NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
National Research Council (US) Committee on Improving Risk Analysis Approaches Used by the U.S. EPA. Science and Decisions: Advancing Risk Assessment. Washington (DC): National Academies Press (US); 2009.
![Cover of Science and Decisions](/corehtml/pmc/pmcgifs/bookshelf/thumbs/th-nap12209-lrg.png)
Science and Decisions: Advancing Risk Assessment.
Show detailsINTRODUCTION AND DEFINITIONS
In the previous chapters, the committee proposed modifications of multiple risk-assessment steps to provide better insight into the health risks associated with exposure to individual chemicals, including characterization of uncertainty and variability. That reflects the focus of many risk-assessment applications in the Environmental Protection Agency (EPA) and elsewhere, which are often centered on evaluating risks associated with individual chemicals in the context of regulatory requirements or isolated actions, such as the issuance of an air permit for an industrial facility.
However, there is increasing concern among stakeholder groups (especially communities affected by environmental exposure) that such a narrow focus does not accurately capture the risks associated with exposure, given simultaneous exposure to multiple chemical and nonchemical stressors and other factors that could influence vulnerability. More generally, a primary aim of risk assessment should be to inform decision-makers about the public-health implications of various strategies for reducing environmental exposure, and omission of the above factors may not provide the information needed to discriminate among competing options accurately. Without additional modifications, risk assessment might become irrelevant in many decision contexts, and its application might exacerbate the credibility and communication gaps between risk assessors and stakeholders.
In part to address those complex issues, EPA has developed the Framework for Cumulative Risk Assessment (EPA 2003a). Cumulative risk is formally defined as the combination of risks posed by aggregate exposure to multiple agents or stressors in which aggregate exposure is exposure by all routes and pathways and from all sources of each given agent or stressor. Chemical, biologic, radiologic, physical, and psychologic stressors are all acknowledged as affecting human health and are potentially addressed in the multiple-stressor, multiple-effects assessments (Callahan and Sexton 2007). Cumulative risk assessment is therefore defined as analysis, characterization, and possible quantification of the combined risks to health or the environment posed by multiple agents or stressors (EPA 2003a).
As noted recently (Callahan and Sexton 2007), there are four key differences between EPA’s cumulative risk-assessment paradigm and traditional human health risk assessments:
- Cumulative risk assessment is not necessarily quantitative.
- Cumulative risk assessment by definition evaluates the combined effects of multiple stressors rather than focusing on single compounds.
- Cumulative risk assessment focuses on population-based assessments rather than source-based assessments.
- Cumulative risk assessment extends beyond chemicals to include psychosocial, physical, and other factors.
In addition, an explicit component of the cumulative risk-assessment paradigm defined by EPA involves an initial planning, scoping, and problem-formulation phase (EPA 2003a), which the committee previously proposed as an important component of any risk assessment in Chapter 3. That involves bringing risk managers, risk assessors, and various stakeholders together early in the process to determine the major factors to be considered, the decision-making context, the timeline and related depth of analysis, and so forth. Planning and scoping ensure that the right questions are asked in the context of the assessment and that the appropriate suite of stressors is considered (NRC 1996).
The committee acknowledges the conceptual framework and broadened definitions of cumulative risk assessment as constituting a move toward making risk assessments more relevant to decision-making and to the concerns of affected communities. Many components of cumulative risk assessment (such as planning and scoping or explicit consideration of vulnerability) should be considered as standard features of any risk assessment in principle. In practice, however, EPA assessments conducted today can fall short of what is possible and what is supported by the agency’s framework, and this chapter is directed at improvements in agency practice.
The chapter considers in detail some of the specific reasons why cumulative risk assessment might be needed, because the risk-management needs will inform necessary revisions of the analytic framework. First, even if the regulatory decision of interest were related to strategies to address a single chemical with a single route of exposure, consideration of other compounds and other factors may be necessary to inform the decision. Ignoring numerous agents or stressors that affect the same toxic process as the chemical of interest and omitting background processes could lead to risk assessments that, for example, assume population thresholds in circumstances when such thresholds may not exist. That issue has been largely addressed in Chapter 5 in relation to the need to evaluate background exposure and vulnerability factors to determine the likelihood that these factors could “linearize” an otherwise nonlinear mode of action (MOA). We do not treat this issue in further detail in this chapter other than to note that it is a crucial component of cumulative risk assessment and that it leads to potentially important exposure-assessment and epidemiologic and toxicologic data requirements.
Second, as alluded to above, the types of questions that are increasingly being asked of EPA require the tools and concepts of cumulative risk assessment. Communities concerned about environmental toxicants often wish to know whether environmental factors can explain observed or hypothesized disease trends or whether specific facilities are associated with important health burdens (and whether specific interventions could reduce those burdens). The relevance of standard risk-assessment methods in settings with vulnerable populations and multiple coexposures is being challenged by stakeholders, especially those with concerns about environmental justice (Israel 1995; Kuehn 1996). Addressing those issues requires an ability to evaluate multiple agents or stressors simultaneously—to consider exposures not in isolation but in the context of other community exposures and risk factors. In addition, many of the decisions faced by EPA and other stakeholders involve tradeoffs and complex interactions among multiple risk factors, and any analytic tool must be able to address these factors reasonably.
Although we propose in this chapter some modifications of the framework and practice of cumulative risk assessment to help EPA and other stakeholders to determine high-risk populations and discriminate among competing options, we recognize that the topic of cumulative risk assessment raises important questions about the bounds between risk assessment and other lines of evidence that may inform risk-related decisions. As the number and types of stressors and end points under consideration increase, decisions must be made about which dimensions should be considered as components of risk assessment as defined and used by EPA and others and which dimensions should be considered as ancillary information that can inform risk-management decisions but not considered as a components of risk assessment itself. That is in part a semantic distinction, but defining the bounds will be important in articulating recommendations for improving risk-analysis methods in EPA. Similarly, decisions must be made about the levels of complexity and quantification necessary for a given cumulative risk assessment in light of the decision context. This chapter emphasizes methods that can allow for the quantification of human health effects associated with exposure to chemical and nonchemical stressors, but we note that cumulative risk assessment can involve qualitative analyses and is not necessarily quantitative (EPA 2003a; Callahan and Sexton 2007), given that such analyses may be sufficient at times to discriminate among competing risk-management options.
Another boundary issue involves the contexts in which cumulative risk assessment would be able to yield useful information. Some of the questions that communities or other stakeholders are concerned about cannot and should not be answered by risk assessment even if refined techniques addressing cumulative risks are used. For example, questions like “What are the sources of environmental contaminants in our community that may be causing the most health problems?” or “What intervention strategies that we can adopt would most improve community health?” can be answered in principle with risk-assessment methods, but questions like “Should yet one more polluting facility be sited in our community?” or “Should there be mitigation because this low-income population lives much closer to sources of environmental contaminants than high-income populations?” are broader questions than can be answered by cumulative risk assessment alone. Clarifying the types of questions that cumulative risk assessment can and cannot answer but can support will be important in refining the cumulative risk-assessment tools and considering complementary analyses to aid in decision-making.
In this chapter, we briefly discuss some key settings in which cumulative risk assessment has been developed and applied in EPA, focusing on the problem context, the analytic methods used, and refinements that may be warranted. We consider proposed approaches derived from such fields as ecologic risk assessment and social epidemiology to construct cumulative risk models in the light of numerous stressors or end points, while maintaining focus on decisions relevant to EPA. We conclude by providing some specific guidance about how the committee believes that cumulative risk assessment needs to be developed further, including the use of clear and consistent terminology; methods to incorporate interactions between chemical and nonchemical stressors; the use of biomonitoring, epidemiologic, and surveillance data; the need to develop simpler analytic tools to support more wide-ranging analyses; and the related need to engage stakeholders throughout the cumulative risk-assessment process.
HISTORY OF CUMULATIVE RISK ASSESSMENT
The formal cumulative risk-assessment framework at EPA was developed recently, but relevant activity has occurred for decades. This historical overview is not meant to be exhaustive but rather aims to illustrate some of the different ways in which cumulative-risk issues have been addressed at different times in different offices in EPA.
One of the early applications of cumulative risk assessment in EPA was in the context of the Superfund program. Given the focus on specific hazardous-waste sites rather than single compounds, risk assessments need to capture the health effects of simultaneous exposures. EPA issued guidance documents focused on methods for addressing chemical mixtures (EPA 1986), which were relatively undetailed but established the general approach of first looking for evidence of health effects of the mixture of concern, then considering effects of a similar mixture if no such information were available, then addressing pairwise interactions if data were available, and finally presuming additivity if none of the prior information was available. The 1986 guidelines also distinguished between dose additivity (appropriate if the compounds of interest had the same MOA and the same health effects) and response additivity (which presumes independent MOAs). Data were available on some complex mixtures, such as diesel emissions and polychlorinated biphenyls, or mixtures similar to them; but in the majority of cases, dose additivity when the same MOA could be assumed was the default. Analyses of chemical mixtures constitute only one component of cumulative risk assessment, and the Superfund risk assessments did not extend beyond this realm, but the early assessments helped to establish the rationale and framework for consideration of multiple stressors.
Similarly, the 1996 amendments to the Safe Drinking Water Act required consideration of chemical mixtures in drinking water by explicitly stating that EPA shall conduct studies that “develop new approaches to the study of complex mixtures … especially to determine the prospects for synergistic or antagonistic interactions that may affect the shape of the dose-response relationship of the individual chemicals or microbes” (Pub. L. No. 104-182, 104th Cong. [1996]). These approaches have been most commonly applied to disinfection byproducts (DBPs): characterization of multiple routes of exposure to multiple DBPs with the same MOA, physiologically based pharmacokinetic models for each individual DBP, and risk characterization that used relative potency factors to aggregate across constituents (Teuschler et al. 2004). Although aggregate exposure assessments have been thoroughly constructed and the combination of dose addition for chemicals with similar MOAs and response addition for mixtures with different MOAs helped to expand the scope of the assessments, the scope of cumulative risk assessment did not consider nonchemical stressors, and insight about synergistic or antagonistic effects remained minimal. Uncertainty quantification was also minimal, and variability was characterized for some components of the risk assessment (such as heterogeneity in food and water consumption) but not others (such as vulnerability).
