United States Science Advisory EPA-SAB-IHEC-99-007
Environmental Board (1400) December1998
Protection Agency Washington DC ivww.epa.gov/sab
c/EPA AN SAB REPORT: REVIEW OF
DISPROPORTIONATE IMPACT
METHODOLOGIES
A REVIEW BY THE INTEGRATED
HUMAN EXPOSURE COMMITTEE
(IHEC) OF THE SCIENCE ADVISORY
BOARD (SAB)
-------
Decembers, 1998
EPA-SAB-IHEC-99-007
The Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, DC 20460
Subject: Science Advisory Board's Review of the Office of Civil Rights (OCR)
Methodologies for Performing Disproportionate Impact Analyses
Dear Ms. Browner:
The Office of Civil Rights (OCR) is charged with responding to complaints filed under
Title VI of the Civil Rights Act of 1964 that allege discriminatory effects from the issuance of
pollution control permits by states or other governmental bodies that receive financial assistance
from EPA. To address such complaints, OCR has collaborated with several other EPA offices on
the development of two methodologies for performing disproportionate impact analyses. The first
methodology, the Relative Burden Analysis (RBA) [of which there are two versions: the Basic
RBA (BRBA) and the Enhanced RBA (ERBA)] was developed by the Office of Research and
Development. The second methodology, the Cumulative Outdoor Air Toxics Concentration
Exposure Methodology (COATCEM), was developed jointly by the Office of Policy and the
Office of Air and Radiation. It was these two methodologies that the Disproportionate Analysis
Methodologies Panel of the Integrated Human Exposure Committee (IHEC) of the Science
Advisory Board (SAB) reviewed in public session on September 3-4, 1998, addressing 14
questions posed by the Agency in the Charge to the Committee.
The IHEC commends the OCR for its efforts to explore utilization of modeling to estimate
chemical-specific impacts on geographic-specific populations at the census block level as a tool to
assist in Title VI decisions. The IHEC recognizes that the 180-day requirement for responding to
Title VI complaints places significant restraints on the practical ability to use the most
scientifically rigorous methods to address the ultimate complex circumstance — multi-chemical,
multi-source, multi-pathway environments - present in communities. Thus, a truly rigorous risk
assessment approach may not be feasible. Emissions data availability limitations, coupled with the
simplifying assumptions present in the currently available models, necessarily relegate the model
outputs to a limited, secondary role in support of Title VI decisions. The utility of these models is
quite different from the comprehensive risk assessments upon which regulatory decisions
elsewhere in the EPA are based. The outputs from the Relative Burden Analysis (RBA) should be
considered supporting indicators and not primary information upon which to base decisions.
-------
COATCEM methodology is an advancement over the RBA and shows great promise. However,
its application to a Title VI analysis has not been fully developed.
Given the important limitations of both methodologies, the Committee provided specific
guidance for conducting disproportionate impact analyses at the present time. It recommended
that disproportionate impact analyses be conducted in a step-wise manner. The first step should
be to examine site-specific data (including both actual measured emissions and TRI reports) to
determine whether the air pathway and chronic exposure are appropriate surrogates for
community burden calculations and then determine which methodology to select for the initial
analysis. It is appropriate for screening purposes to start with simpler methodologies to identify
the chemicals or classes of chemicals upon which to focus in subsequent analyses. Hence, a
modified form of BRBA which includes a simple distance function might be the first methodology
applied. If more detailed analyses are desirable, ERBA could be used to identify burdens based
upon air dispersion modeling of chemicals from the facilities. To assess whether the risks
associated with the derived burdens are de minimis, risk assessments of the chemicals/classes of
chemicals that are the drivers for the burden values could then be performed.
The report contains a series of important Findings. In addition to the recommendation to
perform disproportionate impact analyses in a step-wise manner, the Committee developed eight
other Recommendations to the Agency, as follows:
a) The relevance of relative burden ratio (RBR) determinations is questionable when
all populations exhibit either de minimis risk or risks demanding action. In the
tiered approach suggested, the Agency should consider the sequence in which
these analyses are performed (i.e., determining the potential "risk" to all
populations before estimating disproportionate impact).
b) The Agency should use the term "toxicity-weighted exposure" instead of "burden."
Further, EPA needs to develop a more specific and consistent definition of
"toxicity-weighted exposure" for the RBA methodology. The policy decision as to
whether the toxicity-weighted exposure is considered adverse should be risk-
based. The current RBA burden analyses will not provide useful information to
accommodate this decision.
c) COATCEM has significant potential and should be developed further. In
calibrating and validating this methodology, the Agency should consider two test
cases - one in which mobile and area sources are not considered significant
contributors to overall exposure (e.g., rural case) and one in which area and mobile
sources are important contributors to overall exposures (e.g., urban case).
d) When evaluating potential risks of emitted chemicals for the purpose of
determining whether or not the cumulative risks are de minimis., cancer risks and
non-cancer health effects should be evaluated separately.
-------
e) Given the large number of uncertainties in both the KB A and COATCEM
methodologies, it is important to perform uncertainty and sensitivity analyses of
each methodology. In the policy decision process, the power of these
methodologies to detect differences in toxicity-weighted exposures should be
compared with an a priori identification of the level of difference that is meaningful
in a regulatory or legal sense. For example, one could define as significant
situations in which the calculated risks are above de minimis levels, and the
toxicity-weighted exposure ratios are larger than the uncertainty factors in the
specific method.
f) An important next step in the validation procedure for both the ERBA and
COATCEM methodologies will be to collect ambient monitoring data at sites
included in an analysis of disproportionate impacts using these methodologies, in
order to compare measured concentrations of chemicals with model-estimated
concentrations.
g) Considerations of acute exposure impacts, including irritation and odor, should be
included, to the extent that methodologies are available to address such effects.
h) In interacting with the residents of the communities being studied, it is critical to
maintain good communications and to convey information on the studies in an
understandable and complete manner, making sure that the uses and limitations of
the methodologies are adequately addressed. Special care should be taken to
explain the difference between "toxicity-weighted exposure" and "risk".
We appreciate the opportunity to review the two methodologies and look forward to your
response to the recommendations we have made.
Sincerely,
/signed/ /signed/
Dr. Joan M. Daisey, Chair Dr. Henry A. Anderson, Chair
Science Advisory Board Integrated Human Exposure Committee
Science Advisory Board
/signed/
Dr. Thomas E. McKone, Co-Chair
Integrated Human Exposure Committee
Science Advisory Board
-------
NOTICE
This report has been written as part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the Administrator
and other officials of the Environmental Protection Agency. The Board is structured to provide
balanced, expert assessment of scientific matters related to problems facing the Agency. This
report has not been reviewed for approval by the Agency and, hence, the contents of this report
do not necessarily represent the views and policies of the Environmental Protection Agency, nor
of other agencies in the Executive Branch of the Federal government, nor does mention of trade
names or commercial products constitute a recommendation for use.
-------
ABSTRACT
The Disproportionate Impact Analysis Methodologies Panel of the USEPA Science
Advisory Board (SAB) Integrated Human Exposure Committee (IHEC) met in public session on
September 3-4, 1998 to review the Agency's proposed methods for calculating disproportionate
impacts of air emissions on surrounding populations of different race, color, or national origin.
The Agency is developing these methodologies in connection with Title VI of the Civil Rights Act
of 1964 (as amended).
The Panel reviewed two methods: the first was the Relative Burden Analysis (RBA) [in
two versions: the Basic RBA (BRBA) and the Enhanced RBA (ERBA)]; the second was the
Cumulative Outdoor Air Toxics Concentration Exposure Methodology (COATCEM). The
former has been applied on a trial basis to a site in Louisiana; the latter has not yet been applied to
a particular site.
The Panel commends the Agency for these initial efforts in trying to determine analytically
disproportionate impacts. However, each of the two methods has its limitations in terms of
accuracy, uncertainty, data availability, resources, and level of development. The report contains
a number of findings, nine specific recommendations, including suggested guidance for moving
forward in this important area, and detailed responses to the 14 Charge questions, .
