountability

There are many research, modeling, and evaluation activities that health scientists use to make links between human disease burden and exposures to pollutants. In setting the goals for the work reported here, the authors and the workgroup make a distinction among four approaches, two that have been used historically — epidemiology and risk assessment — and two that are emerging — health tracking and accountability. The workgroup found it important to articulate the distinctions among these different approaches both to facilitate the workgroup process and to make sense of the workshop findings. The distinctions that the workgroup established among these four approaches for linking disease and exposure are provided in Table 1.

Specific Substances of Interest

Three air contaminants are of interest in the context of this health surveillance and research investigation. These are PM, ozone, and toxic air pollutants. On the basis of its authority under the Clean Air Act Amendments of 1990, the EPA established a list of 188 toxic compounds. Thus toxic air pollutants are clearly multi-component in nature whereas PM and ozone are generally considered single compounds for regulatory purposes. But in the case of PM, recent data on health outcomes suggest that coarse, fine, and ultrafine PM may have different health impacts and could be better characterized as a multi-component pollutant in the future as more information becomes available. In the following paragraphs, we (the authors) discuss the important properties of these substances and their link to human disease.

Particulate Matter

PM measurements are driven largely by efforts to meet target limits for the mass concentration of PM in the urban atmosphere. Important for this process is the selection of a metric by which PM concentration is measured. Some form of mass measurement has been used as the metric for PM pollution for at least four decades. Over the last two decades, epidemiologic studies have suggested that fine particles are likely more culpable for adverse health effects, and regulations have shifted to reflect this. As a result, EPA has modified standards to include fine aerosol as a separate class of pollutant.

However, PM is not a single entity. Crustal materials — those dispersed from the soil surface — consisting primarily of carbonates and silicates are likely to have completely different health outcomes when compared to fine or ultrafine particles produced by combustion processes (Spiro and Sligliani, 2003). Diesel exhaust, for example, contains PM with a large fraction of organic material in contrast to the inorganic PM produced in smelting operations. These differing chemical compositions for PM suggest now that determining health outcomes for PM requires alternative strategies that focus on the chemical composition of the PM itself. Sorting PM into different size classes can help address this issue. For example, combustion-related PM sources such as diesel exhaust typically contribute to the fine components of PM (USEPA, 2004). However, to address specific health outcomes, such as respiratory or cardiovascular diseases, better speciation of PM is necessary.

Ozone

Ozone, a molecule consisting of three oxygen atoms, is a potent oxidant due, in part, to its reactive nature. Ozone reacts with sunlight to produce free atomic oxygen [O(³P)] or reacts with other chemicals in the air to produce odd oxygen species, which then begin a chain of reactions that leads to photochemical smog. This reactive behavior contributes to the formation of numerous other species ranging from aldehydes and ketones, to peroxyl radicals and excited intermediates (Seinfeld and Pandos, 1998; Finlayson-Pitts and Pitts, 2000; Spiro and Sligliani, 2003). Thus measuring the principal oxidant species, ozone, can give information on other reactive contaminants in the atmosphere. In this way, ozone can be a chemical surrogate for all oxidizing species to some extent. Ozone is typically monitored in urban settings subject to photochemical smog incidents but usually only during the "ozone season" that typically spans the months of May through September in the Unites States.

Exposure to air containing pure ozone in laboratory settings has demonstrated a relationship between health outcomes and exposure in both animals and human subjects (Hackney et al., 1976; Hackney et al., 1977; Linn et al., 1978, 1982, 1983, 1988; Bates, 1995; Lippmann and Schlesinger, 2000). However, in the ambient environment, exposure to other species, most notably the oxidants found in polluted environments, may also be responsible for adverse health outcomes (Ostro, 1993; Catalano et al., 1996; Burnett et al., 1997; Peters et al., 1999; Kinney and Lippmann, 2000; McConnell et al., 2002; Bernstein et al., 2004; Ruidavets et al., 2005). Ozone measurement systems can be quite specific to the O₃ species, although interferences by water vapor and aromatic hydrocarbons may occur for some analyzers (USEPA, 1997; NARSTO, 1999). Because of the nature of the photochemical equilibrium of ozone in the ambient environment, epidemiologic investigations of the health effects of ozone may in effect be using ozone as an exposure surrogate for these other species, some of which could be more harmful than ozone itself.

Toxic Air Pollutants

The class of air contaminants referred to as toxic air pollutants or "air toxics" contains numerous chemical compounds with widely varying sources, environmental fate, exposure pathways, and health outcomes. Unlike PM, few health scientists in the last 20 years have suggested that this class of substances can be aggregated into a single entity. Data-collection efforts are often broad-based, with whole air samples being captured and analyzed for many compounds simultaneously (see for example, Hayes, 1989; Eschenbacher et al., 1995; Wallace and Pellizzari, 1995; Delfino, 2002; Kinney et al., 2002; Weisel, 2002; Phillips et al., 2005; Xue et al., 2005).

The health effects of toxic air pollutants are more difficult to study than those of PM and ozone. Although it is relatively easy to examine the effects of compound-specific exposures, in ambient air, one encounters different mixtures of toxic air pollutants, with transportation sources, industrial sources, and natural sources contributing to varying degrees depending upon location and time of year. The analysis of the impact of exposure to air toxics on health remains challenging because of this variability. It may be reasonable to expect nonadditive effects on health outcomes associated with simultaneous exposure to a mixture of these compounds. Research is currently focused on understanding these effects, but results from these short-term exposure studies may not be representative for assessing chronic effects such as cancer, which results from long-term exposures to air toxics. Detecting additive health effects, whether synergistic or antagonistic, currently poses significant challenges.

The Workgroup Process

As leaders of the exposure workgroup, we relied on consensus discussions to identify and evaluate key issues. At the beginning of the workgroup process, the following four questions were presented to the workgroup.

1. What are the limitations, barriers, and challenges associated with the use of existing environmental, exposure, and health databases or models, and methodologies used to link them, in air pollution health studies including accountability, epidemiologic research, and EPHT?
2. How can we overcome these limitations, barriers, and challenges?
3. What are the important emerging health effects and air quality issues for accountability, air-pollution health effect research, and EPHT?
4. In the order of priority, what are the short- and long-term activities to improve the use and linkage of environmental exposure, health databases, or models for accountability, air-pollution health effect research, and EPHT?

The group leaders guided the workgroup effort to apply these questions systematically to PM, ozone, and toxic air pollutants. The group approached this goal by first defining a conceptual framework (Figure 1) within which to address the question of how exposure relates to disease and then used this framework to establish, refine, and address a list of key issues. The group then worked to find overlaps and gaps. A revised list was prepared and again evaluated for overlaps and gaps. The group repeated this process until they achieved consensus. These issues were then used to identify and evaluate methods to overcome these limitations, barriers, and challenges.

The workgroup process identified eight key issues based on consideration of the above four questions. These issues are presented in the next section. During their deliberations to address these four questions, the workgroup discussed the capabilities of new sources of information such as satellite and other remote sensing data and the reliability of modeling exposure from air quality data collected in support of compliance or exposure monitoring. They evaluated technical barriers or communication difficulties in sharing the necessary exposure information between the different research organizations, institutions, or programs. They considered a range of actions from simple, short-term solutions to complex and long-term strategies. The workgroup gave particular attention to the types of data and models needed for source-to-dose characterization at neighborhood, urban, and regional scales. They also considered the use of geographic information system (GIS)-based inhalation exposure models for application to PM, ozone, and toxic air pollutants and the impact of location, mobility, activity (breathing rates and microenvironment), and demographics on estimated inhalation rates for PM, ozone, and toxic air pollutants. The workgroup considered specific actions such as the use of census data to locate populations relative to air pollution concentrations, with adjustments to account for population mobility. Finally, they considered the feasibility and value of constructing population exposure distributions by combining (1) spatially and temporally resolved estimates of ambient concentrations of specific air pollutants; (2) geo-coded time–location–activity survey data; (3) specific exposure microenvironments; and (4) breathing rates, which vary by age, gender, and activity level.

