Health Effects Institute

 

 

 

 

 

 

Charlestown Navy Yard

120 Second Avenue

Boston, MA 02129 USA

tel:  +1 617 886 9330 x311

fax: +1 617 886 9335

http://www.healtheffects.org

 

 

Testimony of

Robert O’Keefe

Vice President, Health Effects Institute

 

Before the

 

Senate Committee on Environment and Public Works

 

Hearing on

 

Health Effects of Particulate Matter

October 2, 2002

 

 

            Mr. Chairman, members of the Committee, it is a pleasure to have this chance to appear before you to share the perspective of the Health Effects Institute on what we have learned and what we still need to learn about the health effects of particulate matter.  For the record, I am Robert O’Keefe, Vice President of the Health Effects Institute, an independent research institute funded jointly and equally by the US EPA and industry to provide impartial and high quality science on the health effects of air pollution.

 

The Data We Had in 1997 – Short and Long Term Epidemiology

 

            In 1997, the US EPA promulgated a new set of National Ambient Air Quality Standards (NAAQS) for fine particulate matter (PM2.5).  In large measure, that action was based on two types of epidemiology studies:

 

·       There were nearly 40 short term studies that found a statistical relationship between daily changes in air pollution and daily small but relatively consistent increases in daily levels of death, hospitalization, and illness (e.g. 1% to 2% increases in mortality for every 10 microgram/cubic meter increase in PM10);

 

·       Two long term “cohort” studies – the Harvard Six Cities Study and the Pope/American Cancer Society Study - that tracked selected populations of people in a series of more- and less- polluted cities, and found that those who lived in the most polluted cities had between a 17% and 26% higher risk of premature death than those who lived in the least polluted cities.

 

These studies suggested that a measurable portion of mortality and respiratory and cardiac illness in the United States might be attributable to fine particle air pollution, and based on them, EPA set the new, more stringent NAAQS for PM2.5.  At the same time, there were a number of questions about these studies, key among them:

 

·       The individual short term studies were done by diverse investigators using somewhat different methods – would a more systematic study find the same results?

 

·       Could other pollutants, which occur along with PM2.5 , be more likely to be responsible for the increased mortality?

 

·       Did the deaths measured in these short-term studies represent substantial losses of life years, or the advancing of death for critically ill people by a few days?

 

·       Did the exposures measured in these studies – at central air pollution monitors – accurately represent the exposures of people who in general spend most of their time indoors?

 

·       Could the Harvard Six Cities Study and the American Cancer Society Study, whose data had only been analyzed by the original investigators, stand up to intensive scrutiny and analysis from new, independent investigators?  Could there be other differences between the cities (e.g. differences in socioeconomic status or health care) that would also explain the differences in mortality?

 

In addition to these questions about the epidemiology, there were also questions about the relative toxicity of the many different components of the complex PM mixture, and about the possible biological mechanisms that might explain the epidemiology results, questions that were laid out in a 1998 priority research agenda by the National Academy of Sciences Committee on Research Priorities for Airborne Particulate Matter.

 

What Have We Learned Since 1997?

 

Since 1997, substantial new research has been undertaken to advance our understanding of the health effects of PM.  As one part of the larger effort undertaken, HEI has invested in some 40 epidemiology, exposure, and toxicology studies to test the validity of the original studies, and to begin to answer the remaining questions.

 

Key among HEI’s work have been two efforts to determine the validity of the short and long term epidemiology studies – the National Morbidity, Mortality, and Air Pollution Study (or NMMAPS), and the Reanalysis of the Harvard Six Cities and American Cancer Society studies.

 

The National Morbidity, Mortality, and Air Pollution Study (or NMMAPS)

 

NMMAPS is a systematic study of air pollution, weather and mortality in the 90 largest cities in the United States, conducted - under the oversight, quality assurance procedures, and review of HEI - by investigators at Johns Hopkins University.  NMMAPS also included similar analyses of air pollution and elderly hospitalization, conducted in 14 US cities by investigators at Harvard University.

