A Risk-Benefit Framework for Assessing Intentional Human Dosing Studies
As discussed in Chapter 2, the regulatory framework for human research imposes a number of fundamental conditions: (1) exposure of participants to any risk must be scientifically necessary; (2) risks to participants must be minimized; (3) the potential benefits from the research must justify any risks participants may face; (4) selection of participants must be equitable; (5) participants must give informed consent; and (6) an independent board must give prior approval to the research design and monitor compliance with procedures to protect participants. Research results submitted to the Environmental Protection Agency (EPA) must satisfy these conditions as a minimum condition for acceptability. This chapter examines the risks and benefits of intentional human dosing studies and considers when the benefits may justify the risks.
Comparing risks and benefits in human experiments is a critical and often difficult task. The National Bioethics Advisory Commission (NBAC) observed that “there are no clear criteria for IRBs to use in judging whether the risks of research are reasonable in relation to what might be gained by the research participants” (NBAC, 2001, 69). The task is particularly difficult in the case of human studies submitted to EPA for regulatory purposes, because the benefits of the research typically accrue not to the study participants, but to society at large, calling for an especially cautious approach in applying general principles. The committee decided that it could a provide a framework for clarifying some specific issues regarding the use of intentional human dosing studies for EPA regulatory decision-mak-
ing purposes, but it made no pretense of being able to resolve all of the nettlesome issues, especially the potentially wide range of study-specific risk-benefit comparisons that might be raised in this context. These ultimately must be resolved through publicly transparent policy deliberations and through the case-by-case decisions made by duly constituted review bodies.
POTENTIAL BENEFITS FROM INTENTIONAL HUMAN DOSING STUDIES
The Common Rule under which EPA conducts and sponsors studies requires that “risks to subjects” must be “reasonable in relation to anticipated benefits, if any, to subjects, and the importance of the knowledge that may reasonably be expected to result” (40 CFR §26.111(a)(2)). NBAC interpreted the basic ethical framework guiding human research as requiring independent review to “ensure that risks are reasonable in relation to potential personal and societal benefits” (NBAC, 2001, 3).
As indicated by these formulations of the risk-benefit requirement, potential or anticipated benefits from studies involving humans can be divided into two broad types—personal and societal. Potential personal benefits are those that may accrue to an individual by virtue of participating in the experiment. Potential societal benefits are those that accrue to the society as a whole or to groups within a society by virtue of the application of the scientific results of the study.
For example, placebo-controlled Phase 3 drug trials are designed to test the effectiveness of a drug. If the drug proves effective, at least some of the participants have the prospect of receiving direct medical benefit from the new treatment. Both intervention and control participants also may have the prospect of gaining other personal benefits, although such benefits would not result from receiving the drug being studied. For example, participants may benefit from increased knowledge about their condition from the medical evaluation that is included in the study.
There are many clinical trials, however, that are not intended to offer direct clinical benefits to participants. Phase 1 drug trials, for example, are designed to test for side effects of a drug and to establish dosing regimens. These trials often enroll healthy individuals who do not suffer from the condition the drug is intended to treat. These participants will receive no direct medical benefits from receiving the drugs during the trial. Nonetheless, carefully designed and conducted Phase 1 trials with healthy volunteers have been considered ethically acceptable. When risks are minimized, some risks to informed and consenting participants can be and are considered reasonable in light of the potential societal benefits that may result from the study.
Experiments involving intentional human dosing that are conducted for EPA’s regulatory purposes do not present the possibility of providing any health-related direct personal benefits to participants. As described in Chapter 1, when EPA implements statutes requiring risk assessment, the health effects information used in such assessments contributes to improving the understanding of the adverse effects of environmental toxicants, but it does not produce personal benefits to those who participate in the experiments. Air chamber studies of the kind EPA has conducted in the Air Office can occasionally be exceptions to this general rule, because participants who experience angina pain, for example, may benefit from learning more about the circumstances in which they experience such effects. Pesticide-related studies, however, are designed to detect either adverse effects or effects on normal physiologic reactions, or they are designed to study the pharmacokinetics of a chemical in the human body. Secondary benefits might accrue to participants, for example, who receive a comprehensive medical screening evaluation as a condition of participation. However, the possibility of gaining such benefits does not result from the administration of the chemical, and it is not integral to the goals of the study.
Payment for Participation
Paying research participants, which is a common and longstanding practice in the United States, provides a form of personal benefit. Although payments are made in part to compensate participants for the inconvenience they may experience, they also appear to aid in study recruitment. The value assigned to financial compensation of research participants in the risk-benefit analysis has been controversial among ethicists and other experts. The committee did not undertake an in-depth analysis of the issues involved (although Chapter 5 includes a discussion of the role of payment in an individual’s decision to participate in a research study). Nonetheless, acknowledging the controversy over how compensation affects the overall risk-benefit assessment seems necessary in light of the near universal practice of paying volunteers for their participation in the third-party studies submitted to EPA that were reviewed by the committee.
With regard to compensation, NBAC’s report on Ethical and Policy Issues in Research Involving Human Participants illustrates one significant viewpoint. NBAC expresses the concern that treating compensation as a benefit for purposes of the Common Rule’s balancing of risks and benefits “would inappropriately skew judgments concerning risks and potential benefits, because nearly any level of research risk could be offset by such
gains if they were significant enough—for example, if participants were promised large sums of money for participating in the research” (NBAC, 2001, 74). In light of this concern, NBAC urged that compensation not be considered a benefit for purposes of an Institutional Review Board’s (IRB’s) weighing of the risks and benefits of a research proposal. This result strikes some as counterintuitive, ignoring the undeniable fact that from the perspective of a prospective participant, compensation can and often does count as a benefit, and even one that may tip the balance in the individual’s decision-making process regarding whether or not to participate.1
Qualms about the correct treatment of compensation partially reflect an interest in preserving a central feature of both the Common Rule and other statements of principles regarding human research: the requirement that not one but two affirmative judgments must be made in order for research to be designated as appropriate—one by an individual when providing informed consent, and the other by an independent body evaluating risks and benefits, as well as other features of the research protocol. Under this structure, it should be possible for a potential research participant to give informed consent but for the IRB to consider the protocol unacceptable because of its risks. If the amount of compensation could count as a benefit in the IRB’s assessment, just as it might play a part in the individual’s decision to participate, the two judgments would become difficult to distinguish. The independent assessment contemplated by the Common Rule seems designed to reflect broader social norms regarding acceptable research, norms that cannot be offset by the promise of greater payment to participants.
