Review | Published:

Semi-continuous speciation analyses for ambient air particulate matter: An urgent need for health effects studies

Journal of Exposure Science and Environmental Epidemiology volume 19, pages 235247 (2009) | Download Citation

On the basis of an invited lecture entitled: “Air Quality and Health Researchers Working Together: Stories of Success”, and invited commentary at the Workshop closing session, both presented the EPA/HEI Second Workshop to Discuss Key Issues on Ambient Air Quality and Health Research at Research Triangle Park, NC, 16 and 17 April 2008.

Subjects

Abstract

One of the most urgent needs for future progress in reducing the substantial impacts of ambient air particulate matter (PM) on human health is to determine which of its components are having the greatest effects. The EPA's Speciation Trends Network (STN) has been operating since 2000. It generates 24-h average fine PM component concentrations for sulfate and nitrate ions, elemental and organic carbon (EC/OC), and many elements on an every third or sixth day basis for one or a few sites in most large US cities. To date, a small number of research studies, summarized in this paper, have used available STN and other supplemental data to identify and quantify the influences of specific components or source-related mixtures on measures of health-related impacts. These pioneering studies have demonstrated the potential utility of using such data in analyses that can provide a sound basis for guiding future research and control activities on those PM sources that have the greatest public health relevance. Unfortunately, the STN data collection methods used are expensive, and data have therefore been too sparse for studies of short-term health effects, where semi-continuous data, or at least daily 24-h concentration data are needed, as well as for regional concentration distributions that are needed for definitive analyses. Furthermore, because of cost considerations, there is virtually no prospect of collecting the data needed by the health researchers for more definitive analyses as long as there is continued reliance on current FRM sampling and analysis methodologies. At the second EPA-HEI Workshop on “Air Quality and Health Researchers Working Together” in RTP, NC on 16 and 17 April 2008, many participants concluded that it was both desirable, and possibly technically and economically feasible, to re-equip the STN sites with an automated system of semi-continuous monitors for sulfate, nitrate, EC, OC, and semi-continuous multistage PM samplers for non-volatile elements, providing continuous records of PM components with an averaging time of 6 h for both thoracic coarse mode PM, fine PM, and perhaps ultrafine PM as well. The availability of such data would greatly accelerate the accumulation of knowledge on PM component exposure–response relationships that would provide a sound basis more targeted air quality standards and pollution control measures.

Introduction

The mass concentration of airborne air particulate matter (PM) in particles below 2.5 μm in aerodynamic diameter (PM2.5) has been significantly associated with excess annual mortality from cardiac disease and lung cancer (Laden et al., 2000; Pope et al., 2002, 2004), whereas larger ambient air particles that deposit in the thorax (PM10−2.5) are associated with pulmonary system irritation (US EPA, 2004), and ultrafine particles (PM<0.15) have been associated with lung inflammation (US EPA, 2004). Evidence for short-term health effects has come primarily from studies of human populations using public health records, for example daily rates of mortality and/or hospital admissions in regression analyses in which measurements of PM2.5 mass concentrations at central monitoring sites were used to represent mean population exposures. Furthermore, in most multiple pollutant regression analyses that included pollutant gases such as ozone (O3), nitrogen dioxide (NO2), sulfur dioxide (SO2), and carbon monoxide (CO), the associations were generally strongest for PM2.5 (US EPA, 2004). The effects coefficients for PM mass are highly variable from place-to-place, and from time-to-time, as are the chemical compositions of the PM mixtures. This makes it highly likely that PM toxicity varies with PM composition. However, until the establishment, in 2000, of EPA's Speciation Trends Network (STN) for PM2.5, the lack of any routine data on PM2.5 speciation greatly limited the opportunities to examine the associations between PM2.5 components and health-related indices in human populations. Even now, the limitations of the STN program, that is, only one or a few sites in each city, and a sampling frequency limited to 1 in 3 or 1 in 6 days, severely hinder productive options for time-series epidemiology, especially with regard to evaluating lag structures and distributed lags. Even now, there are no data speciation data for PM10−2.5 or for PM<0.15, although EPA is committed to a limited future program of speciation analyses on PM10 filters that were co-located with PM2.5 STN filters. Analyses of co-located samples would allow for determination of PM10−2.5 species by differences in concentrations. Additional speciation data are available from the Interagency Monitoring of Protected Visual Environments (IMPROVE) network, which is comprised of 160 sampling sites across the United States, mostly in “clean” background areas.

