PROJECT DESCRIPTION

1. OVERALL OBJECTIVES

The Center at Harvard will address key scientific issues regarding the health effects of ambient particles. The specific aims of the Center reflect the National Research Council's ten highest research priorities for ambient particle research (NRC, 1998). To meet these objectives, the Center will focus on the following three research themes: Exposure, Susceptibility, and Biological Mechanisms/Dosimetry. Theme I will investigate human exposures to particles and gaseous co-pollutants in order to differentiate the health effects of particles from outdoor and indoor sources. This theme also will quantify the effect of exposure error for fine particles and their co-pollutants on risk estimates from epidemiological studies. Theme II will use innovative methods to identify individuals who are sensitive to the effects of air pollution, assess whether these individuals are "harvested" by air pollution episodes, and measure the effect of chronic air pollution exposure on the development of chronic diseases. Through studies of animal and human subjects, Theme III will identify the particulate components, or characteristics, and gaseous air pollutants that trigger adverse health effects, as well as differentiate biological mechanisms that may lead to fatal outcomes. Collectively, our proposed projects will address eight out of the ten research priorities included in the NRC report.

By building the Center around the three defined research themes, we will maintain both a common focus and an integrated approach, which will enable us to address key issues relating to the health effects of ambient particles. These three themes will include projects that span several disciplines in which our investigators have expertise. The particular strengths of our research groups include: exposure assessment (Drs. Koutrakis, Suh, and Spengler), particle physics and chemistry (Drs. Koutrakis and Wolfson), epidemiology (Drs. Schwartz, Dockery, Speizer, Gold, and Christiani), dosimetry (Drs. Tsuda and Butler), physiology (Drs. Brain, Godleski, and Butler), cardiology (Drs. Verrier and Stone), pulmonary health (Drs. Speizer, Gold), toxicology (Drs. Godleski, Kobzik, and Kelsey), biostatistics (Drs. Ware and Catalano), risk assessment and public policy (Drs. Evans, Graham, and Hammitt), and air microbiology (Dr. Burge). This investigative group has been collaborating on particle health effects research for more than a decade.

The overall strategy of this Center is to build upon both our previous and ongoing research on particle health effects. This will enable us to maximize the use of data and resources in order to obtain the most useful scientific information to meet our objectives. We will use data from our previous epidemiological studies, as well as personal exposure measurements from our current and future field investigations as the basis for certain projects. The mechanistic and dosimetric studies will be conducted in conjunction with our future inhalation studies. As a result, the ambitious research portfolio outlined in this proposal will be timely and cost-effective.

An overview of the Center for research on ambient particulate matter at Harvard is presented on the following page. As illustrated, the Research Themes and Projects are derived from the current NRC Research Priorities. In turn, the scientific information generated by Harvard and other Centers will serve as a basis for the EPA to establish new public health policies, and for the NRC to redefine particle research needs (Development and Utilization of Scientific Information). As a result of this dynamic process, our research agenda may be redirected to respond to future research needs by refocusing existing projects or developing new ones, as addressed in the Research Coordination and Integration Core. This will ensure that our Center always will be focused on the most critical scientific issues regarding particle health effects.

Figure 1.

2.APPROACH

THEME I: ASSESSING PARTICLE EXPOSURES FOR HEALTH EFFECTS STUDIES

The primary focus of Theme I is to assess human exposures to particles and gaseous co-pollutants in order to better understand their heath effects. As such, research conducted as part of Theme I will contain four main objectives: (1) to characterize the inter- and intra- variability in personal particulate and gaseous exposures (NRC 1, 7); (2) to identify factors affecting the relationship between personal exposures and outdoor levels (NRC 1); (3) to determine the contribution of outdoor and indoor particles to personal particulate exposures (NRC 1, 5); (4) to quantify the effect of measurement error for fine particles and their co-pollutants (coarse mass and the criteria gases) on risk estimates from epidemiological studies (NRC 7, 10); and (5) to differentiate the health effects of particles from outdoor and indoor sources (NRC 5). These objectives will be met by three inter-related research projects, which will make use of our data base of personal, indoor, and outdoor particulate and gaseous exposures.

Outdoor concentrations consistently have been shown to be a relatively poor measure of personal PM10 and PM2.5 exposures. In the Particle Total Exposure Assessment Methodology (PTEAM) study, for example, daytime personal PM10 exposures were found to be 50% higher on average than corresponding ambient levels (Clayton et al., 1993; Thomas et al., 1993), while the Harvard Six Cities study found mean personal PM10 exposures to be more than 100% greater than mean ambient levels (Spengler et al., 1985). In a summer pilot study conducted in Nashville, TN, we showed such personal/outdoor concentration differences for individuals with chronic obstructive pulmonary disease (COPD), a cohort identified by epidemiological studies to be at increased risk of particulate-associated morbidity and mortality (Bahadori et al., 1996). In a follow-up study in Boston, the association between personal exposures and outdoor concentrations was found to be individual-specific. When analyzed longitudinally by individual, the association between personal and outdoor PM2.5 exposures varied widely, with significant associations for approximately half of the monitored individuals (Figure 2). Comparisons of personal PM2.5exposures and indoor concentrations showed significant associations for a slightly larger percentage of the monitored individuals; however, the distribution of the R2 values for these comparisons was similar to that for outdoor concentrations.

Figure 2.

These findings are in marked contrast to those obtained from our earlier study of children's exposures to sulfates (SO42-) (Suh et al., 1993), an important constituent of fine particle mass that lacks indoor sources and thus can be used as a tracer of outdoor PM2.5. Individual comparisons of personal and outdoor concentrations for SO42- yielded R2values that were substantially higher and less variable than those for personal-outdoor and personal-indoor comparisons for PM2.5 (Figure 2). Therefore, these high R2 values for SO42- suggest that outdoor PM2.5 concentrations may be an appropriate measure of exposures to fine particles which originate from outdoors.

Our research projects will examine the validity of outdoor PM2.5 as an exposure measure (Project Ib) with specific emphasis on the contributions of outdoor and indoor particles to personal PM2.5 and PM10 exposures (Project Ia) and their health effects (Project Ic). All three projects will use SO42- as a tracer of outdoor particles and will use statistical and physico-chemical modeling techniques (Koutrakis et al., 1992). As part of these efforts, the projects also will test the alternative hypothesis that exposures to indoor source-related particles contribute to the resultant health effects. The three research projects rely upon our existing exposure database, which includes data from studies funded by the U.S. EPA, HEI, EPRI, and API (Table 1, shown on the next page). The particle exposures in these studies were measured in urban environments characterized by both diverse particle compositions and meteorological conditions (Boston, MA, Nashville, TN, Atlanta, GA, Baltimore, MD, and Los Angeles, CA). Data from these studies will provide a comprehensive database that will be leveraged by the Center to test the specific hypotheses of Projects Ib and Ic that could not be gleaned from these studies individually.

Project Ia: Assessing Human Exposures to Particulate and Gaseous Air Pollutants
Background and significance.
Outdoor PM10 and PM2.5 concentrations have consistently been shown to be relatively poor predictors of personal exposures, as the relationship between personal PM10 and PM2.5 exposures and outdoor concentrations has been found to exhibit substantial inter- and intra-personal variability (Clayton et al., 1993, Thomas et al., 1993, Bahadori et al., 1996). This inter- and intra-personal variability has been attributed to differences in activity patterns and home characteristics, such as air exchange rates and indoor source emissions (Bahadori et al., 1996). Of these factors, indoor emissions from sources such as cooking and cleaning have the most direct impact on the personal- outdoor relationship. Indoor emissions often result in higher indoor as compared to outdoor particulate levels and in indoor levels that are uncorrelated with those outdoors (Abt et al., 1997). These indoor/outdoor concentration differences are largely responsible for the observed variability in the relationship between personal exposures and outdoor concentrations. Since these differences are less pronounced for particles that originate outdoors (Suh et al., 1993, Suh et al., 1994), the association between personal exposures and outdoor concentrations should be stronger for particles that originate from outdoor as compared to indoor sources.

Table 1.

Study design. In this project, we will use a comprehensive data base (Table 1) to test the hypothesis that personal particle exposures are higher than the corresponding outdoor concentrations due to the contribution of microenvironmental sources. These data will include several thousand simultaneous outdoor, indoor, and personal PM2.5 measurements that were obtained from diverse geographical locations for several potentially sensitive subgroups (including those with COPD, myocardial infarctions (MI), senior citizens, children, and asthmatics). In many of these studies, measurements of PM2.5 were or will be made simultaneously with those of PM10 and the gases CO, SO2, NO2, and O3. Many of the collected particulate samples were or will be analyzed for elemental concentrations using X-Ray Fluorescence (XRF). Information on air exchange rates, home characteristics, and time/activity patterns were or will be collected for each individual. Symptom diaries will be collected for the indoor/outdoor particle health effects study (Project Ic). The study designs, protocols, quality assurance narratives, time activity diaries, questionnaires, and human subject committee approvals for these studies are well established and are available on request.

Data analysis. Personal and indoor concentrations of outdoor source particles will be determined for this analysis using SO42- (estimated as the sulfur concentration determined by XRF) as a tracer of outdoor pollution. SO42- is an ideal tracer of outdoor particles since it is a major constituent of outdoor PM2.5 in the Eastern United States and it lacks indoor sources. The ratio of indoor to outdoor SO42- will be assumed to equal the effective penetration rate of all fine particles from the outdoors. Similarly, the ratio of personal to outdoor SO42- will be assumed to equal the contribution of outdoor PM2.5 to personal exposures. The outdoor PM2.5concentration will be multiplied by the respective ratios to estimate the contribution of outdoor particles to indoor levels and personal exposures. Any remaining indoor or personal PM2.5 will be assumed to originate from indoor sources.

The central tendency and variability in the contribution of outdoor particles to both personal and indoor concentrations will be examined using descriptive statistics (mean, standard deviation, distribution percentiles) and graphical displays. Interest will focus on how this contribution varies by season (subject-based data), within season (subject-based data), city, and sampling duration (12- vs. 24-hour). These analyses will be used to reveal general trends in the data as well as to indicate whether the concentration data is skewed, suggesting the need for data transformation or non-parametric statistical approaches. Personal exposure levels will be compared for particles of outdoor and indoor origin and for gases across seasons using t-tests or their non-parametric equivalents. This analysis will specifically examine the influence of activity pattern and housing ventilation characteristics on seasonal variability in personal exposures.

The relationship between personal exposures and corresponding outdoor levels will be evaluated for particles of outdoor and indoor origin and for gases using Spearman correlation coefficients, linear and multivariate regression techniques, and generalized linear modeling techniques. For these analyses, data will be analyzed stratified by city, season, and sensitive sub-group. These variables, as well as activity patterns and housing characteristics, will be included as parameters in the regression, generalized linear, and mixed models to determine their influence on personal exposures and their relationship to outdoor concentrations. Generalized estimating equations and mixed models (Diggle et al., 1994) will be used to account for correlations that result from repeated measurements. These models will be used to determine the inter- and intra-variability in the relationship between personal exposures and outdoor concentrations for both particles and gases.

General project information. This project will be led by Drs. Koutrakis and Suh, both of whom have extensive experience in particle exposure assessment and are the Principal Investigators of the studies summarized in Table 1. Dr. Catalano, who works closely with Drs. Koutrakis and Suh on the analysis of our exposure data sets, also will participate in this analysis. We will start this project immediately and expect to complete our analyses within the first three years.

Project Ib: Quantifying Exposure Error and its Effect on Epidemiological Studies
Background and significance.
Much of the controversy surrounding epidemiological studies of particles is focused on exposure error and its effect on the observed associations between exposure and effect. This controversy is centered primarily on two aspects of exposure error: (1) the use of outdoor concentrations measured at the SAM site to estimate exposures in epidemiological studies; and (2) differences in measurement error for PM2.5 as compared to other pollutants. Both may impact the observed exposure-effect associations; however, there is wide disagreement about the magnitude and the direction of their impact. This disagreement has been difficult to resolve, since little to no quantitative information exists about the magnitude and variability of the exposure error for PM2.5 and its co- pollutants (coarse mass, CO, SO2, NO2, and O3).

Study design. The hypothesis to be tested in this project is that exposure errors cause upward bias in the estimates of the association between exposure and effect. The project will use data collected in our previous exposure and measurement studies. These data sets will be used to characterize four sources of exposure error: instrument error, spatial variation in ambient PM2.5, indoor/outdoor differences, and personal factors. Instrument error will be examined using data from our Fine Particulate Sampling Methodology study, which examined particle mass losses due to volatilization from filters. This study was conducted in five cities (Boston, MA, Chicago, IL, Riverside, CA, Dallas, TX, and Phoenix, AZ), each of which is characterized by different particle compositions and climates. PM2.5 levels were measured using several sampling techniques with approximately 40 sets of PM2.5 samples collected in each city. Error due to spatial variation will be examined using data from our Metropolitan Acid Aerosol Characterization study conducted in four eastern U.S. cities (Philadelphia, PA, Washington, DC, Boston, MA, and Nashville, TN). In each city, 24-h PM2.5and PM10 concentrations were measured either daily or on alternate days at multiple SAM sites located throughout the city (Suh et al., 1995, Burton et al., 1996, Suh et al., 1997, Wilson and Suh, 1997). Error due to indoor/outdoor differences and personal factors will be analyzed using data from our exposure studies (Table 1) and the EPA-sponsored PTEAM study. Based on this combined data base: (1) sources of exposure error will be characterized; (2) the city- and population-specific effects of exposure error on risk estimates will be determined; (3) the impact of differential exposure error for PM2.5 and its co-pollutants (coarse mass, CO, SO2, NO2, and O3) on risk estimates can be quantified; and (4) data from our Six Cities study can be re-analyzed.

Data analysis. Multiple existing data sets will be used to characterize sources of exposure error for PM2.5, and to examine the likelihood that these errors cause upward bias in the estimates of the association between exposure and effect. Once the effect of measurement error on risk estimates is quantified, data from the Six Cities study will be re- analyzed to determine: (1) the impact of differential measurement error on regression coefficients for PM2.5; and (2) how much of the observed differences in regression coefficients can be explained by exposure error. These analyses will be performed using a statistical model of exposure error based on Poisson regression. This model will be similar to that for a linear model, which assumes that the true exposure-effect association for a population (n) is:

(1) Equation

where YT is the effect on day t, yit is the effect in subject i on day t, xit is the true exposure of subject i on day t (or appropriate lag), and it is the error term. Since exposures are estimated in epidemiological studies using single SAM site measurements (zt) and not xit, the effect of exposure error can be determined by expanding the exposure term (Zeger et al., 1998):

(2) Equation

Substituting:

(3) Equation

Since the term Equation

equals zero, the above equation indicates that only the term represengin the classical eror or Equation

can bias the exposure effect association.