In the end, the committee did not attempt to resolve definitively the extent to which compensation should be considered a personal benefit for purposes of the independent appraisal of whether a study’s benefits justify the risks involved. Committee members did agree, however, that if compensation were the only benefit of an intentional human dosing study, this would be inadequate to justify any risk. Because generally in human studies conducted for EPA regulatory purposes there are no other personal benefits to participants beyond compensation (however that is judged as a benefit),2 the justification for such studies depends on the presence of sufficient societal benefits to justify the risks.
NBAC also expressed concern that high levels of compensation would undermine informed consent by “induc[ing] participants to enroll without carefully considering the risks involved in participation” (NBAC, 2001, 74). See Chapter 5 with regard to this aspect of the compensation controversy.
There may be extraordinary cases in which personal benefits are present, but such studies would have to exhibit some distinctive feature not present in the studies that the committee reviewed.
Identifying and assessing the societal benefits of intentional human dosing studies and then comparing those benefits to the risks to participants are a controversial and complex process. In the context of the pesticide program, the Joint Subcommittee of the EPA Science Advisory Board (SAB) and the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) Scientific Advisory Panel (SAP) (discussed in Chapter 1) concluded that, in order for such studies to be ethically justified, “the information expected to be gained must promise reasonable health benefits to the individual or society at large,” and that even then such studies should be considered only if they meet conditions that the report described as ranging from “rigorous to severe” (EPA, 2000, 3). The requirement that the study should hold the promise of providing health benefits means that the SAB/SAP subcommittee “would not support human dosing that intended to bring about increased allowable residue levels” for a pesticide (EPA, 2000, 26), because no health benefits are achieved when tolerance levels for a pesticide are raised. Two members of the subcommittee of 13 filed a minority report expressing still greater reservations about intentional dosing studies. They contended “that no limited human study will provide information about safe levels of intake of pesticides by humans, especially humans” (EPA, 2000, C-1). Those who signed the minority report apparently concluded that the type of studies the subcommittee had been asked to review could never be conducted ethically.
For reasons explained below, this committee does not agree with the SAB/SAP subcommittee in two important respects. First, the committee believes that environmental as well as health benefits should be considered. Second, studies meeting the six conditions imposed by the regulatory framework on human research noted at the beginning of this chapter, but whose results do not promise health or environmental benefits may be acceptable if (1) there is a sound scientific basis for concluding that the exposure during the study to the chemical being tested will not harm research participants and (2) the study would make an important contribution to the scientific quality of a regulatory decision, whether that decision is to decrease or increase an allowable residue level (which, of course, cannot be known with certainty until the study is conducted).
This conclusion hinges on the committee’s determination regarding what constitutes a societal benefit for purposes of evaluating the ethical validity of human dosing studies conducted for EPA regulatory purposes. For many of the same reasons discussed earlier in the context of personal benefits, compensation to participants plays no role in assessing societal benefits. However, the committee identifies two distinct types of societal benefits that might accrue: (1) improving the scientific basis for imple-
menting congressionally mandated regulatory frameworks with all of the community benefits that this implies and (2) human health or environmental benefits that might result from the use of human data in setting regulatory standards.
As noted in Chapter 1, the environmental toxicants over which EPA has jurisdiction under its statutes pose regulatory policy challenges because they produce risks and they are produced for or released as a result of activities that society values. The policy challenge society faces is developing an acceptable means for resolving the clash of interests or values produced by this dilemma. It is, furthermore, a difficult policy challenge, because people disagree over the value of the activity that generates the toxicant, over how essential it is to carry out the activity in a way that generates toxicants, over how much risk is produced, over how to value that risk, and over how all of these considerations should be weighed in the ultimate resolution. Nonetheless, the competing values must be resolved—even as these subjects continue to be debated—and the resolution is created through legislation and legislatively mandated administrative decision processes.
In a functioning democracy, the particular resolution embodied in statutes, regulations, and administrative procedures should be accorded legitimacy, even as efforts may be made by some to change the law. Bringing policy as implemented into closer alignment with policy as enacted, therefore, confers greater legitimacy to government decisions, which is a societal benefit, regardless of whether the result of human testing is to make the regulatory standard more or less stringent. If a different legislative resolution occurs, bringing policy implementation into closer alignment with that different resolution will be what produces a societal benefit. Acknowledging the societal benefit of an improved scientific basis for making decisions obviously does not resolve legislative or public controversy over how the risks of toxicants should be regulated.
A second type of societal benefit consists of benefits to human health or the environment that result from the implementation of a regulatory standard. For example, an air pollution standard that is made more stringent on the basis of human studies provides a health benefit to those who will be protected from the adverse respiratory effects of a pollutant. Because some pesticides contribute to disease control, a risk evaluation that allows the use of such pesticides may produce a public health benefit. Other pesticides do not generate such public health benefits. Recognizing the possible differences in such downstream consequences is a necessary step when comparing the risks and benefits that may result from intentional human dosing studies.
It is important to note that it is not clear whether or how such issues should be considered within the current regulatory framework for hu-
man research. Specifically, the Common Rule at 40 CFR 26.111(2) states that “the IRB should not consider possible long-range effects of applying knowledge gained in the research … as among those research risks that fall within the purview of its responsibility.” It is not apparent from the text of the Common Rule what this language is intended to mean in the context of studies conducted for EPA regulatory purposes. The committee believes that, in the EPA context, considering the benefits associated with the kinds of uses to which tested substances will be put is no less relevant to an IRB review than are the anticipated health-related uses to which a tested pharmaceutical will be put when an IRB is reviewing a drug trial.
The following sections discuss both types of societal benefits—improving the scientific basis for implementing legislation and human health or environmental benefits.
RELIABILITY IN IMPLEMENTING THE CURRENT REGULATORY FRAMEWORK
As a society, we currently employ a variety of approaches to accommodate public health concerns raised by the use of environmental toxicants that are associated with useful activities. Some, such as the Emergency Planning and Community Right to Know Act (42 U.S.C. 116), require sources of pollutants to report the amount of particular harmful substances released into the environment. Some, such as the new source performance standards of the Clean Air Act (42 U.S.C. 85), require sources of pollutants to reduce the release of specific harmful substances to levels attainable through the application of pollution abatement technology that EPA has judged practicable, or best economically achievable, or best available, or that meets some other technology standard established by the statute. Others, such as the ambient air quality standard-setting provisions of the Clean Air Act or the tolerance setting process under the Food Quality Protection Act of 1996 and the Federal Food, Drug, and Cosmetic Act (FFDCA), mandate that levels of particular harmful substances should not exceed the levels judged to be low enough to protect humans from specified adverse health effects. Still others, such as the registration process under FIFRA, require EPA to balance the adverse effects and the beneficial effects of permitting the environmental release of harmful substances.