Information Needs

In the framework outline shown in Figure 1, hazard and exposure information relevant to health tracking includes source/emissions data, environmental monitoring data, biological monitoring data, time–activity–location data for the exposed population, and other available and relevant data (e.g., from questionnaires and diaries, including past and current exposures and exposure factors). In spite of the clear need for health tracking to make use of environmental factors to classify populations with respect to hazards, there is limited capacity for such activities within communities, state agencies, and federal agencies. The need to build this capacity across a broad range of agencies and communities was a key factor in defining the workshop evaluation strategy.

Emerging Issues

Dose reconstruction is becoming important in a number of health studies. The ability to back-calculate the magnitude and source of exposure adds value to epidemiology studies as well as health tracking and accountability studies. But reliable dose reconstruction requires a combination of information on sources, exposures, and markers of dose (see, e.g., McKone et al., 2007). It is important to recognize that even when there is an abundance of information on exposure biomarkers, there may not be sufficient information to establish source-to-dose links. For example, Sohn et al. (2004) have illustrated some of the problems in dose reconstruction even in cases where there is good temporal data on tissue concentrations. As noted by Ryan et al. (2007), there is a continuum of information that must be assembled to use exposure biomarker data to establish source-to-dose links. Improving this process is an emerging area of research in health effects research.

Recent studies reveal that the spatial and population resolutions required for exposure tracking depend on whether one is tracking ambient source pollutants, such as those found in ambient air, or tracking surface water or indoor/product-based pollutants, such as those found in food, soil, consumer products, indoor releases and so on (Lobscheid and McKone, 2004; Marshall et al., 2005; McKone et al., 2007). There is heterogeneity of ambient pollutant intake due to the geographic distribution of sources, variation among pollutants in dispersion patterns from the same source, population mobility, and indoor/outdoor exposure levels. In addition, many contaminants of interest in environmental health research can be associated with both regional concentrations and local- or even indoor-source peak levels. For example, people are exposed to polycyclic aromatic hydrocarbons (PAHs) not only as a result of living in a region with a large number of sources, but also as a result of their proximity to roadways and to neighborhood-scale wood burning (Lobscheid and McKone, 2004). Food residues and indoor releases from cooking may provide additional important exposure pathways to some of these substances.

In addition to the difficulty of linking exposure with proximity to pollutant emissions, there are complex links between exposure and health effects. For many pollutants, there are significant differences between exposure and intake, intake and dose, and dose and effect. Using ambient pollutants as an example, one notes that variations in breathing rates, diet, activity health status, and genetic susceptibility can cause individuals with the same exposure to respond differently. Thus, differences in activity levels, metabolic rates, health status, and other factors can contribute to significant differences in health impacts for individuals experiencing very similar pollutant exposures. This means that to improve the links between exposure and disease at a population scale, one will need to collect, store, and evaluate this type of information at a level that provides detailed sub-population and even individual data on these factors.

Results

In this section, we identify the eight key issues identified by the workshop and provide the workshop findings about each of these key issues in the context of limitations, barriers, and challenges. We then provide and set priorities for actions to understand limitations, remove barriers, and confront challenges.

Findings on the Key Issues

Differential exposures

Exposures to a number of air pollutants are differential in the sense that exposure is not proportional to the relative outdoor concentrations of these pollutants. Time–activity allocations and mobility (e.g., commuting) and microenvironmental and building infiltration factors are among the main factors that modify human exposures to outdoor pollutants. Very reactive gaseous contaminants such as ozone often display indoor–outdoor ratios that are very low due to their reactions with surfaces in indoor environments as well as indoor, gas-phase reactions. Typical indoor–outdoor ratios for ozone range from 0.1 to 0.4 with the low end of these values found in air-conditioned buildings in summer and the high end observed when windows are open and residences are well ventilated (Romieu et al., 1998; Lee et al., 1999, 2004; Geyh et al., 2000). Gaseous pollutants of intermediate reactivity, including NO₂, display indoor–outdoor ratios of about 0.5 in the absence of sources indoor (Ryan et al., 1988; Spengler et al., 1994; Rojas-Bracho et al., 2002). Although there are few sources of ozone indoors, NO₂ is produced indoors by combustion, so the presence of combustion appliances strongly affects this ratio. Nonreactive gas-phase contaminants display indoor–outdoor ratios approaching unity. On the basis of several studies done in locations around the world, typical ratios measured for benzene, for example, center on 0.9 (Son et al., 2003; Adgate et al., 2004; Serrano-Trespalacios et al., 2004). Again, the presence of sources, for example, an attached garage, can markedly affect this ratio. PM can be partially filtered by the building envelope resulting in indoor–outdoor ratios less than unity. Various studies have found indoor–outdoor ratios for PM_2.5 to center on 0.6 (Geller et al., 2002; Bennett and Koutrakis, 2006). Such ratios can, again, be markedly affected by indoor sources including cooking activities, use of tobacco products, cleaning activities (Singer et al., 2006), and general activity associated with house occupants. Because of this, researchers often use tracers unique to outdoor air, for example, sulfate, to measure effective penetration of PM.

Although the summary numbers presented here may indicate that the exposure experienced by individuals could still be approximated using a constant multiple of outdoor concentrations, the variability of time spent indoors and outdoors as well as the variability in these penetration ratios precludes such a simple analysis. Personal activities, including amounts of time spent in each microenvironment, as well as differential use of products that produce indoor sources suggest that the relationship between ambient air concentrations of these contaminants and personal exposures is not simple. Personal exposures experienced by individuals in a population can be lower, approximately equal, or higher than those inferred from ambient pollutant concentrations. Thus, to assess such exposures, exposure researchers need detailed measurement of personal exposures or more detailed exposure models that are based on knowledge of individual activities as well as variations in indoor and ambient concentrations. Population exposures may average out these differences, but exposures experienced by individuals, including those in "small" epidemiologic investigations, are likely to be strongly influenced by variations among individuals' nonambient contributions to overall exposure.

The impact of location and mobility

Mobility and time spent in locations away from a primary residence greatly influence people's exposures to a number of pollutant emissions — both indoors and outdoors. This is particularly important when one has geographically based exposure information rather than personal exposure monitoring data. The majority of exposure information for PM, ozone, and toxic air pollutants comes from stationary monitoring data, pollutant transport models and/or some combination of monitoring data, and model results. However, regardless of the approach used to obtain concentration data, when population exposures are linked to a specific location (census track, zip code, street address) without adjustments for mobility and location changes, estimated individual exposures can differ significantly from actual exposures. For example, near- and on-roadway exposures to motor-vehicle pollutants will be greater than the exposures to same pollutants while subjects are indoors at home, at work, or at school. Location and mobility can be captured by the use of personal monitoring — but this remains expensive, intrusive, and time consuming. Therefore, the near-term likelihood is low for obtaining the large quantities of personal exposure data needed for health impact studies. The remaining alternative is the use of stochastic exposure models that make use of hierarchical Bayesian or other "ground-truthing" methods to calibrate exposure modeling with appropriate exposure indicators. For example, Marshall et al. (2005) made use of regional air pollution model combined with driving diaries in the South Coast (Los Angeles area) air basin to adjust exposure estimates to account for mobility and time spent in locations away from home relative to exposures assigned based on residence location.

Biologically relevant timescales

Biologically relevant timescales of exposure to pollutants vary among the different classes of pollutants (e.g., minutes for some VOCs to hours or days or years for PM). Thus, measurement and exposure characterization for these different air pollutants have to be consistent with appropriate biological and dose–response periods.