This model, with data from our exposure and measurement studies, will be used to characterize four sources of exposure error for PMa2.5: (1) instrument error; (2) spatial variation in ambient concentrations; (3) indoor/outdoor differences; and (4) personal factors. The model also will be used to determine the contribution of each of these sources to classical error. Source-specific definitions for zt and xit will be established. For example, for the analysis of instrument error, the terms zt and xit will be defined as the measured and the corrected PM2.5 concentration, respectively. For the analysis of spatial variation, the terms zt and xit will be the SAM and the satellite site measurements, respectively. Correlations between zt and Equation

will be determined for each error source. If zt and Equation

are uncorrelated, the term Equation

will be a roughly unbiased estimate of the risk associated with particulate exposures (Zeger et al., 1998).

Results from these analyses will be used to quantify the overall impact of exposure error on the exposure-effect association for PM2.5. Monte Carlo random sampling techniques will be used to estimate numerically the bias distribution and to determine the likelihood that the observed effect estimates are biased upwards. Each of the above analyses will be performed on the entire data set and on stratified data to determine whether exposure error differs by population sub-group or geographic location. The impact of differential exposure error on the exposure-effect association will be determined using simultaneous measurements of PM2.5 , coarse mass and the criteria gases, and using multi-pollutant statistical models similar to those described above. Resultant error models will be used to re-analyze data from epidemiological studies published by our investigators to determine whether the observed differences in risk estimates for different cities and sub-populations can be explained by exposure error.

General project information. This project will be led by Drs. Suh, Schwartz, and Evans. Together, these investigators have extensive experience in particle exposure assessment, epidemiology, and exposure error. They will be assisted by a doctoral student. Because much of the data for Project Ib already has been collected and the statistical models and methods to analyze exposure error have been developed, we plan to start this project immediately and expect to complete our analysis of exposure error within the first four years.

Project Ic: Differentiating Health Effects of Particles from Outdoor and Indoor Sources Background and significance. Evidence from recent animal and epidemiological studies suggests that observed associations between ambient PM2.5 and increased mortality and morbidity may occur through particle-mediated impacts on cardiovascular function (Killingsworth et al., 1997; Godleski, 1998; Peters et al., 1998; Shy et al., 1998). In our panel study of active elderly adults in Boston, we found associations between a 14 g/m3 rise in 4- hr PM2.5 and a 15% reduction in r-MSSD, a measure of heart rate variability (HRV) that has been shown to predict arrhythmic complications and mortality after MI (Gold et al., 1998). We found similar results in another study of elderly subjects in Utah Valley (Pope et al., 1998).

The toxic agent responsible for the observed adverse effects is not known. Although outdoor PM2.5 was associated with the observed effects, the adverse effects posed by outdoor PM2.5 exposures may be enhanced by exposures to indoor particles. Several sources of PM2.5, such as smoking and cooking, exist indoors. These sources can cause indoor PM2.5 concentrations to be substantially higher than those outdoors, with 24-h mean levels as much as two to three times higher than corresponding outdoor levels (Abt et al., 1997). The impacts of indoor source-related particles on health effects thus needs to be assessed directly. This will enhance our understanding of outdoor particle health effects and define whether individuals should be further protected from exposures to indoor generated particles.

Study design. The joint and individual effects of outdoor and indoor particulate pollution on heart rate and HRV will be examined for 27 people living in Boston, MA (15 MI patients, 6 individuals with COPD, and 6 asthmatics), and for 15 individuals with COPD living in Los Angeles, CA. Twelve-hour indoor, outdoor, and personal PM2.5, PM10 and gaseous (CO, SO2, NO2, O3) exposures will be measured for each participant throughout two seven-day monitoring periods (summer and winter) (Table 1). The technician will administer a brief questionnaire regarding cardiopulmonary symptoms and medication use. Heart rate and HRV will be monitored by a Holter monitor (Marquette SEER-MC) for a 20-minute period each morning. The participant will be asked to relax for 15 minutes and to perform approximately 3 minutes (20 respiratory cycles) of controlled breathing. Holter tapes initially will be analyzed for heart rate and HRV, but will be stored for possible future analysis of electrocardiogram (ECG) morphology.

Data analysis. The associations between pollution, heart rate, and HRV will be evaluated for the 588 repeated measurements using both mixed effects and fixed effects statistical models. Using fixed effects models, we will fit an individual intercept for each subject, while adjusting for time-varying co-variates and individual traits, such as medication use. Mixed models with random subject effects models will be used to: (1) evaluate the sensitivity of air pollution results to the model choice; and (2) define the primary effects and interactions with PM2.5 of subject characteristics (e.g., MI history, asthma, or COPD). Each of the above analyses will be performed using four particle measurements: (1) outdoor PM2.5 concentrations; (2) indoor source-related indoor PM2.5 concentrations; (3) outdoor source- related personal PM2.5; and (4) indoor source-related personal PM2.5. (The contribution of outdoor particles to indoor PM2.5 concentrations and personal PM2.5 exposures will be established as part of Project Ia using sulfate as a tracer of outdoor particles.) Analyses will be performed using the 588 repeated measurements of heart rate and HRV. This sample size should provide us with ample power to perform the analyses, since our pilot study of the elderly in Boston demonstrated significant associations (p<0.01) between PM2.5 and reduced HRV with only 150-160 observations.

General project information. This project will be led by Drs. Gold, Stone, Koutrakis, and Suh, who together have extensive experience in exposure assessment and epidemiology. Project Ic will be conducted in conjunction with an exposure assessment study sponsored by the EPA. As a result, monitoring will occur during the period of the summer of 1999 through the winter of 2001. We expect to complete data analysis for the project by year four.

THEME II: IDENTIFYING POPULATIONS SUSCEPTIBLE TO THE HEALTH EFFECTS OF PARTICULATE AIR POLLUTION Research conducted under Theme II will address the following specific aims: (1) to identify the human sub-populations which are most susceptible to acute and chronic adverse health effects of particulate exposures (NRC 4); (2) to determine the most important physical and chemical characteristics of particles which produce health effects among susceptible sub-populations (NRC 2, NRC 5); and (3) to investigate interactions of particulate matter and gaseous co-pollutants producing short- and long-term adverse health effects (NRC 6).

Time-series epidemiologic studies suggested that the effects of short-term particle exposures on daily mortality and morbidity were the most profound in a subset of the population which is particularly susceptible. The effects of short-term particle exposures on mortality were more severe among the elderly (>65 yrs) as compared to younger ages (Schwartz and Dockery, 1992). Increased hospital admissions for chronic obstructive pulmonary disease (COPD) and pneumonia were observed among the elderly following particle exposures (Schwartz, 1994a,b). Acute air pollution exposures cannot produce new cases of COPD or pneumonia, but it is possible (and indeed likely) that they can exacerbate pre-existing disease enough to send these sensitive individuals to the hospital. In controlled exposures of rats to concentrated ambient particles, Godleski and colleagues (1996) observed acute deaths among rats with induced chronic respiratory disease: 37% among rats with induced chronic bronchitis, and 19% among rats with monocrotyline induced pulmonary inflammation, compared to no mortality among normal rats (Figure 3).

Figure 3.

In human studies, sensitive subgroups which are potentially responsive to particulate air pollution have been identified through prospective analyses. A recent analysis by Schwartz and colleagues from Barcelona (Sunyer et al., 1998) used prospective follow- up to identify sensitive subgroups within the general population. Individuals who visited Barcelona emergency rooms for COPD during the period of 1985-1989 were tracked through the mortality registry in the years 1990-1995. The risk of acute mortality associated with particulate air pollution for this subgroup (COPD in Figure 4) was higher than for the general population (Total in Figure 4) of Barcelona. Moreover, subjects who came to the emergency department on high air pollution days (Responsive in Figure 4) had an even higher pollution-related risk, suggesting that there is a more specific subgroup of individuals with COPD which is susceptible to air pollution.

Figure 4.

The projects included in Theme II will use these and other innovative methods such as those developed in the Barcelona Study to identify individuals with existing cardiovascular and respiratory diseases. These individuals will be followed to assess the effects of acute and chronic particulate exposures on the risk of death and morbidity. Because of the history of longitudinal air pollution studies at Harvard, the Center has a wealth of data from multiple health effects studies (Table 2 ) which will be the basis of four integrated, multidisciplinary studies that will be conducted as a part of Theme II. Each of these projects will examine susceptibility to particulate air pollution in human populations. Project IIa will define a cohort of elderly subjects with pre-existing respiratory, cardiovascular, or diabetic disease based on MEDICARE hospitalization records. The effect of acute exposure on mortality in this sample will be estimated as a function of pre-existing conditions. In Project IIb, the time scale of the association between mortality and air pollution will be assessed. We will attempt to differentiate short-term changes in life- expectancy (days of life saved) from longer term effects (weeks, months, or years of life saved). We will assess whether the observed associations of short-term exposures are restricted to a small set of susceptible, frail subjects with pre-existing co-morbidities. Project IIc will assess the chronic effects of air pollution exposure on adults from the Six Cities Study. The assessment of survival in this population of adults has provided some of the most compelling evidence for significantly shorter life expectancy associated with particulate air pollution exposures (Dockery et al., 1993). Continued follow-up of this population will allow estimation of the distribution of years of life lost and disease incidence associated with exposures. Project IId will identify the characteristics of particles associated with respiratory illness and lower lung function in children. This project will extend the previous analyses of children from 29 communities across North America to include measures of city-specific particle composition. The relationships between health effects and specific particulate constituents will be quantified. Each of these projects will investigate the specific characteristics of ambient particles most strongly associated with observed health effects. In addition, each project will assess the role of gaseous co-pollutants as confounders, that is alternative predictors, and also as modifiers of the effects of particles. Because these projects build on existing data resources already being used by the Harvard investigators, the objectives will be addressed in a timely and cost-efficient manner.

Table 2.

Project IIa: Examining Conditions in the Elderly which Predispose Towards an Acute Adverse Effect of Particulate Exposures

Background and significance.
Time series studies of the association of particle exposures with daily mortality have consistently shown stronger associations in the elderly as compared to younger populations (Schwartz and Dockery, 1992). Among the elderly, particulate air pollution episodes are associated with increased hospital admissions for COPD and pneumonia (Schwartz 1994a,b, and 1995). This suggests that pre-existing respiratory conditions (either chronic or reversible) are exacerbated by acute exposures to particulate air pollution. Elderly individuals also have increased rates of hospitalization for cardiovascular diagnoses (ischemia, heart failure, and dysrythmias) associated with particulate air pollution episodes (Schwartz, 1995 and 1997). It is likely that these associations represent exacerbation by particles of pre-existing cardiovascular disease. Indeed, Godleski (1998) has shown in a dog model that occlusion of a coronary artery prior to exposure to concentrated ambient particles leads to an enhanced adverse cardiovascular response during and after exposure. The risk of cardiovascular disease is increased two- to four-fold among diabetic patients as compared to matched controls (Savage, 1996). This project will test the hypothesis that patients with pre-existing respiratory, cardiovascular, or diabetic conditions have an enhanced mortality response to particle exposures (fine or coarse fractions). In addition, it will separately assess the effects of gaseous co-pollutants as alternative predictors of mortality (confounders) and the degree to which they modify response to particulate matter (effect modification).

Study design. While time series studies have shown important associations between both daily mortality and morbidity and air pollution, these studies have not been able to link mortality with prior morbidity to provide an assessment of the characteristics of people being affected. Innovative alternative study designs are required to identify the characteristics which make an individual susceptible to sudden death in response to particulate air pollution exposures. In this project, we will identify elderly subjects with a history of hospital admissions for respiratory and cardiovascular disease and diabetes. These subjects will be tracked prospectively to identify deaths. The effect of acute air pollution exposures on these cohorts will be assessed using case-crossover methods. Thus, patients known to have chronic diseases will be followed prospectively to quantify whether they are more responsive to particulate air pollution episodes than the rest of the population. These cohorts are identifiable, since health care costs for the elderly (65+ years) are paid by the Health Care Financing Administration (HCFA). We have purchased from HCFA tape copies of all MEDICARE reimbursements of hospital admissions for respiratory and cardiovascular diagnoses for the entire United States. Patients are identified by Social Security Number, sex, age, and county of residence. Within the selected communities, three cohorts will be selected based on these HCFA hospital admission records for the period 1986 through 1990: patients with hospital admissions for (1) MIs (ICD9 410); (2) COPD (ICD9 490-496, except 493); and (3) diabetes (ICD9 250). These subjects will be tracked for deaths during the period 1991 through 1997 by searching the National Death Index (NDI). Subjects will be matched by Social Security Number, sex, estimated birth year, and county of residence (death) determined from HCFA files. For subjects matching date and place (county) of death, the underlying and contributing cause of death will be retrieved from the NDI.

The proposed study will initially be based on HCFA and air pollution data from seven metropolitan areas (Boston, MA, New Haven, CT, Chicago, IL, Provo-Orem, UT, Salt Lake, UT, Minneapolis-St Paul, MN, and Spokane, WA) for which we already have obtained data on daily measurements of PM10 and criteria gases. Boston will be one of the locations to facilitate linkage of these measures of human response to the same particles in other projects (Projects Ia, Ic, IIIa and IIIc). In the second phase of the project, data from the new EPA particle speciation monitor network will be used to select communities with contrasting fine and coarse particle mass, metal content, and elemental and organic carbon profiles. Finally, weather data will be available from Earth Info CD-ROMs of the National Weather Service Surface Observations.