KEYWORDS: Title VI, environmental justice, RBA, COATCEM, disproportionate
impact, Shintech
-------
U.S. Environmental Protection Agency
Science Advisory Board
Integrated Human Exposure Committee
Disproportionate Impact Analysis Methodology Review Panel
Chair
Dr. Henry A. Anderson, Chief Medical Officer, Wisconsin Bureau of Public Health, Madison,
WI
Co-Chair
Dr. Thomas E. McKone, Staff Scientist/Adjunct Professor, School of Public Health, University
of California, Berkeley, CA
Members
Dr. Paul Bailey, Toxicology Consultant, Mobil Business Resources Corporation, Product
Stewardship & Toxicology, Paulsboro, NJ
Dr. Michael D. Lebowitz, Professor of Medicine and Epidemiology, Arizona Prevention Center,
University of Arizona College of Medicine, Tucson, AZ
Dr. Kai-Shen Liu, Epidemiologist, California Department of Health Services, Berkeley, CA
Dr. Maria Morandi, Assistant Professor, University of Texas Health Science Center at Houston,
School of Public Health, Houston, TX
Dr. Jerome O. Nriagu, Professor, The University of Michigan, School of Public Health,
Department of Environmental and Industrial Health, Ann Arbor, MI
Dr. Barbara Petersen,President, Novigen Sciences, Inc., Washington, DC
Mr. Ron White, Deputy Director, National Programs and Director, Environmental Health,
American Lung Association, Washington, DC
SAB Liaison Member
Dr. Ishwar Murarka, Chief Scientist and President, Ish, Inc., Cupertino, CA [Past Chair, SAB
Environmental Engineering Committee; Member, SAB Research Strategies Advisory
Committee]
Consultants
Dr. Annette Guiseppi-Elie, Senior Staff Environmental Scientist, Exxon Biomedical Sciences,
Inc., East Millstone, NJ
in
-------
Dr. Rogene Henderson, Senior Scientist, Lovelace Respiratory Research Institute, Albuquerque,
NM
Dr. Michael Jayjock, Senior Research Fellow, Rohm and Haas Co., Research Laboratories,
Spring House, PA
Dr. Maurice Knuckles, Assistant Professor and Director, Division of Environmental Health,
Department of Family and Preventative Medicine, Meharry Medical College, Nashville,
TN
Dr. Marinelle Payton, Instructor in Medicine and Occupational Medicine, Environmental and
Occupational Medicine, Harvard Medical School, Harvard School of Public Health,
Boston, MA
Science Advisory Board Staff
Ms. Roslyn Edson, Designated Federal Officer, U.S. Environmental Protection Agency, Science
Advisory Board (1400), 401 M Street, SW, Washington, DC 20460
Dr. Dorothy Canter, Co-Designated Federal Officer (working on this review) and Special
Assistant to the Assistant Administrator, Office of Solid Waste and Emergency Response,
401 M Street, SW, Washington, DC 20460
Ms. Wanda Fields, Management Assistant, U.S. Environmental Protection Agency, Science
Advisory Board (1400), 401 M Street, S.W., Washington, DC 20460
IV
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 5
2.1 Background 5
2.2 Charge 7
2.3 SAB Review Process 8
3. OVERALL FINDINGS AND RECOMMENDATIONS 9
3.1 Introduction 9
3.2 Findings 14
3.2.1 Definition of burden 14
3.2.2 Projecting de minimis risks from relative burden ratios 14
3.2.3 Strengths and weaknesses of individual methodologies 14
3.2.4 Limitations common to both methodologies 16
3.3 Specific IHEC Recommendations 17
4. RESPONSE TO CHARGE QUESTIONS 20
4.1 Question 1 20
4.1.1 Findings and Recommendations 20
4.2 Question 2 21
4.2.1 Findings and Recommendations 21
4.3 Question 3 21
4.3.1 Findings and Recommendations 22
4.4 Question 4 22
4.4.1 Findings and Recommendations 22
4.5 Question 5 23
4.5.1 Findings and Recommendations 23
4.6 Question 6 23
4.6.1 Findings and Recommendations 24
4.7 Question 7 24
4.7.1 Findings and Recommendations 24
4.8 Question 8 25
4.8.1 Findings and Recommendations 25
4.9 Question 9 26
4.9.1 Findings and Recommendations 26
4.10 Question 10 26
4.10.1 Findings and Recommendations 26
4.11 Question 11 27
4.11.1 Findings and Recommendations 27
-------
4.12 Question 12 28
4.12.1 Findings and Recommendations 28
4.13 Question 13 29
4.13.1 Findings and Recommendations 29
4.14 Question 14 30
4.14.1 Findings and Recommendations 30
VI
-------
1. EXECUTIVE SUMMARY
The Office of Civil Rights (OCR) is charged with responding to complaints filed under
Title VI of the Civil Rights Act of 1964 that allege discriminatory effects from the issuance of
pollution control permits by states or other governmental bodies that receive financial assistance
from EPA. To address such complaints, OCR has collaborated with several other EPA offices on
the development of two methodologies for performing disproportionate impact analyses. The first
methodology, entitled the Relative Burden Analysis (RBA), was developed by the Office of
Research and Development; the second methodology, entitled the Cumulative Outdoor Air Toxics
Concentration Exposure Methodology (COATCEM), was developed jointly by the Office of
Policy and the Office of Air and Radiation.
The RBA methodology estimates differential "burdens" to various populations from air
emissions from stationary point sources. It consists of two versions, the Basic RBA (BRBA) and
the Enhanced RBA (ERBA), and uses as the starting point of the analysis the annualized estimates
of emissions reported to the Toxics Release Inventory by specific facilities. It employs the
toxicity weights for individual chemicals from the Office of Pollution Prevention and Toxics Risk
Screening Environmental Indicators (RSEI) methodology to generate a "pseudo-chemical" for the
facilities of concern. The RBA uses the calculated burdens for specific populations to generate
rough indications of risk. The RBA has been applied on a trial basis to a site in Louisiana.
The COATCEM methodology is designed to estimate risks from three sources of air
emissions; namely, stationary, area and mobile sources. It uses air dispersion modeling to develop
exposure concentrations from these sources to populations at specific locations. It includes three
measures of differential impacts among populations: cancer risk, non-cancer toxicity hazard ratio,
and total benchmark exceedances. COATCEM evaluates hazardous air pollutants with known
toxicity values (i.e., cancer slope factors and reference concentrations) individually. To date,
COATCEM has not been applied to an actual case.
OCR plans to use these methodologies to generate information on substantial differences
in impact between populations from specific facilities, potential harm due to such differential
impacts, and an estimate of overall cumulative levels of risks from multiple emitting sources in the
areas of concern. Given the significant scientific and technical issues involved in developing such
methodologies and the precedent-setting nature of performing these analyses, the OCR requested
a review by the EPA Science Advisory Board (SAB) of the RBA and COATCEM methodologies.
Consequently, the SAB Integrated Human Exposures Committee (IHEC) conducted a review of
the two methodologies on September 3-4, 1998, and addressed the 14 questions posed by the
Agency in the charge to the Committee.
The IHEC commends the OCR for its efforts to explore utilization of modeling to estimate
chemical-specific impacts to geographic-specific populations. The IHEC recognizes that the 180-
day requirement for responding to Title VI complaints places significant restraints on the practical
-------
ability to use the most scientifically rigorous methods to address the extremely complex multi-
chemical, multi-source environments present in communities. Thus, a risk assessment approach
may not be feasible. Within the identified programmatic constraints (time, utilization of existing
nationwide data), the OCR has made a strong start, and IHEC encourages OCR to continue its
efforts.
The IHEC concluded that each of the methodologies has specific strengths and
weaknesses. The RBA methodology is simple, transparent, easy to use and understand, and can
be easily updated. It has been utilized in one instance and can be performed in the 180-day time
limit for responding to Title VI complaints. The ERBA is an improvement over the BRBA
because it uses an air dispersion model to estimate concentrations of the pseudo-chemical at
specific distances from the various facilities.
The Committee concluded that the COATCEM methodology has significant potential for
future use. It is more comprehensive and more resource intensive than the RBA methodology. It
evaluates emissions from point, area and mobile sources, and therefore it should represent a more
realistic evaluation of cumulative exposures, particularly in urban areas. It estimates air
concentrations for individual chemicals and evaluates separately cumulative cancer risks and non-
cancer health effects of the chemicals. Given COATCEM's complexity, it remains to be seen
whether it could be utilized within the 180-day limit established for addressing Title VI
complaints. Although the most promising of the methodologies reviewed, COATCEM needs
further development if it is to be used by OCR.
Importantly, the Committee concluded that the definition of "burden" in both versions of
the RBA is not sufficiently precise. It is critical that an understandable and meaningful definition
for burden be developed. The term "toxicity-weighted exposure" should be considered as a
replacement term for burden, since burden can be confused with concepts of internal dosimetry
such as "body burden," and to many people it implies "risk."
The Committee concluded that a fundamental problem in the RBA methodology was an
inability to extrapolate from the relative burden ratio (RBR), or toxicity-weighted exposure of one
population in a given area divided by the toxicity-weighted exposure of a second population in the
same area, to potential harm or risk. The issue ofde minimis risk cannot be addressed without
risk-based analyses. Unless burden is defined to include evaluations of potential risks, the concept
ofde minimis risk based upon the RBR is not meaningful. Hence, as presently defined, the RBR
cannot be used to identify de minimis risks.
Given the important limitations of both methodologies, the Committee provided specific
guidance for conducting disproportionate impact analyses at the present time. It recommended
that disproportionate impact analyses be conducted in a step-wise manner. The first step should
be to examine site-specific data (including both actual measured emissions and TRI reports) to
determine whether the air pathway and chronic exposure are appropriate surrogates for
community burden calculations and then determine which methodology to select for the initial
-------
analysis. It is appropriate for screening purposes to start with simpler methodologies to identify
the chemicals or classes of chemicals upon which to focus in subsequent analyses. Hence, a
modified form of BRBA which includes a simple distance function might be the first methodology
applied. If more detailed analyses are desirable, ERBA could be used to identify toxicity-
weighted exposures based upon air dispersion modeling of chemicals from the facilities. To
assess whether the risks associated with the derived burdens are de minimis, risk assessments of
the chemicals/classes of chemicals that are the drivers for the toxicity-weighted exposure values
could then be performed.
