National Academies Press: OpenBook

Risk Assessment in the Federal Government: Managing the Process (1983)

Chapter: 2 Inference Guidelines for Risk Assessment

« Previous: 1 The Nature of Risk Assessment
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 51
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 52
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 53
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 54
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 55
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 56
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 57
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 58
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 59
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 60
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 61
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 62
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 63
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 64
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 65
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 66
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 67
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 68
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 69
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 70
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 71
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 72
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 73
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 74
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 75
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 76
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 77
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 78
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 79
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 80
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 81
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 82
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 83
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 84
Suggested Citation:"2 Inference Guidelines for Risk Assessment." National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: The National Academies Press. doi: 10.17226/366.
×
Page 85

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

II Inference GuideNnes for Risk Assessment INTRODUCTION Al DEFINITIONS An inference guideline* is an explicit statement of a predetermined choice among the options that arise in inferring human risk from data that are not fully ade quate or not drawn directly from human experience. A guideline might, for example, specify the mathematical model to be used to estimate the ef feats of exposure at low doses from observations based on higher doses. The most Important feature of guideline use i s that it change s the risk assessment process from one In which inference options are selected on a substance by-substance basis to one in which they are selected once and thereafter *The Committee hopes to avoid any misunderstanding result- ing from its use of the terms inference guideline and guideline (used for brevity in lieu of inference quide- l~ne). This terminology is potentially confusing, because Guidelines cat be understood as codified principles ad- dressed to a particular subject matter, risk assessment, or as describing the legal weight of any codified stand cards or principles. For the Committee, it has the former meaning. Inference guidelines are the principles followed by risk assessors in interpreting and reaching judgments based on scientific data. (Thus, our inference guides lines are d, stinct from the standards for toxicologic and other testing standards that many regulatory agencies and sc, entif ic bodies have adopted to govern, or at least influence, the generation of data later used in risk assessment. ) For many lawyers, the term guideline connotes the weight to be given to any set of codified principles, not ~1

~2 applied to an entire series of chemicals. In the absence of guidelines, assessors may well select the same infer- ence options for substance after substance in a given agency program, and a common set of inference options may emerge, In canon law fashion, from their consistent application In the program. But even the continued use of the same set of inference options over time does not necessarily imply that the assessors would make the same choices for every substance. Furthermore, outsiders would have no way of knowing what the common set is. In contrast, the use of guidelines makes more evident the generic choice oF inference options, which we have seen in Chapter 7, is based on both scientific and risk assessment policy considerations. HISTORY OF 'Gil USE OF GUIDErINES ~~F~.~v ~V~1nATION GUIDELINES FOR ~ The development and use of guidelines by a regulatory agency first became of major importance after Congress only those addressed to risk assessment, in legal pro- ceedings. The Food and Drug Administration, for example, has defined a guideline as an official pronouncement of the agency describing a procedure that satisfies legal requirements, but is not mandated by law. A more complete treatment of the distinction between binding regulations and other formal agency pronouncements appears in the section of this chapter entitled Degree to Which Guide- lines May Be Binding on an Agency and a Regulated Party. Tne Committee has used the term guideline to describe the principles by which risk assessments are to be per- formed, because that is the term Congress used in the legislation that authorized this study. The Committee was asked to consider the feasibility of establishing uniform frisk assessment guidelines." There is no ev~- dence that Congress was aware of the different meanings of the term. It obviously was seeking advice about the intellectual and scientific bases for codified principles for risk assessment, not the appropriate legal form for their adoption. Faced with possible confusion no matter which terminology it chose, the Committee has retained the language that Congress itself used to describe our inquiry.

~3 enacted amendments to the Federal Food, Drug, and Cos- metics Act in the 1950s and early 1960s. These laws, as applied to noncarcinogenic agents, required that food additives, color additives, drugs for animals, and pesti- cides be shown to be safe under their intended conditions of use before premarket approval by the Food and Drug Administration (FDA). The agency developed guidelines to provide a systematic way to deal with the legal requires meets embodied in the amendments. Although guidelines for the conduct of various types of toxicity tests received greatest notice, some attention was given to the problem of data interpretation for inferring human risk. For example, a 1959 publication written by several members of the FDA Division of Pharmacology, Appraisal of the Safety of Chemicals in Foods, Drums, and Cosmetics, is devoted primarily JO toxicity testing methods, but cons tains one chapter called Some Interpretative Problems in Evaluating the Safety of Food Additives. (Lehman et al., 1959). Although that publication, which served as a guide for both FOA and the regulated industry for at least a decade, was never published as a regulation, it was widely accepted by affected industrial concerns. In all cases except that of carcinogens, establishment of acceptable intakes was accomplished by applying saf ety factors to experimentally derived no~observe~effect exposures. Testing involved mostly the use of acute and subchronic (90-day) animal tests, although some long-term tests were required. The use of safety factors to estab- lish acceptable intakes was also recommended by the Food Protection Committee-of the National Research Council (ENS, 1970) and adopted by the Joint Food and Agri- cultural Organization and World Health Organization Expert Committees on Food Additives (1972) add Pesticide Residues (1965). This approach continues to be used for noncarcinogenic food additives and pesticides and, in slightly modified form, to define acceptable exposures to occupational and various environmental pollutants. These methods of assigning acceptable limits of expc- sure imply that the application of safety factors of various magnitudes to experimentally derived no-observed- effect em osures will ensure low risk. The acceptable exposure, whether expressed as an acceptable daily intake for a food additive or as a permissible exposure limit for an occupational agent, is derived by imposing untested assumptions (e.g., about the likely nature of dose response relations at low doses3 and by drawing inferences from sparse data. Safety evaluation schemes Inky therefore

54 be classif fed as a set of guidelines that e~hasi ze test- ing methods heavily and that afford methods of inference only scant attention. Recent efforts have dealt more directly with developing guidelines for risk assessment of noncarcinogenic effects. The Environmental Protection Agency (EPA) has proposed guidelines for chemical mutagenesis (EPA, 1980a) and has collected public ancients on them, but has yet to publish a f inal rul e. In addition, Me agency cosponsored two conferences with Oak Ridge National Laboratory on risk assessment methods for reproductive and teratogenic effects; the proceedings of the conferences have been published (ORNL/EPA, 1982). The Interagency Regulatory Liaison Group began to develop guidelines for risk assessment of reproductive and teratogenic effects, but the effort ceased with the disbanding of the group in 1981. The March of Domes Birth Defects Foundation (1981) has published the proceedings of a conference dealing with guidelines for studies of Herman populations exposed to mutagenic and reproductive hazards. Despite the increasing interest in noncarcinogenic effects, methods of estimating the risk of these effects have not been the subject of major public and scientific debate; attention has been devoted mainly to carcinogenic risk assessment. Much more critical review of the inferential methods for assigning risks to noncarcinogenic agents is warranted. GUIDELINES FOR CARCINOGENIC RISI; Ontil the late 1950s, few agents, either chemical or physical, had been regulated in this country on the basis of their carcinogenic action. One important regulated agent was ionizing radiation. Permissible exposures to radiation were set in a manner similar to that for noncar- cinogenic agents, by application of safety factors applied to specified exposures. In the debate over health effects of radioactive fallout from atomic weapons tests in the l95Os, evidence to support a nonthreshold theory for cancer induction emerged. Evidence was also accumulated to indicate that the nonthreshold theory might be appli- cable to chemical carcinogens. It was in this context that Congress enacted statutes* in the 1950s and early l *The enactment of these statutes did not necessarily bring a unique new concept to FDA. In the early 1950s,