Data analysis. The likelihood of death as a function of acute particulate air pollution exposures will be assessed using the case-crossover methodology (Neas et al., 1995). The case crossover study was developed by Maclure (1991) as a variant of the case-control study to examine the effects of short-term exposures. In this design, exposures immediately before the event (the case period) are compared to exposures in other, non-case periods (control periods). That is, person time is sampled for the same individual for the case (death) and control (non- death) periods. Each subject serves as his/her own control, providing perfect matching on all subject characteristics that vary slowly over time. We have demonstrated in a simulation study (Bateson and Schwartz, 1998) that this case-crossover method controls for seasonal confounding by design, and, with suitable modification, also controls for any long-term time trends. Because the analysis is on an individual level, rather than on a population level, we can look for differential effects of pollution by subject characteristic, such as age, race, sex, and number of previous hospital visits for the specified illnesses. We also can match socioeconomic data by residential zip code of each subject to see if various measures of deprivation in the neighborhood of residence are risk modifiers. In addition to examining chronic conditions as predisposing to the effects of airborne particles, we will examine whether acute respiratory infections as contributing causes of death are effect modifiers for air pollution within our cohorts. Finally, to facilitate control of potential confounding by co- pollutants, communities (in the second phase) will be selected with comparable particle mass concentrations, but with contrasting criteria gas concentrations and particle chemical composition.

General project information. This project will be led by Dr. Schwartz. Drs. Ware and Zanobetti (post-doctoral fellow) will participate in the data analysis. The first phase will be completed within the first two years, while the second will be conducted during the last three years when the data from the EPA monitoring will become available.

Project IIb: Assessing Life-Shortening Associated with Exposure to Particulate Matter
Background and significance.
Time series studies have reported that daily mortality increases with increased particle concentrations on the same and preceding days. Moreover, associations are strongest among those with pre-existing diseases. The excess numbers of deaths associated with typical air pollution concentrations are small. These observations suggest that short-term air pollution episodes are restricted to a small pool of frail individuals. Air pollution episodes may be advancing the date of death by only a few days among those individuals who would have died anyway from natural causes. This "harvesting" of the frail population would imply that air pollution episodes produce only trivial decreases in life expectancy and would have little public health significance.

Since the data used in time-series analyses on individuals do not contain identifying information, the actual amount of life shortening attributable to air pollution exposures could not be estimated. However, two recent analyses (Zeger et al., 1998; Schwartz, 1998) assessed the effects of air pollution exposures at multiple time scales, thus allowing the harvesting effect to be examined. If the harvesting effect is substantial, the association between air pollution and daily deaths would be concentrated in high frequency fluctuations (time scales of only a few days) and substantially diminished at lower frequencies (time scales of weeks to months). In a recent analysis of Boston daily mortality, Schwartz (1998) showed that the effect of particulate exposures on COPD deaths decreases with longer time periods, suggesting that particles may indeed be harvesting a frail population of COPD patients. On the other hand, air pollution-related deaths from ischemic heart disease increased with longer time periods (Figure 5), suggesting that the effects on cardiovascular disease are not restricted to a frail population, and indeed, deaths are occurring in people who would have been expected to have substantial additional years of life expectancy.

Figure 5.

This project will apply these methods to different cities across the United States for which particle data are available. The effects of air pollution on mortality will be quantified as a function of time scales from a few days to several months. Harvesting will be examined for all causes of mortality and for specific causes, including pneumonia, COPD, ischemic heart disease, and heart failure. Analyses will be conducted in multiple cities chosen for the heterogeneity in particle characteristics and the concentrations of gaseous co-pollutants.

Study design. If air pollution were only advancing deaths by a few days, then we would expect an increase in daily deaths due to air pollution to be followed shortly by a decline. If we average over a week, the two effects would cancel out (or partially cancel out if some of the deaths are brought forward by a longer period). Hence, if we apply a smooth (or moving average) to the outcome, and regress these smoothed data against air pollution, the short-term harvesting would have been canceled out. Therefore, if we decompose the association between air pollution and daily deaths into different averaging periods, then any association that occurs at longer time periods (lower frequencies) will be free of this short- term harvesting. If we expect harvesting to occur within a few weeks, rather than a few days, we need to adjust our definition of lower frequency. This analysis will define whether most of the life shortening due to air pollution is short-term (days or weeks) or represents longer effects. Air pollution is associated with a worsening of illness as well as mortality. It is possible that for some illnesses air pollution increases the size of the pool of persons at risk of dying from that illness. In the meantime, the size of the pool may decrease, due to air pollution-related deaths. Hence, the sign of the harvesting effect is not determined a priori, and may vary by illness. Therefore, we will conduct these analyses separately for different causes of death. We will analyze data from several locations in order to quantify the difference between particles and gases. Daily death counts initially will be constructed for Chicago, Minneapolis-St. Paul, and Seattle from the National Center for Health Statistics detailed mortality tapes for the years 1988-1993. All non-external causes and cause specific deaths will be tabulated for each day in each city. The specific causes examined will be pneumonia (ICD9 480-487), COPD (ICD9 490-496 except 493), ischemic heart disease (ICD9 410-414), dysrhythmia (ICD9 427), and heart failure (ICD9 428). Separate counts by race, sex, and age strata (<65, 65-74, 75+) also will be computed. Weather data will be obtained from the nearest NOAA weather station, and air pollution data will be obtained from the Environmental Protection Agency's AIRS database. All three study locations have at least one PM10 monitor and criteria gas monitors that operate on a daily basis. In a second phase we will investigate additional cities as new exposure data from the new supersite network becomes available.

Data analysis. Cleveland's STL algorithm (Cleveland et al., 1990) was designed to separate time series into different components with different characteristic time scales. We will use the STL algorithm to resolve air pollution, daily deaths, and weather from our study locations into two time series: one reflecting seasonal and longer fluctuations, and one reflecting intermediate term fluctuations which average out the very short-term patterns. By varying the definition of intermediate term, it will be possible to look at harvesting on different time scales. Specifically, a loess smooth with a window of 120 days will be used to fit the seasonal and long-term time trends in all analyses. To remove short-term harvesting, a second smooth will be applied to the data after the seasonal adjustment. We will examine smoothing windows of 15, 30, 45, and 60 days. The longer windows are able to smooth over harvesting on a longer scale than the shorter ones. The particle concentrations and weather data will be decomposed similarly, and then log linear regressions will be fit to examine the association between particle concentrations and daily deaths, independent of season and long-term time trends, for each of the different midscale smoothing windows. Temperature and humidity will be controlled in all regressions. Generalized additive models will be used to fit nonparametric smoothed functions of the weather variables to account for nonlinearities and extreme temperature effects. The results for each of these windows will be compared to results in the baseline regressions using unfiltered data. Effect size estimates that decrease with longer averaging windows will indicate the existence of a harvesting effect. The lack of such a decrease will indicate that harvesting occurs on longer time scales than the window.

General project information. This project will be led by Dr. Schwartz. He will be assisted by Dr. Zanobetti, and a doctoral student will participate in the analysis.

Project IIc: Investigating Chronic Effects of Exposure to Particulate Matter
Background and significance.
Arguably the most quoted papers on the health effects of air pollution are the two reports describing the association of particulate air pollution exposures with survival of adult population samples followed prospectively in the Six Cities Study (Dockery et al., 1993) and the American Cancer Society study (Pope et al., 1995). The directness and simplicity of the design of these studies makes the results particularly accessible. The prospective follow-up of well-defined populations directly addresses the fundamental question regarding the effect of cumulative exposure to particulate air pollution on incidence of disease and life expectancy. Knowing the baseline health status of the individual study participants, it is possible to directly assess the characteristics which put individuals at risk for responding to the effects of particulate air pollution exposures. These data provide a base for additional follow-up and analysis to address the long-term effects of particulate exposures which would not be feasible with any other design. Project IIc will: (1) extend follow-up of the Six Cities Study cohort up to twenty-four years; (2) assess the cumulative effect of long-term exposures on the incidence of lung cancer, nonmalignant respiratory disease, cardiovascular disease, and cause-specific mortality; (3) estimate the distribution of years of life lost associated with particulate exposure; and (4) determine the pre-existing conditions which predispose individuals to decreased survival and increased incident disease associated with exposures.

Study design. In the mid-1970s (1974-77), a cohort of 8,412 adults (25-74 yrs) was recruited in six Midwestern and eastern cities. Baseline respiratory health status, including pulmonary function, was assessed for each subject, plus individual determinants of respiratory health status, including smoking history, occupational exposure, residential history, and treatment for chronic cardiovascular disease. Vital status follow-up, including annual mailings and searches of the National Death Index (NDI), continued through 1990. Project IIc would continue follow-up of these populations starting with the last contact in 1990. NDI searches would be undertaken for the years since 1990. Underlying and contributing cause of death will be retrieved from the NDI for confirmed matches. In addition to vital status (death) monitoring of this population, we also would begin an active mailed inquiry regarding incident diseases, including COPD, chronic cardiovascular disease, and respiratory cancers. For all participants not identified as deceased by NDI search, we will mail a brief questionnaire to the last known address. For those returned for "Addressee Unknown," we will search national phone databanks, and follow-up with inquiries of contacts identified by the participants at their last follow-up visit. Participants still not found after this multi- level search will be tracked through alternative databanks, including Social Security Administration payments, Health Care Finance Administration payments, motor vehicle license lists, and ultimately commercial tracking services. Thus, despite a gap of approximately eight years since the last contact, we expect to obtain near complete-vital status. HCFA data also will be used to identify any hospital admissions for members of the cohort who are aged 65 or more (2,690 in 1990). The risk of hospital admissions for selected conditions (MIs, heart failure, COPD, pneumonia, etc.) will be examined as a function of individual predictors (including baseline pulmonary function) and long-term exposure to air pollution.

Data analysis. Mortality and disease incidence data will be analyzed using survival analysis methods. Two parameter Weibull models will be used to estimate the distribution of years of life lost. Sensitivity of the result to specific survival analysis methods will be assessed. Competing causes of death will be considered in the estimates of life years lost. Particulate and gaseous air pollutant exposures in each of the Six Cities will be updated from 1990 based on routine monitoring data from the EPA AIRS database. Susceptible populations will be identified by stratified analyses based on history of chronic disease determined from prior questionnaires. The estimates of life-shortening in the Six Cities cohort will be extrapolated to the U.S. population based on the distribution of individual demographic and lifestyle risk factors. Data from the third National Health and Nutrition Examination Survey (NHANES III), a sample of the health status of the entire U.S. population, will be used to define the distribution of demographic, health, and lifestyle characteristics.

General project information. This project will be led by Dr. Dockery, the Principal Investigator of the Six Cities study, who will direct the follow-up of the adults in the Six Cities cohort. Dr. Koutrakis will be responsible for particle chemical analysis. Drs. Ware and Schwartz will be in charge of the data analysis. Dr. Ware was one of the Principal Investigators of the Six Cities study and is familiar with the data sets. They will be assisted by a doctoral student.

Project IId: Determining the Effects of Particle Characteristics on Respiratory Health of Children
Background and significance.
Between 1988 and 1991, we assessed the respiratory health of 13,369 fourth and fifth grade school children from twenty-four communities in the United States and Canada. We found increased rates of bronchitis and lower lung function associated with fine particle mass, sulfate, and acidity (Dockery et al., 1996, Raizenne et al., 1996). Although there was a clear association between low lung function (or bronchitis) and fine particle exposures (Figure 6), there also was considerable variation about the fitted line, suggesting that alternative measures of particulate air pollution exposure might be more predictive. Respiratory health of approximately 9,500 fourth and fifth grade school children also was measured in five urban areas Philadelphia, PA (1992/93), Washington, DC (1993/94), Nashville, TN (1994/95), Boston, MA (1995/96), and Phoenix, AZ (1996/97). Particle measurement methods were similar to those used in the Twenty-Four Cities study so that these populations could be pooled in future analyses. Particle monitoring in the Twenty- Four Cities and Five Cities studies included every other day measurements of PM10 and PM2.5 as well as fine particle sulfate and acidity for approximately 12 months (Spengler et al., 1996). Thus, additional data on particle characteristics, plus the larger sample from the combined studies (over 20,000 children from 29 communities across the U.S. and Canada), provide a unique opportunity to assess the effects of chronic exposure to specific particle components on respiratory health of children.

Figure 6.

Study design. We propose to expand the chemical characterization of particle exposures by conducting additional analyses of the archived Teflon filter samples collected in the Twenty-Four and Five Cities studies. As mentioned above, samples have already been analyzed for fine particle mass, sulfate, and acidity. Based on current hypotheses regarding the mechanism of action of particles and the specific characteristic of particles responsible for adverse effects, these archived samples would be analyzed for metals and elemental carbon, as follows: (1) trace element analyses will be conducted using X-Ray fluorescence; (2) water soluble metal concentrations will be determined by ICPMS; of particular interest will be the concentrations of Fe, Ni, Cu, Mn, Mg, V and, Se; and (3) elemental carbon concentrations will be determined using light absorption (Campbell et al 1995). Although daily carbon measurements will have significant variability, we expect that the yearly averages will be sufficiently accurate to rank the different cities. These additional analyses will make it possible to speciate particles into major constituents (carbon, metals, sulfate, ammonium, and acidity) for a large number of cities in a cost-effective manner.

Data analysis. City-specific mean concentrations of total metals and, separately, water-soluble metals will be compared to city-specific lung function and respiratory symptom prevalence, adjusted for individual characteristics. Previous analyses of city- specific health outcomes have demonstrated a larger between-city variance than would be predicted by inter-individual variance (Ware et al., 1986; Dockery et al., 1989). We will use a two-step analysis to correct for excess between-city variation. In the first step, city- specific adjusted lung function (or logits of respiratory symptoms) will be calculated by adjusting for each child demographic, anthropometric, and life-style risk factors. In the second stage, the city-specific adjusted mean lung function (or logits of symptoms) will be regressed against the city-specific annual mean concentrations of individual air pollutants using weights inversely proportional to the sum of the between-city and within-city variances. The effects between pollutants will be compared based on the estimated effect for an inter-quartile range increment in each pollutant, and the t-statistic of the estimated effect (a dimensionless number). Comparisons across these children in twenty-nine cities will provide insights into the role of long-term exposures to specific components of the particle mix in producing observed differences in the respiratory health of children.

General project information. This project will be led by Drs. Dockery and Speizer, the Principal Investigators of the Twenty-Four Cities and Five Cities studies. Drs. Koutrakis and Suh will be responsible for the analyses of the archived particle filters from these studies. Dr. Ware will participate in the data analysis, since he has worked closely with Drs. Dockery and Speizer on this analysis in the past and is familiar with the data sets to be used for this project. A doctoral student will participate in the analysis.