The primary objective of the April 2008 Workshop was to discuss specific options and recommendations for concrete steps that EPA and other organizations could take to revise their ambient air monitoring programs in order to advance health effects research for the criteria air pollutants. Other objectives included: to reexamine and assess progress to date on key issues identified at the earlier workshop sponsored by the Health Effects Institute (HEI) and EPA1 and in follow-up discussions with the EPA–PM Center Directors, HEI National Particle Component Toxicity (NPACT) Directors, and other researchers; and to provide constructive feedback on draft “white papers” developed to aid in a common understanding of the issues under discussion:

  • Chemical Speciation Network (CSN), including Carbon Monitoring Issues

  • Access to EPA's air quality data for health researchers

  • Air quality sampling: benefits and costs of daily health-targeted monitors for fine particle components

  • Long-term communication strategies for improving interactions between health researchers and air quality staff on changes in ambient air monitoring specifically:

    • Network design and site selection approval

    • Methods implementation

Health effects research using daily and speciation monitoring: success stories

Background

It is important to remember that EPA's Air Information Retrieval System (AIRS) exists primarily as a regulatory tool to determine compliance with National Ambient Air Quality Standards (NAAQS), and to assist State Agency regulators in devising State Implementation Plans (SIPs) that can reduce or eliminate non-compliance, that is, exceedances of the mass-based NAAQS. Advantages of AIRS data for health effects research include: accessible information on site location, site characteristics, a high level of quality assurance, and inclusion of all measurement data, not just those on exceedances, as in water quality and other EPA measurement programs.

Despite all of its limitations in terms of frequency of measurements, and of spatial coverage, the peer-reviewed epidemiological literature based on AIRS data for PM mass has greatly helped EPA in the development of better targeted and more health-protective PM NAAQS in 1987, 1997, and 2007, and has enabled health researchers to better refine their needs for additional monitoring data that will enable them to better quantify exposure–response relationships for PM components. Successful applications of the STN data for associating PM components that are especially toxic have relatively rare.

Most of the “success stories” that follow have relied on sources of speciation data other than those currently available from the STN. They are included herein to the extent that they demonstrate how an enhanced STN could enable health effects researchers to conduct more comprehensive studies on the effects of PM components, and to generate new insights on the nature of the underlying biology.

Types of Speciation Data Currently Available for Health Effects Research

For some PM components that account for significant mass fractions, there are semi-continuous direct-reading air quality monitors. These include monitors for sulfate and nitrate ions, as well as thermal ramp—optical density meters calibrated to generate data on elemental and organic carbon (EC/OC). The STN and IMPROVE networks provide EC and OC data, as well as data on elements in PM samples collected for 24-h on Teflon sampling filters and measured by non-destructive X-ray fluorescence (XRF) analysis. Problems in interpretation may arise from not knowing the chemical form(s) associated with each element, and from readings of some elements being near or below the lower detection limits.

Source-Apportionment Analyses

Although it may be possible to identify a particular chemical entity in ambient air PM with particular health effects, as for nickel (Ni) with cardiac function in a mouse model of atherosclerosis (Lippmann et al., 2006), this situation has, until now, been the exception, rather than the rule. More commonly, it has been possible to use single and/or multiple elements as tracers of PM arising from specific source categories through factor analysis. The major source categories contributing to the mass of ambient air PM2.5, that is, coal combustion, residual oil combustion, traffic movements, resuspended soil, metals refineries and processing, and secondary PM2.5 formed within the atmosphere, have sufficiently different mixture compositions to be separated analytically using source-apportionment models (Thurston et al., 2005).

Spatial U.S. Variability of PM2.5 Factors Confirms Their Source Interpretations

As shown in Figure 1 (Bell et al., 2007), the mass concentrations obtained from source apportionments can be visually depicted as geographic distributions over a map of the US. The map of residual mortality risk produced by Burnett et al. (2001) in Figure 2, when matched with the sulfate concentration map in Figure 1, provides support for the hypothesis that an influential source category, such as coal combustion, may be responsible for excess annual mortality.

Figure 1
Figure 1

Sulfate particulate matter (PM)2.5 average (μg/m3) for 187 US counties, 2000–2005 (adapted from Bell et al., 2007).

Figure 2
Figure 2

Residual mortality risks in the ACS cohort study after controlling for individual risk factors (adapted from Burnett et al., 2001).

Success stories from the NYU PM center and NPACT research

Daily speciation data suitable for determining acute responses to peaking exposures to PM2.5 and its components are not routinely available. Such data were, however, available during a series of daily 6-h weekday exposures of a mouse model of atherosclerosis to concentrated airborne particles (CAPs) in Tuxedo, NY, a small rural community that is 80 km northwest (upwind) of NYC. As shown in Figure 3 (Lippmann et al., 2006), for those mice with implanted electrocardiographic (ECG) monitors, there were much larger than normal excursions of heart rate (HR) and HR variability (HRV) on 14 days during the series, and that these 14 days had abnormally low concentrations of PM2.5 and sulfur, but extraordinarily high concentrations of Ni, Cr, and Fe on the basis of the speciation data. Back-trajectory analysis indicated the air masses on those 14 unusual air quality days had all passed close by the largest Ni smelter in North America, whose 1000-ft tall discharge stack facilitated long-range airborne transport.

