-------
2.2.3. Constructing the Analysis Plan
The analysis plan is the final stage of the planning and scoping process (see discussion in
USEPA, 1998b). It describes how hypotheses about the relationships among the sources,
stressors, exposure conditions, populations, and adverse effects/endpoints (see box) presented in
the conceptual model and narrative will be considered during the risk analysis phase of the
assessment. The plan includes the rationale for which relationships (referred to as "risk
hypotheses" in USEPA, 1998b) are addressed, methods, models, and a discussion of data gaps
and uncertainties. It also may include a comparison between the level of confidence needed for
the management decision and that expected from alternative analyses in order to determine data
needs and evaluate which analytical
approach is best. In some cases, a phased
or tiered risk assessment approach can
facilitate management decisions,
particularly in cases involving minimal
data sets.
Important Details for an Analysis Plan
Sources:
Identification of sources to be included and methods and
associated data for including them.
Stressors:
Identification of stressors to be included and methods and
associated data for including them.
Clarification of direct- and indirect-acting stressors.
Exposure Conditions:
Specification of exposure conditions to be assessed, along
with methods.
Populations:
Identification of the populations on which analysis will
focus.
Endpoints or Adverse Effects:
Identification of one or more unique, well-defined
endpoint for analysis. Note that a concept such as "health
of the community" is not a well-defined endpoint.
Identification of linkages between assessment endpoints
and measurable attributes.
Specification of those endpoints or exposures that will be
measured directly and those that will be estimated or for
which surrogates will be used.
Identification of common endpoints/effects for groups of
stressors for which risks or impacts are to be combined.
Description of methods to be employed for combining
risks in terms of endpoints.
The analysis plan provides a
synopsis of measures that will be used to
evaluate risk hypotheses (as shown in
Appendix D). The plan is strongest when
it contains explicit statements of how
measures were selected, what adverse
effect (or assessment endpoint) they are
intended to evaluate, and which analyses
they support. Uncertainties associated
with selected measures and analyses and
plans for addressing them should be
included in the plan when possible. The
analysis plan can be a brief summary of
the key components of the risk
assessment and how each component will
be measured or calculated.
In a cumulative risk assessment, a
key aspect is considering whether and
how multiple stressors interact or act
together in contributing to risks; thus,
some early thought should be given to the
strategy for addressing this aspect of the
assessment. The strategy should address
methods to be employed for considering
potential joint action of multiple stressors
on a single endpoint as well as whether
28
-------
the assessment will attempt to describe
of this issue in the analysis plan can i
Chapter 3).
cumulative impact on multiple endpoints. The discussion
include both qualitative and quantitative approaches (see
As with the conceptual model, societal importance, complexity, and available data and
resources will determine the degree of sophistication and detail needed in the analysis plan. Key
data gaps should be identified. The plan should also include thoughts about how to fill the
information needs in the near term using existing information, in the midterm by conducting tests
with currently available methods to provide data on the agent(s) of interest, and over the long
term to develop better, more realistic understandings of exposure and effects and more realistic
test methods to evaluate agents of concern. The plan should explain how measures were
selected, what they are intended to evaluate, and which analyses they support. Uncertainties
associated with selected measures and analyses and plans for addressing them should also be
explicitly stated.
The analysis plan should include (where feasible) milestones for completing the risk
assessment. The plan may be revisited and revised periodically. If new information is acquired,
such revisions may refine hypotheses of exposure and toxicity, modify the risk hypotheses
addressed, or compare public concerns with the projected risk management options.
2.2.4. An Early Look at Uncertainty
In preparing the conceptual model and analysis plan, there should be some early thinking
about uncertainty: In Section 4.2.1, there is a discussion of different types of uncertainty that
should be considered in the analysis: (l);parameter uncertainly (uncertainty about technical,
scientific, economic, and political quantities), (2) model uncertainty (uncertainty about the
appropriate functional form of technical; scientific, economic, and political models), and (3)
disagreements among experts (e.g., about the values of quantities or the functional form of
models, as when different health scientists use different forms of dose-response models). These
considerations are important for interpreting the results of the study and should be considered in
the selection of methods as part of the planning, scoping, and problem formulation process.
The first of these uncertainties facing the planning team is the so-called epistemological
uncertainty (not yet even knowing what questions to ask). It is likely that in planning any
complex assessment, some questions will only become evident after the data collection or
analysis has begun. It is therefore important that the planning team make provisions for
revisiting the analysis plan—or even the conceptual model—at intervals during the process.
Even more helpful would be an agreed-upon mechanism for changing the analysis plan or
conceptual model before the need for revision arises, as it almost assuredly will.
The second general aspect of uncertainty that should be dealt with in the planning,
scoping, and problem formulation phase is "acceptable uncertainty." How much uncertainty is
the planning team willing to accept in the results of the study? Typically, this is a very difficult
question for risk assessors and decisionniakers to answer, but it is a key question that
enormously affects the cost and usefulness of the study.
29
-------
At one end of this acceptable uncertainty spectrum are risk assessments that are based
only on readily available information. At the other end is an assessment that starts with carefully
reasoned and detailed quality assurance parameters, leading to specific data being accepted or
rejected for the study based on predetermined quality assurance guidelines. This process in turn
leads to results with known and acceptable uncertainties, but it may either require expensive data
collection or cause the study to fail when none of the data meet the quality assurance
requirements. The planning team should decide where on this spectrum it wants to be for the
study under consideration and whether the results will allow meaningful decisions.
Again, the decision on acceptable uncertainty is difficult, but consideration early in the
process will improve the potential for producing an analysis suited to the needs of the
stakeholders.
2.3. Ecological versus Human Health versus "Integrated" Cumulative Risk Assessment
Cumulative risk assessments may include both human health and ecological aspects.
Several reports have dealt with cumulative ecological risk assessment in some detail (e.g., Foran
and Ferenc 1999; Ferenc and Foran 2000; USEPA 1998b). USEPA (2002c) noted some of the
major differences between human health and ecological assessments (see list below), and these
differences need to be considered when planning a cumulative risk assessment that includes both
aspects:
• Ecological systems are not as well known biologically as are human health systems,
either at the population and at the individual level;
• For this reason, and because biological communities and ecosystems are inherently
more complex, ecological risk assessment requires more preliminary analysis and
deliberation regarding endpoints and protective standards;
• Ecosystems, habitats, and ecological communities have traits and properties that
individuals do not or that are not applicable to individuals or populations;
• Ecological risk assessment has been generally applied to populations, not individuals,
whereas the reverse is true for human health risk assessments; and
• Ecological risk assessment should assess risk at multiple levels or organization, that
is, the molecule, cell, organism, population, community, and ecosystem.
The World Health Organization (WHO, 2001) has published approaches to integrating
human health and ecological risk assessments to improve data quality and understanding of
cumulative risks for decision making. The organization's approach includes an integrated
framework (modified from USEPA, 1998b) and case studies.
Many tribal cultures view ecological and human health in an integrated way such that
they cannot be easily separated. Similarly, there is some effort (especially in Canada) toward
integrating human health and ecological assessment as well as decisionmaking in a field known
as "strategic environmental assessment" (Bonnell and Storey, 2000). This approach has not been
applied widely in the United States, and it remains to be seen how it will develop in the next few
years.
30
-------
2.4. The Final Step Before the Analysis Phase: Discussion of Possible Outcomes
i
Before the analytical efforts of the cumulative risk assessment are started, it is useful for
the entire team to hold some preliminary! discussions about the possible results and their
implications. Given that statutory mandates, regulations, property rights, or due process may
constrain or define most or all acceptability criteria, what conclusions of the team will be
associated with various results or risk levels? For example, for a risk assessment team with
members from the community, industry, and local and other government entities, what would
happen if the assessment shows risk levejls to be "low"? Would members accept this?
Conversely, if "unacceptable" risks are determined, will all team members accept the results and
their potential responsibility to do something about that risk? Do team members understand the
limitations of the information to be generated?
Discussions like these will help dptermine whether the assessment can really address the
questions of the team. If not, the assessment may not be worth doing as planned. If members of
the team will not accept the possibility of a range of results, then it is important to reopen the
entire planning and scoping discussion before anything is done in the analysis phase, because the
planning and scoping phase has not been satisfactorily completed. Although it is not necessary
to have unanimity among stakeholders before proceeding with the plan, knowing where some of
the potential disagreements may occur after the analysis and risk characterization phases are
started allows the stakeholders as a group to plan beforehand for how such disagreements will be
addressed, should they occur. Although it is possible to ensure that all stakeholders have been
heard and their opinions given due consideration and weight, that does not necessarily mean that
all of them will get what they want. :
i
As an example, USEPA (2000f) is a case study where the stakeholders thought they had
agreement on roles, responsibilities, and approach, only to find that the group acrimoniously
splintered after the analysis results came back. The Baltimore report contains valuable lessons
learned in the area of stakeholder disagreements and agendas and can provide some insight for
planning teams.9 i
Discussions just prior to the analysis phase may lead to an assessment that is very
different from the one originally envisioned. For example, in the case of the cumulative risk
initiative for Cook County (IL) and Lake County (IN) (see box on next page) the original plan
was for a quantitative cumulative risk assessment, but because of the lack of some critical
information, the scope was changed. This led to an assessment that, although not as broad as in
the original plan—and that did not even directly calculate risk—had better stakeholder buy-in
and a better chance of success in providing useful information.
Finally, it should be acknowledged by all practitioners of cumulative risk assessment that
in the current state of the science there will be limitations in methods and data available. It will
9 This case study, along with several others, will be examined more fully in followup work to this
framework report. .
!
I 31
-------
Cumulative Risk Initiative (CRI) for Cook County (IL) and Lake County (IN)
(formerly the Chicago Cumulative Risk Initiative, CCRI)
CRI BACKGROUND AND OVERVIEW
In 1995 the Chicago Legal Clinic and 11 Chicago-area community advocacy groups filed a petition under the
Toxic Substances Control Act requesting that the EPA Administrator prohibit or further regulate emissions
from eight proposed or constructed incinerators in the Chicago metropolitan area and Northwest Indiana. The
petitioners believed that neither current statutes nor local siting laws adequately addressed cumulative impacts
of multiple sources of toxic pollutants in a geographic area. They requested that the Administrator restrict
emissions of dioxins, furans, mercury, lead, and cadmium from these sources. In May 1996 the petition was
withdrawn in response to an EPA offer to participate in an investigation of multimedia pollutant impacts in
Cook County, Illinois, and Lake County, Indiana. This effort became the CRI. A CRI is an attempt to
investigate cumulative loadings and hazards from pollutant sources, to develop community-based activities to
help address these concerns, and to use analytic results to help prioritize use of regulatory agency resources
EPA and the petitioners agreed to a four-phase project: (1) an environmental loadings profile (EPA 74/-K-1-
002); (2) a petitioner risk workshop (completed); (3) a hazard screening assessment (peer review draft available
January 2002); and (4) a risk-hazard management response.
HAZARD SCREENING ASSESSMENT
The CRI hazard screening assessment was authored primarily by Argonne National Laboratory, with input from
local State and Federal participants. Reflecting stakeholder deliberations, the report focuses on cumulative
hazard (not "risk" as typically defined by EPA) associated with noncriteria air pollutants ("air toxics ) in the
two-county study area. It relies on "off-the-shelf air pollutant information, including EPA's Toxics Release
Inventory Cumulative Exposure Project, Regional Air Pollutant Inventory Development System, and outdoor
air monitoring data. Emission estimates are "toxicity weighted," and modeled/monitored outdoor air pollutant
concentrations are compared with reference values to develop hazard index-like ratios. The ratios or toxicity-
weighted emission estimates are used to derive indicators of cumulative hazard and then mapped over study
area locations To identify geographic areas where potentially elevated hazards and individuals with potentially
greater susceptibility are collocated, another part of the study assembles pollutant hazard information and data
on existing human disease rates and indicators.
PRELIMINARY LESSONS LEARNED
1 A major planning/scoping/problem formulation effort by a broad group of stakeholders narrowed the scope
of the CRI hazard screening assessment and seemed to increase stakeholder "buy-in" with the process. This
was valuable, given the complexity, expense, effort, time requirement, and difficulty encountered in addressing
even the narrowed scope.
2 Large data gaps make risk and hazard assessment of environmentally relevant chemical exposures highly
uncertain even for single agents. Expanded assessments that address cumulative risk considerations (e.g.
mixtures, developmental toxicity, nonchemical agents) are a better match for real-world circumstances but
require acknowledgment of even more uncertainty.
3. Obtaining and managing input from a large group of technical stakeholders is cumbersome and time-
consuming, but that group's perspective and expertise greatly improved the CRI assessment.
4 Given that the National Research Council's 1983 four-step "framework" required several years for broad
use and acceptance in the United States, the greater complexity of cumulative risk (for CRI, cumulative hazard)
assessment suggests that an equally long period may be needed for terminology standardization, refinement of
approaches, and development of consensus methods.
-------
be important to identify these limitation^ and discuss them frankly in the cumulative risk
assessment report. Data limitations may be somewhat mitigated by qualitative information; the
collection of qualitative data may be valuable in cumulative risk assessment. Still, limitations in
methods or data should not be seen as a;convenient reason for completely ignoring or not posing
questions for which stakeholders may b£ seeking answers. Lack of an appropriate methodology
may indeed be a reason why certain questions cannot be addressed in the analysis phase, but
capturing the questions and having some discussion about why the questions could not be
addressed in the assessment is often helpful.
33
-------
3. THE ANALYSIS PHASE
The analysis phase (Figure 3-1) is primarily an analytic process in which risk experts
apply risk assessment approaches to evaluating the problem at hand.10 The risk assessment
paradigm most widely used by risk assessors during the past two decades was first documented
by the National Research Council (NRC) (NRC, 1983). It consists of four parts: hazard
identification, dose-response assessment, exposure assessment, and risk characterization. This
paradigm was developed when almost all risk assessments were being conducted on single
chemicals. Nevertheless, it is a useful place to start when considering cumulative risks.
This framework follows the NRC risk assessment paradigm in all respects except that the
exposure and hazard/dose-response components should be evaluated together rather than
separately. As a prerequisite to using this framework, assessors considering cumulative risk
assessments should be familiar with the 1983 NRC risk paradigm as well as the various EPA risk
assessment guidelines (see text box titled "EPA's Risk Assessment Guidelines" in Section 1.1).
In both single-stressor and multiple-stressor risk assessments, the analyst will look at
hazard and dose-resp'onse relevant to the stressor(s) of interest and perform an analysis of
exposure(s) to those stressor(s). This chapter begins with a basic discussion of this general
process and its basic ingredients (Section 3.1). The second part of this chapter (Section 3.2)
discusses some of the situations that arise in cumulative risk assessment, methods currently
available for addressing them, steps in the process, and some limitations to these methods.
Finally, Section 3.3 identifies areas of ongoing work that are particularly relevant to cumulative
risk assessment.
3.1. General Process
In developing the conceptual model and analysis plan (see Section 2.2), the scope of the
assessment was specified (see example in box on page 36). Some of the aspects of scope include
stressors, sources, pathways and media, exposure routes, populations and subpopulations,
endpoints, and measures.
The analysis plan should specify how data, modeling, or assumptions will be
obtained, performed, or defined for all of the details concerning the characterization of exposure
of the defined population and subpopulations to the defined set of stressors. Additionally, the
analysis plan specifies the strategy for obtaining and considering hazard and dose-response
10 Although the analysis phase is primarily an analytic process, with heavy emphasis on the role of the
scientist, risk assessor, or other technical expert, other stakeholders can be involved in various ways, as agreed upon
before the analysis phase begins. Some roles'stakeholders might have in the analysis phase include (1) suggesting
sources of data or providing data for the assessment; (2) helping clarify issues identified during problem formulation;
(3) working alongside the risk assessment experts to see what data and assumptions are being used and why and to
better understand how the risk assessment process works; and (4) suggesting alternate scenarios that may reflect
more realistic exposure conditions in the community. A variety of roles for stakeholders in the analysis phase can be
proposed and adapted for the particular circumstances of the individual case, assuming that the roles can be agreed
upon by the team.
34
-------
Planning, Scoping, &
Problem Formulation
Interrelation ancf""^,,
Pjsk Characterization.
Integration of Exposure, Hazard, and Dose-Response Information
Considering: — < - _„ ' •
-Time-flelated Aspects
- Vulnerability ; .
i - - SuBpopuktions with Special Exposure
z ~ 1 * ^ *a " ^r ~> " '•
Single Stressor Information
- Toxicologic Independence
- Toxicologic Similarity
Kf
Multiple Stressors Information
- Stressor.Interactions
- Joint Chemical Toxicity
JQ
O
s1
t
Measures and Metrics
- Decision Indices - Common Metric
- Probabilistic Approaches - Biomarkers
- Qualitative Approaches
Figure 3-1. The Analysis phase.
information for these stressors and the method for combining the exposure information with the
hazard and dose-response information to; generate risk estimates or measures. As the risk
analysis is refined, it may be appropriate!to revisit and refine the exposure, hazard, and dose-
response information in an iterative fashion.
In the integration of exposure, hazard, and dose-response information for a cumulative
risk assessment, several aspects of the assessment may be particularly important. These include
multiple-stressor hazard, dose-response 4nd exposure issues, exposure time or duration-related
issues, vulnerability (including susceptibility) of the study population along with the influencing
factors (including life stage), and subpopulations with special exposures. These items are
discussed in the following section along with the currently recognized methods for evaluating
the toxicity or risk associated with mixtures.
35
-------
The area of identifying and
assessing risk to susceptible
subpopulations has an increased profile in
cumulative risk assessments. A variety of
factors may be influential in affecting
population susceptibility. The extent to
which these can be considered will be
heavily dependent on existing knowledge
and available information.
3.2. Available Methods and Approaches
Many aspects of traditional risk
assessment methodology apply to
cumulative risk assessment. Predicting
cumulative risk of multiple stressors,
however, has required the development of
additional specific methods or approaches.
Additionally, there are some aspects of risk
assessment that, although common to both
single-stressor and multiple-stressor
assessments, may increase in complexity
or significance in a cumulative risk
assessment.
Scope of EPA's National-Scale
Assessment for Hazardous Air Pollutants
(also see Figure 2-3):
Stressors 33 priority urban hazardous air
pollutants (HAPs)
Sources Major industrial, small "area,"
mobile (on- and off-road), and
extrinsic "background" in air
Pathways/media Outdoor air, indoor air
Microenvironments
Routes Inhalation
Subpopulations General population only
Endpoints Cancers, developmental, central
nervous system, kidney, liver,
respiratory effects
Metrics For cancer: distribution of high-end
cancer risk estimates, predicted
percent of population within predicted
cancer risk ranges, predicted number
of cancer cases, HAP-specific and
cumulative
For other effects: distribution of
estimated hazard index values and
estimated percent of population within
specified ranges of index values
Although the aspects common to single-stressor and multiple-stressor assessments may
be many (e.g., the added dimension of multiple stressors influences consideration of stressor
sources, routes of exposure, environmental media/pathways, and other factors), several examples
are cited here. As one example, the assessment of the dose-response relationship and the
corresponding characterization of exposures in terms of duration, timing relevant to life stage,
and exposure history gain an additional dimension with the need to consider them cumulatively
in some way. The consideration of population susceptibility (as a part of vulnerability), as
recommended by EPA (USEPA 1995a, b, 2000c), also increases in complexity. A third example
of a complicating aspect in cumulative risk assessment is the consideration of subpopulations
that have particularly distinctive exposures. These examples are further discussed in Section
3.2.1.
Although it is beyond the scope of this framework report to describe all risk methods in
detail, Appendix B lists a variety of resources relevant to various exposure assessment methods.
Relatively speaking, there is a great deal of information on assessing human and environmental
exposures to chemical stressors and there is some information on biological and radiological
stressors, but there is comparatively little information on many other types of stressors.
36
-------
The most prominent aspect of cumulative risk assessment is often the prediction of the
combined-effects of multiple stressors. past and current activities in the development of
approaches for predicting risk of multiple stressors are described in USEPA (1986b, 2000e).
Concepts, approaches, or methods described in these documents or elsewhere are discussed in
section 3.2.2, with clarification of their applicability, limitations and notable points regarding
interpretation of the results they produce.
I
3.2.1. Examples of Increased Complexity of Cumulative Risk Assessment
Cumulative risk assessments can be quite complex (see text box on the following page
for an example). Three factors that can increase complexity in a cumulative risk assessment are
(1) time-related aspects, (2) vulnerability (including susceptibility), and (3) subpopulations with
special or particularly distinctive exposures. All three are relevant in single-stressor
assessments, but they have the potential; to be more complicated in multiple-stressor assessments
i
Time-related aspects. The issue of repeated exposures to a single stressor or exposures
to multiple stressors that may vary in time dimensions may have implications for susceptibility,
which, consequently, has implications for the dose-response relationship. Traditionally in dose-
response assessment, there is an inherent presumption that, for many stressors and effects, it is
the aggregate exposure (the combination of intensity and duration) to which the organism
responds (e.g., Haber, 1924). Thus dose-response assessments based on one pattern of exposure
(e.g., 6 hours per day, 5 days per week over a lifetime) are routinely applied to the assessment of
risk associated with a variety of patterns of exposure.
In the case of linear carcinogens, this aggregate exposure assumption has been carried as
an explicit assumption in the risk assessment step. Regardless of the details of the exposure
circumstances in the study on which the; cancer potency was based, it is assumed that there is a
linear relationship between amounts of exposure and associated cancer risk. For nonlinear
carcinogens11, and conceivably for linear carcinogens, if data indicate deviation from the
assumption that cancer risk is proportional to lifetime dose, the details and sequence of exposure
may be important, both in developing the dose-response relationship and in predicting risk
associated with exposures and life stages of interest.
Because some chemicals may haVe the ability to affect an organism's response to other
chemicals, consideration of the time sequence of exposure may take on an additional layer of
complexity in multiple-chemical cumulative risk assessments. For example, persons with ;
relevant past exposures might have increased susceptibility to the effects of a particular chemical
due to a previous exposure to the same—or a second—chemical.
11 The draft cancer guidelines (USEPA,| 19991) explicitly recognize the potential for nonlinear dose-
response. It is only in the case where nonlinear response is modeled that time sequence of exposure can be
considered in the risk assessment. ;
; 37
-------
The National-Scale Air Toxics Assessment
The National Air Toxics Assessment (NATA), which is based on 1996 emissions data is an ongoing series of
studies—some of which are completed—that will ultimately provide results that are useful in understanding the
quality of air and its possible effect on human health nationwide. The assessment includes 32 air toxics (a
subset of EPA's list of 188 air toxics) and also diesel paniculate matter (which is used as a surrogate measure for
diesel exhaust). Specifically, the assessment consists of four steps that will produce nationwide estimates of (1)
the release of these pollutants into the air from various sources, (2) the concentration of these compounds in the
air, (3) the exposure of populations to this air, and (4) the risk of both cancer and noncancer health effects
resulting from this exposure.
Purpose: The results of the national-scale assessment will provide important information to help EPA continue
to develop and implement various aspects of the national air toxics program. They will not be used directly to
regulate sources of air toxics emissions. Although regulatory priority setting will be informed by this and future
national assessments, risk-based regulations will be based on more refined and source-specific data and
assessment tools. More specifically, the assessment results will help identify air toxics of greatest potential
concern, characterize the relative contributions to air toxics concentrations and population exposures of different
types of air toxics emissions sources (e.g., major, mobile), and set priorities for the collection of additional air
toxics data and research to improve estimates of air toxics concentrations and their potential public health
impacts. Important additional data collection activities will include upgrading emission inventory information,
ambient air toxics monitoring, and information on adverse effects to health and the environment; establishing a
baseline for tracking trends over time in modeled ambient concentrations of air toxics; and establishing a
baseline for measuring progress toward meeting goals for inhalation risk reduction from ambient air toxics.