THEME III: BIOLOGICAL MECHANISMS/DOSIMETRY

The objectives of Theme III are to identify the particulate and gaseous air pollutants responsible for increased cardiac vulnerability as an adverse health effect and to define the biological mechanisms that lead to this outcome. As part of this theme, we specifically propose to: (1) identify the physical and chemical properties of particulate matter which are responsible for the observed adverse health effects (NRC 5); (2) determine whether gaseous co-pollutants exacerbate the effects of particles (NRC 7); (3) investigate the biological mechanisms by which particulate matter produces mortality and acute or chronic morbidity (NRC 9); and (4) examine particle deposition patterns and fate in the respiratory tract (NRC 6).

These objectives will be addressed in three research projects which explore both the components of air pollution that cause adverse health effects and the biological mechanisms that may lead to fatal outcomes. To date, our studies have made it possible to explore and define both cardiac and pulmonary responses to inhaled fly ash and concentrated ambient particles (Killingsworth et al., 1997; Godleski, 1998, Godleski and Clarke, 1998). The most consistent and dramatic of these responses was the change in the time to ST segment elevation in dogs with coronary occlusion. Specifically, we have employed short-term occlusion of the left anterior descending coronary artery with a permanently implanted occluder. The protocol involved three five-minute occlusions. The first, a conditioning occlusion, was done the morning of the planned exposure. The second occlusion was carried out 20 minutes later, followed by a six-hour exposure either to concentrated air particles or particle-free air (sham exposure). The animals were awake and free to move during exposures but were tethered via exposure tubing connected to permanent tracheostomies. During the occlusions and exposures, continuous ECG data were collected using a Holter monitor and respiratory data were measured from the tracheostomy tube. After the particle or sham exposure, the third coronary occlusion was carried out again for five minutes. Figure 7 illustrates the magnitude of ST segment and T-wave change during an occlusion. A simple measurement which emerged from this experiment is the response time from occlusion to ST segment elevation. The post- exposure time differences for CAPs and sham animals are illustrated in Figure 8. The CAPs-exposed dogs showed ischemically-induced ST segment elevation almost one minute sooner than their sham-exposed counterparts (Figure 8) (Godleski, 1998). These cardiac changes, which have the potential to be associated with arrhythmias and death, provide a plausible mechanism by which ambient particulate exposures can lead to fatal outcomes in certain individuals.

Figure7.

Figure8.

The three projects proposed under this theme will build upon the findings from our ongoing animal studies. In Project IIIa, we will use our dog model of coronary occlusion (Nearing and Verrier, 1992; Godleski et al., 1998; Godleski et al., 1997) and our recently- developed coarse and ultrafine ambient particle concentration technologies to examine the role of particle size (coarse, fine, and ultrafine), particle composition (elemental and organic carbon, sulfates, and trace metals), and gaseous co-pollutants (O3, CO, NO2, and SO2) on cardiac sensitivity to ischemia. The hypotheses to be tested are: (1) ultrafine particles are more toxic than fine particles, which are more toxic than coarse particles; (2) O3 and CO, and to a lesser extent SO2, and NO2, enhance the response to particles; and (3) transition metals have a greater impact than other particle constituents. We will explore biological mechanisms of the response of the occlusion model with pharmacological and/or surgical intervention as a means to abrogate or enhance autonomic and/or inflammatory mechanisms. These are the most likely pathways by which particles might accelerate or increase sensitivity to the ischemic response. The hypothesis to be tested is that inflammatory and vagally-mediated mechanisms are primary response effectors. Project IIIa will expand the scope of our current animal studies, which focus exclusively on the effects of fine particles and their components on the cardiac and respiratory health of animals (NIEHS and EPA grants), and on the development of dose response relationships (HEI grant). Project IIIb will use the dog coronary occlusion models, along with innovative deposition models, to determine the effect of particle deposition and dose on the variability in biological response. The applicability of results from Projects IIIa and IIIb to humans will be investigated in concurrent human studies in Project IIIc. This project will examine the relationship between ambient particulate exposures and plasma viscosity (and other markers of coronary occlusion risk), as well as heart rate and HRV. Findings from Project IIIc will be compared to those from our concurrent animal studies. Central to each of these projects will be our ability to investigate the effects of individual particle components, since results from our current animal studies indicate that particle toxicity varies with particle component, i.e., transition metals produce the greatest adverse effects (Godleski et al., 1998).

Project IIIa: Differentiating the Roles of Particle Size, Particle Composition, and Gaseous Co-Pollutants on Cardiac Ischemia
Background and significance.
Epidemiological studies have shown associations between air pollution and cardiac health (Burnett et al., 1995; Schwartz and Morris, 1995; Schwartz et al., 1997). The change in time to detectable ischemia provides a reproducible experimental measurement that has direct bearing on the severity of an ischemic episode. This electrocardiographic measurement, representing complex biological processes in the heart, provides an endpoint which can be used to define the most important particle constituents or characteristics. Thus, this endpoint can be used to determine the roles of particle size and composition as well as the concomitant gaseous co-pollutants, none of which have been adequately characterized. Toxicological studies, for example, have not yet investigated the health effects of combined exposures to concentrated ambient particulate matter and the criteria gases (CO, SO2, NO2, and O3). Also, the health effects of these criteria gases have not been studied singularly or in combination with particles using the cardiac ischemia model. Therefore, controlled laboratory studies of real world ambient particle exposures with gaseous co-pollutant exposures are needed to examine whether criteria gases exacerbate the effects of particles.

Similarly, studies examining the effects of particle size need to be performed using meaningful outcome parameters. Exposures to ultrafine particles have been linked to a variety of adverse health effects in toxicological studies of laboratory-generated, multi- component ultrafine particles (<0.1 m) (Oberdoerster et al., 1995) and in epidemiological studies (Peters et al., 1998). These effects, however, were not observed in controlled laboratory studies of single component, artificial ultrafine particles (Ziesenis et al., 1998), suggesting that the toxicity of ultrafine particles can best be determined using real world exposures. In addition, although epidemiological studies have not demonstrated direct effects from coarse particle (2.5 to 10 m) exposures, toxicological studies of these particles are needed for setting standards related to this particle size fraction.

The biological mechanism by which particles shorten the time to ST segment elevations has not yet been identified. Since many of the identified adverse cardiac effects involve increased vagal influence on the cardiac and respiratory system (Godleski et al., 1998), pharmacological intervention with agents which affect the vagal response may improve our ability to identify the specific biological mechanism. Acetylcholine, which normally promotes vasodilation, has been shown in cases with minimal arteriosclerosis to result in a vasoconstrictor effect (Ludmer et al., 1986). Thus, it should be the initial target of mechanistic assessment of this biological effect. Other mechanisms for which data supports additional studies are inflammation and sympathetic nervous system responses (Killingsworth et al., 1997; Godleski and Clarke, 1998).

Study design. The proposed study will investigate the effects of: (1) concomitant gaseous co-pollutant exposures; (2) ultrafine and coarse particles; (3) individual particle constituents; and (4) pharmacological. Although the specific objectives of each of these components differ, each uses the cardiac ischemia model and our recently-developed concentrator technologies to expose dogs to ambient particles. The exposure and health monitoring for each of these experiments will follow similar protocols, where dogs are exposed to concentrated air particles (ultrafine, fine, or coarse particles), criteria gaseous pollutants, or filtered air via tracheostomy. Particle mass and black carbon will be measured continuously during exposure. Also, integrated samples will be collected for particle mass, size, acidity, sulfate, nitrate, and trace element analysis. The canine subjects will undergo coronary occlusion to induce acute myocardial ischemia (without any evidence of myocardial infarction) (Godleski et al., 1998). A balloon occluder will be surgically implanted around the left anterior descending coronary artery two weeks prior to the experiment. After recovery, experiments will be carried out in which the occluder will be inflated for five minutes twice pre-exposure and once post-exposure to induce coronary occlusion. During occlusion and exposure, the ECG will be recorded continuously. Breathing parameters will be recorded continuously via differential pressure measurements from the tracheostomy tube. Bronchoalveolar lavage (BAL) and transbronchial biopsy will be used to assess pulmonary inflammatory responses in selected animals, and peripheral blood white blood cell count and differential to assess systemic responses in all studies.

Effect of gaseous co-pollutants: Each exposure will include four canines equipped with an implanted balloon occluder on a coronary artery. The protocol will utilize 2 five- minute occlusions prior to exposure and a five-minute occlusion at the end of the six-hour exposure. On any given day, two chambers (A and B) will be used to expose each of the four subjects to either of the following atmospheres: fine CAPs (chamber A), sham (chamber A), fine CAPs plus gas (chamber B), and sham plus gas (chamber B). Over a two- week period, each dog will undergo one day of each type of exposure. Thus, each dog will have a total of 12 occlusions, which is feasible based on our experience. ECG data from leads V4 and V5 will be collected continuously during the occlusions and exposure. Each breath and each heart beat will be analyzed throughout the six-hour exposure to determine HRV, and T-wave alternans, as well as changes in ECG morphology, and respiratory frequency, volumes, and flow parameters using the Buxco system. This experiment will be repeated for three groups of four dogs for each gas. In these studies, exposures to one of the four criteria gases (O3, CO, SO2, and NO2) will be administered simultaneously with those for the particles. As in the case of particulate exposures, administered gas concentrations will be relatively higher than those encountered under a typical polluted atmosphere (200, 10,000, 50, and 100 ppb, respectively). Depending on response, additional experiments will employ higher or lower concentrations of gaseous pollutants. Continuous measurements of gas concentrations will be conducted throughout the exposures to ensure that the desired dose remains constant. In addition to the continuous measurements, which will be assessed as per existing protocols, the critical measurement in which we expect to define differences is the time to ST change, as illustrated above.

Effect of particle size: Four canines with implanted balloon occluders on their coronary arteries will have pre- and post-exposure occlusion, as described above. On any given day, the exposure for a given dog will be either: fine CAPs (chamber A), fine sham (chamber A), ultrafine CAPs (chamber B), and ultrafine sham (chamber B). Over a two- week period, each dog will undergo one day of each exposure. Thus, each dog will have a total of 12 occlusions. This study also will involve three groups of 4 dogs. The same design will be used to compare the effects of coarse versus fine particles. Exposures to the different particle sizes will be performed using our recently-developed fine (Sioutas et al., 1996), ultrafine (Sioutas and Koutrakis, 1996), and coarse (Koutrakis et al., 1998) particle concentrator techniques. The ultrafine particle concentrator uses a small amount of water vapor to grow ultrafines to larger sizes (approximately 1-2 m). These larger particles are subsequently concentrated using the fine particle concentrator which is based on the virtual impactor technique. Particles are then dried to return them to the ultrafine mode using a diffusion dryer. For the proposed ultrafine studies, a large humidification system will be used at the inlet of the existing fine particle concentrator. Ultrafine particle concentrations will be increased by factor of approximately 60. The coarse particle concentrator, which also was recently developed (operational in early 1999) (Koutrakis et al., 1998), will be used to increase the concentration of coarse particles by a factor of approximately 50.

Pharmacologic assessment of the mechanism responsible for shortening of the time to ST elevation: Similarly to the previous studies, four canines equipped with an implanted balloon occluder will undergo occlusions pre- and post-exposure. Studies will be carried out on the exposure combination that provides the most dramatic change in the time to ST elevation. This could be the most toxic CAPs size fraction or CAPs/Gaseous co-pollutant combination. On any given day, the exposure for a given dog will be either: CAPs (chamber A), sham (chamber A), CAPs plus pharmacologic agent (chamber B), and sham plus pharmacologic agent (chamber B). Specific agents to be used include: (1) the cardioselective blocker atenolol (1mg/kg, i.v.) (Mantelli et al., 1995; Lombardi et al., 1983; Ghaleh et al., 1995), which provides nearly total blockade of beta1-receptors; (2) glycopyrrolate (100 g/kg, i.v.) (Kasanuki et al., 1997; Kovach et al., 1998; and Magi et al., 1983), which is used for vagal blockade and a more rapid onset with a longer duration than atropine; and (3) Prednisone (1 mg/kg/day), a prototypical broad-spectrum anti-inflammatory agent (McDonald and Langston, 1995). Carprofen (4 mg/kg/day) is an alternative anti-inflammatory agent because it is relatively free of side-effects and is preferred in veterinary use. Sympathetic agonist studies will use isoproterenal (1 g/kg, i.v.) and parasympathetic agonist studies will use acelytcholine (100 g/kg). Vagotomy and stellectomy are alternative blockade approaches and will be used to confirm findings. Over a two week period, each dog will undergo one day of each exposure. Thus, each dog in these studies also will have a total of 12 occlusions and will undergo one day of each type of exposure. Mechanistic studies will utilize similar protocols and experimental design to those for the previous exposures described above and will be performed after those investigating particle co-pollutants and size.

Particle composition: The effects of particle composition will be examined as part of our ongoing research projects and as part of those proposed above. All exposure studies will include continuous and integrated sampling to determine particle mass (gravimetric and continuous monitoring); sulfate, nitrate, and ammonium (ion chromatography); acidity (pH); trace elements (X-Ray Fluorescence); elemental carbon (thermal analysis/aethalometry); organic carbon (thermal analysis and gas chromatography/mass spectrometry); and measurement of endotoxin.

Data analysis. Based on our studies to date, exposing three groups of four dogs will provide sufficient power to detect differences in response, even with occasional days with no detectable CAPs effect (Godleski et al., 1998). This is due to the use of: (1) a very sensitive animal model; (2) exposures to concentrated ambient particles; (3) continuous cardiac and respiratory measurements; and (4) the crossover design, which controls for dog and day variability. Changes in time to ST elevation before and after each exposure will be compared to the integrated six-hour particle measurements. This time measurement difference will be assessed using Student's paired or unpaired t-test. For each of the four exposure scenarios, continuous measurements of cardiac and respiratory parameters will be compared. Semiparametric models (Speckman, 1988; Hastie and Tibshirani, 1993; Hobert et al., 1997) will be used to measure the effect of CAPs exposure on respiratory and heart function while controlling for nuisance nonlinear trends in the measurements over each six-hour exposure. Cardiac and respiratory effects will be related to particle size, particle composition, and gaseous pollutant concentrations. In the case of gaseous pollutants, no-effect concentrations for each gas will be determined singly and in combination with particles. Similar approaches will be used in the pharmacologic interventional studies. If an agent blocks the effect of exposure and its corresponding agonist enhances that effect, a specific biologic mechanism can be inferred.