The advantages and disadvantages of each of these approaches have been debated at great length.3 Whatever approach Congress has chosen,
For a useful summary, see Office of Technology Assessment, Environmental Policy Tools (1995), available at www.wws.princeton.edu/~ota/ns20/alpha_f.html.
a constant has been the need to develop factual information for its implementation, much of which is scientific in nature. Major differences among the diverse approaches include the particular type of scientific information needed and the conclusions that must be reached in order to implement them. Health-based approaches—such as the ambient air quality setting process of the Clean Air Act or the process for setting tolerances for pesticide use on food on the basis of a “reasonable certainty of no harm,” and the risk-benefit balancing approaches, such as the licensing process for nonfood use pesticides—require information that relates exposure to the substance to types and levels of harm. In other words, they require some assessment of the risks associated with the substance. Where the risks to humans are among those that need to be assessed—as they are in the cases of the Clean Air Act and FIFRA—then information that could predict potential human responses to exposure is relevant to that risk assessment, especially in the hazard identification and dose-response assessment components. (The general risk-assessment framework is described in Chapter 1.)
Those who assess risk try to provide information to risk managers that rests on reliable science, information that typically is drawn primarily from animal toxicity studies. However, except in cases in which the specific risk of concern is directly measurable in humans, nothing guarantees that science at any point produces correct answers. As the U.S. Supreme Court has recently remarked, “it would be unreasonable to conclude that [scientific conclusions] must be ‘known’ to a certainty; arguably, there are no certainties in science” (Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 590 ). In many cases, again in which the effect of concern cannot be directly measured in human studies, all that science can provide is a determination supported by a broad segment of the scientific community—where one exists—that a particular approach or finding represents the best understanding at any particular point in time.
In many instances, reputable science will be unable to generate all the findings necessary to give risk managers a completely science-based set of findings on which to predicate the public health and welfare decisions that must be made. For almost all risk assessments involving toxicants, current scientific knowledge is insufficient to reach a definitive conclusion with regard to some of the questions such assessments raise, because in almost all cases, human toxicity, especially long-term toxicity, is difficult or impossible to study directly.
For example, scientific studies could establish that a toxicant causes malignant tumors in several animal species. What do these results say about the carcinogenic potential of the substance in humans? The animals were exposed to high levels of the substance for a prolonged period. What do these high-exposure results say about the ability of the substance to
produce malignant tumors at the relatively low and often intermittent doses to which humans are exposed? Is there a level of exposure below which the substance does not have any potential to cause adverse effects? Or can we describe with certainty an exposure that produces a specific level of risk that we would consider acceptable? These kinds of questions cannot be answered with certainty by today’s science. Yet some judgment about the answers must be reached by agency decision makers who have been charged with making regulatory determinations based on an assessment of risks.
Making a regulatory judgment (including a decision to do nothing), cannot be delayed until science makes all the predicate findings. Some determination of how to weigh useful activities and the risks they create must occur now, even though it may need to be changed later. Therefore, public policy decisions regarding these questions will have to be made even in the presence of important gaps in knowledge that cannot be filled by science at this time. These decisions will necessarily involve both scientific findings and judgments about how the gaps in knowledge should be filled.
The 1983 National Research Council report, Risk Assessment in the Federal Government: Managing the Process, identified 51 different places in a routine risk assessment where the exercise of judgment can be required to bridge a gap between what science is currently prepared to accept as a valid finding and the next step in the analytic process that constitutes the risk assessment (NRC, 1983, 33-37). The report referred to these gap-filling judgments as “inferential bridges.” EPA refers to them as “default assumptions”—that is, the risks that will be presumed to exist in the absence of other data. These assumptions reflect the agency’s current thinking regarding how a question should be answered in the absence of additional scientific evidence indicating that a different answer would be better. Thus, for example, in the absence of scientific evidence showing that there is a safe, nonzero level of exposure to a human carcinogen, EPA’s default assumption is that there is no such level. Accordingly, any human exposure to such a substance is assumed for regulatory decision-making purposes to create some risk of contracting cancer (NRC, 1994).
These default assumptions allow risk-management decisions that are based on human risk assessments to be made under conditions of uncertainty, an unavoidable necessity. At the same time, the presence of such default assumptions produces another imperative. When possible, the default assumptions should be adjusted and eventually replaced with findings or judgments that are rooted in improved scientific understanding. This imperative is implicit in the commitment to use the best available science in making risk-management decisions. The committee determined that this commitment is sound, subject, as discussed below, to the
equally important and often overriding need to protect participants in research. Such a commitment to the use of the best available science has been made and consistently reaffirmed by all three branches of government—executive, legislative, and judicial—and accordingly it should be considered by EPA in assessing the benefits of all scientific results, including those involving intentional human dosing studies.
Executive Order 12866, issued initially by President Clinton in 1993 and revised by President Bush in 2003 (without changes to the relevant portions), directs each administrative agency to “base its decisions on the best reasonably obtainable scientific, technical, economic, and other information concerning the need for, and consequences of, the intended regulation” (E.O. 12866 §(b)(7)). EPA has a similar longstanding and publicly stated commitment to using the best available scientific information. In 1991, the agency issued a mission statement that included the commitment to ensure that “national efforts to reduce environmental risk are based on the best available scientific information communicated clearly to the public” (EPA, 1991). A year later, an expert panel on science at EPA reiterated this commitment in a report to Administrator William Reilly (EPA, 1992). In 1994, a policy guideline from EPA Administrator Carol Browner stated that “EPA strives to ensure that the scientific and technical underpinnings of its decisions meet two important criteria: they should be based upon the best current knowledge from science, engineering and other domains of technical expertise; and they should be judged credible by those who deal with the Agency” (EPA, 1994). The agency’s current mission statement also commits the agency to ensuring that “[n]ational efforts to reduce environmental risk are based on the best available scientific information.”4
Default assumptions in risk assessments are needed in areas in which science has not progressed sufficiently to provide an answer to a question that is a necessary part of a risk assessment. When an answer to such a question becomes available, however, the general imperative of using the best available science implies that this answer should replace the default assumption. Numerous specific pronouncements by the agency regarding default assumptions bear out the desirability of replacing default assumptions with scientific results. For example, the Draft Water Quality Criteria Methodology Revisions state that “When adequate data are available they are used to make accurate exposure predictions for the population(s) of concern. When this is not possible, a series of qualitative alternatives is proposed using less adequate data or default assumptions
Available at www.epa.gov/history/org/origins/mission.htm.
that allow for the inadequacies of the data while protecting human health” (EPA, 1998a).