Timescales for exposure and health effects have often been characterized as acute, subchronic, and chronic (see, e.g., USEPA, 2001). Acute exposures are those that occur over short timescales lasting from a few seconds to a few hours. Examples include exposures under accidental release scenarios, photochemical smog incidents, or incidents such as the London Fog episode. Such episodes are characterized by short-duration and high-pollution concentrations that usually dissipate to background levels after a relatively short period. Effects may be catastrophic, such as those that might occur with an acute carbon monoxide exposure episode or the occurrence of a fatal asthma attack. Less catastrophic effects may result in visits to the hospital emergency department, simple irritation, or no clinically observable effect. Because short-duration, high-impact pollutant concentrations are very difficult to predict, measurement of such exposures is problematic.

Subchronic effects occur over timescales ranging from a few days to about 30 days duration. Synoptic and some seasonal pollution effects may occur over this type of timescale (Barry and Chorley, 1998). Health outcomes of interest in such situations are less well defined than those for acute changes in exposure.

Chronic exposures occur over longer durations up to and including lifetime exposure. The health impacts of interest here are chronic diseases and those, such as cancer, with a long induction period. In such cases, it is of interest to measure exposures integrated over long periods. Such a need suggests alternative measurement strategies differing in intent and, perhaps, instrumentation, from those used in acute investigations.

In developing measurement strategies, the biological outcome of interest must be considered. For example, if acute effects are to be understood properly, then fine time-resolution data must be collected. One solution, albeit an expensive and burdensome one, is to monitor for the contaminant of interest on a real-time basis. This is a costly undertaking for some pollutants in that most of the measurements can be at background levels with only a small number of such measurements truly of concern. On the other end of the timescale, chronic exposures and concomitant health outcomes do not need such detailed time resolution whereas short-term studies of a few days, or even a few weeks, duration do not supply the information needed to assess the risk or impact of a lifetime of exposures. For example, measurement of toxic air pollution exposures for a week during the summer months gives little insight into the lifetime exposure likely experienced by any individual. Further, measurement in a single location, given the mobility of the population (Chapin, 1974; Klepeis et al., 2001), may be insufficient to determine health impacts associated with the "true" exposures.

The development of a measurement strategy for health research studies must account for biologically relevant timescales for data collection. If short-term effects are the target, then real-time data collection may be the only solution. For chronic effects, excursions of short duration, even if quite spectacular, are unlikely to influence long-term exposure. Selecting an optimum measurement or modeling-based exposure prediction strategy has to be based on consideration of the timescale over which the biological effect occurs.

Age-specific exposure issues

Exposure issues are not the same for all ages. As an example, for children and older adults, exposures are quite different than those of younger adults. These issues are important for understanding exposure vulnerability, as well as biological sensitivity or susceptibility of population groups to air pollutants.

Monitoring strategies must take into account the target population. Children, adolescents, adults, and the elderly have vastly different exposure profiles due to different activities they undertake as well as vastly different susceptibility to identical exposures. Although younger children spend the majority of their time in their own homes, older children may spend much more time outdoors. Adults spend a significant fraction of their time in work environments, some with significant occupational exposures (Klepeis et al., 2001). Children, especially the very young, may have incompletely developed immune and neurological systems. For example, exposures to lead in the very young may result in irreversible cognitive impairment whereas similar exposure to an adult would not give rise to such adverse consequences (Needleman, 1993; Muldoon et al., 1996; Lanphear et al., 2005). Similarly, in the elderly, exposure to contaminant concentrations, similar to those of younger adults, may lead to very different outcomes. An elderly person, with reduced lung capacity or a compromised cardiovascular system, may exhibit catastrophic effects from exposure to PM or carbon monoxide at levels that would result in no observable difficulties for a robust young adult. There is a large literature on this topic (Fung et al., 2006; Gong et al., 2005; Martins et al., 2006; Sarnat et al., 2006; Vallejo et al., 2006). The workgroup also considered that an immune-compromised adult may suffer substantially greater adverse effects from exposure to the same level of contaminant than his or her healthy counterpart, but did not identify a study to confirm this.

Health scientists developing plans for an exposure evaluation must recognize the age-specific exposure differences within a population and account for these differences in health studies. This is necessary to protect both the vulnerable members of society as well as those who are robust. A properly designed monitoring program takes this into account. Air pollution standards are often derived to protect sensitive subgroups within the population. These are often children, the elderly, and those with chronic conditions that compromise their ability to mitigate exposure effects. As discussed more fully in the accompanying Symposium paper by Thurston et al. (2008), a number of diseases of emerging importance such as, multiple sclerosis, autism spectrum disorders, and other immunological disorders, pose significant challenges to health researchers due to long latency between exposure and diagnosis.

Exposures to multiple pollutants

In an epidemiologic analysis of air-pollution health effects, the contributions from PM and other co-pollutants (e.g. ozone, NO₂, and combustion-related air toxics) can be interwined. Moreover, differences in PM size and chemical composition are important factors for cross-sectional analysis of PM and co-pollutant health effects. Both the spatial and temporal differences in sources and concentrations of air pollutants of health concern can be quite complex across large metropolitan or urban areas. Within cities, the spatial variability of concentrations of PM and its species near roadways tends to be large. In particular, near-roadway concentrations of ultrafine PM, combustion-related gases, and air toxics are often influenced by mobile source emissions. Human contact with these localized peaks in concentrations tends to occur more during commuting or walking near busy roads. However, homes, schools, or workplaces located near these roads may also experience greater levels of ambient pollutant infiltration indoors. Consequently, source-specific contributions of PM and other pollutants may vary significantly by location (e.g., ambient outdoors, indoors, commuting, etc.), by season, and type of microenvironment. Diurnal or temporal variability in the contributions from different sources of PM and air toxics influence not only the composition of air pollution mixtures in various microenvironments but also their relative toxicity to humans. Human exposures to these pollutants depend strongly on the behavioral patterns of individuals. Therefore, the covariance between these behavioral factors and concentrations of PM, ozone, and air toxics can be quite complex and variable among individuals. Hence, relating stationary outdoor air pollution monitoring data to realistic exposures of individuals or to population exposures to multiple pollutants requires the use of sophisticated measurement and modeling information. Monitoring of multiple pollutants in key outdoor and indoor microenvironments, along with personal exposure measurements, is often needed to generate the necessary information. Typically, these data are then used in conjunction with either land-use regression modeling (Jerrett et al., 2005; Ross et al., 2006) or air quality dispersion modeling (Isakov et al., 2006) in estimating the concentrations of pollutants of interest at different outdoor human receptor locations. Personal exposure estimates may then be produced by using either empirical or mechanistic human exposure models. The mechanistic exposure models, such as the SHEDS model (Burke et al., 2001), incorporate information on sources and concentrations of pollutants in different microenvironments, with corresponding human contact data derived from available time–activity diaries. The application of these more refined exposure estimation methodologies in multi-pollutant health effect studies has been considered more recently (English et al., 1999; Isakov and Özkaynak, 2007).

Tracking effect modifiers

An effect modifier in the context of an epidemiologic health study refers to a variable that influences the magnitude of the association between an exposure measure and the health outcome studied. Some exposure-related factors may also influence either the composition or the toxicity of the selected or indicator pollutant. Some examples of potential effect modifiers that show up in health effect studies for PM and toxic air pollutants are (1) prior or concurrent exposures to air pollutants that are not recorded, (2) spatial and temporal variation in concentrations and/or composition of PM and toxic air pollutants by region, (3) temporal changes in the residential indoor–outdoor air exchanges influencing infiltration of outdoor pollutants indoors, (4) differences among households in air conditioning and window usage, (5) differences among households in combustion appliance type and use, presence of attached garages, consumer product use, exposure to second-hand smoke, and (6) age, occupation, and susceptibility-based exposure differences among the study subjects.

These factors may play an important role in the outcome of the research, tracking, and accountability investigations considered here. In particular, when any of these evaluations are based on either geographic or temporal contrasts between different population groups, the variations in the effect modifiers could influence either the results or the interpretation of study findings. Consequently, there is a need to understand the impacts of these and other likely effect modifiers to assess their contribution to the health end points appropriately across populations, study areas, and time.