General project information. Project IIIa encompasses a series of comprehensive animal studies which were presented together; however, this project will require a large fraction of the Center resources (20-25%) and will be critical for the prioritization and integration of our research activities. This project will be directed by Dr. Godleski who is the Principal Investigator of most of our ongoing particle toxicology studies. He will be assisted by Drs. Koutrakis (particle generation and characterization), Verrier (cardiology), Kobzik (lung pathology), and Catalano (biostatistics). A large group of talented investigators will participate in this project, such as Drs. Reinisch, Wolfson, Lawrence, Clark, Lovett, Gazula, and Mr. Ferguson. Their role in the project is described in the budget justification section.

Project IIIb: Assessing Deposition of Ambient Particles in the Lung
Background and significance.
Investigation of particle deposition is critical in our efforts to link biological response to particle exposure. A key to understanding deposition patterns in the alveolar regions of the lung is the mixing of inhaled aerosols with residual gas. Recent experimental studies have shown that fine and ultrafine particles mix appreciably with alveolar gas, and deposit on the lung periphery (Heyder et al., 1988, Shultz et al., 1992 and 1995, Tsuda et al., 1995a and 1998, Butler and Tsuda, 1997 and Darquenne et al., 1997). In an attempt to explain differences between classical deposition theories (Morrow et al., 1996; Davis et al., 1972; Watson et al., 1974; Ultman et al., 1985) and current experimental evidence, we have proposed a new mechanism for fine particle deposition in the deep lung, which we have called "stretch and fold" convective mixing mechanism (Tsuda et al., 1995b and 1998, Butler and Tsuda, 1997). Figure 9 demonstrates the highly convoluted "stretch and fold" irreversible convective patterns seen in the acinar zone of rat lungs. This picture is a cross section of a small airway filled with white latex solution. A single infusion of a blue latex solution shows the pattern of "stretch and fold" mixing when breath mixes with residual air. These patterns are very different from those of laminar or turbulent steady flows inside tubes which are commonly used by classical theories. "Stretch and fold" convective flow patterns result in perturbations of streamlines that evolve exponentially with each breath. Cycle-by-cycle narrowing of lateral distances between streamlines causes mixing to take place. This mechanism has the potential of inducing appreciable deposition of fine and ultrafine particles explaining their deposition in the deep lung (Butler and Tsuda, 1997, Tsuda et al., 1998). The main objective of this project is to use in situ continuous respiratory and total deposition measurements to develop a new regional deposition model, which will be based on our recent theoretical calculations (Tsuda et al., 1994ab, 1995ab and 1998, Butler and Tsuda, 1997, Haber and Tsuda, 1998).

Figure 9.

Study design. (1) Dosimetry and respiratory measurements: During our canine exposures via tracheostomy (Project IIIa), in situ particle deposition and respiratory measurements will be made. For each breath, the tidal volume, VT, breath frequency, f, and instantaneous inspiratory and expiratory flow rates, Vi, Ve, will be recorded as part of our routine measurements. Based on our previous exposure studies (Godleski et al., 1998), we expect to record approximately 5,000 -20,000 breath-by-breath data points during the six- hour exposures. One-minute average particle concentrations and size distributions in inhaled (Cin) and exhaled (Cout,) air will be determined. For measurements of exhaled air, a one-way differential flow valve will connect the tracheostomy tube to a small chamber. The one-way valve will be used to isolate the exhaled air while the small chamber will be used to allow for fluctuations in the flow of exhaled air. A combination of a differential mobility analyzer (for particle diameters up to 0.7 m), and particle size analyzer (for particles larger than 0.5 m) will be used to measure particle concentration and size at the exit of this chamber. (Correction for particle losses inside the chamber may be necessary.) Six-hour integrated samples of the inhaled and exhaled air will be collected to determine the average dose for the different particle components, such as ions, metals, and carbon.

(2) Lung deposition model: Using the respiratory parameters and the inhaled and exhaled particle concentrations obtained during the canine exposures, particle deposition in the intrathoracic airways will be modeled. Based on the flow regimes, the transport and deposition of particles in the intrathoracic airways falls into three compartments (Pedley et al., 1977): (1) large airways (turbulent flow; high Reynolds number); (2) medium and small airways with laminar and secondary flows [10 less than Re less than 1000]; and (3) alveolated terminal airways (low Re number). The volume of the conducting compartment will be considered constant and all ventilatory volumes will be accommodated in the parenchymal compartment. The deposition efficiencies of each of these flow regimes will be estimated. The parenchymal compartment is expected to be the primary site for deposition of fine and ultrafine particles. We will first solve the differential mass conservation equations for each compartment as a function of the experimentally- measured breathing parameters and inhaled concentrations, and the unknown deposition efficiencies. The values of the deposition efficiencies will be determined by minimizing the sum of squared residuals (difference between predicted and measured expired particle concentrations). This problem requires numerical integration of the governing equations, since they are driven by changing patterns of ventilation. Also, nonlinear least squares estimation techniques will be used to calculate the deposition efficiencies, since these appear nonlinearly in the solution of the coupled set of conservative equations. Uniqueness and resolution of the estimated efficiencies will be analyzed as well.

Data analysis. Using the respiratory and particle continuous measurements, the deposition efficiency in each of the three intrathoracic compartments will be computed and compared with corresponding physiological parameters. One of the objectives of this analysis will be to identify mechanisms responsible for regional deposition. Regional deposition efficiencies will be examined as a function of respiratory parameters, with particular attention to the frequency, f, dependence. This will make it possible to test whether parenchymal deposition depends on f1/2, as predicted by the classical "diffusive" deposition theory, or is independent of f, which is consistent with the "stretch and fold" convective mixing model. In an effort to identify the mechanisms responsible for particle deposition at different sites, the dependence of regional deposition efficiencies on particle physico-chemical characteristics, such as size, number, composition, and hygroscopicity, will be investigated. Finally, dose calculations will be used to quantify the extent to which particle dose explains variability in exposure/biological response relationships in Project IIIa.

General project information. This project will be led by Dr. Tsuda, who is an expert in particle dosimetry. Dr. Butler, also an expert in this field, will participate in this project. Dr. Godleski will oversee all the exposure and biological aspects of the project. Dr. Koutrakis will be responsible for all particle measurements. The laboratory studies related to this project will be performed by the same research group presented in Project IIIa.

Project IIIc: Relating Changes in Blood Viscosity, Other Clotting Parameters, Heart Rate and Heart Rate Variability to Particulate and Criteria Gas Exposures
Background and significance. In the 1952 London smog episode and in other subsequent episodes, the highest levels of mortality occurred as a result of sudden death, which is presumed to be related to acute MI or arrhythmia. As mentioned above, both animal (Godleski et al., 1998) and human (Gold, 1998, Creason et al., 1998) studies have found associations between heart rate and HRV and particulate exposures. Peters et al. (1998) recently showed that air pollution levels were associated with increased plasma viscosity in the Augsburg cohort of the MONICA study (Peters et al., 1998). These findings suggest that coronary artery obstruction is associated with exposures to particles. To test this hypothesis, it will be necessary to demonstrate associations between exposures and events consistent with the development of coronary artery obstruction. This would require comprehensive health monitoring of free living individuals and detailed characterization of their exposures. The Normative Aging Study (NAS) cohort is such a population. The main objective of Project IIIc is to investigate associations of selected inflammatory and blood clotting parameters in free living humans with particle and criteria gas exposures. Our hypothesis is that particle exposures can induce systemic inflammation manifested by activation of clotting parameters. Heart rate and HRV will be another measure of systemically-induced cardiac effects, and their associations with particle exposures will be examined. This will enable use to make comparisons between the effects observed in the animal and human health studies.

Study design. The Normative Aging Study has followed a population of approximately 2,280 males since 1963. Subjects were originally selected to be free of chronic disease. These men have been examined every five years until they reached the age of 55, and every three years after the age of 55. The sample for the proposed study will consist of: (1) NAS participants who will report for their regularly scheduled medical examination between October 1, 1999 and September 30, 2003; and (2) their wives. Based on the number of subjects examined between October 1993 and September 1997, we expect that approximately 1,170 men (390/year, ranging in age from 51 to 92 years, mean age of 69 years) will report for examination during this period. In addition to studying the sample of 1,039 men, we plan to identify a subsample of married men whose wives will be willing to participate. We expect 904 of the 1,039 men to be living with their wives. We further expect 678 (75%) of the 904 wives to participate in this study (a realistic percentage based on our recent experience recruiting subjects for similar studies involving physiologic testing). This investigation will make use of data normally collected during triennial NAS examinations, including smoking history, socioeconomic status, anthropometry (body measurements), respiratory symptoms (currently assessed by standardized ATS-DLD questionnaire), and pulmonary function. Objective measurements of atopy have been conducted since April 1984, and include skin testing, blood eosinophil counts, and serum concentrations of IgE. Nonspecific airway responsiveness has been assessed by a methacholine challenge test since April 1984. The later examinations include routine health evaluations, pulmonary function, methacholine challenge studies, and ECG examinations, as well as routine blood examination, measurements of fibrinogen and other clotting factors. Detailed dietary assessment, as well as measurements of body burden of lead and autonomic function, have been carried out as part of the most recent examinations. Approximately two subjects are screened per day as part of the routine assessments. Blood will be obtained for a series of clotting parameters, including fibrinogen, clotting time, whole blood viscosity, platelet counts, and factors VII and IX. Additional data will include smoking and residential history, plus detailed medical and drug intake history. Five-minute rhythm strips in the supine position after resting for 10 minutes, along with a full ECG, will be obtained. For these data sets, we will pay one-third ($300,000) of the total cost. The remaining amount will be covered by the NAS study (for further details see the Budget Justification). This project will be conducted in parallel with the animal exposure studies; thus, both projects (IIIa and IIIc) will use the same particle measurements. Particle and criteria gas concentrations will be measured daily (at the roof of our school near downtown Boston). These measurements also will be used for the animal studies and have been described above. Our previous studies in Boston and other northeastern metropolitan areas have shown that fine particle concentrations present little spatial variability; therefore, it is expected that our monitoring site will be representative of the study catchment area (Burton et al., 1996; Suh et al., 1997).

Data analysis. Daily measurements of heart rate, HRV and clotting factors will be regressed against individual risk factors (diet, medications, smoking history, baseline ECG classifications according to the Minnesota Codes) and daily particle/gas exposure data (particle mass, size, and composition and criteria gases). Generalized additive models will be used to assure adequate control for weather, season, and day of the week using smoothing functions of those variables. The potential for a threshold will be examined using a smoothed function of particle exposure. Effect modification by age, medication use, smoking history, dietary antioxidant, and baseline ECG status also will be examined. Based on the study by Peters et al. (1998), it is anticipated that the large sample size of approximately 1,500 subjects will be sufficient to detect differences in clotting and cardiac function that may be associated with changes in exposure. Exposure/responses for the human and animal studies will be compared and will be related to different exposure parameters, such as particle mass, size, composition and criteria pollutant concentrations. The collective analysis of the data from Projects IIIa, b, and c will enable us to elucidate the role of particle characteristics, dosimetry and gaseous co-pollutants and to explain the variability in exposure/response relationships. Ultimately this will help us to enhance our understanding of particle health effects.

General project information. This project will be led by Dr. Speizer who is an expert in the field of air quality health effects. He will collaborate with Dr. Pentel Vokonas, the Principal Investigator of the NAS study who will be responsible for the collection of health data (see attached letter from Massachusetts Veterans Epidemiology Research and Information Center). Dr. Koutrakis will be responsible for the air quality measurements. Drs. Speizer and Godleski will collaborate on the data analysis to make comparisons between the health outcomes from Projects IIIa and IIIc. Drs. Schwartz and Ware will also participate in the statistical analysis.

ADMINISTRATIVE CORE

The Administrative Core will be responsible for the overall coordination of the Center. This core will includes administrative support, core secretarial support, fiscal management, central data management, quality assurance, internal Executive and External Advisory Committee meeting schedules, meeting organization, report preparation, and communication with EPA. Dr. Petros Koutrakis will oversee all administrative and fiscal issues for the Center. He will be assisted by Christina DiMeo, the Center Administrator. All fiscal and administrative duties will be directed by Linda Fox, Administrator of the Environmental Science and Engineering Program. She will provide quarterly budget summaries and projected expenditures. Ms. Fox will be assisted by May Seto, her financial administrative assistant, and also by Brenda Barrett and Carla Silva, Administrators of the Environmental Epidemiology and Physiology programs, respectively. (Note that their efforts are not included in the budget.) The research staff at Harvard University has considerable experience working with large exposure, epidemiological, and toxicological data sets. While the principal investigators will have responsibility for processing and analyzing data for their individual projects, data from each of the projects also will be stored and managed centrally, with Dr. Helen Suh overseeing the data management and supervising the data manager (to be named). Finally, George Allen, a Senior Environmental Engineer, will be the Center QA/QC Officer. He has considerable experience preparing and reviewing laboratory and field protocols and quality control procedures.

RESEARCH COORDINATION CORE

The Research Coordination Core will be responsible for defining, coordinating, and integrating all research conducted as part of the Center and will be led by Dr. Dockery. A key aspect of the Center is its intention to have an ongoing and continual evaluation of research needs and priorities. To achieve this goal, the Center will include a rigorous and multi-phased research coordination and evaluation process, which was developed based on our experience from over twenty years of multi-disciplinary collaborations in air pollution health effects. This coordination and evaluation process will draw from experts from a wide range of disciplines at the Harvard School of Public Health and the Medical School, as well as from experts from outside agencies, universities, and other organizations to provide focused and timely responses to current and evolving questions about airborne particulate matter. Specifically, experts from six internal and external groups will contribute to the research coordination and evaluation process and will determine the direction and coordination of PM research that is conducted at the Center. These groups include: the National Research Council, the External Science Advisory Committee, the consortium of EPA Airborne Particulate Matter Centers, the Working Group on PM Exposures and Health Effects, the Working Group on Research Strategy Evaluation, and the Steering Committee at the center at Harvard. Their contribution to the research coordination and evaluation process are described below.