The Committee developed eight other recommendations to the Agency, as follows:
a) The relevance of RBR determinations is questionable when all populations exhibit
either de minimis risk or risks demanding action. In the tiered approach
suggested, the Agency should consider the sequence in which these analyses are
performed (i.e., determining the potential "risk" to all populations before
estimating disproportionate impact).
b) The Agency should use the term "toxicity-weighted exposure" instead of "burden."
Further, EPA needs to develop a more specific and consistent definition of
"toxicity-weighted exposure" for the RBA methodology. The policy decision as to
whether the toxicity-weighted exposure is considered adverse should be risk-
based. The current RBA burden analyses will not provide useful information to
accommodate this decision.
c) COATCEM has significant potential and should be developed further. In
calibrating and validating this methodology, the Agency should consider two test
cases: one in which mobile and area sources are not considered significant
contributors to overall exposure (e.g., rural case), and one in which area and
mobile sources are important contributors to overall exposures (e.g., urban case).
d) When evaluating potential risks of emitted chemicals for the purpose of
determining whether or not the cumulative risks are de minimis, cancer risks and
non-cancer health effects should be evaluated separately.
e) Given the large number of uncertainties in both the RBA and COATCEM
methodologies, it is important to perform uncertainty and sensitivity analyses of
each methodology. In the policy decision process, the power of these
methodologies to detect differences in toxicity-weighted exposures should be
compared with an a priori identification of the level of difference that is
meaningful in a regulatory or legal sense. For example, one could define as
significant situations in which the calculated risks are above de minimis levels, and
the toxicity-weighted exposure ratios are larger than the uncertainty factors in the
specific method.
-------
f) An important next step in the validation procedure for both the ERBA and
COATCEM methodologies will be to collect ambient monitoring data at sites
included in an analysis of disproportionate impacts using these methodologies, in
order to compare measured concentrations of chemicals with model estimated
concentrations.
g) Considerations of acute exposure impacts, including irritation and odor, should be
included, to the extent that methodologies are available to address acute effects.
h) In interacting with the residents of the communities being studied, it is critical to
maintain good communications and to convey information on the studies in an
understandable and complete manner, making sure that the uses and limitations of
the methodologies are adequately addressed. Special care should be taken to
explain the difference between "toxicity-weighted exposure" and "risk".
-------
2. INTRODUCTION
2.1 Background
The background for the project was effectively provided in the documentation that
provided the Charge to the SAB; specifically, "Questions for the Science Advisory Board on the
Title VI Relative Burden Analyses and Cumulative Outdoor Air Toxics Concentrations and
Exposure Methodology". That background is quoted as follows:
"Title VI of the Civil Rights Act of 1964 as amended (Title VI) prohibits recipients of
Federal financial assistance (such as state environmental departments) from discriminating on the
basis of race, color, or national origin in their programs or activities. Title VI requires Federal
agencies that provide financial assistance, including the Environmental Protection Agency (EPA),
to ensure that recipients of Federal financial assistance do not discriminate on the basis of race,
color, or national origin. Discrimination can result from policies and practices that are neutral on
their face, but have the effect of discriminating. In addition to prohibiting intentional
discrimination, EPA's Title VI regulations (40 C.F.R. Part 7) prohibit facially-neutral policies or
practices that result in a disparate adverse impact, unless it is shown that they are justified and that
there is no less discriminatory alternative.
"Since 1993, EPA has received an increasing number of Title VI complaints that allege
violations of EPA's discriminatory effects regulations from the issuance of pollution control
permits by EPA recipients. EPA's Office of Civil Rights (OCR) currently has 15 open
investigations, as well as 12 awaiting processing, of complaints which allege discriminatory effects
of permitting decisions. On February 5, 1998, EPA released its Interim Guidance for
Investigating Title VI Administrative Complaints Challenging Permits (Title VI Interim Guidance)
which is an internal guidance document that describes how OCR will process these types of
complaints. Generally, Title VI complaints are subject to the following process: 1) initial finding
of disparate impact, 2) presentation of rebuttal evidence, 3) identification of legitimate
justifications, and 4) identification of less discriminatory alternatives. EPA is currently focused on
developing sound methods for establishing the first element of this process - the initial finding of
disparate impact. OCR is interested in developing tools that can be used repeatedly with some
ease so that ultimately they may be used by recipients and others as a means of identifying
potential Title VI disparate impacts in the context of individual permit decisions.
"The investigation and resolution of Title VI complaints regarding potential discriminatory
effects of environmental permitting decisions is precedent-setting and may have implications on
how recipient agencies implement their environmental permitting programs to ensure no person is
discriminated against based on race, color, or national origin. As a result, the issue of how to
measure disparate adverse impacts from permitted facilities has had high visibility in the news
media, as well as generated interest and debate within the industrial, state/local government, and
environmental justice communities.
-------
"A. Context for Assessing Title VI Violations
"EPA's Title VI discriminatory effects regulations are violated if facially-neutral policies or
practices result in a disparate adverse impact, unless it is shown that they are justified and that
there is no less discriminatory alternative.
"Most of the Title VI administrative complaints filed with OCR under the discriminatory
effects standard have involved the issuance of permits for facilities ordinarily considered to be
undesirable, including hazardous and municipal waste landfills and incinerators. Some complaints
have involved permits for product manufacturing facilities. These activities, requiring
environmental pollution control permits, may have both positive impacts (e.g., economic
development, necessary services, employment opportunities) and negative impacts (e.g., pollutant
emissions and discharges, noise, odors, accidents) upon the surrounding areas and nearby
populations. OCR recognizes that positive impacts can and often do result from the operation of
such facilities and that such positive impacts can be considered in the justification phase of the
Title VI analysis. However, the particular methods and analytic tools discussed herein are solely
to measure and analyze specific negative (undesirable) impacts on the surrounding community.
"The Title VI Interim Guidance states that investigations will include an evaluation of
permitted facilities 'which together present cumulative burden or which reflect a pattern of
disparate impact.' OCR anticipates that many of these Title VI investigations will involve
evaluating aggregated, or cumulative, impacts on population subgroups defined by race, color, or
national origin. The range of permitted facilities within the scope of these investigations is
potentially broad. To determine whether the operation of permitted facilities poses, as an initial
matter, a disparate adverse impact based on race, color, or national origin within surrounding
populations, OCR needs a method of measuring or estimating the difference in the impact
between population subgroups. First, OCR needs to determine whether population subgroups
defined by race, color, or national origin are experiencing a substantial difference in impact (i.e.,
disparity of impact). Second, OCR will determine whether the impact experienced by such
population subgroups are at or above a level of concern (as opposed to de minimis) so as to be
considered adverse (i.e., the impact is harmful). The determination of whether an identified
disparity is substantial and whether the impact is at or above a level of concern are policy issues.
However, the methods and tools used to identify the disparity and to measure the impact used in
these policy decisions are scientific in nature and should be subjected to peer review.
"B. Approaches to Analyzing Impacts of Permitted Toxic Air Emissions
"In developing the methodologies described below, EPA considered the number and types
of facilities potentially involved in these complaints, regulatory time constraints for resolving Title
VI administrative complaints (i.e., 180 days), the type of data likely to be made available in such a
time frame, as well as resource implications for OCR and recipient agencies. These
considerations limit the nature and level of analysis that can be performed and preclude the
conduct of detailed risk assessments.
-------
"Impacts from permitted industrial activity tend to be distributed in certain geographic
patterns relative to the facilities themselves. Permitted air emissions (stack and fugitive emissions)
impact surrounding populations via local wind patterns and inversely with distance from the
facility. To determine whether permitted "toxic" air pollution emissions have a disparate adverse
impact on population subgroups defined by race, color, or national origin, EPA has developed and
applied relative burden analyses [KB A] in an investigation of a Title VI administrative complaint
that alleges the Louisiana Department of Environmental Quality's issuance of a Title V permit
under the Clean Air Act to Shintech, Inc. for a proposed polyvinyl chloride facility will result in
discriminatory effects (i.e., an unjustified disparate adverse impact) on African Americans. The
analyses assist EPA in determining the average burden from these combined toxicity-weighted
Toxics Release Inventory (TRI) air emissions upon one population subgroup compared to another
and to obtain rough estimates of cumulative risk in areas proximate to the permitted facilities.
"In addition, EPA proposes to use, in the Louisiana complaint investigation, the
Cumulative Outdoor Air Toxics Concentration and Exposure Methodology (COATCEM). This
methodology is similar to the Office of Policy (formerly, Office of Policy, Planning and
Evaluation) Cumulative Exposure Project (CEP) methodology that was reviewed by the Science
Advisory Board (SAB) in 1996 (SAB, 1996). Like the CEP, the proposed COATCEM approach
uses a Gaussian dispersion model to analyze outdoor air concentrations of hazardous air
pollutants (HAPs) over large areas from point sources, mobile sources, and area sources in
combination. Emissions estimates to be used in the model include data for approximately 115
individual HAPs from the 1996 Louisiana Toxics Emissions Data Inventory (TEDI) and, for
mobile sources and area sources, from EPA's National Toxics Inventory. Modeled concentrations
for individual HAPs will be compared with health benchmark concentrations for the HAPs to
develop several multiple-pollutant metrics. The technical approach for COATCEM will be similar
to that for CEP, but will be done on a census block level of resolution, while the CEP was
conducted at the census tract level.