The Four Steps: The national-scale assessment includes the following four major steps for assessing air toxics
across the contiguous United States (also Puerto Rico and the Virgin Islands).
(1) Compiling a 1996 national emissions inventory of air toxics emissions from outdoor sources. The types of
emissions sources in the inventory include major stationary sources (e.g., large waste incinerators and factories),
area and other sources (e.g., dry cleaners, small manufacturers, wildfires), and both onroad and nonroad mobile
sources (e.g., cars, trucks, boats). EPA made some modifications to the 1996 National Toxics Inventory to
prepare the emissions for computer modeling.
(2) Estimating 1996 ambient concentrations bqsed on the 1996 emissions as input to an air dispersion model
(the ASPEN model). As part of this modeling exercise, EPA compared estimated ambient concentrations to
available ambient air toxics monitoring data to evaluate model performance.
(3) Estimating 1996 population exposures based on a screening-level inhalation exposure model (HAPEM4)
and the estimated ambient concentrations (from the ASPEN model) as input to the exposure model. Estimating
exposure is a key step in determining potential health risk. People move around from one location to another,
outside to inside, etc., so exposure is not the same as concentration at a static site. People also breathe at
different rates depending on their activity levels, so the amount of air they take in varies. For these reasons, the
average concentration of a pollutant that people breathe (i.e., exposure concentration) may be significantly
higher or lower than the concentration at a fixed location (i.e., ambient concentration).
(4) Characterizing 1996 potential public health risks due to inhalation of air toxics. This includes both cancer
and noncancer effects using available information on air toxics health effects, current EPA risk assessment and
risk characterization guidelines, and estimated population exposures. Using the toxicological independence
formula and the default assumption of additivity of risks (USEPA, 1986b, 2000e), this assessment combines
cancer risk estimates by summing them for certain weight-of-evidence groupings and also across all groupings.
For noncancer effects, the assessment assumes dose additivity and aggregates or sums hazard quotients for
individual air toxics that affect the same organ or organ system (USEPA, 2000e), in this case combining air
toxics that act as respiratory irritants.
38
-------
These considerations suggest that for cumulative risk assessment, chemical exposures
need to be characterized in terms of which other chemicals are present, and when. As noted in
ILSI (1999), "Data collected specifically to support a cumulative exposure assessment should
conserve the covariance and dependency structures associated with the chemicals of concern." It
is important to note, however, that the level of detail to which exposures are characterized should
be closely tied to the level of detail of information available in the dose-response assessment,
because a lack of corresponding detail in the dose-response assessment can pose a limitation on
the interpretation and usefulness of detailed exposure estimates.
Cumulative risk assessment can present challenges in matching exposure estimates with
dose-response relationships. Ideally, the dose-response assessment will indicate whether the
time sequence for the chemical(s) or stressors of interest in the assessment is important for risk
estimation. In cumulative assessments involving chemicals for which the time sequence of
exposure is important, it may be necessary to characterize the details and sequence of exposure
to the exposed population (see text box'on the following page), so that there will be a match in
not only the form, but also in the assumptions between the dose-response relationship and the
exposure/dose estimate. ''
i
Vulnerability. One of the concepts that can be used in risk assessments (both for human
health and ecological assessments) is that of vulnerability of the population or ecosystem.
Vulnerability has been a common topic: in socioeconomic and environmental studies. The
European Commission's TEMRAP (Th'e European Multi-Hazard Risk Assessment Project),
studying vulnerability to natural disasters such as floods, windstorms, fires, earthquakes, and
others, defines vulnerability as "the intrinsic predisposition of an exposed element [organism,
population, or ecologically valuable entity] to be at risk of suffering losses (life,;health, cultural
or economic) upon the occurrence of an event of [a specific] intensity" (European Commission,
2000). Kasperson et al. (1995) defines vulnerability as "The propensity of social or ecological.
systems to suffer harm from external stresses and perturbations. Involves the sensitivity to
exposures and adaptive measures to anticipate and reduce future harm." Kasperson (2000)
identified four types of vulnerability, discussed further below.
The Agency's risk characterization policy and guidance (USEPA, 2000c) touches on this
concept by recommending that risk assessments "address or provide descriptions of [risk
to]...important subgroups of the population, such as highly exposed or highly susceptible
groups." Further, the Agency's guidance on planning and scoping for cumulative risk
assessments (USEPA, 1995b) recognizes the importance of "defining the characteristics of the
population at risk, which include individuals or sensitive subgroups...." That guidance also
recognizes the potential importance of Other social, economic, behavioral, or psychological
stressors that may contribute to adverse ^health effects (e.g., existing health condition, anxiety,
nutritional status, crime, and congestion). As discussed below, the ways in which the Agency
and others describe these concepts in the context of human health risk assessment overlap the
various ways described by Kasperson (2000) in which human and biological ecosystems,
communities, and populations may be vulnerable: susceptibility/sensitivity, differential
exposure, differential preparedness, and differential ability to recover.
39
-------
Examples of Exposure Models that Consider Time Aspects
Calcndex (Novigen Sciences, Inc) integrates different pathways (e.g., dietary [food and water] and residential)
and routes of exposure (oral, dermal, inhalation) using a calendar-based probabilistic approach. One of the
important factors of this approach is that it provides estimates of risk that reflect aggregate and cumulative
exposure to discrete individuals, with exposure pathways and routes appropriately linked for the scenarios being
assessed. Calendex also allows one to estimate exposure before and after the use of a chemical, as well as
during degradation periods. Calendar-based assessments maintain the integrity of the individual by capturing
the location of the exposed individual, the time of year in which he or she was exposed, and the patterns of
exposure. Calendex also allows for a variety of time-breakout options for the analysis of exposure.
APEX - The Air Pollution Exposure (APEX) model is based on the probabilistic National Ambient Air Quality
Standards exposure model (pNEM) for carbon monoxide (Johnson et al., 2000). This model mimics the basic
abilities of the pNEM/CO model; it calculates the distributions of human exposure to selected airborne
pollutants within a selected study area as a function of time. As a dose model (for carbon monoxide), it
calculates the pollutant dose within the body, specifically summarized by the blood carboxyhemoglobin
(COHb) concentration. APEX is a cohort-microenvironment exposure model in that it combines daily activity
diaries to form a composite year-long activity pattern that represents specific population cohorts as they move
from one microenvironmentto another. A cohort consists of a subset of the population that is expected to have
somewhat similar activity (and hence exposure) patterns; it is formed by combining, demographic groups and
geographic locations (districts). Once each cohort has been modeled and its relative size determined, an
exposure distribution for the entire population can be assembled. A microenvironment is a description of the
immediate surroundings of an individual that serves as an indicator of exposure (e.g., inside a residence, school,
or car; outdoors; etc.). APEX has been developed as one of the inhalation exposure models accessible in the
Exposure Event Module of the Total Risk Integrated Methodology (TRIM.Expo) for assessment of exposures to
either criteria or hazardous air pollutants (USEPA, 1999J)
Other models include Lifeline, developed under a cooperative agreement between EPA/OPP and Hampshire
Research Institute (HRI, 1999,2000); Stochastic Human Exposure and Dose Simulation (SHEDS), under
development by EPA's Office of Research and Development (Zartarian et al., 2000), and Cumulative and
Aggregate Risk Evaluation System (CARES), under development by member companies of the American
Crop Protection Association (ACPA, 1999) along with Residential Exposure Year (RExY), which is being
developed by Infoscientific.com.
The first of Kasperson's categories is susceptibility or sensitivity. Although these two
words may have slightly different meanings, they are often used interchangeably. They refer to
an increased likelihood of sustaining an adverse effect, and they are often discussed in terms of
relationship to a factor describing a human subpopulation. For example, susceptible persons or
populations may be those who are significantly more liable than the general population to be
affected by a stressor due to life stage (e.g., children, the elderly, or pregnant women), genetic
polymorphisms (e.g., the small but significant percentage of the population who have genetic
susceptibilities), prior immune reactions (e.g., individuals who have been "sensitized" to a
particular chemical), disease state (e.g., asthmatics), or prior damage to cells or systems (e.g.,
individuals with damaged ear structures due to prior exposure to toluene, making them more
sensitive to damage by high noise levels) (Morata et al., 1997). Confronted with equal
concentrations of a chemical for equal durations, for example, a biologically susceptible or
40
-------
sensitive individual may show effects, whereas the typical individual within the population
would have no or less severe effects. This category would also include generation-skipping
• effects. Although we generally do not hkve a lot of data available on this topic, susceptibilities
or sensitivities may also exist among races or genders.
Kasperson's second category of vulnerability is differential exposure. Although it is
obvious from examining a dose-response curve that two individuals at different exposure levels
may have a different likelihood of effects, this category extends to differences in historical
exposure, body burden, and background'exposure, which are sometimes overlooked in an
assessment. When looking at the dose-response curves for a typical individual and an individual
vulnerable due to differential exposure, the curves may be the same, but the vulnerable
individual may be currently at a higher dose due to greater current or prior exposure and body
burden, so an increment of additional exposure may (due to slope of the curve at that point)
produce a more pronounced effect than in a typical individual.
Kasperson's third category of vulnerability is differential preparedness to withstand the
insult of the stressor. This is linked to wliat kind of coping systems and resources an individual,
population, or community has: the more {prepared, the less vulnerable. As an example, consider
two individuals, one of whom has had a phildhood disease immunization shot and the other has
not. The two may be exposed to the same insult, but due to a difference in preparedness, the
effects on the person with the immunization shot may be much milder or nonexistent. As
another example, hurricanes typically cause less damage to boarded up homes than they do to
homes without this reinforcement, even though the weather insult to both homes may be the
same. j
Kasperson's fourth category is the differential ability to recover from the effects of the
stressor. This again is linked to what kind of coping systems and resources an individual,
population, or community has. One aspect of differential ability to recover is illustrated by
differing survival rates for the same disease (e.g., Lantz et al., 1998). Put in terms of progression
of disease, for example, two persons in an early stage of cancer have different prospects for
recovery if one is treated immediately while the other does not have access to, or does not trust,
health care. On the ecological side, opportunistic infections in marine mammals12 appear to be
related to accumulation of polychlorinated biphenyls and organotin compounds, which cause an
immunosuppression response in laboratory animals (Tanabe, 1998).
Preparedness and ability to recover are often crucial factors in ecological assessments. In
human health assessments, lack of access to health care, income differences, unemployment, or
lack of insurance, for example, may affect a community's ability to prepare for or recover from
a stressor. -
i
Cumulative risk assessments may be uniquely suited to addressing the issues related to
vulnerability. In order to do so, however, there should be some relationship between the factors
12 That is, infections easily warded off by healthy marine mammals.
41
-------
discussed above and changes in risk. Many of these factors have not yet been extensively
developed beyond correlations between mortality rates and several socioeconomic factors, such
as income (e.g., Lynch et al., 1998). Susceptibility has been more developed than the other
factors, and current approaches implemented by EPA and others to address risk of noncancer
endpoints routinely employ a 10-fold factor to address heterogeneity in sensitivity. Variability
with regard to susceptibility is discussed in detail by NRC (1994), and the current state of
knowledge concerning epidemiologically based (e.g., oncogene-specific) risk factors provides
empirical data upon which at least crude estimates of the magnitude of heterogeneity in
susceptibility to toxic response can be based. However, much research in this area remains to be
done.
Subpopulations with Special Exposures. Certain subpopulations can be highly exposed
to stressors because of geographic proximity to the sources of these stressors, coincident direct
or indirect occupational exposures, activity patterns, or a combination of these factors. The
Agency's risk characterization policy and guidance (USEPA, 2000c) includes recognition of the
need for risk information to include, as available, information on highly exposed subgroups.
Accordingly, risk assessments, including cumulative assessments, may need to put special
emphasis on identifying and evaluating these subpopulations.
Subpopulations at risk of high exposure due to geographic proximity could include
workers at a facility that is a source of a stressor or residents near such sources. Specific
examples might be people living downwind from a coal-burning power plant, those near and
using a polluted water body (e.g., for fishing or recreation), or those living or working near
roadways with high levels of vehicular traffic. Occupational exposures may be either direct
(occurring in the workplace) or indirect (occurring at home). Indirect occupational exposures
include those experienced by family members who may be exposed to occupational chemicals
brought into the house by the worker (e.g., on clothing). Thus, workers or family members may
be subject to greater exposures than others in the population who do not have this additional
burden.
Examples of subpopulations at high exposure due to activity patterns may include people
who exercise heavily in polluted air, recreational or subsistence fishers or hunters who consume
large quantities offish or game, farmers or others who get a large proportion of their food from a
location near a source of pollution and live in areas with high pesticide use, individuals who
have long commutes in automobiles, or children (because they consume a larger amount of food,
drink, and air relative to their body weight and because of additional exposure routes such as
incidental soil ingestion). Additionally, some subpopulations may be affected by the combined
impact of high geographic exposure and high-exposure activity patterns (e.g., runners who run
along heavily traveled roadways and people who fish for food in heavily polluted urban rivers).
It is important to recognize that some heavily exposed populations may also be
particularly vulnerable or susceptible to the effects associated with the stressors of concern.
Examples of those who could be particularly vulnerable to certain stressors include children
during certain stages of development, people with chronic respiratory problems, the elderly, and
those who are economically disadvantaged and do not have access to medical care. A
42
-------
cumulative risk assessment may need to take into account potential combinations of high
exposure and high vulnerability, but few, if any, methods are currently available and accepted to
address the combined effects of exposure and vulnerability. This is an important area for further
research and methods development.
3.2.2. Approaches for Predicting Risk of Multiple Stressors
i
Combination toxicology (Carpy et al., 2000) is the study of the toxicity of mixtures. In
such studies, one may either measure the mixture toxicity directly (whole mixture toxicity), or
one may develop an estimate of the combined toxicity from information on the multiple
component stressors acting in concert with each other. (Toxicity of chemical mixtures has also
been modeled on a physiologically based pharmacokinetic basis [e.g., Haddad et al., 2000,
2001].) If evaluated using its component chemicals, the mixture toxicity data set should be
treated only as a snapshot of a multidimensional dose-response relationship, because the joint
toxicity and interactions can change witfji changes in exposure route, duration, relative
proportions of the components, or the effect being tracked. The application of such a data set to
a specific situation then requires careful matching of the test mixture composition and exposure
conditions to those of the target situation.. In whole mixture toxicity, once the mixture toxicity is
known, a risk evaluation can be done onithe mixture using the 1983 NRC risk assessment
paradigm. On the other hand, component-based mixture assessments are rarely evaluated using
the strict NRC paradigm, because the exposure and toxicity information should be compatible,
requiring some iteration to obtain toxicity information that is relevant to the actual exposure
estimates (USEPA, 2000e). ;
To address concerns over health risks from multichemical exposures, EPA issued
guidelines for health risk from exposure to chemical mixtures in 1986 (USEPA, 1986b). The
guidelines described broad concepts related to mixtures exposure and toxicity and included few
specific procedures. In 1989, EPA published guidance for the Superfund program on hazardous
waste that gave practical steps for condupting a mixtures risk assessment (USEPA, 1989a). Also
in 1989, EPA published the revised document on the use of TEFs for characterizing health risks
of the class of toxicologically similar chemicals that included the dibenzodioxins and
dibenzofurans (USEPA, 1989b). In 1990, EPA published a technical support document to
provide more detailed information on toxicity of whole mixtures and on toxicologic interactions
(e.g., synergism) between chemicals in a;two-chemical mixture (USEPA, 1990a). Whole-
mixture assessments, toxicologic independence and similarity, and risk methods using
toxicologic interactions are discussed at length in USEPA (2000e).
Risk assessment for mixtures usually involves substantial uncertainty. If the mixture is
treated as a single complex substance, these uncertainties range from inexact descriptions of
exposure to inadequate toxicity information. When viewed as a collection of a few component
chemicals, the uncertainties also include the generally poor understanding of the magnitude and
43
-------
nature of toxicologic interactions, especially those involving three or more chemicals. Because
of these uncertainties, the assessment of health risk from chemical mixtures should include a
thorough discussion of all assumptions and the identification, when possible, of the major
sources of uncertainty.
3.2.2.1. Single stressor information
Assessments that evaluate the risk from a single stressor do not fall into the category, of
cumulative risk assessments under the definition given in Section 1.3, whether these single-
stressor assessments address a single (dominant) endpoint or multiple endpoints or whether the
exposures are simple or complex (e.g., multisource, multipathway, multiroute exposure). Some
may be termed "aggregate risk assessments" by extension of the FQPA terminology. They can,
however, provide useful information for cumulative assessments.
A cumulative risk assessment considers the joint impact of multiple stressors. Studies on
individual stressors can, however, provide informative qualitative information for multistressor
assessments, particularly regarding hazard identification. The collection of single-stressor
effects can indicate the variety of types of adverse effects likely to result from the stressor
combination, although perhaps not the magnitude or extent of the effects. Factors affecting
population susceptibility to the individual chemicals are also likely to be important with the
combined exposure. To go further in terms of quantitative risk assessment requires
consideration of the potential for joint toxicity. For most exposure situations, hazard and dose-
response studies of all of the joint effects from the multiple stressors will not be available, so that
conclusions will have to be based at least partly on the single stressor information.
Exposure assessments for single stressors also need further consideration before they can
be used to characterize long-term exposure to all the stressors by all pathways. Transport and
environmental transformation of a chemical can be influenced by the presence of other
chemicals. Consequently, both the exposure levels and the relative proportions of chemicals at
future times may not correspond well to present measurements of a combination of chemicals
unless these influences are taken into account. In addition, exposure to one stressor may
influence the uptake of a second stressor. For example, a nonchemical stressor that increases
ventilation rate will increase the inhalation uptake of airborne chemicals.
Toxicologic independence. Two situations allow plausible approximations of the joint
exposure-response relationship using only the single stressor information: toxicologic
independence and toxicologic similarity (USEPA, 2000e). In the case of toxicologic
independence, if the toxicity modes of action are biologically independent, then as long as there
are no pre-toxicity interactions (e.g., metabolic inhibition, influence on uptake), the single
stressor information is sufficient to approximate the joint exposure-response relationship. When
the effects from two or more stressors are different, the cumulative response, if toxicologically
independent, is merely all the single-stressor responses, as if the other stressors were not present.
For example, joint but low exposure to heat (causing minor elevated heart rate and toluene
(causing minor hearing loss) would be expected to cause both the minor heart rate elevation and
minor hearing loss, but to the same extent as expected for each stressor alone. If each stressor is
44
-------
below its toxicity threshold, then for stressors exhibiting toxicologic independence, there will be
no estimated cumulative response, because the set of individual responses is then a collection of
zeros, ' . ... .
i
When the single stressor and cumulative toxicities are each represented by a frequency or
probability for affected individuals—also termed a probabilistic risk—then independence means
that "response addition," as defined in USEPA (2000e), can be applied for each adverse effect
that the stressors have in common. When all the single-stressor risks are low, the joint risk of a
common effect under response addition pan be approximated by the simple sum of the single-
stressor risks. For example, if reproductive toxicity is the general effect common to the multiple
chemicals, the cumulative risk of reproductive effects (at low single-chemical risk levels) is
approximately the sum of the single-chemical reproductive risks. Risk addition under
independence places no constraints on the individual chemical dose-response curves.
Toxicologic Similarity. In the second situation, the stressors are grouped according to
the common mode of action for each effect of concern determined in the planning and scoping
phase (USEPA, 2002a). For all effects caused by that mode of action, "dose addition" (USEPA,
2000e) can be applied to the stressor group. Thus far, this approach has been used only with
combinations of lexicologically similar chemicals, not with combinations of chemicals with
other kinds of stressors such as radiation;, physical factors, or health status. With similar
chemicals, each chemical exposure is converted into the equivalent exposure level of one of the
chemicals, called the index chemical. The joint toxicity or risk from the combined exposure is
then estimated by determining the effects or risk for that equivalent exposure level using the
dose-response information for the index chemical. For example, with the dioxins and furans (see
text box on next page), each congener exposure level is converted into its equivalent exposure as
the index chemical, 2,3,7,8-TCDD (USE|PA, 1989b).
Although the assumption itself is not complicated, the decision to assume toxicologic
similarity can be complicated, depending on the level of assessment decided on in the planning
and scoping phase and described in the analysis plan. The implementation used in Superfund
assessments (USEPA, 1989a, Part D) is a rough approximation to dose addition where a
hazardindex is determined whenever chemicals have a common target organ. The
implementation by the Office of Pesticide Programs in support of FQPA (USEPA, 2002a) is
much more extensive and requires knowledge of modes of action in order to calculate the
Relative Potency Factors (RPFs) for the effect of concern (see example in Appendix E). The
TEF method used for the dioxins is a special case of the RPF method (see Appendix E); it
requires the most toxicologic similarity because the similarity applies to every toxic effect by
any type of exposure (USEPA, 2000e). j
Single stressor information can also be used with dissimilar chemicals to gauge the
potential for toxicologic interaction. For Example, chemicals with long whole-body half-lives or
long tissue residence times have the potential to be present in those tissues at the same time.
Such overlapping exposures can result in a higher effective tissue dose, altered tissue doses
caused by toxicokinetic interactions, or altered toxicity from interacting toxic mechanisms.
45
-------
Toxicologic Similarity: The Dioxin Reassessment
Scientists from EPA, other Federal agencies, and the general scientific community have been involved in a
comprehensive reassessment of dioxin exposure and human health effects since 1991 (USEPA, 2002d). The
final dioxin reassessment will consist of three parts. Part 1: Estimating Exposure to Dioxin-Like Compounds
will include four volumes that focus on sources, levels of dioxin-like compounds in environmental media, and
human exposures. Part 2: Human Health Assessment Document for 2,3,7,8-Tetrachlorodibemo-p-Dwxm
(TCDD) and Related Compounds will consist of two volumes that include information on critical human health
endpoints, mode of action, pharmacokinetics, dose-response, and toxicity equivalence factors (TEFs) Part 3:
Integrated Summary and Risk Characterization for 2,3,7,8-Tetrachlorodibemo-p-Dioxin (TCDD) and Related
Compounds will be a stand-alone document. In this summary and characterization, key findings pertinent to
understanding the potential hazards and risks of dioxins are described and integrated, including a discussion of
all important assumptions and uncertainties.
237 8-Tetrachlorodibenzo-p-dioxin (TCDD) is highly toxic to many animal species, producing a variety of
cancer and noncancer effects. Other 2,3,7,8-substituted polychlorinated dibenzo-^-dioxins (PCDDs) and
dibenzofurans (PCDFs) and coplanar polychlorinated biphenyls (PCBs) exhibit similar effects, albeit at different
doses and with different degrees of confidence in the database. The similarities in toxicity between species and
across different dioxin congeners stem from a common mode of action via initial binding to the aryl hydrocarbon
(Ah) receptor. This common mode of action is supported by consistency in effects evident from data from
multiple congeners. This has led to an international scientific consensus that it is prudent science policy to use
the concept of TEFs to sum the contributions of individual PCDD, PCDF, and coplanar PCB congeners with
dioxin-like activity (van den Berg et al., 1998). The data supportive of dioxin-like toxicity, both cancer and
noncancer, are strongest for those congeners that are the major contributors to the risk to human populations. In
addressing receptor-mediated responses resulting from complex mixtures of dioxin-hke congeners, this
assessment has provided a basis for the use of integrated measures of dose, such as average body burden, as more
appropriate default metrics than daily intake. The Agency recognizes, however, that the final choice of an
appropriate dose metric may depend on the endpoint under evaluation.