National Research Council. The National Research Council panel on Research Priorities for Airborne Particulate Matter was and will continue to be an important source of direction for the Center. In fact, the Center's proposed ten initial projects were developed to address the immediate research priorities listed in NRC's report entitled "Immediate Priorities and Long-Range Research Portfolio." The Center is fortunate to have two investigators, Drs. Koutrakis and Speizer, as members of the NRC Expert panel. As a result, direct and immediate access to panel deliberations will be available to Center investigators and can be used to decide the Center's future research priorities and directions.

External Advisory Committee. A multi-disciplinary External Advisory Committee of distinguished scientists will be established to provide input into both ongoing and future research directions. The Advisory Committee will be comprised of experts in a range of disciplines, including atmospheric chemistry, exposure and risk assessment, policy, biostatistics, epidemiology, and toxicology. Some committee members will be recruited from other PM Centers to foster and facilitate exchange and collaborations, including experts from the fields of Exposure and Risk Assessment, Atmospheric Chemistry, Epidemiology, Toxicology, Biostatistics, Cardiac and Respiratory Health, and Public Policy. The Committee will meet annually for two days to formally review the Center activities. The first meeting day will be devoted to the traditional presentation of study designs and results, and will be followed by a structured workshop on the second day to define research needs and priorities. This workshop would include both the Committee members and the Center investigators.

Consortium of EPA Airborne Particulate Matter Centers. As one of EPA's PM Centers, we will approach the other Centers to form a consortium of PM Centers. The specific aims of this Consortium will be to ensure that research in each Center is coordinated with, complementary to, and not redundant with that of other EPA PM Centers, and to facilitate rapid dissemination of research findings and other information between Centers, the EPA, the rest of the scientific community, and the lay public. To achieve these specific aims, the Consortium will organize an annual colloquium to review particulate matter research. As part of this colloquium, representatives from each of the EPA-sponsored Centers would participate in a structured meeting to plan and coordinate research activities. Through the Consortium, the EPA Centers will establish linked sites on the Web to provide descriptions of current research programs, downloadable copies of research reports and publications, and access to extended summaries and original data.

Harvard Working Group on PM Health Effects. A previously-established Working Group on PM Health Effects will continue to meet bi-weekly to encourage informal interactions between the Center investigators. This Working Group, which was formed several years ago, includes experts in exposure and risk assessment, epidemiology, toxicology, clinical medicine, and physiology. The proposed PM Center at Harvard will establish a formal structure for this Working group and will provide core support for its activities. Activities will include a monthly series of informal Work-in-Progress seminars by Center investigators and formal presentations by local and invited experts. To date, major advances have been made as a result of the informal interactions among the members of the Working Group. For example, results from our observational studies led us to design animal inhalation studies to CAPs. These laboratory studies not only validated our epidemiological study findings, but also enabled us to generate specific hypotheses regarding the effects of particles on cardiac health. Subsequent panel studies of the elderly have found associations between heart rate and HRV and particle exposures, and have confirmed results from our laboratory studies. Furthermore, because of the close collaboration between the toxicology and particle chemistry groups, we have preliminary results regarding the association between particle health effects and specific constituents of fine particulate matter. Similarly, the development of the Ambient Particle Concentrator has spawned new research projects in both toxicology and epidemiology.

Working Group on Evaluation of Research Strategies. We propose to use a formal decision and value of information analysis of particulate matter control and research to guide our decisions about future research activities. This analysis will be based on the concept of the value of information (i.e., the expected value of the likely consequences of suboptimal decisions) as a measure of the costs of current levels of uncertainty. To implement our future research evaluation process, we will characterize quantitatively the current risk uncertainties and will develop estimates of the informativeness and cost of alternative research strategies. Uncertainty in risk estimates will be examined based on two broad uncertainty sources: parameter and model uncertainty. Parameter uncertainty, of which the slope of the dose-response function of particle exposure and mortality is an example, will be determined based on well- developed and widely accepted approaches.

Most of these approaches are based on frequentist notions of probability and standard approaches for analysis of the propagation of uncertainty (e.g., Gauss' law, lognormal error analysis, or Monte Carlo simulation). Model uncertainty will be determined through the use of formally elicited expert judgment. These methods will first be applied within our Working Group on Particle Health Effects to identify the specific areas which should be targeted as research priorities. Subsequently, they will be applied to the External Advisory Committee in a workshop following their initial meeting and also to the Consortium of EPA-sponsored PM Centers. In each case, the traditional assessment and the structure assessment of the Evaluation of Research Strategies Working Group will be made available to the Steering Committee (see below).

The Evaluation of Research Strategies Working Group, which will be directed by Drs. Evans, Graham, and Hammitt, has successfully applied this approach in several other settings (Finkel and Evans, 1987, Evans et al., 1988, Evans et al., 1992, Taylor et al., 1993, Thompson and Evans, 1997) and is currently engaged in several preliminary analyses in support of this effort. For example, we are in the process of developing estimates of the cost and effectiveness of fine particle control strategies. We have developed a novel approach to characterize the impact of emissions reductions on population exposure (Evans et al., 1998). In addition, we have constructed a preliminary probability tree to use in characterization of the uncertainty in estimates of mortality from acute exposures to fine particles, and have completed a preliminary decision analysis and value of information analysis of the entire problem.

Steering Committee at the PM Center at Harvard. The PM Center at Harvard will be directed by a Steering Committee consisting of the Center Director, the two Co-Directors, and the Principal Investigators of the research projects and cores. Dr. Koutrakis, the Principal Investigator, will chair the Steering Committee. The Steering Committee will be responsible for the overall direction, coordination, and integration of the research conducted by the Center. It will establish research priorities and directions based on recommendations from external groups, including the National Research Council, the External Advisory Committee, the Consortium of PM Centers, the Harvard Working Groups on Particle Health Effects and Evaluation of Research Strategies. The Steering Committee will meet at least quarterly to monitor progress, identify new research initiatives, and coordinate research with other Centers.

ANALYTICAL AND FACILITIES CORE

This core will be led by Dr. Godleski. Most of the proposed research will be centered at the Harvard School of Public Health, located in the Longwood Medical Area of Boston. A portion of the work will be carried out at the Beth Israel Deaconess Medical Center, Brigham and Women's Hospital, Veterans Hospital, and the Harvard Medical School. All medical area resources, including those of Harvard Medical School, Children's, Dana-Farber Hospital, Beth Israel Deaconess Medical Center, and Countway Library are available to investigators involved in this Center. We provide a brief description of the available facilities below. (i) Animal Facilities: The complex consists of a total of 4092 ft2 for housing animals and an animal surgery suite. Fully-equipped operating rooms are available for the dog surgery proposed. The facility is AALAC accredited and staff veterinarians are available for consultation. (ii) Environmental Chemistry Laboratory: All particle characterizations of airborne samples from the animal exposures will be conducted by this laboratory (7,000 ft2 ). Equipment that is currently available to be used includes: continuous monitors for particle size and number, particle mass, sulfate, carbon, criteria gases, temperature and relative humidity, computer controlled data acquisition systems necessary for automatic data collection, a facility for the complete range of calibrations for all of the continuous monitors, based on the requirements of and in collaboration with the EPA. For performing chemical analysis of the particle samples we have a GC/MS for organics, Plasma Emission Spectroscopy for elements, HPLC for particulate-phase polycyclic aromatic hydrocarbons, a pH-meter for aerosol acidity, and IC for ions. Gravimetric analysis of the particulate matter collected on filters will be done in a temperature and relative humidity controlled room. For bioaerosol measurement, bacterial and fungus culture, and endotoxin quantification, a new environmental microbiology laboratory (1600 ft2) is available. Finally, all sampling equipment for personal measurements, microenvironmental and outdoor measurements is available. (iii) Laboratory for Inhalation Toxicology and Animal Cardiac and Pulmonary Function: Major equipment located in this room (over 700 ft2) includes Harvard Ambient Air Particle Concentrators with 30X and 100X concentrating ability, multiple exposure chambers (1 m3 and 0.1 m3), particle and criteria gas monitors, oximeter for continuous O2 saturation monitoring, calibrating spirometer, Buxco pulmonary function system, fluid filled pressure transducers, differential pressure transducers, and Holter monitors. (iv) Electron Microscopy Facility: This facility (470 ft2) includes a Zeiss energy filtering transmission electron microscope, an Amray 100 Scanning EM with ion pump, Kevex Energy Dispersive X-Ray Analysis system, and Robinson backscatter detector. (v) Laboratories for Biochemistry and Cell Physiology: This facility includes three fully-equipped laboratories (1900 ft2) with standard utilities, fume hoods, and a tissue culture area. (vi) Molecular Biology Laboratory: One room of 400 ft2 with supporting facilities fully equipped for molecular techniques. (vii) Computer Facilities: The computer facilities available for the Center include a network of three linked Sparc workstations, including an Ultrasparc with 256 megabytes of RAM and 16 gigabytes of disk storage, two 8mm tape drives, and a CDROM recorder. These workstations all have SAS and Splus, and are networked to all of the researcher's PC's. In addition, the Kresge Center for Environmental Health has a second Ultrasparc Server that is also available for the PM center's use. Marquette Medical (Mars) Work Stations are available for ECG analyses. (viii) Surgical Facilities: A fully-equipped, newly- constructed septic surgical suite with appropriate ventilators, physiologic monitors, and complete inventory of surgical instruments and gowns is available. An adjoining room contains a state-of-the-art fluoroscope, recorders, and computers.

TECHNOLOGY DEVELOPMENT AND TRANSFER CORE

The Technology Core will be led by Dr. Koutrakis, who has been involved in the development of many particle instruments and holds several patents. This Core will encompass the development and transfer of technologies and methods in three areas: particle measurement and generation, statistical methods, and cardiovascular measurement methods: (i) Particle Measurement and Generation Methods. The core will support the exposure assessment studies through the development and provision of state- of-the-art personal, microenvironmental, and outdoor particulate and gaseous samplers. Also, this core will support toxicological studies by operating and servicing the concentrator and by developing new concentrator technologies. We expect to continue the development of particle samplers and generating systems. As they are developed, they will be made available to other Centers and research groups. Our expertise in this area has been demonstrated through the development of a number of sampling devices that can be used to collect particulate and gaseous air pollutants, such as the Harvard Impactor (Marple et al., 1987), the Harvard/EPA Annular Denuder System (Koutrakis et al., 1988), the passive ozone sampler (Koutrakis et al., 1993), the personal gas/particle sampler (Koutrakis et al., 1989), the personal multi-pollutant sampler (Koutrakis et al., 1998), the continuous sulfate monitor (Allen et al., 1984), and the continuous ambient particulate mass monitor (Koutrakis et al., 1996). Each of the developed methods are commercially available and have been used extensively for outdoor, indoor, and personal monitoring by investigators located throughout the world. In addition, the recent development of the Fine Ambient Particle Concentrator (FAPC) (Sioutas et al., 1995) has made it possible for the first time to conduct human and animal inhalation studies of ambient particles. These studies already have substantially improved our understanding of particle health effects. Many other research groups will use this technology in their own research, including the U.S. EPA, the University of Michigan, the University of Southern California, the University of Toronto, the Netherlands National Institute of Health (RIVM), and the University of So Paolo. This technology has been expanded to include a recently- developed Coarse Particle Concentrator for RIVM (Koutrakis et al., 1998) and a high volume particle collector to collect large quantities of size fractionated particles for toxicological studies for the Finnish Institute of Public Health. (ii) Statistical Methods. We have introduced the use of Poisson time series into environmental epidemiology, as well as the use of generalized estimating equations, nonparametric smoothing, diary studies, and generalized additive models (Schwartz et al., 1991). More recently, we have pioneered the use of the case-crossover methodology in environmental epidemiology and have made methodological advances on the harvesting question (Neas et al., 1995). We have always had an active technology transfer program, with faculty members serving as external advisors for the major European air pollution studies, teaching short courses in the U.S. and abroad, working with researchers in Mexico City, So Paolo, and Santiago, and welcoming visitors for periods of a few weeks to a year to work with us and learn our methods. We plan on continuing this general approach and collaborating closely with other PM Centers to discuss new methodologies and issues. (iii) Cardiovascular Measurement Methods. Heart rate variability methods employed in the Center will be made available to other members of the research community. Our laboratory employs MARS Workstations (Marquette Medical Systems, Milwaukee, WI) for downloading and annotating ambulatory ECGs. The MARS system permits both raw ECG and annotations to be exported in the de facto standard "MIT" format. Use of files adhering to the MIT standards assures maximum interchangeability among diverse sources of ECG. Time- domain HRV metrics will be implemented in ANSIIC to ensure portability. Future routines for frequency-domain HRV and ECG morphology analysis will be developed along similar lines with an eye toward inter-operability. Programs will be made available by request through post or electronic mail, as well as by anonymous Internet FTP access. Each program will be fully documented with a user's manual and description of the algorithm(s) used.

3. EXPECTED BENEFITS
The Center will address critical questions about the health effects of ambient particles using input from multiple disciplines and continual evaluation of research findings and needs. Through its three themes, research conducted by the Center will be comprehensive, including projects that in total address eight of the ten research priorities identified by the NRC. These projects will use a wide variety of innovative measurement and analysis techniques to examine diverse questions about the potential toxic agent(s), susceptible populations, and biological mechanisms of ambient particles. Since each of the Center's projects leverages existing research conducted by Center investigators, the projects are both cost-effective and expansive. Existing data and resources will be maximized in order to obtain the most relevant and essential scientific information. For example, our Center will use data from our earlier epidemiological studies to examine the effects of chronic exposures to ambient particles, an area that has historically been difficult to examine due to cost concerns. Similarly, the Center's leveraging of our existing toxicological and exposure studies makes particle deposition and indoor particle health effect studies possible, respectively.

Each of our research projects will involve investigators from wide range of disciplines. The inter-disciplinary nature of the Center will enable us to integrate findings from a variety of studies as they emerge and, combined with structured methods to evaluate research needs and priorities, will allow the Center to remain flexible and at the forefront of the health effects debate on particles. Findings from the Center will have important ramifications to environmental policy and control. Our findings will improve our ability to set appropriate and effective air quality standards for particulate matter. Moreover, they will help us determine the relevant particle parameter and the appropriate level to regulate. This will, in turn, allow effective particle control strategies to be determined and ensure the protection of public health from particle exposures.