"OCR plans to use both methodologies, as appropriate, to generate information about
disparity of impacts, potential harm of toxic air emissions from a variety of facilities, and estimates
of the overall cumulative background levels of risk from sources in the area of concern. This
information will supplement other evidence regarding other impacts gathered during the course of
an investigation and will be useful to consider in making the policy-level decisions that must be
made in Title VI cases."
2.2 Charge
The Charge to the SAB consisted of 14 specific questions that are detailed in Section 4.0
(Responses to Charge Questions) below. The questions addressed specific aspects of the relative
burden analyses (RBA) and Cumulative Outdoor Air Toxics Concentration and Exposure
Methodology (COATCEM) methodologies. The intent was to focus the SAB review on
particular technical aspects of the methodologies. In preparing this report, the SAB was guided
by the Charge questions, which are answered in detail in Section 4.0, but was not limited by the
-------
questions. In Section 3.0 (Overall Findings and Recommendations) below, the SAB has offered
advice on more general matters beyond the Charge and summarized its review in a set of overall
Recommendations.
2.3 SAB Review Process
At the request of the EPA Office of Civil Rights (OCR), the Disproportionate Impact
Analysis Methodology Panel of the Integrated Human Exposure Committee (IHEC) of the
Science Advisory Board (SAB) met on September 3-4, 1998, to review two methodologies
developed to analyze potential disproportionate impacts to populations in specified areas from air
emissions from various sources. The Committee evaluated the document on the methodologies,
as well as submitted supporting data, and considered written comments submitted by interested
parties and oral comments delivered by the public at the meeting.
The Charge and this report are limited to an evaluation of scientific merits of two data
modeling methodologies. At the time of the review, the Agency had not yet developed a policy
on what model outcomes would constitute substantial differences in impact. A thorough
understanding of the science and its limitations should assist the Agency in developing its
interpretive policy.
The IHEC draft report was made available to the Agency and the public prior to the
October 28-29, 1998 SAB Executive Committee (EC) meeting. The EC reviewed the report for
matters of fact, clarity, and completeness in addressing the Charge. Following incorporation of
EC comments, the report is being sent to the Administrator.
-------
3. OVERALL FINDINGS AND RECOMMENDATIONS
3.1 Introduction
The IHEC commends the OCR for its efforts to explore utilization of modeling to estimate
chemical-specific impacts on geographic-specific populations at the census block level as a tool to
assist in Title VI decisions. The IHEC recognizes that the 180-day requirement for responding to
Title VI complaints places significant restraints on the practical ability to use the most
scientifically rigorous methods to address the ultimate complex circumstances - multi-chemical,
multi-source, multi-pathway environments — present in communities. Thus, a "rigorous" risk
assessment approach may not be feasible. Emissions data availability limitations coupled with the
simplifying assumptions present in the currently available models necessarily relegate the model
outputs to a limited, secondary role in support of Title VI decisions. The utility of these models is
quite different from the comprehensive risk assessments upon which regulatory decisions
elsewhere in the EPA are based. The outputs from the Relative Burden Analysis (RB A) should be
considered supporting indicators and not primary information upon which to base decisions.
COATCEM methodology is an advancement over the RBA and shows great promise. However
its application to a Title VI analysis has not been fully developed.
Within the identified programmatic constraints (time, utilization of existing nationwide
data), the OCR has made a strong good start, and IHEC encourages OCR to not only continue its
efforts to further develop modeling tools applicable to their needs but also to concomitantly
carefully define how such model results will contribute to Title VI decisions.
The two methodologies reviewed by the IHEC are the Relative Burden Analysis (RBA),
developed by the Office of Research and Development (ORD), and the Cumulative Outdoor Air
Toxics Concentration and Exposure Methodology (COATCEM), developed jointly by the Office
of Air and Radiation and the Office of Policy. The RBA has contains two versions, the Basic
Relative Burden Analysis (BRBA) and the Enhanced Relative Burden Analysis (ERBA). Certain
technical flaws have been identified that need to be addressed before either of the proposed
methods would be considered viable for immediate use.
To address potential disproportionate impacts among populations, both variations of the
RBA methodology estimate the "burden" to specified populations within the vicinity of a group of
permitted facilities. In RBA the burden is the measure of intensity. To calculate burden, both
BRBA and ERBA convert chemicals known to be emitted from the areas of concern into a
pseudo-chemical using relative toxicity weighting factors. In creating the pseudo-chemical,
cancer risks and non-cancer toxicities of individual chemicals are combined using a defined
conversion factor.
The COATCEM methodology calculates three different metrics of the modeled air toxics
to deal with differential impacts among populations; they are cancer risk, non-cancer toxicity
-------
hazard ratio, and total benchmark exceedances. COATCEM evaluates hazardous air pollutants
with known toxicity values (i.e., cancer slope factors and reference concentrations) individually.
Table 1 summarizes relevant information on the two methodologies, including BRBA and
ERBA.
10
-------
TABLE 1: COMPARISON OF DISPROPORTIONATE IMPACT ANALYSIS METHODOLOGIES
ATTRIBUTE
Medium of Releases
Sources Evaluated
Point Sources Evaluated
Source of Emissions Data*
Type of Emissions Data
Exposure Pathways
Evaluated
Air Dispersion Modeling
Ambient Distribution of
Emissions from Each Facility
Nature of Dispersion of
Emitted Chemicals
RBA
BASIC (BRBA)
Air
Point Sources
Stack, Fugitive
1995 TRI Data; some 1996
TEDI** Data
Mainly Estimated/Some
Monitored
Inhalation
None
Uniformly Distributed in
Circle of Specified Radius
with Center at Facility
Not Applicable
ENHANCED (ERBA)
Air
Point Sources
Stack, Fugitive
1995 TRI Data; some 1996
TEDI Data
Mainly Estimated/Some
Monitored
Inhalation
ISC-LT
Function of Air Dispersion
Modeling
All assumed to disperse in
the same way***
COATCEM
Air
Point, Area and Mobile
Sources
Stack
1996 TEDI Data,
supplemented by TRI Data;
National Toxics Inventory
Estimated
Inhalation
ISC-ST
Function of Air Dispersion
Modeling
All assumed to disperse in
the same way***
11
-------
ATTRIBUTE
Consideration of
Deposition/Re-Emission
Comparison of Calculated
Chemical Concentrations
with Monitored
Concentrations
Chemicals Evaluated
Total Number of Chemicals
that could be Evaluated
Approach to Chemical
Toxicity
Evaluation of Chronic
Effects
Combination of Cancer
Risks and Non-Cancer
Health Effects
Evaluation of Acute Health
Effects
RBA
BASIC (BRBA)
No
No
TRI Chemicals
>650
Risk Screening
Environmental Indicators
toxicity weights of chemicals
to develop pseudo-chemical
Yes
Yes
No
ENHANCED (ERBA)
No
No
TRI Chemicals
>650
Risk Screening
Environmental Indicators
toxicity weights of chemicals
to develop pseudo-chemical
Yes
Yes
No
COATCEM
No
Some
HAPs, chemicals identified
as important in RBA process
115 (HAPs with toxicity
benchmarks)
Use of cancer and non-
cancer toxicity benchmarks
for chemicals with sufficient
toxicity data
Yes
No
No
12
-------
ATTRIBUTE
Use of GIS
Measure(s) of
Disproportionate Impacts to
Populations in Given Area
Application of Methodology
at Specified Location
Uncertainty Analysis Done
Sensitivity Analysis Done
Time Frame for Performing
Analysis
RBA
BASIC (BRBA)
Source Locations, 1990
Census Data
Relative Burden Ratio
Yes
No
No
<180 days
ENHANCED (ERBA)
Source Locations, 1990
Census Data
Relative Burden Ratio
Yes
No
No
<180 days
COATCEM
Source Locations, 1990
Census Data
Modeled concentrations
compared to (1) Cancer
Benchmark Concentrations
(2) Non-Cancer Benchmark
Concentrations; Total
Number of Benchmark
Exceedances
No
No
No
May be significantly longer
than 180 days
* Emissions data listed in the table apply only to Louisiana case study. For other sites, different data sets should be used.
** TEDI is Louisiana's Toxic Emissions Data Inventory system. In 1996 226 facilities reported to TEDI.
*** Assumption used for Louisiana case study
13
-------
3.2 Findings
3.2.1 Definition of burden
The Committee concluded that the definition of burden in both versions of the KB A is not
sufficiently precise. In fact, burden was defined differently in the background document and in the
presentation on the KB A methodology at the September 3-4, 1998, IHEC meeting. It is critical
that an understandable and meaningful definition for burden be developed. In addition, the term
"toxicity-weighted exposure" should be considered as a replacement term for burden, since
burden can be confused with concepts of internal dosimetry such as "body burden" or "lung
burden" and often is interpreted as a risk estimate.