Figure 3
Figure 3

Comparisons of concentrated airborne particles (CAPs) mass, CAPs component concentrations, heart rate, and heart rate variability for ApoE−/− mice undergoing daily 6-h exposures over 6 months in Tuxedo, NY, for 14 days with northwest winds vs the 89 days with all other wind directions (from Lippmann et al., 2006, courtesy Environ Health Perspect).

I was aware of suggestive evidence from research on the health benefits of a mandated switch to low-sulfur fuels in Hong Kong in 1990 that were associated with substantial reductions in SO2 (Hedley et al., 2002) as well as of V and Ni (Hedley et al., 2004). I wondered whether Ni might have been responsible for the notably high daily mortality associated with PM10 in NYC (Dominici et al., 2003). The PM10 mortality coefficient for NYC in the NMMAPS study re-analysis was 3.8 × higher than the national average, whereas Ni was 9.5 × higher than average for the 60 NMMAPS communities with speciation data for 2000–2003.

Figure 4 shows the resulting difference in the PM10 mortality risk estimates (in percent per 10 μg/m3 increase in PM10) per 5th to 95th percentile difference in the PM2.5 available. For example, for Ni and V, the PM10 risk coefficients (per 10 μg/m3) were high (0.6) in the MSAs where Ni and V were significantly high (95th percentile), compared to the MSAs where Ni was low (5th percentile). These differences in magnitude were not small, as the nationwide combined estimate for the 90 MSAs in the NMMAPS study was 0.21. Ni and V, which are most strongly associated with residual oil combustion effluent, showed the strongest predictions of the variation in PM10 risk estimates across the NMMAPS MSAs, followed by elevated, but non-significant increases above 0.21 that were associated with EC, Zn, SO42−, Cu, Pb, and OC. The metals most closely associated with resuspended soil, that is, Al and Si, had the lowest values, suggesting that they were unlikely to be influential on daily mortality. Thus, PM2.5 components appear to explain some of the MSA-to-MSA variation in the NMMAPS PM10 daily mortality risk coefficients. A subsequent study by Dominici et al. (2007), using additional years and a few more cities of speciation data, confirmed the Lippmann et al. (2006) findings, but noted that, without data from NYC, the association of daily mortality with Ni was no longer statistically significant. On the basis that NYC, with both its very large population and extraordinarily high concentrations of airborne Ni, is driving the statistically significant associations between daily mortality and Ni, I concluded that a closer examination of the causes and effects of high concentrations of Ni in NYC warranted further examination. As described in a recent paper (Peltier et al., in press), the NYU PM center assembled all available STN, IMPROVE, and NYU's own research study data on Ni concentrations in the ambient air of NYC and the States of New Jersey and Connecticut. It is notable that there are no STN or IMPROVE sites in Brooklyn, Staten Island, or most of Queens, where the majority of the population lives.

Figure 4
Figure 4

Differences in daily mortality risk coefficients per the 5th to 95th percentile difference in FPM component concentrations across NMMAP MSAs (for the 60 MSAs for which FPM speciation data were available—courtesy Environ Health Perspect).

Figure 5 clearly demonstrates that Ni concentrations at all sites in NYC have a pronounced seasonal pattern, with much higher concentrations in the winter than in the summer. It is also notable that the lowest Ni concentrations in NYC were at Canal Street on the Hudson River shore in Lower Manhattan, and at Queens College in a neighborhood of low-rise buildings in eastern Queens. Finally, it should be noted that the ratio of Ni to V is unusually high at these NYC sites. By contrast, in Figure 6, showing data from sites to the west in New Jersey, Ni concentrations are about an order of magnitude lower than in NYC, the strong seasonality is missing, and the ratio of Ni to V is much lower. These data, and those from the few sites in Connecticut are summarized in Table 1. Note that at the remotely located IMPROVE sites at Mohawk Mountain in northwest Connecticut and at Brigantine on the Atlantic Ocean shore in southern New Jersey, the Ni concentrations were below 1.5 ng/m3, whereas at the Chester, NJ, STN site on the western outer edge of the NYC metropolitan area, they averaged 2.3 ng/m3. The extremely high concentrations of Ni at STN sites in NYC in the winter (>18 ng/m3), and the spatial and seasonal gradients within NYC, are consistent with the widespread use of residual oil for space and domestic water heating in older large commercial and residential buildings in Manhattan and the Bronx. We have initiated a study to determine the spatial gradients throughout NYC, including the low-rise areas of Brooklyn and Queens, where most NYC residents live, and plan to determine if seasonal rates of hospital admissions for Medicare patients are correlated with neighborhood concentrations of Ni or other measured PM components.

Figure 5
Figure 5

Smoothed time series of nickel and nickel/vanadium ratio for sites within New York City—from Peltier et al. (in press).

Figure 6
Figure 6

As in Figure 5, a time series of nickel and nickel/vanadium ratio for sites in the New Jersey area. This also includes a site in Sterling Forest (Tuxedo, NY)—from Peltier et al. (in press).

