The Time Has Come for Reconsidering the Role of Generic Default Assumptions Based on "Conservative Policy Choice" in Scientific Risk Assessments

Robert C. Barnard, Roger O. McClellan, and Donald L. Morgan

Robert C. Barnard is Of Counsel to, and Donald L. Morgan is Senior Partner in, the Washington, D.C., office of the firm Cleary, Gottlieb, Steen & Hamilton. Roger O. McClellan, D.V.M., D.B.A.T., D.A.B.V.T., is formerly President of the Chemical Industry Institute of Toxicology and past Chairman of the EPA Clean Air Act Scientific Advisory Committee. He is an Adjunct Professor at Duke University, the University of North Carolina, and the University of New Mexico. He is now a consultant and is active in a number of scientific advisory committees.

[31 ELR 10873]

The use of default assumptions in risk assessment originated in unusual conditions. In the early 1970s, President Richard M. Nixon had declared a war on cancer. The U.S. Environmental Protection Agency (EPA) was a new agency created to deal with environmental hazards. While cancer was a well-known problem, scientific understanding of its causes were limited, and experimental methods were in an early stage of development. Although cancer was known to represent perhaps as many as 200 diseases involving disturbances in cell growth and death, some scientists proposed to reduce existing knowledge regarding causation to simplified generic statements referred to as "principles" applicable to the entire broad class of cancer.

In its 1976 Interim Procedures and Guidelines, EPA adopted the concept of "General Principles Concerning the Assessment of Carcinogenic Data."¹ The interim guidelines created a Cancer Assessment Group to review the data and provide advice on risk assessment. It was in that atmosphere that the default assumptions were developed. Scientific understanding of carcinogenesis has advanced significantly since then, but EPA continues to use the 1970s generic default assumptions. Nevertheless, EPA has sent some mixed signals about continuing their use.

It has become traditional for each new administration to set forth its guiding regulatory philosophy and principles. The purpose of this Dialogue is to urge the Bush Administration to seek a solution to the long-standing debate on the role of assumptions in risk assessment, particularly the role of generic default assumptions based on a precautionary policy referred to as "conservatism." Conservatism in risk assessment refers to "science policy" assumptions that are chosen to fill gaps and uncertainties in data and to guide the selection of data and methodologies used in risk assessment to avoid underestimating risk.

Because the conservative policy choices in each step in the risk estimation process are multiplied together, using the conservative upper bound estimates at each step in the risk assessment process enhances the calculated risk level. As a consequence, the calculated final risk value can significantly overestimate the real risk. For example, when referring to the default linearized multistage model, the currently effective 1986 EPA cancer guidelines state:

It should be emphasized that the linearized multistate model leads to an upper limit to the risk that is consistent with some proposed mechanisms of carcinogenesis. Such an estimate, however, does not necessarily give a realistic prediction of risk. The true value of the risk is unknown and may be as low as zero.²

Moreover, the nature and magnitude of the overestimate is not disclosed to the risk manager or to the public. Thus, the undisclosed consequences of using default assumptions are not consistent with the precepts of full disclosure and transparency.

The Clean Air Act (CAA) Amendments of 1990 directed EPA to have the National Academy of Sciences (NAS) (through the National Research Council (NRC)) review EPA's risk assessment methods and to suggest improvements in such methods. The NAS Committee had available during its deliberations an early draft of EPA's proposed revision of its 1986 guidelines for cancer risk assessment.³ The NAS report, Science and Judgment in Risk Assessment, was published in 1994.⁴ The NAS noted that "EPA's risk assessment practices rest heavily on inference guidelines, or, as they are often called 'default options.'"⁵ The NAS report reviewed the debate concerning the default options and reached two very important conclusions:

1. EPA should have principles for choosing default options and when to depart from them. The choice of such principles goes beyond science and inevitably involves policy choices and how to balance them.⁶

2. Although the choice of the "level of conservatism" is a risk manager's prerogative, managers might be operating in the dark about how "conservative" the choices are if the uncertainty (and hence the degree by which the estimate may fall above or [31 ELR 10874] below the true value) is ignored or assumed, rather than calculated.⁷

EPA then issued its proposed revisions to the cancer guidelines in 1996 in response to the NAS report.⁸ The proposed guidelines have been extensively commented on and have been subjected to peer review. Although the proposed guidelines have been used by EPA on an ad hoc basis,⁹ they have not been formally finalized and issued. Thus, the reconsideration of science policy generic default assumptions remains an open issue.

After a brief discussion of the origin of the generic defaults, this Dialogue will review the recommendations set forth in the NAS' Science and Judgment in Risk Assessment report, comments in the Office of Management and Budget's (OMB's) interagency report to implement President Clinton's Executive Order No. 12866, and the report of the Presidential/Congressional Commission on Risk Assessment and Risk Management. This review demonstrates that it is time to review the role of default assumptions. Although assumptions may be necessary to deal with data uncertainties, EPA should follow the advice of the NAS and set forth the scientific and policy basis for each default option. We urge that preference be given to incorporating available scientific information. Use of default assumptions should be justified scientifically and used only in the absence of relevant data. An uncertainty analysis should be made to identify and estimate the impact of the default assumptions on the estimated level of risk.