The Guidelines for Neurotoxicity Risk Assessment are similar, stating that “default assumptions should not be applied indiscriminately. First, all available mechanistic and pharmacokinetic data should be considered. If these data indicate that an alternative assumption is appropriate or if they obviate the need for applying an assumption, such information should be used in risk assessment.” (EPA, 1998b). Finally, EPA’s Proposed Guidelines for Carcinogen Risk Assessment state that “EPA’s 1986 guidelines for cancer risk assessment … were developed in response to [the Red Book]. The guidelines contained a number of default assumptions. They also encouraged research and analysis that would lead to new risk assessment methods and data and anticipated that these would replace defaults” (EPA, 1996).
There is strong evidence that Congress has consistently shared with the executive branch the view that when science is to be relied on to supply information pertinent to a regulatory decision, the best available science should be employed. Sometimes it has stated this view explicitly, as in the 1996 amendments to the Safe Drinking Water Act. There, Congress has provided that “to the degree that an [EPA] action is based on science, the Administrator shall use:
the best available, peer-reviewed science and supporting studies conducted in accordance with sound and objective scientific practices; and
data collected by accepted methods or best available methods (if the reliability of the method and the nature of the decision justifies use of the data) (42 U.S.C. §300g-1(b)(3)(A)).
Examples of statutory language requiring the use of the best available science can be found in older statutes as well. The Asbestos School Hazard Abatement Reauthorization Act of 1989 requires that when EPA provides information to schools about the hazards of asbestos, “[s]uch information or advisory shall be based on the best scientific evidence…” (15 U.S.C. §2643).
Congress’s commitment to using the best science available is not limited to actions taken by EPA. For example, the Endangered Species Act was amended in 1978 to include the instruction to all federal agencies that “each agency shall use the best scientific and commercial data available” in ensuring that any action by an agency will not threaten the existence of an endangered species (16 U.S.C. §1536(a)(2)). Dating back to 1970, the Occupational Health and Safety Act contains provisions regarding the protection of workers from exposure to toxicants that require protective measures to be based on the “best available evidence” (29 U.S.C. §655(b)(5)).
Besides these statutory instructions, the scientific advisory panels and peer review procedures established by law under many of the statutes EPA administers provide further evidence of Congress’s appreciation of the value of improving the quality of scientific findings that inform agency decision making. Under FIFRA, for example, the Administrator is to “solicit from the [scientific] advisory panel comments, evaluations and recommendations for operating guidelines to improve the effectiveness of scientific analyses made by personnel of the Environmental Protection Agency that lead to decisions by the Administrator in carrying out the provisions of this subchapter” (7 U.S.C. §136w(d)(1)).
Not all of the statutes EPA administers explicitly invoke the use of the “best available science,” or something equivalent in statutory language, but these illustrations show that many statutes reflect the conviction that when scientific judgments are called for, better science is preferred.5 There is no reason to believe that when statutory language does not contain such explicit language, the presumption that the best available science should be employed should be any different. Regardless of the statute and the science involved, improvements in the accuracy and reliability of the science improve the quality of information that is relied on for making ultimate regulatory decisions. It is hard to imagine that Congress would not consider the improvement of the quality of scientific information to be a benefit to the regulatory processes that it has asked EPA to implement. The committee thus concludes that when Congress has enacted regulatory processes that rely on science, improving the science those processes employ serves to implement the resolution of competing interests.
Balancing the interests of the various parties affected by EPA’s statutory requirements can lead to the development of regulations that effectively establish legal rights and responsibilities. Parties can correctly insist that the rights and responsibilities that are ultimately established and enforced by EPA through its regulations implement the public policy that Congress has enacted. In our democratic system of government, Congress codifies a particular balance when it writes binding law, and that balance should be observed by administrative agencies until it is changed.
Even in cases where the agency employs elements of discretion in fine-tuning the ultimate regulations, that discretion should be based on the best relevant and available understanding of the information Congress has directed the agency to take into account. In the case of risk assessments and the regulatory decisions that employ them, this information includes the scientific components of the assessment. The more accurate the science-based components of the regulatory systems EPA administers under these statutes, the better informed EPA’s exercise of discretion will be.
In addition to its stated general preference for replacing default assumptions with scientific findings, EPA has expressed a specific preference for supplementing animal data with human data when conducting human risk assessments. EPA has said that it looks to human data whenever possible in completing human risk assessments: “If adequate human studies (confirmed for validity and applicability) exist, these studies are given first priority in the dose-response assessment, and animal toxicity studies are used as supportive evidence” (EPA, 1989). Often, such data can be obtained from epidemiological studies, which do not involve the intentional dosing of research participants, but rather evaluate the effects of exposures that have occurred in an occupational setting or because of the peculiarities of a specific geographical setting.6 Regardless of the origins of such human data, “risk assessments based on human data have the advantage of avoiding the problems inherent in interspecies extrapolation” (EPA, 1993).
The default assumptions that are of particular relevance to the issues raised by third-party intentional human dosing studies are those that bridge gaps between animal results and estimates of effects in humans. In the context of FIFRA, for example, EPA has routinely divided the calculated “safe” dose for animals by a factor of 10, to account for the possibility that humans are more sensitive to the substance being tested than are
the animal species. Third-party submitters of human dosing studies have been particularly interested in modifying this default assumption by introducing data obtained directly from human studies.
The benefits to the regulatory process of improved science are generally accepted without question in areas of risk assessment that do not involve the deliberate exposure of humans to toxicants. For example, the fate and transport studies central to defining the nature and extent of human exposure require an understanding of how substances released into the environment move in that environment, interact with other substances, and eventually come into contact with humans, whether through dermal contact, inhalation, or ingestion. Answering these questions involved in fate and transport studies involves applying knowledge in fields such as hydrology and chemistry. There is little controversy regarding the idea that improving the accuracy and reliability of the science benefits the risk-management process by providing the best answers to scientific questions that can be provided at the time.