In this study 237 8-TCDD was chosen as the index chemical, and the other dibenzo-p-dioxins and
dibenzofurans and coplanar PCB doses were adjusted to 2,3,7,8-TCDD-equivalent toxicities so the doses could
be added.
When a careful evaluation indicates no internal dose overlap, including metabolites, the single
exposures might be considered independently.
3.2.2.2. Information on stressor interactions and multiple exposures
One important simplification that is common in single-stressor human health assessments
is the separate evaluation of many of the key steps. That is, simplifying assumptions have often
been made regarding many characteristics of exposure (e.g., continuous vs. intermittent
variations in magnitude). For a given exposure route, for example, only one dose-response curve
may be used for the bounding case of setting a cleanup or action level of exposure and also the
predictive case of estimating existing risk. These simplifying assumptions allow the dose-
response step to be performed in isolation from the exposure assessment step, with the two'steps
executed in either order. For health-protective action levels, one may use bounds, such as the
upper bounds on toxic potency and exposure and lower bounds on the resulting acceptable
46
-------
exposure level. Such bounds may be much easier to calculate, but they may be more difficult to
interpret in terms of the uncertainties, likelihood, and closeness to the best or central estimate.
The incorporation of multiple chemicals, other stressors, and multiple exposure
conditions obviously complicates the assessment and the use of simplifying assumptions.13 In
cumulative assessments, performing the exposure and dose-response steps of the risk assessment
paradigm separately is an approximation; that obviously invokes a simplifying assumption. If the
dose-response data do not represent the same conditions as the exposure being assessed, an
extrapolation has to be made, which introduces additional uncertainty that should be clearly
stated. Joint or cumulative toxicity depends on the total dose or exposure, relative exposure
levels, and the many characteristics of exposure (e.g., duration, continuous vs. intermittent
presence, route, co-occurrence with other chemicals). In many cases, the complexities
introduced by multiple stressors will not allow use of some of the common simplifying
assumptions of single-stressor assessments. For example, toxicologic interactions have been
shown to change when the same doses are used but the sequence of exposure is reversed (i.e.,
chemical B then A instead of A then B), so that the exposure and dose-response analysis should
be compatible and performed together. :
i
Nonchemical stressors (e.g., biological entities or even physical stressors such as noise)
can also interact with chemicals to change the risks either that would cause separately. For
example, chemicals such as toluene can damage the auditory system and have been shown to
potentiate the effects of a physical stressor, noise, on hearing loss (Morata et al., 1997; Morata,
2000). For aquatic organisms, the toxicity of polyaromatic hydrocarbons increased with
exposure to ultraviolet radiation (Oris an$ Geisy, 1985).
Toxicity and interaction data that cover the full range of exposures for the exposure-
response relationship for the mixture of interest is usually impossible to obtain because of limits
on budget and other resources. More feasible approaches to cumulative risk characterization,
beyond that with various simplifying assumptions, then require close matching of the exposure
and dose-response steps to minimize the data requirements. In many cases, screening-level
ranking may be the only practical assessment. In some cases there will be sufficient information
for some quantitative evaluation of cumulative health risks that reflect both the complex
exposures and toxicologic interactions. ,
"Joint chemical toxicity" means the outcome of exposure to multiple chemicals that
includes the single-chemical effects along with any toxicologic interactions. Chemical
interactions can be divided into two major categories: those resulting from a toxicokinetic mode
of action and those resulting from a toxicOdynamic mode of action (USEPA, 2000e).
Toxicokinetic modes of interaction involve alterations in metabolism or disposition of the toxic
chemicals, for example, by the induction or inhibition of enzymes involved in xenobiotic
activation and detoxification. Toxicodynamic modes of interaction include those processes that
13 For ecological risk assessments, Gentile et al. (1999) review the theory and several methods for
evaluating stressor/response linkages and stressorinteractions for multiple stressors.
i 47
-------
affect a tissue's response or susceptibility to toxic injury. A simplifying observation is that most
interactions seem to involve pharmacokinetics. Unfortunately, most studies of toxicologic
interaction to date have involved only two chemicals, and few have quantified the magnitude of
the interaction or its dependence on exposure conditions.
Toxicologic interactions are commonly described in terms ofsynergism or antagonism.
These terms are only marginally useful, in part because the underlying toxicologic concepts are
only defined for two-chemical mixtures, and most environmental and occupational exposures are
to mixtures of many more chemicals. Further, the mathematical characterizations ofsynergism
and antagonism are inextricably linked to the prevailing definition of "no interaction," instead of
to some intrinsic toxicologic property (Hertzberg and MacDonell, 2002). EPA (USEPA, 2000e)
has selected "dose addition" as the primary "no interaction" definition for mixture risk
assessment, so that synergism would represent observed toxic effects that exceed those predicted
from dose addition . The EPA mixture risk guidance also describes a modified hazard index that
incorporates evidence of pairwise toxicologic interactions but notes that the pairwise evidence
may be specific to the exposure conditions of the study. The guidance further encourages
development of full biomathematical models for the joint toxicity—such as those based on
pharmacokinetics—so that qualitative interaction labels such as synergism are replaced by
quantitative estimates of mixture response that directly reflect the actual environmental exposure
levels.
3.2.2.3. Decision indices
Among the complexities of cumulative risk assessments is the frequent need to combine
widely differing types of data. Exposure data for some stressors may be available only as time-
weighted averages, whereas other data reflect daily human activity patterns. Toxicity data for
some chemicals may allow estimation of probabilistic risk for one endpoint while providing only
qualitative descriptions of other endpoints. It is possible to develop the risk characterization
using the original information in a matrix, but such a summary will be difficult to evaluate and
communicate. One approach to diverse multivariate data that has been used successfully for
weather forecasting is the decision index, with examples such as the smog index, the pollen
count, and the mold index commonly used to assist in public and personal decisions about
environmental exposure. A similar approach can be taken for cumulative risk assessment
(Hertzberg^2000).
The advantage of a decision index is the simplicity in converting highly multivariate
technical information into a single number. The most common example used for cumulative
health risk is the hazard index for mixture risk (see box on next page). Although specific for a
single affected target organ, each hazard index reflects multiple studies of multiple chemicals,
often involving multiple test animal species and test exposures and highly varied measures of
toxicity.
48
-------
The Hazard Index
The hazard index (HI) for oral exposure is
implemented by Superfund assessors by the formula:
ffl = sum[ HQ ] = sum[ E/Rfp, ]
where Ej an(j RfDj are the daily exposure and
reference dose of chemical j.
The RfD is itself a kind of decision index in that it
reflects a dose that is selected to be sufficiently low
that any toxic effects are judged highly unlikely. All
available dose-response data on all effects are
considered in determining each RfD. Uncertainties
in the RfD will differ across the chemicals, making
the uncertainty in HI difficult to characterize.
The main disadvantage of a simple
index is that the uncertainties in its : ,
calculation are largely hidden. Another key
disadvantage is in quantifying what are often
scientific judgments. For example, the ;
hazard index implemented under Superfund
(USEPA, 1989a) is a number whose decision
threshold is usually given as 1.0, so thatjwhen
the hazard index is greater than 1, additional
action is indicated. The actual value of a
hazard index is not that informative: a value
of 6 is not necessarily twice as bad as a value
of 3. This is partly due to the uncertainty
factors necessary to develop the reference
dose (RfD) or the reference concentration
(RfC). The total value of these factors can be
as low as 3 or as high as 3000, depending on
the data upon which the RfD or RfC for a specific chemical is based (Barnes et al., 1988; Beck et
al., 1993; USEPA, 1994; Dourson et al.,'1996; USEPA, 2002e).
i
One alternative for addressing multiple effects is to recast these qualitative judgments in
terms of severity categories or levels of concern (e.g., high/medium/low) and then use statistical
methods such as categorical regression that use only the ordering of the severity scores but not
their actual values. The result is not a risk of a particular toxic effect but rather a risk of
exceeding a certain minimum toxic severity level or level of minimal concern (Hertzberg, 1989;
Guth et al., 1993). In the best situations,; such as the EPA interaction-based hazard index
(USEPA, 2000e), the decision index formula is modular, so component pieces can be evaluated
separately for accuracy and improvements in one area can be easily incorporated to give an
improved index. !
Another example of a decision index with more overt display of its diverse parts is the
Hazard Ranking System (HRS) (47 Fed. Reg. 31219, dated July 16, 1982, and amended 55 Fed.
Reg. 51532, dated December 14, 1990), a formula developed for characterizing the relative
hazards of a particular waste site. These jhazards were highly diverse and include corrosivity,
explosivity, toxicity, and soil conditions.. As with the hazard index, different uncertainties in the
components make the uncertainty of the HRS index difficult to describe. Instead of merely
presenting the index as a number, a graphical presentation such as the star plots of multivariate
data could be used (Chambers et al., 1983; Hertzberg, 2000), where each arm of the star
represents one of the sub-indices. Although this approach shows the relative contribution of
each factor, it again hides the uncertainties of the factors as well as of the HRS index itself.
Hybrid methods that combine judgment with numerical descriptions of risk or dose-
response have also been used for complex risk assessments. The EPA interaction-based hazard
index (USEPA, 2000e) and the mixture risk approaches of the Agency for Toxic Substances and
Disease Registry (ATSDR) (Hansen, et at, 1998) both include a judgmental weight-of-evidence
i 49
-------
(WOE) score to reflect the strength of evidence for toxicologic interactions and relevance to
human health risk. The ATSDR WOE is used in communicating risks and intervention options,
whereas the EPA WOE is used to calculate a modified hazard index. A slightly different
approach is the Integral Search System database program for combinations of carcinogens (Woo
et al., 1994) by which available studies on pairwise interactions of carcinogenicity are used to
modify the risk range of the combination from that predicted by response addition (USEPA,
2000e). In all these cases, scientific judgment is used to alter the risk description or quantitative
estimate, but only in terms of an approximate risk interval or a decision threshold.
3.2.2.4. Probabilistic approaches
The recent report by Bogen (2001) illustrates an alternative probabilistic approach to
noncancer endpoints in which methods used for integrated quantitative treatment of uncertainty
and variability are made consistent with those used for probabilistic assessment of cancer risk.
This report addresses many issues concerning the implementation of probabilistic methods for
noncancer endpoints and cites a number of related references (e.g., Lewis, 1993; Dourson et al.,
1997; Slob and Pieters, 1998).
Any approach to cumulative risk assessment should carefully define the set of relevant
endpoints. Precisely how this is done has important logical and practical implications for how
the cumulative risk may be calculated and interpreted. For example, the risk of inducing a given
endpoint may differ among different people in a population at risk for some endpoints, (e.g.,
cancer conditional on all carcinogen exposures) but may be unaffected by interindividual
variability (e.g., in exposure or susceptibility) for other endpoints (such as ecological or aesthetic
effects). Defining the latter risks in terms of individual risk per se will thus complicate
calculating cumulative risk if a probabilistic approach to cumulative risk assessment is used and
perhaps if other approaches are used as well.
In contrast, the probabilistic approach to cumulative risk assessment may be facilitated
by defining the risk of a given endpoint in terms of population risk, that is, in terms of the
predicted number of cases of that endpoint. Alternatively (or additionally), similar simplification
can be achieved for all heterogeneous endpoints by defining the risk only with respect to those
persons in the population at risk who are reasonably maximally exposed (e.g., individuals
adjacent to a proposed source) or to those persons who will incur the greatest increased risk (e.g.,
persons at a vulnerable life stage, such as children, or other members of a sensitive
subpopulation who might be located adjacent to a proposed source).
3.3. Areas of Complexity and Current Research
One reason for the somewhat limited availability of cumulative risk assessments may be
the accompanying complexity that arises in various aspects of the assessment. Some of this
complexity is discussed in the previous section, along with currently available methods specific
to human health risk assessment. In this section, some areas where research is ongoing are
discussed, and some existing methods for quantitatively assessing multiple types of risk or
hazard using a single metric are described.
50
-------
3.3.1. Interactions Between Stressors and Other Factors
In identifying and characterizing susceptible subpopulations, it may be important to
consider a variety of factors, such as current physical and mental health status and past exposure
histories, that may exacerbate the effects of the stressors of interest. Social factors such as
community property values, source of income, level of income, and standard of living may also
affect vulnerability of subpopulations to certain other stressors. Risks associated with chemical
or biological stressors may be significantly affected by "vulnerability factors" such as lack of
health care or genetic predisposition to some diseases and effects. Community traditions and
beliefs may affect activity patterns and behaviors and therefore affect exposure to stressors as
well as the acceptability of the risk management options. Depending on the scope of the
assessment and the stressors included, lifestyle factors such as smoking and nutritional habits,
among others, may be important to susceptibility.
I
In what could be characterized as an exploration of how somewhat abstract factors may
affect susceptibility, ATSDR held an expert panel workshop on the subject of psychological
responses to hazardous substances (ATSDR, 1995). In its report, the panel noted that there is "a
significant lack of information" about how often communities near hazardous waste sites or
spills suffer chronic stress reactions, but that psychological stress causes both psychological
changes that can be measured by self-reports and objective tests as well as physical changes such
as increased blood pressure and heart rate and biochemical parameters such as changes in stress
hormones. Assessing the levels of stress and their potential contribution to risk is difficult for a
variety of reasons. The report notes that
i
unlike the damage and injuries caused by a natural disaster, many toxic
substances are invisible to the senses.... In the face of no external cues and
uncertain circumstances, each person affected by a hazardous exposure develops
their own beliefs about the nature of the resultant harm. These beliefs are based
on the facts available to them, prfe-existing opinions, cultural factors, sensory
cues, and the beliefs of leaders and others in the community.... Unlike a natural
disaster, which hits and has a low point after which recovery can begin, the
response to a hazardous waste site can take 12 to 20 years.
Although the ATSDR report indicates that stress related to hazardous chemicals in the
community can show measurable physical effects, it stopped short of saying that long-term
health effects from this stress can be converted to risk estimates at this time. One of the
questions the panel was asked to address' was, "Given what is known regarding the psychology
of stress, are there interactions between chronic stress and exposure to neurotoxicants that could
shift the dose-response curve for neurotoxins?" The panel concluded:
A methodology does not exist that would allow for discrimination between stress
or neurotoxicant-mediated effects in community-based studies.... Experimental
animal data exist to suggest that stress levels can modulate a toxic response;
however, the question of specificity remains. Given that stress can induce or
unmask a latent effect of a toxicant, there is the possibility that chronic stress
51
-------
could alter basal levels of neurofunctioning and shift the threshold for
neurotoxicity. Indeed, one may find a shift in the dose response to a
neurotoxicant; however, a specific effect of the neurotoxicant should be examined
in greater detail than the generalized non-specific endpoints. Detecting such a
shift would require the knowledge of toxicant-specific biological mechanisms of
actions, which most often are not known.
The ATSDR report contains many suggestions for research to fill data gaps in this area,
and scientists may make significant progress in the coming years.
"Quality-of-life" issues may also influence risk to health or the environment, and
evaluating those issues may require an approach that differs from the traditional NRC risk
paradigm. Although a cumulative human health or ecological health risk assessment is not a
cumulative impact analysis such as is conducted under NEPA, changes in quality-of-life factors
may affect the vulnerability of a population to health or ecological risks and consequently may
be part of the considerations in a cumulative risk assessment. Because few, if any, established
and accepted relationships are currently available to quantitatively link quality-of-life factors and
health or ecological risk, this further research in this area may prove valuable.
To evaluate the effects on human or ecological health from these types of stressors, a
more deliberative approach (in the analytical-deliberative process) is needed than is used in, say,
cancer risk analysis. EPA (USEPA, 1993b) suggests a six-step process that may help
characterize quality-of-life factors, some of which may be relevant to -the assessment (e.g., in
considering population susceptibility). An example of a set of quality-of-life criteria developed
by the State of Vermont's Agency of Natural Resources is provided in Appendix F; however, it
should be noted that quality-of-life issues can encompass much more than the criteria mentioned
in this example. Some human health or ecological cumulative risk assessments may consider
quality-of-life factors as having a role in susceptibility to the stressors being assessed.
3.3.2. The Promise of Biomarkers and Biomonitoring
There are a variety of measures that are inherently cumulative. These include biomarkers
(they give the full effect or full exposure, regardless of source) and measures of the incidence
and prevalence of disease in a community. The latter give an indication of the total effect of
multiple sources of exposure. In light of our understanding of the multifactorial basis of disease,
a public health approach that says "regardless of the cause, a community has x level of disease"
can be informative. Such statistics can be compared across geographical areas that have
different sources or groups that have different levels of vulnerability. The approach is based
strongly in the field of epidemiology. Indeed, the most often-heard critique of
epidemiology—that it is the prevalence or incidence of disease documented as a function of the
combined effect of many exposures (over time and/or space)—is exactly what makes it so well
suited for cumulative risk assessment. It is likely that epidemiological concepts will figure
prominently in cumulative risk assessment, both in identifying the underlying vulnerability of a
population and in generating hypotheses regarding the determination of relative contributions of
multiple stressors (IPCS, 1983).
52
-------
Sources of data include cross-sectional analyses that determine prevalence levels as well
as basic surveillance techniques. With respect to the latter, The Pew Environmental Health
Commission (http://pewenvirohealth.jhsph.edu/html/home/home.html) has recently completed a
series of reports that document the extent of national- and state-level resources for chronic
disease surveillance. Reports focus on the type of surveillance systems needed as well as the
status of registries for birth defects and asthma. Health Track (http://health-track.org/ and
http://healthyamericans.org/) is the outgrowth of that research; it is devoted to tracking and
monitoring chronic disease to help communities to identify patterns of health problems.
Like epidemiologic data, some biomarkers reflect the cumulative history of individuals
and populations. The use of biomarkers, is based on the concept that the biological unit (organ,
body, etc.) can be an effective and accurate element for integrating the aggregate exposures or
doses or cumulative risks. Using biological measurements—biomarkers—to determine prior
exposures (biomarkers of exposure) or the current health status of individuals (biomarkers of
effect) holds some promise for cumulative risk assessments of the future (IPCS, 1993, 2001).
Use of biomarkers for a group of chemicals or stressors that act upon individuals in the same
way can give the assessor a picture of where an individual currently falls on the continuum from
exposure to effects, making it much easier to predict risks if additional exposure occurs.
A few biomarkers (or even a single one) can possibly represent exposure to a suite of
chemicals. Although this approach reduces the analytical burden and simplifies the process of
estimating cumulative risk, it loses some of the advantages of single-chemical assessment
(especially being able to quickly discern the importance of different pathways and routes of
exposure contributing to the risk).
Biomarkers have a number of advantages; one disadvantage, however, is that they
generally cannot link an effect to any particular exposure. For example, information on the
cumulative risks in a local population of a group of chemicals that are toxic to the liver might be
provided by selective liver function tests, but causal inferences would have to take account of
many other factors that may affect liver function. Likewise, body burden data for chlorinated
dioxins and related compounds may show that exposure has occurred, but assumptions would
need to be made as to the pathways, route, and timing of exposures and scenarios developed for
future exposures if risks are to be estimated. For a regulatory agency such as EPA, a decision to
act to reduce risk often depends on separating contributions from exposure pathways so that
effective policies can be determined. '.
One of the benefits of the biomarker approach—the development of data that show the
actual current exposure and risk status of a population—is also its major impediment: it can
require extensive (or for humans, possibly invasive) monitoring. Monitoring data can be not
only costly, but also difficult to obtain. This approach uses primarily measurement methods; it
can also be used to develop statements of probability of adverse effects of additional incremental
exposures. This approach holds great promise for simplifying cumulative risk assessments, but
few methods exist at this time for such applications. Although this framework report provides
53
-------
only a cursory discussion of the biomarker approach, it is hoped that the planned guidelines for
cumulative risk assessment will discuss this approach in greater detail.
3.3.3. A Single Metric for Multiple Types of Hazard
The most complex cumulative risk assessments will evaluate both multiple exposures
(potentially, multiple sources, stressors, pathways, and durations) and multiple effects. Ideally
this evaluation would provide projections regarding the potential for a particular complex
exposure to cause particular effects to different physiological systems and also provide an
integration of these projections into a qualitative characterization of overall potential impact to
human health. Some applications have attempted to integrate the potential impacts across the
different physiological systems. Approaches vary from treating the assessment as a number of
multistressor, single-effect assessments, where the risks are combined only at the final step, to
assessments that are more integrated throughout all the steps in the assessment process.
For example, cumulative ecological risk assessments conducted in the Columbia River
Basin and the Chesapeake Bay (Barnthouse et al., 2000) focused on a number of observed
adverse conditions, and then determined from among all the possible stressors, which particular
combination was most influential in creating those conditions. Stressors such as over harvesting
of natural resources; modification of natural hydrology; land use change; point-source and
nonpoint-source pollution, including toxic chemicals; and the presence of exotic species were
analyzed, with the goal of designing effective restoration strategies to eliminate or ameliorate the
conditions.
If it is considered desirable for the assessment, an important activity may be to determine
how (if at all possible) to combine risks from different effects—or the even-more-problematic
disparate measures of risk—and present them in an integrated manner. Depending on the
purpose and risk management objectives (see Section 2.1.1), some cumulative risk assessments
may employ some sort of single, common metric to describe overall risk.
One—but certainly not the only—approach to simplifying this problem is to collapse this
"n-dimensional matrix" of hazards and risks into a few or even a single measure (Murray, 1994).
However, this requires converting the various measures of risk into a common metric or
otherwise translating them into a common scale or index. Some methods for combining
disparate measures of risk are briefly described below.
54
-------
3.3.3.1. Creating a common metric
As discussed earlier in this chapter, there are several different theoretical approaches to
cumulative risk assessment. Some of triem require synthesizing a risk estimate (or risk
indication) by "adding up" risks from different parts of the risk picture. Actual mathematical
addition, of course, requires a "commoiji denominator," or a common metric. Frequently used
common metrics are risk, money, time, land effort. Finding a common metric for dissimilar risks
(e.g., cancer vs. noncancer, human vs. ecological) is not strictly an analytic process, because
some judgments should be made as to how to link two or more separate scales of risks. These
judgments often involve subjective values, and because of this, it is a deliberative process.
EPA's Office of Pollution Prevention and Toxics has released a CD-ROM (USEPA,
1999i)14 that shows an example of combining different effects into a common metric and the
consequent judgment needed to achieve a common metric. In this model, emissions for both
carcinogens and noncarcinogens are weighted by a toxicity factor so that they can be combined
in a risk-based screening "score" for a particular geographic area. The scale for this weight for
carcinogens is related to the unit risk fadtor, and the weight for the noncarcinogens is based on
the RfD. According to the authors, it isipossible to link these two different scales by making a
deliberative judgment or assumption as to their relationship. They note that in their case, "when
combining cancer and noncancer endpoints, it is assumed that exposure at the RfD is equivalent
to a 2.5 x 10'4 cancer risk" (Bouwes and Hassur, 1998; USEPA, 1998h).