Background Information on Generic Default Options

Generic default assumptions in scientific risk assessment had their origin in the 1970s when the regulatory agencies, i.e., EPA, the U.S. Food and Drug Administration, the Occupational Safety and Health Administration, and the Consumer Product Safety Commission, were developing their guidelines for scientific risk assessment with a focus on cancer. At that time the focus at EPA was on pesticides, and much of the debate centered on the testimony of Dr. Umberto Saffiotti, a National Cancer Institute scientist and a witness in an administrative pesticide proceeding in which he reduced the science of carcinogenesis to what he called "general principles" of carcinogenesis. In a subsequent administrative proceeding, EPA attorneys moved that the "principles" presented in the prior hearing should not require further proof and should be accepted as facts subject to judicial notice.¹⁰ The hearing officer rejected the motion. Nevertheless, EPA in its interim cancer guidelines, which were used until 1986, adopted the concept of "General Principles Concerning the Assessment of Carcinogenic Data" that ultimately led to the default options.¹¹

The Red Book

Much of the rationale for the use of assumptions in risk assessment is traceable to the NAS 1983 report, Risk Assessment in the Federal Government: Managing the Process (familiarly known as the Red Book).¹² The Red Book established the basic principles of the separation between scientific risk assessment and risk management, and identified four steps in risk assessment: hazard assessment, exposure assessment, dose response assessment, and risk characterization.

Referring to risk assessment, the Red Book stressed, "the dominant analytic difficulty is the pervasive uncertainty,"¹³ and to "make judgments amid such uncertainty, risk assessors must rely on a series of assumptions."¹⁴

Under the heading "Scientific and Policy Judgments in Risk Assessment," the Red Book distinguished between scientific principles to be addressed in evaluating a particular subject matter and "inference options." "Inference options are choices based on a mixture of scientific fact and consensus, on informed scientific judgment, and on policy determination based on 'conservatism.'"¹⁵

The Red Book commented on the nature of the policy involved in the choice of inference assumptions:

A review of the list of components [in the risk assessment process] reveals that many components lack definitive scientific answers, that the degree of scientific consensus concerning the best answer varies (some are more controversial among scientists than others), and that the inference options available differ in their degree of conservatism. The choices encountered in risk assessment rest, to varying degrees, on a mixture of scientific fact and consensus, on informed scientific judgment, and on policy determinations (the appropriate degree of conservatism).¹⁶

The NAS did not specify how to determine the "appropriate degree of conservatism," nor did it consider how to evaluate the impact of the conservative choices on the calculated risk value. Perhaps more importantly, the Red Book did not consider how the conservatism impact on the risk estimate [31 ELR 10875] could be disclosed and the process made transparent to the risk manager and the public.

The Red Book, however, did recognize that policy choices based on "conservatism" raise significant questions of intrusion on risk management:

That a scientist makes the choices does not render the judgments devoid of policy implications. Scientists differ in their opinions of the validity of various options, even if they are not consciously choosing to be more or less conservative …. From the regulatory official's point of view, the same choice may appear to be a value decision as to how conservative regulatory policy should be, given the lack of empirical basics for choice.

A risk assessor, in the absence of a clear indication based on science, could choose a particular approach (e.g., the use of an extrapolation model) solely on the basis of the degree to which it is conservative, i.e., on the basis of policy implications. Furthermore, a desire to err on the side of overprotection of public health by increasing the estimate of risk could lead an assessor to chose the most conservative assumptions throughout the process for components on which science does not indicate a preferred choice.¹⁷

The default options are used in the first three stages of risk assessment (hazard identification, exposure assessment, and dose response assessment). The conservative bias from default options is multiplied in the risk assessment process, which can result in a significant overestimate of the level of risk; that overestimate is not disclosed in the fourth stage of risk assessment—risk characterization. Thus, as matters now stand, the default assumption "conservative" bias impacts the whole regulatory process. The economist relies on the risk assessment in assessing costs and benefits, and the risk manager and the public must make decisions without being aware of the magnitude of the conservative bias.

Despite the potential intrusion of the policy-chosen inference options on the risk manager's functions, the Red Book did not attempt to describe what is meant by the term "appropriate level of conservatism." Nor did the Red Book propose the use of an uncertainty analysis to provide information on the impact of policy-selected assumptions on the level of the estimated risk. Thus, the impact of an "appropriate" conservative policy choice was obscured and not presented to the risk manager or the public.

Major Review of Default Options

Science and Judgment in Risk Assessment

The NAS 1994 report, Science and Judgment in Risk Assessment, is a very important scientific critique of default options. This report contains a short history of the development of the use of safety factors and the efforts to develop default assumptions. The report reviewed the Red Book's views on generic policy inferences and recommended that EPA should have "guiding principles for choosing default options and for judging when and how to depart from them."¹⁸ It was in this context that the NAS made the statement that the "choice of such principles goes beyond science":

The choice of such principles goes beyond science and inevitably involves policy choices on how to balance such criteria. After extensive discussion the committee found that it could not reach consensus as to what the principles should be or whether it is appropriate for this committee to recommend principles.¹⁹

The NAS committee explained that it had had an extensive debate focused on default options. "EPA's risk assessment practices rest heavily on inference 'guidelines' or, as they are often called 'default options.'"²⁰ (An appendix to the report cites two papers addressing the use of default assumptions: one paper prepared by Adam Finkel argued for the use of "plausible conservatism" in selecting defaults; the second paper, prepared by Roger McClellan and Warner North, argued for the full use of scientific information in risk assessment.) The report lists eight of the principal default options taken from EPA's 1986 cancer guidelines.²¹ The list illustrates the very wide scope of the generic options and the potential impact on the risk assessment process.