The critical difference between improving the exposure assessment component of a risk assessment through better fate and transport models and improving the dose-response component of that assessment through human studies is not that the first supplies a benefit to the regulatory process and the second does not. Both provide benefits in the form of better estimates to use in the risk-management process; however, this certainly does not mean that it should be federal policy to pursue either of these benefits indiscriminately. To say that a piece of information supplies a benefit is not the same as saying that we should acquire the information regardless of the costs. A major commitment involved in ensuring the ethical treatment of research participants is being prepared to reject research that would produce beneficial information if that research exposes humans to unjustified risks. The difference between improving an exposure assessment and improving a dose-response assessment is that the former typically does not expose humans to health risks, while the latter, if it is to be accomplished by experimentation directly on humans, potentially does. This difference obviously has tremendous significance and a profound effect on how one should approach evaluations of those studies. In terms of the risk-benefit calculus that would be applied to judging the ethical acceptability of a human study, the way to take this difference into account is first by making a careful determination of what the risks are and then weighing those risks against any benefits that might result from the study.
The committee concludes that it is a matter of established and sound public policy that the use of the best available science—including the replacement of default assumptions with reliable scientific information—constitutes a societal benefit.
As discussed earlier, the Common Rule requires that there should be not only an expectation of benefit resulting from the proposed research, measured for present purposes by the “importance” of the knowledge to be gained, but that risks to participants should be considered as well and that these risks should be “reasonable” in relation to the importance of the knowledge. This standard requires that the risks and benefits of a study be evaluated and then compared.
Building on these principles, the following section provides the committee’s perspective on how the balancing of risks and benefits should be approached with respect to human studies conducted for EPA regulatory purposes.
BALANCING RISKS AND BENEFITS
Of the three basic ethical principles governing the protection of research participants—respect for persons, beneficence, and justice—beneficence is the one that, in the context of this report, requires the greatest exercise of subjective judgment with the least amount of guidance from established policy or precedent. Informed consent (respect for persons) and the fair distribution of the benefits and burdens of human research (justice) both are important and challenging, but it is possible to delineate reasonably objective decision rules to guide their application.
Beneficence is the ethical principle that requires considering the well-being of the research participant and ensuring that possible risks are minimized and that any risks that remain are justified by the potential benefits of the research (National Commission, 1979). Beneficence thus requires a subjective balancing judgment. Moreover, in the context of human research conducted to inform EPA’s regulatory decision making, beneficence requires balancing anticipated risks to the participant against potential benefits to society in order to assure that the risks are justified by the benefits. To paraphrase the Common Rule, the risks must be “reasonable” in relation to the importance to society of the knowledge produced by the research. There are no formulas for determining whether a risk to an individual is justified by a benefit to society.
Independent review of human research, such as is conducted through local IRBs, is essential to ensuring that all three of the key ethical principles are being followed. In the case of clinical research on therapeutic products, IRBs have considerable experience in balancing risks and benefits and are also familiar with certain kinds of studies in which the benefit does not accrue directly to study participants, such as pharmacokinetic (PK) and pharmacokinetic-pharmacodynamic (PK/PD) studies, other mechanistic studies, and Phase 1 studies. In these cases, the kind of information to be obtained and its usefulness are relatively familiar. In
the cases addressed in this report, however, concerning human studies conducted to inform EPA’s regulatory decision making, most IRBs have little or no experience in weighing the kinds of benefits that might arise against the risks. This is one reason why the committee recommends later in this report (see Chapter 6) that there should be a role for a centralized review body operating under EPA’s auspices to review human studies conducted for EPA regulatory purposes.
In the next section, the committee provides an overview of the kinds of risks and benefits that such studies may present. It also provides some perspectives on how the risks and benefits might be balanced to determine whether a study comports with the principle of beneficence.
Assessing the Risks
The Common Rule requires investigators and IRBs to identify, analyze, and assess risks, and investigators to disclose risks to potential research participants. The term “risk” refers to the probability of a harm occurring and includes consideration of both the magnitude of a particular harm and the probability of its occurrence. Because both the risks and the benefits of research are not known in advance and can only be projected or predicted, the proper comparison is not between risks and benefits but rather between anticipated risks and potential benefits.
Under the Common Rule, IRBs evaluating research protocols are required to (1) classify risks (as minimal or greater than minimal), (2) ensure that “risks to subjects are minimized,” and (3) determine that risks are reasonable in relation to probable benefits to research participants and/or the “importance” of the reasonably expected knowledge. Each task poses important challenges for IRBs. The first two are discussed here, and the third—the balancing of risks and benefits—is discussed later in this section.
The distinction in the Common Rule between minimal and greater-than-minimal risk provides a sorting mechanism that enables IRBs to attend more closely to protocols that involve greater risks. A classification as minimal risk is a necessary (but not sufficient) condition for a protocol’s expedited review, rather than full convened IRB review, and for a waiver or modification of the elements of informed consent or of the documentation of informed consent. Another category, “a minor increase over minimal risk,” has been adopted by the Department of Health and Human Services (DHHS) and by the U.S. Department of Education for research involving children (Subpart D of the DHHS version of the Common Rule, 45 CFR 46).
The minimal risk standard encompasses studies whose risks are so low that customary IRB review and even some elements of informed con-
sent can be bypassed. Most studies that qualify as minimal risk under the Common Rule involve no active intervention affecting the research participant—that is, they are observational or epidemiological rather than invasive. A study of postexposure pesticide levels might belong in the minimal risk category. Although the language of minimal risk is widely used in the United States and in international discussions, its interpretation varies, especially in cases that involve some active intervention. According to the Common Rule, “Minimal risk means that the probability and magnitude of harm or discomfort anticipated in the research are not greater in and of themselves than those ordinarily encountered in daily life or during the performance of routine physical or psychological examinations or tests” (40 CFR 26.102(h)(1)). Even with this definition as a guide, minimal risk is not construed in consistent ways among federal agencies or by different IRBs. In view of these inconsistencies, NBAC proposed that:
IRBs should use a standard related to the risks of daily life that are familiar to the general population for determining whether the level of risk is minimal or more than minimal. The standard should not refer to the particular risks encountered by particular persons or groups [emphasis added]. It should refer, therefore, to common risks—for example, driving to work, crossing the street, getting a blood test, or answering questions over the telephone (2001, 83).
NBAC made this distinction because people who face inherently risky situations, by virtue of, for example, illness or occupation, should not be allowed to face higher risks in research than others, except in exceptional circumstances (e.g., compassionate use of experimental treatments in the terminally ill).