Obviously, as Bouwes and Hassur acknowledge, equating an HQ15 value of 1.0 (i.e.,
exposure is at the RfD) with a cancer risk of 2.5 x 10'4 is a judgment that is outside the strictly
analytic part of an assessment; the equating of the two points in the respective scales represents a
value judgment and as such can be debated. Therefore, this particular part of the assessment is
deliberative in nature. In making these types of judgments in a risk assessment, some care
should be taken not to lose information in the aggregation, especially if all stakeholders do not
agree on the relative tradeoffs necessary to arrive at the common scale of risk. If there is
disagreement on constructing the scale, pr even if more clarity is desired in the final report, the
disaggregated risks can also be presented. Equity issues may also arise here, making it necessary
to break out risks into relative burdens fpr different subpopulation.
In most cases, construction of a single scale for different types of endpoints will involve
comparative risk, a field where different types of risks or endpoints are ranked, compared,
weighted, or converted to a scale on the basis of the judgments and values of the persons doing
the assessments (USEPA, 1990b, 1993b, 1998f, 1999J). Groups of stakeholders such as are
14 As of this writing, version 2.0 is in beta test. Details are available at
.
15 A hazard quotient, or HQ, in this context
chemical divided by the RfD (or RfC) for that.chemical
generally expected to be without effect during a
is the estimated exposure or dose level for a given individual
cal. Values of less than 1.0 for HQ indicate levels that are
lifetime.
55
-------
gathered for cumulative risk assessment can provide ranking of various effects in terms of
importance even if the effects cannot be put on a single scale or metric. This information may
subsequently be used by decisionmakers for dealing with "worst risks first."
There have been some attempts to allow for transparent and quantitative incorporation of
values into a common metric. One example flows from the suggestion that "time is the unit of
measure for the burden of disease," whether the disease results in disability or premature
mortality (Murray, 1994). On this premise, economic analyses of the costs and benefits of
disease intervention strategies have used quality-adjusted life years (QALYs) and disability-
adjusted life years (DALYs) as the metrics for the adverse effects of disease. These metrics are
intended to reflect the years of life spent in disease states (considering the variation in severity of
effects) and the years of life lost due to premature mortality resulting from disease as a surrogate
measure for risk from a variety of different types of effect.
Even if this conversion of effects into QALYs or DALYs is successful, for diseases that
result in periods of morbidity and disability (but not death), weighting factors (based on
judgments) are used to equate time spent in various disease states with years lost to mortality. In
this way, dissimilar adverse effects can be combined to provide a single measure of disease
burden. However, it should be noted that aggregation of effects in this manner obscures the
meaning of the final measure. QALYs and DALYs do not represent an actual shortening of the
life span but are indicators of the overall degradation of well-being that results from various
disease states.
Experience with applying such measures as QALYs and DALYs to environmental risk
problems is extremely limited. Some very early methods development work has been initiated
that explores the use of QALYs for combining microbial and disinfection by-product risks
(USEPA, 1998f). However, some concerns have been raised about the adequacy of such
measures', especially when integrated with economic information for decision making (USEPA,
2000g). Further methods development work is needed to improve the usefulness of QALYs and
DALYs for environmental risk assessments, especially with respect to the incorporation of
uncertainty (USEPA, 1999J).
Categorical regression may provide another tool for combining disparate effects using a
common metric. In this approach, adverse effects are assigned to severity categories (again, a
judgment making the process deliberative) and the ordered categories are regressed against
increasing dose (Teuschler et al, 1999). This use of categorical regression puts definite limits on
the interpretation of the results. Because the toxicities are only represented by categories and
judgment is used to place the observed response into a severity category, the results are rather
coarse. But because the analysis is almost totally empirical—that is, no low-dose extrapolation
is required—the results can still be quite useful.
EPA has also used decision indices (see Section 3.2.2.3) that are based on dissimilar
measures, and although they do not produce risk estimates, they can still prove useful. The
approach involves developing a composite score—or index—from measures of various risk
dimensions. Various environmental risk indices have been developed and applied to ranking and
56
-------
comparative analyses. Often, these indices use surrogate measures for risk rather than actual
calculations of the probability of adverse effects. One such index is the HRS, which is used to
place uncontrolled waste sites on the National Priorities List for Superfund. This index is based
on the likelihood of off-site movement of waste, the toxicity of the waste, and the people and
sensitive environments that may be affected. It also uses corrosivity, toxicity, fire hazard, and
other factors, which are scored and combined into one numerical indicator of overall hazard
potential. Such an approach for a composite index has been suggested for the communication of
cumulative risk (Hertzberg, 2000). |
Fischhoff et al. (1984) provides an example of this approach as applied to the evaluation
of energy technologies. In this case, disparate risks are assigned a score from a fixed scale (e.g.,
from 0, representing no risk, to 100, representing the worst risk for that dimension). The scores
are then weighted to reflect value judgments about the importance of the various risk
dimensions, and the composite score is calculated by summing the individual weighted scores.
Again, the aggregation of dissimilar adverse effects obscures the meaning of the final score,
making this approach more appropriate for ranking and comparative analyses.
Recently, EPA has been working on several index-based approaches to dealing with
cumulative risk issues. EPA Region 3 ahd the Office of Research and Development have been
jointly working to develop a potential risk indexing system (USEPA, 1993c, 1995d, 1997c).
This index also uses a vulnerability index, and it gauges the overall well-being of a locale and
various subpopulations. Again, the volume and toxicity of released stressors serve as surrogate
measures of risk in developing this index.
Combining the diverse effects and risk using either common metrics or indices has both
pros and cons. A weakness of the index approach is that, by aggregating dissimilar information,
information is "lost," and the meaning of the final score can be obscured. However, both
approaches have one strength in common: the ability to incorporate social values into the risk
assessment in an explicit and quantitative manner. For example, in the derivation of DALYs,
weights can be used to reflect the different social roles people play as they age (Murray, 1994).
In the composite scores developed by Fischhoff et al. (1984), public concern was incorporated as
an adverse effect. The ability to incorporate issues suach as public concern into the composite
scores is an important feature for methods that will be applied to cumulative risk assessments,
especially for communities. Given that pumulative assessments have a community/population
focus, the ability to incorporate social values into an overall assessment of well-being will be
critical. \
3.3.3.2. General issues regarding a single metric
As described above, each approach to portraying the results of a cumulative risk
assessment has desirable and undesirably features. Although common metrics and indices can
incorporate social values in an explicit and quantitative manner, the meaning of the final
measure can be obscured by the aggregation of dissimilar effects. The abstract nature of the final
measure could lead to difficulties when communicating the results of the assessment to the
public. The use of graphical and mapping techniques do not necessarily overcome
; 57 •
-------
communication problems. Although these techniques do not have some of the problems
associated with the mathematical aggregation of dissimilar effects, it still may be difficult, for
example, to accurately describe the information that a graphic is intended to convey.
Because we have relatively little experience in combining different types of risk, a key
issue is the need for methods development in this area. The approaches described above indicate
a beginning. Additional exploratory work is needed, however, to further develop existing
methods and to find additional methods that are flexible, that can incorporate social values, that
are easy to communicate, and that provide an integrated portrayal of the overall well-being of a
community and its various subpopulations.
3.3.4. Qualitative Approaches
There will be cases where cumulative risk cannot be quantified in any meaningful or
reliable way. Qualitative approaches can be valuable for cumulative risk assessment and, in the
near term, they may be the only practical way to address many of the complexities involved.
Qualitative approaches may be used as a way to overcome the complexity and data deficiencies
that hinder quantitative approaches. In many assessments, risk may not be a quantifiable
variable.
For these cases, there may be qualitative approaches that provide some insight. Broad
indicators related to exposure in complex ways (e.g., production volumes, emissions
inventories, environmental concentrations, etc.) and indicators of toxicity can be communicated
using geographic information systems. Displaying complex, multidimensional matrices on a
map can help in visualizing locations of areas with multiple stressors. Furthermore,
geographically based measures of hazard are potentially useful cumulative measures; although
they do not provide information on the risks, the locations of hazards can be used as an indicator
of aggregate exposures and, thus, cumulative risks from all of the potential chemicals associated
with that site. The environmental justice literature has used this approach.
Quantitative results might eventually be reduced to a more qualitative scale (high,
medium, or low), or the qualitative results could provide "comments" tacked onto the
quantitative results. The assessment might simply raise red flags associated with specific issues
(e.g. density of emitters in a community, presence of minority populations, special exposure
pathways, etc); a high number of such flags would indicate unacceptable cumulative risk, even if
the risk is not quantified. This approach has been used in the European Union (CEU, 1996), and
its experience in using qualitative methods for permitting suggests that "qualitative" is not
"irrational." Other relevant tools include expert judgment techniques, focus groups, opinion
surveys, citizen juries, and alternative dispute resolution.
58
-------
4. THE RISK CHARACTERIZATION PHASE
The last phase of cumulative risk assessment, risk characterization, integrates and
interprets the results of the analysis phase and addresses the problem(s) formulated in the
planning and scoping phase (Figure 4-1). It should describe the qualitative and/or quantitative
risk assessment results; list the important assumptions, limitations, and uncertainties associated
•with those results; and discuss the ultimate use of the analytic-deliberative outcomes. Given the
complexity of cumulative risk issues and the need for clarity and transparency in risk
characterization, such "full disclosure" presents a major communication challenge..
There is a substantial analytical component of the risk characterization phase, but there is
also a considerable need for deliberation!. At a minimum, stakeholders in this phase should (1)
understand the outcome of the cumulative risk assessment, (2) ask questions about how best to
frame the interpretation, and (3) confirm!tnat tne cumulative risk assessment met the goals set in
the problem formulation, or if not, why hot. As in the previous phase, the stakeholders' role is
only limited by what is proposed and agreed upon in the particular case being assessed.
Risk estimation in a cumulative risk assessment will involve some combination of risks,
whether the risks from different stressors cause similar effects or different types of effects. The
stressors themselves may be similar or Widely different. Combinations of many types of
stressors that have different endpoints will quickly cause the risk estimation step to become very
complex and difficult. j
Because of its potential complexity, and because in some cases the cumulative risk
assessment will be dealing methodologically with "uncharted territory," it is very important that
the planning, conduct, analysis, and characterization of an assessment be transparent. As stated
by the Office of Management and Budget (OMB, 2002), the "benefit of transparency is that the
public will be able to assess how much ah agency's analytic result hinges on the specific analytic
choices made by the, agency." The process, methodology, data, assumptions, and selection
among alternate interpretations should be very carefully documented and very clearly stated.
This is noted again in the next section. t
59
-------
Planning, Scoping, and
Problem Formulation
Analysis
iterpretation and
RtelTCfiafacterization
Risk Description
- Central Tendency and High-End
Individual Risk
- Population Risk
- Risk to Important Subpopulations
Uncertainty Analysis
-Being Explicit about Uncertainty
- Uncertainty and Variability
- Uncertainty and Risk Addition
- Sensitivity Analysis
\L
Information Provided by Cumulative Risk Assessment
Using the Results of Cumulative Risk Assessment
Figure 4-1. The Risk Characterization phase.
60
-------
4.1. Risk Description
The ultimate product in the risk assessment process is the risk characterization, in which
the information from all the steps is integrated and an overall conclusion about risk that is
complete, informative, and useful for decisionmakers is synthesized. The nature of the risk
characterization will depend on the information available, the regulatory application of the risk
information, and the resources available (including time). It is important to identify and discuss
all major issues associated with determining
the nature and extent of the risk. Further,
USEPA (1995a) specifies that a risk
characterization "be prepared in a manner
that is clear, transparent, reasonable, and
consistent with other risk characterizations of
similar scope prepared across programs in
the Agency." In short, estimates of health
risk are to be presented in the context of
uncertainties and limitations in the data and
methodology.
Risk Characterization Guiding Principles
Regarding information content and uncertainty aspects:
*• The risk characterization integrates the information
from the exposure and dose-response assessments
using a combination of qualitative information,
quantitative information, and information regarding
uncertainties. ' |
> The risk characterization includes a discussiqn of •
uncertainly and variability. !
>• Well-balanced risk characterizations present risk
conclusions and information regarding the strengths
and limitations of the .assessment for other risk
assessors, EPA decisionmakers, and the publfc.
Regarding risk descriptors: '
* Information about the distribution of individual
exposures is important to communicating the;
results of a risk assessment.
> Information about population exposure leads 'to
another important way to describe risk.
*• Information about the distribution of exposurfe and
risk for different subgroups of the population are
important components of a risk assessment, j
* Situation-specific information adds perspective on
possible future events or regulatory options. '.
>• An evaluation of the uncertainty in the risk Jl
descriptors is an important component of the i
uncertainty discussion in the assessment.
Source: USEPA, 1995b |
USEPA (1995b) lists several guiding
principles for defining risk characterization
in the context of risk assessment (see text
box), both with respect to information
content and uncertainty aspects and with
respect to descriptions of risk. EPA has also
published a handbook on risk
characterization (USEPA, 2000c).
Risk assessments are intended to
address or provide descriptions of risk to one
or more of the following: (1) people exposed
at average levels and people in the high-end
portions of the risk distribution,
(2) the exposed population as a whole, and
(3) important subgroups or life stage strata of
the population (e.g., children) or other highly
susceptible groups or individuals, if known.
Risk predictions for sensitive subpopulations
are a subset of population risks. Sensitive
subpopulations consist of a specific set of
individuals who are particularly susceptible
to adverse health effects because of
physiological (e.g., age, gender, pre-existing
61
-------
Some Thoughts on Risk Characterization
Understanding Risk (NRC, 1996) focuses on risk
characterization and reaches the following
conclusions:
1. Risk characterization should be a- decision-driven
activity, directed towards informing choices and
solving problems. The view of risk characterization
as a translation or summary is seriously deficient...
Risk characterization should not be an activity added
at the end of risk analysis; rather, its needs should
largely determine the scope and nature of risk
analysis.
2. Coping with a risk situation requires a broad
understanding of the relevant losses, harms, or
consequences to the interested and affected parties. ,
A risk characterization must address what the
interested and affected parties believe to be at risk in
the particular situation, and it must incorporate their
perspectives and specialized knowledge.
3. Risk characterization is the outcome of an analytic-
deliberative process.... Analysis and deliberation can
be thought of as two complementary approaches to
gaining knowledge about the world, forming
understandings on the basis of knowledge, and
reaching agreement among people.
4. The analytic-deliberative process leading to a risk
characterization should include early and explicit
attention to problem formulation.
5. The analytic-deliberative process should be
mutual and recursive,... A recurring criticism of risk
characterization is that the underlying analysis failed
to pay adequate attention to questions of central
concern to some of the interested and affected parties.
This is not so much a failure of analysis as a failure to
integrate it with broadly based deliberation: the
analysis was not framed by adequate understanding
about what should be analyzed.... Structuring an
effective analytic-deliberative process for informing a
risk decision is not a matter for a recipe. Every step
involves judgment, and the right choices are situation
dependent. Still, it is possible to identify objectives
that also serve as criteria for judging success:
Getting the science right. The underlying analysis
meets high scientific standards in terms of
measurement, analytic methods, databases used,
plausibility of assumptions, and respectfulness of
both the magnitude and character of uncertainty....
Getting the right science. The analysis has addressed
the significant risk-related concerns of public officials
and the spectrum of interested and affected parties,
such as risks to health, economic well-being, and • .
ecological and social values, with analytic priorities
having been set so as to emphasize the issues most
relevant to the decision.
Getting the right participation. The analytic-
deliberative process has had sufficiently broad
participation to ensure that the important, decision-
relevant information enters the process, that all
important perspectives are considered, and that the
parties' legitimate concerns about inclusiveness and
openness are met.
Getting the participation right. The analytic-
deliberative process satisfies the decision makers and
interested and affected parties that it is responsive to
their needs: that their information, viewpoints, and
concerns have been adequately represented and taken
into account; that they have been adequately consulted;
and that their participation has been able to affect the
way risk problems are defined and understood.
Developing an accurate, balanced, and informative
synthesis. The risk characterization presents the state
of knowledge, uncertainty, and disagreement about the
risk situation to reflect the range of relevant knowledge
and perspectives and satisfies the parties to a decision
that they have been adequately informed within the
limits of available knowledge.
6. Those responsible for a risk characterization should
begin by developing a diagnosis of the decision situation
so that they can better match the analytic-deliberative
process leading to the characterization to the needs of
the decision, particularly in terms of level and intensity
of effort and presentation of parties.... Diagnosis of risk
decision situations should follow eight steps: (1)
diagnose the kinds of risk and the state of knowledge,
(2) describe the legal mandate, (3) describe the purpose
of the risk decision, (4) describe the affected parties
and anticipate public reactions, (5) estimate resource
needs and timetable, (6) plan for organizational needs,
(7) develop a preliminary process design, and (8)
summarize and discuss the diagnosis with the
responsible organization.
-------
r
conditions), socioeconomic (e.g., nutrition), or demographic variables or because of significantly
greater levels of exposure (USEPA, 1992a). Subpopulations can be defined using age, race,
gender, and other factors. If enough information is available, a quantitative risk estimate for a
subpopulation can be developed; if not, then any qualitative information about subpopulations
gathered during hazard identification should be summarized as part of the risk characterization
(USEPA, 2000c). The box on the previous page summarizes some of the points made in
Understanding Risk (NRC, 1996), which devotes a great deal of discussion to risk
characterization. Risk characterization ijs most efficiently conducted with early and continued
attention to the risk characterization step in the risk assessment process (NRC, 1996- USEPA
2000c).
4.2. Uncertainty Analysis
Morgan and Henrion (1990) note that, historically, the most common approach to
uncertainty in policy analysis (including in risk assessment) has been to ignore it.. In a section
titled "Why Consider Uncertainty?" they advance three primary reasons, all of which are
especially relevant to an analytic-deliberative process such as cumulative risk assessment. They
suggest that it is important to worry about uncertainty
r
• "when one is performing an analysis in which people's attitude toward risk is likely
to be important, for example, when people display significant risk aversion;
• "when one is performing an analysis in which uncertain information from different
sources must be combined. The precision of each source should help determine its
weighting in the combination; and
• "when a decision must be made about whether to expend resources to acquire
additional information. In general, the greater the uncertainty, the greater the
expected value of additional information."
Morgan and Henrion provide
ten requirements for good policy
analysis, and although all are
commendable and several have been
discussed elsewhere in this framework
report, we should look more closely at
numbers 6-8 in the box at right for
some insight into uncertainty analysis.
There are many resources
available that talk in detail about how
to perform uncertainty analysis (e.g.,
USEPA, 1997b; Morgan and Henrion,
1990). Although detailed instruction is
beyond the scope of this framework
report, we believe that a discussion of
some general principles is in order.
Morgan and Henrion's Ten Requirements
for Good Policy Analysis
1." Do your homework with literature, experts, and users.
2. Let the problem drive the analysis.
3. Make the analysis as simple as possible, but no simpler.
4. Identify all significant assumptions.
5. Be explicit about decision criteria and policy strategies.
6. Be explicit about uncertainties.
7. Perform systematic sensitivity and uncertainty analysis.
8. Iteratively refine the problem statement and the analysis.
9. Document clearly and completely.
10. Expose the work to peer review.
Source: Morgan and Henrion, 1990
63
-------
4.2.1. Assumptions in the Assessment
Cumulative risk assessment will typically be used in a decision-making process to help
inform the decisionmaker(s). For this reason, it is important that the decisionmakers be made
explicitly aware of any assumptions that may significantly affect the conclusions of the analysis
(requirement number 4 in the box. on previous page). Morgan and Henrion (1990) suggest that
these assumptions include:
• the main policy concerns, issues, or decisions that prompted the assessment,
• the evaluation criteria to be used to define issues of concern or options,
• the scope and boundaries of the assessment and ways in which alternate selections
might influence the conclusions reached,
• soft or intangible issues that are ignored or inadequately dealt with in the quantitative
analysis (e.g., intrinsic value of wilderness, equity of distribution of risks and
benefits), .
- approximations introduced by the level of aggregation or by level of detail in models,
• value judgments and tradeoffs, and
• the objective function used, including methods of combining ratings on multiple
criteria (or combining risk scales).
Identifying significant assumptions can often highlight "soft" uncertainties that are not
easily quantified and are therefore often left out of a quantitative uncertainty analysis.
Nevertheless, these "soft" uncertainties can often contribute more to the overall uncertainty of
the assessment than .the factors more easily quantified.
Morgan and Henrion's sixth requirement for good policy analysis (see box on previous
page) includes three types of uncertainty that analysts should explicitly address:
• Uncertainty about technical, scientific, economic, and political quantities (e.g.,
quantities such as rate constants often lend themselves to quantitative uncertainty
estimates relatively easily);
• Uncertainty about the appropriate functional form of technical, scientific, economic,
and political models (e.g., are the models used, such as dose-response models,
biologically sound?); and
• Disagreements among experts about the values of quantities or the functional torm ot
models (e.g., different health scientists using different forms of dose-response
models).
In requirement number 7, Morgan and Henrion suggest that an assessor should find out
which assumptions and uncertainties may significantly alter the conclusions, and that this
process can be conducted using sensitivity and uncertainty analysis. Techniques include:
• Deterministic, one-at-a-time analysis of each factor, holding all others constant at
nominal values;
64
-------
• Deterministic joint analysis, changing the values of more than one factor at a time;
• Parametric analysis, moving :one or a few inputs across reasonably selected ranges to
observe the shape of the response; and
• Probabilistic analysis, using correlation, rank correlation, regression, or other means
to examine how much of the uncertainty in the conclusions is attributable to which
inputs. j
Finally, Morgan and Henrion answer the question of why we should consider uncertainty
analysis with the following point. "Policy analysts have a professional and ethical responsibility
to present not just 'answers' but also a clear and explicit statement of the implications and
limitations of their work. Attempts to fully characterize and deal with important associated
uncertainties help them to execute this responsibility better."
!
4.2.2. Uncertainty and Variability
NRC (1994) notes a clear difference between uncertainty and variability and recommends
that the distinction between these two be maintained:
A distinction between uncertainty (i.e., degree of potential error) and inter-
individual variability (i.e., population heterogeneity) is generally required if the
resulting quantitative risk characterization is to be optimally useful for regulatory
. purposes, particularly insofar as risk
characterizations are treated
quantitatively. The distinction
between uncertainty and individual
variability ought to be maintained
rigorously at the level of separate risk-
assessment components (e.g., ambient
concentration, uptake, and potency) as
well as at the level of an integrated
risk characterization.
The Cumulative Exposure Project
EPA's Cumulative Exposure Project (CEP),' : • •
completed in 1998, modeled 1990 outdoor '
concentrations of hazardous air pollutants (HAjPs)
across the United States, which were combined with
unit risk estimates to estimate the potential increase
in excess cancer risk from multiple HAPs. The
cancer risks of different HAPs were assumed to be
additive and were summed across pollutants in: each
census tract to estimate a total cancer risk in each
census tract. :
Consideration of some specific uncertainties, i
including underestimation of ambient concentrations,
combining upper 95% confidence bound potency
estimates, and changes to potency estimates, found
that cancer risk may be underestimated by 15% or
overestimated by 40-50%. Other unanalyzed
uncertainties could make these under- or
overestimates larger.
Source: Woodruff et al., 2000
Variability and uncertainty have been
treated separately and distinctly in single-
chemical assessments such as the assessment
of TCE in ground water at Beale Air Force
Base in California (Bogen, 2001). The
treatment of variability and uncertainty will
also be an important issue in cumulative risk
assessments, although at the time of this
writing there are no good examples of an
elegant treatment of this issue for cumulative
risk.
65
-------
4.2.3. Uncertainty and Risk Addition
Calculating individual stressor risks and then combining them largely presents the same
challenges as combination toxicology but also adds some statistical stumbling blocks. Toxicity
addition, independence, synergism, or antagonism still need to be evaluated, but because risk
estimates for various stressors are often presented as values on the same numeric scale (e.g., as
cancer probabilities), cancer risks are often simply added together.