In the section entitled "Problems With EPA's Current Approach to Uncertainty," the NAS stated that the choice of the level of conservatism is a risk management problem:

Probably most fundamentally, without uncertainty analysis it can be quite difficult to determine the conservatism of an estimate. In an ideal risk assessment, a complete uncertainty analysis would provide a risk manager with the ability to estimate the risk for each person in a given population in both actual and projected scenarios of exposure; it would also estimate uncertainty in each prediction in quantitative, probabilistic terms. But even a less exhaustive treatment of uncertainty will serve an important service: it can reveal whether the point estimate [31 ELR 10876] used to summarize the uncertain risk is "conservative," and, if so, to what extent. Although the choice of the "level of conservatism" is a risk-management problem, managers may be operating in the dark about how "conservative" these choices are if the uncertainty (and hence the degree to which the risk estimate may fall above or below the true value) is ignored or assumed, rather than calculated.²²

The NAS recommended that EPA conduct an uncertainty analysis in order that the risk manager can be informed of the impact of conservative choices in the estimated risk level:

. EPA should conduct formal uncertaintyanalyses, which can show where additional research might solve major uncertainties and where it might not.

. EPA should consider in its risk assessment the limits of scientific knowledge, the remaining uncertainties, and the desire to identify errors of either overestimation or underestimation.

. EPA should develop guidelines for quantifying and communicating uncertainty, e.g., for models and data sets, as it occurs in each step in the risk assessment process.²³

Chapters 7 to 12 in the NAS report contain a careful discussion of uncertainties, and the need for identification and the means of measuring the uncertainties qualitatively and quantitatively. The following recommendations illustrate the approach recommended by the NAS:

. EPA should rigorously establish the predictive accuracy and uncertainty of its methods and models, and the quality of data used in risk assessment.

. EPA should develop guidelines for the amount and quality of emission information for particular risk assessments and for estimating and reporting uncertainty in emission estimates, e.g., the predictive accuracy and uncertainty associated with such use of Human Exposure Model (HEM) for exposure assessment.²⁴

The NAS made a number of recommendations to reduce the use of default options, such as an iterative risk assessment process in which the defaults are used in the initial screening stage to determine whether a more complete risk assessment is needed. If the risk suggests regulatory attention, then more complete risk assessments using all the data should be used, including data developed as a result of subsequent research.

The report states that the guidelines do not set out the defaults "with ideal clarity"; "nor has the agency explicitly articulated the scientific and policy basis for its options."²⁵ It recommends that EPA "develop guidelines for quantifying and communicating uncertainty (e.g. for models and data sets) as it occurs into each step in the risk assessment process."²⁶ The report also recommended several procedures for reducing reliance on defaults and strongly urged that EPA publish the scientific basis for the use of defaults and formulate criteria for departure from the defaults.

In response to the 1994 NAS report, EPA published the 1996 draft of proposed revised cancer guidelines.²⁷ The draft guidelines did not address the NAS conclusion that principles on which default assumptions are based are beyond science, nor did it respond to the NAS conclusion that the level of conservatism is a risk management problem and that managers would be operating in the dark if not informed of how conservative estimates are. The fact that EPA has not finalized the proposed cancer risk assessment guidelines emphasizes the importance of reconsideration of the use of generic default options by the new administration.

The need for timely completion of the guidelines revision process is critical in view of EPA's use of the proposed revised guidelines in an ad hoc manner for several important regulatory decisions.

Executive Order No. 12866 and the OMB Interagency Memorandum

The Clinton Administration during its first year in office undertook a very complete examination of what was called "best practices" for preparing scientific risk assessments and cost-benefit analyses of proposed regulations. The "best practices" did not involve a specific evaluation of default assumptions but did address the issue of conservatism and its impact on the regulatory process.

Executive Order No. 12866 set forth the Clinton Administration's regulatory philosophy and principles.²⁸ It stated broad principles relating to the review of regulations. It directed that in deciding to regulate, "the agencies should assess all costs and benefits of available regulatory alternatives …" and "shall … propose or adopt a regulation only on a reasoned determination that the intended regulation justify its costs."²⁹ An interdepartmental committee was formed to develop "best practices" for implementing the order. The report of that interagency committee was designated as an OMB document and issued as a White House press release on January 22, 1996.³⁰

The OMB report addressed risk assessment and risk management issues that are relevant to the generic default question. The OMB report pointed out that risks, costs, and benefits must be expressed as probability distributions because of their uncertainty:

The term "risk" in this document refers generally to a probability distribution over a set of outcomes ….

