National Academies Press: OpenBook

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

Chapter: 3 Organizational Arrangements for Risk Assessment

« Previous: 2 Inference Guidelines for Risk Assessment
Suggested Citation:"3 Organizational Arrangements 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 86
Suggested Citation:"3 Organizational Arrangements 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 87
Suggested Citation:"3 Organizational Arrangements 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 88
Suggested Citation:"3 Organizational Arrangements 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 89
Suggested Citation:"3 Organizational Arrangements 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 90
Suggested Citation:"3 Organizational Arrangements 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 91
Suggested Citation:"3 Organizational Arrangements 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 92
Suggested Citation:"3 Organizational Arrangements 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 93
Suggested Citation:"3 Organizational Arrangements 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 94
Suggested Citation:"3 Organizational Arrangements 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 95
Suggested Citation:"3 Organizational Arrangements 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 96
Suggested Citation:"3 Organizational Arrangements 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 97
Suggested Citation:"3 Organizational Arrangements 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 98
Suggested Citation:"3 Organizational Arrangements 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 99
Suggested Citation:"3 Organizational Arrangements 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 100
Suggested Citation:"3 Organizational Arrangements 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 101
Suggested Citation:"3 Organizational Arrangements 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 102
Suggested Citation:"3 Organizational Arrangements 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 103
Suggested Citation:"3 Organizational Arrangements 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 104
Suggested Citation:"3 Organizational Arrangements 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 105
Suggested Citation:"3 Organizational Arrangements 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 106
Suggested Citation:"3 Organizational Arrangements 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 107
Suggested Citation:"3 Organizational Arrangements 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 108
Suggested Citation:"3 Organizational Arrangements 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 109
Suggested Citation:"3 Organizational Arrangements 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 110
Suggested Citation:"3 Organizational Arrangements 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 111
Suggested Citation:"3 Organizational Arrangements 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 112
Suggested Citation:"3 Organizational Arrangements 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 113
Suggested Citation:"3 Organizational Arrangements 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 114
Suggested Citation:"3 Organizational Arrangements 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 115
Suggested Citation:"3 Organizational Arrangements 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 116
Suggested Citation:"3 Organizational Arrangements 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 117
Suggested Citation:"3 Organizational Arrangements 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 118
Suggested Citation:"3 Organizational Arrangements 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 119
Suggested Citation:"3 Organizational Arrangements 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 120
Suggested Citation:"3 Organizational Arrangements 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 121
Suggested Citation:"3 Organizational Arrangements 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 122
Suggested Citation:"3 Organizational Arrangements 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 123
Suggested Citation:"3 Organizational Arrangements 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 124
Suggested Citation:"3 Organizational Arrangements 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 125
Suggested Citation:"3 Organizational Arrangements 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 126
Suggested Citation:"3 Organizational Arrangements 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 127
Suggested Citation:"3 Organizational Arrangements 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 128
Suggested Citation:"3 Organizational Arrangements 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 129
Suggested Citation:"3 Organizational Arrangements 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 130
Suggested Citation:"3 Organizational Arrangements 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 131
Suggested Citation:"3 Organizational Arrangements 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 132
Suggested Citation:"3 Organizational Arrangements 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 133
Suggested Citation:"3 Organizational Arrangements 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 134
Suggested Citation:"3 Organizational Arrangements 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 135
Suggested Citation:"3 Organizational Arrangements 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 136
Suggested Citation:"3 Organizational Arrangements 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 137
Suggested Citation:"3 Organizational Arrangements 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 138
Suggested Citation:"3 Organizational Arrangements 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 139
Suggested Citation:"3 Organizational Arrangements 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 140
Suggested Citation:"3 Organizational Arrangements 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 141
Suggested Citation:"3 Organizational Arrangements 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 142
Suggested Citation:"3 Organizational Arrangements 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 143
Suggested Citation:"3 Organizational Arrangements 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 144
Suggested Citation:"3 Organizational Arrangements 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 145
Suggested Citation:"3 Organizational Arrangements 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 146
Suggested Citation:"3 Organizational Arrangements 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 147
Suggested Citation:"3 Organizational Arrangements 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 148
Suggested Citation:"3 Organizational Arrangements 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 149

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.