~5 1960s that required FDA to ban Me use of food and color additives shown to be carcinogenic. The aSsm~tion, which differed from that underlying safety evaluation of noncarcinogens, was that no exposure could be presumed safe. Thus, a full risk assessment scheme was not needed for carcinogens. identification. The process stopped at hazard Many factors contributed to the later use of dose- response assessment, exposure assessment, and risk char- acterization to determine quantitative estimates of risk. One of these may have been the growing perception during the 1960s and 1970s that many kinds of risk could not be eliminated completely without unacceptable social and economic consequences. New laws reflecting this belief were enacted, and some agencies were required to balance the risk posed by carcinogenic agents against their per- ceived benefits. Quantitative risk assessment was the system developed to estimate the risk side of the bal- ance. Over a period of 2 decades, various expert come mittens sponsored by government agencies and other organizations published numerous reports dealing with carcinogenicity evaluation. Most of these were state- of-the art reports on aspects of carcinogenicity ~nfer- ence, and many suggested guidelines for hazard identifiers tion. More recent reports have dealt explicitly with quantitative risk assessment. The impetus for producing these reports was probably a belief in the federal re- search and regulatory communities that some scientific principles related to carcinogenic~ty data evaluation had to be continually reexamined and reaffirmed. This belief pervaded the public-health establishment not only in the United States, but also in other countries and in the United Nations. The following discussion examines efforts to develop and apply guidelines for the evaluation of carcinoge- nicity data by the federal regulatory agencies and the International Agency for Research on Cancer over the last decade--efforts that developed out of 2 decades of scientific consensus-building. before the' ~ enactment, the agency had prohibited three food additives on the grounds that they were found to be carcinogenic in test animals.

56 International Agency for Research on Cancer (IARC) In 1971, the International Agency for Research on Cancer (IARC), an agency of the World Bealth organization, began publication of a series of monographs on known and sus- pected carcinogens. These monographs are prepared by international groups of experts assembled by IARC, who critically review pertinent literature and draw conclu- sions regarding the carcinogenic~ty of various substances. The results of IARC reviews and evaluations are widely accepted. The guidelines used for evaluation by the IARC expert committees are set forth in the monographs. They = ~ a —~ are expressed in very general berms gnu "~= `~ "~ , ~ ~ only six components of hazard identification, completely covered in six pages of text. A major feature of the guidelines is the presentation of criteria that classify the evidence of suspected carcinogens as surf icient or limited. The IARC allows the expert committees consid- erable latitude to evaluate many inference options on a case-by-case basis, although the agency appears to insist on adherence to the few stated guidelines. Food and Drug Administration The 1958 Food Additives Amendment to the Food, Drug, and Cosmetics Act prohibited the use of Food additives found to be carcinogenic. The law was also interpreted as prohibiting MA approval of any drug, for use in animal s produced for human food, that had been shown to cause cancer. In 1962, by congressional amendment, EDA was permitted to approve the use of a carcinogenic animal drug ~f tne agency was convinced that no residue of a drug would be found in edible tissues of the treated animals. Congress specified that FOA was to prescribe the analytic methods for verifying the absence of residues. This directive proved to be unworkable, for . _ , two reasons: progress In analytic chemistry was so Haiku that approved methods of analysis quickly Became ODSo~e~e and improved detection methods showed that no drug admini- stered to animals is ever entirely absent from animal tissues. The problem of enforcing the 1962 amendment was highlighted in the early 1970s, when die~Ylsti~estrol residues were discovered in beef liver with highly sew sitive, but as yet unapproved, analytic methods. In an attempt to provide a consistent and predictable procedure for approving methods to search for drug resz-

~7 dues, FDA proposed sensitivity-of-method guidelines in the form of regulations (FDA, 1973, 1977, 1979b). Rather than gear criteria to an analytic technique, the agency def fined its standards in terms of r isk. It proposed that any assay approved for controlling a carcinogenic drug must be capable of measuring residues that present more than an insignif icant risk of cancer, and specif fed a 10-6 lifetime risk of cancer as a quantitative criter- ion of insignificance. If a drug sponsor could provide a detection method capable of measuring residues posing a risk of this magnitude or greater, FDA would ignore resi- dues that could not be detected with this method. Thus, FDA found guidelines for quantitative estimation of risk necessary. FDA's sensitivity-of-method guidelines are unique in several ways. They address a narrow though complex set of issues encountered in regulating a single class of products, animal drugs. Although they deal to ~ large extent with testing, they were the first to address quantitative risk assessment methods, listing assumptions for dose~response assessment, exposure assessment, and risk characterization. And they are the only guidelines that att^ - t to establish a definition of significant risk. The guidelines have yet to be adopted, a decade . after they were first proposed, but the agency has applied the methods of quantitative risk assessment embodied in the sensit~vity-of-method document in connection with the regulation not only of animal drugs, but also of food contaminants, such as aflatoxin (EDA, 1979a) and trace constituents of some additives (EDA, 1982~). The sensit~v~ty-of-method guidelines were proposed as regu- lations, as were the cancer guidelines of the Occupational Safety and Bealth Administration (OSElA). In both cases, regulation engendered substantial controversy. The major debate over the sensitivity-of-method guidelines has dealt not so much with risk assessment or the definition of significant risk as with the amount and cost of testing that FDA would require from industry before product approval. Environmental Protection Agencv During the early to middle 1970s, EPA initiated actions to prohibit or restrict the use of several pesticides. The agency lacked internal procedures for assessing carcinogenic risk and relied heavily on the judgment of scientists outside EPA. Attorneys for EPA, in summar-

~8 icing the testimony of their expert witnesses during administrative hearings on actions against the pestz- cides, set forth several statements that, in the legal brief, were referred to as cancer print (EPA, 1972, 1975). They conveyed the idea that the only acceptable degree of regulation would be a total ban on exposures. The principles, perceived as EPA's cancer policy, incurred wide criticism from the scientific community, the private sector, and Congress. The Practicability of achieving Cow rick On a broad scale for a large number of economi- cally important chemicals became increasingly apparent. In response to this new perception, and perhaps out of a desire to avoid misunderstanding of its cancer policy, the EPA became the first agency to adopt formal guidelines embracing a two-step process of risk assessment (EPA, 1976). The first step is a determination of whether a particular substance constitutes a cancer risk (hazard identification). The second step is a determination of what regulatory action, if any, should be taken to reduce the ~ 1stc . As part of the second step, the agency explicit itly endorses the use of quantitative risk assessment as the means of determining He magnitude of the likely impact of a potential human carcinogen on public health. These guidelines were not published as regulations and "niov fairly wide acceptance from most interested par- _ , ~ ties. As stated in the preface to the guidelines, they were published to Improve agency procedures, to provide public notice of the approach that EPA would take, and to stimulate commentary from all sources on that approach. The guidelines were probably more important as a state- ment of a novel approach to risk assessment than for their content. They are grate general, cover less than a page of Federal Resister text, and address only a few co~po- nents of hazard identification, dose-response assessment, exposure assessment, and risk characterization. More detailed guidelines that specify assumptions for the choice of extrapolation models, scaling factors, and other elements of dose~response assessment were published in 3980 by program offices in EPA (ERA, 1980~). These rely in part on the Interagency Regulatory Liaison Group (IRKING) guidelines {IBLG, 1979a) and are currently undergoing review. recall - Final s:~fetv and Health Admit In 1977r OSEA published guidelines in a proposed regular tion, ~Identification, Classification, and Regulation of