Often risks, benefits and costs are measured imperfectly because key parameters are not known precisely; instead economic analysis must rely on statistical probability distribution for the values of parameters ….³¹

Risk assessment should be conceptually appropriate to the needs of the risk manager:

The risk assessment should generate a credible, objective, realistic, and scientifically balanced analysis, present [31 ELR 10877] information on hazard, dose response and exposure; … and explain the confidence in each assessment by clearly delineating strengths, uncertainties and assumptions, along with the impacts of these factors on the risk assessments.³²

As for conservatism, the OMB commented:

The data, assumptions, models and inferences used in risk assessment to construct quantitative characterizations of the probabilities of occurrence of health, safety, or ecological effects should not reflect unstated or unsupported preferences for protecting public health and the environment, or unstated safety factors to account for uncertainty and unmeasured variability. Such procedures may introduce levels of conservatism that cumulate across assumptions and make it difficult for the decision maker to evaluate the magnitude of the risks involved.³³

On the need to disclose the effects of assumptions that are necessary in the absence of adequate valid data, the OMB counseled:

Alternate models and assumptions should be used in risk assessment as needed to provide decision makers with information on the robustness of risk estimates and estimates of regulatory impacts.³⁴

The OMB also noted that the "best practices" included further directions on the presentation of data and assumptions:

Where feasible, data and assumptions should be presented in a manner that permits quantitative evaluation of their incremental efforts. The cumulative effects of assumptions and inferences should also be evaluated ….

The material provided should permit the reader to replicate the analysis and quantify the effects of key assumptions. Such analyses are becoming increasingly easy to perform because of advances in computing power and new methodological developments. Thus the level and scope of disclosure and transparency should increase over time.³⁵

The OMB report stresses the importance of full disclosure and transparency:

The treatment of uncertainty in developing risk benefit and cost information must be guided by the principles of full disclosure and transparency as with other elements of an EA [Economic Analysis]. Data, models, and their implications for risk assessment should be identified in the risk characterization. Inferences and assumptions should be identified and evaluated explicitly, together with adequate justification of choices made, and assessment of those choices on the analyses ….³⁶

The OMB report further commented on how uncertainty can be calculated and disclosed. Overall uncertainty is typically a consequence of uncertainty about many different factors. Appropriate statistical techniques should be used to combine uncertainties about separate factors into an overall probability distribution for a risk. When such techniques cannot be used, other methods may be useful for providing more complete information.

The OMB report discussed various procedures for use in evaluating uncertainty and the impact of assumptions including Monte Carlo analysis, sensitivity analysis, delphi methods, and meta-analysis.

Overall the OMB report reflects care and thought in addressing the importance of a transparent and full disclosure of the impact of assumptions to assure that risk analysis provides a sound basis for economic analysis and for use by the risk manager.

The Report of the Presidential/Congressional Commission on Risk Assessment and Risk Management

The Presidential/Congressional Commission on Risk Assessment and Risk Management was established pursuant to CAA § 303 to address risk assessment and risk management issues relating to federal laws to prevent cancer and other chronic health effects. The commission issued its report in 1997.³⁷ The commission proposed a new Risk Management Framework with the overall goal of sustainable development:

We recognize that it is time to modify the traditional approaches to assessing and redefining risks that have relied on a chemical-by-chemical, medium-by-medium, risk-by-risk strategy. While risk assessment has been growing more complex and sophisticated, this output of risk assessment for the regulatory process often seems too focused on refining assumption-laden-mathematical estimates of small risks associated with exposure to individual chemicals rather than on the overall goal—risk reduction and improved health status.³⁸

In order to obtain that goal, the commission recommended:

We call for explicit description of assumptions, data sources, sources of uncertainties, and distribution of costs and benefits across society associated with economic analysis, in parallel with descriptions associated with risk assessments.³⁹

The commission pointed out that economic analysis relies on knowledge of a distribution of risks:

Economic analysis relies not on point estimates of risks, but on entire probability distribution of potential costs and benefits for an entire affected population, which cannot be extrapolated from an upper bound estimate of individual risk …. If a scientific assessment of risk provides only information on the upper bounds of hazards the economic analysis will either overstate the net benefits to the general population or be relevant only to the tail of the risk distribution.⁴⁰

The commission commented on the NAS' Science and Judgment in Risk Assessment report. Under the heading [31 ELR 10878] "Maximal Use of Scientific Information Versus Plausible Conservatism," the commission commented:

The NRC report stated that EPA operates in a decision-making context that imposes pressures on the conduct of risk assessments and that these contextual pressures have lead to recurrent problems of scientific credibility. Criticisms of EPA's risk assessment focus on three basic decision-making structural and functional problems:

. Unjustified conservatism, often manifested as unwillingness to accept new data or abandon default options.

. Undue reliance on point estimates generated by risk assessment.

. Lack of appropriate conservatism to accommodate such issues as synergism, human variability and ad hoc departures from established procedures.

The NRC report pointed out that whereas EPA's practices rely heavily on default options, EPA has never articulated the scientific or policy basis of those options. Because of limitations on time, resources, scientific knowledge, and available data, the report concluded that EPA should generally retain its conservatism, default-based approach to risk assessments for screening analysis in standard setting. The authors offered several recommendations to make this approach more effective:

. Use an iterative approach to risk assessment.