An important recent example of cumulative risk assessment was related to the Food Quality Protection Act (FQPA), which explicitly required EPA to assess aggregate exposures to pesticides across multiple exposure routes and to consider the cumulative effects of exposures to pesticides with the same MOAs (Pub. L. No. 104-170, 104 Cong. [1996]). Key work completed to date has included a cumulative risk assessment of organophosphorus (OP) pesticides (EPA 2006a). Given the fact that the OP pesticides have a common MOA (inhibition of cholinesterase activity), a cumulative assessment of all pesticides in the family was used. Components of the analysis that deviated from single-chemical risk assessment included consideration of coexposures through various exposure pathways (that is, in the case of a given food item, which pesticides are likely to be found together), consideration of aggregate exposures across multiple pathways, and calculation of relative potency factors to allow cumulative noncancer hazard indexes to be calculated. That work produced among the most detailed and comprehensive cumulative risk assessments conducted to date. However, no evidence was available to determine potential deviations from dose additivity, to incorporate pharmacokinetics explicitly into the dose-response assessment, or to consider interactions with nonchemical stressors or vulnerability other than mandated safety factors of 10 for infants and children. In addition, uncertainty quantification was not extensive, and the focus on margin-of-exposure calculations for individual routes of exposure makes it difficult to quantify the magnitude of harm at various exposure levels (as discussed in Chapter 5). As a general point, most publications in the peer-reviewed literature related to cumulative risk assessment have focused on pesticide health risks both because of the structure of the FQPA and because of availability of data on pesticides.
A final example of cumulative risk assessment in EPA is the National-Scale Air Toxics Assessment, an attempt to estimate the cancer and noncancer health effects of joint exposure to air toxics across the United States. The most recent assessment (EPA 2006b) considered 177 air toxics, used atmospheric-dispersion models to estimate concentrations on the basis of a national emissions inventory, linked the concentrations to population exposure, and estimated health risks. Cancer risks were calculated individually for each compound, given inhalation unit risks from EPA’s Integrated Risk Information System database and other resources; synergistic and antagonistic effects were not considered. Noncancer effects were determined by estimating reference concentrations (RfCs) and adding the hazard quotients of individual compounds that had similar adverse health effects (not necessarily similar MOAs). Thus, the analysis clearly captured multiple agents or stressors, but, like the previous applications, did not introduce evidence beyond simple additivity, did not consider nonchemical stressors or vulnerability, and did not provide extensive insight about uncertainties. The study is also an example of the importance of characterizing exposures to multiple compounds in the current and modified noncancer risk-assessment frameworks: acrolein concentrations exceeded the RfC for a majority of the U.S. population, and this implies that other respiratory irritants (in spite of being below their individual RfCs) were considered to contribute to population health risks.
Thus, in part because of the risk-management questions and regulatory issues historically facing EPA, cumulative risk assessments to date have largely focused on aggregate exposure assessment and have generally not considered nonchemical stressors. However, in segments of EPA and the stakeholder community interested in environmental justice, discussions about cumulative risk assessment have focused on different dimensions of the methodology and extended beyond aggregate chemical-exposure issues. For example, a 2004 National Environmental Justice Advisory Council (NEJAC) report provided guidance about the short-term and long-term actions that EPA should take to implement the concepts in its Framework for Cumulative Risk Assessment with a focus on environmental justice (NEJAC 2004; Hynes and Lopez 2007). Among the important insights in the report were
- The need to distinguish between cumulative risks and cumulative impacts; although the report does not formally define these terms, both are mentioned explicitly throughout.
- The importance of considering nonchemical stressors in the context of a community assessment.
- The significance of vulnerability as a critical component of cumulative risk assessment, including differential sensitivity and susceptibility, differential exposure, differential preparedness to respond to an environmental insult, and differential ability to recover from the effects of an insult or stressor.
- The significance of community-based participatory research to implement cumulative risk assessment, both for capacity-building and to incorporate local data and knowledge into the analysis.
- The need to avoid analytic complexity that seriously delays decision-making and, in parallel, the value of efficient screening and priority-setting tools that can be used by all stakeholders and the necessity of qualitative information in domains where quantitative assessment is not likely in the near term.
The NEJAC report emphasized risks to communities, so some of the components (such as community-based participatory research) may not be applicable to national-scale or other broad-based cumulative risk assessments. Although cumulative risk assessment and community-based risk assessment have many features in common, they are not identical. Other components emphasized in the NEJAC report (such as explicit consideration of vulnerability and having a level of analytic complexity appropriate for the decision context) can be generalized beyond cumulative risk assessment to all forms of risk assessment, as stated in earlier chapters (such as Chapters 3 and 5). Regardless, the NEJAC report emphasized that multiple stakeholders perceive that the potential of cumulative risk assessment as articulated by EPA has not yet been met, primarily because many of the dimensions beyond aggregate chemical exposure assessment have not been formally incorporated.
Related to those issues are recent efforts at EPA to develop tools and techniques for community-based risk assessment, including assessment in the Community Action for a Renewed Environment program (EPA 2008a). Resources and simplified approaches for risk-based priority-setting are made available to communities (EPA 2004), but the approaches do not yet consider key dimensions of cumulative risk, such as nonchemical stressors, vulnerability, or multiple routes of exposure.
A final setting outside EPA in which the general concepts of cumulative risk assessment have been applied is the assessment of the global burden of disease related to environmental and other risk factors. It may not be directly relevant to EPA, given the primary focus on multifactorial global risk rankings (including many nonenvironmental stressors), but it provides some additional lessons related to the analytic challenges and potential information value of assessments that consider an array of diverse risk factors. As articulated by Ezzati et al. (2003), these global burden of disease analyses estimate the population attributable fractions associated with various risk factors, defined as the proportional reductions in population disease or mortality that would occur if exposure to a given risk factor were reduced to an alternative exposure scenario. The risk factors in question are as varied as diet, physical activity, smoking, and environmental and occupational exposures. Given the number of factors considered and the desire to develop indicators applicable to numerous countries (Ezzati et al. 2003), the methods used in connection with any individual risk factor were relatively simple. For example, the burden of disease associated with urban air pollution was estimated on the basis of particulate-matter concentrations, and the concentration-response function from a cohort mortality study in the United States was applied to all countries included in the analysis. The analytic methods took account of potential interactions between risk factors and distinguished between situations in which the direct effects of a risk factor are mediated through intermediate factors, in which effect modification occurs, and in which effects may be independent but exposures may be correlated. The analyses demonstrated approaches in which relatively simplified exposure and dose-response assessment could be applied to yield insight about relative contributions to disease patterns and approaches by which interactions among risk factors could be considered. However, it is important to note the considerable opportunities for mischaracterization of factors when attributable-risk methods are used (Cox 1984, 1987; Greenland and Robins 1988; Greenland 1999; Greenland and Robins 2000), and these issues may grow in significance when the marginal benefits of control strategies are considered.
In conclusion, cumulative risk assessment has been applied in EPA and elsewhere in an increasing number of contexts over the past two decades, and, given the recent development of the Framework for Cumulative Risk Assessment and growing interest in numerous arms of EPA, the applications are expected to grow. The studies have generally been thorough in modeling distributions of aggregate exposures (albeit with limited characterization of uncertainty), and the approach to evaluate cumulative risk posed by multiple chemicals with similar MOAs has been developed reasonably as well (although with generally modest treatment of synergistic and antagonistic effects). However, cumulative risk assessments have generally not yet reached the potential implied by the stated definition; there has been less than optimal formal consideration of nonchemical stressors, aspects of vulnerability, background processes, and other factors that could be of interest to stakeholders concerned about effects of cumulative exposures. Stakeholder involvement has not been as comprehensive as guidelines would indicate would be optimal in most of the above applications, and the tools have not yet been developed to allow communities to engage in even simplified cumulative risk assessment (screening methods are generally restricted to single media and standard risk-assessment practice). Cumulative risk assessment has also been used to determine the risks posed by baseline exposures rather than the benefits of various risk-management strategies, and this use has implications for the methods developed and their interpretations.
Some of the omissions can be attributed to the fact that formal consideration of numerous simultaneous chemical, physical, and psychosocial exposures with evaluation of background disease processes and other dimensions of vulnerability could quickly become analytically intractable if the standard risk-assessment paradigm is followed, both because of the computational burden and because of the likelihood that important exposure and dose-response data will be missing. That points toward the need for simplification of risk-assessment tools in the spirit of iterative risk assessment, and it emphasizes that cumulative human health risk assessment could learn a great deal from such fields as ecologic risk assessment and social epidemiology, which have had to grapple with similar issues related to evaluation of the effects of numerous stressors on defined populations or geographic areas. The expanded scope of cumulative risk assessment that would be theoretically desired includes many elements outside EPA’s standard practice, expertise, and regulatory functions, so there is clearly a need to define carefully how nonchemical stressors and aspects of vulnerability should most appropriately be considered. The following sections present approaches that can be used to expand the scope of cumulative risk assessment while keeping in mind the need for timeliness and EPA’s regulatory mandates, in part by developing screening tools and by orienting analyses around well-defined risk-management objectives.
APPROACHES TO CUMULATIVE RISK ASSESSMENT
From the definitions and examples above, it is clear that cumulative risk assessment has a broad scope and an extremely ambitious mandate. In fact, it is difficult to imagine any risk assessment in which it would not be important to understand the effects of coexposures to agents or stressors that have similar MOAs (as articulated in Chapter 5) or to identify characteristics of the affected populations that could contribute to vulnerability to a given exposure. That is salient in a context of risk management, in which numerous chemical and nonchemical stressors could be simultaneously affected. The critical challenge from the perspective of the risk assessor is to devise an analytic scope and a level of complexity that are appropriate to the context in which cumulative risk assessment is used. Following some of the approaches outlined below could allow EPA to incorporate the aforementioned dimensions of cumulative risk assessment.
A few general approaches have been proposed in the literature; the most appropriate approach clearly is driven by the problem and decision context. Using approaches from ecologic risk assessment, Menzie et al. (2007) develop one type of application, an effects-based assessment. In this case, epidemiologic analyses or general surveillance data provide an indication that a defined population may be at increased risk, and the objective of the analysis is to determine which stressors influence the observed effects. An effects-based assessment is retrospective, so it does not fit neatly into a risk-management framework in which various control options are being weighed; but there are contexts in which strategies would be developed around specific end points, and many of the methods could be generalized to other approaches (including stressor-based assessments, as described below).
