Epidemiology and Risk Assessment

JONATHAN M. SAMET
THOMAS A. BURKE

Overview and Definitions

Risk assessment is now a widely used term for a systematic approach to characterizing the risks
posed to individuals and populations by environmental pollutants and other potentially adverse
exposures. Risk assessment is increasingly applied as a translational tool for moving from research
findings to the implementation and evaluation of policies. In the United States, its use is either explicitly
or implicitly required by a number of federal statutes (Table 5-1), and its application worldwide is
mounting. This chapter provides an introduction to risk assessment, focusing on those aspects of its
methodology most pertinent to epidemiologists and addressing the use of epidemiologic data in risk
assessment. The topic is assuming ever greater relevance for epidemiologists as the findings of
epidemiologic research are incorporated into risk assessments both to determine the existence of a
hazard and to gauge its extent of the hazard.

A seminal 1983 National Research Council report, Risk Assessment in the Federal Government:
Managing the Process (often called the "Red Book" because of its cover), defined risk assessment as "the
use of the factual base to define the health effects of exposure of individuals or populations to
hazardous materials and situations (National Research Council [NRC] 1983)." This conceptualization of
risk assessment is both qualitative and quantitative, although quantitative risk assessment should be
considered as a component of risk assessment in its broadest context. The term "risk," as used in the
context of risk assessment, conveys the same meaning as in its standard epidemiologic formulation: the
probability of an event, e.g., disease occurrence, taking place. Depending on the context, risks sustained
by individuals or by populations may be of interest. Some other health-related applications.

Table 5-1. Principal US Environmental Health And Safety Laws: Agencies And Mandates Related To Risk

Statute Responsible Agency Mandate

Federal Food, Drug, and Department of Health and Prohibit, inter alia, distribution
Cosmetic Act, 21 U.S.C. 301 el Human services of foods, food and color
seq. Environmental Protection additives,
Agency
of the term risk, as in "risk adjustment" for underlying disease severity, are not related to risk
assessment.

The 1983 National Research Council report explicitly positioned risk assessment as a tool for
translating the findings of research into science-based risk management strategies (Figure 5-1). Risk
assessment evaluates and incorporates the findings of all relevant lines of investigation, from the
molecular to population levels, through the application of a systematic process with four sequential
steps: hazard identification, dose-response assessment, exposure assessment, and risk characterization
(Table 5-2). If there is no positive determination of the existence of a hazard, then the subsequent steps
are not warranted. Risk assessment also provides a comprehensive framework for bringing together all
relevant information on the existence of a hazard to health and on the magnitude of the hazard. Thus,
the hazard identification step could involve consideration of structure-activity relationships for a toxin,
laboratory findings from in vitro and in vivo experiments, and epidemiologic evidence, dose-response
assessment may also draw on multiple types of data. While the figure separates research from risk
assessment, there is continued interplay between researchers and risk assessors as key gaps in evidence
are identified and research is initiated to address them.

Risk assessments are performed by a variety of institutions. Some are conducted within federal
and state agencies because of mandated requirements (Table 5-1). For example, the Occupational
Safety and Health Administration uses risk assessment to establish that the agent considered for
regulation causes "significant risk of harm," as mandated by the Supreme Court's "Benzene" decision. In
this significant Supreme Court decision, the Occupational Safety and Health Administration was required
to conduct a risk assessment for exposure of workers to benzene in order to show that benzene caused
a significant risk to health. The Environmental Protection Agency, under the 1990 Clean Air Act
amendments, has been required to evaluate the risks of 189 hazardous air pollutants to ensure that
exposures to populations have "an ample margin of safety." Private sector entities including pollutant
emitting industries may also use risk assessment to determine potential consequences of exposures to
workers or to the general population from processes that may lead to environmental contamination.
Voluntary health organizations, such as the American Lung Association, or environmental organizations
may themselves apply risk assessment methods to gauge the magnitude of hazards posed by
environmental toxins and then use the results to promote prevention and influence the public and
policy-makers.