. Provide justification for defaults and establish a procedure that permits departure from defaults.

. When communicating information to decision makers and the public, identify the sources and magnitude of the uncertainty associated with risk estimates.⁴¹

The commission made the following comments:

The Commission concurs that default assumptions are a necessary part of the conduct of risk assessment. Risk assessments make predictions about the unknowable by using inferences that have not been or cannot be adequately tested with the scientific method. In the absence of adequate scientific information, science and policy based assumptions are appropriate. The Commission supports the goal of transparency and believes that assumptions used in risk assessments and uncertainty associated with their results should be clearly identified and justified.⁴²

There is general agreement on the principles of separation of risk assessment and risk management. There is also general agreement that the choice of the level of conservatism is a risk management problem.⁴³ The NAS report on science and judgment and the Presidential/Congressional Commission on Risk Assessment and Risk Management report recommend a limited use of default options with an accompanying formal uncertainty analysis to identify and quantify the impact of the uncertainties on the calculated risk level from the default options.

EPA's Equivocal Response to Major Criticisms of Default Options

EPA has sent mixed signals regarding the criticism and recommendations by the NAS and the Presidential/Congressional Commission on Risk Assessment and Risk Management relating to the use of default options. The 1996 draft cancer guidelines discussed above have never been finalized. There are other signals, but they don't provide a thorough review of default options that we argue is called for by the new administration. A brief review of other EPA signals underlines the need for reconsideration.

Default as "Long Held EPA Policy"—The Chlorine Council Decision

The decision of the U.S. Court of Appeals for the District of Columbia Circuit in Chlorine Chemistry Council v. U.S. Environmental Protection Agency⁴⁴ involved the Safe Drinking Water Act (SDWA).⁴⁵ Under that Act, EPA sets standards for drinking water contaminants using "the best available science." For each contaminant EPA sets a maximum contaminant level goal (MCLG) at which no known adverse effects occur and which allows an adequate margin of safety. The enforceable standard known as the maximum contaminant level (MCL) takes practical considerations into account and is to be set "as close to the MCLG as is feasible."⁴⁶

In 1994, EPA issued a proposed rule for four compounds known as trihalomethanes that are the byproduct of chlorination used to assure the microbial safety of drinking water. On the basis of data from rodents exposed to high levels, EPA concluded that the drinking water contaminant chloroform, a trihalomethane, was a probable human carcinogen. On July 29, 1994, EPA issued a proposed rule on trihalomethanes and proposed a zero MCLG for chloroform based on the default no-threshold assumption that any exposure to a carcinogen poses a risk of cancer.

The district court described the series of events that lead to the major issue:

In March 1998 EPA concluded that chloroform, a drinking water contaminant, exhibits a "nonlinear mode of carcinogenic action." … In other words, exposure to chloroform below some threshold level pose no risk of cancer. But in promulgating the MCLG it retained the existing standard of zero, which was based on the previously held assumption that there was no safe threshold…. EPA justified its action on a variety of grounds, including an alleged need to consult the report of its Science Advisory Board ("SAB"), which would not be available until after the statutory deadline for rulemaking had expired. Petitioners … petitioned this court for review, arguing that EPA violated the statutory mandate to use the "best available" evidence when implementing the provisions of the Safe Drinking Water Act.⁴⁷

The court continued:

[31 ELR 10879]

When EPA came to promulgate its final rule in December 1998, however, its MCLG was again zero…. It stuck with 1994's zero level despite its explicit statement that it now "believed that the underlying science for using a nonlinear extrapolation approach to evaluate the carcinogenic risk from chloroform is well founded." … It justified the action on the basis that "additional deliberations with the Agency's SAP on the analytical approach used" and on the underlying scientific evidence were needed "prior to departure from a long-held EPA policy."⁴⁸

After briefing was complete but before oral argument, EPA filed a motion for a voluntary remand to consider the Science Advisory Board's (SAB's) report on chloroform. Plaintiffs objected and the court refused. On the day of oral argument, EPA notified the court that it had consulted with the SAB and no longer believed it should continue its original decision and moved to vacate the MCLG.

Plaintiffs objected and the court held that EPA "had openly overrode" the best available scientific evidence. The court found the setting of the zero MCLG to be arbitrary and capricious and in excess of EPA's statutory authority.

Risk Characterization

In February 1995, EPA published Guidance for Risk Characterization, which included some clarifications and changes to give more prominence to certain issues, such as the need to explain the use of default assumptions, in response to the NAS Science and Judgment in Risk Assessment report.⁴⁹ This guidance document states that risk assessment is to be conducted "without considering issues such as cost, feasibility, or how the scientific analysis might influence the regulatory or site specific decisions."⁵⁰ Risk assessors are charged with, among other matters, "explaining confidence in each assessment by clearly delineating strengths, uncertainties and assumptions, along with the impact of those factors (e.g., confidence limits, use of conservative/non-conservative assumptions) on the overall assessment."⁵¹

EPA added:

Matters such as risk assessment priorities, degree of conservatism, and acceptability of particular risk levels are reserved to decision makers who are charged with making decisions regarding protection of public health.⁵²