Tracking exposures for long-latency outcomes

Some diseases are separated by long latency periods between the environmental exposures that potentially give rise to or promote a disease and the frank presentation of the disease. These diseases provide a particular challenge to epidemiology, health tracking, accountability, and risk assessment. Among the diseases with very long latency periods (of the order of decades) are cancer, amyotrophic lateral sclerosis, and a large number of age-related diseases (heart disease, osteoarthritis, diabetes, skin disorders, eye disorders, etc.). To explore any potential link between onset of these diseases and environmental exposures in early life requires long-term exposure tracking. Many other diseases can have latency periods of the order of years that add to the complexity of constructing any postulated exposure–disease link (Thurston et al., 2008). This is particularly important in cases in which disease incidence might be enhanced by both recent and long-term exposures (years or more cumulative exposures). Examples of diseases in this category include early childhood diseases, development of asthma, autoimmune diseases, autism spectrum disorders, and neurological disorders such as, multiple sclerosis and attention-deficit hyperactivity disorder.

There are a number of options for confronting the complexity of tracking exposures for long-latency outcomes. One approach is to be resourceful in methods used to reconstruct historical exposures. An example is the enormous efforts that have been expended for the Hiroshima/Nagasaki populations with regard to radiation-exposure reconstruction (Marchetti and Straume, 1996). Another approach is to make use of long-lived biomarkers of exposure. Here again the radiation community has set the pace (Lucas, 1997), but there have been efforts of this sort for chemicals (Chen and McKone, 2001). Another option is to make use of early markers of effect that link both to recent exposures and to occurrence of disease later in life or in offspring (Ryan et al., 2007). One example of this is the use of chromosomal aberrations that are key precursors to later disease (Chen and McKone, 2001).

Short-scale exposure-outcome tracking

The etiology of many chronic diseases (such as cardiovascular disease, asthma, and cancer) is complex and involves a mix of genetic and environmental factors interacting with each other over hours, days, months, or years. Recent studies have demonstrated clear links between short-term increases in exposure to PM and increases in incidence of cardiovascular events (Zanobetti and Schwartz, 2005; Pope et al., 2006), changes in cardiopulmonary markers of inflammation (Pope et al., 2004; Rückerl et al., 2006), and in asthma (Gilmour et al., 2006). Efforts to track short-term peaks in exposure to pollutants such as PM, ozone, and toxic air pollutants are constrained both by the technical feasibility of deploying monitoring networks with sufficient scale and coverage to capture the appropriate level of population, spatial, and temporal variation and by the lack of financial resources to support and maintain such a network. Clearly, there is a need to build capacity to deploy, maintain, and track a network of sensors that can track short-term (hours to days) exposure profiles for PM, ozone, and toxic air pollutants.

Confronting Limitations, Barriers, and Challenges

Limitations, barriers, and challenges: PM

The workshop process also identified several issues that impose limitations, barriers, and challenges to efforts to improve studies on the health impacts of PM. Key among these is better understanding of PM speciation and how this relates to health impacts for all types of health impact studies — epidemiology, risk assessment, health tracking, and accountability. Similarly, better resolution of spatial and temporal variations of PM concentrations is important to all types of health studies. This includes variations that derive from land use (such as urban/rural differential and near or far from roadway classifications), seasonal variations, and transient events. For example, the emerging evidence for links between cardiac disease and short-term PM exposure places high value on future studies that match transient events to specific populations such as populations served by a particular hospital (Künzli et al., 2005). This appears to also be relevant to the exacerbation of asthma and the ability to collect hospital data on an exposed population, or in some cases, to use school questionnaires (Mortimer et al., 2004, Roberts et al., 2006).

Limitations, barriers, and challenges: ozone

The workshop process identified several issues that impose limitations, barriers, and challenges to efforts to improve studies on the health impacts of ozone. A key problem is the paucity of methods and opportunities to link medical surveillance to individual measures of exposure. There is a lack of personal exposure information for ozone. Ambient ozone measurements have not been linked well to specific microenvironments resulting in a significant barrier to improving the resolution of exposure–disease links. Because current ozone monitoring programs are motivated by compliance with the ambient standard, and they usually operate from May through September, there are very few ozone data available for characterizing annual exposures. This is a barrier to studies that need data on combined exposures to ozone and PM during winter months. But this barrier is easily overcome by simply monitoring ozone together with PM throughout the year at existing sites. Other challenges for ozone are improving the geographic coverage of exposure data and making upstream links to the ultimate sources that account for observed ozone concentrations. There is also the challenge of making more effective use of models and monitoring data to map out ozone exposures in space and time. Finally, there is the challenge of getting more information about indoor microenvironments — including residences, schools, and vehicles. Ozone concentrations are generally much lower indoors. To address these limitations, one requires more ozone measurements in these particular microenvironments, together with activity-based exposure models that can be calibrated against these data. Without this information, large uncertainties will remain along with the potential for significant exposure misclassification.

Limitations, barriers, and challenges: toxic air pollutants

The workshop process also identified several issues that impose limitations, barriers, and challenges to efforts to improve studies of the health impacts of toxic air pollutants. Differential exposure and how this relates to health impacts are important for all types of health impact studies — epidemiology, risk assessment, health tracking, and accountability. As was the case for PM, better resolution of spatial and temporal variations of toxic air pollutant concentrations is important for all types of health studies. This includes variations that derive from geographical location (such as urban, rural, proximity to roadways or point sources), seasonal variations, and transient events. Because toxic air pollution includes a mixture of different substances, it is important to have better information on how the composition of this mixture may vary among different generic geographic locations (urban, suburban, rural) as well as among different specific locations (Boston, Detroit, Los Angeles, Oakland, etc.). A further challenge in the case of toxic air pollutants is that many of these pollutants reach humans through multiple pathways — making geographic and even site-specific air monitoring data less useful for exposure classification. For example, many toxic air pollutants are emitted to the indoor environment by cooking, smoking, or wood fires (Zhang and Smith, 2003), and some of the larger and more persistent PAHs enter humans primarily through ingestion and the dietary pathway (Lobscheid et al., 2004). In these cases, there is not only the potential for exposure misclassification but also the problem of inconsistencies in exposure estimated or inferred from biomonitoring (e.g., hydroxy-PAH levels in urine) and exposure inferred from ambient measurements. In addition, the uptake and toxicodynamics of air toxics such as PAHs may differ significantly based on whether the PAHs are ingested, inhaled, or absorbed dermally.

Defining and Setting Priorities for Action

The workshop participants not only identified limitations, barriers, and challenges but also identified efforts to overcome these issues, and then set priorities among these efforts. In this section, we summarize priorities for short-term and long-term goals to improve PM, ozone, and toxic air-pollutant exposure data for accountability, air-pollution health effect research, EPHT, and public health surveillance studies.

We consider the recommendations below the first step in a long-term process for improving the contributions of exposure information to health studies. It is important to note that even though the workshop process did set priorities, the participants found themselves constrained by lack of information about the scope of the misclassification problem. Therefore, participants identified the following as their first priorities:

• Evaluate the nature and magnitude of exposure misclassification problem for different types of health effects studies. This requires some exploratory case studies that illustrate how additional exposure information can improve the statistical power to test hypothesized associations in health surveillance research.
• Continue to convene these types of symposia and workgroup efforts. The efforts here are a starting point but cannot end with one symposium.
• The health science community must become more aware of and active in practicing homologous matching — that is matching the space and timescales of pollutants, diseases, and populations. Each pollutant has a characteristic reach and effective lifetime. Human populations also have characteristic geographic regions in which they migrate and temporal activity scales. Diseases also play out with complex time histories. The link between sources of pollutants and any disease outcome requires better sense of how the overlap of spatial and temporal patterns among pollutants, populations, and diseases plays out to make exposure–disease connections.