~9 Toxic Substances Posing a Potential Occupational Risk of Cancers (OSBA, 1977); after extensive administrative hearings, it published a final rule in 1980 (OSHA, 1980). The guidelines proved to be highly controversial, and the hearings were marked by vigorous debate on almost every component of risk assessment covered by the guidelines. The OSEA rule, written by agency staff, was a detailed scientific and regulatory document that took several hunt dred pages of Federal Register text and addressed almost every component of hazard identification. The final rule did not address exposure assessment and red ected the use of dose~response assessment for any regulatory purpose except priority-setting. The main purposes of the guide- lines. as stated in the preface, were to streamline the process of risk assessment, to speed up regulation, and to reduce the workload of agency staff. Another purpose was to foster continuity of approach, even in the face of changes of poll cy-makers. OSE1A stat f perceived that the case~by-case approach to risk assessment was long and t~e-consuming, because the same controversial questions had to be argued each time a chemical was under cons~d- eration for regulation. The agency believed Mat the generic approach to risk assessment would reduce debate on these questions; the controversial issues could be decided once, incorporated into guidelines, and applied to all chemicals. For reasons of efficiency, the guide- lines.were written in language that permitted little deviation from the judgments embodied in them. Because they were written as regulations, regulated parties were required to abide by them. The agency has not used the rule as a basis for any published scientific assessment of carcinogenic hazard. It was revised in 1981 (OSAKA, 1981) to accommodate the Supreme Court's ruling on ben- zene, which required that OLGA make a finding that the risk to workers in the absence of regulation was s~gnifi- cant and would be reduced by the proposed standard. But this change and additional amendments were recently with- drawn, and the entire policy is under reconsideration (OSEA, 1982). ~ , ~ ,= ~ Consumer Product Safetv Commission The Consumer Product Safety Commission (CPSC) proposed cancer guidelines dealing mainly with hazard identifica- t~on (CPSC, 1978). Ten components related to that step were addressed in several pages of Federal Resister text.

60 Some minor attention was Given to exposure assessment and dose-sesponse assessment, for przority-setting purposes only. The rationale for publishing the guidelines, as stated in the preface of that document, was to establish CPSC's general principles and to solicit comments on them, to assist the general public and the regulated industries in understanding standards that CPSC would apply and reg- ulatory actions that it was likely to take, and to set forth its approach to some issues that tended to recur i n each case. The guidelines had no regulatory status; they were a statement of selected inference options to which the agency would adhere. The CPSC guidelines were never used; they were challenged In court, and the court ruled that CPSC had promulgated them illegally inasmuch as they were adopted without an opportunity for public comment. Furthermore, at that tome CPSC had decided to rely on the guidelines of IRLG. Interagency Regulatory Liaison Grout The four agencies represented in IRLG undertook the task or developing guidelines to n ensure that the regulatory agencies evaluate carcinogenic risk consistently. n In 1979, after an 18-month interagency effort, FROG pub- l~shed a report, unscientific Bases for Identification of Potential Carcinogens and Estimation of Risk.. The report was prepared by personnel of CPSC, EPA, For, and OSHA, with the assistance of senior scientists from the National Cancer Institute and the Nationa} Institute of Environs mental Health Sciences. It was published in a scientific journal (IRLG, 1979~) and in the Federal Register (IRLG, 1979a); IRVING requested public comment in the Federal Register. The IRVING report was said to represent an inter- agency consensus on Me scientific aspects of carcinogenic risk assessment.* It was the most comprehensive set of guidelines that had been developed for agency use, addres- sing most components of hazard identification and dose- ~esponse assessment with some general discussion of *Because rule~making was under way in connection with its cancer policy, OSLO declined to participate in the IRKING notice and comment procedure. After the report was completed, the Food Safety Quality Service of the O.S. Department of Agriculture joined IRKING and participated in the notice and comment.

61 osure assessment and risk characterization; it had, however, no official legal status. The report was note- worthy, in that it constituted the first evidence that all the federal regulatory agencies agreed on the infer- ence options applicable to the identification of carcin- ogenic hazards and measurement of risks. The document made clear, however, that not all the agencies were bound to conduct quantitative assessments; it stated only that, if such assessments were to be conducted, they would be conducted uniformly. This language was probably a cone cession to OSBA's view, as expressed in its cancer policy, that quantitative risk assessment was to play no more than a priority-setting role in that agency's regulatory activities. Almost immediately after its publication, the IRLG report was adopted by the Pres~dent's Regulatory Council and incorporated as the scientific basis of the Council's government-wide statement on regulation of chemical carcinogens. The Council viewed the IRLG guide- lines as a major step in reducing inconsistency, dupli- cation of effort, and lack of coordination among agencies in carcinogenic risk assessment (Regulatory Council, 1979). The scientific aspects of the final OSBA cancer policy, which was written to allow less latitude in the choice of inference options, were, nevertheless, za general agrees ment with the I=G guidelines. CPSC and EPA stated that they relied on the IN'G document, but the degree to which they rely on it today is not clear. FDA }was made no statement other than that associated with Me document's initial publication; in fact, in a recent proposal con- cerning the application of risk asses~'ent to a class of trace constituents of additives, F ~ die not even cite the IRLG document as a reference (FDA, 1982b). Although IRLG received a great deal of public comment on the guide lines, no report of the agencies' review of these comments has appeared. In fact, the document was heavily critz- cized by industry, because it was published In its final form and adopted before the c~ents could be analyzed and revisions incorporated. The Reagan Administration has off icially disbanded the entire 7BLG effort, so it Is unlikely that review of the public comments will ever occur. Although the IRLG charter was not renewed, a similar group has been established, but one that is coordinated by the White Bouse Office of Science and Technology Policy. This group has prepared a draft document on the scientific basis of risk assessment and has distributed

62 it for comment (OSTP, 1982) . The group anti cipates that this document may serve as a reference point for later development of general guidelines for the agencies. VARIATION IN 'em; FORM OF GUIDELINES CO~RE9=SIVENE:SS . Guidelines developed by agencies in the past have varied in ache extent to which they have addressed each of the steps of risk assessment. lances guidelines address only hazard identification; OSTIA's guidelines (1980) dealt mainly win hazard identification, with some ct~scusszon of dose~response assessment and none of exposure assess' ment and risk characterization; and IBI`G's guidelines focused in detail on hazard identification and dose- response assessment, with some discussion of exposure assessment and risk characterization. Guidelines also have varied in the extent to which they have addressed the components of the risk assessment steps. ~AI='s guidelines address a small number of components. Study of the latest IARC monograph {1982) reveals only six selected options that deal with inference of risk: treatment of benign versus malignant turners, the choice of statistical methods for application of data from an, ma1 studies, the relevance of negative results of epi- demiologic studies, the evaluation of tumors that occur spontaneously, the utility of shor~term tests, and the overall weighting of evidence. The OSLO {1980) and IRLG documents, in contrast, each discussed and embraced over _ 20 selected options dealing with hazard zuentzz~ca=~. EXACT OF DETAIL Guidelines have differed not only in their comprehensive- ness, but also in the detail with which they have treated specific components of risk assessment. When the content of a guideline is detailed, the assessor is presented with more complete information than would be available from a more general guideline. For example, the statement In IARG's guidelines on benign tumors is general, compared with that in the IRLG guidelines. IARC concludes briefly: If a substance is found to induce only benign tumours in experimental animals, it should be

~3 suspected of being a carcinogen and requires further investigation. The IKLG document made a similar statement, but in addition elaborated on several issues relevant to the evaluation of benign tumors that are not mentioned by lARC--e.g., evaluation of tumor incidence when both benign and malignant tumors are present; a listing of tumor types commonly observed as benign in test animals, but known to progress to frank malignant stages; evaluation of the quality of a histologic examination that might be prep sensed as evidence: and an illustrative example of the . dependence of response on the genetic characteristics of the rest animal. The additional material could have been used by an assessor, particularly one not familiar with the newest information on benign tumors, to ensure that a more thorough analysis of the relevant issues had been performed. FLEXIBILITY Detail can often be confused with inflexibility, and it is important to make a distinction between these charac-- teristics. Certainly, detailed guidelines can be inf}ex- ible if the detail is designed to limit agency discretion, and thus public debate, on an issue that is subject to multiple sczentif ic interpretations . However, detailed guidelines can have quite a different effect if their intent is to provide an assessor with background informa- tion that describes the complexity of an issue, with nuances that may influence particular judgments, or with examples of cases that are legitimate exceptions to the general rule. As descried in Chapter I, almost all components of risk assessment theoretically Trace one or more infer- ence options. For exe - le. in determining which doses response curve to choose, the biologically plausible inference options may include the linear, multistage, sublinear, and threshold models. A guideline usllall y prefers one option, although some guidelines permit the selection of more than one or of all the options. The preferred inference opt ion may be viewed as a default option, i.e., Be option chosen on the basis of risk assessment policy that appears to be the best choice in the absence of data to the contrary. A guideline may be said to be flexible according to the degree to which it