Risk management follows and builds from risk assessment. Risk management involves the
evaluation of alternative regulatory actions and the selection of the strategy to be applied. Risk
communication is the transmission of the findings of risk assessments to the many "stake-holders" who
need to know the results to participate in the policy-making process and to the general public. In this
formalism, and in practice to some degree, those performing the risk assessment—risk assessors—and
those managing the risks— risk managers—are separate groups of professionals and distinct from the
researchers who develop the data used in risk assessments. At present, risk assessors come from diverse
disciplinary backgrounds and many have moved from scientific- or policy-based positions into
conducting risk assessments.
 Risk Risk
Research
Assesment Management
Laboratory and field Development of
observations of adverse health regulatory actions
effects and exposures to Hazard
particular agents Identification

Information on extrapolation Evaluation of public

methods for high to low dose Dose-Response Risk health, economic,
and animal to human Assesment Characterization social, political,
consequences of
regulatory actions

Field measurements, estimated

exposures characterization of Exposure
populations Assesment Agency decisions and
actions
Figure 5-1. Schematic relations among research, risk assessment, and risk management. Source: NCR
(1983)

Table 5-2. The "Red Book" Paradigm: the Four Steps of Risk Assessment

Hazard identification: A review of the relevant biological and chemical

Dose response:
Exposure assessment:
Risk characterization:
information bearing on whether or not an agent
may pose a carcinogenic hazard and whether
toxic effects in one setting will occur in other
settings
The process of quantifying a dosage and evaluating
its relationship to the incidence of adverse
health effects response
The determination or estimation (qualitative or
quantitative) of the magnitude, duration, and
route of exposure
An integration and summary of hazard
identification, dose-response assessment, and
exposure assessment presented with
assumptions and uncertainties. This final step
provides an estimate of the risk to public health
and a framework to define the significance of
the risk

Epidemiologists have diverse potential roles in risk assessment and risk management. First, the
findings of epidemiologic studies may be key in the steps of hazard identification and dose-response
assessment, and may also contribute to exposure assessment. Consequently, epidemiologists may
become partners in the conduct of a risk assessment and assist in assuring that the findings of
epidemiologic studies have been interpreted appropriately by risk assessors. This type of engagement by
epidemiologists may involve committee service, consultation, peer review, or testimony. Second,
epidemiologists may assume the roles of risk assessors or risk managers, leaving behind the primary role
of epidemiologic researcher. Finally, epidemiologists may be called on to devise surveillance systems to
document the consequences of an implemented risk management strategy.

While epidemiologists and epidemiologic data may have prominent roles in risk assessments,
the epidemiologic literature contains surprisingly few discussions of risk assessment. The topic was
treated directly in a 1985 conference (Gordis 1988) and again in a 1994 conference (Graham 1995).
Hertz- Picciotto (1995) provided a cogent summary in a recent review, emphasizing the use of
epidemiologic data to characterize dose-response relationships. Risk assessment in general is covered in
a series of landmark reports on the topic by committees of the US National Research Council: the 1983
report, Risk Assessment in the Federal Government: Managing the Process (NRC 1983); a 1989 report on
risk communication, Improving Risk Communication (NRC 1989); a 1994 report that comprehensively
evaluated risk assessment methods, Science and Judgment in Risk Assessment (known as the "Blue
Book") (NRC 1994a); and a 1996 report on risk characterization, Understanding Risk. Informing Decisions
in a Democratic Society (NRC 1996). Other National Research Council reports address risks of specific
agents. Rodricks (1992) offers a readable introduction that emphasizes toxicologic approaches in
Calculated Risks. The journal Risk Analysis focuses on the topic.