In 2001, EPA released its Risk Characterization Guide.⁵³ The document states that "effective risk characterization depends on Transparency, Clarity, Consistency and Reasonableness (TCCR)."⁵⁴ Under the heading "What Are Criteria for Transparency," EPA states:

Transparency achieves full disclosure in terms of:

(a) the assessment approach employed

(b) use of assumptions and their impact on the assessment

(c) plausible alternatives and the choices made among those alternatives

(d) the use of models in measurements and their impact on the assessment

(e) plausible alternatives and the choice among those alternatives

(f) impact of one choice vs. another on the assessment

(g) significant data gaps and their implication for the assessment

(h) the scientific conclusions identified separately from default assumptions and policy calls

(i) the major risk conclusions and the assessor's confidence and uncertainties in them

(j) the relative strength of each risk assessment component and its impact on overall assessment (e.g., the case for the agent proposing a hazard is strong, but the overall risk assessment is weak because the exposure is weak).⁵⁵

Despite these comments on risk characterization, EPA did not note any proposed changes in its default assumptions, and it summarized the responsibility of the assessor, among other things, to "describe the uncertainties in the risk assessment and the default positionused to address those uncertainties."⁵⁶

A Glance at the Future

Default options were developed to respond to uncertainties and gaps in data. Recent significant scientific advances, however, have opened the possibility of important improvements in risk assessment methodology. Recent scientific advances in genomics and the function of ribosomes raise the possibility that significant new data on the function of genes and the importance of proteins produced by ribosomes in maintaining homeostasis in both the entire organism and in cells could lead to significant improvements in risk assessment procedures. Procedures for utilizing these new data are being evaluated with a new "Toxicogenomics" initiative established by the National Institute for Environmental and Health Sciences. A brief comment on the new biological science revolution is helpful to understand the potential impact on the role of default assumptions in risk assessment.

In the December 22, 2000, issue of Science, the journal announced that the number one scientific breakthrough of the year was the success in determining the virtually complete genome sequence of the mouse and the 95% completion of the human genome sequence.⁵⁷ The number two breakthrough—"Ribosome's Revelations"—is the understanding of the function of the cellular ribosomes in catalyzing the production of proteins coded by the genes. The proteins and their interactions are essential to maintaining homeostasis of the whole organism and the cells.⁵⁸

The gene sequencing efforts of government, industry, and private laboratories have produced a remarkable advance in scientific understanding. Previously the study of genes was [31 ELR 10880] a slow, tedious process of attempting to identify and characterize single genes. In June 2000, the publicly funded Genome Project and the private company Celera announced the virtual completion of the sequencing of the human genome. Science commented that "this breakthrough might well be the breakthrough for the decade, perhaps even the century."⁵⁹ A more complete picture of these major advances appeared in simultaneous publications: the private Celera gene sequence program results were published in the April 15, 2001, issue of Science, and the public consortium gene sequence results were published in the April 15, 2001, issue of Nature.⁶⁰ The results of both programs have contributed to the results of other mapping programs.

The ribosome advances relate to the operative arm of the genes. Cells are "protein factories." Ribosomes are macromolecules made up of units of ribonucleic acid (RNA). Ribosomes catalyze the assembly of proteins from amino acids to form proteins that are genetically encoded by the genes through the RNA. "Biologists have been eagerly awaiting a close up of this complex for cues how it makes protein with such stunning accuracy."⁶¹ The proteins are essential to the complex control and feedback that assures that the proteins are produced that are necessary to maintain an ordered metabolism in a changing environment. Mapping the families of human and animal proteins is underway. A high-speed industry effort to uncover the structure of proteins is also underway. These advances in scientific knowledge open the door to a new integrative approach to the essential life functions.

These new branches of science—functional genomics, proteomics, and protein structure—are already enabling the pharmaceutical, agricultural, and health industries to take advantage of this stunning new source of biological data. Governments and private industries have already developed enormous libraries of data. The database generated by the public Genome Project is freely available. Private industry databases are generally available for a fee. As noted above, the regulatory agencies are also in the forefront of examining how the growing volume of data can be used in risk assessment. The scientific journals are replete with articles on the biological revolution.

This greatly simplified glance at these new scientific breakthroughs is included to indicate that these advances have the potential to lead to significant improvements in risk assessment. The availability of data at the molecular level promises to close many of the data gaps and greatly improve the understanding of disease causation, including the role of exogenous factors in the environment or workplace.

It is fair to say, however, that the door has been opened but much needs to be done before this kind of data can be incorporated in risk assessment. When the genomic and protein data that relate specifically to adverse consequences from exposure to a potential toxicant are identified, the need for assumptions or generic defaults on mechanisms of action and cause in risk assessment could be greatly reduced. The data may also explain differences between toxicity observed in laboratory animals and observed in humans, and it may reveal dose response information at levels of exposure more typical of those encountered by humans as distinguished from high exposure levels typically used in animal experiments. An important obstacle will be the demand for privacy of tests and test results.

Genomic test methods have improved dramatically. It is possible to assemble gene arrays so that many thousands of genes for testing can be included in a single array. A clear challenge will exist to distinguish in both laboratory animals and humans whether observed gene changes—either up regulated (increased activity), or down regulated (reduced activity)—are indicative of adverse effects, harbingers of adverse effects that may be observed with longer periods of exposure and observation, or indicate homeostatic changes reflective of physiological adaptation to an exogenous agent.