64 allows the default option to be superseded by another inference option as a result of convincing scientific evidence. * Comparison of I=G's guidelines with Oh's guidelines illustrates how comprehensive and detailed guidelines have varied in flexibility. On the issue of benign versus malignant t'',nors, IRLG' s guideline stated: The induction of benign neoplasms would, therefore, h" Considered evidence of carcinogenic activity unless definitive evidence is provided that the test chemical is incapable of inducing malignant neoplasm. The guideline did not attempt to define the type of defier itive evidence that would be needed to demonstrate that a latest chemical is incapable of inducing malignant new plasma.. In contrast, OSLO created strict minimal criteria for acceptance of such evidence: (i) Benign tumors. Results based on the ~nduc- tion of benign or malignant tumors, or both, will be used to establish a qualitative inference of carcinogenic hazard to workers. Arguments that substances that induce benign tars do not present a carcinogenic risk to workers will be considered only if evidence that meets the criteria set forth in 1990.144(e) is provided. Section 1990.144(e) stated: (e) Benign tumors. The Secretary will consider evidence that the substance subject to Me rules making proceeding is capable only of inducing benign tenors in human or experimental animals provided that the evidence for the specific substance meets the following criteria: Criteria. (i) Data are available from at }east two well-conducted bioassays in each of two species of mammals (or from equivalent evidence in more than two species). *Flexibility is also intimately related to the legal weight that the agency desires a guideline to have; the implications for flexibility of adopting guidelines under different legal authorities are reviewed In the next section.

65 (ii) Each of the bioassays to be considered has been conducted for the full lifetime of the experimental animals. (ii`) The relevant tissue slides are made avail- able to OSLO of its designee and the diagnoses of the tumors as benign are made by at least one qualified pathologist who has personally examined each of the slides and who provides specific diagnostic criteria and descriptions; and (iv) All of the induced tumors must be shown to belong to a type which is known not to progress to malignancy or to be at a benign stage when observed. In the latter case, data must be presented to show that multiple sections of the affected organ(s) were adequately examined to search for invasion of the tumor cells into adjacent tissue, and that multiple sections of other organs were adequately examined to search for tumor metastases. By leaving open the type of evidence needed to supe~- sede the default option (benign tumors should be consid- ered evidence of carcinogenic activity), FROG allowed more flexibility than OSLO. In no case did the IRLG guidelines attempt to restrict the type of evidence that would be needed for acceptance Of alternative interpretations. In contrast, OSLO speci- fied minimal criteria for acceptance of alternative interpretations on the issues of negative epidemiologic studies, proof of metabolic differences between animals and humans, and rejection of the use of data from testing at high doses. By invoking such criteria, OSLO attempted to limit the definition of acceptable interpretation and, in so doing, eliminate or reduce scientific debate on controversial issues in its rule-mak~ng proceedings. IRLG also created flexibility by not choosing a default option, i.e., by citing a range of possible inference options to be used in a risk assessment. The statement on ~nterspecies conversion factors illustrates this point: Several species-conversion factors should be con- sidered In estimating risk levels for humans from data obtained in another species. All OSHA guidelines were restricted to the choice of a single inference option.

66 DEGREE TO WErIC}] G13IDELI~:S I} ON AN AGI:~CY ANr A P<EG=ATED PARTY .. . . The guidelines developed by or for regulatory agencies may vary in their legal status and thus in their procedural Implications. For example, OSEA's guidelines (1980) appeared as regulations formally published, after oppor- tunity for public comment, in the Federal Resister. }n contrast, EPA's guidelines (1976), although eventually printed in the Federal Resister, have never been offici- ally subjected to public comment and do not purport to be regulations. To appreciate the practical differences among the approaches that an agency might follow, it is useful to distinguish three types of administrative documents: regulations (or, synonymously, rules), established pro- ceaures (a term we have devised to refer to agency pro- nouncements that are in some contexts referred to as guidelines), and recommendations. There is no single authoritative definition of the latter two types of document. The discussion here is an attempt to reflect common understanding; it draws as well on the practice, but not the terminology of one agency, FDA (1982a).* An illustration will illuminate the practical differences among these three types of documents. Suppose that an agency decides to adopt, as one of its risk assessment guidelines, the default option that benign tumors should be aggregated with malignant tumors in determining whether a mammalian bioassay demonstrates that an agent causes cancer in the test species. This guideline could be adopted as a regulation, as what we term an established procedure, or simply as a recommendation. For internal purposes, it is not likely to matter which form the *FDA officially recognizes three types of documents: binding regulations, guidelines, and recommendations . That terminology is potentially confusing here, because we have Liven guidelines a special meaning, connoting codified principles for rise assessment, anal Aye from FDA's legal definition. The reader is referred to the footnote at the beginning of this chapter for a more complete treatment of this discrepancy. We have there- fore coined the substitute phrase established procedures, to describe any standards of criteria for fulfilling a regulatory requirement that the agency commits itself to follow until they are formally revoked or revised.

67 agency' s guideli ne takes. If the guideline is understood to represent prevailing agency policy, the agency's mana- gers can assume that assessors will adhere to it in eval- uating test data, regardless of its form. Important differences will be observed, however, In the guideline' s impact on interested third parties. I~ the guideline were adopted as a regulation, it would be reciprocally binding. Neither the agency nor any pri- vate party would be free to argue in a regulatory proceed- ing that benign and malignant tutors should never be aggregated or should not in a particular instance be aggregated; the agency's regulation would render such arguments legally irrelevant. It Is precisely this effect of regulations--i.e., their treatment of previously cone tested (and in theory still contestable) issues as author- itatively resolved--that OSEA sought when it published its risk assessment guidelines as regulations. If the guidelines were merely a recommendation, manu- facturers of chemicals under evaluation would not be bound by it. They could argue, to the agency or in court, that _ _ ~~ a .: __.__— _ _ ~ benign tumors should never he aggregates WllU l~`y6~- tumors or that they should not be aggregated in a particu~ lar case. They might not convince the agency, but the agency could not lawfully refuse to consider their argu- _ ~ ~ :__ ~~ -_A ~ has? mint meets or reject e~rz~ence supporting cnem, ~~" "~ ~^y~- convince a court that the agency guide~ine~-i~e., its choice of inference options--~s wrong generally or inapt placable in a particular case. If the guideline were an established agency procedure, a private party could similarly argue that it is wrong generally or inapplicable to a particular case. An established procedure does not, therefore, preclude efforts by third parties to treat the benign-versus- malignant issue as an open question. The difference between a recommendation and an established procedure lies in the latter's effect on the agency itself. An agency can depart from a recommendation at any time. Under FDA's practicer however, it may not depart from an established procedure unless it has previously announced that it no longer regards the procedure as sound. an other words, such an established procedure is binding on the agency until formally revoked or changed, and third parties can rely on it and insist that the agency adhere to it.* *The practical effects of the legal distinctions drawn here are possibly overdrawn. The flexibility accorded by