The Evolution of Risk Assessment

The formal characteristics of risk assessment have a brief history (NRC 1994a; Rodricks 1992). While
many core concepts had been developed earlier, the origins of contemporary risk assessment can be
traced to the 1970s when new environmental regulations called for information on risks in order to set
policy. Even earlier, however, the need to protect the general public and workers had led to the
development of methods for setting exposure limits that inherently involved risk estimation. To protect
workers, particularly against short-term toxicity, exposure limits were set that were below levels known
or considered likely to have adverse effects. For example, Threshold Limit Values (TLVs) were first set by
the American Conference of Government Industrial Hygienists (ACGIH). For foods, acceptable daily
intakes (ADIs) of pesticides and food additives were set based on animal assays. The no-observed-effect
level (NOEL) in the assay, subsequently modified to the no-observed-adverse-effect level (NOAEL), was
divided by a safety factor to yield the ADI for people. For foods, a safety factor of 100 was assumed.
These approaches implicitly assume that there is a threshold level of exposure or dose that must be
exceeded for an adverse effect to occur at an unacceptable frequency.

In the 1960s and 1970s, mounting concern about environmental carcinogens accelerated the
use of risk assessment by federal agencies, including the Food and Drug Administration, the
Environmental Protection Agency, and the Occupational Safety and Health Administration. The widening
use of risk assessment, as regulators attempted to manage increasing numbers of chemicals, motivated
the Food and Drug Administration to support the National Research Council Committee, which wrote
the Red Book. The committee had three principal objectives: (1) "to assess the merits of separating the
analytic functions of developing risk assessments from the regulatory functions of making policy
decisions," (2) "to consider the feasibility of designating a single organization to do risk assessments for
all regulatory agencies," and (3) "to consider the feasibility of developing uniform risk assessment
guidelines for use by all regulatory agencies" (NRC 1983).

The committee's response to this charge continues to set the framework for risk assessment and
risk management. The committee recommended a clear conceptual distinction between risk assessment
and risk management (Figure 5-1) and formalized the risk assessment process into the four-step
paradigm (Table 5-2). Paramount for epidemiologists, the report further distinguished research from risk
assessment. The report acknowledged that uncertainties affect risk assessments and that gaps in
knowledge need to be filled by making choices among plausible options, termed "inference options."
The committee also called for the development of uniform guidelines for selecting among inference
options. The committee recognized that the inference option selected could carry policy implications.

Subsequent to the Red Book, use of risk assessment at the federal and state levels increased.
Guidelines for carcinogen assessment and other types of toxicity were published by the Environmental
Protection Agency (US Environmental Protection Agency [US EPA] 1986), which also developed
guidelines for exposure assessment (US EPA 1992). Risk assessment was used as a priority-setting tool
by the Department of Energy in implementing clean-up programs at its nuclear sites; and the Agency for
Toxic Substances and Disease Registry applied risk assessment approaches to contaminated sites
throughout the United States. The Environmental Protection Agency took a risk-based approach in
attempting to assign priorities to the many environmental hazards that it faced. A 1987 report,
Unfinished Business, provided the findings of agency staff on the relative importance of a listing of 31
hazards in four categories of risk: human cancer risk, human noncancer risk, ecological risk, and welfare
risk (US EPA 1987). This type of ranking was needed to consider strategic options, and the report
represented one of the first comprehensive exercises in risk ranking. In the 1990 follow-up report,
Reducing Risk: Setting Priorities and Strategies for Environmental Protection, a committee addressed the
data and methodologies needed for risk ranking (US EPA 1990). The committee's recommendations
emphasized the pervasive need for information on risk in setting environmental policies.

The Clean Air Act Amendments of 1990 required a review by the National Academy of Sciences
of methods used by the Environmental Protection Agency to estimate risk. The review, published in
1994 and entitled Science and Judgment in Risk Assessment (NRC 1994a), provides a summary of the
state of the art in risk assessment as of the early 1990s. It recommended the continued use of risk
assessment but called for an iterative approach that better blends risk assessment with risk
management. The report identified many gaps in the data needed for risk assessment and in the
methods and assumptions made by the agency in conducting its risk assessments. A chapter addressed
use of observational evidence.