The committee finds the experience with the minimal risk concept in the context of clinical research uninformative for purposes of assessing the ethical validity of the types of human studies most likely to be conducted for EPA regulatory purposes. Even though some of the intentional dosing studies conducted for these purposes pose no identifiable risk to participants, the committee is reluctant to consider any toxicant dosing study a minimal risk study within the meaning of the Common Rule. Importantly, the committee concludes that any human dosing study conducted for EPA regulatory purposes, regardless of how safe it may appear to be, and even if it could be judged by some to pose minimal risk under the Common Rule, should be reviewed both by an IRB and by the Human Studies Review Board recommended in Chapter 6. This will ensure that the health of participants is in fact protected. It also reflects the need for careful review to ensure that a proposed study provides the specialized form of societal benefit—improving the scientific quality of regulatory decision making—potentially associated with studies conducted for
EPA purposes. Thus, even if a human dosing study conducted for EPA regulatory purposes could be deemed to pose minimal risk, that finding would not, under the committee’s recommendations, have the practical consequences it has for the more typical human research study evaluated under the Common Rule.
For these reasons, in describing the range of risks posed by the human dosing studies addressed in this report and how those anticipated risks might be balanced against potential benefits, the committee does not use the terminology of “minimal risk” or “minor increase over minimal risk.” Rather, the committee uses other terms that describe the anticipated risk or lack thereof, as discussed below, but will do so free of the implications the term “minimal risk” carries as applied in other settings under the Common Rule. This approach also is intended to better reflect the nature of the human dosing studies that are generally conducted for EPA purposes and the range of possible human responses to chemical exposures.
Exposure to any chemical substance, whether of natural or industrial origin, can cause alterations of many types in the biological structures and functions of living organisms, including humans. These alterations vary among chemicals and also with the conditions of exposure (with conditions referring to the magnitude, duration, and route of exposure). In addition, for most chemicals there are ranges of doses outside of which no biological change in structure or function can be detected using the best available scientific technology. For example, as discussed in Chapter 3, PK studies of toxicants, which are intended to document how a chemical is normally metabolized by the body rather than to elicit any response or alteration, are often conducted at doses that are not expected to cause any significant or even detectable alteration in biological structures or functions. In other intentional human dosing studies involving pesticides, the objective is to elicit some biological response and to identify a dose at which the response did not occur. In such studies, the maximum studied dose at which no biological changes can be observed (always relative to a control group) is referred to as the “no observed effect level,” or NOELHU, but to determine the NOELHU rigorously it is necessary to find the dose at which the effect is seen, a “lowest observed effect level,” or LOELHU. In other studies the effect investigated could be relatively mild but nonetheless undesirable (i.e., perceived as an adverse event). The committee saw no examples of studies conducted for EPA that were intended to provoke more serious adverse effects. However, one study did produce an effect larger than expected at the midrange of the intended dosing schedule and was stopped.
With this as background, and based on its review of the kinds of human dosing studies that have been and are likely in the future to be sub-
mitted to EPA, the committee identifies three categories of anticipated risk associated with such studies.
The first category of risk includes studies that pose no identifiable risk to participants. This category includes PK studies conducted at low doses that delineate uptake and disposition of a chemical and its metabolites, but are expected, based on extensive previous testing in animals, to have no biological effect on the participant, as discussed in Chapter 3. Although it is not possible to prove the total absence of risk with absolute certainty, low-dose PK studies of the kind noted here and discussed in Chapter 3 are typically conducted at levels far below those that have been or would be judged safe under the legal safety standard in FFDCA for pesticide residues in food and are as close to being risk free as any human dosing study can be.
The second category of risk includes studies that elicit a biological response, but ones that are not in any way adverse to the participant, such that, on the basis of ample scientific evidence, experts would conclude that there is a reasonable certainty of no harm to study participants. This category includes the PD studies discussed in Chapter 3 in which the observable changes serve as indicators or biomarkers of exposure, but are immediately reversible upon cessation of exposure and would be expected to have no consequence to the health of the individual experiencing them. Changes, for example, in cholinesterase activity in blood would be rapidly reversible and at low exposure would not be associated with any adverse effect. Detectable but clinically insignificant changes in blood pressure or heart rate in normotensive individuals would similarly be considered nonadverse and are often categorized as indicators of exposure to a chemical, rather than as evidence of toxicity. In some cases, it is clear that those biological changes, while not adverse in themselves, are sensitive indicators of a process that would be adverse if the effect were greater (e.g., from a larger dose) or if dosing were prolonged. Cholinesterase inhibition studies on organophosphate (OP) pesticides are examples of such cases, as discussed further in Chapter 7.
The third category of risk includes studies at doses that elicit a biological response in either the structure or function of the organism that is potentially detrimental to health, or adverse, and thus poses an identifiable risk to study participants. This category theoretically encompasses a wide range of risks. It includes potential risks posed in the PD studies discussed in Chapter 3 that have adverse effects that are non-trivial but transitory and expected ultimately not to be harmful to the study participants, in the sense of causing any lasting impairment or pain. Examples of such transitory, non-trivial symptoms that are adverse but not ultimately harmful include headache, nausea, or temporary irritation to the eyes or airways. These are symptoms sometimes seen in air pollution studies.
Studies whose adverse effects are not transitory and thus may result in lasting harm are also included in this category, but the committee is unaware of human dosing studies conducted for EPA regulatory purposes that were anticipated to result in lasting harm to participants. Such studies are clearly not allowed. Even when the adverse effect is transitory, however, such studies pose an identifiable risk of immediate harm and, especially if conducted in vulnerable populations, pose a significantly greater risk of unexpected lasting harm than the studies in the first two risk categories.
Assessing the Benefits
Any human dosing study, regardless of its risk category, must have a useful purpose and convey some benefit to the participants and/or society. As discussed earlier, the committee concludes that under the risk-benefit balancing required by the principle of beneficence and the Common Rule, personal benefits to participants are insufficient by themselves to justify human dosing studies conducted for EPA regulatory purposes. This means that risks to participants imposed by human dosing studies must be justified by the societal benefits that are anticipated to come from a successful study, if they are to be justified at all. In this respect, human dosing studies are similar to Phase 1 drug trials.
The committee also concludes that in order to generate societal benefits at all, human dosing studies must (1) be performed in a context in which there is a clearly defined regulatory objective and a critical, unanswered question or other compelling scientific need that cannot be satisfied with animal data and (2) be designed with the requisite statistical power and other design features required to meet that regulatory objective and scientific need. These are threshold requirements that any human dosing study must meet. The steps that study sponsors must take to satisfy this threshold test are discussed in Chapter 3.