68 There is another important difference between regula- tzons and established procedures or, indeed, rec~mmendar tions. To adopt regulations that have the reciprocally binding effects described above, an agency must follow the procedures prescribed by the Administrative Procedure Act, or by its own statute, for rule-making. At a mini- mum, these procedures include publication of a proposal, an opportunity for the submission of public comments, and promulgation of a Final document that discusses and responds to all significant comments. The process can be long and acrimonious, and that helps to explain why agent cies sometimes choose not to adopt policies, particularly those addressing complex issues, in the form of regula- tions. The same process must be followed to effect changes in regulations once adopted, and that inhibits rapid response to changes in scientific understanding. AR<;UME:NTS FOR AND AGAINST - I: USE OF G=D=I=S The advantages and disadvantages listed below constitute an inventory of arguments that have been brought forward by the proponents and critics of guidelines for risk assessment. In most cases, an argument Is most convincing for guidelines of a particular form and content, rather than for guidelines in general. For these cases, the characteristics of guidelines that would support or refute an argument are indicated. - any set of guidelines depends as much on the las~guage chosen as on the legal form in which they appear. Suppose that an agency's default option is: Ordinarily benign and malignant tumors shall be equated and their sum used to determine the significance of observed effects, unless (a) new data suggest the inappropriateness of this peace tice generally, or (3~) results from the test in question or other tests of the compound make aggregation inappro~ priate in the particular case. ~ This text anticipates exceptions, and would not prevent either the agency or a third party from taking a different view about the meaning of a particular test, whether i t appeared as a regulation or in some other form.

69 ADVANTAGES OF GUIDELINE USE Separation of Risk Assessment from Risk Manaqeseent Proponents of guidelines argue that their use would help to separate risk assessment from other parts of the Begun latory process. They contend that, when selected infer- ence options are clearly delineated in a formal document, risk assessments will not likely be influenced to f' t prior conclusions about regulation of a particular subs stance. The use of guidelines can also dispel the appear- ance of such inf luence when, in f act, there is none . Agencies can defend their assessments on the grounds that they always do them in the way set forth in the guides 1, nest Compared with reliance on Me ad hoc selection of inference options, the use of guidelines could reduce the controversy focused on individual assessments. Debate will shift to the more general discussion of the generic choice of inference options addressed in the guidelines. Guidelines that are c`~rehens~ve and derailed will define and bracket the components of risk assessment most come pletely and explicitly. Thus, such guidelines could probably provide the sharpest distinction between risk assessment and risk management. Quality Control Proponents of guidelines argue that their use would ensure the application of selected inference options based on the informed judgment of experts. A single risk assessment requires knowledge in diverse fields, such as epidemiol- ogy, hiostatist~cs, toxicology, biochemistry, chemistry, and clinical medicine. Generally, assessors have advanced expertise in no more than a few fields. Guidelines could help to bridge gaps in knowledge by ensuring that deci- sions are based on judgments formulated by experts in each subject. Guidelines could also help to ensure that assessors apply judgments that are In accord with current scientific thinking in each field. This argument high- lights the importance of including experts from a wide range of scientific disciplines in the formulation of guidelines. Furthermore, it suggests that guidelines should be reviewed periodically so that new scientific developments can be accommodated. Proponents believe that comprehensive, detailed guides lines would be most helpful in providing guidance to

70 assessors. Comprehensiveness is necessary to provide guidance on all or most of the components of risk assess- ment. Detailed guidelines could provide an assessor with an expert's insight into aspects of risk assessment that require special consideration. Bow flexibility could affect quality control is not clear; however, a flexible framework could have a positive effect, especially if guidelines can help an assessor to know when exceptional or novel scientific evidence should be admitted. Consistency Almost all guideline documents have stated, in their introductions, that consistency is a major rationale for guideline use. Consistency in risk assessment is ~mpor- tant to the agencies, because it helps to ensure fairness and rationality by precluding the arbitrary application of selected inference options that differ from one time to the next. Consistency also permits comparison of risks associated with different chemicals, and this is useful for priority~setting and for facilitating regulatory decision-making. When the same set of guidelines is applied uniformly by all the agencies, government-wide consistency may be ~proved. This has Important impli- cations for interagency coordination and for reducing the possibility that risk assessments by different agencies will be pitted against each other during litigation on a given chemical. Guidelines of a type that fosters consis- tency among agencies have yet to be adopted and used. In the absence of such guidelines, there are increased oppor- tunities for inconsistency in the choice of inference options available for each risk assessment component and in the conclusions based on those choices. Proponents of guidelines contend it is often difficult even to know whether there is consistency among risk assessments, because of lack of explicit documentation of inference options used. Comprehensive, detailed guidelines applied uniformly across the agencies appear to be ache most suitable form for reducing inconsistency. To ensure thoroughness and clarity in drawing conclusions, assessors should eyelid itly document the use of such guidelines in their reports. Flexibility does not imply inconsistency in the applica- tion of risk assessment policy. The same inference options can be applied consistently, except in instance s where convincing contrary scientif ic evidence is prey

71 sensed. When such evidence is available, the choice of different inference options has a scientific rationale and does not imply an arbitrary shift in risk assessment policy. It is not the same kind of inconsistency as that which can occur when, for example, one assessor uses a speciestcrhuman conversion factor based on surface-area ratios and another, for no better scientific reason, uses a factor based on body-weight ratios. P~edictabiliLv Proponents of guidelines argue that the private sector should be told expl~cz~ly which inference options the government will select to evaluate healthreffects data. Industry needs this information to assess its own activ- ities and testing programs. Without uniformly applied guidelines, a regulated party may have to call on the agencies for judgments on numerous issues and have no assurance that the judgments will not change unexpectedly or that one agency's judgment will be consistent with another ' s. }ndustry representatives have stated their preference for uniform federal guidelines (although they have been much more cautious In discussing the content of and legal weight given to the guidelines. Consider, for example, the following comment by the American Industrial Health Council, regarding the publication of the IRLG cancer guidelines (AIEC, 1979): The report is a significant step toward the formulation of a national cancer policy. AIEC supports the report's stated objective of ensuring that regulatory agencies evaluate carcinogenic risks consistently. We strongly urge that this initial step be followed up so that a national cancer policy is developed and conflicting policies among the regulatory agencies are minimized. This point of view takes on added significance in view of the increasing desire of some states to develop their own cancer policies. Six states have initiated programs thus far, and California has already published its own guide- lines (State of California, 1982a, b). For the private sector to have to contend with a range of different poli- cies in different states would clearly be disadvantageous and burdensome. A federal cancer policy could serve as a model to the states and foster a more uniform approach to risk assessment.

72 Proponents believe that the most useful guidelines in gauging government actions would be detailed and compre- hensive. Although flexibility may undermine predictabil- z~y, it is reasonable to assume that industry would welcome such a trade-off. Guidelines published as estab- lished procedures would be the best option, for the regulatory agencies would not change their procedures without formal notice, but the procedures would not be binding on the regulated parties. Evolutionarv Improvement of the Risk Assessment Process Proponents of guidelines argue that their use provides a locus for debate, examination, and revision of the selected inference options generally used in risk assess meet. By contrast, the argument proceeds, when chemicals are evaluated on an ad hoc basis, the focus of debate is shifted from generic issues to case-specific issues, and the methods and assumptions of risk assessment are obscured from critical view. Over the last decade, new and refined techniques of risk assessment have emerged. Two important examples are the use of short-te~m in vitro tests to infer carcino- genicity and mutagenicity and the use of dose-response assessment to estimate the magnitude of human risk at low doses. Guidelines may have contributed to the evolution of both by proposing generic interpretations that would be evaluated and tested both in theory and in the lab- oratory. The choice of a low~dose extrapolation model is a specific example. The first guidelines (FDA, 1973) proposed the use of the Mantel-Bryan model. This choice was the subject of much debate (~DA, 1977, 19?9b); newer guidelines have suggested that this model has been dis- counted by tne agencies, an part because it is essentially empirical and lacks biologic relevance with respect to current knowledge about carcinogenesis (GREG, 19795; EPA, 1980a). Furthermore, the debate over an appropriate model helped to foster a major research effort. m e EDt1 experiment, also known as the ~megamouse study,. involved the testing of 24,000 female mice given known carcinogens at low doses in an attempt to determine the shape of the response curve at low doses. Guidelines that are comprehensive and detailed would invite the most opportunity for debate and evolutionary refinement.