Table 1: Nickel and vanadium mean concentrations (±standard error), their ratio, and the increase of winter Ni above summer Ni in NYC and vicinity

Success story from the Harvard PM center

Franklin et al. (2008) assembled daily mortality and PM2.5 data for at least 4 years for 25 US cities (2000–2005), and developed city-specific daily mortality coefficients for each city. They then used speciation data that were available for at least 2 years, plus daily weather data, to examine the possible roles of each of the available PM2.5 components as a modifier of the apparent effect of PM2.5. They also examined whether they could identify a cluster of components that could account for most or all of the intercity heterogeneity of daily mortality coefficients.

They did regression Poisson regression analyses, controlling for lags, temperature, season, year, mean seasonal concentration ratios for PM2.5 and components, influence of components on PM2.5 mortality coefficients, and potential confounding factors: SES, age, temperature, air conditioning, and so on.

As shown in Table 2 Al, As, Ni, Si, and SO4 were significant as effect modifiers for PM2.5 association with excess daily mortality in 25 US cities and Al, Ni, and either SO42− or As, can account for 100% of the heterogeneity of the PM2.5 association with daily mortality. V was not a significant effect modifier (suggesting that the associations of health effects of residual oil fly ash (ROFA) are due mostly to Ni). There were significant associations with Al (and Si), but no significant associations with OC and EC (suggesting that health effects associated with proximity to major roadways may be due more to resuspended surface dust than to vehicle exhaust).

Table 2: Effect modification of composition on the estimated percent increase in mortality with a 10-μg/m3 increase in PM2.5 (from Franklin et al. 2008)

Success story from the Rochester PM center

Yue et al. (2007, 2008) used PM component data that were collected specifically for the purpose of their study of the role of community air pollution on prolonged repolarization and increased levels of inflammation in male coronary artery disease patients in Erfurt, Germany. They utilized air quality data resources that were collected between September 1997 and August 2001. As shown in Figure 7, particle counts (PCs) were made in size intervals ranging from 0.01 to 3.0 μm. Air pollutant concentration data were available for: O3, NO, NO2, CO, SO2, SO4, OC, and EC.

Figure 7
Figure 7

Particle-size profiles of five particle source factors obtained by positive matrix factorization (PMF) in Erfurt, Germany, in the Yue et al. (2007) study.

They performed source identification by Positive Matrix Factorization, using particle count data within specific size ranges of ultrafine particles (UFP) to identify five specific source categories: (1) airborne soil; (2) UFP from local traffic; (3) secondary aerosol from local fuel combustion; (4) particles from remote traffic sources; and (5) secondary aerosol from multiple sources. The health-related parameters measured in their subjects (men, average age=66 years) were, for ECG, QT interval, and T-wave amplitude; and for blood: von Willebrand factor (vWF) and C-reactive protein (CRP). The results of this source apportionment were used to explore adverse health effects of source-specific PM (Yue et al., 2007). From their linear and logistic regression models, they found that an increase in QT interval and a decrease in T-wave amplitude were associated with traffic-related particles exposure during 0–23 h before the ECG recordings. Both traffic-related particles and combustion-generated aerosols, at different exposure lags, were related to an increase in the inflammatory marker vWF. All of the source types showed positive associations with CRP levels above the 90th percentile (8.5 mg/l). This combination of results suggested that traffic-related and combustion-generated particles had stronger adverse health impacts with regard to cardiac effects, and that particles from other sources induced acute phase responses. This was the first instance of using source apportionment based on particle size distributions in health effects modeling.

Success story from the Southern California PM center

Delfino et al. (2008) measured circulating biomarkers of inflammation, antioxidant activity, and platelet activation to determine their association with ultrafine particles and primary combustion aerosols in elderly subjects with a history of coronary artery disease (13 male and 17 female subjects, average age=86 years, non-smokers with coronary artery disease). They used air quality data resources that were available for July 2005–February 2006 in San Gabriel Valley, CA. These were: hourly total PC data; hourly EC and OC data; 24-h concentration data for PM0.25, PM0.25−2.5, PM2.5−10, O3, NO2, and CO. They also estimated primary and secondary OC, for both indoor and outdoor microenvironments. The health-related parameters measured in weekly blood samples were: CRP, fibrinogen, IL-6, TNFα, VCAM-1, ICAM-1, fibrin D-dimer, sP-selectin, MPO, SOD, and GPx-1. There were also weekly measurements of exhaled breath NO, a marker of pulmonary inflammation.

As shown in Figure 8, there were significant positive associations of a biomarker of systemic inflammation (IL-6) with PM, which were largely driven by markers of primary combustion sources (EC, BC, primary OC, and PC), but not with total OC or secondary OC. This suggests that aged and more oxidized OC in secondary OC is less toxic than freshly generated OC associated with combustion effluents. Outdoor PM concentrations were associated with indoor exposures to PM of outdoor origin.