Because cancer slope factors are not "most probable estimates," but rather 95% upper
confidence levels, adding traditional risk levels can cause the resulting sum to overestimate a
95% upper confidence level risk for a mixture. There have been several recent papers discussing
this problem and how it may affect the resulting estimates. Kodell and Chen (1994) looked at
several binary mixtures and calculated that the summation of individual upper 95% confidence
intervals for chlorobenzene and hexachlorobenzene would overestimate the upper-bound risk of
a binary mixture of these compounds by 2 to 6%, whereas for chlorobenzene and TCE the
overestimate would be in the range of 12 to 15%. Seed et al. (1995) noted that, "in most cases,
the magnitude of the difference in cancer risk estimates calculated by [Kodell and Chen's]
various methods will be greatest for mixtures of equipotent compounds. However, even for
mixtures of equipotent compounds, the differences in joint risk estimated by summing the upper
95% confidence levels...are not great."
After analyzing four cases, Cogliano (1997) concluded that "as the number of risk
estimates increases, their sum becomes increasingly improbable, but not misleading." For
example, in adding 20 different cancer risk estimates based on a 95% upper bound, the resulting
sum of the upper bounds was no more than 2.2 times the true upper bound. Cogliano went on to
suggest that, for certain cases not involving synergistic or antagonistic interactions, "depending
on the number of carcinogens and the shape of the underlying risk distributions, division by a
factor of 2 can be sufficient to convert a sum of upper bounds into a plausible upper bound for
the overall risk."
The assumption of toxicologic independence (see Section 3.2.2) may not be a bad one if
other evidence supports it, but it should be addressed in the assessment if used (i.e., if risks are
added). Although some scientists believe that toxicologic interactions are of minor consequence
at concentrations observed in the environment (see discussion in USEPA, 2000e), the scientific
evidence for such an assumption has not been firmly established.
Notwithstanding the statistical limitations of adding traditional risk estimates and the
implicit assumption that the toxicities will be additive16 (i.e., no interactions such as synergism
or antagonism occur), the numerical ease for combining risks in this way may make it the most
16 At risk levels often seen with pollutant concentrations observed in the environment, the combined risks
calculated assuming "response additivity" (that is, each component acts as if the other were not present) are
approximately the same as with dose additivity (USEPA, 2000e).
66
-------
popular method for approximating cumulative risks in the short term, at least at a screening level
of assessment. , '•
4.2.4. Other Cumulative Risk Assessment Uncertainties
This framework does not, and cahnot, provide an exhaustive list of uncertainties unique
to cumulative risk assessment. Without!question, however, there will be uncertainties inherent in
a cumulative risk analysis that have not been as important in traditional assessments. As an
example, because cumulative risk assessments can be geographically based and GIS technology
seems to be a potentially useful tool for displaying results, there will be issues concerning how to
present uncertainty information, for example, by overlaying impacts or risks for several
chemicals, on a GIS map. j
Specific uncertainties can arise when adding doses for chemicals that operate by the same
mode of action, such as the organophosphorous (OP) pesticides. In this case, USEPA (2002c)
notes that uncertainties arise in estimating the RPFs of the OP pesticides. These RPF
uncertainties can be partitioned into three groups: those that are basic (e.g., uncertainty in the
dose-response relationship for the reference chemical), those that deal with chemicals in relation
to one another (relative potencies of other chemicals relative to the reference chemical), and
those concerning joint mode of action (e.g., members of the common mechanism group may
have other modes of action that are not fully captured via the common-mechanism potency
calculation). As risk assessors develop more experience with cumulative risk assessments, many
more of these uncertainties may arise, but it is not possible to foresee all of them.
i;
4.3. Information Provided by Cumulative Risk Assessment
It is important to clarify how cumulative risk assessment and this framework report relate
to community assessments and community decision making. Certainly, the Agency's risk
characterization handbook (USEPA, 2000c) emphasizes that whatever information is imparted
should be transparent, clear, consistent, and reasonable. For example, if it is known that the
results of a particular cumulative risk assessment will be severely limited because of a lack of
data or available methods, it may be advisable to start with a screening analysis to set priorities
for a subsequent study that is more detailed and focused. In simple terms, what can a cumulative
risk assessment tell us, and what can't it tell us?
4.3.1. Making Sense of Multiple-Stressor Effects
The information provided by cumulative risk assessment is only a portion of that needed
by communities and governments to make informed decisions about risks. There are almost
always a multitude of factors that affect health in a community (e.g., crime, drugs, health care
access, vehicle safety, climate, infectious disease, diet) that may not have been considered within
the scope of a given cumulative risk assessment. Community decision making will typically take
into account risks to the environment as well as consideration of historical and cultural values
and questions of fairness and distribution of risks. The methodology is not currently available to
understand how these factors (or stressors) may affect cumulative health risk.
67
-------
Additionally, benefits such as jobs and useful products or services that may be associated
with chemical or other stressor exposures may be important contexts for decisions on the risks
considered in cumulative risk assessments.
This framework report is not an attempt to lay out protocols to address all the risks or
considerations that are needed to adequately inform community decisions. Rather, its focus is on
describing various aspects of cumulative risk, whether or not the methods or data currently exist
to adequately analyze or evaluate those aspects of the assessment. It devotes considerable time
to a discussion on improving the methods for a single part on the broader picture: characterizing
health risks associated with exposures to multiple chemicals via multiple routes. Because of the
limitations of the current state of the science, cumulative risk assessments in the near future will
not be able to adequately answer all the questions posed by stakeholders or interested parties.
This does not mean, however, that they would not be useful in providing insights to some of the
questions asked; in fact, cumulative risk assessment may be the best tool available to address
certain questions dealing with multiple-stressor impacts.
4.3.2. Cumulative Risk Assessments in a Public Health Context
The public often asks—in a variety of ways—for clarification of the relationship between
environmental pollution (and risk assessments concerning it) and public health. Although
cumulative risk assessment holds the promise of better public health-related information for
communities, it is not a panacea. To draw relationships between environmental pollutant
exposures and disease incidence, a body of epidemiological study is necessary. Trying to "work
backwards" from health statistics to risk factors requires full knowledge of the risk factors
associated with the relevant disease(s). This is challenging under the best of circumstances, with
good data; many times it is not possible with the data at hand.
Health statistics, including death rates and incidence of various diseases, illustrate the
impact of a variety of risk factors (e.g., smoking as well as environmental pollutants) and risk
reduction factors (e.g., exercise and good nutrition as well as pollution control measures).
Indeed, population health statistics are reflective of all risk and risk reduction factors in a
population's history to date. Even the best cumulative risk assessment, given today's state of the
science, could not include an evaluation of the magnitude and interactions of all stressors and .
their effects. At best, the risk estimates of a cumulative risk assessment will reflect some of the
risks that may be reflected in community health statistics. With rare exceptions17, cumulative
risk assessment estimates would not be expected to match exactly with community health
statistics, even for specific health endpoints such as specific cancers.
17 It is conceivable that high risks to rare specific effects could be comparable for a risk assessment and
community health statistics, given current state of the art. To be sure this is not coincidental, a substantial effort to
match risk assessment scenarios with actual histories or exposures would have to be made.
68 :
-------
4.3.3. How the Scope and Purpose of the Assessment Affect Results
Historically, the Agency's risk assessments have focused on assessing the risks from
environmental pollutants to public health or the environment, usually for the purposes of
prioritizing risk management activities or triggering regulatory action. Given the need for public
health-protective decisions, traditional risk assessment tools usually focus on predicting high
ends of the risk distribution. Also, the traditional tools are not designed to predict risk of
diseases other than cancer. Additionally, the many environmental pollutants make up only some
of the categories of risks to public health. Although quite adequate for their original purpose,
when the results of these types of assessments are viewed from another perspective, such as that
of a community concerned about the cumulative health impacts of five industrial and commercial
facilities within a two-block area, they may not be useful.
The Agency is doing more place-based human health and ecological assessments (i.e.,
compared to source- or media-specific assessments) than in the past, but it will be some time
before they become commonplace. Consistent with good practices for planning and scoping,
they often may be driven by specific risk-management needs. The desired objectives and purpose
of parties who were outside the process may differ from those for which the assessment was
designed. For this reason, users of cumulative risk assessments are advised to carefully study the
scope and purpose of the assessment at hand as well as the analysis plan and resulting
characterization to determine whether it is suitable (or partly suitable) to answer questions
outside its stated objectives and purpose.
4.3.4. Documenting Stakeholder Input
Somewhere in the discussion of how the assessment meets or does not meet the
objectives laid out in the planning and scoping phase, it is useful to document how stakeholder
input has influenced the process, noting also those suggestions that were not included and why.
This documentation supports stakeholder participation and provides assurance that individuals
have been heard. {
4.4. Using the Results of the Assessment
• Once the results of an assessment are in hand, the assessment participants will usually
focus primarily on the communication and use of those results. The intended use of the
assessment was considered at the beginning, in the problem formulation phase, both to plan the
assessment work and to set the stage for possible actions that might be taken at this point. A
detailed discussion of the communication and use of the results of a cumulative risk assessment
is beyond the scope of this document, but it should be noted that in deciding on a course of
action, considerations other than the results of the assessment will also need to be taken into
account.
If the goals of a cumulative risk analysis are to estimate the risk from multichemical and
multipathway exposure to people living: within a geographical area of concern, tnen an
69
-------
important objective in presenting the results is to identify the major risk contributors in order to
understand the sources, pathways, and stressors that contribute most to that overall risk. The
results of a cumulative risk assessment provide an additional tool for the risk manager, one that
permits a more complete accounting and more explicit analysis to target follow-up risk
mitigation strategies toward those stressors that most contribute to the population's risk.
If action to mitigate or prevent risk is the goal of the stakeholders, then the options for
action discussed in the planning of the assessment can be re-evaluated in light of the results of
the assessment. Some questions that might arise from this re-evaluation include: "Is regulatory
authority available to address concerns or are voluntary actions better suited to address the
risks?" or "Can the concerns be addressed by the stakeholders involved in the assessment or are
the options for mitigation and prevention beyond the scope of their control?" In the latter case,
for example, siting issues are usually decided locally and may be within the authority of the
participants of a local assessment. In contrast, risk from mobile sources or acid rain are likely to
require action that is beyond the scope of a single local community. In that case, taking action
will require working with other communities and is likely to take more time. Discussion of the
options available for addressing the results of a risk assessment will help to keep expectations in
line with possibilities.
With regard to taking—or not taking—action after a cumulative risk assessment has been
interpreted, the team may benefit from lessons learned by others, just as in the planning, scoping,
and problem formulation phase. In early 2002 the European Environment Agency (EEA, 2001)
released an extensive study of 12 classic case studies in human and environmental health
protection and the lessons learned from them (see text box on the next page). The report is
available on the Internet and should be food for thought for any group contemplating protective
actions, particularly for community assessments.
Finally, it is important to keep in mind that the results of the risk assessment are only one
of the factors to be considered in making a decision on action to address the risk. Risk
information can make an important and valued contribution to the decision-making process, but
it cannot by itself determine the decision. Factors such as the availability of resources for
change; perceived fairness; politics; business and employment; quality-of-life issues; the
religious, cultural, aesthetics, or social values of a community; or concern for future generations
may also influence decisions.
In the siting example mentioned above, the assessment may determine that the new
facility does not significantly increase risk to the community but a decision not to site the facility
might still be made on the basis of a quality-of-life issue that is unrelated to risk. Or, a
community may decide that the economic' and employment benefits outweigh the risks
associated with the siting. Other risk factors not considered in the assessment may also enter
into the decision-making process, including both the environmental risks not covered in the
cumulative risk assessment and the nonenvironmental risks that may affect a community. With
limited resources, a community may use all available risk information to most effectively target
its resources.
70
-------
The European Environment Agency's
12 Lessons Learned Late
• Acknowledge and respond to ignorance, as well as
uncertainty and risk, in technology appraisal and
public policy-making.
• Provide adequate long-term environmental and
health monitoring and research into early warnings.
• Identify and work to reduce blind spots and gaps in
scientific knowledge.
• Identify and reduce interdisciplinary obstacles to
learning. :
• Ensure that real world conditions are adequately
accounted for in regulatory appraisal.
• Systematically scrutinize the claimed justifications
and benefits Alongside the potential risks.
• Evaluate a range of alternative options for meeting
needs alongside the option under appraisal, and
promote more robust, diverse and adaptable
technologies |so as to minimize the costs of
surprises and maximize the benefits of innovation.
• Ensure use of "lay" and local knowledge as well as
relevant specialist expertise in the appraisal.
• Take full account of the assumptions and values of
different social groups.
• Maintain regulatory independence from interested
parties while; retaining an inclusive approach to
information and opinion gathering.
• Identify and reduce institutional obstacles to
learning and action.
• Avoid "paralysis by analysis" by acting to reduce
potential harm when there are reasonable grounds
for concern.
Source: EEA, 2001
71
-------
5. GLOSSARY
Adverse effect - A biochemical change, functional impairment, or pathological lesion that either
singly or in combination adversely affects the performance of the whole organism or reduces an
organism's ability to respond to an additional environmental challenge.
Agent - a chemical, physical, or biological entity that may cause deleterious effects in an
organism after the organism is exposed to it.
Aggregate exposure - The combined exposure of an individual (or defined population) to a
specific agent or stressor via relevant routes, pathways, and sources.
Aggregate risk - The risk resulting from aggregate exposure to a single agent or stressor.
Benchmark dose (BMD) - The dose producing a predetermined, altered response for an effect.
A BMD,0, for example, would be calculated on the basis of a benchmark response of 10%.
Benchmark response (BMR) - A predetermined level of altered response or risk at which the
benchmark dose is calculated. Typically, the BMRs used are 1%, 5%, or 10%.
Conceptual model - A written description and/or a visual representation of actual or predicted
relationships between humans or ecological entities and the chemicals or other stressors to which
they may be exposed.
Cumulative risk - The combined risks from aggregate exposures to multiple agents or stressors.
Cumulative risk assessment - An analysis, characterization, and possible quantification of the
combined risks to health or the environment from multiple agents or stressors.
Dose additivity - In a'mixture, when each chemical behaves as a concentration or dilution of
every other chemical. The response of the combination of chemicals is the response expected
from the equivalent dose of an index chemical (the chemical selected as a basis for
standardization of toxicity of components in a mixture). The equivalent dose is the sum of
component doses scaled by their toxic potency relative to the index chemical. For example, for
chlorinated dibenzodioxins (CDDs), 2,3,7,8-TCDD is selected as the index chemical; other CDD
concentrations are adjusted for their potency relative to 2,3,7,8-TCDD and then treated as if they
were 2,3 J,8-TCDD "equivalents."
Dose-response relationship - A relationship between (1) the dose, either "administered dose" or
absorbed dose and (2) the extent of toxic injury produced by that chemical or agent. Response
can be expressed either as the severity of injury or the proportion of exposed subjects affected.
Endpoint - An observable or measurable biological or chemical event that is used as an index of
the effect of a stressor on a cell, tissue, organ, organism, etc.
72
-------
Exposure pathway - The physical course that a chemical or pollutant takes from the source to
the organism exposed. !
Exposure route - The way a chemical o|r pollutant enters an organism after contact, for example,
by ingestion, inhalation, or dermal absorption.
Lowest-observed-adverse-effect level (LOAEL) - The lowest dose or exposure level at which
there is a statistically or biologically significant increase in the frequency or severity of an
adverse effect in the exposed population as compared with an appropriate, unexpbsed control
group. i
i
i
Model - A mathematical representation
-------
from zero (representing the certainty that the probability of harm is no greater than the
background probability) to one (representing the certainty that harm will occur).
Stakeholder - An interested or affected party in an ongoing or contemplated project (usually
involving a group or team planning the project, analyzing one or more problems, and making
decisions for possible actions on the basis of the interpretation of that analysis).
Stressor - Any physical, chemical, or biological entity that can induce an adverse response. A
stressor may also be the lack of an essential entity, such as a habitat.
74
-------
REFERENCES I .
ACGIH (American Conference of Government Industrial Hygienists), 1998. Threshold Limit
Values for Chemical Substances and Physical Agents. Cincinnati, OH.
ACPA (American Crop Protection Association), 1999. "Cumulative & Aggregate Risk
Evaluation System: CARES. Conceptual Model." Washington, DC. Internet:
www.alphacares.org/. .
i
AIChE (American Institute of Chemical ;Engineers), 1992. Guidelines for Hazard Evaluation
Procedures, 2nd Edition with Worked Examples. Center for Chemical Process Safety, New York,
NY. ;
AIChE, 1996. Guidelines for Use of Vapor Cloud Dispersion Models, 2nd Ed. Center for
Chemical Process Safety, New York, NY-
!
I
AIHA (American Industrial Hygiene Association), 2000. Emergency Response Planning
Guidelines Series. Emergency Response Planning Guidelines Committee, Fairfax, VA.
i
Albert, Roy E., 1999. Unifying the standard setting process for carcinogens and non-
carcinogens. Applied Occupational and Environmental Hygiene 14:742-747.
Albert, Roy E., Joellen Lewtas, Stephen C. Nesnow, Todd W. Thorslund, and Elizabeth L.
Anderson, 1983. Comparative potency method for cancer risk assessment application to diesel
particulate emissions. Risk Analysis 3:101-117.
American Heart Association, 2000. Stroke Risk Factors. Internet: www.americanheart.org.
Ashford, N.A., D. Hattis, E.M. Zolt, J.I. Katz, G.R. Heaton, and W.C. Priest, 1981. "Evaluating
Chemical Regulations: Trade-Off Analysis and Impact Assessment for Environmental Decision-
Making. Massachusetts Institute of Technology, Center for Policy Alternatives. Cambridge,
Massachusetts. CPA-80-13 (NTIS #PB81-195067).
ATSDR (Agency for Toxic Substances and Disease Registry), 1995. Report of the Expert Panel
Workshop on the Psychological Responses to Hazardous Substances. U.S. Department of Health
and Human Services. Atlanta, Georgia. Internet: www.atsdr.cdc.gov/HEC/PRHS/.
i
Barnes, Donald G., Ann Alford-Stevens,| Linda Birnbaum, Frederick W. Kutz, William Wood,
and Dorothy Patton, 1991. Toxic equivalency factors for PCBs? Quality Assurance 1:70-81.
Barnes, Donald G., and Michael L. Dour;son, 1988. Reference dose (RfD): description and use
in health risk assessments. Regulatory Toxicology and Pharmacology 8:471-486.
75
-------
Barnthouse, Lawrence W., David R. Marmorek, and Calvin N. Peters, 2000. Assessment of
multiple stresses at regional scales. In: Multiple Stressors in Ecological Risk and Impact
Assessment: Approaches to Risk Estimation. (Ferenc and Foran, Eds.) Society of Environmental
Toxicology and Chemistry, SETAC Press, Pensacola, Florida. ISBN 1-880611-40-6.
Beck, Barbara D., Rory B. Connolly, Michael L. Dourson, Daniel Guth, Dale Hattis, Carole
Kimmel, and Stephen C. Lewis, 1993. Improvements in quantitative noncancer risk assessment.
Fundamental and Applied Toxicology, 20:1-14.
Berglund, Marika, Carl-Gustaf Blinder, and Lars Jarup, 2001. Human Exposure Assessment: An
Introduction. World Health Organization, Geneva. WHO/SDE/OEH/01.3.
Bogen, K.T., 2001. "Methods for Addressing Uncertainty and Variability to Characterize
Potential Health Risk from Trichloroethylene-Contaminated Ground Water at Beale Air Force
Base in California: Integration of Uncertainty and Variability in Pharmacokinetics and Dose-
Response." Lawrence Livermore National Laboratory, U.S. Department of Energy, Livermore,
CA. UCRL-ID-135978 Rev 1.
Bonnell, Steve, and Keith Storey, 2000. Addressing cumulative effects through strategic
environmental assessment: A case study of small hydro development in Newfoundland, Canada.
Journal of Environmental Assessment Policy and Management2: 477-499.
Bouwes, Nicolaas W. and Steven M. Hassur, 1998. "OPPT's Risk-Screening Environmental
Indicators: Toxic Weights for Toxic Release Inventory (TRI) Chemicals and Chemical
Categories." Office of Pollution Prevention and Toxic Substances, Office of Prevention,
Pesticides, and Toxic Substances, U.S. Environmental Protection Agency. Washington, DC.
April 28, 1998.
Bullard, Robert D., 1990. Dumping in Dixie: Race, Class, and Environmental Quality.
Westview Press, Boulder, CO. ISBN 0-8133-7954-7.
Carpy, Serge A., Werner Kobel, and John Doe, 2000. Health risk of low-dose pesticides
mixtures: A review of the 1985-1998 literature on combination toxicology and health risk
assessment. Journal of Toxicology and Environmental Health Part B, 3:1-25.
CEU (Council of the European Union), 1996. Council directive concerning integrated pollution
prevention and control. Brussels. 96/61/EC dated 24 September 1996, EU official Journal L
257, 10/10/1996; 0026-0040.
CEQ (Council on Environmental Quality), 1997. "Considering Cumulative Effects Under the
National Environmental Policy Act." Executive Office of the President, Washington, DC.
76
-------
Chambers, J.M., W.S. Cleveland, B. Kleiner, and P.A. Tukey. 1983. Graphical methods for data
analysis. Wadsworth;International Group, Belmont, CA, 145-171.
i
Chess, C., andK. Purcell, 1999. Public Participation and the Environment: Do We Know What
Works? Environmental Science & Technology 33(16):2685-2692.
. Clemen, Robert T., 1996. Making Hard'^Decisions: An Introduction to Decision Analysis. 2nd Ed.
Duxbury Press, Wadsworth Publishing Co., Belmont, CA. ISBN 0-534-26034-9. p. 5.
Cogliano, Vincent James, 1997. Plausible upper bounds: Are their sums plausible? Risk Analysis
17:77-84.
i
Cohen, Bernard L., 1991. Catalog of risks extended and updated. Health Physics 61:317-335.
DOT (U.S. Department of Transportation), 1998. "High-Speed Ground Transportation Noise and
Vibration Impact Assessment." Final Draft. Office of Railroad Development, Federal Railroad
Administration. Washington, DC. Report No. 293630-1 Internet: .
http://projectl .parsons.com/ptgnechsr/noisejrnanual.htm.
Dourson, M.L., S.P. Felter, and D. Robinson. 1996. Evolution of science-based uncertainty
factors in noncancer risk assessment. Regulatory Toxicology and Pharmacology 24:108-120.
Dourson, Michael L., Linda K. Teuschler, Patrick R. Durkin, and William M. Stiteler, 1997.
Categorical regression of toxicity data: A case study using aldicarb. Regulatory Toxicology and
Pharmacology 25:121-129. :-.. .
Eadon, George A., Laurence S. Kaminsky, Jay Silkworth, K. Aldous, D. Milker, P. O'Keefe, R.
Smith, J. Gierthy, J. Hawley, N. Kim, et al., 1986. Calculation of 2,3,7,8-TCDD equivalent
concentrations of complex environmental contaminant mixtures. Environmental Health
Perspectives 70:221-227. \
EEA (European Environment Agency), 2001. "Late Lessons from Early Warnings: The
Precautionary Principle 1896-2000." Copenhagen. Environmental Issue Report No. 22 (Released
January 10, 2002) Internet: http://reports.eea.eu.int/environmental_issue_report_2001_22/en.
Environmental Defense, 2001. "Scorecatd.". Internet: http://www.scorecard.org.