73 Public Understanding Because risk assessment is complex, it is easy to parody and demean the process. For example, the decision to label soft drinks containing saccharin was satirized in several highly publicized jokes, e.g., Caution: Sac- char~n is hazardous to your ratn and Shrink 800 bottles of pop a day and get cancer." Proponents of guidelines argue that comprehensive, detailed guidelines setting forth the scientific and policy bases of risk assessment could improve public understanding and help to dispel the Impression that government actions are based on tenuous and inadequate reasoning. Administrative Efficiency Some contend that when risk assessments are performed on a chemical-by-chemical basis without the use of guide- lines, too many agency resources are devoted to reargu- ment of the same issues with regulated parties. For example: should animal carcinogenicity data be used to assess human risk? Should data on animals with a high incidence of spontaneous tumors be considered valid? Should benign tumors be assigned the same weight as malignant ones? Which statistical methods should be applied? Guidelines could reduce repetitious discussion by specifying which types of interpretations are acceptable, given the current state of scientific understanding. OSLO, in its reidentification, Classification, and Regulation of Potential Carcinogens. (1980), registered concern about its efficiency Only seven rule~making proceedings completed in 9 years) and cited one major reason for its low productivity: The necessity to resolve basic scientific policy issues anew, in each rulemalcing, has increased the burden on the Department of Labor and members of the scientific community called upon to address these widely accepted policies. Moreover, rel~ti- gation of these issues in the federal courts has also drained staff tome and energy and has inhib- ited OSLO initiatives while its policy determina- tions were repeatedly relitigated.

74 OSHA maintained that the adoption of cancer guidelines was vital to efficient regulation: OSEA believes that this general policy and process cure will facilitate Me sifting through the evidence concerning substances which may be interpreted to be potential carcinogens. . . Without such a system and appropriate criteria, OSHA believes that this task cannot be accom~ plished in a tamely And efficient manner. Efficiency could best be served by guidelines that are comprehensive, detailed, and inflexible and are adopted as regulations binding on all parties, but this would entail other costs. The disadvantages of such guidelines are described in some of the agruments cited in the following discussion. DISADVANTAGES OF GUIDELINE USE Oversimplification The adoption of guidelines may foster a cookbook approach to risk assessment. The more assessors look at chemicals from a generic point of view, the less they are able to draw distinctions among them on We basis of specific data. The critics' ultimate concern is that blind adher- ence to guidelines might cause scientific information relevant to a particular chemical to be arbitrarily cast aside because it has not been accommodated in the guidelines. The following underlined phrases are examples of guides lines that critics believe may lead to oversimplification: . ream ad: she mans ':-n~itive species to determine risk. Critics contend that, if information shows that metabolic similarity to humans is greater for a species that is less sensitive, data on this species may be preferable. · Absence of a threshold for carci~o2:~gi~. Critics argue that tumors may be induced by a genetic mechanism or by an epigenetic mech~"imu. In the latter case, a threshold may exist. · Unoualif fed acceptance of posits ve results at

75 hiqh-dose testing. Critics believe that validity should depend on whether there is a pharmacokinetic difference between high and low dose. Special consideration should be given to whether detoxifying or repair processes are saturated and to whether competing metabolic pathways are involved and become saturated. Another potential problem is the lack of attention to weighting of evidence. For example, a guideline may simply state that appositive results in animal tests should always outweigh negative results. This does not take into account the quality and statistical power of the different tests; i- -~1a Fonder the anti tude that such considerations are of minor Importance. To a large extent, the strength of such criticisms depends on the form and contents of the guidelines. Those which are comprehensive and leave little latitude for exceptional cases tend to maximize the problem of oversi~pl~f ication; those which are f lexible could be most effective in mitigating the problem. In addition, guidelines may explicitly direct the assessor to consider the weight of evidence of a given test result. For example. the IRLG guideline stated that positive results ~ _ . _ = a _ = ~ = _ should supersede negative results, out aauea a caveat ; n If the positive result is itself not fully conclusive or if reasons exist for c~estzon~ng Its validity as evidence of carcinogenicity, Me result is generally classified as 'inconclusive' or 'only suggestive' even in the absence of other negative results. ~ Detailed guidelines can reduce the possibility of over- simnlif ~ cation if the intent of detail Is to capture for the assessor the cc$pplex~ty of the issue addressee. For example, a guideline might state the scientific basis for the chosen inference option, Me kinds of evidence that are typically applicable, circumstances in which accept tance of exceptional evidence may be appropriate, and other rationales for choosing a particular inference option. Regardless of Me form of a guideline, there are some parts of risk assessment, particularly those dealing with the quality of data and the magnitude of uncertainty, that defy or at least resist generic interpretation. Individ- ual judgment Is most important in such cases. A guide line should not be viewed as a formula for producing risk assessments without the need for such judgment. /

76 Mixing of Scientific Knowledge and Risk Assessment Policy Gu Sidelines unavoidably embody both science if ic knowledge and risk assessment policy. In the past, regulatory agencies typically used a conservative approach in the development of risk assessment policy, as in the choice of the most sensitive species, use of the most conserva~ tive dose~response curve, and ache lack of acceptance of negative epidemiologic data. Industry has been highly critical of this approach. Some representatives believe that risk assessment should be solely a scientific func- tion and should be separated from policy decisions. Consider, for example, the American Industrial Bealth Councilts criticism of the IRVING guidelines (ATTIC, 1980): When the IREG report speaks of the importance of using conservative methods or assumptions so as not to underestimate human risk, the report is mixing regulatory considerations into the so lent tific function. The scientific determination should be made separately from the regulatory determinations. On the basis of the best sczenr tific estimate of the real risk, the regulatory agency can then consider costs, benefits and other elements that enter into a regulatory determination. Furthermore, there is a fear that the mixing will go unrecognized outside the scientific community (AIEC, 1980): When value judgments are formalized by the selec- tion, for ~conservative. reasons, of a mathematical model or an assumption used for extrapolating human risk, the fact that value judgments have been made escapes the regulator and the public. The first criticism appears to miss the crucial fact that risk assessment must always include policy, as well as The Important issues are what the risk assess- ~nent policy content is and whether it will be applied consistently or not. The second criticism Is most appli- cable to guidelines that permit an agency to represent as science the conclusions that have been reached in part on the basis of policy considerations. The argument is less applicable to guidelines that explicitly distinguish between scientific knowledge and risk assessment policy

77 and direct the assessor to address such distinctions when reaching conclusions. Furthermore, it is not clear that risk assessment performed on an ad hoc basis would reduce the opportunity for unrecognized mixing of science and policy; indeed, carefully designed guidelines could help to inhibit such mixing. Guidelines very different from the kinds described could be designed to be devoid of Disk assessment policy choices. They would state the scientifically plausible inference options for each risk assessment component without attempting to select or even suggest a preferred inference option. Bowever, a risk assessment based on such guidelines (containing all the plausible options for perhaps 40 components) could result in such a wide range of risk estimates that the analysis would not be useful to a regulator or to the public. Furthermore, regulators could reach conclusions based on the ad hoc exercise of risk assessment policy decisions. Misallocation of Agency Resources to Development and Amendment of Guidelines Critics contend that the dedication of time and resources to the process of guideline development and amendment detracts from an agency's ability to conduct regulatory activities. For example, OS~A's cancer guidelines required 3 years of effort before promulgation of the final rule In January 1980. The full rule~making record eventually exceeded 250,000 pages. OSEA itself offered some 45 witnesses who addressed the scientific content and the policy implications of the proposal, and a much larger number of witnesses appeared in behalf of other participants. The final policy consisted of more than 280 Federal Resister pages of preamble and a dozen pages of regulatory text. Notwithstanding this intensive effort, the guidelines have yet to be applied, and new leadership at OSHA is in the process of reevaluating some provisions of the standard. The procedures required by the Administrative Proce- dure Act are so elaborate that development and amendment of guidelines written as regulations are expected to demand more intensive effort than guidelines written as established procedures or recommendations. Regardless of the legal status given to the guidelines, their stability over time is susceptible to major changes in policy stances. However, guidelines that clearly distinguish