Menzie et al. recommend that risk assessors begin with a conceptual model that considers the subset of stressors that are plausibly associated with the health outcomes or other effects of interest. That step would dovetail with the proposed MOA assessment steps proposed in Chapter 5, including MOA evaluation, background and vulnerability assessment, and selection of a conceptual model, but beginning with the health outcome rather than the individual chemical. The next step proposed by Menzie et al. would be a screening assessment to determine a manageable number of factors that are most likely to contribute substantially to the observed effects; this is based in part on simple comparisons with reference values or discussions with stakeholders, and it may be a crucial element of the planning and scoping for the analysis. Stressors are then evaluated individually, then in combination without consideration of interactions, and finally with consideration of interactions and a reliance in part on standard epidemiologic techniques. Although many characteristics are shared by this approach and epidemiologic assessment, this is not identical with proposing that a formal site-specific epidemiologic investigation be conducted. In many community circumstances, epidemiologic investigations will not have adequate statistical power to link defined environmental exposures with observed health outcomes. However, epidemiologic concepts could be useful in framing the analysis and providing insight into the subset of stressors that merit more careful consideration, and knowledge could be leveraged from previously conducted epidemiologic studies. The primary value of this approach is that it emphasizes the need for characterization of coexposures and background processes that could influence the health outcomes of interest and the need to conduct initial screening assessments to construct an analytically tractable model.
A more common approach to risk management would be a stressor-based assessment, in which the cumulative risk assessment is initiated not by questions about the stressors that may explain observed or hypothesized health effects but by questions about the effects that may be associated (generally in a prospective assessment) with a defined set of stressors. A stressor-based assessment would often arise in a source-oriented analysis, in which stakeholders wish to assess the effects of a source (or the benefits of control strategies that address the source) but want to take account of the full array of chemical and nonchemical stressors that have similar health effects. The framework proposed (Menzie et al. 2007) begins with a conceptual model and involves a screening assessment followed by consideration of individual stressors followed by interactions among stressors, but a stressor-based assessment begins with the stressors and identification of the populations and end points that would be influenced by them. The MOA assessment steps outlined above would be central to this process, in that they would help to characterize the end points of interest, the related stressors, and factors that could influence variability in response to the stressors.
An important modification in the approach to cumulative risk assessment that could potentially alleviate some of the analytic challenges would involve an orientation around evaluation of risk-management options rather than characterization of problems (see Chapter 8 for a more extensive discussion of this proposed framework). The approaches presented above and most previous case examples would help to determine which stressors are of greatest concern with respect to a defined outcome in a defined subpopulation or what the burden of disease is in the context of simultaneous exposure to a number of stressors. However, cumulative risk assessment would be most valuable to both communities and decision-makers when it can provide information about the health implications of alternative control options. For example, a community may be choosing among alternatives for drinking-water disinfection, and it would be important to consider the effects of the changes in concentrations of all disinfection byproducts jointly, to consider simultaneous exposure to a number of waterborne pathogens, to consider all routes of exposure to key compounds of interest, and to identify vulnerable populations. Many of the analytic tools would be similar, but in a decision context different factors may be correlated or affected on the margin from those when baseline conditions are considered, and the stressors that are important to include may also differ. In other words, it is important to include a stressor only to the extent that it will influence the estimated benefits of a control strategy either in its estimation or in its interpretation. In principle, focusing on stressors relevant to risk-management strategies will help to ensure that analyses are aligned with EPA’s mandated focus on chemical or biologic stressors while acknowledging the influence of nonchemical stressors. A modified version of the stressor-based paradigm from Menzie et al. oriented around discriminating among risk-management options is presented in Table 7-1.
TABLE 7-1
Modified Version of Stressor-Based Cumulative-Risk-Assessment Approach from Menzie et al. (2007) Oriented Around Discriminating among Risk-Management Options.
Following that approach would have multiple fringe benefits. For example, evaluating background exposures and vulnerability factors will not only allow cumulative risk assessment after the committee’s proposed revisions to the cancer and noncancer dose-response assessment paradigm (Chapter 5) but will also provide information that can be used in environmental-justice analyses focused on inequality in outcomes and help to bring risk assessment and environmental justice into a single analytic framework (Levy et al. 2006; Morello-Frosch and Jesdale 2006). The geospatial components of the exposure and vulnerability assessment could be mapped to communicate key information to stakeholders, who would be engaged throughout the analytic process in a community risk setting. Most important, as alluded to above, the approach would potentially result in a need to model only a subset of stressors formally; the remainder would contribute to a general understanding about background processes but would otherwise not need to be quantitatively characterized to determine the benefits of risk-management options.
In spite of the benefits, there clearly are limitations of both the bottom-up stressor-based and top-down effects-based approaches. In cumulative risk assessment, the scope and complexity of the problem can quickly exceed the capacity of stressor-based analyses, although the approach outlined above can help to maintain focus on the key stressors. Given the analytic challenges, there is a temptation to think that effects-based analyses would be more practical even though risk-management decisions are often stressor-based. However, the size and subtlety of the effects are generally beyond the reach of standard epidemiologic tools. The relative influence of stressor-based vs effects-based analyses clearly will depend on the problem framework, including the decision context and the geographic scale of the analysis.
In addition, although the proposed approaches provide guidance on how a complex system can be systematically evaluated to develop an analytically tractable cumulative risk assessment, data limitations may make quantitative analyses impractical for some cumula tive risk assessments. In ecologic risk assessment, a rank-oriented approach has been used in a relative-risk model (RRM) to account for the fact that addressing cumulative effects of multiple chemical and nonchemical stressors may not otherwise be viable. The RRM was developed to evaluate simultaneously and comparatively the risk posed by multiple, dissimilar stressors to multiple receptors in heterogeneous environments on landscape scales. It was first developed in 1997 for an ecologic risk assessment of chemical stressors at Port Valdez, AK (Landis and Wiegers 1997) and later applied successfully to other risk assessments of ecosystems on various scales and with other stressors and end points (Landis et al. 2000; Obery and Landis 2002). One of its specific strengths is an ability to incorporate stakeholders’ values readily in evaluating risks in multiple geographic areas with multiple stressors, habitats, and receptors. Although originally designed for ecologic concerns, risk to humans can be readily accommodated in its flexible framework.
Similarly, in the realm of social epidemiology, the complexities of simultaneous exposures to numerous physical and social environmental factors have been addressed in some applications with cumulative risk models based on summing dichotomous classifications (for example, 1 if more than one standard deviation above the mean for a given risk factor, otherwise 0) for numerous risk factors of interest. Those indicators are acknowledged as not capturing the relative weights of the various factors, but they avoid the need for numerous multiplicative interaction models and have been shown to be more predictive of health end points than single-risk-factor models (Evans 2003). When data are sufficient, more refined approaches based on relative risks rather than simply distributions of exposures may be useful.
A disadvantage of the approaches is their focus on ranking and scoring systems where weights do not necessarily correspond with relative risks, which can be difficult to interpret in situations where different risk-management strategies lead to different combinations of risk factor reductions without one strategy leading to greater reductions for all risk factors. Practices that move away from quantitative risk characterizations within a core component of risk assessment should be considered and implemented judiciously because the applicability and interpretability of the resulting assessments in a decision context can be severely limited. At a minimum, ranking approaches should be evaluated for their sensitivity to key input assumptions, and in settings where quantitative information is available, these approaches could be helpful in initial assessments for organizing information and determining whether a solution can be easily chosen or more complex analysis is needed to distinguish among options.
KEY CONCERNS AND PROPOSED MODIFICATIONS
The EPA cumulative risk-assessment paradigm recognizes an important issue and provides a useful conceptual framework, but substantial logistical barriers remain, and some core issues are largely unaddressed by the current framework. For example, as articulated by EPA (2003a), that about 20,000 pesticide products are on the market and 80,000 existing chemicals are on the Toxic Substances Control Act inventory makes it impractical to try to account for all relevant synergisms and antagonisms. More broadly, cumulative risk assessment requires extensive information beyond chemical toxicity and MOAs, including aggregate exposure data and information on population characteristics and nonchemical stressors. Therefore, EPA concludes in its Framework for Cumulative Risk Assessment that “identification of critical information and research needs may be the primary result of many cumulative risk assessment endeavors” (EPA 2003a, p. xii).
That statement may be correct, and it does reflect one important aim of risk assessment (to provide insight about key uncertainties that should be addressed to discriminate among risk-management options), but it implies that cumulative risk assessment would be largely uninformative for near-term decision-making, and this is a matter of concern, given the salience of the questions asked by cumulative risk assessment from the perspective of many stakeholders. The committee feels that the conclusion understates the value of less complex but more wide-ranging risk assessments and ignores the fact that an analysis focused on specific mitigation measures in a community will potentially have a more narrow scope than an attempt to characterize relative contributors to the burden of disease (as described in Table 7-1). That is, although there may be numerous theoretical combinations of exposures, only a subset will be relevant in choosing among various intervention options for a well-defined problem.
We propose below a series of short-term and long-term efforts, focusing on measures that could enhance the utility of cumulative risk assessment in the context of environmental decision-making.
Clarification of Terminology
Although the definition of cumulative risk assessment as articulated by EPA is comprehensive and well crafted, the fact that a cumulative risk assessment as defined (including nonchemical stressors and vulnerability) has never been done in the agency raises questions about whether the definition is practical in the near term without some modifications of current practice. The Framework for Cumulative Risk Assessment was published relatively recently, but research and regulatory action related to cumulative risks have been conducted for decades without much advancement beyond chemical stressors in a small number of contexts. In addition, the ways in which cumulative risk assessment is being considered vary greatly among offices in EPA and among different stakeholder groups, and this indicates the need for greater clarity in its aims and scope.
We propose that EPA explicitly define and maintain a conceptual distinction among cumulative risk assessment, cumulative impact assessment, and community-based risk assessment, which overlap but are conflated in many discussions. The terms have been defined (CEQ 1997) and recently discussed (NEJAC 2004), but a clear and consistent delineation of EPA’s interpretation of the boundaries and degree of overlap would help to reduce confusion about the intended scope of any given assessment.