3.2.2 Projecting de minimis risks from relative burden ratios
In the KB A methodology the relative burden ratio (KBR) of two populations in a given
area is derived by dividing the average burden of one population in the area (e.g., African
Americans) by the average burden of the second population (e.g., non-African Americans). The
Committee concluded that there is a fundamental problem in the RBA methodology with
extrapolating from the RBR to potential harm or risk. The issue ofde minimis risk cannot be
addressed without risk based analyses. Unless burden is defined to include evaluations of
potential risks, the concept ofde minimis risk based upon the RBR is not meaningful. A possible
way to derive risks from burdens would be to determine the set of chemicals which drive the
burden values and then perform a risk assessment of that subset of chemicals. Barring that, it is
necessary to keep risk and burden separate. Hence, as presently defined, the RBR cannot be used
to identify de minimis risks.
From a decision policy perspective much remains to be done. It is not clear how the
proposed analyses will assist interpretation of a particular RBR. To inform policy, statistical
analytic methods will need to be developed to determine whether an RBR of 1.1, 2.0, or 5.0 is
significantly different from an RBR of 1.0. Will RBRs only be calculated when risks exceed de
minimis^ What consideration will be given to RBRs which are less than 1.0? Could they be
indications of reverse disproportionate impacts? In cases in which calculated risks are available
and indicate that risks for all of the populations are below de minimis levels, the relevance of the
RBRs, regardless of the magnitude of the values, may be questionable and, at worst, misleading to
the public.
3.2.3 Strengths and weaknesses of individual methodologies
To be useful in the time frame allowed, these models all begin with comprehensive
simplifying assumptions (such as all emissions are additive, health outcomes of concern can be
represented by TRI annual average data, all chemicals disperse in a similar manner, etc.) The
relevance of these assumptions to a specific geographic location's emission impacts must be
carefully evaluated before modeling begins, to assure that the broad uncertainties introduced by
14
-------
the assumptions are manageable. Thus, if the emissions data from the TRI in a locality are flawed,
or TRI does not represent the chemicals of concern in a community, use of the models would be
inappropriate. Similarly, the model outputs will not be able to address acute, peak effects on a
community or quality-of-life issues such as odor and irritation. The Agency will need other tools
to address such issues.
a) RBA methodology (BRBA and ERBA forms^ - The Committee noted that the
BRBA is simple, transparent, easy to use and understand, and can be easily
updated. It has been utilized in one instance and can be performed in the 180-day
time limit for responding to Title VI complaints.
In its current form the BRBA has a fundamental weakness which significantly
limits its utility. It assumes that all emissions from a particular source are
distributed uniformly within circles of specified distances (e.g., 2 miles or 4 miles)
from that source and that every person living within a given circle is exposed to the
same ambient concentration from the source. It does not consider distance from
the emitting facility, wind direction and velocity, or stack height. The Committee
concluded that, without some modifications, the BRBA is too simplistic and the
output not scientifically defensible, even as a first order screening tool. To be of
value, the BRBA needs, at a minimum, to include some consideration of dispersion
(for example, an inverse distance function).
The ERBA is an improvement over the BRBA because it incorporates an air
dispersion model, the Industrial Source Complex - Long Term (ISC-LT) model, to
estimate concentrations of the pseudo-chemical at specific distances from the
various facilities. Like BRBA, it is easy to use and understand and can be easily
updated. It should be able to be utilized within the 180-day limit for responding to
Title VI complaints.
Combining all emissions into one pseudo-chemical which is then modeled in the
RBA makes the screening process manageable but leads to a number of significant
limitations and potential unintended consequences because of the type and
magnitude of the differences among chemicals (acute or chronic toxicity,
atmospheric reactivity, etc). An approach which would still allow a manageable
number of analyses might be to group chemicals into categories, depending upon
their chemical/physical properties, and thus generate a set of pseudo-chemicals
with similar properties. The role of atmospheric degradation of specific chemicals
should be considered.
Combining cancer risk and non-cancer chronic health effects adds uncertainty.
Cumulative cancer risks and cumulative non-cancer health effects could be
estimated separately and would provide more transparent results. Further,
different chemicals have different effects on different organ systems; adding
15
-------
reference concentrations for chemicals which act upon different organ systems can
be problematic.
Another limitation of the KB A methodology is that it is designed only to evaluate
air emissions from point sources; it does not take into account emissions from area
and mobile sources. The ISC-LT model is not predictive in all circumstances,
especially in urban environments with tall buildings.
b) COATCEM Methodology - The Committee concluded that the COATCEM
methodology has significant potential for future use. It is more comprehensive and
more resource intensive than the RBA methodology. It evaluates emissions from
point, area and mobile sources, so should represent a more realistic evaluation of
cumulative exposures, particularly in urban areas. It estimates air concentrations
for individual hazardous air pollutants (HAPs) and evaluates separately cumulative
cancer risks and non-cancer health effects of the chemicals.
However, COATCEM has not yet been applied to an actual case. Therefore, no
conclusions can be reached as to whether it is an improvement over the RBA.
There are presently no outcomes to compare with those of the RBA methodology,
nor are there estimated ambient concentrations of emitted chemicals to compare
with monitored concentrations of those chemicals.
The models used in COATCEM are relatively large and complex, which makes
sensitivity and uncertainty analyses difficult and time-consuming. The quality of
the emissions data for area and mobile sources will depend on the data in the 1996
National Toxics Inventory, once that data base has been completed. Given
COATCEM's complexity, it remains to be seen whether it could be utilized within
the 180-day limit established for addressing Title VI complaints.
Although the most promising of the methodologies reviewed, COATCEM needs
further development if it is to be used by OCR. The Agency priority for further
development and application of COATCEM is somewhat unclear, since the
development to date of the methodology has been overseen by two EPA offices.
3.2.4 Limitations common to both methodologies
Both the RBA and COATCEM methodologies evaluate only exposures through direct
contact with ambient air inhalation. Neither takes into account exposures from the drinking
water, soil or food chain pathways due to air emissions; exposures from surface water discharges,
underground injection or on-site land releases are also not considered.
Although both the ERBA and COATCEM methodologies use air dispersion modeling
[ERBA uses the ISC-LT model, while COATCEM uses Industrial Source Complex - Short Term
16
-------
(ISC-ST)] to estimate air concentrations in the vicinity of facilities with point source air
emissions, both use annualized emissions data. They do not take into account short term
excursions from steady state levels. As a result there could be acute exposures that may be
significantly higher than the calculated steady state levels. Neither ERBA nor COATCEM
evaluate deposition transfers to other environmental media of emitted chemicals or subsequent re-
emission of these chemicals. In addition, both methodologies assume that all emitted chemicals
disperse in the same manner. They do not take into account that some emitted chemicals are
stable while others are reactive. In addition, they do not address the fact that certain chemicals
are released in the vapor phase, while others are associated with particles.
The RBA and COATCEM methodologies only evaluate risks from chronic exposures.
Given that acute, intermittent exposures can be significantly higher than chronic exposures, and
are often the source of community complaints about decreased quality of life, failure to consider
acute risks is a serious limitation of both methodologies.
Neither of the methodologies consider length of residence of persons within the census
blocks or their activity patterns, including time spent indoors versus outdoors. Nor do they
consider population mobility such as the percentage of persons employed at other locations.
Both methodologies utilize Toxics Release Inventory (TRI) data on annual air releases
reported by specific facilities. These data are useful but have certain limitations, since they are
self-reported by facilities and are often based upon estimates rather than upon monitored
emissions. Moreover, not all types of facilities are required to report release data to TRI, nor are
all chemicals emitted from a facility required to be reported. In addition, specific chemicals at
some of the facilities may be used only at certain times during the year, so using average annual
emission rates for such chemicals is not appropriate.
3.3 Specific IHEC Recommendations
a) All disproportionate impact analyses should be conducted in a stepwise manner.
The first step should be to examine site-specific data (including both actual
measured emissions and TRI reports) to determine whether the air pathway and
chronic exposure are appropriate surrogates for community burden calculations
and then determine which methodology to select for the initial analysis. It is
appropriate for screening purposes to start with simpler methodologies to identify
the chemicals or classes of chemicals upon which to focus in subsequent analyses.
Hence, a modified form of BRBA which includes a simple distance function might
be the first methodology applied. If more detailed analyses are desirable, ERB A
could be used to identify burdens based upon air dispersion modeling of chemicals
from the facilities. To assess whether the risks associated with the derived burdens
are de minimis, risk assessments of the chemicals/classes of chemicals that are the
drivers for the burden values could then be performed.
17
-------
b) The relevance of RBR determinations is questionable when all populations exhibit
either de minimis risk or risks demanding action. In the tiered approach
suggested, the Agency should consider the sequence in which these analyses, i.e.,
determining the potential "risk" to all populations before estimating
disproportionate impact, are performed.
c) The Agency should use the term "toxicity-weighted exposure" instead of "burden."