78 scientific knowledge from risk assessment policy judg- ments could provide a locus for facilitating changes in policy orientation. They would define elements of risk assessment policy that are amenable to change and scienr tific elements that should not be changed for policy reasons. When risk assessment is done on an ad hoc basis, such distinctions may not exist. Freezing of Science Critics believe that guidelines would hinder the timely incorporation of important new scientific evidence during standard-setting. The Dow Chemical Company raised Ones concern about OSEA's cancer guidelines (°SBAr 1980): The record . . . has now made it clear that there ~ s absolutely no assurance that the latest sc' em tific evidence in the field will be permitted to be applied under the proposal to any given regulation of a specific chemical substance. OSLO responded to this criticism by incorporate ng three amendment procedures into its cancer policy: a genera' review of the guidelines every 3 years by the directors of the National Cancer Institute, the National Institute of Environmental Health Sciences, and the National Instz- tute for Occupational Safety and Bealth; recommendations at any time from the National Cancer Institute, the National Institute of Environmental Bealth Sciences, or the National Institute for Occupational Safety and Health; and petitions from the public. Final amendments would occur only through formal, independent rule making, to ensure that major changes in the guidelines would not be made during the litigation of individual cases. In incus try's perception, ache amendment provision did not answer its initial criticism. The A3aeric~" Industrial Bealth Council characterized the amendment procedures as "a t~me-consuming and ponderous mechanism for incorporating into the regulatory standards newly available evidence or data concerning heretofore unresolved issues" (Osaka, 1980). This argument is most applicable to guidelines that are adopted as regulations and to those which are compre- hensive and inflexible. When guidelines are flexible and adopted as established procedures or recommendations, the rapid incorporation of novel scientific information is

79 more easily accommodated. The intent of flexibility is to allow the acceptance of exceptional evidence based on convincing scientif tic justif ication. In the case of established procedures or recommendations, changes in guidelines coul d occur without the necessity of a lengthy rule~making process. CONCLUSIONS . On the basis of its review of the historical record of guideline development and use and its evaluation of the arguments for and against guideline use, the Committee has drawn several conclusions. 1. All agencies have found it necessary to write guidelines, in part, to make their choice of inference options more evident to the public. However, the apuli- cation of inference options to specific risk assessments has been marked by a general lack of explicitness. Because of the lack of explicitness in identifying the choice of inference options in specific risk assessments, it has often been difficult to know whether assessors adhere to guidelines. Within a given program, a consis- tent set of selected inference options may emerge over t ~e. However, the degree of consistency Tong programs and agencies is not well defined. 2. Agency Guidelines have varied markedly in form and content. Without a deliberate coordinating effort, there is no reason to assume that Guidelines will became more nearly uniform. Although the scientific bases or cancer guidelines developed in the past by the agencies have been generally consistent, the degree to which the guidelines are compre- hensive, detailed, flexible, and legally binding has varied widely. EPA's guidelines are statements of broad principles covering a few components in the four steps of risk assessment; they have no regulatory status. OSEA's guidelines were comprehensive and detailed and dealt mainly with hazard identification; they were regulations. CPSC's guidelines were not comprehensive and dealt mainly with hazard identification; they had no regulatory status. FDA's proposed sensitivity~of-method guidelines are cow prehensive and detailed for dose-response assessment and exposure assessment; they are regulations. The formation of the IRLG caused the agencies to adopt a single set of

80 guidelines for the first tome, but, since its disbanding in 1981, there has been no Further progress on guideline development.* 3 Uniform Sidelines for risk assessment (except for ~ ~ ~ , exposure assessment) are Feasible and desir ~ le. Guidelines are feasible. The Committee believes that current statutory requirements would not prevent the use of uniform guidelines. Regulators administer laws reflec- ting social policies that suggest different degrees of acceptable risk. She argue that uniform guidelines would keep regulators from applying different standards of risk that were based on these laws. Bowever, regulators can apply such standards on the basis of risk management decisions after completion of the risk assessment. Fur- thermore, feasibility has already been demonstrated by the adoption of the IRLG guidelines. Uniform guidelines are desirable for several reasons. First, the use of different guidelines by the agencies could undermine the credibility of their risk assessments. Critics of an agency risk assessment might argue persuar sively that another agency estimates risk differently, on the basis of a different set of inference options. Second, almost every regulated chemical is in the juris- dict~on of two or more agencies, and the possibility of duplication of effort in performing risk asses-~ents on a given chemical could be minimized if the guidelines were applied uniformly. Adoption of uniform guidelines could foster joint risk assessment elms my "y=~- ^..~_ ested in regulating the same chemical; or one agency could rely on the assessment of another agency. Through such cooperative efforts, a small agency like CPSC:, which lacks the scientific capability~of EPA and F=, could gain help in evaluating complex data. Third, government- wide guidelines could help industry to gauge government actions asked to define the types of data and interpreta~ tions relevant to industries' own testing program. Fourth, federal policy could orchestrate efforts toward uniformity among the states. a_ w~ ~__; ~- ~—for— _ *The Office of Science and Technology Policy (OSTP), with agency participation, has written a document describing the scientific basis of risk assessment. OSTP envisions the ultimate evolution of a set of principles for risk assessment from this document.

81 E - osure guidelines, in contrast with guidelines for other risk assessment steps, are not now readily amenable to uniform application in the various agencies. Apart from EPA, the agencies have rather narrowly defined interests regarding exposure, i.e., foods and drugs at FOA, consumer products at CPSC, and occupational hazards at OSEA. Whereas guidelines for the identification of hazard and for the quantitative estimation of risk in test animals may be commonly applied, no such con basis exists for applying exposure assessment guidelines. 4 Even well-desi~ned guidelines mav be unsuccessful unless: . . Attention is Liven to the process by which they are developed. Thev can accommodate change. · They are viewed as valuable f Nebulas for Producing risk assessments. Because guidelines must include both scientific knowl- edge and policy judgments, designing a development pro- cedure is a cliff icult task. Risk assessment requires advanced knowledge in a number of disciplines, and guide- l~nes should be formulated in part on the basis of the best possible scientific expertness in those disciplines. The best mechanism for determining risk assessment policy must be carefully defined. Because of the necessity of considering policy aspects in guidelines, duly appointed public officials must take responsibility for the policy implications. A major goal of the development process should be the assurance that the guidelines preserve a sharp distinction between scientific knowledge and risk assessment policy. The Committee believes that guidelines should be capable of accommodating evolving scientific concepts in two ways. First, they should be periodically reviewed and, if necessary, revised. Second, they should permit acceptance of new evidence that differs from what was previously perceived as the general case, when scientifi- cally justifiable. Bowever, an unavoidable trade-off results from the use of such flexible guidelines: predictability and consistency may be reduced for the sake of flexibility. Every risk assessment involves consideration of case specific factors, such as the quality of the data or the overall strength of the evidence. These factors cannot