The committee proposes that cumulative risk assessment be defined as evaluating an array of stressors (chemical and nonchemical) to characterize—quantitatively to the extent possible—human health or ecologic effects, taking account of such factors as vulnerability and background exposures. Cumulative impact assessment would consider a wider array of end points, including effects on historical resources, quality of life, community structure, and cultural practices (CEQ 1997), some of which may not lend themselves to quantification following the Risk Assessment in the Federal Government: Managing the Process (NRC 1983; the Red Book) paradigm and are beyond the scope of the present report. Community-based risk assessment would follow the practices and principles of community-based participatory research (CBPR), involving active engagement of the community throughout the entire assessment process (Israel et al. 1998).
Although those are conceptually distinct definitions, there will be overlaps in practice. For example, it will often be desirable to use CBPR approaches in cumulative risk assessments, although in principle a community-based risk assessment might not address cumulative risks, and a cumulative risk assessment (such as the pesticide analyses under the FQPA) may not always follow CBPR approaches. Similarly, cumulative impact assessments would generally include the outputs of cumulative risk assessment and other considerations; but, depending on the nature of the decision, the quantitative cumulative risk component may have more or less significance in a cumulative impact assessment.
The definition of cumulative risk assessment above is meant to be functionally identical with that of cumulative risk assessment in the Framework for Cumulative Risk Assessment (EPA 2003a) and that of cumulative impact assessment1 by the California Environmental Protection Agency (CalEPA 2005). This difference in nomenclature further emphasizes the need for clear definitions. In addition, although it is preferable to have quantitative information as the primary health risk-assessment output, it will often be useful to provide qualitative information about potential health effects when risks cannot be fully quantified and to have terminology that distinguishes the full discussion of possible health effects from the myriad other effects that may be considered in a cumulative impact assessment and that may be important for a decision at hand.
We further propose that EPA apply the term cumulative risk assessment only to an analysis that considers in some capacity all the components mentioned in EPA’s definition of cumulative risk assessment. An analysis that does not consider nonchemical stressors, that considers only a subset of routes and pathways of exposure, or that does not consider vulnerability should not be termed a cumulative risk assessment. This does not imply that all cumulative risk assessments will formally quantify all of these dimensions - if an initial screening assessment or qualitative examination demonstrates that it is not necessary to consider nonchemical stressors, vulnerability, or specified routes of exposure given a defined decision context, they need not be included in the final assessment for it to be deemed a cumulative risk assessment. That may appear to be a largely semantic distinction, but it emphasizes the primary aims and objectives of cumulative risk assessment and would encourage EPA and other investigators to develop methods to address the aforementioned elements when relevant to a regulatory decision. The committee recognizes that these modified definitions may run counter to the language in the FQPA and elsewhere; this may make redefinition impractical in the near term, but this inconsistency within the agency reinforces the need for greater clarity. Following these modified definitions, many of the previous assessments by EPA and others would be more appropriately termed mixture risk assessments, inasmuch as they consider aggregate exposures to multiple chemicals in the same family but do not consider other components mentioned above. To be clear, this does not imply that such assessments were not well done or informative for policy decisions, as analyses of the effects of chemical mixtures can have great utility, but simply that they do not answer the same questions asked by cumulative risk assessment.
More generally, EPA should emphasize that even cumulative impact assessment cannot by itself bridge the gap between community concerns about environmental risks and decisions made by EPA and other stakeholders. Some communities are concerned principally about the cumulative burden of environmental exposures or the local burden of disease, but others may be more concerned about unfairness in siting processes, ensuring that low-socio-economic-status (low-SES) communities are at the table with other stakeholders articulating their concerns, and so forth. Some of those concerns can be addressed through cumulative impact assessment, but not all of them. EPA should recognize that cumulative impact assessment has the potential to greatly inform concerns related to outcomes but cannot by itself address concerns about process (although, as articulated later, stakeholder involvement is a crucial component of cumulative risk assessment and cumulative impact assessment, which could help to address some process concerns). The clarification about the decision contexts in which cumulative impact assessment will and will not be useful should provide more realistic expectations on the part of all stakeholders.
Integrating Nonchemical Stressors
In spite of the fact that cumulative risk assessment by definition considers psychosocial, physical, and other factors, no cumulative risk assessments by EPA have formally incorporated nonchemical stressors. That may be in large part because data have been inadequate and because many nonchemical stressors are beyond EPA’s regulatory mandate, but the omission means that cumulative risk assessment has a much narrower scope than originally expected or desired by many stakeholders. Moreover, as illustrated in the global analyses of burden of disease described above, data are available on the effects of a number of dietary, physical, and psychosocial risk factors, and extensive exposure data are available on many of these stressors. In addition, ecologic risk assessments commonly apply methods that simultaneously consider numerous chemical and nonchemical stressors in a single assessment in spite of the complexity of the system and the limitations of data availability. In this sec tion, we give examples of some data sources that EPA could use to incorporate nonchemical stressors and use a case example to demonstrate the utility of a cumulative risk assessment that includes nonchemical stressors.
An initial recommendation is that EPA develop databases and default approaches that would allow the incorporation of key nonchemical stressors in the absence of population-specific data. From an exposure perspective, a parallel effort would be the Exposure Factors Handbook (EPA 1997), which synthesizes extensive data from disparate sources to allow default estimates of activity patterns and intake rates for defined subpopulations. EPA should work to synthesize and develop datasets related to exposures to nonchemical stressors that influence similar health end points as key chemical stressors to allow these factors to be readily incorporated into cumulative risk assessments in settings where population-specific assessment is infeasible or impractical. Emphasis should be on characterization of distributions for key subpopulations and on evaluation of correlations between factors to allow more realistic assessments. For some factors (for example, smoking, diet, and alcohol consumption), extensive data are already available from other sources, such as the National Health and Nutrition Examination Survey (NHANES), but would need to be compiled and processed in a format suitable for cumulative risk assessment. For example, cumulative risk assessments may require information about correlations between exposures to chemical and nonchemical stressors (cross-sectional or longitudinal), which may not generally be calculated and compiled for other purposes. Factors such as temperature and humidity (which may interact with air pollution effects) and various infectious agents would similarly have readily-available data sets which may require additional analysis to be incorporated into cumulative risk assessments. In general, EPA should collaborate with other agencies and organizations with more expertise in nonchemical stressors to build these databases.
For other factors (such as psychosocial stress), additional methodologic research and data-collection efforts would potentially be needed. With individual stressors for which exposures could be quantified, EPA should compile relevant data related to socioeconomic status (SES), which may serve as a proxy for numerous individual risk factors (O’Neill et al. 2003) and may be a more direct measure of vulnerability than could reasonably be assembled by looking at all relevant individual risk factors. The key is to understand correlations between SES and exposure-related activities and later the degree to which SES acts as an effect modifier for given chemical stressors and health outcomes. Efforts such as these may be beyond the expertise and purview of EPA in the near term, and knowledge of other agencies (such as CDC) and stakeholders should be leveraged.
Incorporating nonchemical stressors also requires information on modes of action among disparate types of exposures. EPA not only should focus on pharmacokinetic and pharmacodynamic models and approaches typically used in MOA determinations for chemicals (following the modified approach proposed in Chapter 5) but should make use of epidemiologic evidence on effect modification when it is available. For example, there may be epidemiologic studies that demonstrate differential relative risks by SES (for example, risk of death related to particulate matter) or interactions between smoking status and chemical exposures (for example, to radon). The importance of epidemiologic evidence can be seen by considering socioeconomic factors and stressors, which could not be incorporated by using evidence only from animal bioassays. Although direct epidemiologic evidence may not be available on a specific chemical of interest, insight from similar compounds may provide useful default assumptions about interactions between chemical and nonchemical stressors. The potential importance of epidemiology in cumulative risk assessment is discussed in more detail later in this chapter.
To illustrate how a cumulative risk assessment could in principle capture both chemical and nonchemical stressors while maintaining focus on the subset of stressors influenced by risk-management strategies under study, we provide an illustrative example. Suppose that the risk-management decisions in question were related to various strategies to reduce the public-health effects of airport emissions on the surrounding communities. Some of the strategies (such as changes in fuel composition or control technologies) would influence only air-pollution exposures and related health risks, and others (such as changes in flight paths or runway use) could also influence noise exposures and related psychosocial stress. The committee recognizes that such an example does not neatly correspond with EPA’s regulatory mandates and would cross the jurisdictional boundaries of multiple agencies; this example is simply meant to illustrate the steps that would need to be taken within a cumulative risk assessment.
Following the paradigm proposed in Table 7-1, the first step in a stressor-based assessment oriented around risk-management options would involve building a conceptual model to provide insight into the various stressors of concern and their linkages with the health outcomes of interest. Given a focus on evaluating the benefits of proposed risk-management strategies rather than burden-of-disease assessments, the stressors of interest should include either the ones that would be influenced differentially by potential risk-management strategies or the ones that would not be influenced by the strategies but would have a quantitative influence on risk estimates.
In this case, it clearly would be important to include psychosocial stress as a key non-chemical stressor in at least two dimensions. First, it would be important to know whether the effect of air-pollution exposure reductions depended at all on the level of psychosocial stress (related to noise and other causes). That would be important even for the interventions that did not influence psychosocial stress, provided that the level of psychosocial stress influenced the effects of changes in air pollution (that is, by contributing to background processes or acting as an effect modifier). Second, it would be important to develop the quantitative relationship between interventions and levels of psychosocial stress, as a potential cobenefit of risk-management strategies targeted at air pollutant emissions. If the effect of air pollution were independent of the level of psychosocial stress and the interventions did not have any differential influence on psychosocial stress, it would not be an important stressor to consider in this decision context even if it were an important contributor to the general burden of disease.
Given that structure, the approach in Table 7-1 involves a MOA assessment and consideration of background exposures that may affect the same health outcome. A comprehensive evaluation for this case is beyond the scope of the present report, but one example could involve cardiovascular disease as a significant end point of concern and hypertension as the mechanistic link between the various stressors and this end point. Previous studies (Evans et al. 1998) have demonstrated that airport noise and the associated stress can increase blood pressure (and epinephrine, norepinephrine, and cortisol). Air pollution has similarly been associated with blood pressure (Künzli and Tager 2005), and this indicates that both exposures would be important to model, given either an underlying model linking hypertension and cardiovascular end points or a quantal cutpoint for hypertension itself. It would also be necessary to be able to model the relationship between the risk-management strategies and exposures to both air pollution and noise. Following the conceptual model makes this relatively straightforward: methods are readily available to model the influence of airport activities on noise (which could be presumed to be a surrogate for airport-related psychosocial stress), and the aforementioned studies can link noise with such key health-relevant end points as blood pressure. Thus, a physiologically based conceptual model can readily incorporate nonchemical stressors into the cumulative risk assessment.