PM: priority activities

For PM and for the range of health research studies considered here, the long- and short-term strategies include the effective use of models and monitoring. In contrast to ozone and toxic air pollutants, the existing monitoring network for PM has more spatial and temporal resolution. Moreover, there have been many studies of indoor/outdoor relationships for PM exposure. But information on speciation is limited. In the short term, models can be used to make more relevant but less reliable (relative to monitoring data) empirical estimates of the variation of PM speciation among different generic (urban/rural) and specific geographic regions. For example, Lobscheid et al. (2007) have demonstrated that monitoring data for PM2.5, which has more spatial and temporal resolution than does PAH measurement data, can be calibrated to make proxy estimates of PAH composition of the particles.

Accountability and risk assessment can profit from improved modeling along with efforts to capture better both variability and uncertainty in the PM exposure estimates. Air pollution epidemiology and risk assessment will benefit from early efforts to select a limited set of locations for detailed speciation. This effort can identify whether, and how much, exposure difference there is among different generic locations, for example, coastal/inland, near road/away from road. In addition, this effort could also inform both addition exposure research and improved monitoring strategies. Finally, this effort should strive to characterize different geographical scales — from community to urban, regional, and even continental background.

Health tracking studies can benefit from efforts to improve the scope of PM monitoring and exposure tracking. One area that can be addressed in the short term is the need to track exposures during short-term, high-exposure events. For example, rapidly deployable monitors in mobile systems can be used to track PM concentrations during forest fires and agricultural burning. There is also the potential for stratified deployment of monitors for unique events — such as heat waves or weather conditions that trap high PM levels.

In the long term, the broad range of health-research efforts will profit significantly from (i) increasing the number of continuous monitoring sites, (ii) conducting routine ultrafine monitoring at key sites, and (iii) the colocation PM and PM speciation monitoring.

Ozone: priority activities

The workshop participants ranked priority activities in terms of context of the various health research areas — air pollution epidemiology, risk assessment, health tracking, and accountability. In the case of ozone and for the full range of health research studies, a short-term priority is the need to integrate models and limited measurements to provide more information on indoor-environment exposures, and a short- but also longer-term priority is more personal exposure measurements. There is also a need for capturing baseline ambient data throughout the year and in a broader range of environments — not just summertime measurements in urban hot spots.

For air pollution epidemiology, risk assessment and accountability, exposure uncertainty is an important limitation and a barrier to improved classification. The first priority is to characterize and communicate this uncertainty for specific geographic areas. This effort will help researchers understand and confront the reliability of exposure information and the level of confidence placed in these data for health research. Better communicating uncertainty about ozone should also give insight on the best short- and long-term strategies for improving the reliability of exposure information.

For health tracking, the highest priority is expanded monitoring to characterize better ozone exposures for specific communities and sub-populations. This is particularly important for communities whose health status is being monitored as part of an existing community health study. But health tracking can also profit in the short term from systematic integration of modeling and limited monitoring data. For example, the EPA-CDC PHASE project has demonstrated how hierarchical Bayesian methods (Fuentes and Raftery, 2005) can improve the temporal and spatial resolution of air pollution exposure information (Boothe et al., 2005; Haley et al., 2007). But so far, this approach has been retrospective. Accountability and risk assessment are health studies that require these types of efforts for making projections of future exposures.

Toxic air pollutants: priority activities

The need among all health research studies for better spatial and temporal resolution of toxic air-pollutant exposures can be addressed in the short term with modeling improvements but in the long term will require more personal, community, and food monitoring. As in the case of PM, in the short term, improved modeling efforts can provide preliminary, but less reliable improvements in spatial/temporal resolution relative to a long-term effort at personal monitoring. For example, the hierarchical Bayesian method used in the EPA-CDC PHASE project has demonstrated improvement in the spatial and temporal resolution of modeling in a first set of case studies. However, the US-wide spatial/temporal resolution provided by the PHASE approach may be limited in certain geographical areas and for some pollutants, for the purposes of tracking and epidemiology studies. This issue may be addressed either by increasing the spatial resolution in the PHASE process or by employing hybrid modeling approaches, which combine local and regional scale modeling results (e.g., Isakov and Özkaynak, 2007) relevant to each application. The approach used in the EPA-CDC PHASE project or similar approaches could also be extended to toxic air pollutants if routine measurement of toxic air pollutants were to be made. Also, in the short term, models can be used to assess differential exposures attributable to both indoor and dietary sources of toxic air pollutants. In the long-term, there is a need for a systematic effort to monitor and model population variability in cumulative intake of toxic air pollutants from ambient air, indoor air, water, and food. An important resource for this effort is the NHANES biomonitoring data. Although NHANES data do not capture regional variability, they do provide information on the magnitude and variation of exposures to many pollutants including toxic air pollutants. Interpretation of the biomonitoring data can, however, be problematic, especially for those air toxics that have short lifetimes in the body.

An emerging opportunity for increasing the utility of the NHANES data is the development of state-level biomonitoring programs. California has now established a state-level biomonitoring effort and we expect that other states will follow. The ability to contrast biomonitoring data from a national-scale and state or local-scale population provides an important opportunity to detect and understand competing exposure pathways (see for example, McKone et al., 2007).

Overarching Issues

One of the overarching lessons of this symposium is that obtaining better exposure information for disease surveillance, risk assessment, accountability, and tracking requires both setting goals for what is needed and mapping out the pathway from current capabilities to these goals. Meeting these long-term goals requires definition of incremental steps that provide useful information for the interim and move the endeavor toward the long-term goals. Meeting these goals may also require increased resources and research infrastructure — issues that must be addressed by environmental and health agencies, policy makers, and legislatures at both the state and federal level.

The need for better health indicators at the community scale often motivated the types of health research that were considered in this symposium. We recognize that communications between the community members and health researchers is an important aspect of both exposure and health surveillance research. Detailed discussion of these issues and how they relate to exposure information was not the key goal of this symposium, which was more focused on technical improvements. But the symposium participants recognized and noted that, while technical methods and scientific findings inform the health research process, they often fail to provide information that is relevant or meaningful to stakeholders. Work to date has focused on the technical aspects of obtaining and linking data, and less emphasis has been placed on the significance of representing and interpreting such data for nontechnical audiences. This issue relates both to transparency and the need to share information effectively and in a timely manner with other agencies and organizations.

Another overarching theme among the three different pollutants and the different health study approaches is the need for integration among alternate exposure-assessment approaches. For example, different groups may advocate exposure indicators, biomonitoring, mapping methods (e.g., GIS), modeling, environmental media monitoring, and/or personal exposure modeling. However, the workgroup discussions and observations of emerging research reveal that the greatest progress comes from integrations among two or more of these efforts (see e.g., Nuckols et al., 2004; Sohn et al., 2004; McKone et al., 2007).

In the results above, we distinguish the findings among three pollutants — PM, ozone, and toxic air pollutants — and four health research areas — air pollution epidemiology, risk assessment, health tracking, and accountability. This approach is useful for finding important differences. For example, consider the expanding role of models. They are most useful for accountability but helpful also for epidemiology and tracking. Models can be used to detect small changes in exposure and health benefits that an epidemiology study could not find. Researchers must be careful to recognize that, although models can be set up to provide detailed results in space and time, model and parameter uncertainties may limit the reliability of such detailed results.

There are overlapping research needs and opportunities among these different health research strategies — epidemiology, health tracking, and accountability. When these health research strategies are applied to PM, ozone and toxic air pollutants, accurate classification of individual exposures must include adjustments based on a geographical measure of proximity to hazard. Personal measures of exposure will fail to correlate with local source contributions from multiple sources and locations, including indoor sources, unless adjustments are made for contributions from indoor and near-home sources, such as near-roadway sources. There is a need to improve methods for combining the geographic distributions of environmental factors, with information on human population distributions and human activities to identify and evaluate those factors that provide reliable indicators of the magnitude and source of population exposures. Finally, reflecting back on Figure 1, we note the need to consider the continuum of indicators from emissions to exposures to dose to disease to identify better the critical exposure-related information that is most relevant to advancing our knowledge on air-pollution health impacts.