Figure 8
Figure 8

Significant source-related elevations in IL-6 with outdoor carbonaceous aerosols and particle number (per IQR) in the Delfino et al. (2008) study.

Particle number (PN) concentrations, and quasi-ultrafine concentrations were more strongly associated with biomarkers than were accumulation mode PM and secondary organic aerosol (SOA).

The inverse association of a biomarker of antioxidant activity (SOD) with PM was also largely driven by markers of primary combustion sources (EC, BC, primary OC). Inactivation of antioxidant enzymes (SOD, GPx-1) by ROS, RNS, or electrophiles may be one mechanism of pollutant-induced systemic inflammation and thrombosis.

Success story from the Seattle PM Center

Allen et al. (2008) reported on the associations between PM components (PM2.5 of outdoor origin, PM2.5 of indoor origin, BC, and levoglucosan, a marker for wood smoke) and indices of pulmonary responses (exhaled NO, FEV1, FVC, PEF, and MEF) in a panel study of residents of Seattle, WA.

As shown in Figure 9, there were significant increases in exhaled NO associated with personal exposure to BC and PM2.5, as well as to outdoor PM2.5. Significant reductions in PEF and MEF were associated with outdoor concentrations of PM2.5, BC, and levoglucosan. Reduced FEV1 was associated only with outdoor levoglucosan.

Figure 9
Figure 9

Results of 1999–2001 panel study in Seattle, WA, from Allen et al. (2008)—courtesy Inhal Toxicol.

Success story from collaborations among the PM centers and EPA's national laboratories

Ito et al. (2006) reported on the analyses of the associations between daily mortality in Washington, DC, and components based on PM2.5 sources using nine different source-apportionment methods and Mar et al. (2006) reported on comparable analyses for Phoenix, AZ. Ito et al. (2006) used PM2.5 speciation data for every third day in Washington that were collected from 1988 through 1997. Mar et al. (2006) used analyses of PM2.5 components in Phoenix, AZ, collected daily from 1995 through 1998. Both used nine different source-apportionment methods performed by teams at: Clarkson University, University of Washington, University of Southern California, University of Rochester, New York University, and Harvard University.

For Washington, significant associations of the apportioned anthropogenic PM2.5 source categories were secondary sulfate (lag 3), primary coal combustion (lag 3), and resuspended soil (various lags), Some models also showed significant associations for traffic. The associations were similar for total mortality and for cardiovascular and cardiorespiratory causes.

Of the apportioned anthropogenic PM2.5 source categories in Phoenix, secondary sulfate (lag 0), traffic (lag 1), and copper smelter-derived particles (lag 0) were the sources most consistently associated with cardiovascular mortality.

Other recent success stories relating PM sources to health effects

Andersen et al. (2007) examined associations between PM10 and PM10 components (biomass, secondary, oil, crustal, sea salt, and traffic) and daily hospital admissions over a 6-year period (1999–2004) in Copenhagen. For adults over 65 years of age, PM10 was significantly associated with admissions for cardiovascular and respiratory causes. In two-pollutant models with PM10 and a specific PM10 source category for cardiovascular admissions, PM10-crustal was significant whereas PM10 was not. In two-pollutant models with PM10 and a specific PM10 source category for respiratory admissions, PM10-biomass was significant whereas PM10 was not. For both categories of admissions, PM10 remained significant whereas the categories of oil, salt, and traffic were not.

Sarnat et al. (2008) studied the influence of PM2.5 sources (gas engines, diesel engines, wood smoke, resuspended soil, secondary sulfate, secondary nitrate, cement kiln, railroad, and metal processing) on cardiorespiratory morbidity in Atlanta, GA, for 4 years (1999–2002). There were clear positive associations between PM2.5 attributed to mobile sources and biomass burning and emergency department (ED) visits for cardiovascular disease. For the summer months, PM2.5 sulfate was significantly associated with respiratory ED visits. The results were similar for different source-apportionment methods.

STN data have been sufficient to enable research investigators at NYU (Lippmann et al., 2006) and Johns Hopkins (Bell et al., 2007) to gain new insights into national variations in source-related PM components. Daily speciation data collected for 3 years in Sterling Forest by NYU investigators (Figure 3) enabled them to identify acute effects of Ni that were attributable to a distant upwind source on cardiac function in a mouse model of atherosclerosis.

STN data have been sufficient to enable research investigators at NYU (Lippmann et al., 2006), and Harvard (Franklin et al., 2008) to gain new insights into the influences of PM2.5 components (e.g., Ni, Al, and SO4) and/or tracers on daily mortality and other health-related outcomes in human populations.

A panel study of Seattle residents (Allen et al., 2008) had increased exhaled NO in association with personal exposure to BC and PM2.5, and with ambient PM2.5; decreases in PEF and MMEF were associated with ambient BC and PM2.5, whereas ambient levoglucosan was associated with reduced FEV1 as well as PEF and MMEF.