That example was simplified and did not formally go through all the steps in Table 7-1; for example, characterizing the baseline distribution of blood pressures in the population would be necessary, as would characterizing the distribution of other underlying disease processes to which the stressors could contribute. Other issues are potentially raised by the above approach, such as a focus on only pathways that are well understood and quantifiable, as well as the complexity of a real-world case that would potentially involve multiple federal agencies. However, in spite of those concerns, this simple example demonstrates the general feasibility of the approach and highlights that a focus on specific risk-management strategies would greatly narrow the scope of the analysis. Often, more epidemiologic evidence is available on nonchemical stressors than on chemical stressors, so inclusion of nonchemical stressors may be plausible in many contexts.
The inclusion of nonchemical stressors as outlined above can lead to more informative assessments and correspondingly better decisions if used appropriately but can run the risk of contributing to less informative assessments if used in the wrong way. Information on the varied risk factors should not be used solely for risk comparisons that are uninformative from the perspective of the decisions faced by EPA. For example, if the inputs for cumulative risk assessment are used not to determine the impacts of alternative risk-management strategies but to determine contributors to disease burdens in a community, analyses may find that cigarette-smoking confers a greater disease burden than outdoor exposures to air toxics. Even setting aside the risk-communication limitations of such a comparison (given the different nature of the risks), the comparison is largely uninformative from the perspective of EPA, industry, or other agency decision-making. In other words, it is difficult to imagine a context in which EPA must decide whether to require industrial facilities to install pollution-control devices or to lobby other agencies to increase funding for smoking-cessation efforts. The problem would be avoided by the framework proposed in detail in Chapter 8, in that a focus on options to achieve a defined objective (that is, a functional-unit definition) would make these sorts of burden-of-disease comparisons less relevant. The simple fact that stressors other than chemicals may contribute a substantial portion of the burden of disease in a community does not by itself imply that reduction of chemical exposures would not have net benefits that would exceed the costs, and an emphasis on this comparative dimension of the analysis will only widen the gulf between risk assessors and community stakeholders. This is not to say that there is no rationale for risk communication efforts that attempt to contextualize risk assessment outputs by comparison with other risk factors, but simply to emphasize that such comparisons should not be the primary intent of cumulative risk assessment.
In addition, especially with nonchemical stressors, such as psychosocial stress, analytic boundaries need to be carefully established. If the existence of industrial facilities or other environmental problems serves as a social stressor, control strategies could reduce both chemical exposures and psychosocial stress (provided that the affected community perceived the reduction as important and substantive). The Agency for Toxic Substances and Disease Registry (ATSDR) recently (Tucker 2002) emphasized the psychosocial ramifications of living near hazardous-waste sites and the potential need to consider psychosocial factors in remediation decisions, although these factors are rarely formally quantified or characterized. That raises the broader question of whether stress related to an environmental exposure should “count” as part of quantifying the benefits of an intervention. Counting those benefits would in principle provide a more accurate estimate of benefits, but one could imagine a situation in an extreme case in which a community is greatly concerned about a chemical in its drinking water that has no direct effects on health but in which an intervention measure could result in health benefits through the reduction of psychosocial stress. That would be somewhat more important than a placebo effect, but it would be awkward to estimate health benefits associated with controlling a benign chemical. Such extreme cases should be avoided by well-formulated problem scoping and risk-management option development, but the example highlights the importance of stakeholder involvement at multiple stages in the assessment process.
Finally, even with the triage indicated in Table 7-1, addition of all relevant chemical and nonchemical stressors runs the risk of making the assessment analytically intractable and impossible to complete in a limited amount of time and of jeopardizing timely decision-making. In addition to limiting the number of stressors under consideration, there is a need for relatively simple risk-assessment methods that can be applied to address the stressors in a timely fashion; this is discussed in more detail later.
In summary, approaches to incorporate nonchemical stressors into cumulative risk assessment are feasible in the near term although there are many situations in which site-specific data needs may not be met. We recommend that EPA start to address nonchemical stressors in settings in which sufficient epidemiologic or pharmacokinetic and pharmacodynamic data are available to understand interactions with chemical stressors, following the tiered strategy articulated by Menzie et al. (2007) and reoriented in Table 7-1 to focus on discriminating among risk-management options. Databases and default approaches should be developed regarding exposure patterns and plausible interactions with chemical stressors. In the long term, we recommend that EPA and other agencies invest in research related to interactions between chemical and nonchemical stressors, including epidemiologic investigations and pharmacokinetic and pharmacodynamic or other study types as relevant. The direction of the research should be informed by pending risk-management decisions in which the agency identifies critical data gaps that impede decision-making in specific contexts rather than broadly considering all the combinations of chemical and nonchemical stressors that could potentially be investigated.
Role of Biomonitoring
As summarized recently (Ryan et al. 2007), biomonitoring has a potentially important role in cumulative risk assessment, with significant roles to be played by biomarkers of exposure, susceptibility, and effect. For example, if multiple stressors are thought to influence acetylcholinesterase inhibition (that is, in the case of OP pesticides), simultaneous collection of compound-specific biomarkers, nonspecific biomarkers of the OP family, and biomarkers of effect can provide insight into the joint effects of these exposures. Collection of biologic samples can allow characterization of simultaneous exposure to multiple stressors, which may be difficult to determine accurately by modeling exposures to each of the compounds individually.
Ryan et al. (2007) view the primary capabilities of biomonitoring in the framework of cumulative risk assessment as the ability to disaggregate disease burden into specific risk factors and the ability to infer contributions of different sources and pathways. The former approach provides one route for effects-based or burden-of-disease assessments, and the latter approach can in principle inform stressor-based and later cumulative risk assessments focused on interventions.
A potential limitation of biomonitoring data is the difficulty of linking biomarkers to contributions from individual sources of emissions. Even if the distribution of biomarkers of exposure or effect is well characterized for a defined subpopulation, including an understanding of routes of exposure and contributing source categories, it is difficult to model how changes in emissions from a small number of identified sources would influence the distribution. Biomarkers may therefore be suitable for developing mechanistic understanding and contributing to effects-based cumulative risk assessment but may be of limited use to stressor-based cumulative risk assessment directly in a risk-management context, especially in situations with relatively small marginal changes in exposures. Research efforts related to reverse dosimetry (Sohn et al. 2004; Tan et al. 2006) indicate a possible approach to reconstructing exposures from dose data, but such methods are not sensitive enough to determine marginal changes in emissions from individual facilities and therefore may not be suitable for discriminating among risk-management options for more narrowly-defined or community-scale control strategies. In this context, biomonitoring may be most useful as a validation check against modeled doses or as an input to epidemiologic investigations.
Regardless, the existence of the Centers for Disease Control and Prevention (CDC) large-scale biomarker databases, the Third National Report on Human Exposure to Envi ronmental Chemicals (CDC 2005), indicates that data on the distribution of doses among representative samples of the U.S. population are increasingly available. The full set of data available through the NHANES could also provide a means of characterizing correlations between biomarkers for chemical and nonchemical stressors, demographic predictors of magnitudes of those stressors, and other relationships that could form the basis of a cumulative risk assessment. Thus, although it seems unlikely, because of both cost and limited interpretability, that biomarkers could be used directly to quantify the benefits of control strategies leading to marginal changes in exposures, biomarker studies can provide enhanced mechanistic understanding of the relationships among chemical and nonchemical stressors, and insight about highly-exposed populations or source category contributions that can allow for the development of targeted control strategies.
Role of Epidemiology and Surveillance Data
The cumulative risk-assessment paradigm, given its focus on communities or defined populations and consideration of such nonchemical stressors as SES and access to health care, lends itself to being informed by epidemiology. In fact, many of the key interactions among chemical and nonchemical stressors, given numerous simultaneous coexposures, would be impossible to capture in toxicologic studies. The call for more “realistic” risk assessment in community settings is in part a call for better epidemiology that can characterize the effects of varied coexposures in the presence of background processes and differences in vulnerability. This raises the question of whether sufficient epidemiologic information is available, or could be developed, to enable EPA to generate cumulative risk assessments that include physical, chemical, biologic, and social factors with a sufficient degree of scientific plausibility. This section briefly provides examples of advances in epidemiologic methods that show promise for improving the information base needed for the advancement of cumulative risk assessment, and in parallel it describes the role that surveillance data and systems could play in facilitating the transition from single chemical risk assessment to cumulative risk assessment.
At the outset, limitations of epidemiology in the context of cumulative risk assessment must be acknowledged. Because of relatively low ambient exposures, multiple concurrent exposures, weak statistical power, exposure misclassification, and other issues, it is often difficult for epidemiology to capture main effects, let alone interaction effects, of environmental exposures. In spite of those limitations, there is growing epidemiologic evidence of interactions between environmental stressors and place-based and individual-based psychosocial stressors, driven in part by the spatial and demographic concordance between physical and chemical environmental exposures and socioeconomic stressors (IOM 1999; O’Neill et al. 2003; Clougherty et al. 2007). The evidence adds to historical examples of well-documented interactions between environmental and nonenvironmental risk factors in humans, such as synergistic effects between radon or asbestos and cigarette-smoking. In addition, by definition, reliance on epidemiology reduces the ability to be preventive and to evaluate the risk of new stressors to which humans have not yet been exposed. Epidemiology is best suited to cumulative risk assessments directed at remediation of existing problems, which would be expected to be the majority of applications, given the inherent focus on populations at risk.
Two growing categories of inquiry in epidemiology may help to bolster the evidence base and inform cumulative risk assessment. Problems of characterizing exposure and outcomes in observational epidemiology have generated increasing attention to molecular epidemiology, which involves incorporating biologic events at the physiologic, cellular, and molecular levels into epidemiologic studies. Aside from enhancing the biologic understanding of epidemiologic findings, the biomarkers used in molecular epidemiology can be used in some circumstances to reconstruct exposure (albeit with some of the limitations listed above). The combination of better exposure assessment and better understanding of disease pathways helps to reduce their misclassification in epidemiologic studies. That provides better statistical power and biologic insight that can improve characterization of potential synergies among risk factors and factors that contribute to vulnerability, including age, sex, inherited genetic variation, nutrition, and pre-existing health impairments. Such studies, although it may be difficult to apply them directly to quantitative population risk assessment, may have a greater likelihood of detecting subtle effects in relatively small populations and demonstrating the biologic plausibility of synergistic relationships.