Conclusion

It seems clear that the stage has been set for a reconsideration of the role of default assumptions in risk assessment. Although EPA in the Chlorine Council decision characterized defaults as "long held EPA policy," the critical reviews of the use of default assumptions by the NAS, the OMB document, and the Presidential/Congressional Commission on Risk Assessment and Risk Management support the use of substance-by-substance data. The NAS' Science and Judgment in Risk Assessment report concluded that the "degree of conservatism" in regulatory decisions is a risk management responsibility. EPA, in its Guidance for Risk Characterization report, agrees and proposes that the impact of assumptions on the risk assessment should be made transparent in the risk characterization.

There is general agreement that the role of defaults should be significantly reduced. In response to the critical review and recommendations, new EPA risk assessment cancer guidelines should set forth clearly the science and policy basis for defaults and criteria for departure from defaults. EPA should be required to explain the science that justifies use of a default assumption and should, in the interest of transparency, conduct an uncertainty analysis and disclose the impact on the risk assessment of both the default and plausible alternative assumptions in each case. With that information, the risk manager can make an informed decision regarding the level of conservatism appropriate for each regulatory decision.

While the growing mass of genomic and proteomic knowledge may not yet be ready for general integration into the risk assessment process, it seems reasonable that as data from the biological revolution become available, the need for default assumptions should be greatly reduced. The EPA carcinogenic guidelines should give priority to the integration of new scientific data and understanding on causation into risk assessment as the information becomes available.

The risk manager and the public are entitled to the best estimate and the distribution of estimates of the true risk value as supported by the available science. It is no longer reasonable for EPA to state, as it did in the 1986 risk assessment guidelines, that the default linearized multistage procedure "does not necessarily give a realistic picture of the risk, the true value of the risk is unknown and may be as low as zero."⁶² While the distribution of estimates includes "zero," it should be explicitly stated.

1. See U.S. EPA, Health Risk and Economic Impact Assessments of Suspected Carcinogens: Interim Procedure and Guidelines, 41 Fed. Reg. 21402, 21404 (May 25, 1976).

2. U.S. EPA, Guidelines for Carcinogen Risk Assessment, 51 Fed. Reg. 33992, 33998 (Sept. 24, 1986) (emphasis added).

3. The draft guidelines supplied to the NAS were not published.

4. NAS, SCIENCE AND JUDGMENT IN RISK ASSESSMENT (National Academy Press 1994) [hereinafter NAS SCIENCE AND JUDGMENT REPORT]. One of the authors (McClellan) was a member of the NAS committee and published a commentary on the report, Roger O. McClellan, Science and Judgment in Risk Assessment, 20 REG. TOXICOLOGY & PHARMACOLOGY 2142 (1994).

5. NAS SCIENCE AND JUDGMENT REPORT, supra note 4, at 6-1.

6. Id. at E-6 to E-7 (emphasis added).

7. Id. at 9-7.

8. U.S. EPA, Proposed Guidelines for Carcinogenic Risk Assessment, 61 Fed. Reg. 17960 (Apr. 23, 1996).

9. A good source of information on the developments and controversies during the development period is found in CONGRESSIONAL OFFICE OF TECH. ASSESSMENT, TECHNOLOGIES FOR DETERMINING CANCER RISKS FROM THE ENVIRONMENT (1986). The history is also reviewed in the NAS SCIENCE AND JUDGMENT REPORT, supra note 4.

10. EPA Moves to Have 17 Cancer Principles Officially Noticed in Mirex Case, PESTICIDE CHEM. NEWS, Sept. 17, 1975, at 10-12; EPA Continues Effort to Prevent Mirex Questions From Going to NAS Committee, PESTICIDE CHEM. NEWS, Sept. 24, 1975, at 27.

On September 19, 1975, Dr. Frank J. Rauscher, Director of the National Cancer Program, convened the Subcommittee of the National Cancer Institute to consider general guidelines for interpreting carcinogenesis data. The subcommittee concluded:

The criteria adopted are based upon those generally used in the application of the scientific method to any question. In assembling these criteria, the Subcommittee reiterates that there is no simple and universal definition of either carcinogenesis or neoplasia. The criteria which are described are general guidelines and no [sic] rigid, universal criteria.

SUBCOMM. ON ENVTL. CARCINOGENESIS, GENERAL CRITERIA FOR ASSESSING EVIDENCE OF CARCINOGENESIS OF CHEMICAL SUBSTANCES: REPORT OF THE SUBCOMMITTEE ON ENVIRONMENTAL CARCINOGENESIS: NATIONAL CANCER ADVISORY BOARD, 58 J. NAT'L CANCER INST. 461-65 (1977).

11. 41 Fed. Reg. at 21402.

12. NRC, RISK ASSESSMENT IN THE FEDERAL GOVERNMENT: MANAGING THE PROCESS (National Academy Press 1983) [hereinafter RED BOOK].