Further, EPA needs to develop a more specific and consistent definition of
"toxicity-weighted exposure" for the RBA methodology. The policy decision as to
whether the toxicity-weighted exposure is considered adverse should be risk-
based. The current RBA burden analyses will not provide useful information to
accommodate this decision.
d) COATCEM has significant potential and should be developed further. In
calibrating and validating this methodology, the Agency should consider two test
cases - one in which mobile and area sources are not considered significant
contributors to overall exposure (e.g., rural case) and one in which area and mobile
sources are important contributors to overall exposures (e.g., urban case).
e) When evaluating potential risks of emitted chemicals for the purpose of
determining whether or not the cumulative risks are de minimis, cancer risks and
non-cancer health effects should be evaluated separately.
f) Given the large number of uncertainties in both the RBA and COATCEM
methodologies, it is important to perform uncertainty and sensitivity analyses of
each methodology. In the policy decision process, the power of these
methodologies to detect differences in "toxicity-weighted exposures" should be
compared with an a priori identification of the level of difference that is meaningful
in a regulatory or legal sense. For example, one could define as significant
situations in which the calculated risks are above the de minimis levels and the
toxicity-weighted exposure ratios are larger than the uncertainty factors in the
specific method.
g) An important next step in the validation procedure for both the ERBA and
COATCEM methodologies will be to collect ambient monitoring data at sites
included in an analysis of disproportionate impacts using these methodologies, in
order to compare measured concentrations of chemicals with model estimated
concentrations.
h) Considerations of acute exposure impacts, including irritation and odor, should be
included, to the extent that methodologies are available to address such effects.
18
-------
i) In interacting with the residents of the communities being studied, it is critical to
maintain good communications and to convey information on the studies in an
understandable and complete manner, making sure that the uses and limitations of
the methodologies are adequately addressed. Special care should be taken to
explain the difference between "toxicity-weighted exposure" and "risk".
19
-------
4. RESPONSE TO CHARGE QUESTIONS
The IHEC addressed the following 14 specific charge questions.
4.1 Question 1
The Risk Screening Environmental Indicators (RSEI) toxicity weights that Office of
Pollution Prevention and Toxics (OPPT) developed have been reviewed and commented
upon by the SAB within the past year (EPA-SAB-EEC-98-007). OPPT has addressed the
major concerns of the SAB as to having the weights ordered on a continuous scale directly
related to their toxicity values rather than in order of magnitude "bins" and avoiding
truncation of the value range. The use of these weights for the specific purpose of doing
relative burden analyses in the way outlined in the review document has not been
commented upon by the SAB. What are the strengths and weaknesses of this approach,
which applies the toxicity weights to a number of chemicals released into the air, for the
purpose of developing a burden measure?
4.1.1 Findings and Recommendations
The strengths of using the toxicity weights for individual chemicals from the Risk
Screening Environmental Indicators methodology to develop a "pseudo-chemical" in the RBA are
the simplicity of the approach and the ability to account for different degrees of toxicity of
multiple chemicals in mixtures. This makes the modeling manageable within the 180 day
requirement. In addition, data exist to assign toxicity weights to a number of chemicals of
concern.
However, the simplicity of the approach is also its greatest weakness. The model assumes
that in mixtures of environmental chemicals, the risks from the individual chemicals are additive.
This assumption is problematic when considering the universe of chemicals, since it does not take
into account possible synergism or antagonism within mixtures. However, it is a commonly used
assumption in assessing risks from mixtures because of the extremely limited data base on toxicity
of specific mixtures. Moreover, from a practical standpoint it is better than not taking cumulative
toxicity into account.
The use of TRI data in the RBA is a limitation since such data are self-reported by
facilities and are often based upon estimates rather than upon monitored emissions. Moreover,
not all types of facilities are required to report release data to TRI, nor are all chemicals emitted
from a facility reported to the TRI data base. In addition, specific chemicals at some of the
facilities may be used only at certain times during the year, so using average annual emission rates
for such chemicals is not appropriate.
Another option discussed was the use of maximum emissions levels established in the air
permits of facilities. However, these data may be even less representative of actual emissions.
20
-------
It is important to obtain ambient monitoring data to test the validity of the model
estimates. Also a validation effort is needed in which several different exposure scenarios are
tested. The methodology only considers exposures to air; exposures through other pathways are
not evaluated. The approach does not consider the average duration of residence in the census
blocks. The toxicity weights are strictly for chronic health effects and do not take into account
acute effects, which may well be more significant. Also, ecological effects are not considered.
4.2 Question 2
The Basic Relative Burden Analysis (BRBA) method is relatively simple and may not
consider important parameters such as relative proximity, weather, stack height. Please
provide comment on the strengths, weaknesses, and utility of the "basic" method in
estimating the distribution of burden to areas proximate to facilities with air emissions.
4.2.1 Findings and Recommendations
The reviewers agreed that BRBA is simple and easy to use, but concluded that it has a
number of weaknesses. A fundamental weakness is that BRBA does not consider distance from
the plant or wind direction and velocity, all important variables. Any approach used for assessing
differences in exposures to chemicals in air should include an inverse distance function. It also
does not consider atmospheric stability. The TRI data are useful but have numerous limitations.
At a minimum these limitations must be discussed in the background document. Other sources of
emissions data need to be examined. For example, most major corporations have meteorological
towers which collect meteorologic data. Some companies measure the concentrations of emitted
chemicals at plant boundaries. On a site-specific basis, these data might be obtainable within the
180 day evaluation period.
Given the large number of uncertainties in the emissions levels, it is important to perform
an uncertainty analysis. Regardless of the results from the BRBA, it is necessary to do further
analyses because BRBA, in its present form, cannot be used to conclude that a particular
population group is more affected than another. Thus, the methodology is currently not a useful
screening tool.
4.3 Question 3
The Enhanced Relative Burden Analysis (ERBA) method was an extension of the BRBA
by using the Industrial Source Complex—Long Term, Version 2 (ISCLT2), a standard air
model, to model the toxicity-weighted air emissions from each facility. The
toxicity-weighted air emissions are modeled as if they were one "pseudo-chemical,"'
although stack and fugitive emissions were treated separately for each facility. This
approach has been adopted in order to make more manageable the screening evaluation of
potentially hundreds of chemicals and multiple sources. Please provide comment on the
utility and limitations of modeling several chemicals simultaneously as one
pseudo-chemical with the model. If individual chemical properties would make this
modeling method problematic, which classes of air release chemicals are likely to need to
21
-------
be modeled separately? Within the relatively small geographic areas analyzed, will
atmospheric degradation play a major factor in the analysis?
4.3.1 Findings and Recommendations
The ERBA is more useful than BRBA because it takes into account air dispersion of the
emitted species. However, it does not consider mass or density and treats all chemicals as stable
and equally dispersable. A potentially serious limitation is the lack of evaluation of coupled
transport in air. Also stack and fugitive emissions need to be considered separately since their
effects will occur at different distances from the plant. Fugitive emissions consist mainly of
volatile chemicals which may or may not be reactive; they will generally produce more local
effects. Emissions from tall stacks will lead to exposures further from the plant. Atmospheric
degradation and fallout are also important issues. ERBA specifically excludes evaluation of
potential accidental releases, which can significantly increase short-term exposure. It would be
beneficial to use a more sophisticated model.
Modeling one pseudo-chemical makes the screening process more manageable. However,
the approach has significant limitations. The chemicals should be separated into categories
depending upon their chemical/physical properties. For example, persistent compounds should be
considered separately from those compounds which are rapidly cleared. The kinetics of the
chemical in both the atmosphere and the exposed person should be taken into account. The role
of atmospheric degradation of specific chemicals should be considered. Further, different
chemicals have different effects on different organ systems. Combining cancer risk and non-
cancer health effects is a problem. Overall cancer risks and overall potential for non-cancer health
effects should be estimated separately. Another important weakness is the lack of evaluation of
acute effects.
4.4 Question 4
In the ERBA method, modeling of the air emissions was truncated at 2, 4, or 6 miles. For
example, in the 4-mile run, burden was added to census blocks within 4 miles from each
facility, but not beyond that, and correspondingly for the 2- and 6-mile runs.
Computationally, the number of census blocks potentially affected increases dramatically
with increasing radius from the facility and the burden values drop off as the radius
increases. (For example, with 314 facilities in Louisiana, the total number of census
block-facility combinations within 6 miles of any facility was over 300,000.) What are the
strengths and weaknesses of limiting the modeling to a certain radius from the facility for
the purpose of evaluating burden, and specifically, 2, 4, or 6 miles?
4.4.1 Findings and Recommendations
Insufficient information was presented on the ERBA model to conclude whether limiting
the modeling to a specific distance, such as 2, 4, or 6 miles, is scientifically appropriate. One
needs to know what factors make the most difference. It is well known that most dispersion
22
-------
models are not effective beyond six miles from the source. One also needs to know the dominant
chemicals from each site and how far they disperse. A sensitivity analysis is needed to understand
the process before technical advice can be given on the appropriate distance at which to truncate
the analysis. Moreover, determining the appropriate distance is a policy decision.