A somewhat different direction of epidemiologic inquiry potentially informative for cumulative risk assessment involves the emerging field of social epidemiology, which has shed light on the relations between social factors and disease in populations (Kaufman and Cooper 1999). There is little room for disagreement about the importance of “social factors” as predictors of health risks; the consistent documentation of these patterns in a wide variety of outcomes is an important achievement of health and medical science. Of significance for cumulative risk assessment is the recent work of social epidemiologists who are examining the biologic underpinnings of social factors and considering interactions with environmental exposures (Berkman and Glass 2000). Aside from elucidating those interactions, social epidemiology may provide methodologic lessons for cumulative risk assessment in general; as mentioned above, methods have been developed to characterize cumulative risks (Evans 2003), and studies addressing allostatic load (the long-term effect of the various physiologic responses to stress) have both considered the effects of numerous stressors and developed measures of allostatic load that integrate multiple outcomes (McEwen 1998).
To benefit from developments in molecular and social epidemiology and related sciences and technology with the potential to reduce exposure-measurement error (that is, environmental sensors, biologic sensors, and geographic information systems), there will need to be greater interactions between epidemiologic research and risk assessment, as opposed to treating risk assessment simply as an end user of epidemiologic output. Epidemiologic studies conducted with cumulative risk assessment in mind may use different exposure-assessment and analytic strategies from those used by epidemiologic studies conducted for other purposes. For example, an epidemiologic analysis done for its own sake will tend to focus on disentangling the contributions of individual risk factors in the presence of potential confounding, whereas an epidemiologic analysis done for cumulative risk assessment might characterize the risks of defined “bundles” of exposures without further decomposition.
The interaction between epidemiology and cumulative risk assessment can be enhanced as risk assessments identify key uncertainties related to interactions among chemical and nonchemical stressors, shaping the research agenda and stimulating demand for more relevant (to risk assessment) epidemiologic research. In general, as mentioned above, EPA and other agencies should pursue a long-term research agenda related to enhanced epidemiologic insight into interactions among chemical and nonchemical stressors and in the short term should work to develop internal capacity in a variety of epidemiologic disciplines to foster the development of new methods and knowledge.
Although epidemiologic approaches may improve understanding of the effects of exposure to multiple stressors, for effects-based assessments, surveillance data may be needed both to identify the at-risk populations and to characterize patterns of disease and background exposures. Surveillance for various diseases is well established in the public-health system, including monitoring networks and registries that collect data in several ways. For example, nearly all states have some form of infectious-disease and chronic-disease reporting laws that require hospitals, physicians, or schools to report cases that are considered to be of public-health importance to the state or to CDC. Such information is available at various levels of spatial resolution, influenced in part by confidentiality considerations and by the nature and prevalence of the disease in question. In addition, federal agencies, such as CDC, maintain active or passive surveillance on a wide variety of diseases and health-status measures for populations in various geographic areas. A relatively new component of public-health surveillance involves biosurveillance, the early detection of abnormal disease patterns and nontraditional early disease indicators, such as pharmaceutical sales, school and work absences, and cases of animal disease.
Another form of surveillance system is the toxic-substance registry. As mandated by Superfund legislation, the ATSDR established a National Exposure Registry (ATSDR 2008) with the goal of assessing and evaluating relationships between adverse health effects and exposure to hazardous waste, particularly between chronic health effects and long-term, low-level chemical exposure. For example, NER’s trichloroethylene subregistry has been used to demonstrate increased rates of hearing impairment and other conditions associated with historical exposure to trichloroethylene.
Those surveillance systems have substantial utility in some contexts but have been limited in multiple respects in the context of environmental risk factors. In particular, little information has been routinely and systematically collected on many health outcomes potentially linked to environmental pollutants, such as birth defects, developmental disorders, childhood leukemia, and lupus. More generally, many chronic diseases (such as diabetes and asthma) have not been given sufficient attention. In addition, given numerous data streams, it has been difficult to relate members of populations included in one health-information system to members in another system.
For those reasons, CDC in 2001 began the development of a health-tracking network to monitor the prevalence of chronic conditions of potential interest for human health risk assessment. Known as the Environmental Public Health Tracking (EPHT) Program, its purpose is to provide information from a nationwide network of integrated health and environmental data to be used as the basis of risk assessment and risk management. An important distinction between EPHT and traditional surveillance is the emphasis on data integration across health, human-exposure, and hazard-information systems, which will enhance efforts of risk assessors to evaluate the spatial and temporal relations between environmental factors and health outcomes. If the EPHT surveillance systems were linked with registries from private health-care organizations, more comprehensive disease-prevalence estimates could be readily obtained.
Of particular interest to the cumulative risk-assessment process is the potential of EPHT to identify susceptible populations and to provide an important foundation for environmen tal epidemiology addressing chemical and nonchemical stressors. Developing the relations between environmental and health outcomes will require individual-level data not routinely collected by any surveillance system, so there will be the need for both targeted research and methods for data linkage with the EPHT Program. In general, the goals of EPHT are ambitious and resources are limited, in particular for data-linkage efforts that are expensive in both time and money (Kyle et al. 2006). Investing more resources in EPHT could be a useful mechanism to develop the information base necessary for cumulative risk assessment or community-based risk assessment.
Need for Simpler Analytic Tools
Given the breadth of exposure pathways and types of stressors considered in cumulative risk assessment, there is a danger that it could become analytically intractable and therefore uninformative for making decisions in a timely fashion. Application of more advanced methods for dose-response assessment as proposed in Chapters 4 and 5 would appear to make this issue even more problematic. The problem is more acute in community-based risk assessments, in which the sheer number of communities and environmental risks that could potentially be evaluated could quickly outstrip the available resources for conducting such analyses and in which the CBPR emphasis implies that analytic tools should be able to be understood and implemented by community stakeholders. It should be clear that not all decisions will need to be informed by the most advanced analytic methods (see Chapters 3 and 8), just as not all risk-management decisions will necessarily involve quantifying all theoretical dimensions of cumulative risk assessment.
To enhance the utility of cumulative risk assessment, there will need to be increased reliance on relatively simple methods to determine whether more refined methods are required or information is adequate to inform policy decisions. Developing simpler tools seems to contradict the complexity of cumulative risks, but methods can be developed that capture the breadth of chemical and nonchemical stressors with less computational burden, at least for initial screening calculations. There will also need to be techniques to develop indicators or ranking approaches that could categorize the benefits of different strategies ordinally as has been done in ecologic risk assessment; for example, Thomas (2005) has shown that the RRM, a rank-based method, can be used to analyze alternative decisions involving multiple stressors and receptors on various spatial scales. The critical issue is to ensure that any simplified methods used in the context of cumulative risk assessment retain the key attributes of quantitative risk assessment, that is, consideration of both exposure and toxicity, notions of probability rather than just possibility, and information about the severity of health effects. It will be difficult to interpret outputs that do not retain those features, especially in the contexts of tradeoffs or comparisons with control costs.
While development of simpler approaches will not be straightforward, fields such as ecologic risk assessment and life cycle analysis have successfully developed and utilized tools to address similar concerns, and these methods will be relevant to cumulative risk assessment. One example focused on exposure assessment comes from the field of intake-fraction estimation (Bennett et al. 2002a). An intake fraction is the population exposure per unit of emission from a defined source or source category. Intake fractions are generally derived from dispersion modeling or from the combination of monitoring data and emissions-inventory assessment, in either case linked with population patterns. They therefore use detailed information about exposures but summarize this information as single unitless measures directly interpretable for risk assessment; in cases in which the dose-response function is linear in the range of exposures of interest or is well defined and nonlinear, intake fractions can be used directly to estimate population health risks. Intake fractions vary with the compound, source, and setting, but values can be extrapolated to unstudied settings given known characteristics of the setting (such as population density). Intake fractions have been adopted by the life-cycle analysis community for incorporating population-exposure concepts in settings where more complex modeling is implausible and where the alternative is priority-setting with no consideration of exposure (Bennett et al. 2002b; Evans et al. 2002). As another example of simplified methods for exposure assessment in the context of screening-level risk assessment, the Community Air Screening How-To Manual (EPA 2004) includes look-up tables for concentration effects, given stack characteristics and distance from a source.
Although those approaches address only exposure assessment, they provide useful lessons about how simpler methods can be applied to yield reasonable and timely insight without sacrificing the critical components of quantitative risk assessment. The concept of using a limited number of more extensive analyses to determine approximate relationships for an unstudied setting can be extended to exposures to nonchemical stressors or interactions among compounds. This can provide effective defaults in the absence of more detailed site-specific data. The committee therefore recommends that EPA develop guidelines and methods for less analytically complex cumulative risk assessments to be used for screening assessments. The guidelines should give insight into approaches for choosing the appropriate level of analytic complexity and into recommended methods for simplified assessments, including both exposure assessment and dose-response assessment. The selection of the appropriate analytic model would be a component of the planning and scoping and problem-formulation steps and would be driven by the risk-management decisions at hand and the priorities of the various stakeholders. In other words, drawing on the example above, simplifying exposure assessment by using intake fractions is valuable only if total population benefits without distributional considerations were the measure of interest to risk managers. The simplified tools would need to be tailored to the decision context and the outputs of interest.
The databases, methods and other modeling resources developed by EPA for less analytically complex cumulative risk assessments would have an important ancillary benefit. Local community participation could be greatly enhanced if analytic tools were easier to understand or, ideally, could be used by community groups and other stakeholders to determine the benefits of control strategies in a cumulative risk context quickly but reasonably. That is clearly difficult given the numerous decision contexts and types of models required, but examples could be drawn from the life-cycle analysis community, in which generally applicable software packages and on-line resources have been developed that can be used by people who lack expertise in the specific scientific disciplines that underlie life-cycle impact assessment. The general issue of the need for and approaches to enhancing stakeholder involvement in cumulative risk assessment is discussed in more detail in the next section.