 

In brief, this systematic and rigorous study found a consistent relationship between PM10 and mortality in the 90 largest cities of an approximately 0.4% increase in mortality for every 10 micrograms increase in PM10 .  This level of effect was about half the size of that found in the earlier study, but as the graph in my testimony illustrates, this effect was not substantially affected by any of the other gaseous air pollutants.  (See Figure 1)  The NMMAPS investigators also found that at least a portion of the mortality was not solely frail people dying a few days early, but deaths advanced 30 days or more, and conducted analyses that suggested that errors from using centrally monitored air pollution to estimate exposure were not likely to change the basic results.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Fig.1 NMMAPS Analyses of PM vs. Other Pollutants

 

At the same time, this first nationwide analysis found differences in levels of effect across the US, suggesting that other factors, perhaps different mixes of pollution, could contribute along with particles to the effect. (See Figure 2) Overall, the NMMAPS analyses provided greater confidence in the results of the short-term epidemiology.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Fig.2 NMMAPS Regional Analyses

 

 

NMMAPS Update: This past Spring,  members of the original team of investigators  from the Johns Hopkins School of Public Health were conducting additional analysis on their findings of an association between daily changes in air pollution and mortality. In the course of testing these analyses against different assumptions and examining the methods used they identified a generally unknown aspect of S-Plus, a statistical software package widely used by air pollution and other investigators to fit general additive models (GAMs) to data. In NMMAPS the investigators found that the result of using this approach was to overstate the effect estimates in this study. Upon notification of these new findings, HEI mobilized its NMMAPS Review Panel, Chaired by Dr. Sverre Vedal of the National Jewish Medical and Research Center in Denver. The panel provided initial peer review of the work of the investigators to apply alternative analytic techniques to the data to correct for this effect. In brief the Panel found that:

 

·       most of the raw, unadjusted individual city estimates changed, with an increased number of estimates that were negative or zero;

·       the mean effect estimate shifted from .41 increase in mortality for every 10 micrograms increase in PM10 (in the original study) to .21 percent (in the revised analysis);

·       in the revised results, PM10 effect estimates are unaffected by the addition of co-pollutants such as ozone.  (see Figure 3 below)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Fig. 3 Revised NMMAPS Analysis of PM vs. Other Pollutants

 

 

The HEI Panel continues to review the work of the investigator teams from both Johns Hopkins and Harvard to recalculate key analyses in the studies and provide comprehensive HEI peer review and commentary. A final report is expected in January. In addition, a number of other studies cited in EPA’s current draft Criteria Document for Particulate Matter also use this software and may be affected in similar or different ways. To assess the nature and extent of this effect, US EPA and its Clean Air Scientific Advisory Committee (CASAC) at its July meeting set out a multi-step process to identify studies central to the NAAQS and recalculate key analyses in these studies. HEI, at the request of EPA and CASAC will play a central role in the review of these analyses.       

 

The Reanalysis of the Harvard Six Cities and American Cancer Society Studies

 

In addition to NMMAPS, and in response to requests from Congress, US EPA, industry and others, HEI convened a detailed reanalysis of the Harvard Six Cities and American Cancer Society studies.  Given full access to the entire medical and air pollution data base from the original investigators, HEI’s Expert Panel selected an entirely new team of investigators, conducted a detailed quality assurance audit of the data and replication analyses, and then implemented a large number of sensitivity

analyses to test whether some other difference between the most and least polluted cities (e.g. differences in the quality of medical care) could explain the increased mortality risk.

 

In brief, the reanalysis assured the quality of the data, replicated the original results, and tested those results against alternative risk models and analytical approaches without substantively altering the original findings of an association between sulfates (a form of particles created in the atmosphere from coal combustion and other emissions) and fine particles (PM2.5) and mortality (Table 1 below).

 

 

Analysis

PM2.5

Sulfates

Original

1.17(1.08,1.27)

1.15(1.08,1.22)

Full

1.18(1.09,1.26)

1.15(1.09,1.21)

Extended

1.18(1.09,1.26)

1.15(1.09,1.21)

 

Table 1. Relative Risk of Mortality for those living in most polluted city in ACS Study for original analysis and reanalyses (e.g .in original analysis those living in city with the highest PM2.5 had a 17% higher risk of mortality)

 

At the same time, the reanalyses extended and challenged our understanding of the original results:

 

·       the effects on mortality appeared to increase for those with less education (and likely therefore of lower socioeconomic status;

 

·       when the correlations among cities near one another were considered, the effects of fine particles remained but were diminished; and

 

·       an association between sulfur dioxide (SO2) and mortality (but not other pollutants) was observed and persisted when other variables were included.