Notes

Disclaimer

The United States Environmental Protection Agency through its Office of Research and Development partially funded and collaborated in the research described here under contract no. EP06D000722 to Dr. P. Barry Ryan and through Interagency Agreement DW-89-93058201-1 with Lawrence Berkeley National Laboratory through the US Department of Energy under Contract Grant no. DE-AC02-05CH11231. It has been subjected to Agency review and approved for publication. This work was also supported by Cooperative Agreement number U19/EH000097-03 from the US Centers for Disease Control and Prevention (CDC).

References

1. Adgate J.L., Eberly L.E., Stroebel C., Pellizzari E.D., and Sexton K. Personal, indoor, and outdoor VOC exposures in a probability sample of children. J Expo Anal Environ Epidemiol 2004: 14: S4–S13. | Article | PubMed | ChemPort |
2. Barry R., and Chorley R. Atmosphere, Weather, and Climate. Rutledge, London and New York, 1998.
3. Bates D.V. Adverse health impacts of air pollution—Continuing problems. Scandi J Work Environ Health 1995: 21(6): 405–411. | ChemPort |
4. Bennett D.H., and Koutrakis P. Determining the infiltration of outdoor particles in the indoor environment using a dynamic model. J Aerosol Sci 2006: 37(6): 766–785. | Article | ChemPort |
5. Bernstein J.A., Alexis N., Barnes C., Bernstein I.L., Bernstein J.A., Nel A., Peden D., Diaz-Sanchez D., Tarlo S.M., and Williams P.B. Health effects of air pollution. J Allergy Clin Immunol 2004: 114(5): 1116–1123. | Article | PubMed |
6. Boothe V., Dimmick W.F., and Talbot T.O. Relating air quality to environmental public health tracking data. In: Aral M.M., Brebbia, C.A., Maslia, M.L., and Sinks, T. (Eds.). Environmental Exposure and Health. Wessex Institute Transactions on Ecology and the Environment, WIT Press: Southhampton, UK, 2005: 85, pp 43–52. | ChemPort |
7. Burke J.M., Zufall M.J., and Özkaynak H. A population exposure model for particulate matter: case study results for PM2.5 in Philadelphia, PA. J Expo Anal Environ Epidemiol 2001: 11: 470–489. | Article | PubMed | ISI | ChemPort |
8. Burnett R.T., Cakmak S., Brook J.R., and Krewski D. The role of particulate size and chemistry in the association between summertime ambient air pollution and hospitalization for cardiorespiratory diseases. Environ Health Perspect 1997: 105(6): 614–620. | Article | PubMed | ISI | ChemPort |
9. Catalano P.J., Ryan L.M., and Kaden D.A. Statistical design aspects of the NTP/HEI collaborative study on the health effects of chronic ozone inhalation. Inhal Toxicol 1996: 8(3): 229–249. | Article | ChemPort |
10. Chapin F. Human Activity Patterns in the City. John Wiley & Sons, New York, 1974.
11. Chen W.C., and McKone T.E. Chronic health risks from aggregate exposures to ionizing radiation and chemicals: scientific basis for an assessment framework. Risk Anal 2001: 21(1): 25–42. | Article | PubMed | ChemPort |
12. Delfino R.J. Epidemiologic evidence for asthma and exposure to air toxics: linkages between occupational, indoor, and community air pollution research. Environ Health Perspect 2002: 110(Suppl 4): 573–589. | PubMed | ChemPort |
13. English P., Neutra R., Scalf R., Sullivan M., Waller L., and Zhu L. Examining associations between childhood asthma and traffic flow using a geographic information system. Environ Health Perspect 1999: 107(9): 761–767. | Article | PubMed | ISI | ChemPort |
14. Eschenbacher W.L., Holian A., and Campion R.J. Air toxics and asthma: impacts and end points. Environ Health Perspect 1995: 103(Suppl 6): 209–211. | Article | PubMed |
15. Finlayson-Pitts B., and Pitts J.N. Chemistry of the Upper and Lower Atmopshere: Theory, Experiments, and Applications. Academic Press, San Diego, 2000.
16. Fuentes M., and Raftery A.E. Model evaluation and spatial interpolation by Bayesian combination of observations with outputs from numerical models. Biometrics 2005: 61: 36–45. | Article | PubMed |
17. Fung K.Y., Khan S., Krewski D., and Chen Y. Association between air pollution and multiple respiratory hospitalizations among the elderly in Vancouver, Canada. Inhal Toxicol 2006: 18(13): 1005–1011. | Article | PubMed | ChemPort |
18. Geller M.D., Chang M.H., Sioutas C., Ostro B.D., and Lipsett M.J. Indoor/outdoor relationship and chemical composition of fine and coarse particles in the southern California deserts. Atmos Environ 2002: 36(6): 1099–1110. | Article | ISI | ChemPort |
19. Geyh A.S., Xue J.P., Özkaynak H., and Spengler J.D. 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(3): 265–270. | Article | PubMed | ISI | ChemPort |
20. Gilmour I.M., Jaakkola M.S., London S.J., Andre E., Nel A.E., and Rogers C.A. How exposure to environmental tobacco smoke, outdoor air pollutants, and increased pollen burdens influences the incidence of asthma. Environ Health Perspect 2006: 114: 627–633. | PubMed |
21. Gong Jr H., Linn W.S., Clark K.W., Anderson K.R., Geller M.D., and Sioutas C. Respiratory responses to exposures with fine particulates and nitrogen dioxide in the elderly with and without COPD. Inhal Toxicol 2005: 17(3): 123–132. | Article | PubMed | ChemPort |
22. Hackney J.D., Linn W.S., Buckley R.D., and Hislop H.J. Studies in adaption to ambient oxidant air pollution: effects of ozone exposure in Los Angeles residents vs. new arrivals. Environ Health Perspect 1976: 18: 141–146. | Article | PubMed | ChemPort |
23. Hackney J.D., Linn W.S., Karuza S.K., Buckley R.D., Law D.C., Bates D.V., Hazucha M., Pengelly L.D., and Silverman F. Efects of ozone exposure in Canadians and Southern Californians. Evidence for adaptation? Arch Environ Health 1977: 32(3): 110–116. | PubMed | ChemPort |
24. Haley V., Bateson T., Boothe V., Talbot T., Dimmick F., Paulu C., and Warner M., Technical Background Report: Case-Crossover Analysis of Acute Health Effects and Ambient Air Quality for Environmental Public Health Tracking. Submitted to U.S. Centers for Disease Control and Prevention, Environmental Health Tracking Branch, Division of Environmental Hazards and Health Effects, Atlanta, GA January 9, 2007.
25. Hayes S.R. Estimating the effect of being indoors on total personal exposure to outdoor air pollution. JAPCA 1989: 39(11): 1453–1461. | PubMed | ChemPort |
26. Isakov V., Graham S., Burke J., and Özkaynak H. Linking air quality and exposure models. EM Magazine 2006, 26–29.
27. Isakov V, and Özkaynak H A modeling methodology to support evaluation of public health impacts of air pollution reduction programs. Proceedings of the 29th International Technical Meeting on Air Pollution Modeling, Aveiro, Portugal, Sept 24-28, 2007.
28. Jerrett M., Arain A., Kanaroglou P., Beckerman B., Potoglou D., Sahsuvaroglu T., Morrison J., and Giovis C. A review and evaluation of intraurban air pollution exposure models. J Expo Anal Environ Epidemiol 2005: 15(2): 185–204. | Article | PubMed | ChemPort |