NYU investigators used STN data (Figures 5 and 6) have been used to demonstrate the nature and substantial extent of spatial and seasonal variability of Ni in a region (New York Metro) with an atypical source category (space heating with residual oil combustion).

Hourly data on PN concentration within size intervals within the range from 0.01 to 3.0 μm (Figure 7), when used in conjunction with data on the concentrations of O3, NO, NO2, CO, SO2, SO42−, OC, and EC in Erfurt, Germany, enabled the Rochester PM Center to identify five major PM source categories and their significant associations with ECG and blood parameters in a panel of coronary arterial disease patients.

Daily data on PM0.25, PM0.25−2.5, and PM2.5−10, when used in conjunction with hourly total PC, EC, O3, NO2, CO, primary and secondary OC, enabled the Southern California PM Center to find positive associations for EC, BC, and primary OC with a biomarker of systemic inflammation (Figure 8). The associations with PC and PM0.25 were greater than those with accumulation mode PM.

Closing remarks based on the workshop discussions

The Workshop did not have a closing panel discussion or mechanism for coming to consensus views on the next steps on enhanced air quality monitoring for health effects studies, leaving decisions on that to its EPA sponsor. I took advantage of the invitation to make some observations on the state of our knowledge concerning AQS data resources and their usage in health effects studies at the final Workshop session. What follows are my own personal perspectives on the lessons I learned at the Workshop and on the next steps that I would advocate.

In my view, the success stories that I reviewed in the opening session have shown that:

  • there is a great potential for non-routine measurement data on PM2.5 composition and size distributions to elucidate the substantial differences of specific components and PM sources in different airsheds.

  • there are significant and provocative associations of PM2.5 components: Ni, Al, As, SO42−, and Particle Counts [PCs] in specific size ranges, with health-related parameters in human populations and animal models.

  • STN data, which are currently limited to every third or sixth day at a very limited number of sites, have not been useful for studies for studies of acute responses in humans.

  • improvements can be made in the STN in order to enhance the number of routine measurements of air quality parameters being made, and their frequency that will enable health researchers to perform studies to characterize the nature and extent of the effect of PM components on human health.

Based on presentations at the Workshop, I came to some conclusions about the key topics that were discussed in some detail. The following are the key issues, and my conclusions concerning what is currently known and not known about them:

(1) Discussion Topic: EC and OC Measurements

Question posed: Will a new STN protocol for EC/OC help?

Answer: It is not clear, since there were, at the end of the session, several unresolved issues.

  • While EC showed more consistent associations with health-related effects than did OC, it was not clear if this was due to measurement method related differences for OC.

  • Is OC too broad an index? It may be that the effects really attributable to primary OC are masked by the presence of less toxic secondary OC.

  • Are the apparent effects of EC due to the carbon core particles? They may be due to the adsorption of gaseous reaction products of atmospheric chemistry, which are carried into the deep lung airways on EC surfaces.

  • Will transition to clean diesel engine technology be sufficient for reducing health risks associated with EC? This is a possibility, but a speculative one at this time.

(2) Discussion Topic: Can We Identify Causal PM Components for Health Effects Associated with Proximity to Heavy Traffic?

Answer: We don’t know yet, since there were, at the end of the session, several unresolved questions. These are:

  • Are UFP important in terms of particle count – or in terms of their chemical components?

  • Is the frequently observed association of health effects with proximity to heavy traffic due to: tailpipe emissions (e.g., EC and primary OC), or to Al and/or Zn from engine block wear?

  • Could the frequently observed association of health effects with proximity to heavy traffic due to components of re-suspended road dusts (e.g., Al, Si, latex, road salt, engine oil)?

(3) Discussion Topic: What Accounts for Associations of Sulfate Ion with Health Effects?

Question posed: What is sulfate ion a surrogate for?

Answer: We don’t yet know, since there were, at the end of the session, several unresolved questions. These included:

*Is it H+ per se, acting as an irritant, or via its role in the solubilization of toxic metals?

*Is SO42− serving as a surrogate marker for peroxides and their irritant effects?

*Is sulfate ion serving as a marker for coal and residual oil combustion effluents?

(4) Discussion Topic: What Accounts for the Toxicity of ROFA (Residual Oil Fly Ash)?

Question posed: Is Ni a causal factor for ROFA toxicity, or is it a surrogate marker for co-pollutants in ROFA?

Answer: Ni may be a causal factor for cardiovascular disease by itself, but there remained, at the end of the session, several unresolved questions: If Ni is not a causal factor for a wide range of effects associated with ROFA exposures, does it serve as a surrogate marker for:

  • V, which is more closely associated with pulmonary effects?

  • Other metals that may modulate toxicity?

(5) Discussion Topic: Are Current Measurement Methods and Temporal Resolution for PM Components Adequate to Support Health Effects Research?