4.5 Question 5
Please provide comment on the strengths and weaknesses of the ERBA methods for
analyzing the relative burdens from airborne emissions from nearby facilities for one
population subgroup versus another in populations proximate to fixed air emissions
sources?
4.5.1 Findings and Recommendations
The ERBA method for analyzing the relative burdens is an improvement over the BRBA
method and is more scientifically defensible. A number of problems, however, exist with this
method. It assumes an equal distribution of the population within each census block, which is
often not the case. It also assumes that all the emitted chemicals disperse as if they were one
chemical. The chemicals should be divided into groups with similar properties and then modeled.
For example, the reactivity of the chemicals and their mode of transport in the atmosphere will
affect their dispersion. Furthermore, the toxicity should be computed at the receptors rather than
at the source of the emissions. The ERBA method uses standard models with default parameters,
but it is used at sites where conditions do not meet the default assumptions. The method should
evaluate site-specific conditions rather than model defaults. Sensitivity analyses should be
performed using different values for various parameters. Uncertainty analyses should also be
conducted.
Whenever the RBR is calculated, the uncertainties should also be discussed. Although the
RBR is mathematically acceptable, its value and meaning are questionable. Calculation of a
relative burden without consideration of risks to different populations is not likely not produce
meaningful answers as to differential health impacts on these populations. Like BRBA, the ERBA
method is a screening method. If the ERBA calculations indicate potential differential impacts,
further analyses will be needed, which use more data and more realistic assumptions.
4.6 Question 6
The average toxicity weighted concentration, or burden, for each census block has been
calculated. Please provide comment on the strengths and weaknesses of additional
information which can be derived from the BRBA and ERBA methods, such as ranking
census blocks in the state or smaller geographic area by average burden value or
comparing the average burden in blocks near one facility to those near another for the
purpose of identifying potential problem areas.
23
-------
4.6.1 Findings and Recommendations
Regarding the average toxicity-weighted exposure, or burden, a mapping of burdens in a
given area should be performed, using GIS and different colors in the census blocks, depending
upon the calculated burden. Percentiles, rather than average values, should be developed for the
census blocks. Based upon the color distribution, one could estimate the most impacted, as well
as the least impacted, census blocks.
There are a number of weaknesses in the KB A approach. It is not known how the
weighting factors are assigned when aggregating the block burden to obtain the average at one
point in the census block. Performing an inter-block comparison should be undertaken only with
great caution since burdens at different points could be the same, even though the chemical
mixtures could be quite different. Moreover, the criterion for baseline burden has not been
developed, so it is not clear how to compare two separate areas, both of which are heavily
polluted.
Identifying everything in terms of burden without addressing the potential effects on
human health muddies the issues. Toxicity considerations need to be incorporated more fully in
the analyses, with at least a consideration of whether there is, or is not, a potential health hazard.
The Agency should use the term "toxicity-weighted exposure" instead of "burden."
4.7 Question 7
What are the strengths and weaknesses of the BRBA methodology for assessing relative
impacts on population subgroups?
4.7.1 Findings and Recommendations
Advantages of the BRBA methodology are that the method is relatively easy to perform
and to explain, that the models and data that are used can be easily updated, and that sensitivity
and uncertainty analyses can be incorporated into model development. It can be performed in the
180-day time limit for responding to Title VI complaints. However, the simplicity of the
methodology is also a basic weakness.
In BRBA a circle is drawn and the toxicity-weighted exposure is distributed uniformly to
the population within the circle. This yields a set of relationships that are too simplistic. The
methodology does not consider meteorology, wind speed and direction, stack height, and
deposition/re-emission. A radial distribution function is needed, even for such a basic
methodology. Geographic information system (GIS) mapping is a useful tool for managing data
but still leaves the need to collect data on populations, such as duration at current residences and
activity patterns. Census data should be reviewed for frequency of moving and the percent of
employed persons.
24
-------
The BRBA methodology uses air exposures as the surrogate for exposures from all
pathways. This could lead to an underestimate of exposures. There is also no temporal relation
for the concentration which is expressed as an annual average; emissions excursions could lead to
acute health effects which are not considered. In addition, most people are concerned about their
neighborhood, not just about the particular census block in which they live. Questions exist as to
the completeness and quality of the TRI data base and the variations from State to State. Using a
composite pseudo-chemical does not take into account differences in reactivity of chemicals and
in their atmospheric transport.
The Agency should use the term "toxicity-weighted exposure" instead of "burden."
Further, EPA needs to develop a more specific and consistent definition of "toxicity-weighted
exposure" for the KB A methodology. Without a meaningful definition of "toxicity-weighted
exposure" or "burden", it is difficult to evaluate the BRBA method. A real exposure assessment
should be performed. Determining "toxicity-weighted exposures" or "burden" in census blocks
may not be meaningful, particularly for those blocks in which there is no evidence of health effects
or minimal projected risks, or in which there exist risks of significant magnitude to demand action.
Ultimately, the important endpoint is potential for health effects. If "toxicity-weighted exposure"
is to be estimated, then "toxicity-weighted exposure" values for all census blocks in a State should
be calculated, and a relative ranking of census blocks in the State should be developed. Then one
can compare the ranking of particular blocks with the percentage of specific populations living in
those blocks.
In its present form BRBA is not a useful first-order screening tool. Until such time as an
inverse distance function is incorporated into the methodology to account for dispersion,
undertaking additional analyses using this methodology is problematic.
4.8 Question 8
What are the strengths and weaknesses of the ERBA methodology assessing relative
impacts on population subgroups?
4.8.1 Findings and Recommendations
The main strength of ERBA as compared to BRBA is that distance from the facility is
taken into account in developing measures of burden or "toxicity-weighted exposure". A steady
state air dispersion model (ISC-LT) is used to calculate air concentrations of the pseudo-chemical
at various locations. Wind speed and direction, and stack height are taken into account. As with
BRBA, updating of ERBA is an easy task, and uncertainty and sensitivity analyses will be easy to
perform.
However, deposition/re-emission is not evaluated. Also one must be clear in presentations
to the public on the air dispersion model; otherwise, the function of the model may be overlooked.
Aside from including a distance function based upon air dispersion modeling, ERBA suffers from
a number of the same problems as does the BRBA version.
25
-------
4.9 Question 9
Please provide comment on the appropriateness of the review document's interpretation of
the Relative Burden Ratio, given the methodology and data used?
4.9.1 Findings and Recommendations
Regarding the interpretation of the relative burden ratio (RBR), the background document
does an excellent job of stating the problems. What is missing, however, is how to interpret
different relative burden ratios. For example, what is the significance of 1.1 versus 1.0? Is 2.0
significantly different from 1.0? Even if the RBRs are determined to be significantly different in a
mathematical sense, the EPA needs to determine the level of difference that is meaningful in a
regulatory sense. The RBR is a screening tool; it cannot be used in risk assessment and should
only be used with caution.
The definition of "burden" in the background document is different from that given in the
presentation to the IHEC. The Committee concluded that "toxicity-weighted exposure" is a
better term than "burden" since burden can be confused with concepts such as body burden.
A reality check on the RBA methodology would be to collect from State monitoring
stations ambient air concentration data of chemicals considered in ERB A and then compare the
concentrations of the chemicals from the monitoring data and the ERB A calculations.
4.10 Question 10
Please provide comment on the strengths and weaknesses of the ERB A method of
estimating general risk and hazard numbers from concentration burdens and its utility for
screening out de minimis burdens.
4.10.1 Findings and Recommendations
ERB A provides a more accurate estimate of concentrations of the pseudo-chemical in air
than does BRBA. It is a relatively simple approach, which could be performed within the 180 day
time frame of Title VI complaints, and it should be generally understandable by the public.
Like BRBA, however, it does not take into account emissions from area and mobile
sources. This is a problem for screening out de minimis burdens since ERBA does not take into
account cumulative exposures. Thus, it is not a sequential screening tool. In rural locations, area
and mobile sources may not be an important issue. But in cities, area and mobile sources are
significant contributors to cumulative exposure and must be considered.
Another problem with ERBA is that it does not subdivide the emitted chemicals into
subclasses according to how they behave in the environment. Different areas surrounding an
emissions source will be impacted more by particular classes of chemicals. One should develop
different risk indices for the various classes of chemicals.
26
-------
For screening purposes only, it is reasonable to lump cancer risks and non-cancer health
effects together. However, to make recommendations about de minimis risks, cancer risks and
non-cancer health effects need to be considered separately. One can add "toxi city-weighted
exposures" or "burdens", but cannot add cancer and non-cancer effects. Also de minimis risks
may vary by nature of the population in specific areas (e.g., Hispanic families generally have larger
families than Anglos in the same area).
There is a fundamental problem with extrapolating from RBR to potential harm or risk.
Unless one can derive health risks from burdens or toxi city-weighted exposures, the concept of a
de minimis risk based upon the RBR is meaningless. To estimate risks, one must evaluate the
risks of individual chemicals, or of several classes of similarly acting chemicals. A possible way to
derive risks from burdens would be to determine the subset of chemicals which drive the burden
values and then perform risk assessments of this subset of chemicals. Barring that, it is necessary
to keep risk and burden separate.