Need for Stakeholder Involvement
The issue of increased stakeholder involvement in the risk-assessment process has been discussed at length in previous National Research Council reports and EPA guidance documents. The committee agrees with many of the core principles articulated in those reports, such as the mutual and recursive analytic-deliberative process articulated in Understanding Risk Informing Decisions in a Democratic Society (NRC 1996) and the need for stakeholder participation throughout the risk-assessment process, including participation in planning and scoping and in problem formulation (EPA 2003b). A key insight from the previous reports is that stakeholder involvement should go well beyond risk communication or risk characterization and should include substantive involvement in the assessment process (often following CBPR principles) and explicit attempts to build capacity to ensure that all stakeholders have an equal opportunity to participate substantively in collaborative problem-solving (NEJAC 2004). That is not simply a means of improving public relations and acceptability of risk-assessment outputs but a means of enhancing the technical quality of the analysis and ensuring that risk-management strategies are reasonable and well developed.
The cumulative risk-assessment framework further emphasizes the value of bringing stakeholders together at the outset, devising clear and explicit project planning and scoping, and focusing on a specific decision problem to guide the analysis. However, the added complexity of cumulative risk assessment creates some substantial barriers: if there is to be substantive stakeholder involvement, all parties must have access to and in-depth understanding of relevant databases, models, and information resources. It is not realistic to hope that all stakeholders will become expert risk assessors, but the use of simpler analytic tools, as proposed above, may provide some of the necessary resources for community members and other stakeholders to understand and participate in the analytic portions of an assessment.
In addition to models for cumulative risk assessment, information resources would need to be developed to allow stakeholders to be sufficiently informed to participate in the process. EPA has developed a substantial array of public resources and databases, but none provides adequate information to allow stakeholders to understand the intricacies of cumulative risks in specific communities or subpopulations. For example, EPA has made available such public resources as Envirofacts (EPA 2007a), EnviroMapper (EPA 2006c), and TRI Explorer (EPA 2007b), which provide extensive information about the locations of key emission sites for any given ZIP code, information about environmental-justice assessments, and links to related concentration data. However, none of the available resources provides the information or tools needed for stakeholders to understand their cumulative risks associated with chemical and nonchemical stressors or, more important, the potential benefits associated with specific control strategies. Models of the benefits of control strategies may be beyond the scope of on-line resources, but well-developed and publicly available databases could provide both the foundation for cumulative risk models and the information for communities to use in understanding their exposures and background disease patterns. Linking environmental databases described above with surveillance-system data in a framework of geographic information systems would be a good starting point for such efforts, using high spatial resolution to provide maximal insight into community-scale risks.
EPA has numerous programs and guidance documents related to stakeholder involvement (EPA 2008b), whose formal evaluation is beyond the scope of this chapter. The committee recommends that EPA adhere to its guidance when conducting cumulative risk assessments, including planning and budgeting for public and other stakeholder involvement, working to identify interested parties, providing financial or technical assistance and resources to facilitate involvement, providing information and outreach materials, engaging in other activities to build community capacity to participate in the process, involving the public in the decision process at a stage where substantive input can be made, and formally evaluating the process to ensure that adequate stakeholder participation (in depth and breadth) has been incorporated (EPA 2003b).
RECOMMENDATIONS
The committee recommends the following short-term and long-term actions to enhance the utility of cumulative risk assessment for discriminating among risk-management options:
- EPA should maintain the core definitional components of cumulative risk assessment from its 2003 framework document—including planning, scoping, and problem-formulation phases; explicit consideration of vulnerability; and the use of screening tools and other methods to ensure analytic complexity appropriate for the decision context. The analytic structure of ecologic risk assessment should continue to serve as an important guide for human health cumulative risk assessment, given the conceptual similarities.
- EPA should use a revised framework for risk-based decision making (see Chapter 8), focused on discriminating among risk-management options, to narrow the scope of cumulative risk assessments to those stressors that would be influenced by risk-management options or would modify the risks of other stressors influenced by risk-management options. This would allow for the inclusion of nonchemical stressors within a decision framework relevant to EPA. For stressor-based assessments, EPA should follow a tiered assessment strategy that parallels the mode-of-action and background-process determination to ascertain the subset of stressors that would substantially influence the benefits of proposed risk-management strategies.
- EPA should explicitly define and maintain conceptual distinctions among cumulative risk assessment, cumulative impact assessment, and community-based risk assessment to avoid confusion about the scope of work expected of a given assessment. These definitions should be consistently used and applied across the agency.
- In the near term, EPA should develop databases and default approaches to allow the incorporation of key nonchemical stressors in cumulative risk assessments in the absence of population-specific data, considering exposure patterns, contributions to relevant background processes, and interactions with chemical stressors. EPA should use existing nationally representative biomarker and surveillance databases and databases related to nonchemical stressors to help to construct the approaches, leveraging insight from social epidemiology and ecologic risk assessment.
- In the long term, EPA should invest in research programs and develop internal capacity related to interactions between chemical and nonchemical stressors, including epidemiologic investigations with sufficient power to evaluate interactions and physiologically based pharmacokinetic and other study types as relevant. Given the need for substantial epidemiologic research conducted in a form and direction suitable for cumulative risk assessment, EPA should build internal capacity in various epidemiologic disciplines and ensure close collaboration between epidemiologists and risk assessors. EPA should also develop partnerships with other federal agencies with expertise related to nonchemical stressors, and should work with these agencies on large-scale cumulative risk assessments that cross jurisdictional boundaries.
- In the process of refining cumulative risk assessment, EPA should focus on development of guidelines and methods for simplified analytic tools that could allow screening-level cumulative risk assessment and could provide tools for communities and other stakeholders to use in conducting assessments. These tools can be used as the foundation of an enhanced stakeholder-participation process that builds on current guidance but expands it by providing cumulative risk models that can be applied and interpreted by nonpractitioners. EPA should work to ensure that cumulative risk assessments both guide future information and research needs and inform near-term decisions, recognizing that decisions must be made with incomplete information.
REFERENCES
- ATSDR (Agency for Toxic Substances and Disease Registry). 2008. National Exposure Registry [online]. Available: http://www
.atsdr.cdc.gov/ner/index.html [accessed Aug. 12, 2008]. - Bennett, D.H., T.E. McKone, J.S. Evans, W.W. Nazaroff, M.D. Margni, O. Jolliet, and K.R. Smith. 2002. a. Defining intake fraction. Environ. Sci. Technol. 36(9):207A-211A. [PubMed: 12026996]
- Bennett, D.H., M.D. Margni, T.E. McKone, and O. Jolliet. 2002. b. Intake fraction for multimedia pollutants: A tool for life cycle analysis and comparative risk assessment. Risk Anal. 22(5):905-918. [PubMed: 12442988]
- Berkman, L.F., and T.A. Glass. 2000. Social integration, social networks, social support and health. Pp. 137-173 in Social Epidemiology, L.F. Berkman, editor; , and I. Kawachi, editor. , eds. New York, NY: Oxford University Press.
- CalEPA (California Environmental Protection Agency). 2005. Cal/EPA EJ Action Plan Pilot Projects Addressing Cumulative Impacts and Precautionary Approach. California Environmental Protection Agency. March 25, 2005 [online]. Available: http://www
.calepa.ca .gov/envjustice/ActionPlan/ [accessed Jan. 28, 2008]. - Callahan, M.A., and K. Sexton. 2007. If cumulative risk assessment is the answer, what is the question? Environ. Health Perspect. 115(5):799-806. [PMC free article: PMC1867996] [PubMed: 17520071]
- CDC (Centers for Disease Control and Prevention). 2005. Third National Report on Human Exposure to Environmental Chemicals. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, Atlanta, GA: [online]. Available: http://www
.jhsph.edu /ephtcenter/Third%20Report.pdf [accessed Jan. 24, 2008]. - CEQ (Council on Environmental Quality). 1997. Considering Cumulative Effects under the National Policy Act. Council on Environmental Quality, Executive Office of the President, Washington, DC. January 1997 [online]. Available: http://www
.nepa.gov/nepa /ccenepa/ccenepa.htm [accessed Jan. 28, 2008]. - Clougherty, J.E., J.I. Levy, L.D. Kubzansky, P.B. Ryan, S.F. Suglia, M.J. Canner, and R.J. Wright. 2007. Synergistic effects of traffic-related air pollution and exposure to violence on urban asthma etiology. Environ. Health Perspect. 115(8):1140-1146. [PMC free article: PMC1940095] [PubMed: 17687439]
- Cox, L.A., Jr. 1984. Probability of causation and the attributable proportion of risk. Risk Anal. 4(3):221-230.