European Commission, 2000. "The European Multi-Hazard Risk Assessment Project
(T.E.M.R.A.P.)" Directorate General XII; for Science Research and Development, Environment
and Climate - 1994/1998 - Climatology and Natural Hazard, European Commission. Brussels,
Belgium. Internet: http://phypc9.geo.ulg.ac.be/nouveau/temrap/MainWebPage.htm.
77
-------
Ferenc, Susan A., and Jeffrey A. Foran (Eds.), 2000. Multiple Stressors in Ecological Risk and
Impact Assessment: Approaches to Risk Estimation. Society of Environmental Toxicology and
Chemistry, SETAC Press, Pensacola, FL. ISBN 1-880611-40-6.
Fischhoff, Baruch, S. Watson, and C. Hope, 1984. Defining risk. Policy Sciences 17:123-139.
Foran, Jeffrey A., and Susan A. Ferenc (Eds.), 1999. Multiple Stressors in Ecological Risk and
Impact Assessment. Society of Environmental Toxicology and Chemistry, SETAC Press,
Pensacola, Florida. ISBN 1-880611-32-5.
Freeman, A. Myrick, III, 1999. Economics, Incentives, and Environmental Regulation. In:
Environmental Policy in the 1990s: Reform or Reaction? Fourth Ed., Chapter 9. Norman J. Vig
and Michael E. Kraft [Eds.]. Congressional Quarterly Press, Washington, DC. ISBN 1-56802-
341-3.
Frewer, L., 1999. Risk Perception, Social Trust, and Public Participation in Strategic Decision
Making: Implications for Emerging Technologies. Ambio 28(6):569-574.
GAO (U.S. General Accounting Office), 1983. "Siting of Hazardous Waste Landfills and Their
Correlation with Racial and Economic Status of Surrounding Communities." Washington, DC.
GAO/RCED 83-168.
Gentile, John H., Keith R. Solomon, Jonathan B. Butcher, Michael C. Harrass, Wayne G. Landis,
Michael Power, Barnett A. Rattner, William J. Warren-Hicks, and Robert Wegner, 1999.
Linking Stressors and ecological responses'. In: Multiple Stressors in Ecological Risk and Impact
Assessment. Jeffrey Foran and Susan Ferenc, Eds. Society of Environmental Toxicology and
Chemistry, SETAC Press, Pensacola, FL. ISBN 1-880611-32-5.
Gum, Daniel J., Richard C. Hertzberg , and Annie M. Jarabek, 1993. Exposure-response
analysis: modeling severity against concentration and duration. In: "Improvements in
Quantitative Noncancer Risk Assessment: Symposium Overview," Society of Toxicology
meeting; February 1992; Seattle, WA.
Haber F 1924. Zur Geschichte des Gaskrieges [On the History of Gas Warfare], in: Funf
Vortra'ge aus den Jahren 1920-1923 (Five Lectures from the Years 1920-1923), Springer, Berlin,
76-92.
Haddad, S., G. Charest-Tardif, and K. Krishnan, 2000. Physiologically based modeling of the
maximal effect of metabolic interactions on the kinetics of components of complex chemical
mixtures. Journal of Toxicology and Environmental Health, Part A 61:209-223.
Haddad, S., M. Beliveau, R. Tardif, and K. Krishnan, 2001. A PBPK modeling-based approach
to account for interactions in the health risk assessment of chemical mixtures. Toxicological
Sciences 63:125-131.
78
-------
Hampshire Research Institute, 1999. "Review of an Aggregate Exposure Assessment Tool."
Presentation to the U.S. EPA FIFRA Scientific Advisory Panel, September, 1999 Hampshire
Research Institute, Alexandria, VA. Internet:
www.epa.goy/scipoly/sap/1999/september/aggbkgd.pdf.
Hampshire Research Institute, 2000. "Oyerview of the Fundamentals of Version 1.0 of LifeLine
Software for Modeling Aggregate and Cumulative Exposures to Pesticides." Hampshire
Research Institute, Alexandria, VA. Internet: www.epa.gov/scipoly/sap/2000/september
/finaljfundamentals .pdf.
!
Hansen, Hugh, Christopher T. De Rosa, Hana Pohl, Michael Fay, and Moiz M. Mumtaz, 1998.
Public health challenges posed by chemical mixtures. Environ Health Perspect 106:1271-1280.
!
Hattis, Dale and Rob L. Gobel, 1994. Current Priority-Setting Methodology: Too Little
Rationality or Too Much? In: Adam M. Finkel and Dominic Golding [Eds.]. Worst Things First?
The Debate over Risk-Based National Environmental Priorities. Resources for the Future.
Washington, DC. ISBN 0-915707-76-4. :
Hertzberg, Richard C., 1989. Fitting a model to categorical response data with application to
species extrapolation. Health Physics, 57(Supplement 1):405^09.
Hertzberg, Richard C., 2000. "Communicating cumulative risk: sound bites from chaos."
Luncheon address, presented at the conference, Toxicology and Risk Assessment Approaches for
the 21 st Century, April 10-13, 2000, Kings Island, OH.
Hertzberg, Richard C., and M.M. MacDonell, 2002. Synergy and other ineffective mixture risk
definitions. The Science of the Total Environment 288:31-42.
HUD (U.S. Department of Housing and Urban Development), 1991. "The Noise Guidebook."
Office of Community Planning and Development, Washington, DC. HUD-953-CPD(l).
ILSI (International Life Sciences Institute), 1998. Aggregate Exposure Assessment. Washington,
DC. ISBN 1-57881-040-X. ;
i
ILSI, 1999. A Framework for Cumulative Risk Assessment. Washington, DC. ISBN 1-57881-
055-8. ' ; •
ILSI, 2000. Revised Framework for Microbial Risk Assessment. Washington, DC. ISBN 1-
57881-081-7. I
ILSI, 2001. Aggregate Exposure Assessment: Model Evaluation and Refinement Workshop
Report. Washington, DC. ISBN 1-57881^120-1.
79
-------
IPCS (International Programme on Chemical Safety), 1983. Environmental Health Criteria 27.
Guidelines on Studies in Environmental Epidemiology. International Programme on Chemical
Safety. World Health Organization. Geneva. ISBN 92 4 154087 7.
IPCS, 1993. Environmental Health Criteria 155. Biomarkers in Risk Assessment: Concepts and
Principles. World Health Organization, Geneva. ISBN 92 4 157155 1.
IPCS, 2001. Environmental Health Criteria 222. Biomarkers in Risk Assessment: Validity and
Validation. World Health Organization, Geneva. ISBN 92 4 157222 1.
Johnson, Ted, Gary Mihlan, Jacky LaPointe, Kris Fletcher, Jim Capel, Arlene Rosenbaum,
Jonathan Cohen, and Pat Stiefer, 2000. "Estimation of Carbon Monoxide Exposures and
Associated Carboxyhemoglobin Levels for Residents of Denver and Los Angeles using
pNEM/CO (version 2.1)." Draft report prepared for Office of Air Quality Planning and
Standards, Office of Air and Radiation, U.S. Environmental Protection Agency. Research
Triangle Park, NC. Contract No. 68-D6-0064.
Kasperson, Jeanne X., Roger E. Kasperson, and Billie L. Turner, II (Eds.), 1995. Regions at
Risk: Comparisons of Threatened Environments. United Nations University Press, Tokyo.
Kasperson, Roger E., 2000. Personal communication to Michael Callahan, Chair of the
Cumulative Risk Technical Panel, as part of an EPA Science Advisory Board Consultation on
the draft Framework for Cumulative Risk Assessment.
Kodell, Ralph L., and James J. Chen, 1994. Reducing conservatism in risk estimation for
mixtures of carcinogens. Risk Analysis 14:327-332.
Kroschwitz, Jacqueline I., and Mary Howe-Grant (Eds.), 1994. Kirk-Othmer Encyclopedia of
Chemical Technology. 4th Ed. John Wiley and Sons, New York. ISBN 0-47152-677-0.
Laden, F., L.M. Neas, D.W. Dockery, and J. Schwartz, 2000. Association of fine particulate
matter from different sources with daily mortality in six U.S. Cities. Environmental Health
Perspectives 108:941-947.
Lantz, Paula M, James S. House, James M. Lepkowski, David R. Williams, Richard P. Mero,
and Jieming Chen, 1998. Socioeconomic factors, health behaviors, and mortality: Results from a
nationally representative prospective study of US adults. Journal of the American Medical
Association 279:1703-1709.
Lewis, S.C., 1993. Reducing uncertainty with adjustment factors: Improvements in quantitative
noncancer risk assessment. Fundamentals of Applied Toxicology 20:2-4.
80
-------
Lewtas, Joellen, 1985. Development of a comparative potency method for cancer risk
assessment of complex mixtures using short-term in vivo and in vitro bioassays. Toxicol. Ind.
Health 1:193-203. ;
Lewtas, Joellen, 1988. Genotoxicity of complex mixtures: Strategies for the identification and
comparative assessment of airborne mutagens and carcinogens from combustion sources.
Fundamentals of Applied Toxicology 10;571-589.
Lynch, John W., George A. Kaplan, Elsie R. Pamuk, Richard D. Cohen, Katherine E. Heck,
Jennifer L. Balfour, and Irene H. Yen, 1998. Income inequality and mortality in metropolitan
areas of the United States. American Journal of Public Health 88:1074-1080.
Morata, Thais C., 2000. "Suggested Guidelines to Studying the Combined Effects of
Occupational Exposure to Noise and Chfemicals on Hearing." National Institute for Occupational
Safety and Health. Cincinnati, OH. Dated February, 2000. Internet:
http://www.ucl.ac.uk/noiseandhealth/WJ>3%20draft%20protocol%20-%20Morata.doc.
Morata, T.C., A.C. Fiorini, P.M. Fischer;, S. Colacioppo, K.M. Wallingford, E.F. Kreig, D.E.
Dunn, L. Gozzoli, M.A. Padrao, and C.L.G. Cesar, 1997. Toluene-induced hearing loss among
rotogravure printing workers. Scandanavian Journal of Work, Environment & Health
23:289-298. ' ' |-
Morgan, M. Granger, and Max Henrion, 1990. Uncertainty: A Guide to Dealing with
Uncertainty in Quantitative Risk and Policy Analysis. Cambridge University Press, New York,
NY. ISBN 0-521-36542-2. \
Murray, C., 1994. Quantifying the burden of disease: the technical basis for disability-adjusted
life years. Bulletin of the World Health Organization 72(3):429-445.
NAPA (National Academy of Public Administration), 1995. Setting Priorities, Getting Results:
A New Direction for EPA. Washington, DC. LCCN 95-68048.
NIOSH (National Institute for Occupational Safety and Health), 1997, "NIOSH Pocket Guide to
Chemical Hazards." Centers for Disease-Control and Prevention. Washington, DC. NTIS PB 97-
177-604. !
. [
NOAA (National Oceanic and Atmospheric Administration), 1999. "An Assessment of Lower
Snake River Hydrosystem Alternatives on Survival and Recovery of Snake River Salmonids."
National Marine Fisheries Service. Seattle, WA. Dated April 14, 1999.
NRC (National Research Council), 1983;. Risk Assessment in the Federal Government:
Managing the Process. Committee on thfe Institutional Means for Assessments of Risk to Public
Health, Commission on Life Sciences. National Academy Press, Washington, DC. ISBN 0-309-
03349-7.
81
-------
NRC, 1993. Pesticides in the Diets of Infants and Children. National Academy Press,
Washington, DC.
NRC, 1994. Science and Judgment in Risk Assessment. Committee on Risk Assessment of
Hazardous Air Pollutants, Board on Environmental Sciences and Technology, Commission on
Life Sciences. National Academy Press, Washington, DC. ISBN 0-309-04894-X.
NRC, 1996. Understanding Risk: Informing Decisions in a Democratic Society. Committee on
Risk Characterization, Commission on Behavioral and Social Sciences and Education. National
Academy Press, Washington, DC. ISBN 0-309-05396-X.
OMB (U.S. Office of Management and Budget), 2000. "Guidelines to Standardize Measures of
Costs and Benefits and the Format of Accounting Statements." Memo dated March 22, 2000,
from Jacob J. Lew, Director, to Heads of Departments and Agencies. Washington, DC.
OMB, 2002. "Guidelines for Ensuring and Maximizing the Quality, Objectivity, Utility, and
Integrity of Information Disseminated by Federal Agencies." Washington, DC. Federal Register
67:369-378.
Oris, J.T., and J.P. Geisy, 1985. The photo-enhanced toxicity of anthacene to juvenile sunfish
(Lepomis spp.). Aquatic Toxicology 6:133-146.
Osowski, Sharon L., Joseph D. Swick, Jr., Gerald R. Carney, Hector B. Pena, Jeffrey E.
Danielson, and David A. Parrish, 2001. A watershed-based cumulative risk impact analysis:
Environmental vulnerability and impact criteria. Environmental Monitoring and Assessment
66:159-185.
Perry, Robert H., Don W. Green, and James O. Maloney (Eds.), 1997. Perry's Chemical
Engineers'Handbook. 7th Ed. McGraw Hill Professional Publishing, New York. ISBN 0-07049-
841-5.
PCCRARM (Presidential/Congressional Commission on Risk Assessment and Risk
Management), 1997. Risk Assessment and Risk Management in Regulatory Decision-Making.
Washington, DC.
Seed, Jennifer, Ronald P. Brown, Stephen S. Olin, and Jeffery A. Foran, 1995. Chemical
mixtures: Current risk assessment methodologies and future directions. Regulatory Toxicology
and Pharmacology 22:76-94.
Sexton, Ken, David E. Kleffman, and Michael A. Callahan, 1995. An introduction to the
National Human Exposure Assessment Survey (NHEXAS) and related phase I field studies. J.
Expos. Anal. Environ. Epidem. 5:229-232. Related papers are in the same issue of Journal of
Exposure Analysis and Environmental Epidemiology 5:233-444.
82
-------
Slob, W., and M.N. Pieters. 1998. A probabilistic approach for deriving acceptable human intake
limits and human health risks from toxicological studies: General framework. Risk Anal.
18:787-798. ''•-..
l' i
State of Vermont, 1991. "Environment 1991: Risks td Vermont and Vermonters." Agency of
Natural Resources. Waterbury, VT. ' j'
i
Tanabe, Shinsuke, 1998. Butyltin contamination in marine mammals: A review. Marine
Pollution Bulletin 39:62-72. '
i
Teuschler, Linda K., Michael L. Doursoh, William M. Stiteler, Peter McClure, and Heather
Tully, 1999. Health risks above the reference dose for multiple chemicals. Regulatory
Toxicology and Pharmacology 30:S 19—S26.
Thomas, J., 1995. Public Participation in Public Decisions: New Skills and Strategies for Public
Managers. Jossey-Bass, San Francisco, pA.
United Church of Christ, 1987. Toxic Waste and Race in the United States: A National Report on
the Racial and Socio-Economic Characteristics of Communities with Hazardous Waste Sites.
United Church of Christ Commission for Racial Justice. New York, NY.
USEPA (Environmental Protection Agency), 1986a. "The Risk Assessment Guidelines of 1986."
Risk Assessment Forum, Office of Research and Development. Washington, DC. EPA/600/8-
87/045. : ,
USEPA, 1986b. "Guidelines for the Health Risk Assessment of Chemical Mixtures." Risk
Assessment Forum, Office of Research sind Development. Washington, DC. EPA/630/R-98/002.
USEPA, 1986c. "Guidelines for Mutagenicity Risk Assessment." Risk Assessment Forum,
Office of Research and Development. Washington, DC. EPA/630/R-98/003.
i
USEPA, 1986d. "Guidelines for Carcinogen Risk Assessment." Risk Assessment Forum, Office
of Research and Development. Washington, DC. EPA/63O/R-98/001.
USEPA, 1987. "The Total Exposure Assessment Methodology (TEAM) Study." Office of Acid
Deposition, Environmental Monitoring and Quality Assurance, Office of Research and
Development. Washington, DC. EPA/60JO/6-87/002.
USEPA, 1988. "Federal Guidance Report No. 11: Limiting Values of Radionuclide Intake and
Air Concentration and Dose Conversion Factors for Inhalation, Submersion, and Ingestion."
Office of Radiation Programs. Washington, DC. EPA-520/1-88-020.
83
-------
USEPA, 1989a. "Risk Assessment Guidance for Superfund, Volume 1: Human Health
Evaluation Manual. Office of Emergency and Remedial Response, Office of Solid Waste and
Emergency Response. Washington, DC. EPA/540/1 -89/002.
USEPA, 1989b. "Interim Procedures for Estimating Risks Associated with Exposure to
Chlorinated dibenzo-p-dioxins and -dibenzofurans (CDDs and CDFs) and 1989 update. Risk
Assessment Forum, Office of Research and Development Washington, DC. EPA 625/3-89/016.
USEPA, 1990a. "Technical Support Document on Health Assessment of Chemical Mixtures."
Office of Research and Development. Washington, DC. EPA/600/8-90/064.
USEPA, 1990b. "Reducing Risk: Setting Priorities and Strategies for Environmental Protection."
Science Advisory Board. Washington, DC. SAB-EC-90-021.
USEPA, 199la. "Guidelines for Developmental Toxicity Risk Assessment." Risk Assessment
Forum, Office of Research and Development. Washington, DC. EPA/600/FR-91/001.
USEPA, 1991b. "Locational Data Policy." Office of Policy, Planning, and Evaluation.
Washington, DC. IRM Policy Manual 2100-CGH2, dated April 8, 1991.
USEPA, 1992a. "Safeguarding the Future: Credible Science, Credible Decisions. A Report of the
Expert Panel on the Role of Science at EPA." Washington, DC. EPA/600/9-91/050
USEPA, 1992b. "Framework for Ecological Risk Assessment." Risk Assessment Forum, Office
of Research and Development, Science Advisory Board. Washington, DC. EPA/630/R-92/Q01.
USEPA, 1992c. "Guidelines for Exposure Assessment." Risk Assessment Forum, Office of
Research and Development. Washington, DC. EPA/600/Z-92/001.
USEPA, 1993a. "Guidance on the Application of Refined Dispersion Models for
Hazardous/Toxic Air Release." Office of Air Quality Planning and Standards, Office of Air and
Radiation. Research Triangle Park, NC. EPA 454/R-93-002. !
USEPA, 1993b. "A Guidebook to Comparing Risks and Setting Environmental Priorities."
Office of Policy, Planning and Evaluation. Washington, DC. EPA 230-B-93-003.
USEPA, 1993c. "Chemical Indexing System for the Toxic Chemical Release Inventory, Part I:
Chronic Index." Air, Radiation and Toxics Division, EPA Region III. Philadelphia, PA.
EPA/903/R-93/002.
USEPA, 1993d. "Federal Guidance Report No. 12: External Exposure to Radionuclides in Air,
Water, and Soil." Office of Air and Radiation. Washington, DC. EPA-402-R-93-081.
84
-------
USEPA, 1994. "Methods for Derivation !of Inhalation Reference Concentrations and Application
of Inhalation Dosimetry." Office of Health and Environmental Assessment, Office of Research
and Development. Research Triangle Park, NC. EPA/600/8-90/066F.
USEPA, 1995a. "Policy for Risk Characterization." Memorandum from U.S. Environmental
Protection Agency Administrator Carol M. Browner, dated March 21, 1995. Washington, DC.
USEPA, 1995b. "Guidance for Risk Characterization." Policy paper dated February, 1995.
Science Policy Council. Washington, DC.
USEPA, 1995c. "The Use of the Benchmark Dose Approach in Health Risk Assessment." Risk
Assessment Forum, Office of Research and Development. Washington, DC. EPA/630/R-94/007.
USEPA, 1995d. "Chemical Indexing System for the Toxic Chemical Release Inventory, Part I:
Chronic Index; Addendum." Air, Radiation and Toxics Division, EPA Region III. Philadelphia,
PA. EPA/903/R-93/002a (August, 1995).
i ,
USEPA, 1995e. "Compilation of Air Pollutant Emission Factors, Volume I: Stationary Point and
Area Sources." 5th Ed. Office of Air Quality Planning and Standards, Office of Air and
Radiation. Research Triangle Park, NC. EPA AP-42.
i
USEPA, 1996a. "RAGS Reform Stakeholder Forums: Synopsis of Participants' Comments. San
Francisco, California, October 30-November, 1,1996, and Washington, DC, November 6-
November 8, 1996." Office of Emergency and Remedial Response, Washington, DC.
USEPA, 1996b. "Guidelines for Reproductive Toxicity Risk Assessment." Risk Assessment
Forum, Office of Research and Development. Washington, DC. EPA/630/R-96/009.
USEPA, 1996c. "Proposed Guidelines fpr Carcinogen Risk Assessment." Risk Assessment
Forum, Office of Research and Development. Washington, DC. EPA/600/P-92/003C.
USEPA, 1996d. "Compilation of Air Pollutant Emission Factors, Volume I: Stationary Point and
Area Sources." 5th Ed. Supplements A & B. Office of Air Quality Planning and Standards,
Office of Air and Radiation. Research Triangle Park, NC. EPA AP-42.
USEPA, 1996e. "Radiation Exposure and Risk Assessment Manual (RERAM)." Office of Air
and Radiation. Washington, DC. EPA 4Q2-R-96-016..
85
-------
USEPA, 1997a. "Guidance on Cumulative Risk Assessment, Part 1. Planning and Scoping."
Science Policy Council. Washington, DC. Attachment to memo dated July 3, 1997 from the
Administrator, Carol Browner, and Deputy Administrator, Fred Hansen, titled "Cumulative Risk
Assessment Guidance-Phase I Planning and Scoping."
USEPA, 1997b. "Guiding Principles for Monte Carlo Analysis." Risk Assessment Forum,
Office of Research and Development, Washington, DC. EPA/63 O/R-97/001.
USEPA, 1997c. "Chemical Indexing System, Part II: Vulnerability Index." Waste and
Chemicals Management Division, EPA Region III. Philadelphia, PA. EPA/903/R-97/021.
USEPA, 1997d. "Compilation of Air Pollutant Emission Factors, Volume I: Stationary Point and
Area Sources." Fifth Edition, Supplement C. Office of Air Quality Planning and Standards,
Office of Air and Radiation. Research Triangle Park, NC. EPA AP-42.
USEPA, 1998a. "Risk Assessment Guidance for Superfund (RAGS) Stakeholder Forum:
Synopsis of Participants' Comments. Atlanta, Georgia, March 2-4, 1998." Office of Emergency
and Remedial Response. Washington, DC.
USEPA, 1998b. "Guidelines for Ecological Risk Assessment." Risk Assessment Forum, Office
of Research and Development, Washington, DC. EPA/630/R-95/002F.
USEPA, 1998c. "Guidelines for Neurotoxicity Risk Assessment." Risk Assessment Forum,
Office of Research and Development. Washington, DC. EPA/630/R-95/00 IF.
USEPA, 1998d. "Methods for Exposure-Response Analysis for Acute Inhalation Exposure to
Chemicals: Development of the Acute Reference Exposure." Review Draft. Office of Research
and Development. Washington, DC. EPA/600/R-98/051.
USEPA, 1998e. "General Guidance for Risk Management Programs (40 CFR Part 68)."
Chemical Emergency Preparedness and Prevention Office, Office of Solid Waste and Emergency
Response. Washington, DC. EPA 550-B-98-003.
USEPA, 1998f. "Comparative Risk Framework: Methodology and Case Study." SAB Review
Draft dated November 9, 1998. National Center for Environmental Assessment, Office of
Research and Development. Cincinnati, OH.