 

In conclusion: the reanalysis identified relatively robust associations of mortality with fine particles, sulfate, and sulfur dioxide, and tested those associations in nearly every possible manner within the limitations of the data sets.

 

Key Question for the Longer Term:  Are All Particles Created Equal?

 

            To date, most analyses of the effects of particulate matter have focused on the mass of PM.  Particles are, however, a complex mixture of pollutants, and over the longer term, it will be important to understand whether all particles have similar levels of toxicity, or whether some particles, and therefore some sources, contribute higher toxicity, and should be more stringently controlled.  While there are many actions underway already to reduce overall particle levels - for example to control diesel vehicle PM emissions and nitrogen oxide emissions (a precursor of nitrates) from power plants - in the years to come, it will be especially important to develop the most cost-effective control strategies aimed at the most toxic sources, or at the most toxic components of those sources’ emissions.  This will be a critical area for new research.

 

            There are a number of components of PM that could cause toxicity.  At a multi-disciplinary NARSTO/EPA workshop in July, 1998, the following key PM characteristics and components were identified:

 

·       PM mass

·       PM particle size, surface area

·       Ultra fine PM

·       Reactive transition metals

·       Organic compounds (e.g. diesel PM)

·       Acids

·       Biogenic particles

·       Sulfates and nitrates (e.g. from SO2 and NOx)

·       Peroxides

·       Soot

·       Co-pollutants - SO2, CO, Ozone, etc.

 

Research studies are now underway at EPA, HEI, EPRI, NIEHS, and other research institutions to begin to identify the relative toxicity of some of these components.  Initial indication of the potency of some of these elements (e.g. the metals attached to PM) are beginning to emerge.  In some cases, studies have looked at effects of emissions from power plants.  Some studies have not found effects from exposure to sulfates; however other studies, including the reanalysis and toxicology studies, have found effects of sulfates and other potential emissions such as fly ash.  Ultimately, identifying whether one or more of these components is especially toxic will require a systematic, multidisciplinary effort.

 

To address these questions, the HEI Review Committee, in April 2002, issued the second in its HEI Perspectives series, entitled Understanding the Health Effects of Components of the Particulate Matter Mix: Progress and Next Steps.  This review, which I have provided to your staff and is available on the HEI web site at http://www.healtheffects.org/Pubs/Perspectives-2.pdf, summarizes recent HEI and other research on the effects of different components of the mix.  It also lays out a systematic effort necessary to achieve a better understanding, including:

 

·       Parallel epidemiology studies in carefully selected, representative cities throughout the US, with detailed daily characterization of the particle mixture;

 

·       Companion toxicology studies using concentrated ambient particles, source-specific particles, and model particles to test the full range of health endpoints and mechanisms for each particle type.

 

Many elements of such an effort are currently underway in the EPA research program and other efforts.  A more systematic approach will require substantial resources dedicated over the next decade.  However, the result of such an effort could be a better-focused and more cost-effective path to improved public health.

 

Conclusion: Progress and Next Steps

 

            In conclusion, we have made much progress in the last five years, especially in testing the validity of the short and long term epidemiology studies which served as the primary basis for the setting of the 1997 NAAQS for particulate matter.  We have tested a number of possible confounding factors, explored whether errors in measuring exposure might explain the relationships between PM and health, and analyzed whether different statistical techniques could change the results.  In reviewing the latest evidence, the HEI Review Committee concluded “epidemiologic evidence of PM’s effects on mortality and morbidity persists even when alternative explanations have been largely addressed”.  Based on this evidence, a number of initial control measures are now moving forward.

 

            At the same time, important new questions have arisen. In the near term it is necessary to complete the reassessment of NMMAPS and identify, reassess and provide peer review for other key studies that use GAM. Over the longer term, other important questions also remain, especially concerning the comparative toxicity of different components and sources of the PM mixture.  Much research is underway to understand this important question and to inform and target future strategies for control of those emissions that may be most responsible.  Only through a systematic effort to test and compare the toxicity of these diverse particles will we be able to have the best chance of answering these key questions for the future.

 

            Thank you again for the opportunity to present this testimony.  I would be pleased to answer any questions you might have.