- Cox, L.A., Jr. 1987. Statistical issues in the estimation of assigned shares for carcinogenesis liability. Risk Anal. 7(1):71-80. [PubMed: 3615994]
- EPA (U.S. Environmental Protection Agency). 1986. Guidelines for Health Risk Assessment of Chemical Mixtures. EPA/630/R-98/002. Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, DC. September 1986 [online]. Available: http://www
.epa.gov/ncea /raf/pdfs/chem_mix/chemmix_1986.pdf [accessed Jan. 7, 2008]. - EPA (U.S. Environmental Protection Agency). 1997. Exposure Factors Handbook. EPA/600/P-95/002F. National Center for Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC: [online]. Available: http://www
.epa.gov/ncea/efh/ [accessed Jan. 28, 2008]. - EPA (U.S. Environmental Protection Agency). 2003. a. Framework for Cumulative Risk Assessment. EPA/600/P-02/001F. National Center for Environmental Assessment, Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, DC: [online]. Available: http://cfpub
.epa.gov /ncea/cfm/recordisplay.cfm?deid=54944 [accessed Jan. 4, 2008]. - EPA (U.S. Environmental Protection Agency). 2003. b. Public Involment Policy of the U.S. Environmental Protection Agency. EPA 233-B-03-002. Office of Policy, Economics and Innovation, U.S. Environmental Protection Agency. May 2003 [online]. Available: http://www
.epa.gov/stakeholders /pdf/policy2003.pdf [accessed Jan. 28, 2008]. - EPA (U.S. Environmental Protection Agency). 2004. Community Air Screening How-To Manual, A Step-by-Step Guide to Using Risk-Based Screening to Identify Priorities for Improving Outdoor Air Quality. EPA 744-B-04-001. U.S. Environmental Protection Agency, Washington, DC: [online]. Available: http://www
.epa.gov/oppt /cahp/pubs/community _air_screening_how-to_manual.pdf [accessed Jan. 28, 2008]. - EPA (U.S. Environmental Protection Agency). 2006. a. Organophosphorus Cumulative Risk Assessment-2006 Update. Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, DC. August 2006 [online]. Available: http://www
.epa.gov/pesticides /cumulative/2006-op/index.htm [accessed Jan. 28, 2008]. - EPA (U.S. Environmental Protection Agency). 2006. b. 1999 National-Scale Air Toxics Assessment. Technology Transfer Network, U.S. Environmental Protection Agency; [online]. Available: http://www
.epa.gov/ttn/atw/nata1999/ [accessed Jan. 28, 2008]. - EPA (U.S. Environmental Protection Agency). 2006. c. EnviroMapper. U.S. Environmental Protection Agency; [online]. Available: http://www
.epa.gov/enviro/html/em/ accessed Jan. 29, 2008]. - EPA (U.S. Environmental Protection Agency). 2007. a. Envirofacts Data Warehouse. U.S. Environmental Protection Agency; [online]. Available: http://www
.epa.gov/enviro/ [accessed Jan. 29, 2008]. - EPA (U.S. Environmental Protection Agency). 2007. b. TRI Explorer. Toxics Release Inventory, U.S. Environmental Protection Agency; [online]. Available: http://www
.epa.gov/triexplorer/ [accessed Jan. 29, 2008]. - EPA (U.S. Environmental Protection Agency). 2008. a. Community Action for a Renewed Environment (CARE). Office of Air and Radiation, U.S. Environmental Protection Agency; [online]. Available: http://www
.epa.gov/air/care/index.htm [accessed Jan. 29, 2008]. - EPA (U.S. Environmental Protection Agency). 2008. b. Tools for Public Involvement. U.S. Environmental Protection Agency; [online]. Available: http://www
.epa.gov/stakeholders /involvework.htm [accessed Jan. 29, 2008]. - Evans, G.W. 2003. A multimethodological analysis of cumulative risk and allostatic load among rural children. Dev. Psychol. 39(5):924-933. [PubMed: 12952404]
- Evans, G.W., M. Bullinger, and S. Hygge. 1998. Chronic noise exposure and physiological response: A prospective study of children living under environmental stress. Psychol. Sci. 9(1):75-77.
- Evans, J.S., S.K. Wolff, K. Phonboon, J.I. Levy, and K.R. Smith. 2002. Exposure efficiency: An idea whose time has come? Chemosphere 49(9):1075-1091. [PubMed: 12492166]
- Ezzati, M., S.V. Hoorn, A. Rodgers, A.D. Lopez, C.D. Mathers, and C.J. Murray. 2003. Estimates of global and regional potential health gains from reducing multiple major risk factors. Lancet 362(9380):271-280. [PubMed: 12892956]
- Greenland, S. 1999. Relation of probability of causation to relative risk and doubling dose: A methodologic error that has become a social problem. Am. J. Public Health 89(8):1166-1169. [PMC free article: PMC1508676] [PubMed: 10432900]
- Greenland, S., and J.M. Robins. 1988. Conceptual problems in the definition and interpretation of attributable fractions. Am. J. Epidemiol. 128(6):1185-1197. [PubMed: 3057878]
- Greenland, S., and J.M. Robins. 2000. Epidemiology, justice and the probability of causation. Jurimetrics 40(3):321-340.
- Hynes, H.P., and R. Lopez. 2007. Cumulative risk and a call for action in environmental justice communities. J. Health Disparities Res. Pract. 1(2):29-57.
- IOM (Institute of Medicine). 1999. Toward Environmental Justice: Research, Education and Health Policy Needs. Washington, DC: National Academy Press. [PubMed: 23035313]
- Israel, B.A., A.J. Schulz, E.A. Parker, and A.B. Becker. 1998. Review of community-based research: Assessing partnership approaches to improving public health. Annu. Rev. Public Health 19:173-202. [PubMed: 9611617]
- Israel, B.D. 1995. An environmental justice critique of risk assessment. New York U. Environ. Law J. 3(2):469-522.
- Kaufman, J.S., and R.S. Cooper. 1999. Seeking causal explanations in social epidemiology. Am. J. Epidemiol. 150 (2):113-120. [PubMed: 10412955]
- Kuehn, R.R. 1996. The environmental justice implications of quantitative risk assessment. U. Illinois Law Rev. 1996(1):103-172.
- Künzli, N., and I.B. Tager. 2005. Air pollution: From lung to heart. Swiss Med. Wkly. 135(47-48):697-702. [PubMed: 16511705]
- Kyle, A.D., J.R. Balmes, P.A. Buffler, and P.R. Lee. 2006. Integrating research, surveillance, and practice in environmental public health tracking. Environ. Health Perspect. 114(7):980-984. [PMC free article: PMC1513336] [PubMed: 16835047]
- Landis, W.G., and J.A. Wiegers. 1997. Design considerations and a suggested approach for regional and comparative ecological risk assessment. Hum. Ecol. Risk Assess. 3(3):287-297.
- Landis, W.G., M. Luxon, and L.R. Bodensteiner. 2000. Design of a relative risk model regional-scale risk assessment with conformational sampling for the Willamette and McKenzie Rivers, Oregon. Pp. 67-88 in Environmental Toxicology and Risk Assessment: Recent Achievements in Environmental Fate and Transport, Vol. 9., F.T. Prince, editor; , K.V. Brix, editor; , and N.K. Lane, editor. , eds. STP1381. West Conshohocken, PA: American Society for Testing and Materials.
- Levy, J.I., S.M. Chemerynski, and J.L. Tuchmann. 2006. Incorporating concepts of inequality and inequity into health benefits analysis. Int. J. Equity Health 5:2. [PMC free article: PMC1459160] [PubMed: 16569243]
- McEwen, B.S. 1998. Protective and damaging effects of stress mediators. N. Engl. J. Med. 338(3):171-179. [PubMed: 9428819]
- Menzie, C.A., M.M. MacDonell, and M. Mumtaz. 2007. A phased approach for assessing combined effects from multiple stressors. Environ. Health Perspect. 115(5):807-816. [PMC free article: PMC1868003] [PubMed: 17520072]
- Morello-Frosch, R., and B.M. Jesdale. 2006. Separate and unequal: Residential segregation and estimated cancer risks associated with ambient air toxics in U.S. metropolitan areas. Environ. Health Perspect. 114(3):386-393. [PMC free article: PMC1392233] [PubMed: 16507462]
- NEJAC (National Environmental Justice Advisory Council). 2004. Ensuring Risk Reduction in Communities with Multiple Stressors: Environmental Justice and Cumulative Risks/Impacts. National Environmental Justice Advisory Council, Cumulative Risks/Impacts Working Group. December 2004 [online]. Available: http://www
.epa.gov/enforcement /resources /publications/ej/nejac /nejac-cum-risk-rpt-122104.pdf [accessed Jan. 29, 2008]. - NRC (National Research Council). 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: National Academy Press. [PubMed: 25032414]
- NRC (National Research Council). 1996. Understanding Risk: Informing Decisions in a Democratic Society. Washington, DC: National Academy Press.
- Obery, A.M., and W.G. Landis. 2002. A regional multiple stressors assessment of the Codorus Creek watershed applying the relative risk model. Hum. Ecol. Risk Assess. 8(2):405-428.
- O’Neill, M.S., M. Jerrett, I. Kawachi, J.I. Levy, A.J. Cohen, N. Gouveia, P. Wilkinson, T. Fletcher, L. Cifuentes, and J. Schwartz. 2003. Health, wealth, and air pollution: Advancing theory and methods. Environ. Health Perspect. 111(16):1861-1870. [PMC free article: PMC1241758] [PubMed: 14644658]
- Ryan, P.B., T.A. Burke, E.A. Cohen Hubal, J.J. Cura, and T.E. McKone. 2007. Using biomarkers to inform cumulative risk assessment. Environ. Health Perspect. 115(5):833-840. [PMC free article: PMC1867975] [PubMed: 17520075]
- Sohn, M.D., T.E. McKone, and J.N. Blancato. 2004. Reconstructing population exposures from dose biomarkers: Inhalation of trichloroethylene (TCE) as a case study. J. Expo. Anal. Environ. Epidemiol. 14(3):204-213. [PubMed: 15141149]
- Tan, Y.M., K.H. Liao, R.B. Conolly, B.C. Blount, A.M. Mason, and H.J. Clewell. 2006. Use of a physiologically based pharmacokinetic model to identify exposures consistent with human biomonitoring data for chloroform. J. Toxicol. Environ. Health A 69(18):1727-1756. [PubMed: 16864423]
- Teuschler, L.K., G.E. Rice, C.R. Wilkes, J.C. Lipscomb, and F.W. Power. 2004. A feasibility study of cumulative risk assessment methods for drinking water disinfection by-product mixtures. J. Toxicol. Environ. Health A 67(8-10):755-777. [PubMed: 15192867]
- Thomas, J.F. 2005. Codorus Creek: Use of the relative risk model ecological risk assessment as a predictive model for decision-making. Pp. 143-158 in Regional Scale Ecological Risk Assessment: Using the Relative Risk Model, W.G. Landis, editor. , ed. Boca Raton, FL: CRC Press.
- Tucker, P. 2002. Report of the Expert Panel Workshop on the Psychological Responses to Hazardous Substances. Office of the Director, Division of Health Education and Promotion, Agency for Toxic Substances and Disease Registry, Atlanta, GA: [online]. Available: http://www
.atsdr.cdc .gov/HEC/PRHS/psych5ed.pdf [accessed Oct. 19, 2007].
Footnotes
- 1
As defined by the California EPA, cumulative impact means exposures, public health, or environmental effects from the combined emissions and discharges in a geographic area, including environmental pollution from all sources, whether single or multi-media, routinely, accidentally, or otherwise released. Impacts will take into account sensitive populations and socio-economic factors, where applicable and to the extent data are available (CalEPA 2005).
- Implementing Cumulative Risk Assessment - Science and DecisionsImplementing Cumulative Risk Assessment - Science and Decisions
Your browsing activity is empty.
Activity recording is turned off.
See more...