29. Kinney P.L., Chillrud S.N., Ramstrom S., Ross J., and Spengler J.D. Exposures to multiple air toxics in New York City. Environ Health Perspect 2002: 110(Suppl. 4): 539–546. | PubMed | ISI | ChemPort |
30. Kinney P.L., and Lippmann M. Respiratory effects of seasonal exposures to ozone and particles. Arch Environ Health 2000: 55(3): 210–216. | PubMed | ChemPort |
31. Klepeis N.E., Nelson W.C., Ott W.R., Robinson J.P., Tsang A.M., Switzer P., Behar J.V., Hern S.C., and Engelmann W.H. The National Human Activity Pattern Survey (NHAPS): A resource for assessing exposure to environmental pollutants. J Expo Anal Environ Epidemiol 2001: 11(3): 231–252. | Article | PubMed | ISI | ChemPort |
32. Künzli N., Jerrett M., Mack W.J., Beckerman B., LaBree L., Gilliland F., Thomas D., Peters J., and Hodis H.N. Ambient air pollution and atherosclerosis in Los Angeles. Environ Health Perspect 2005: 113(2): 201–206. | PubMed | ChemPort |
33. Lanphear B.P., Hornung R., Khoury J., Yolton K, Baghurst P., Bellinger D.C., Canfield R.L., Dietrich K.N., Bornschein R., Greene T., Rothenberg S.J., Needleman H.L., Schnaas L., Wasserman G., Graziano J., and Roberts R. Low-level environmental lead exposure and children's intellectual function: an international pooled analysis. Environ Health Perspect 2005: 113(7): 894–899. | PubMed | ChemPort |
34. Last J.M. A Dictionary of Epidemiology. International Epidemiological Association Inc., New York, 2001.
35. Lee K., Parkhurst W.J., Xue J.P., Özkaynak A.H., Neuberg D., and Spengler J.D. Outdoor/indoor/personal ozone exposures of children in Nashville, Tennessee. J Air Waste Manage Assoc 2004: 54(3): 352–359. | ChemPort |
36. Lee K., Vallarino J., Dumyahn T., Özkaynak H., and Spengler J.D. Ozone decay rates in residences. J Air Waste Manage Assoc 1999: 49(10): 1238–1244. | ChemPort |
37. Linn W.S., Avol E.L., Shamoo D.A., Peng R.C., Valencia L.M., Little D.E., and Hackney J.D. Repeated laboratory ozone exposures of volunteer Los Angeles residents: an apparent seasonal variation in response. Toxicol Ind Health 1988: 4(4): 505–520. | PubMed | ChemPort |
38. Linn W.S., Buckley R.D., Spier C.E., Blessey R.L., Jones M.P., Fischer D.A., and Hackney J.D. Health effects of ozone exposure in asthmatics. Am Rev Respir Dis 1978: 117(5): 835–843. | PubMed | ChemPort |
39. Linn W.S., Fischer D.A., Medway D.A., Anzar U.T., Spier C.E., Valencia L.M., Venet T.G., and Hackney J.D. Short-term respiratory effects of 0.12 ppm ozone exposure in volunteers with chronic obstructive pulmonary disease. Am Rev Respir Dis 1982: 125(6): 658–663. | PubMed | ChemPort |
40. Linn W.S., Shamoo D.A., Venet T.G., Spier C.E., Valencia L.M., Anzar U.T., and Hackney J.D. Response to ozone in volunteers with chronic obstructive pulmonary disease. Arch Environ Health 1983: 38(5): 278–283. | PubMed | ChemPort |
41. Lippmann M., and Schlesinger R.B. Toxicological bases for the setting of health-related air pollution standards. Annu Rev Public Health 2000: 21: 309–333. | Article | PubMed | ChemPort |
42. Lobscheid A.B., Maddalena R.L., and McKone T.E. Contribution of locally grown foods in cumulative exposure assessments. J Expo Anal Environ Epidemiol 2004: 14(1): 60–73. | Article | PubMed | ChemPort |
43. Lobscheid A.B., and McKone T.E. Constraining uncertainties about the sources and magnitude of polycyclic aromatic hydrocarbon (PAH) levels in ambient air: the State of Minnesota as a case study. Atmos Environ 2004: 38(33): 5501–5515. | Article | ChemPort |
44. Lobscheid A.B., McKone T.E., and Vallero D. Exploring relationships between outdoor air particulate-associated polycyclic aromatic hydrocarbon and PM2.5: a case study of benzo(a)pyrene in California metropolitan regions. Atmos Environ 2007: 41: 5659–5672. | Article | ChemPort |
45. Lucas J.N. Dose reconstruction for individuals exposed to ionizing radiation using chromosome painting. Radiat Res 1997: 148(5 Suppl): S33–S38. Supplement: Space Radiation Damage and Biodosimetry. | Article | PubMed | ChemPort |
46. Marchetti A.A., and Straume T. A search for neutron reactions that may be useful for Hiroshima dose reconstruction. Appl Radiat Isot 1996: 47(1): 97–103. | Article | ChemPort |
47. Marshall J.D., McKone T.E., Deakin E.A., and Nazaroff W.W. Inhalation of motor vehicle emissions: effects of urban population and area. Atmos Environ 2005: 39(2): 283–295. | Article | ChemPort |
48. Martins L.C., Pereira L.A., Lin C.A., Santos U.P., Prioli G., Luiz Odo C., Saldiva P.H., and Braga A.L. The effects of air pollution on cardiovascular diseases: lag structures. Revista de Saude Publica 2006: 40(4): 677–683. | PubMed |
49. McConnell R., Berhane K., Gilliland F., London S.J., Islam T., Gauderman W.J., Avol E., Margolis H.G., and Peters J.M. Asthma in exercising children exposed to ozone: a cohort study.[see comment][erratum appears in Lancet 2002 Mar 9;359(9309):896]. Lancet 2002: 359(9304): 386–391. | Article | PubMed | ChemPort |
50. McKone T.E., Castorina R., Kuwabara Y., Harnly M.E., Eskenazi B., and Bradman A. Merging models and biomonitoring data to characterize sources and pathways of human exposure to organophosphorous pesticides in the Salinas Valley of California. Environ Sci Technol 2007: 41: 3233–3240. | Article | PubMed | ChemPort |
51. Mortimer K., Matovinovic E., Lurmann F., Funk T., Balmes J.R., Mann J., Margolis H., Hammond S.K., and Tager I.B. Exposure to traffic and the risk of atopy in asthmatic children. Epidemiology 2004: 15(4): S66–S67. | Article |
52. Muldoon S.B., Cauley J.A., Kuller L.H., Morrow L., Needleman H.L., Scott J., and Hooper F.J. Effects of blood lead levels on cognitive function of older women. Neuroepidemiology 1996: 15(2): 62–72. | Article | PubMed | ChemPort |
53. NARSTO NARSTO Measurement Methods Compendium: Ozone Methods. "Issued in 1999 and available from 1999: http://narsto.esd.ornl.gov/Compendium/methods/o3.shtml [Accessed 16 March 2007].
54. Needleman H.L. The current status of childhood low-level lead toxicity. Neurotoxicology 1993: 14(2-3): 161–166. | PubMed | ChemPort |
55. Nuckols J.R., Ward M.H., and Jarup L. Using geographic information systems for exposure assessment in environmental epidemiology studies. Environ Health Perspect 2004: 112(9): 1007–1015. | PubMed | ISI |
56. Ostro B.D. Examining acute health outcomes due to ozone exposure and their subsequent relationship to chronic disease outcomes. Environ Health Perspect 1993: 101: 213–216. | Article | PubMed |
57. Özkaynak H., Palma T., Touma J.S., and Thurman J. Modeling population exposures to outdoor sources of hazardous air pollutants. J Exp Sci Environ Epidemiol 2008a: 18(1): 45–58. | Article | ChemPort |
58. Özkaynak H., Glenn B., Qualters J.R., Strosnider H., McGeehin M.A., and Zenick H. Summary and findings of the EPA and CDC Symposium on Air Pollution Exposure and Health. J Exp Sci Environ Epidemiol 2008b (in press).