13. Id. at 11.

14. Id. at 12.

15. Id. at 36 (emphasis added).

16. Id. (emphasis added).

17. Id. at 36-37.

18. NAS SCIENCE AND JUDGMENT REPORT, supra note 4, at E-6.

19. Id. at E-7 (emphasis added).

20. Id. at 6-1.

21. The eight principle default options are:

. Laboratory animals are a surrogate for humans in assessing cancer risks; positive cancer-bioassay results in laboratory animals are taken as evidence of a chemical's cancer-causing potential in humans.

. Humans are as sensitive as the most sensitive animal species, strain, or sex evaluated in a bioassay with appropriate study-design characteristics.

. Agents that are positive in long-term animal experiments and also show evidence of promoting cocarcinogenic activity should be considered as complete carcinogens.

. Benign tumors are surrogates for malignant tumors, so benign and malignant tumors are added in evaluating whether a chemical is carcinogenic and in assessing its potency.

. Chemicals act like radiation at low exposures (doses) in inducing cancer; i.e., intake of even one molecule of a chemical has an associated probability for cancer induction that can be calculated, so the appropriate model for relating exposure-response relationships is the linearized multistage model.

. Important biological parameters, including the rate of metabolism of chemicals, in humans and laboratory animals are related to body surface area. When extrapolating metabolic data from laboratory animals to humans, one may use the relationship of surface area in the test species to that in humans in modifying the laboratory animal data.

. A given unit of intake of a chemical has the same effect, regardless of the time of its intake; chemical intake is integrated over time, irrespective of intake rate and duration.

. Individual chemicals act independently of other chemicals in inducing cancer when multiple chemicals are taken into the body; when assessing the risks associated with exposures to mixtures of chemicals, one treats the risks additively.

Id. at 6-4 (guideline citations omitted).

22. Id. at 9-7 (emphasis added).

23. Id. at E-11 to E-12.

24. Id. at E-8 to E-9.

25. Id. at E-11.

26. Id.

27. 61 Fed. Reg. at 17960.

28. Regulatory Planning and Review, Exec. Order No. 12866, 58 Fed. Reg. 51735 (Oct. 4, 1993), ADMIN. MAT. 45070.

29. Id. at 51736, ADMIN. MAT. at 45070.

30. U.S. OMB, ECONOMIC ANALYSIS OF FEDERAL REGULATIONS UNDER EXECUTIVE ORDER NO. 12866 (Jan. 11, 1996) available at http://www.whitehouse.gov/omb/inforeg/riaguide.html.

31. Id.

32. Id.

33. Id. (emphasis added).

34. Id.

35. Id.

36. Id. (emphasis in original).

37. The report was issued in two volumes: volume one, PRESIDENTIAL/CONGRESSIONAL COMMISSION ON RISK ASSESSMENT AND RISK MANAGEMENT, FRAMEWORK FOR ENVIRONMENTAL HEALTH RISK MANAGEMENT (1997); and volume two, PRESIDENTIAL/CONGRESSIONAL COMMISSION ON RISK ASSESSMENT AND RISK MANAGEMENT, RISK ASSESSMENT AND RISK MANAGEMENT IN REGULATORY DECISION MAKING (1997). Volume two, labeled "Final Report," was issued after receiving public comments on volume one. The references in this Dialogue are to volume two [hereinafter COMMISSION REPORT].

38. COMMISSION REPORT, supra note 37, at i.

39. Id. at iv.

40. Id. at 97.

41. Id. at 65-166 (emphasis added).

42. Id. at 166 (emphasis added).

43. The Red Book in a footnote stated that in its view, guidelines are not intended to describe the "legal weight of any codified [by guidelines] standards or principles." RED BOOK, supra note 12, at 51 n.*.

44. 206 F.3d 1286, 30 ELR 20473 (D.C. Cir. 2000).

45. The SDWA can be found at 42 U.S.C. §§ 300f to 300j-26, ELR STAT. SDWA §§ 1401-1465.

46. Id. § 300g-1(b)(4)(B), ELR STAT. SDWA § 1412(b)(4)(B).

47. 206 F.3d at 1287, 30 ELR at 20473.

48. Id. at 1288, 30 ELR at 20474 (citations omitted).

49. U.S. EPA, GUIDANCE FOR RISK CHARACTERIZATION (1995).

50. Id. at 2.

51. Id. at 3 (emphasis added).

52. Id.

53. U.S. EPA, RISK CHARACTERIZATION GUIDE (2001), accompanied by, U.S. EPA, OVERVIEW OF THE RISK CHARACTERIZATION HANDBOOK (2001).

54. U.S. EPA, OVERVIEW OF THE RISK CHARACTERIZATION HANDBOOK, supra note 53, at 1.

55. U.S. EPA, RISK CHARACTERIZATION GUIDE, supra note 53, at 15.

56. Id. at 21.

57. News & Editorial Staff, Genomics Comes of Age, 290 SCIENCE 2220 (2000).

58. Id. at 2220-21.

59. Id. at 2221.

60. The Human Genome, 291 SCIENCE 1145 (2001); The Human Genome, 409 NATURE 745 (2001).

61. Genomics Comes of Age, supra note 57, at 2221.

62. U.S. EPA, Guidelines for Carcinogen Risk Assessment, 51 Fed. Reg. at 33998.