4. GENERAL PROJECT INFORMATION
An organizational chart of our Center and information about the direction of the projects and the roles of the individual investigators is included in the budget justification..


REFERENCES:

Abt, E., Suh, H.H., Allen, G., Koutrakis, P. "Characterization of Indoor Particulate Sources in Four Homes in the Boston Metropolitan Area." In Proceedings of the U.S.EPA/AWMA Conference on Measurement of Toxic and Related Air Pollutants, Research Triangle Park, NC, Apr 29-May 1, 1997.

Allen, G.A. "Description of a Continuous Sulfuric Acid/sulfate Monitor." Presented at the U.S. EPA Fourth Annual Symposium on Recent Advances in Pollutant Monitoring of Ambient Air and Stationary Sources, Raleigh, NC, May 1984. Proceedings published in EPA publication #EPA-600/9-84-019.

Bahadori, T., Rojas, L., Suh, H.H., Koutrakis, P. "Characterization of Personal Pm10 and Pm2.5 Exposures: Results from Studies Conducted in Nashville, TN, and Boston, MA." Presented at the Int. Society for Exp. Analysis-Soc. Risk AssessmentConference. New Orleans, LA, Dec 1996.

Bateson, T., Schwartz, J. "Control for Season and Time Trend in Studies of Daily Events Using the Case-crossover Design." In review, 1998.

Burnett, R.T., Dales, R., Krewski, D., Vincent, R., Dann, T., Brook, J.R. "Associations Between Ambient Particulate Sulfate and Admissions to Ontario Hospitals for Cardiac and Respiratory Diseases." Am J Epidemiol, 142 (1): 15-22, 1995.

Burton, R.M., Wilson, W.E., Suh, H.H., and Koutrakis, P. "Spatial Variation in Particulate Concentration Within Metropolitan Philadelphia." Environmental Science & Technology, 30:400-407, 1996.

Burton, R.M., Suh, H.H., Koutrakis, P. "Characterization of Outdoor Particle Concentrations Within Metropolitan Philadelphia." Environ. Sci. Technol. 30, 400, 1996.

Butler, J.P. and Tsuda, A. "Effect of Convective Stretching and Folding on Aerosol Mixing Deep in the Lung, Assessed by Approximate Entropy." J. Appl. Physiol., 83(3):800-809, 1997.

Campbell, D., Copeland, S., Cahill, T. "Measurement of Aerosol Absorption Coefficient from Teflon Filters Using Integrating Plate and Integrating Sphere Techniques." J. Aerosol Science & Tech., 22:287-292, 1995.

Clayton, C.A., Perritt, R.L., Pellizzari, E.D., Thomas, K.W., Whitmore, R.W., Wallace, L.A., Ozkaynak, H., Spengler, J.D. "Particle Total Exposure Assessment Methodology (PTEAM) Study: Distributions of Aerosol and Elemental Concentrations in Personal, Indoor, and Outdoor Air Samples in a Southern California Community." J. Expos. Anal. Environ. Epidem., 3, 227-250, 1993.

Cleveland, R.B., Cleveland, W.S., McRae, J.E. et al. "Seasonal-trend Decomposition Procedure Based on LOESS." J Offic Stat., 6:3-73, 1990. Dakins, M.E., Toll, J.E., Small, M.J., and Brand, K.P. "Risk-Based Environmental Remediation: Bayesian Monte Carlo Analysis and the Expected Value of Sample Information." Risk Analysis, 16, 67-79, 1996.

Darquenne, C., Paiva, M., West, J.B., and Prisk, G.K.. "Effect of Microgravity and Hypergravity on Deposition of 0.5 to 3mm Diameter Aerosol in the Human Lung." J. Appl. Physiol., 83(6):2029-2036, 1997.

Davies, C.N. "Breathing of Half-micron Aerosols Ii. Interpretation of Experimental Results." J. Appl. Physiol., 32(5): 601-611, 1972.

Diggle, P.J., Liang, K.Y., Zeger, S.L. "Analysis of Longitudinal Data." Oxford University Press, Oxford, England, 1994.

Dockery, D.W., Speizer, F.E. et al. "Effects of Inhalable Particles on Respiratory Health of Children." Am. Rev. Respir. Dis., 139:587-594, 1989.

Dockery, D.W., Pope, A.C., 3d, Xu, X., Spengler, J.D., Ware, J.H., Fay, M.E., Ferris, B.G., Jr, Speizer, F.E. "An Association Between Air Pollution and Mortality in Six U.S. Cities." N Engl J Med. 329(24): 1753-1759, Dec. 9, 1993.

Dockery, D.W., Cunningham, J., Damokosh, A.I., Neas, L.M., Spengler, J.D., Koutrakis, P., Ware, J.H., Raizenne, M., Speizer, F.E. "Health Effects of Acid Aerosols on North American Children: Respiratory Symptoms." Environ Health Perspect, 104(5): 500-505, May 1996.

Evans, J.E., Hawkins, N.C. and Graham, J.D. "The Value of Monitoring for Radon in the Home: a Decision Analysis." J Air Pollution Control Assoc., 38(11):1380-1385, 1988.

Evans, J.E., Graham, J.D., Gray, G.M., Silken, R.L., and Smith, W.E. "A Distributional Analysis of the Carcinogenic Potency of Chloroform." Presented at the 1stInternational Conference on the Safety of Water Disinfection: Balancing Chemical and Microbial Risks, Washington, DC, Aug 31- Sep 3, 1992.

Evans, J.E., Gray, G.M., Silken, R.L., Smith, W.E., Valdez-Flores, Carica, and Graham, J.D. "Use of Probabilistic Expert Judgment in Uncertainty Analysis of Carcinogenic Potency." Regulatory Toxicology and Pharmacology, 20(1):15-36, 1994.

Finked, A.M. and Evans, J.E. "Evaluating the Benefits of Uncertainty Reduction in Environmental Health Risk Management." J Air Pollution Control Assoc., 37(10):1164-1171, 1987.

Gaetano, M., Vanoli, E., Stramba-Badiale, M. et al. "Vagal Reflexes and Survival During Acute Myocardial Ischemia in Conscious Dogs with Healed Myocardial Infarction." Am. J. Physiol., 261 (Heart Circ. Physiol. 30):H63-H69, 1991.

Ghaleh, B., Bea, M.L., Dubois-Rande, J.L. et al. "Endothelial Modulation of Beta-adrenergic Dilation of Large Coronary Arteries in Conscious Dogs." Circulation, 92:2627-2635, 1995.

Godleski, J.E., Cedes, C., Cutler, M., Koutrakis, P. "Death from Inhalation of Concentrated Ambient Air Particles in Animal Models of Pulmonary Disease." Proceedings of the Second Colloquium on Particulate Air Pollution and Human Health. Park City, Utah, 4:136-143, May 1-3, 1996.

Godleski, J.J., Sioutas, C., Verrier, R.L., Killingsworth, C.L., Lovett, E., Krishna Murthy, G.G., Hatch, V., Wolfson, J.M., Ferguson, S.T., and Koutrakis, P. "Inhalation Exposure of Canines to Concentrated Ambient Air Particles." American Journal of Respiratory and Critical Care Medicine, 155(4):A246, 1997.

Godleski, J.E., Verrier, R.L., Koutrakis, P. et al. "Mechanisms of Morbidity and Mortality from Exposure to Ambient Air Particulate." Health Effects Institute Report, in revision, 1998.

Godleski, J.J. and Clarke, R.E. "Systemic Responses to Inhaled Ambient Particles: Pathophysiologic Mechanisms of Cardiopulmonary Effects." Inhaled Particles, ed. by J. Heyder & P. Gehr. Lung Biology in Health and Disease Series, Marcel Dekker.

Gold, D.R., Litonjua, A., Schwartz, J., Verrier, M., Milstein, R., Larson, A., Lovett, E., Verrier, R. "Cardiovascular Vulnerability to Particulate Pollution." Am. J. Resp. Crit. Care Med., 157: A261, 1998.

Haler, S., and Tsuda, A. "The Effect of Flow Generated by a Rhymically Expanding Pulmonary Acinus on Aerosol Dynamics." J. Aerosol Sci., 29(3):309-322, 1998.

Hammitt, J.K. and Cave, J.A.K. "Research Planning for Food Safety: A Value-of-Information Approach." RAND. R-3946-ASPE/NCTR. Prepared for the Assistant Secretary for Planning and Evaluation, National Center for Toxicological Research, U.S. Department of Health and Human Services, 1991.

Hammitt, J.K. "Can More Information Increase Uncertainty?" Chance, 8, 15-17, 1995. Hastie, T.J., Tibshirani, R.J. "Generalized Additive Models." London: Chipman & Hall, 1990.

Heyder, J., Blanchard, J.D., Feldman, H.A., and Brain, J.D. "Convective Mixing in Human Respiratory Tract: Estimates with Aerosol Boli." J. Appl. Physiol., 64(3): 1273- 1278, 1988.

Hobert, J.P., Altman, N.S., Schofield, C.L. "Analyses of Fish Species Richness with Spatial Covariates." J.Am. Statistical Assoc., 92:846-854, 1997.

Kasanuki, H., Ohnishi, S., Ohtuka, M., Matsuda, N., Nirei, T., Isogai, R., Shoda, M., Toyoshima, Y., Hosoda, S. "Idiopathic ventricular fibrillation induced with vagal activity in patients without obvious heart disease." Circulation 95:2277-2285, 1997.

Killingsworth, C.R., Alessandrini, F., Murthy, G.G.K., Catalano, P.J., Paulauskis, J.D., and Godleski, J.J. "Inflammation, Chemokine Expression, and Death in Monocrotaline- treated Rats Following Fuel Oil Fly Ash Inhalation." Inhalation Toxicology. 9: 541-565, 1997.

Kirby, D.A., Verrier, R.L. "Differential Effects of Sleep Stage on Coronary Hemodynamic Function." Am J Physiol, 256:H1378-H1383, 1989.

Koutrakis, P., Wolfson, J.M., Slater, J.L., Brauer, M., Spengler, J.D., and Stevens, R.K. "Evaluation of an Annular Denuder/filter Pack System to Collect Acidic Aerosols and Gases." Environmental Science and Technology, 22:1463-1468, 1988.

Koutrakis, P., Fasano, A.M., Slater, J.L., Spengler, J.D., McCarthy, J.F., and Leaderer, B. "Design of a Personal Annular Denuder Sampler to Measure Atmospheric Gases and Particles." Atmospheric Environment, 23: 2767-2773, 1989.

Koutrakis, P., Briggs, S.L.K., and Leaderer, B.P. "Source Apportionment of Indoor Aerosols in Suffolk and Onondaga Counties, New York." Environmental Science and Technology, 26: 521-527, 1992.

Koutrakis, P., Wolfson, J.W., Bunyaviroch, A., Froehlich, S. Hirano, K., and Mulik, J.D. "Measurement of Ambient Ozone Using a Nitrite-coated Filter." Analytical Chemistry, 65: No. 3, 209-214, 1993.

Koutrakis, P., Wang, E., Cedes, C. and Wolfson, M. Continuous Ambient Particle Monitor, patent number 5,571,945, Nov 5, 1996.

Koutrakis, P., Wolfson, J.M., Ferguson, S.T. "Development of a Coarse Particle Concentrator for Inhalation Exposures." J. Aerosol Sci., 29:999-1000, 1998.

Koutrakis, P. "Use of a Multipollutant Sampler for Personal and Microenvironmental Measurements." Submitted to the Journal of the Association for Air Pollution Prevention. Kovach, J.A., Nearing, B.D., Verrier, R.L. "Anger Potentiates Myocardial Ischemia- induced T-wave Alternans and Cardiac Electrical Instability," 1998.

Krupnick, A.J. and Cropper, M.L. "The Effect of Information on Health Risk Valuations." Journal of Risk and Uncertainty, 5:29-48, 1992.

Lave, L.B., Ennever, F.K., Rosenkranz, H.S., and Omenn, G.S. "Information Value of the Rodent Bioassay." Nature, 336, 631-633, 1988.

Lindley, D.V. "Making Decisions, 2nd Ed." John Wiley and Sons, London, 1985.

Linville, C.D. "Mathematical and Computational Techniques for Research Prioritization with an Application to Global Climate Change Research." Doctoral Thesis, Department of Engineering and Public Policy, Carnegie-Mellon University, 1998.

Lombardi, F., Verrier, R.L., Lown, B. "Relationship Between Sympathetic Neural Activity, Coronary Dynamics, and Vulnerability to Ventricular Fibrillation During Myocardial Ischemia and Reperfusion." 1983.

Ludmer, P.L., Selwyn, A.P., Shook, T.L., Wayne, R.R., Mudge, G.H., Alexander, R.W., Ganz, P. "Paradoxical Vasoconstriction Induced by Acetylcholine in Atherosclerotic Coronary Arteries." N Engl J Med, 315:1045-1051, 1986.

Magi, N., Eckberg, D., Sprenkli, J. "Low Dose Atropine Reduces Ventricular Vulnerability in Normal and Ischemic Hearts." Clin. Res. 29:813A, 1983.

Maclure, M. "The Case-Crossover Design: a Method for Studying Transient Effects on the Risk of Acute Events." Am J Epidemiol., 133(2): 144-153, Jan 15, 1991.

Marple, V.A., Rubow, K.L., Turner, W., and Spengler, J. D. "Low Flow Rate Sharp Cut Impactors for Indoor Air Sampling: Design and Calibration." JAPCA, 37, 1303- 1307, 1987.

Mantelli, L., Amerini, S., Ledda, F. "Bradykinin-induced Vasodilation Is Changed to a Vasoconstrictor Response in Vessels of Aged Normotensive and Hypertensive Rats." Inflamm Res, 44:70-73, 1995.

McDonald, R.K., Langston, V.C. "Use of Corticosteroids and Non-steroidal Anti- inflammatory Agents." In: Ettinger, S.J., Feldman, E.C., eds. Textbook of Veterinary Internal Medicine, W.B. Saunders, Philadelphia, 1995.