82 be addressed effectively in guidelines. If assessors were to use guidelines in a strictly mechanical fashion, without recognizing the importance of case-specific judgments, the quality of risk assessments could be diminished. 5. Uniform Guidelines for effects other than cancer are desirable, but typically they would be based on a less extensive scientific data base. The same reasons enunciated for the desirability of cancer guidelines Impel the conclusion that guidelines are needed to guide assessments of other effects. Scienr tific data available on these effects may be organized to provide useful information for assessing case. _ _ _1 In facts guidelines have already been developed for some of these (although never adopted by the agencies), i.e., guidelines for mutagenesis (EPA, 1980; March of Dimes Birth Defects Foundation, 1983~) and guidelines for reproductive and teratogen~c effects tORNL/EPA, 1982 ; March of Domes Birth Defects Foundation, 1981). ~E=NCES AIEC (American Industrial Bealth Council). October 11, 1979. ALEC comments on: A Report of the Interagency Regulatory Liaison Group (IRKING), Work Group on Risk Assessment, entitled Unscientific Basis for Identificar tion of Potential Carcinogens and Estimation of Risks,. p. 2. AIEC (American Industrial Bealth Council). April 3G, 1980. In Review of Public Comments on Statement on Regulation of Chemical Carcinogens; prepared for the O.S. Regulatory Council, Washington, D.C., pp. B-7, B-10. CPSC {Cons'` - er Product Safety Commission). 1978. Interim policy and procedure for classifying, evaluating, and regulating carcinogens in consumer products. Fed. Reg. 43:25658. EPA (Environmental Protection Agency). ~ . · ~ _..__:, ~ April 5, 1972. a- =-: ^= ; _ =~- of annex findings J conclusions and order at 63-64, in re: Stevens Indus., Inc. (consolidated DOT hearings). EPA (environmental Protection Agency). September 5, 1975. Respondents motion to determine whether or not the registrations of mirex should be canceled or amended; Attachment A. Rex ends =~ ^~e ~ && -—Ye—~— —— ~——=-

83 EPA (Environmental Protection Agency). 1976. Health risk and economic impact assessments of suspected carcinogens. Fed. Reg. 41:21402. EPA (Environmental Protection Agency). 1980a. Mutagen- ic~ty risk assessments; proposed guidelines. Fed. Reg. 45:74984. EPA Environmental Protection Agency). Office of Water Regulations and Standards. 1980b. Water quality criteria documents; availability. Fed. Reg. 45: 79350-79353 . EPA (Environmental Protection Agency). Office of Research and Development, Carcinogen Assessment Group. August 8, 1980c. The Carcinogen Assessment Group's method for determining the unit risk estimate for air pollutants. External review copy, prepared for Office of Air Quality Planning and Standards and Office of Air, Noise and Radiation. FDA (Food and Drug Administration). 1973. Chemical compounds in food-producing animals; criteria and procedures for evaluating assays for carcinogenic residues. Fed. Reg. 38:19226. FDA tFood and Drug Administration) . 1977 . Ch^mica 1 compounds in food-producing animals; criteria and procedures for evaluating assays for carcinogenic residues. Fed. Reg. 42:10412. FDA (Food and Drug Administration). 1979a. Assessment of Estimated Risk Resulting from Aflatoxins in Consumer Product Peanut Products and Other Contami- nants. Rock~ille, Md.: Food and Drug Administration. FDA (Food and Drug Administration). 1979b. Chemical compounds in fo~producing animals; criteria and procedures for evaluating assays for carcinogenic residues. Fed. Reg. 44:17070. FDA (Food and Drug Administration) . }982a. Code o Federal Regulations, Title 21, Section 10.90. Washington, D.C.: O.S. Goverrment Printing Office. FOA (Food and Drug Administration). }982b. D & C Green, No. 6, listing as a color additive in externally applied drugs and cosmetics. Fed. Reg. 47:14138. IRAC (International Agency for Research on Cancer). 1982. General principles for evaluating the carcinogenic risk of chemicals; in IAMB monographs on We Evaluation of the Carcinogenic Risk of Chemicals to Pans. IARC, Lyon, France, vol. 29. I=G (Interagency Regulatory Liaison Group), work Group on Risk Assessment. 1979a. Scientific bases for identification of potential carcinogens and estimation of risks. Fed. Reg . 44:39858.

84 IRI'G (Interagency Regulatory Liaison Group), Work Group on Risk Assessment. 19795. Scientific bases for identification of potential carcinogens and estimation of risks. J. Natl. Cancer Inst. 63: 242. Joint Food and Agricultural Organization and World Health Organization Expert Committee on Pesticide Residues. 1965. Evaluati on of the Toxicity of Pesticide Residues in Food: Report of the 2nd Joint Meeting. FAO Meet. Rep. No. PL/1965/10, t~IO/Food Add./26.65. Joint Food and Agricultural Organization and world Health Organization Expert Committee on Food Additives. 1972. Evaluation of Certain Food Additives and the Contaminants Mercury, Lead, and Cadmium. WHO Tech. Rep. Ser. SQ5. Geneva. Lehman, A. J., F. A. Vorhes, fir., L. L. Ramsey, E. C. Hagan, O. G. Fitzhugh, P. J. Schouboe, J. H. Draize, E . I . Goldenthal, W. ~ 'Aguanr~o, E. T. Lang, E . J. Omberger, J. B. Gass, R. E. Zwickey, E. J. Davis, ~ . A. Braun, A. A. Nelson, and B. 3. Vos. 1959. Appraisal of the Safety of Chemicals in Foods, Drugs and Cosmetics. Association Food and Drug Officials of the United States (AFOOUS). March of Dies Birth Defects Foundation. January 26-27, 1981. Guidelines for Studies of Bran Populations Exposed to Mutagenic and Reproductive Hazards; Proceedings of Conference, Washington, O.C. PK:/NAS (National Research CounciV~at~ona1 Academy of Sciences), Food Protection Committee, Food and Nutrition Board. 1970. Evaluating the Safety of Food Chemicals. Washington, D.C.: National Academy? of Sciences. OR~/EPA (Oak Ridge National Laboratory/Environmental Protection Agency). February 1982. Assessment of Risks to Human Reproduction and to Development of Me Euman Conceptus from Exposure to Environmental Substances. ORNL/EIS-197, EPA/9-82-001. .OSEA (Occupational Safety and Health Administration). 1977. Proposed rule: identification, classification and regulation of potential occupational carcinogens. Fed. Reg. 42:54148. O SEA (Occupational Safety and Bealth Administration). 1980. Final rule: identification, classification and regulation of potential occupat tonal carcinogens. Fed. Reg. 45:5001. O SEA (Occupational Safety and Bealth Administration). 1981. Identif ication, class'f icat' on arid regulation of potential occupational carcinogens; conforming deletions. Fed. Reg. 46: 4889 .

85 OSLO (Occupational Safety and Health A - inistratzon). 1982. Identif ication, classif ication and regulation of potential occupational carcinogens. Fed. Reg. 47:187. OSTP (Office of Science and Technology Policy), Regulatory Work Group on Science and Technology, Executive Office of the President. October 1, 1982. Potential Bum an Carcinogens: Methods for Identification and Chateau terization. Part 1: Current Views: Discussion Draft. Regulatory Council. 1979. Statement on regulation of chemical carcinogens: policy and retest for public comment. Fed. Reg. 44: 60038. State of California, Bealth and Welfare Agency. July 1982a. Carcinogen Identification Policy: A Statement of Science as a Basis of Policy. State of California, Bealth and Welfare Agency. October 1982b. Carcinogen Identification Policy: A Statement of Science as a Basis of Policy; Section 2: Methods for Estimating Cancer Risks from Exposures to Carcinogens.

Next: 3 Organizational Arrangements for Risk Assessment »
Risk Assessment in the Federal Government: Managing the Process Get This Book
×
Buy Paperback | $49.95
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The regulation of potentially hazardous substances has become a controversial issue. This volume evaluates past efforts to develop and use risk assessment guidelines, reviews the experience of regulatory agencies with different administrative arrangements for risk assessment, and evaluates various proposals to modify procedures. The book's conclusions and recommendations can be applied across the entire field of environmental health.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!