Studies that satisfy this threshold test have the ability to improve the accuracy of EPA’s regulatory decision making. For the reasons discussed earlier in this chapter, the committee concludes that improving the accuracy of the science employed in regulatory decisions constitutes a societal benefit. Beyond this minimal benefit, however, human dosing studies also can generate different kinds of societal benefits as well, depending on the nature of the scientific question a study seeks to answer, the uses to which study results may be put, and the consequences that may flow from those uses. Thus, just as there is a spectrum of risk categories into which human dosing studies might fall, there is a spectrum of potential societal benefits, which can be categorized roughly as follows.
The first benefit category is the one outlined above in which the study
provides improved accuracy of EPA decision making but conveys no other societal benefit in terms of better protecting human health or the environment. This benefit category provides the minimum benefit required to justify a human dosing study. The third-party studies that have been recently submitted to EPA’s pesticide program aimed at identifying the NOELHUs and LOELHUs for specific OP pesticides are the most prominent examples of studies that may provide benefits in this category. In these cases, extensive animal testing has established that the critical determinant of toxicity (and risk) for the pesticide is cholinesterase inhibition. Thus, use of data on cholinesterase inhibition from humans, assuming they derive from properly designed and executed studies, would improve the scientific accuracy of EPA’s risk assessment. Although such studies conceivably could demonstrate that humans are more sensitive than animals and that the Reference Dose (RfD) derived from human data is lower than one based on animal data, the interest of the study sponsor is to increase the RfD and thus allow for greater use of the pesticide based on a more scientifically accurate risk assessment.
Determining the implications of the benefits from such studies was challenging to the committee, as they could result in the reduction or elimination of the uncertainty factor, which could produce a less stringent regulatory standard. The committee concludes that when a study improves scientific accuracy relevant to regulatory decision making but generates no health or environmental benefits, the benefit of the improved scientific accuracy of decision making can justify the intentional exposure of humans only to the lowest two categories of risk, as outlined above. This means that such studies must pose no identifiable risk because they elicit no biological response or, in the case of studies that elicit a nonadverse response (such as a nonadverse change in a biomarker), there must be a reasonable certainty of no harm to study participants based on a careful review of an adequate body of scientific evidence. The OP-related human dosing studies submitted to EPA’s pesticide program have measured cholinesterase inhibition as a biomarker of exposure and potential toxicity, rather than as a toxic endpoint per se. As explained in Chapter 3, an independent review could conclude that there is a reasonable certainty in such studies of no harm occurring to study participants.
Recommendation 4-1: Value of Studies That Seek to Improve the Accuracy of EPA’s Decisions But Do Not Provide a Public Health or Environmental Benefit
EPA should consider a human dosing study intended to reduce the interspecies uncertainty factor (for example, a study of a biomarker such as cholinesterase inhibition) as conferring a societal benefit only if it was designed and conducted in a manner that would im-
prove the scientific accuracy of EPA’s extrapolation from animal to human data. Because the anticipated benefit would not be as great as that conferred by studies intended to provide a public health or environmental benefit, the study could be justified ethically only if the participants’ exposure to the pesticide could reliably be anticipated to pose no identifiable risk or present a reasonable certainty of no harm to study participants.
The corollary of this recommendation is that a human dosing study on a chemical toxicant that poses an identifiable risk to study participants, even if it involves a transitory adverse effect, can be justified only if the study also provides a benefit to public health or the environment beyond the improvement of the scientific accuracy of the risk assessment underlying EPA decision making about that chemical.
Recommendation 4-2: Value of Studies That Seek to Provide a Potential Public Health or Environmental Benefit
An IRB should be properly constituted to be able to consider whether a study has the potential of providing a clear health or environmental benefit to the community. Such studies could be acceptable even if they involved a somewhat higher level of risk than that posed by studies for which there is no identifiable risk or for which there is a reasonable certainty of no harm. No study is ethically justifiable if it is expected to cause lasting harm to study participants.
There are a number of ways in which a human dosing study could provide benefits to society beyond the minimum benefit of improving the accuracy of regulatory decision making, including the following:
• The study results in a more stringent regulatory standard.
Human dosing studies that are reasonably expected to result in more stringent permissible limits for chemicals in the environment or food supply not only improve the scientific quality of the regulatory decision by substituting the more relevant human data for animal data, they also confer a potential public health benefit. Such studies require a risk-benefit balancing, as discussed below, and might be acceptable even if they involved risks somewhat greater than those involved in studies that provide the minimum benefit of improving the scientific accuracy of EPA regulatory decision making.
• The study enables EPA to adopt a public health measure it otherwise could not adopt.
Outside of the pesticides setting, EPA itself has from time to time looked to intentional human dosing studies in its air and water pollution programs, where EPA must marshal the evidence required for risk assessment. In the case of air pollutants, for example, intentional human dosing studies under controlled conditions may be the only or best way to reliably estimate the dose-response relationship in humans. Evidence from such studies could be needed to enable EPA to set standards it might not otherwise have been able to set or to set standards that are more fully protective of public health. In such cases, the knowledge derived from the study would have the important societal benefit of improving the population’s health. The magnitude of the benefit would depend on the importance of the risk being addressed by the standard and how critical having human data would be to setting the standard at a level that protects health.
• The study supports approval of a product that protects public health.
Pesticidal products that are used to control or eliminate disease vectors (such as mosquito or tick control agents) can confer important health benefits to society. If human research were required to understand the risks posed by such products and thus support their regulatory approval by EPA, such research would provide an important health benefit. As in the previous category of benefits, the size of the benefit would depend on the risk being addressed and the importance of the study.
• The study improves the scientific accuracy of risk assessment for a class of chemicals and/or EPA decisions.
Studies may have consequences for scientific knowledge that extend beyond the making of any single regulatory decision. Such consequences could occur, for example, if a human study revealed information about the proper extrapolation of animal results to humans that could be applied to an entire class or category of substances, such as the OP pesticides, that operate through a common mechanism of toxicity. By expanding the scope of the benefit to a larger class of EPA decisions without increasing the number of study participants, a study can provide benefits beyond those provided by one whose relevance does not extend beyond a single regulatory decision.