Morgan, M.G., Morris, S.C., Meier, A.K., and Shenk, D.L. "A Probabilistic Methodology for Estimating Air Pollution Health Effects from Coal-Fired Power Plants." Energy Systems and Policy, 2, 287-310, 1978.

Morgan, M.G. and Henrion, M. "Uncertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis." Cambridge University Press, 1990.

Morgan, M.G. and Keith, D.W. "Climate Change: Subjective Judgments by Climate Experts." Environmental Science and Technology, 29, 468A-476A, 1995.

Morrow, P.E., Chairman, Task Group on Lung Dynamics. "Deposition and Retention Models for Internal Dosimetry of the Human Respiratory Tract." Health Phys., 12:173-207, 1966.

Mulatto, G., Beria, G., Sala, S., Bonazzi, O., Schwartz, P.J. "Quantitative Analysis of T Wave Abnormalities and Their Prognostic Implications in the Idiopathic Long QT Syndrome." J Amer Coll Cardiol, 23: 296-301, 1994.

Nearing, B.D., Huang, A.H., Verrier, R.L. "Dynamic Tracking of Cardiac Vulnerability by Complex Demodulation of the T-Wave." Science, 252: 437-440, 1991.

Nearing, B.D., Verrier, R.L. "Simultaneous Assessment of Autonomic Regulation and Cardiac Vulnerability During Coronary Occlusion and Reperfusion by Complex Demodulation of Heart Rate Variability and T-wave Alternans." Circulation 86 (suppl I): I-639, 1992.

Nearing, B.D., Oesterle, S.N., Verrier, R.L. "Quantification of Ischaemia-induced Vulnerability by Precordial T-wave Alternans Analysis in Dog and Human." Cardiovasc Res. 28:1440-1449, 1994.

Neas, L.M., Schwartz, J., Dockery, D.W. "Case-crossover Analysis of Mortality and Particulate Air Pollution in Philadelphia, Pennsylvania." Epidemiology, 6(4): S56, 1995.

North, D.W. and Balson, W.E. "Acid Deposition: Decision Framework." Prepared for Electric Power Research Institute. EPRI Research Project # 2156-1, 1982.

North, D.W., Selker, F., and Guardino, T. "Estimating the Value of Research: An Illustrative Calculation for Ingested Inorganic Arsenic." Decision Focus Incorporated, Sept 1993.

Oberdorster, G., Gelein, R.M., Ferin, J., Weiss, B. "Association of Particulate Air Pollution And Acute Mortality: Involvement of Ultrafine Particles?" Inhalat Tox 7: 111- 124, 1995.

Pedley, T.J., Schroter, R.C., and Sudlow, M.F. "Gas Flow and Mixing in the Airways. In: Bioengineering Aspects of the Lung," edited by J.B. West, Chapt.3, 163-265, Lung Biology in Health and Disease, Vol. 3, Marcel Dekker, Inc, NY, 1977.

Peters, A., Perz, S., Doring, A., Stieber, J., Koenig, W., Wichmann, H.E. "Increases in Heart Rate Point Towards Activation of the Autonomic Nervous System During an Air Pollution Episode." Submitted to Lancet, 1998.

Pope, C.A., III, Thun, M.J., Namboodiri, M.M., Dockery, D.W., Evans, J.E., Speizer, F.E., Heath, C.W., Jr. "Particulate Air Pollution as a Predictor of Mortality in a Prospective Study of U.S. Adults." Am J Respir Crit Care Med, 151 (3 Pt 1): 669- 674, Mar 1995.

Pope, C.A., Dockery, D.W., Kanner, R.E., Villegas, G.M., Schwartz, J. "Daily Changes in Oxygen Saturation and Pulse Rate Associated with Particulate Air Pollution and Barometric Pressure." Health Effects Institute Annual Meeting, Apr 1997.

Pope, C.A., Verrier, R.L., Lovett, E.G., Larson, A.C., Raizenne, M.E., Kanner, R.E., Schwartz, J., Villegas, G.M., Dockery, D.W. "Heart Rate Variability Associated with Particulate Air Pollution." American Heart Journal, submitted 1998.

Raiffa, H. "Decision Analysis." Addison-Wesley, Reading, MA, 1968.

Raizenne, M., Neas, L.M., Damokosh, A.I., Dockery, D.W., Spengler, J.D., Koutrakis, P., Ware, J.H., Speizer, F.E. "Health Effects of Acid Aerosols on North American Children: Pulmonary Function." Environ Health Perspect., 104(5): 506-514, May 1996.

Rojas, L., Koutrakis, P., Suh, H.H. "Personal Exposure to Particulate Matter and its Relationship to Outdoor Concentrations in a Group of COPD Patients Living in the Boston Area." Presented at the PM2.5: A Fine Particulate Standard Conference, Long Beach, CA, Jan 28-30, 1998.

Sapin, P.M., Koch, G., Blauwet, M.B. et al. "Identification of False Positive Exercise Test with Use of Electrocardiographic Criteria: a Possible Role for Atrial Repolarization Waves." J Amer Coll Cardiol., 18: 127-135, 1991.

Savage, P.J. "Cardiovascular Complications of Diabetes Mellitus: What We Know about Their Prevention." Ann Intern Med, 124:123-126, 1996.

Schultz, H., Heilmann, P., Hilebrecht, A., Gebhart, J., Meyer, M., Piiper, J., and Heyder, J. "Convective and Diffusive Gas Transport in Canine Intrapulmonary Airways." J. Appl. Physiol., 72:1557-1562, 1992.

Schultz, H., Schultz, A., Eder, G., and Heyder, J. "Influence of Gas Composition on Convective and Diffusive Intrapulmonary Gas Transport." Exp. Lung. Res., 21:853- 876, 1995.

Schwartz, J., Haselblad, V., Pitcher, H. "Air Pollution and Morbidity: a Further Analysis of the Los Angeles Student Nurses Data." JAPCA, 38:158-162, 1988.

Schwartz, J., Katz, S., Fegley, R., Tockman, M. "Analysis of Spirometric Data from a National Sample of Healthy 6-24 Year Olds." Am Rev Respir Dis., 138:1405-1414, 1988.

Schwartz, J., Dockery, D.W., Sypij, D., Ware, J., Zeger, S., Spengler, J., Speizer, F.E., Ferris, B.G., Jr. "Daily Diaries of Respiratory Symptoms and Air Pollution: Methodological Issues and Results." Environ Health Perspect., 90:181-188, 1991.

Schwartz, J. "Particulate Air Pollution and Daily Mortality in Detroit." Environ Res.,56:204-213, 1991.

Schwartz, J., Dockery, D.W. "Increased Mortality in Philadelphia Associated with Daily Air Pollution Concentrations." Am Rev Respir Dis, 145(3): 600-604, Mar 1992.

Schwartz, J., Dockery, D.W. "Particulate Air Pollution and Daily Mortality in Steubenville, Ohio." Am J Epid, 135:12-20, 1992.

Schwartz, J. "Air Pollution and Daily Mortality in Birmingham, Alabama." Am J Epid, 137:1136-1147, 1993.

Schwartz, J. "Air Pollution and Hospital Admissions for the Elderly in Birmingham, Alabama." Am J Epid., 139:589-598, 1994.

Schwartz, J. "Air Pollution and Hospital Admissions for the Elderly in Detroit, MI." Am J Respir Crit Care Med, 150:648-655, 1994.

Schwartz, J. "Short Term Fluctuations in Air Pollution and Hospital Admissions of the Elderly for Respiratory Disease." Thorax, 50:531-538, 1995.

Schwartz, J., Morris, R. "Air Pollution and Hospital Admissions for Cardiovascular Disease in Detroit, Michigan." Am J Epid., 50:23-35, 1995.

Schwartz, J. "Air Pollution and Hospital Admissions for Cardiovascular Disease in Tucson." Epid., 8:371-377, 1997.

Schwartz, J. "Harvesting and Long Term Exposure Effects in the Relationship Between Air Pollution and Mortality." In review, 1998.

Shy, C.M., Creason, J., Williams, R. et al. "Physiological Response in the Elderly Exposed to Ambient Particulate Air Pollution: a Pilot Study." Health Effects Institute Annual Meeting, April 1998.

Sioutas, C., Koutrakis, P., and Burton, R.M. "A Technique to Expose Animals to Concentrated Fine Ambient Aerosols." Environmental Health Perspectives, 103:172-177, 1995.

Smith, V.K. "Environmental Risk Perception and Valuation:Conventional Versus Prospective Preference Theory," in Bromley, D.W. and Segerson, K. "The Social Response to Environmental Risk." Kluwer Academic Publishers, 1992.

Speckman, P. "Kernel Smoothing in Partial Linear Models." Journal of the Royal Statistical Society, Ser. B 50 413-436.

Spengler, J.D., Treitman, R.D., Tosteson, T.D., Mage, D.T., Soczek, M.L. "Personal Exposures to Respirable Particulates and Implications for Air Pollution Epidemiology." Environ. Sci. Technol., 19, 700, 1985.

Suh, H.H., Koutrakis, P., Spengler, J.D. "Interactions Between Aerosol Acidity and Ammonia in Indoor Environments." J. Expos. Anal. Environ. Epidem., 4, 1, 1994.

Suh, H.H., Koutrakis, P., Allen, G.A., Burton, R.M. "Spatial Variation in Outdoor Acidic Sulfate and Ammonia Concentrations Within Metropolitan Philadelphia." JAWMA, 45, 442, 1995.

Suh, H.H., Koutrakis, P., Nishioka, Y., and Burton, R.M. "The New Metropolitan Acid Aerosol Characterization Study: Results for the Summer 1994 Washington, DC Field Study." Environmental Health Perspectives, 105, 8: 826-834, 1997.

Sunyer, J., Schwartz, J., Tobias, A., MacFarlane, D., Garcia, J., Ant, J.M. "Patients with Chronic Obstructive Pulmonary Disease Are a Susceptible Population of Dying Due to Urban Particles." In review.

Taylor, A.C., Evans, J.E., and McKone, T.E. "The Value of Animal Test Information in Environmental Control Decisions." Risk Analysis, 13(4):403-412, 1993.

Thomas, K.W., Pellizzari, E.D., Clayton, C.A., Whitaker, D.W., Shores, R.C., Spengler, J.D., Ozkaynak, H., Froelich, S.E., Wallace, L.A. "Particle Total Exposure Assessment Methodology (PTEAM) Study: Method Performance and Data Quality for Personal, Indoor, and Outdoor Monitoring." J. Expos. Anal. Environ. Epidem., 3, 203-226, 1993.

Thompson, K.M., and Evans, J.E. "The Value of Improved National Exposure Information for Perchloroethlene (Perc): A Case Study for Dry Cleaners." Risk Analysis, 17(2):253-271, 1997.

Tsuda, A., Butler, J.P., and Fredberg, J.E. "Effects of Alveolated Duct Structure on Aerosol Kinetics. I. Diffusional Deposition in the Absence of Gravity." J. Appl. Physiol., 76(6):2497-2509, 1994a.

Tsuda, A., Butler, J.P., and Fredberg, J.E. "Effects of Alveolated Duct Structure on Aerosol Kinetics. Ii. Gravitational Sedimentation and Inertial Impaction." J. Appl. Physiol., 76(6):2510-2516, 1994b.

Tsuda, A., Butler, J.P., Henry, F.S., Otani, Y., and Lehr, J. "Chaotic Mixing of Acinar Flow in Rhythmically Expanding Pulmonary Acinus." Am. J. Respir. Crit. Care Med., 151(4):A232, 1995a.

Tsuda, A., Henry, F.S., and Butler, J.P. "Chaotic Mixing of Alveolated Duct Flow in Rhythmically Expanding Pulmonary Acinus." J. Appl. Physiol., 79(3):1055-1063, 1995b.

Tsuda, A., Otani, Y., and Butler, J.P. "Acinar Flow Irreversibility Caused by Perturbations in Reversible Alveolar Wall Motion." J. Appl. Physiol., 1998 (Accepted).

Ultman, J.E. "Gas Transport in the Conducting Airways, In: Gas Mixing and Distribution in the Lung," edited by L.A. Engel and M. Paiva, ch.3, 63-136, Lung Biology in Health and Disease, vol. 25, Marcel Dekker, Inc, NY, 1985.

Van den Berg, M., Haaksma, J., Brouwer, J. et al. "Heart rate variability in patients with atrial fibrillation is related to Sioutas C, Koutrakis P, Burton RM. 1995b. A technique to expose animals to concentrated fine ambient aerosols." Environ Health Perspect, 103:172-177, 1997.

Vatner, S.F., McRitchie, R.J. "Interaction of the chemoreflex and the pulmonary inflation reflex in the regulation of coronary circulation in conscious dogs." Circ. Res. 37:664-673, 1975.

Viscusi, W.K., Magat, W.A., and Huber, J. "Pricing Environmental Health Risks: Survey Assessments of Risk-Risk and Risk-Dollar Tradeoffs for Chronic Bronchitis." Journal of Environmental Economics and Management, 21, 32-51, 1991.

Viscusi, W.K. "Fatal Tradeoffs: Public and Private Responsibilities for Risk." Oxford, 1992.

Ware, J.H., Ferris, B.G., Jr., Dockery, D.W. et al. "Effects of Ambient Sulfur Oxides and Suspended Particles on Respiratory Health of Preadolescent Children." Am. Rev. Respir. Dis., 133(5):834-892, 1986.

Watson, E.J. "Fluid Flow in a Model Alveolar Sac" (Appendix). J. Appl. Physiol., 37:251, 1974.

Wilson, W.E., Suh, H.H. "Fine and Coarse Particles: Concentration Relationships Relevant to Epidemiologic Studies." JAWMA, 47, 1238-1249, 1997.

Zar, J.H. "Biostatistical Analysis." Prentice-Hall, 1996.

Zeger, S., Dominici, F., Samet, J. "Harvesting Resistant Estimates of Air Pollution Effects on Mortality." In review.

Zeger, S., Thomas, D., Dominici, A., Cohen, A., Schwartz, J., Dockery, D., Samet, J.M. "Measurement Error in Time-series Studies of Air Pollution: Concepts and Consequences." Manuscript in preparation, 1998.

Ziesenis, A., Karg, E., Kreyling, W. et al. "Acute Responses of Rat Lungs to Inhaled Ultrafine Particles." American Journal of Respiratory and Critical Care Medicine,157(3):A151, 1998.