It is also important to inform the communities that the ERBA and BRBA methodologies
do not address health outcomes, only burdens.
4.11 Question 11
The ambient concentration modeling methodology associated with COATCEM is similar
to that used in several previous studies conducted by EPA and reviewed by the SAB (e.g.,
EPA-SAB-fflEC-96-004; EPA-SAB-EEC-98-007). Are there any assumptions or input
data involved in the COATCEM approach which would change the SAB's earlier
judgements? Please provide comment on the strengths and weaknesses of the approach
for assessing concentrations for the disparate impact analysis given the large number of
sources and chemicals considered in the analysis?
4.11.1 Findings and Recommendations
Limited detail has been presented on the COATCEM methodology, so it is difficult to
answer whether there is anything that would change the SAB's earlier judgments on the
Cumulative Exposures Project (CEP). However, the SAB had some reservations concerning the
reliability of the results from the CEP. It was noted that the CEP needed a strategic plan for the
collection of data and calibration and validation of the model. The SAB was also concerned that
the description of the modeling process did not detail its capabilities and limitations and that the
process did not model the uncertainties. Nor did it contain a sensitivity analysis. Based upon the
limited information presented on COATCEM, there is no evidence that COATCEM has yet dealt
with these issues.
Regarding the strengths and weaknesses of COATCEM, it is a more comprehensive and
more resource intensive approach than the RBA method. It includes evaluation of the
contributions of stationary, area and mobile sources. Conceptually, it is moving in the right
27
-------
direction. The level of detail on the 188 HAPs is good. It evaluates cancer risks and non-cancer
health effects separately.
However, the method has never been applied, so it is not possible to assess whether it is
an improvement over the KB A. The models used are relatively large and complex. This makes
performing sensitivity and uncertainty analyses difficult and time-consuming. At present the
method only evaluates HAPs, so some important chemicals are not included, which could result in
an underestimate of risks. Questions exist as to the availability and reliability of the data on area
and mobile sources; when completed, the 1996 National Toxics Inventory may address this issue.
In calibrating and validating the method, one should evaluate two test cases - one in which mobile
and area sources are not important in terms of exposure, and one in which area and mobile
sources are important contributors. Given COATCEM's complexity, it could take more than 180
days to utilize the method in any Title VI case.
It was agreed that there is a trade-off between value added and complexity added in using
a validated version of COATCEM. The value added by COATCEM cannot be analyzed at this
time because there is no outcome to compare to the outcome of ERBA.
COATCEM has significant potential and should be developed further. The priority within
the Agency for further development and application of COATCEM is somewhat unclear, since the
development to date of the methodology has been overseen by two EPA offices.
4.12 Question 12
Please provide comment on the strengths and weaknesses of the COATCEM method for:
(1) evaluating the relative burdens from airborne emissions from nearby facilities for one
group versus another in a population proximate to fixed air emissions sources, and (2) its
utility in screening out de minimis burdens.
4.12.1 Findings and Recommendations
The conceptual framework of COATCEM is scientifically defensible for various
populations and sub-populations. The method calculates cancer risks and non-cancer health
effects separately and estimates the contributions of area and mobile sources to burden as well as
those of point sources. It represents a step into the future. A significant advantage is that
disproportionate impact measures can be converted into risk indices since COATCEM uses more
of a statistical approach than a relative ratio approach. The statistical methods can include effect
modifiers and census data covariates. Because COATCEM is a risk-based approach, it has the
potential to be used to screen out de minimis risks by comparing atmospheric concentrations of
chemicals at specific locations to benchmark concentrations. This is an advantage over the RBR
approach.
The strengths of COATCEM are dependent, however, on the ability to obtain significant
amounts of data. Moreover, like ERBA, it does not take into account atmospheric chemical
28
-------
transformations, deposition, or variations in dispersion modeling around large bodies of water. It
does compare monitored and modeled data, but does not make corrections on the basis of the
monitored data. It only considers HAPs, but other chemicals can be added. Estimates of whether
risks are greater than de minimis levels could be improved by grouping chemicals into appropriate
chemical classes.
4.13 Question 13
The BRBA, ERBA, and COATCEM approaches described in the review document may
be applied to various geographic scales (e.g., national, regional, state, basin, county,
place) and collections of sources. Given the inherent uncertainties described in the review
document, please comment on how the results of the analysis relate to the resolution of
the input data, the varying geographic scales, and numbers of sources being analyzed.
4.13.1 Findings and Recommendations
The uncertainties of BRB A, ERBA and COATCEM are well described in the
background document. Because of the relative magnitude of the uncertainties, the Committee
questions whether the level of spatial detail used in the disproportionate impact analysis is
consistent with the accuracy of the methods. For example, it is not clear why a 50 meter-by-50
meter dense receptor grid is used in ERBA to obtain averaged results within a census block. The
use of a centroid concentration by census blocks, as is done in COATCEM, is appropriate if the
model used is accurate and has been validated at census-block scales. There are situations in
which the use of census blocks can have accuracy problems. For example, if tall buildings are
located within the grid, downwash and channeling of the plume will occur. These processes make
it difficult to model dispersion at small scales. This is especially an issue in cities.
It is not clear that COATCEM can handle large numbers of chemicals. Like RBA,
COATCEM only addresses inhalation exposures.
COATCEM is a variation of the CEP model, which is known to under-predict ambient
chemical concentrations. In order to be protective from a public health standpoint, predicted air
concentrations should be greater than monitored air concentrations.
COATCEM (and ERBA) can model dispersion of emissions out to 30 miles. Point, area,
and mobile emissions sources can all be modeled. It is important that all the input data be
collected during the same time periods. If the data are collected at different times, then it is
difficult to compare census blocks. It is crucial to use actual data from an area such as the
Louisiana Industrial Corridor to validate COATCEM and to compare its outputs with those of
ERBA.
The geographic scale should be defined so as to be able to provide information that would
be requested by discrete communities (e.g., neighborhood associations).
29
-------
4.14 Question 14
Overall, what are the other major uncertainties involved in using the BRBA, ERBA, and
COATCEM methods to address disparate impact issue? Are there situations where these
methods would have to be modified because the models used are not suitable? What
research or improvements in the methodologies would be most helpful to focus upon in
the next few years?
4.14.1 Findings and Recommendations
There are four categories of uncertainty (variability/ heterogeneity, true uncertainty, model
uncertainty, and decision rule uncertainty). Of these, the least important and easiest to deal with
is variability and heterogeneity, while the most important and hardest to address is decision rule
uncertainty. Examples of decision rule uncertainty in the disproportionate impact methodologies
are the definition of relative burden, the location at which to stop evaluating source emissions, and
the decision to combine (or not to combine) cancer risks and non-cancer health effects. To
improve the process, one should start by decreasing decision rule uncertainties. KB A (both the
basic and enhanced versions) have more model uncertainty than does COATCEM, but there are
more inputs to COATCEM.
Comparing the outputs of the KB A and COATCEM methods to actual monitoring data
will provide key information on the degree of uncertainty associated with the air modeling used in
the two methodologies. It is important for the modelers to communicate with the persons who
perform the air monitoring. There is a need to improve the emissions inventories; for COATCEM
to be validated, it will be important for the 1996 National Toxics Inventory to be completed.
Other areas of significant uncertainty are the absence of an assessment of acute health
effects in any of the methodologies and the lack of assessment of exposures other than from the
inhalation pathway. All methodologies should evaluate cancer risks and non-cancer health effects
separately. More than one pseudo-chemical should be evaluated; toward this end, the chemicals
should be categorized into important chemical classes. If, however, only one or two chemicals in
a study area are the primary contributors to exposure concentrations, it will not be necessary to
model all of the chemicals emitted.
Because of the significant uncertainties in any of the estimates of toxi city-weighted
exposure, there is a need for strategies to apply these estimates in the context of the uncertainties.
One strategy would be to define as significant only those toxicity-weighted exposure ratios that
are larger than the uncertainties in the method used to estimate them. Thus, if the method is only
able to make estimates that are good to a factor of 2, a ratio of greater than 2 would be required.
If the method is only good to a factor of 10, a ratio of greater than 10 would be required. In
practice, this would require that relative "burden" ratios could only be considered significant for
situations in which the calculated risks are above de minimis levels, and the ratios are larger than
the uncertainty factors in the specific method.
30
-------
Burden should derive from risk considerations, not vice-versa. First the atmospheric
concentrations of the emitted chemicals at key locations should be calculated. Then health
benchmark values should be used to determine whether estimated risks are below, at, or above de
minimis levels. It is only at this point that burden should be analyzed. When analyzing burden, it
is important to keep in mind that multiple factors contribute to disease states, including lifestyles,
genetic polymorphisms, and nutrition.
The community does not care which model is used; they just want to know whether the
outcome is defensible and to have it be explained in understandable terms.
31
-------
DISTRIBUTION LIST
Administrator
Deputy Administrator
Assistant Administrators
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries
Library of Congress
National Technical Information Service
Office of Technology Assessment
Congressional Research Service
------- |