The Clean Air Act amendments also mandated the establishment of a Commission on Risk
Assessment and Risk Management that would "make a full investigation of the policy implications and
appropriate uses of risk assessment and risk management in regulatory programs under various Federal
laws to prevent cancer and other chronic human health effects which may result from exposure to
hazardous substances." The commission's report comments that risk assessment has become more
refined analytically butmnotes that risk assessments done for regulation tend to give insufficient
attention to risk reduction and improving health (The Presidential/Congressional Commission on Risk
Assessment and Risk Management 1997). It proposes a new framework for risk management that places
collaboration with stakeholders at the center. Risk assessment remains key, but risks should be placed
into the broad context of public health, and comparisons should be made to other risks to the
population.

A similar broadening of the Red Book framework was proposed in the 1996 report of a
committee of the National Research Council: Understanding Risk. Informing Decisions in a Democratic
Society (NRC 1996). This report extended the concept of risk characterization articulated in the Red
Book. Like the draft report of the Commission on Risk Assessment and Risk Management, this report
noted that a broad context needs to be set for risk characterization and recommended broad
participation in risk characterization from all stakeholders. It called for an iterative process of analyses
and deliberation and for determining the concerns and perceived risks of stakeholders as the risk
assessment is initiated. A risk characterization, to be informative, may need to be expressed along
multiple dimensions, and not be limited to a simple numeric expression of harm, e.g., the number of
excess cancers. It should be aimed at informing the decision process and solving problems.

In the mid-1990s, we seem poised for a broadening use of risk assessment in developing public
policies, particularly those involving environmental regulation. In fact, in 1994, the 104th Congress of
the United States began with clamor for "regulatory reform," and most draft reform bills gave risk
assessment a central role in the setting of regulations and in evaluating the costs and benefits of
regulations. There was an accompanying call for "sound science" as the basis for risk assessments. Risk
assessment has also assumed increasing importance as a regulatory tool at state and local levels, and
internationally.

In spite of its brief history, risk assessment has already gained substantial notoriety; so, too, has
the use of epidemiologic data for risk assessment purposes. Examples of controversial and debated
applications of the method and the use of epidemiologic data include the Environmental Protection
Agency's risk assessments for environmental tobacco smoke (US EPA 1992a) and for radon (US EPA
1992b), and the assessment of the risk of workplace exposure to environmental tobacco smoke
conducted by the Occupational Safety and Health \ Administration (US Department of Labor and
Occupational Safety and Health Administration [US OSHA] 1994). Consequently, have been published
evaluations of the utility of epidemiologic data for risk assessment purposes (Graham 1995) and
attempts to develop guidelines for their use (Federal Focus Inc. 1996). These guidelines are covered in
this chapter.

For epidemiologists, the lessons to be learned from this brief review should include recognition
that risk assessment is now ensconced as a policymaking tool and that epidemiologic data may have a
central role in setting policies that have substantial societal implications. Legislative trends indicate that
risk assessment will likely gain prominence as a tool for translating epidemiologic research into public
policy.

Epidemiology and Risk Assessment

The value of epidemiologic data for risk assessment has been widely discussed (Graham 1995). Pundits
argue that epidemiologic data are rarely relevant and too often flawed by poor quality and uncontrolled
biases (Graham 1995). Epidemiologic studies have also been deemed uninformative given the "weak
associations" anticipated for typical levels of exposure to many environmental agents. Proponents of
epidemiology, while acknowledging the limitations of observational studies, advance its strengths: the
investigation of the effects of real exposures as received by the population; the characterization of
effect across the full range of susceptibility in the population; and, above all, the direct relevance of
epidemiologic evidence to public health (Hertz-Picciotto 1995; Burke 1995). The debate on the role of
epidemiologic evidence in risk assessment has proceeded both generally and specifically, and risk
assessment findings for individual agents, such as radon and environmental tobacco smoke, have been
questioned. Guidelines for epidemiologic research to be used in risk assessment have been offered as
one solution to strengthening the evidence base (Hertz-Picciotto 1995; Federal Focus Inc. 1996; Auchter
1995). The guidelines largely echo principles that are already tenets of the field.