USEPA, 1998g. "Report of the Common Sense Initiative Council's Stakeholder Involvement
Work Group." Common Sense Initiative Council. Washington, DC.
USEPA, 1998h. "An SAB Report: Review of Disproportionate Impact Methodologies." Science
Advisory Board. Washington, DC. EPA-SAB-IHEC-99-007. [
86
-------
USEPA, 19981. "Handbook for Air Topics Emission Inventory Development, Volume I:
Stationary Sources." Office of Air Quality Planning and Standards, Office of Air and Radiation.
Research Triangle Park, NC. EPA-454/B-98-002.
i
USEPA, 1998J. "Human Health Risk Assessment Protocol for Hazardous Waste Combustion
Facilities." Peer Review Draft. Office of Solid Waste and Emergency Response. Washington,
DC. EPA530-D-98-001 Internet: www.epa.gov/epaoswer/hazwaste/combust/riskvol.htm.
USEPA, 1998k. "A Common Mechanism of Action: The Organophosphate Pesticides."
Scientific Advisory Panel. Washington, DC. Internet:
www.epa.gov/scipoly/sap/1998/march/comec.htm.
USEPA, 1999a. "Guidelines for Carcinogen Risk Assessment." Review draft dated July, 1999.
Risk Assessment Forum, Office of Research and Development. Washington, DC. NCEA-F-0644.
USEPA, 1999b. "Review of Revised Sections of the Proposed Guidelines for Carcinogen Risk
Assessment." Science Advisory Board. Washington, DC. EPA-SAB-EC-99-015.
USEPA, 1999c. "Risk Assessment Guidance for Superfund: Volume 1 - Human Health
Evaluation Manual. Supplement to Part A: Community Involvement in Superfund Risk
Assessments." Office of Solid Waste and Emergency Response. Washington, DC. EPA 540-R-
98-042/PB99-963303. i
USEPA, 1999d. "Risk Management Program Guidance for Offsite Consequence Analysis."
Chemical Emergency Preparedness and Prevention Office, Office of Solid Waste and Emergency
Response. Washington, DC. EPA 550-B:-99-009.
USEPA, 1999e. "Guideline on Air Quality Models." Office of Air Quality Planning and
Standards, Office of Air and Radiation. Research Triangle Park, NC. 40CFR Ch. I (7-1-99
Edition), Appendix W to Part 51, pages 390^81.
USEPA, 1999f. "EPA's Framework for Community-Based Environmental Protection." Office of
Policy/Office of Reinvention. Washington, DC. EPA 237-K-00-001.
USEPA, 1999g. "Guidance for Performing Aggregate Exposure and Risk Assessments." Office
of Pollution Prevention and Toxic Substances, Office of Pesticide Programs. Washington, DC.
Item 6043, dated October 29, 1999. Available on the internet at:
http://www.epa.gOV/fedrgstr/EPA-PEST/l 999/November/Day-10/6043 .pdf.
USEPA, 1999h. "Guidance for Identifying Pesticide Chemicals and Other Substances that Have
a Common Mechanism of Toxicity." Office of Pesticide Programs, Office of Prevention,
Pesticides, and Toxic Substances. Washington, D.C. Internet:
http://www.epa.gov/fedrgstr/EP A_PEST/1999/February/Day_05/.
87
-------
USEPA, 19991. "Risk-Screening Environmental Indicators: 1988-1997 TRI Data 'Air-Only'
Model." Office of Pollution Prevention and Toxic Substances, Office of Prevention, Pesticides,
and Toxic Substances. Washington, DC. CD-ROM Version 1.0, dated July 6, 1999. (Version 2.0
is now in beta-testing.)
USEPA, 1999J. "An SAB Report on the National Center for Environmental Assessment's
Comparative Risk Framework Methodology." Science Advisory Board. Washington, DC. EPA-
SAB-DWC-99-016.
USEPA, 1999k. "Handbook for Criteria Pollutant Inventory Development: A Beginner's Guide
for Point and Area Sources." Office of Air Quality Planning and Standards, Office of Air and
Radiation. Research Triangle Park, NC. EPA-454/R-99-037.
USEPA, 19991. "Federal Guidance Report No. 13: Cancer Risk Coefficients for Environmental
Exposure to Radionuclides." Office of Air and Radiation. Washington, DC. EPA-402-R-99-001.
USEPA, 2000a. "Toward Integrated Environmental Decision-Making." Science Advisory
Board. Washington, DC. EPA-SAB-EC-00-01 1 .
USEPA, 2000b. "Benchmark Dose Technical Guidance Document" Draft report. Risk
Assessment Forum, Office of Research and Development. Washington, DC. EPA/630/R-00/001.
USEPA, 2000c. "Science Policy Council Handbook: Risk Characterization." Science Policy
Council. Washington, DC. EPA 1 OO-B-00-002.
USEPA, 2000d. "Science Policy Council Handbook: Peer Review." 2nd Edition. Science Policy
Council. Washington, DC. EPA 100-B-OO-OOl.
USEPA, 2000e. "Supplementary Guidance for Conducting Health Risk Assessment of Chemical
Mixtures." Risk Assessment Forum, Office of Research and Development. Washington, DC.
EPA/630/R-00/002.
USEPA, 2000f. "Baltimore Community Environmental Partnership Air Committee Technical
Report. Community Risk-Based Air Screening: A Case Study in Baltimore, MD." Office of
Pollution Prevention and Toxics, Office of Prevention, Pesticides, and Toxic Substances.
Washington, DC. EPA 744-R-00-005.
USEPA, 2000g. "Handbook for Non-Cancer Health Effects Valuation." Non-Cancer Health
Effects Valuation Subcommittee of the EPA Social Science Discussion Group, Science Policy
Council. Washington, DC. Dated November, 2000.
USEPA, 2000h. "AP-42: Compilation of Air Pollutant Emission Factors, Volume II: Mobile
Sources." Office of Transportation and Air Quality, Office of Air and Radiation. Washington,
DC. EPA AP-42, Volume II Internet: www.epa.gov/otaq/ap42.htm.
88
-------
USEPA, 20001. "Catalog of Hazardous and Solid Waste Publications." 13th Edition. Office of
Solid Waste and Emergency Response. Washington, DC. EPA530-B-00-001 Internet:'
www.epa.gov/epaoswer/osw/catalog.htm.
i
USEPA, 2000J. "Guide to Field Storage of Biosolids, Appendix A: Odor Characterization,
Assessment and Sampling." Office of Wastewater Management, Office of Water. Washington,
DC. EPA/832-B-00-007. Internet: www.epa.gov/owm/bio/fsguide/..
i
USEPA, 2001a. Personal communication. Debby Sisco,. Biological and Economic Analysis
Division, Office of Pesticide Programs, Office of Prevention, Pesticides, and Toxic Substances.
Washington, DC. August 1,2001. j
USEPA, 2001b. Personal communication.'Anna Koutlakis, Office of Prevention, Pesticides, and
Toxic Substances. Washington, DC. August 1, 2001.
USEPA, 2001c. "Stakeholder Involvement & Public Participation at the U.S. EPA: Lessons
Learned, Barriers, & Innovative Approaches." Office of Policy, Economics and Innovation.
Washington, DC. EPA-100-R-00-040. |
USEPA, 200Id. "Top 10 Watershed Lessons Learned." Office of Wetlands, Oceans and
Watersheds, Office of Water. Washington, DC. Internet:
http://www.epa.gov/owow/watershed/lessons/top 10.pdf.
USEPA, 200le. "National-Scale Air Topics Assessment for 1996." SAB Review Draft. Office of
Air Quality, Planning and Standards, Office of Air and Radiation. Washington, DC. EPA-453-R-
01-003. :
i
USEPA, 2002a. "Guidance on Cumulative Risk Assessment of Pesticide Chemicals that Have a
Common Mechanism of Toxicity." Office of Pesticide Programs, Office of Prevention,
Pesticides, and Toxic, Substances. Washington, DC. January 14, 2002. Internet:
http://www.epa.gov/pesticides/trac/scienpe/cumulative_guidance.pdf.
USEPA, 2002b. "Lesson Learned on Planning and Scoping for Environmental Risk
Assessments." Science Policy Council. Washington, DC. January, 2002. Internet:
http://www.epa.gov/ORD/spc/2cumrisk.htm.
USEPA, 2002c. "Summary Report of the Technical Peer Review Workshop on the EPA Risk
Assessment Forum Draft Framework for Cumulative Risk Assessment." Risk Assessment
Forum, Washington, DC; EPA/630/R-03/002. Internet: http://www.epa.gov/ncea/raf.
USEPA, 2002d. "Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlorodibenzo
-p-Dioxin (TCDD) and Related Compounds." National Center for Environmental Assessment,
Office of Research and Development. Washington, DC, (to be published).
89
-------
USEPA, 2002e "A Review of the Reference Dose and Reference Concentration Processes." Risk
Assessment Forum, Washington, DC; EPA/630/P-02/002F. Internet: http://www.epa.gov/ncea/raf.
van den Berg, Martin, Linda Birnbaum, Albertus T.C. Bosveld, Bjorn Brunstrom, Philip Cook, Mark
Feeley, John P. Giesy, Annika Hanberg, Ryuichi Hasegawa, Sean W. Kennedy, Timothy Kubiak, John
Christian Larsen, F.X. Rolaf van Leeuwen, A.K. Djien Liem, Cynthia Nolt, Richard E. Peterson,
Lorenz Poellinger, Stephen Safe, Dieter Schrenk, Donald Tillitt, Mats Tysklind, Maged Younes,
Fredrik Wzern, and Tim Zacharewski, 1998. Toxic equivalency factors (TEFs) for PCBs, PCDDs,
PCDFs for humans and wildlife. Environmental Health Perspectives 106:775-792.
WHO (World Health Organization), 2001. "Approaches to Integrated Risk Assessment." International
Programme on Chemical Safety. Geneva. WHO/IPCS/IRA/01/12 Internet:
http://w\vw.who.int/pcs/emerg_main.html.
Woo, Yin-Tak, Fred J. DiCarlo, Joseph C. Arcos, Mary Argus, Greg Polansky, and Jeff DuBose,
1994. Assessment of carcinogenic hazard of chemical mixtures through analysis of binary chemical
interaction data. Environmental Health Perspectives 102 (Supplement 9): 113-118.
Woodruff, Tracey J., Jane Caldwell, Vincent J. Cogliano, and Daniel A. Axelrad, 2000. Estimating
cancer risk from outdoor concentrations of hazardous air pollutants in 1990. Environmental Research
Section A 82:194-206.
Zartarian, Valerie G., Haluk Ozkaynak, Janet M. Burke, Maria J. Zufall, Marc L. Rigas, and Edwin J.
Furtaw, Jr., 2000. "A Modeling Framework for Estimating Children's Residential Exposure and Dose
to Chlorpyrifos via Dermal Residue Contact and Non-Dietary Ingestion." Environmental Health
Perspectives 108:505-514.
90
-------
APPENDIX A: RESEARCH AND DEVELOPMENT NEEDS
Framework for Cumulative Risk Assessment is intended to provide a basic structure for
the issues and define key terms and concepts. In some cases, the concepts introduced in the
framework report require the application of knowledge and methods that are not currently
available. The following is a discussion ;of the needed areas of research and methods
development highlighted within the report that may be most important to an evaluation of
cumulative risks. This is not intended to; be a comprehensive listing of cumulative risk
assessment research needs. ;
i •
i
EPA and other scientists are currently investigating the use of similar approaches for
cancer and noncancer assessments. Although we do not discuss this research need here, it would
be useful to cumulative risk assessment to have similar approaches, and it is a topic of current
discussion within scientific circles (e.g., jAlbert, 1999).
i
Understanding the Timing of Exposure and its Relationship to Effects
A key concept in the definition of cumulative risk is that it represents an accumulation of
risk over time. However, unlike the traditional approach to risk assessment, where exposure
events are summed and averaged over a period of time, cumulative risk assessment involves
developing an understanding of how the sequence and timing of exposures influence the ultimate
risk for effects. For example, for multiple stressors, it is important to understand how prior
exposures to one or several stressors influence the risks from subsequent exposures to the same
or different stressors. In addition, it is injiportant to understand the implications of these
exposures occurring during critical periods of an individual's life (e.g., important periods of
development or periods of disease). Several exposure models are under development that
recognize the need to understand the timing of various exposure events (e.g., Calendex, APEX,
Lifeline, SHEDS, and CARES/RExY). ;
In addition to gaining a better understanding of the sequence and timing of exposures and
their relationship to effects, it is important to understand how acute, nonlethal exposures from
accidents contribute to chronic or long-tdrm effects.
i
Understanding the Composition and Toxicity of Mixtures
Chemical mixtures can change or; degrade over time and space, making the assessment of
exposure a particular challenge. For cumulative risk assessment, the composition of the mixture
at the point of contact with the receptor should be well characterized. Measurement techniques
(at the receptor) and predictive models are both applicable in this characterization.
EPA's guidance for the health risk assessment of chemical mixtures (USEPA, 2000e)
presents approaches for combining the toxicities of multiple chemical stressors. These
approaches necessarily involve a number of simplifying assumptions when the mixtures are
complex. Although the current methods provide a valuable resource for assessing cumulative
91
-------
risks, future cumulative risk assessment will need a more complete understanding of the
interactions among chemicals in complex mixtures. Some current research efforts are seeking to
identify toxicologic principles of joint action that are applicable to mixtures involving many
chemicals.
Applying the Risk Factor Approach to Environmental Health Risks
The risk factor approach has been used in the medical profession to predict the chances
of individuals developing various diseases. It has proved to be a useful approach not only in
assessing certain cumulative risks, but also in communicating with patients. In this approach,
characteristics of a population (e.g., age, ethnicity, personal habits, genetic polymorphisms, prior
diseases, etc.) are correlated with the incidence of disease. For some diseases (e.g., breast
cancer, coronary artery disease, stroke) these correlations are well established. However, there
are substantial data gaps in terms of the role played by exposures to environmental stressors in
the development of human disease, and correlations of environmental exposures with disease
outcomes are generally not available.
Using Biomarkers and Biomonitoring
The use of biomarkers of exposure or effect holds a great deal of promise for cumulative
risk assessment. This approach can provide a method for assessing stressors in groups.
Currently, however, this approach is not practicable when considering a large number of diverse
stressors, because appropriate biomarkers for many types of stressors have not yet been
developed.
Considering Hazards Presented by Nonchemical Stressors
Cumulative risk assessment could encompass the interactions of chemical stressors with
biological, radiological, and other physical stressors; socioeconomic stressors; and lifestyle
conditions. In trying to assess all these different types of stressors, it is helpful to determine
what types of effects the stressors produce and then to try to group stressors by like effects.
Ideally, one would like to know the mechanism or mode of action by which various stressors
cause effects to allow a more refined grouping. Currently, however, there are few methods for
understanding how these disparate stressors interact to result in risk.
Considering Psychological Stress as Part of Cumulative Risk
Psychological stress causes both psychological and physiological changes that can be
measured. However, assessing levels of stress and their potential contribution to risk is difficult
for a variety of reasons. The Agency for Toxic Substances and Disease Registry began the
process of identifying research needs in this area through an expert panel workshop held in 1995.
Considering All Aspects of Vulnerability
92
-------
The issue of the vulnerability of a population can be thought of as having four
components: susceptibility of individuals, differential exposures, differential preparedness to
withstand the insult, and differential ability to recover from effects. Traditional risk assessment
may consider one or more of these categories, but rarely are all considered. The overall
consideration of all four categories may be more important in cumulative risk assessment than in
traditional one-chemical assessments. A cumulative risk assessment, for example, may need to
consider potential combinations of high exposure and high vulnerability across stressors.
Methods development work is needed in this. area. ,
Methods for Combining Different Types .of Risk
Another key concept in the definition of cumulative risk assessment is that such an
assessment represents the combined risk! from multiple stressors. This implies that, in some
cases, it may be necessary to combine disparate measures of risk (i.e., different types of effects)
to simplify the expression of cumulative risks. There have been some attempts to collapse
complex arrays of risk into a few or eveij a single measure. These approaches have involved the
use of common metrics (e.g., quality-adjusted life years, disability-adjusted life years, loss of life
expectancy, etc.) and indices (e.g., hazard ranking system, etc.) and the categorization of effects
(e.g., as for categorical regression). Alternatively, geographic information systems and mapping
techniques can be used to graphically pqrtray integrated information on risks without
mathematically combining disparate measures. Much methods development work remains to be
completed in each of these areas. '•
Development of Default Values for Cumulative Risk Assessments
\
Conventional'risk assessments use a series of default values for screening or other
applications, and it may be necessary to investigate whether certain defaults need to be
established specifically for cumulative risk assessments.
i "
Development of Case Studies and Issue Papers on Specific Cumulative Risk Topics
The more detailed technical issues and methodologies should be developed as a series of
issues papers that would augment the framework report. The level of .detail would, of course,
vary, depending on the topic. The issues papers (or white papers) should also include details on
additional approaches to cumulative risk assessment that are currently being explored (including
screening-level analyses, place-based assessments, comparative risk assessments, National
Environmental Policy Act cumulative effects analyses, and hazard assessments). In addition, the
issues papers could include summaries of case studies of cumulative risk projects that would
extend the framework from theoretical to practical approaches and applications.
93
-------
APPENDIX B: SELECT RESOURCES FOR EXPOSURE AND RISK ASSESSMENT
B.I. Resources Relevant to Chemical Exposures
EPA Guidelines:
Most of EPA's general guidelines are listed in the text box in Section 1.1.
Air-related sources and activities:
EPA's Clearinghouse for Inventories and Emission Factors (CHIEF) website
(wvvw.epa.gov/ttn/chief/) is an excellent starting place. It has many of the relevant
documents on methods and data for constructing emissions inventories available for
download, including Handbook for Criteria Pollutant Inventory Development: A
Beginner's Guide for Point and Area Sources (USEPA, 1999k), Handbook for Air Toxics
Emission Inventory Development, Volume I: Stationary Sources (USEPA., 19981), and the
two volumes and supplement of Compilation of Air Pollutant Emission Factors (for both
stationary and mobile sources) (USEPA, 1995e, 1996d, 1997d, 2000h), as well as many
other documents and software.
EPA's Support Center for Regulatory Air Models (SCRAM) website
(www.epa.gov/ttn/scram/) provides extensive information on the models discussed in
Guideline on Air Quality Models (USEPA, 1999e), including downloadable software and
users guides for many of the models.
The Ambient Monitoring Technology Information Center (AMTIC) website
(www.epa.gov/ttn/amtic/) contains information on monitoring programs and methods and
other monitoring-related information.
The umbrella website for all three of the above is the Technology Transfer Network
(www.epa.gov/ttn/), which also has other useful information and links in addition to
those noted above.
Sources for land and waste-related activities:
EPA's Office of Solid Waste and Emergency Response has compiled an extensive
catalog summarizing their publications (USEPA, 2000i). It has also published a "peer
review draft" document titled Human Health Risk Assessment Protocol for Hazardous
Waste Combustion Facilities (USEPA, 1998J), which deals with how to assess risks from
hazardous waste incinerators. These reports are available on-line.
94
-------
!
Chemical accidents and transportation-related spills:
i
Assessing an accidental chemical release exposure involves several steps. The typical
analytical steps in the overall assessment are process analysis, likelihood or frequency of
accidents, source term modeling, dispersion or consequence modeling, and the exposure
assessment. ;
i
»• The process analysis is a formal, systematic analysis of the process where a
chemical is handled to determine the probabilities and consequences of acute,
catastrophic failures of engineered systems leading to an accidental release of the
chemical. This analysis is often called a process hazards analysis (PHA). Several
formal PHA evaluation techniques are available, including "What-If," "Failure
Mode and Effect Analysis;," "Event-Tree," and "Fault-Tree" analyses (USEPA,
1998e;AIChE, 1992). •
*• The likelihood or frequency of accidents step is an evaluation of each of the
scenarios uncovered in the process analysis step for likelihood or frequency of
occurrence. ; .
I
*• Source term modeling, which estimates the amount or rate of release in case of
accident, is performed once the failure scenarios are determined. A wide variety
of published calculation methods or models are available (USEPA, 1998e, 1999d)
to determine the source terms for an accidental chemical release.
*• Dispersion or consequence modeling is performed once the source terms (rate and
duration of the release) are known. A wide variety of dispersion and consequence
modeling tools, ranging from simple screening models to sophisticated and
complex computer applications, are available for this step (USEPA, 1993a,
1999d; AIChE, 1996). In addition to the source terms generated above, several
other data elements are needed, such as physical/chemical properties (e.g.,
whether the vapor cloud is heavier than air or water reactive), meteorological
conditions (e.g., wind speed and direction, temperature, humidity), and terrain
surrounding the facility (e.g., buildings or valleys that may channel or disperse a
vapor cloud). Physical/chemical properties can be found in-chemical reference
texts such as Kirk-Othmer's Encyclopedia of Chemical Technology (Kroschwitz
and Howe-Grant, 1994), Perry's Chemical Engineers' Handbook (Perry et al.,
1997), on Material Safety Data Sheets (MSDS)18, or Risk Management Guidance
for Offsite Consequence Analysis (USEPA, 1999d). Meteorological conditions
are often collected on-site or at local airports. Information about terrain can be
collected from topological maps or by visual inspection. Guidance on all these
parameters is available in USEPA (1999d).
18
There are many searchable MSDS databases on-line that can be located with most search engines.
! 95
-------
»• The final step is the exposure assessment, which is related to, and builds from, the
dispersion or consequence modeling step. The dispersion or consequence
modeling depends on a health endpoint and the exposure level related to that
endpoint. Besides lethality, concentrations for certain health effects (e.g., odor
thresholds, eye irritation) are available for several common toxic substances
(NIOSH, 1997; ACGIH, 1998; AIHA, 2000).
B.2. Resources Relevant to Exposures to Nonchemical Stressors
Biological stressors:
The International Life Sciences Institute's Risk Science Institute has published a
workshop report entitled "Revised Framework for Microbial Risk Assessment" (ILSI,
2000), which looks at methods for assessing risks to microorganisms such as
Cryptosporidium, which has caused disease outbreaks when it contaminates drinking
water. The methodology is superficially similar to that of a risk assessment conducted
for a chemical pollutant, but only at the most general level. For example, the
characterization of exposure in the ILSI framework differs from that in an environmental
chemical exposure assessment; it includes (1) pathogen characterization, (2) pathogen
occurrence, (3) exposure analysis, and, finally, developing (4) an exposure profile.
Radiological stressors:
EPA's Office of Air and Radiation maintains a web page at
. This page provides (or cites) much of the
needed documentation for performing risk assessments for radionuclides, including
Radiation Exposure and Risk Assessment Manual (RERAM) (USEPA, 1996e) and several
Federal guidance reports (USEPA, 1988,1993d, 19991).
Noise, vibration, and congestion:
The U.S. Department of Housing and Urban Development (HUD) has issued The Noise
Guidebook (HUD, 1991), which implements the existing noise regulations (24 CFR 51-
B) and includes the HUD noise assessment guidelines. (The guidebook is available in
hard copy only.)
The Federal Railroad Administration has developed a manual titled High-Speed Ground
Transportation Noise and Vibration Impact Assessment (DOT, 1998), which provides the
theory, equations, and applications of noise and vibration analysis for high-speed
railroads. Much of the theory and information is also applicable to other noise and
vibration problems. Appendix A of the DOT guide is a general discussion of noise
concepts, with references. The guide is available on-line
(http://projectl .parsons.com/ptgnechsr/noise_manual.htm).