Answer: Most Workshop participants did not think the currently used STN methods and their frequency were adequate. To address the deficiencies of the STN for this purpose, a number of participants recommended phasing out reliance on 24-h filter samples and transitioning to more PM-component continuous monitoring. In this context, the issue of which components should be monitored continuously, and whether satisfactory instruments were available for such monitoring were discussed. These discussions identified several unresolved questions, including:

  • Are BC measurements adequate exposure surrogates for risk assessments for carbonaceous PM?

  • Answer: Yes, in some cases, but probably not in all.

  • If not, does the Sunset semi-continuous EC/OC instrument provide sufficient information on the ambient air concentrations of EC and OC?

  • Answer: That depends on whether there is an important role for OC, which has not yet been established.

  • Are we there yet for reliable semi-continuous measurements of SO42−and NO3?

  • Answer: Yes for SO42−, not yet clear for NO3.

  • Should we encourage use of automated dichotomous (dichot) samplers for semi-continuous measurements of both PM2.5 and PM10–2.5 mass and non-volatile elemental concentrations?

  • Note:The dichot would provide simultaneous samples (on Teflon filters) for PM2.5 and for PM10–2.5, (with about 10% of the PM2.5 on the PM10–2.5 filter). An advantage is that there would be only one sampler to maintain, instead of two.

  • Alternatively, should we switch to the DRUM sampler (a cascade impactor) and Synchrotron XRF analysis for cost-effective semi-continuous measurements of UFP, accumulation mode, and coarse thoracic concentrations? [Note: The Synchrotron is a high energy accelerator at the Berkeley National Laboratory of the Department of Energy.] At the Workshop, William Wilson of EPA estimated that use of the DRUM sampler (Pere-Trepat et al. 2007), with elemental analyses by Synchrotron XRF would permit semi-continuous analyses of 6-h average concentrations for ultrafine, accumulation mode, and coarse thoracic PM components at a cost comparable to that now spent on 24-h Teflon filter samples collected every third day for PM2.5 alone.

  • Answer: If Dr. Wilson's cost estimates are realistic, the data generated using this approach would be an invaluable resource to health effects investigators.

Discussion Topic: Access to Air Quality Network Data

Question posed: Is the existing AIRS database (on PM mass concentrations) adequate for epidemiological research?

The Workshop consensus was: Almost, and getting better.

However, there were, at the end of the session, some unresolved questions, for which there were no clear answers.:

  • Is there currently an adequate capacity for downloading data?

  • Is there adequate labeling of the latest database version for efficient utilization?

  • Can there be access to the full dataset as well as summary data for researchers?

  • Can EPA help researchers find a pathway to databases of state and local agencies?

Discussion Topic: Strategic Approach to Scoping PM Component Measurements Needs in Support of Epidemiological Research

The discussion that followed was largely devoted to framing the questions, i.e.:

(1) How can researchers and regulators characterize spatial and temporal variability of PM components of current concern that are known to be spatially heterogeneous? These include:

  • Traffic-related components: UFP, EC, primary OC, Si, Al.

  • Ni from ships in seaports, space heating.

(2) How to establish local networks to determine concentration contours for pollutants that are known to be spatially heterogeneous? Needed guidance includes:

  • Locating core and satellite sites in urban areas.

  • Selecting a suite of measurements to be made at each site

Topic: Strategic Approach to Scoping PM Component Measurements Needs in Support of Epidemiological Research

The discussion that followed was largely devoted to framing the questions, i.e.

(1) How to select of a limited number of MSAs for scoping study. This involves:

  • Considering the different PM characteristics by particle size distributions and source-related mixtures.

  • Considering the population size needed for epidemiological study power.

  • Determining whether populations currently under study are suitable.

(2) How to provide support for cohort and other cross-sectional studies of chronic health effects. The first step is to determine:

  • Are data being collected for daily concentrations appropriate for consolidation into long-term average concentrations?

  • If data are not available, or not appropriate, should multi-day samples be collected for economical consideration?

Discussion Topic: Next Steps

My Own Recommendations to EPA:

  • Have Workshop Steering Committee analyze Workshop's lessons learned, and use them to prepare recommendations for next steps.

  • Have the Steering Committee request that EPA's National Exposure Research Library (NERL) conduct feasibility studies of more continuous monitoring methods for PM components to replace analyses based on 24-h filters collected every third or sixth day.

  • Have the Steering committee prepare a request to EPA's Clean Air Scientific Advisory Committee (CASAC) for a scientific review of its recommendation and their capacity to facilitate future epidemiological research on the health effects of PM components.

Notes

  1. 1.

    HEI and EPA co-sponsored a meeting in late 2006 to discuss how the use of the accumulating data derived from nationwide monitoring of fine particulate matter (PM) components can facilitate current and future health effects studies and improve comparisons of risk estimates across studies. The workshop discussions illuminated issues associated with accessing and analyzing monitoring data and identified needs of the health effects research community regarding monitoring of fine particle components. See http://www.healtheffects.org/AQDNov06/AQDWorkshop.html for more information.