59. Pope III C.A., Hansen M.L., Long R.W., Nielsen K.R., Eatough N.L., Wilson W.E., and Eatough D.J. Ambient particulate pollution, heart rate variability, and blood markers of inflammation in a panel of elderly subjects. Environ Health Perspect 2004: 112: 339–345. | PubMed | ISI |
60. Pope III C.A., Muhlstein J.B., May H.T., Renlund D.G., Anderson J.L., and Horne B.D. Ischemic heart disease events triggered by short-term exposure to fine particulate air pollution. Circulation 2006: 114: 2443–2448. | Article | PubMed | ChemPort |
61. Roberts E.M., English P.B., Wong M., Wolff C., Valdez S., Van den Eeden S.K., and Ray G.T. Progress in pediatric asthma surveillance II: geospatial patterns of asthma in Alameda County, California. Prev Chronic Dis 2006: 3(3): A92. | PubMed |
62. Ross Z., English P.B., Scalf R., Gunier R., Smorodinsky S., Wall S., and Jerrett M. Nitrogen dioxide prediction in Southern California using land use regression modeling: potential for environmental health analyses. J Exp Sci Environ Epidemiol 2006: 16(2): 106–114. | Article | ChemPort |
63. Rückerl R., Ibald-Mulli A., Koenig W., Woelke G., Cyrys J., Heinrich J., Cyrys J., Heinrich J., Marder V., Frampton M., Wichmann H.E., and Peters A. Air pollution and markers of inflammation and coagulation in patients with coronary heart disease. Am J Respir Crit Care Med 2006: 173: 432–441. | Article | PubMed |
64. Peters J.M., Avol E., Gauderman W.J., Linn W.S., Navidi W., London S.J., Margolis H., Rappaport E., Vora H., Gong Jr H., and Thomas D.C. A study of twelve Southern California communities with differing levels and types of air pollution. II. Effects on pulmonary function. Am J Respir Crit Care Med 1999: 159(3): 768–775. | PubMed | ISI | ChemPort |
65. Phillips M.L., Esmen N.A., Hall T.A., and Lynch R. Determinants of exposure to volatile organic compounds in four Oklahoma cities. J Expo Anal Environ Epidemiol 2005: 15(1): 35–46. | Article | PubMed | ChemPort |
66. Rojas-Bracho L., Suh H.H., Oyola P., and Koutrakis P. Measurements of children's exposures to particles and nitrogen dioxide in Santiago, Chile. Sci Total Environ 2002: 287(3): 249–264. | Article | PubMed | ISI | ChemPort |
67. Romieu I., Lugo M.C., Colome S., Garcia A.M., Avila M.H., Geyh A., Velasco S.R., and Rendon E.P. Evaluation of indoor ozone concentration and predictors of indoor-outdoor ratio in Mexico City. J Air Waste Manage Assoc 1998: 48(4): 327–335. | ChemPort |
68. Ruidavets J.B., Cournot M., Cassadou S., Giroux M., Meybeck M., and Ferrieres J. Ozone air pollution is associated with acute myocardial infarction. Circulation 2005: 111(5): 563–569. | Article | PubMed | ChemPort |
69. Ryan P.B., Burke T.A., Cohen Hubal E.A., Cura J.J., and McKone T.E. Using biomarkers to inform cumulative risk assessment. Environ Health Perspect 2007: 115(5): 833–840. | PubMed | ChemPort |
70. Ryan P.B., Soczek M.L., Treitman R.D., Spengler J.D., and Billick I.H. The Boston residential NO2 characterization study: II. Survey methodology and population concentration estimates. Atmos Environ 1988: 22: 2115–2125. | Article | ISI | ChemPort |
71. Sarnat S.E., Suh H.H., Coull B.A., Schwartz J., Stone P.H., and Gold D.R. Ambient particulate air pollution and cardiac arrhythmia in a panel of older adults in Steubenville, Ohio. Occup Environ Med 2006: 63(10): 700–706. | Article | PubMed | ChemPort |
72. Seinfeld J.H., and Pandos S.N. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. John Wiley & Sons Inc, New York, 1998.
73. Serrano-Trespalacios P.I., Ryan L., and Spengler J.D. Ambient, indoor and personal exposure relationships of volatile organic compounds in Mexico City Metropolitan Area. J Expo Anal Environ Epidemiol 2004: 14: S118–S132. | Article | PubMed | ChemPort |
74. Singer B.C., Coleman B.K., Destaillats H., Hodgson A.T., Weschler C.J., and Nazaroff W.W. Indoor secondary pollutants from cleaning product and air freshener use in the presence of ozone. Atmos Environ 2006: 40: 6696–6710. | Article | ChemPort |
75. Son B., Breysse P., and Yang W. Volatile organic compounds concentrations in residential indoor and outdoor and its personal exposure in Korea. Environ Int 2003: 29(1): 79–85. | Article | PubMed | ChemPort |
76. Sohn M.D., McKone T.E., and Blancato J.N. Reconstructing population exposures from dose biomarkers: Inhalation of trichloroethylene (TCE) as a case study. J Expo Anal Environ Epidemiol 2004: 14(3): 204–213. | Article | PubMed | ChemPort |
77. Spengler J.D., Schwab M., Ryan P.B., Colome S.D., Wilson A.L., Billick I.H., and Becker E. Personal exposure to nitrogen dioxide in the Los Angeles Basin: study design and results. J Air Waste Manage Assoc 1994: 44: 39–47. | PubMed | ChemPort |
78. Spiro T.G., and Sligliani W.M. Chemistry of the Environment. Prenctice-Hall Inc., Upper Saddle River, New Jersey, 2003.
79. Thurston G.D., Bekkedal M.Y.V., Roberts E., Ito K., Pope A.C., Glenn B., Özkaynak H., and Utell M.J. Emerging Health and Air Quality Issues for Air Pollution Epidemiology, Surveillance and Accountability. J Exp Sci Environ Epidemiol 2008 (in press).
80. USEPA. Determination of Ozone by Ultraviolet Analysis, 1997. Issued in 1997 and available from http://nsdi.epa.gov/ttn/amtic/files/ambient/qaqc/ozone4.pdf [Accessed 16 March 2007].
81. USEPA. 2001 Introduction to Air Toxics Risk Assessment. Issued 2001 and available from rieved April 4, 2007, from http://www.epa.gov/ttn/atw/wks/appx1011.pdf [Accessed 4 April 2007].
82. USEPA. 2004 Criteria Document for Particulate Matter. Issued 2004 and available from http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_cd.html [Accessed 16 March 2007].
83. Vallejo M., Ruiz S., Hermosillo A.G., Borja-Aburto V.H., and Cardenas M. Ambient fine particles modify heart rate variability in young healthy adults. J Expo Sci Environ Epidemiol 2006: 16(2): 125–130. | Article | PubMed | ChemPort |
84. Wallace L.A., and Pellizzari E.D. Recent advances in measuring exhaled breath and estimating exposure and body burden for volatile organic compounds (VOCs). Environ Health Perspect 1995: 103(Suppl 3): 95–98. | Article | PubMed | ChemPort |
85. Weisel C.P. Assessing exposure to air toxics relative to asthma. Environ Health Perspect 2002: 110(Suppl 4): 527–537. | PubMed | ChemPort |
86. Xue J., Liu S.V., Özkaynak H., and Spengler J.D. Parameter evaluation and model validation of ozone exposure assessment using Harvard Southern California Chronic Ozone Exposure Study data. J Air Waste Manage Assoc 2005: 55(10): 1508–1515. | ChemPort |
87. Zanobetti A., and Schwartz J. The effect of particulate air pollution on emergency admissions for myocardial infarction: a multicity case-crossover analysis. Environ Health Perspect 2005: 113: 978–982. | PubMed |
88. Zhang J., and Smith K.R. Indoor air pollution: a global health concern. Br Med Bull 2003: 68: 209–225. | Article | PubMed | ChemPort |

Acknowledgements

We are grateful to the excellent contributions made by the symposium attendees who participated in the exposure break-out group discussions on the second day of the meeting.