96
-------
Odor:
The National Institute of Occupational Health and Safety has done much research on the
interaction of noise with chemical exposures (Morata, 2000).
EPA's Office of Wastewater Management has issued a report titled Guide to Field
Storage ofBiosolids (USEPA, 2000J), which contains an appendix on "Odor
Characterization, Assessment, and Sampling." Odor assessment is an analytic-
deliberative process involving both science-based analytical methods and more
subjective analysis. The appendix of the guide discusses sensory characterization of
odors (character, intensity, pervasiveness, quantity), some practical options for assessing
odors in a community, and the chemistry of odors (including range of odor thresholds). It
also discusses odor sample collection and analysis and has several dozen references for
further information. This report is available on-line (www.epa.gov/owm/bio/fsguide/).
97
-------
19
APPENDIX C: SOME THOUGHTS ON BACKGROUND EXPOSURES
When looking at aggregate exposures or cumulative risks of citizens, "background
exposures" to specific chemicals are no less "real" than the exposures to pollution usually
studied for regulatory purposes. Whereas in historical single-chemical assessments conducted
for limiting pollution, background sources of the chemical were often irrelevant to the questions
being asked of the assessment (or ignored as having negligible effect on risk), background
sources in cumulative risk assessments are rarely irrelevant.20
Background concentrations can be categorized as either naturally occurring, that is,
chemicals that are naturally present in the environment before it was influenced by humans, or
anthropogenic, that is, present in the environment due to historical human-made sources.
Naturally occurring background chemicals may be either localized or ubiquitous. Anthropogenic
background sources can be either localized from a point source or generalized from unidentified
sources or nonpoint sources.
Assessments of morbidity incidence and death rates, market basket surveys, and pesticide
residue surveys also provide information that can be reflective of background chemical
concentrations as well as overt pollution. Background issues extend across all media, beyond
regulated sources, and beyond direct exposure. Many chemicals are naturally present in the
environment (e.g., soils, water, vegetation, and other biota) and are consequently part of dietary,
dermal, and inhalation exposures. In some cases, naturally occurring substances may be present
at levels that exceed health-based or risk-based regulatory standards (e.g., drinking water
standards) or other levels established to protect human health and the environment. Because
cumulative risk assessments are population based, exposures due to naturally occurring
background concentrations should typically be considered important.
19 Several terms are used to discuss background, and there are several ways to describe different aspects of
this issue. It has been suggested (deFur, 2002) that a more appropriate term for present conditions is "ambient," and
that "background" should be reserved for some untouched, even pristine state or condition. Although the Technical
Panel discussed this use of the word "background" as a pristine reference area, the discussion in this appendix is
meant to more closely reflect the way the word is used in practice within EPA. It is acknowledged that not all
programs or scientists even within EPA use this term to mean the same thing.
20 The word "background" is often used to describe exposures to chemicals or other stressors that derive
from sources other than the sources being assessed. For example, in the Agency's assessment of residual risk
associated with hazardous air pollutant emissions from particular categories of sources that remain after the
implementation of technology-based controls, "background" is defined as all hazardous air pollutant exposures (via
inhalation or other routes) not associated with the source(s) being assessed. At a Superfund site, "background
contamination" refers to contamination that is not related to the site release of chemicals, as defined by
Comprehensive, Environmental Response, Compensation and Liability Act (CERCLA) (P.L. 96-510, December 11,
1980, as amended by P.L. 98-802, August 23,1983, and P.L. 99-499, October 17, 1986). Such focusing or
segregation in a risk assessment can be useful to decisions involving pollution sources covered by particular
statutory authorities, but it is typical of a chemically focused assessment rather than a population-focused assessment
such as a cumulative risk assessment.
98
-------
There are several important issues related to natural or anthropogenic background
concentrations in cumulative risk assessment. First, if the risks posed by background
concentrations of certain chemicals are significant (and some may approach or exceed health
reference levels), their exclusion from the cumulative risk estimates and characterization may
seriously distort the portion of the total estimated risk thought to be posed to the population by a
specific evaluated source. A second issue is the problem of whether background chemical
exposures can be clearly distinguished from specific source-related chemicals and how to
quantify these exposures. It may be important in a cumulative risk assessment to estimate
background exposures separately from specific source-related exposures, so that the risk assessor
can provide the community with a more complete picture of both total and known source-related
risks. This also provides a clearer, more'complete picture for making risk management
decisions. Finally, there may be problems in identifying representative geographic areas for
determining background levels for comparison. '
Finally, background exposures for a community or population may also include both .
voluntary and involuntary exposures and subsequent risks. Involuntary exposures are associated
with the naturally occurring or anthropogenic background concentrations described above.
Voluntary exposures, such as are associated with lifestyle decisions, are exposures due to
activities such as smoking, consuming char-grilled meats with polycyclic aromatic
hydrocarbons, or other choice-based exposures and may also sometimes be defined in the
assessment as background exposures if they are not assessed directly in the cumulative risk
assessment. ; .
99
-------
APPENDIX D: EXAMPLES OF OUTLINES OF ANALYSIS PLANS21
D.I. Outline for Human Health Analysis Plan for Pesticides Under the'1996 Food Quality
Protection Act (FQPA)
Risk management/regulatory goal: Protection of the general human population and susceptible
subpopulations to adverse effects from exposure to pesticide "X" under FQPA.
Assessment Endpoints:
- human or animal health status of exposed versus unexposed populations/cohorts/dose
groups
Measures of Effects:
- general types of toxicological effects grouped according to acute, subchronic, and
chronic exposure durations
- organ-specific toxicity such as reproductive effects, developmental effects,
neurotoxicity, developmental neurotoxicity, immunotoxicity, hepatotoxicity,
pulmonary effects, cardiovascular effects, etc.
- general classes of toxic effects such as carcinogenicity, mutagenicity
Measures of Exposure:
- monitoring of food, water, residential, occupational exposures, etc. (direct or surrogate)
- monitoring of biological fluids or biomarkers (blood, urine, DNA or other
macromolecules)
What Can and Cannot Be Done Based on Planning and Scoping
- pathways and relationships to be evaluated
- resource restraints
- milestones for completion of risk assessment
Methods for Conducting Risk Analysis
-RfD ;
- margin of exposure
- probabilistic risk assessment based on dose-response or exposure parameters
- quotients (e.g., ratio of exposure level to toxicity threshold)
- narrative discussions
- other considerations (e.g., mechanisms of action, toxicokinetic models, timing of dose,
sensitive population characteristics)
Data Needs and Uncertainties
21 Conceptual models are not included here.
100
-------
D.2. Outline for Ecological Analysis Plan
Risk management/regulatory goal: Viable, self-sustaining coho salmon population that supports
a subsistence and sport fishery. ',
Assessment endpoints: Coho salmon breeding success, fry survival, and adult return rates.
Measures of Effects: !
- egg and fry response to low dissolved oxygen
- adult behavior in response to obstacles
- spawning behavior and egg survival with changes in sedimentation
- population data over time in relation to fish passage
Measures of Ecosystem and Receptor Characteristics:
- water temperature, water velocity, and physical obstructions
- abundance and distributions of suitable breeding substrate
- abundance and distribution of suitable food sources for fry
- feeding, resting, and breeding behavior
- natural reproduction, growth, and mortality rates
Measures of Exposure: [
- number of hydroelectric dams and associated ease of fish passage
- toxic chemical concentrations in water, sediment, and fish tissue
- nutrient and dissolved oxygen levels in ambient waters
- riparian cover, sediment loading, and water temperature
What Can and Cannot Be Done Based on Planning and Scoping
- pathways and relationships to be evaluated
- resource restraints
- milestones for completion of risk assessment
Methods for Conducting Risk Analysis ;
- quotients I
- narrative discussions i
- stressor-response curves with probabilities
Data Needs and Uncertainties !
101
-------
APPENDIX E: TOXICOLOGIC SIMILARITY—ORGANOPHOSPHORUS
PESTICIDES
The Food Quality Protection Act of 1996 (FQPA) requires that EPA reassess pesticide
tolerances (legal limits for residues in food) that were in effect as of August 1996. As part of the
reassessment, EPA should consider available information concerning the cumulative effects on
human health resulting from exposure to multiple chemicals that have a common mechanism of
toxicity. In this context, pesticides are determined to have a common mechanism of toxicity if
they produce the same toxic effect in the same organ or tissue by essentially the same sequence
of major biochemical events (USEPA, 1999h).
Shortly after enactment of FQPA, EPA began developing new methods and tools that
would allow the consideration of combined risks from exposure to several pesticides via several
pathways and routes of exposure. Actual data sets for organophosphorus (OP) pesticides were
used in pilot analyses to test these methods. The methods and pilot analyses, were subjected to
peer review through the FIFRA Scientific Advisory Panel to ensure the use of sound science. As
part of this ongoing effort, on December 28, 2001 EPA's Office of Pesticide Programs (OPP)
announced the availability of the preliminary organophosphorus cumulative risk assessment
[66FR67249-67250]. The risk assessment is available electronically at
. In preparing the cumulative risk assessment for
the OP pesticides, OPP followed five major steps.
1. Selection of the specific pesticides, pesticide uses, and pathways and routes of exposure to
include in the quantitative analysis.
The selection of the specific OP pesticides began with identifying a "common
mechanism group." This was accomplished following Guidance For Identifying
Pesticide Chemicals And Other Substances That Have A Common Mechanism Of
Toxicity (available at http://www.epa.gov/pesticides/trac/science). All 39 registered OP
pesticides share inhibition of acetylcholinesterase as a common mechanism for causing
adverse effects (USEPA, 1998k).
The common mechanism group was further refined to reflect current use patterns and
information on the detection of residues from USDA's Pesticide Data Program. This
resulted in the following recommendations for quantitative analysis: include 22 OP
pesticides for the food pathway of exposure; 24 OPs for the water pathway and 10 OPs
for residential exposures were identified on the basis of use patterns and their individual
assessments.
102
-------
2. Dose-response analysis for toxic potencies, relative contribution from each OP pesticide, and
selection of an index chemical to use as the point of reference in the dose-response analysis.
To determine the combined risk from multiple OP pesticides, EPA used the relative
potency factor approach (for additional examples of comparative potency approaches, see
Albert et al., 1983; Lewtas, 1985; 1988). The index chemical was selected on the basis of
the quality of the dose-response data. Then the relative potency of each OP pesticide was
estimated by taking the ratio of its toxic potency to that of the index chemical.
In selecting studies for evaluating toxic potencies, EPA used relative potency factors and
points of departure developed from cholinesterase inhibition in rats exposed to pesticides
for 21 days or more. This practice was adopted to reflect cholinesterase inhibition at a
point in the treatment schedule at which a steady state had been achieved. OPP elected to
use data that reflected a steady state in the interest of producing relative potency factors
that are reproducible and reflect less uncertainty due to rapidly changing time-sensitive
measures of cholinesterase. •
Also, EPA considered that people generally have some level of prior exposure to OP
pesticides. Further, the effects of exposure can persist for several days to weeks.
Therefore, people may be more vulnerable to subsequent exposures to OP pesticides than
might be predicted if these prior exposures are not considered.
3. Estimation of the risks associated with all pertinent pathways of exposure in a manner that is
both realistic and reflective of variability due to differences in location, time, and demographic
characteristics of exposed groups. <
Evaluation of the OP pesticide use profiles allowed for the identification of exposure
scenarios that may overlap, co-occur, or vary between chemicals. In addition, the use of
profiles allowed for the identification of populations of potentialconcerh. On the basis
of this analysis, EPA considered 'exposure to OP pesticides in food to be uniform across
the nation (i.e., there are no significant differences in food exposure due to time of year
or geographic location). For the residential and drinking water pathways of exposure,
EPA divided the nation into 12 regions for assessment. This allowed for the
consideration of such factors as the location of vulnerable surface watersheds and region-
specific pest pressures. To estimate risks, EPA used Calendex, a calendar-based
computer model. This model integrates the various pathways of exposure while
simultaneously incorporating the time dimensions of the data. The model produces a
detailed profile of the potential exposure to individuals across a calendar year.
4. Identification of the significant contributors to risk.
Although interpretation of the preliminary organophosphorous cumulative risk
assessment is ongoing, there are some early indications concerning contribution to risk.
The drinking water pathway for exposure does not appear to be a major contributor to the
103
-------
total cumulative risk. Residential exposure appears to be a contributor to risk,
particularly inhalation exposures from certain no-pest strips and crack and crevice
treatments. Childhood exposure from mouthing hands also appears to be a contributor,
but there is a great deal of uncertainty associated with the estimates.
5. Characterization of the confidence in the results and the uncertainties encountered.
In addition to some uncertainties noted above, EPA identified many areas for additional
analysis, including sensitivity analyses on input parameters, verification of residential use
patterns, closer examination of the tails of the food consumption distribution, and
evaluation of the effect of assumptions about residue concentrations in baby foods.
104
-------
APPENDIX F: OTHER TYPES OF CUMULATIVE ASSESSMENTS
Several other types of cumulative assessments are related to the types of human health
and ecological cumulative assessments done by the Agency. It is beyond the scope of this
framework to discuss these in detail, but a short explanation of several other types of cumulative
assessments are given in this appendix.
F.I. Quality-of-Life Assessments
One type of assessment that resembles a cumulative risk assessment—but whose
evaluation may require a different approach from the traditional National Research Council risk
paradigm—is the quality-o&life assessment. These assessments define "harm" to an individual
or community broadly, then evaluate the importance of the various threats of harm to a set of
"quality-of-life" criteria. These assessments do not usually attempt to predict probability that the
harm will occur (as would a cumulative risk assessment), but rather aim to apply the
community's values to deal with the most important perceived threats.
)•
Although a quality-of-life assessment is not a risk assessment in most cases, changes in
quality-of-life factors may affect the vulnerability of a population to health or ecological risks
and consequently may be part of the considerations in a cumulative risk assessment. Because
few, if any, established and accepted relationships are currently available quantitatively linking
quality-of-life factors and health or ecological risk, this is an area in which further research may
prove valuable. '
To evaluate the effects on human or ecological health from these types of impacts, a
more deliberative approach (in the analytical-deliberative process) is needed than is used in, say,
cancer risk analysis. To better help characterize these impacts, EPA's A Guidebook to
Comparing Risks and Setting Environmental Priorities (USEPA, 1993b) suggests a six-step
process in quality-of-life analysis:
1. Identify impacts and determine the values of the community.
2. Identify and define evaluative criteria.
3. Collect and analyze data on impacts.
4. Characterize impacts for all problem areas.
5. Present findings and rank problem areas for quality-of-life impacts.
6. Analyze future environmental conditions and risk management considerations.
105
-------
Vermont's Quality-of-Life Criteria
Impacts on Aesthetics: Reduced visibility, noise,
odors, dust and other unpleasant sensations, and
visual impact from degradation of natural or
agricultural landscapes.
Economic Weil-Being: Higher out-of-pocket
expenses to fix, replace, or buy items or services
(e.g., higher waste disposal fees, cost of replacing a
well, higher housing costs), lower income or higher
taxes paid because of environmental problems, and
health-care costs and lost productivity caused by
environmental problems.
Fairness: Unequal distribution of costs and benefits
(e.g., costs and benefits may be economic, health-
related, aesthetic).
Future Generations: Shifting the costs (e.g.,
economic, health risks, environmental damage) of
today's activities to people not yet able to vote or not
born yet.
Peace of Mind: Feeling threatened by possible
hazards in air or drinking water or potentially risky
structures of facilities (e.g., waste sites, power lines,
nuclear plants), and heightened stress caused by
urbanization, traffic, etc.
Recreation: Loss of access to recreational lands
(public and private) and degraded quality of
recreation experience (e.g., spoiled wilderness,
fished-out streams).
Sense of Community: Rapid growth in population or
number of structures or development that changes the
appearance and feel of a town; loss of mutual respect,
cooperation, ability, or willingness to solve problems
together; individual liberty exercised at the expense
of the community; the loss of Vermont's landscape
and the connection between the people and the land.
Source: State of Vermont, 1991
Quality-of-life impacts are determined
by analyzing a set of criteria developed for
each community, depending on what it
values. Stressors are those things that
threaten to degrade the quality-of-life criteria
for that community. An example of a set of
quality-of-life criteria and their descriptions
is shown in the box on this page. These
criteria were developed by the State of
Vermont's Agency of Natural Resources
(State of Vermont, 1991). Vermont's
experience in evaluating these criteria was
described as a qualitative description of harm
or, in their terms, "risk":
Because most of these seven criteria
are intangible, they are extremely
difficult to measure or quantify. The
Quality-of-Life Work Group
described how each problem area
affects each criterion and how
widespread or intense the effects are.
Although these non-quantitative
descriptions of risk often lack
precision and scientific objectivity,
they focus attention on specific
critical issues and thus are useful tools
for comparing the problems
systematically and consistently. (State
of Vermont, 1991)
Quality-of-life issues can encompass
much more than the criteria shown in the
example and thus may introduce much
additional complexity into the analysis. For
instance, there may be feedback loops that
cannot be easily evaluated, for example, loss
of property value or aesthetics tends to
negatively affect the socioeconomic system,
which tends to increase rates of crime, traffic
accidents, and communicable pathogen transmission, all of which in turn ultimately reflect on
overall community health or ecological risk. Some cumulative risk assessments may
consequently include quality-of-life impacts as indirect measures of health effects if sufficient
links can be established between the two.
106
-------
F.2. Cumulative Impact Assessments
The National Environmental Policy Act (NEPA) defines "cumulative impact" (see box), and has
certain requirements for a cumulative impacts analysis. Although the Council on Environmental
Quality's guidelines for cumulative impact analysis (CEQ, 1997) take a primarily qualitative
approach to the analysis, this is a multiple-stressor, multiple-effect analysis that looks at a
variety of impacts on the environment.
The projects or actions that NEPA addresses can be viewed as sources of stressors.
Under NEPA, a description of the affected environment in an environmental impact assessment
contains four types of information: (1) data on the status of important natural, cultural, social, or
economic resources and systems; (2) data that characterize important environmental or social
stress factors; (3) a description of pertinent regulations, administrative standards, and
development plans; and (4) data on environmental and socioeconomic trends. Health effects on
populations and susceptible individuals are part of the affected environment as considered by the
NEPA cumulative effects analysis, but the NEPA analysis may also consider effects on historic
and archaeological resources, socioeconomic factors such as employment, human community
structure, and quality of life changes. :
Although there is not always a clear relationship between these NEPA cumulative
impacts and effects relevant to human health, the NEPA methods and tools for cumulative
impact analysis may be useful for cumulative risk assessments. For example, cumulative impact
analysis begins with an extensive scoping process and relies on conceptual models to plan the
analysis. NEPA effects data may help risk assessors identify susceptible subpopulations,
environmental pathways, or exposure patterns. •
EPA Region 6 has developed a
system called the Cumulative Risk Index;
Analysis (CRIA), primarily for NEPA-type
assessments (Osowski et al., 2001). The
CRIA contains some 90 criteria with which
to evaluate the health of an area and its ;
ecosystem/human populations. These criteria
help evaluate such diverse factors as human
health, ecosystem health, and environmental
justice considerations. Each criterion, which
leads to an indexing of 1 through 5, has been
through the deliberative process and peer
review and is well documented.
We also acknowledge that other
Federal agencies have been preparing
"cumulative risk analyses" for various
NEPA's "Cumulative Impact" Definition
Council on Environmental Quality Regulation 1508
for Implementing the National Environmental Policy
Act of 1969 [P.L. 91-190, 42 U.S.C. 4321-4347,
January 1, 1970, as amended by P.L. 94-52, July 3,
1975, P.L. 94-83, August 9, 1975, and P.L. 97-258,
§4(b), Sept. 13, 1982] defines "cumulative impact"
as "the impact on the environment which results from
the incremental impact of the action when added to
other past, present, and reasonably foreseeable future
actions regardless of what agency (Federal or
non-Federal) or person undertakes such other actions.
Cumulative impacts can result from individually
minor but collectively significant actions taking place
over a period of time."
Source: CEQ, 1997
107
-------
purposes related to their own mission as part of environmental impact statements (e.g., NOAA, 1999).
F.3. Empirically Derived Medical Models
The medical profession has long used empirically derived models to predict the chances
of particular health effects in individual patients. In this approach, the characteristics of
individuals within the population are correlated with the incidence of specific diseases or effects.
For example, the risk factors for stroke are increasing age, heredity (family history) and race,
prior stroke, high blood pressure, cigarette smoking, diabetes mellitus, carotid and other artery
disease, heart disease, transient ischemic attacks, high red blood cell count, sickle cell anemia,
socioeconomic factors, excessive alcohol consumption, and certain types of drug abuse
(American Heart Association, 2000). Each of these risk factors can be correlated with stroke
incidence, and then the risk of stroke from various combinations of these factors can be explored.
In this way, the analysis is "cumulative," but "risk factors" are not always synonymous with
"stressors."
Physicians use models containing effect-specific risk factors to advise patients of the
probabilities of future effects (e.g., stroke, breast cancer) on the basis of their medical history.
Although the medical data upon which these factors are based have been well developed for
many effects in humans, there are substantial data gaps in terms of the role played by exposures
to many chemicals in the environment in the development of human disease. This empirically
derived medical model approach to cumulative risk may be built on links between risk factors
and effects for better-studied stressors but may be limited or nonexistent for less robust health
effects databases. Although this approach may some day be applicable to human health and
environmental risk assessment such as EPA conducts, at present the data and methods are not
available.
In a larger sense, although empirically derived models may be cumulative risk models,
the approach to determining risk is substantially different from the risk assessment approach
used by EPA, where a combined effect is estimated as the predicted aggregation of the effects of
several different stressors. In an empirical model such as physicians use, the focus is on an
effect of concern, and the model derives the influence of various "stressors" or "risk factors"
from actual observations, usually through the use of multiple regression analyses. Although
ideally the equations derived to represent the influences of various factors on the measured
outcome (the effect of concern) would be causal-predictive models, in practice they are usually
the most parsimonious "best fit" equations that satisfy statistical criteria. The versatility of this
approach, however, is the ability to tease apart contributions of different sources of
environmental exposures of interest. This is illustrated by Laden et al. (2000) in the association
of particulates from different sources with short-term mortality changes. This approach also has
considerable potential to be used in conjunction with biomarkers as dependent variables (Hattis,
2002).
The topic of cumulative risk models will likely be covered in more detail in the future
guidelines for cumulative risk assessment.
108
-------
F.4. Risk Surrogates
Geographic information systems and related mapping techniques (see, e.g.,
Environmental Defense, 2001) appear to hold some promise as tools for presenting integrated
information concerning cumulative risks without mathematically combining disparate measures.
Considerable methods development work remains to be completed.
Not all statements of probability of harm are expressed as probabilities of specific health
effects. Cohen (1991) uses mortality ratios to derive "loss of life expectancy" (LLE) estimates
for a wide, variety of risk-related activities. For example, workers in all occupations have a 60-
day LLE as a result of working, but workers in agriculture have a 320-day LLE and construction
workers a 227-day LLE as a result of their particular occupation. These types of statements are
empirically derived, probability-based statements of harm that do not use "probability of adverse
health effect" as the basis for the risk statement. For estimates such as LLEs, one could
theoretically add up the various activities and the corresponding LLEs in days to estimate a
cumulative risk in terms of loss of life expectancy. These "other" types of risk-surrogate
probability statements could conceivably be used in cumulative risk assessment, although
currently they are not widely used, perhaps due to lack of methods.
109
-------
-------
-------
-------
-------
-------