References

  1. , , , , , , and Changes in lung function and airway inflammation among asthmatic children residing in a woodsmoke-impacted urban area. Inhal Toxicol 2008: 20: 423–433.

  2. , , , , and Ambient particle source apportionment and daily hospital admissions among children and elderly in Copenhagen. J Expo Sci Environ Epidemiol 2007: 17: 625–636.

  3. , , , , and Spatial and temporal variation in PM2.5 chemical composition in the United States for health effect studies. Environ Health Perspect 2007: 115: 989–995.

  4. , , , , , , and The spatial association between community air pollution and mortality: a new method of analyzing correlated geographic cohort data. Environ Health Perspect 2001: 109(Suppl 3): 375–380.

  5. , et al. Circulating biomarkers of inflammation, antioxidant activity, and platelet activation are associated with ultrafine particles and primary combustion aerosols in elderly subjects with a history of coronary artery disease. Environ Health Perspect 2008: 116(7): 898–906.

  6. , , , , and National maps of the effects of particulate matter on mortality: exploring geographic variation. Environ Health Perspect 2003: 111: 39–44.

  7. , , , , , and Does the effect of PM10 on mortality depend on PM nickel and vanadium content? A reanalysis of the NMMAPS data. Environ Health Perspect 2007: 115: 1701–1703.

  8. EPA. Ambient Air Quality Monitoring and Health Research: Summary of April 16–17, 2008 Workshop to Discuss Key Issues. EPA, Research Triangle Park, NC, 2008.

  9. , , and The role of particle composition on the association between PM2.5 and mortality. Epidemioology 2008: 19(5): 680–689.

  10. , , , , , and Cardiorespiratory and all-cause mortality after restrictions on sulphur content of fuel in Hong Kong: an intervention study. Lancet 2002, 1646–1652.

  11. , , and The change in sub-species of particulate matter [PM10] before and after an intervention to restrict sulphur content of fuel in Hong Kong. Poster presented at Better Air Quality/Asian Development Bank meeting at Agra, India 2004.

  12. , , , , , , , , , , and PM source apportionment and health effects. 2. An investigation of inter-method variations in associations between source-apportioned fine particle mass and daily mortality in Washington, DC. J Expos Anal Environ Epidemiol 2006: 15: 1–11.

  13. , , , and Association of fine particulate matter from different sources with daily mortality in six U.S. cities. Environ Health Perspect 2000: 108: 941–947.

  14. , , , , and Cardiovascular effects of nickel in ambient air. Environ Health Perspect 2006: 114: 1662–1669.

  15. , , , , , , , , , , , , and PM source apportionment and health effects. 3. Investigation of inter-method variations in associations between estimated source contributions of PM2.5 and daily mortality in Phoenix, AZ. J Expos Sci Environ Epidemiol 2006: 16: 311–320.

  16. , , , and Residual Oil Combustion: a major source of airborne nickel in New York City. J Expos Sci Environ Epidemiol (in press).

  17. , , , and Source apportionment of time and size resolved ambient particulate matter measured with a rotating drum impactor. Atmos Environ 2007: 41: 5921–5933.

  18. , , , , , . et al. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. J Am Med Assoc 2002: 287: 1132–1141.

  19. , , , , , , et al. Cardiovascular mortality and long-term exposure to particulate air pollution. Epidemiological evidence of general pathophysiological pathways of disease. Circulation 2004: 109: 71–77.

  20. , , , , , , , , and Fine particle sources and cardiorespiratory morbidity: an application of chemical mass balance and factor analytical source-apportionment methods. Environ Health Perspect 2008: 116: 459–466.

  21. , , , , , , , , , , , , , , , and Workgroup Report: Workshop on source apportionment of particulate matter health effects—intercomparison of results and implications. Environ Health Perspect 2005: 113: 1768–1774.

  22. US EPA. Environmental Protection Agency. Air Quality Criteria for Particulate Matter. National Center for Environmental Assessment-RTP Office, Research Triangle Park, NC, report no. EPA/600/P-99/002aF 2004.

  23. , et al. Ambient source-specific particles are associated with prolonged repolarization and increased levels of inflammation in male coronary artery disease patients. Mutat Res 2007: 621: 50–60.

  24. , et al. Source apportionment of ambient fine particle size distribution using positive matrix factorization in Erfurt, Germany. Sci Total Environ 2008: 398(1–3): 133–144.

Download references

Acknowledgements

The preparation of this paper was supported by a research grant from the Health Effects Institute, and is part of a Center Program supported by Grant ES 00260 from the National Institute of Environmental Health Sciences.

Author information

Affiliations

  1. New York University School of Medicine, Tuxedo, New York, USA

    • Morton Lippmann

Authors

  1. Search for Morton Lippmann in:

Corresponding author

Correspondence to Morton Lippmann.

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/jes.2008.65

Further reading