FINDING THE BALANCE
As noted at the beginning of this chapter, determining whether the principle of beneficence has been satisfied requires balancing the anticipated risks to study participants against the anticipated benefits of the study to society. The risks to participants must be reasonable in relation to the societal benefit. In the words of the Common Rule, the risks must be reasonable in relation to the importance of the knowledge that may reasonably be expected to result (40 CFR 26.111 (a) (2)). In the EPA context, if an intentional human dosing study does not have a clearly defined and important regulatory purpose and is not designed adequately to both achieve that purpose and minimize the risks to participants, the study should not be conducted, as such studies needlessly expose humans to health risks. If these threshold requirements are satisfied, risks and benefits can be balanced.
Although the preceding discussion of benefits sheds some light on the judgments that are required to strike an appropriate balance between risks and benefits in the regulatory contexts EPA confronts, the committee recognizes that the balancing requires judgment and that there are no clear rules or formulas. This is why careful independent review of proposed human dosing studies is essential.
At the extremes, the risk-benefit balancing judgment may be relatively easy. In the case of a PK study on a well-tested chemical with established “safe” levels of exposure set through the regulatory process—one that is conducted at dose levels well within the established safe level—there may be no identifiable risk, and the study could be justified if it meets the minimum test for benefits discussed above. Such a study would probably be acceptable if it met a clearly defined regulatory need for the best available scientific evidence. At the other extreme, a study in a medically vulnerable population (e.g., children) that has the potential to cause adverse responses and whose potential to cause lasting harm is uncertain poses risks that would be difficult for potential benefits to outweigh, unless perhaps the substance being tested provided significant health benefits to the study participants or to the class of individuals to which they belong.
The cases between the extremes will be more difficult to evaluate. For example, in an air pollutant air chamber study intended to improve the scientific basis for and health protectiveness of a regulatory standard, how much risk is it reasonable to impose on healthy adults? How great would the potential benefits of the study have to be to justify exposing individuals with impaired pulmonary function? An IRB or other review body would need to consider how important the study would be to the establishment of the standard and whether the risk is reasonable in relation to the societal benefit.
These examples illustrate an important point: Assessing the risk-ben-
efit balance in the case of human studies conducted for EPA regulatory purposes requires careful review and special expertise that is not always available to IRBs. It requires balancing risks to participants against benefits to society that lie in the realm of improved regulatory decision making or in broad public health or environmental impacts. It involves making risk assessments and safety judgments about chemicals that require access to data and expertise that reside at EPA and few other places. It is for these reasons that the committee recommends, in Chapter 6, that local IRB review of human studies conducted for EPA regulatory purposes be supplemented with review by a central, EPA-managed body that has the requisite expertise and that will be publicly accountable for decisions on the ethical acceptability of such studies.
Weighing the risks and benefits that might arise in human experiments is a critical and particularly difficult element of the ethical evaluation of such studies. Potential or anticipated benefits from studies involving humans can be divided into two broad types—personal and societal. Personal benefits are those that may accrue to an individual by virtue of participation in the experiment. Few intentional human dosing studies promise personal gain. Societal benefits are those that accrue to the society as a whole, or to groups within society, by virtue of the application of the scientific results of the study. This calls for an especially cautious approach in applying general principles and in evaluating, in particular cases, whether the rights and welfare of participants have been adequately protected.
The committee assumes that human studies conducted for EPA regulatory purposes do not confer personal benefits on study participants. This means that the risk-benefit balancing required under the principle of beneficence depends on the evaluation of a societal benefit. Although the volunteer’s compensation for participation can be considered a personal benefit at one level, it is properly excluded from the risk-benefit balance for reasons discussed elsewhere in this report.
Benefits do accrue to society, however, when science improves the accuracy of regulatory decisions, including the replacement of default assumptions with reliable scientific information. In the words of the Common Rule, such scientific information has “importance” that should be considered in weighing whether a study is ethically justified. This conclusion is only the starting point, however, for the ethical analysis of human studies. In particular, only risks that are commensurate with this minimal societal benefit can be justified unless additional social benefits also are present. In the case of human dosing studies that provide no further pub-
lic health or environmental benefits, the committee concludes that such studies that pose no identifiable risk to study participants or that present a reasonable certainty of no harm, based on a careful review of an adequate body of scientific evidence, can be justified under restricted conditions and with appropriate oversight and review regardless of whether the information obtained from the study results in a less stringent or more stringent regulatory outcome.
The committee determined that the analysis and conclusions presented in this chapter could clarify some issues regarding the use of studies that deliberately expose participants to toxicants for EPA regulatory decision making purposes, but it does not pretend to resolve here all of the nettlesome issues that arise from intentional human dosing studies. These ultimately must be resolved through EPA’s publicly transparent policy deliberations and through the case-by-case decisions of duly constituted review bodies charged with protecting the interests of participants in particular studies.
Environmental Protection Agency (EPA). 1989. Risk Assessment Guidance for Superfund, Vol. 1: Human Evaluation Manual, Interim Final. EPA/540-1-89/002. Available at www.epa.gov/cgi-bin/claritgw?op-Display&document=clserv:OSWER:1175;&rank=4&template=epa.
EPA. 1991. Strategic Direction for the U.S. Environmental Protection Agency.
EPA. 1992. Safeguarding the Future: Credible Science, Credible Decisions. EPA/600-9-91-050.
EPA. 1993. Reference Dose (RfD): Description and Use in Health Risk Assessments, § 22.214.171.124.1. Background document. Available at www.epa.gov/IRIS/rfd.htm.
EPA. 1994. Peer Review and Peer Involvement at the U.S. Environmental Protection Agency. Available at www.epa.gov/osp/spc/perevmem.htm.
EPA. 1996. Proposed Guidelines for Carcinogen Risk Assessment. Federal Register 61:17960, 17964.
EPA. 1998a. Draft Water Quality Criteria Methodology Revisions: Human Health; Notice. Federal Register 63(157):43755-43828.
EPA. 1998b. Guidelines for Neurotoxicity Risk Assessment. Federal Register 63(93):26926-26954.
EPA. 2000. Comments on the Use of Data from the Testing of Human Subjects: A Report by the Science Advisory Board and the FIFRA Scientific Advisory Panel. EPA-SAB-EC-00-017. Washington, D.C.: EPA. Available at HtmlResAnchor www.epa.gov/science1/pdf/ec0017.pdf.
National Bioethics Advisory Commission (NBAC). 2001. Ethical and Policy Issues in Research Involving Human Participants: Vol. 1. Bethesda, MD: U.S. Government Printing Office.
National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (National Commission). 1979. Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research. Washington, D.C.: U.S. Government Printing Office.
National Research Council (NRC). 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, D.C.: National Academy Press.
NRC. 1994. Science and Judgment in Risk Assessment. Washington, D.C.: National Academy Press.