III Organizalional Arrangements for Risk Assessment The different structures, procedures, and histories of the agencies responsible for regulating toxic substances have produced diversity in their approaches to risk assessment, but common patterns can be discerned, and they permit some broad generalizations about agency organizational arrangements. first, most agencies have exerted little effort to maintain a sharp organizational separation between employees engaged in assessing the risks associated with substances and those responsible for identifying and evaluating regulatory responses. This is not to suggest that the same persons perform both functions; generally, they do not, for agency organizations reflect consider- able specialization, recognizing the distinctive training and capabilities of staff members. However, the two func- tions are often housed in one organizational unit that is responsible for preparing integrated analyses that incor- porate assessments both of risk and of recommended regule- tory responses. Sometimes, risk assessment stat f are employed in an office that is separate from the office of those who formulate and analyze regulatory options, but, with she notable exceptions, this organizational struck ture does not lead to a rigid separation of the two staffs. Second, with the exception of a few experiments in interagency r, sk assessment during the late 1970s and continuing informal exchanges of information, each agency has performed its own assessments of the risks posed by substances that are candidates for regulation. This operational autonomy does not reflect willful ignorance of the activities of sister agencies or indifference to the desirability of consistency in the evaluation of common candidate substances. Rather, it is a product of 86

87 several factors, including the lack of obvious mechanisms for formalized interagency collaboration, the desire of agency po}'cy-makers to reserve authority for policy dis- cretion in reaching conclusions based on risk assessment, the perception that the diversity of types of exposure for which each agency is responsible makes collaborative r isk assessment impractical, and differences in regu~a- tory priorities and schedules. Third, although the four agencies have viewed them-- sei ves as ultimately response b, e for the risk assessments that support their actions, they often extend their own staff resources available for performing risk assessment by relying on consultants and contractors who are closely supervised by agency personnel. Some agencies--notably the Occupational Safety and Health Administration (OSEA) and the National Institute for Occupational Safety and Health (NIOSE) - whose staffs are small or lack needed expertise rely very heavily on nongovernment contractors and outside scientists in the academic community and government research institutions for performance of risk assessments or specific tasks related to risk assessment (such as literature reviews). In addition, outside scientists are often called on to review assessments produced by agency staff. Such consul- tations sometimes take place informally, but of ten through special advisory committees. These committees can be per- manent, such as the Environmental Protection Agency (EPA) Clean Air Science Advisory Committee, or can be created to review particular risk assessments, as is done for many of the Food and Drug Administration (FDA) Bureau of Foods assessments. Some are established by statute, with requirements that they review agency assessments befor e regulations are proposed. Others are created voluntarily by an agency itself. The members of all federal advisory co~t~ees are appointed by the agencies, perhaps with the approval of higher executive-branch authority. Can- d~dates for committee membership usually are identified by agency staff, although some agencies seek nominees from professional organizations and other interested parties. Nominations for some statutorily mandated committees are supplied by an external body, such as the National Academy of Sciences or the National Science Foundation. Advisory panels generally exercise considerable influence and, although legally they are only advisory, share to some extent the agencies' authority to reach conclusions about risk.

88 o to A: A: a: o - N A: as o o ~2 o a: Sit o o Q. a' X IY A: I m ~ ~2 m Y ID ~ v - 3 IV (L) _ tL C) U) ~ P4 o - ~ V X AS ~ o 0 AS U] A: o A: V ~ _ AS 04 - ~) :^ O In ~ _ An ~ ~S ~: P. U) 0 3 I . U) o a ~S] S~ 3 o o V u} ~: 0 c m ~2 tQ ~: s S~ ~ o ~ ~5 t) Q) ~ :~: $5 ~ ~ := Q O U) o ~: O :- ~ ~ O ~ O U] ~ ~ _ :> ~ V 6 0 AS O O {U .! 2 V 3 0 f: ~ ~4 ~ ~5 Q as O ~ ~ ~ ~¢ s`3 m s V ~ Q, S~ c: a' ~ ~ ~ m 0 O c: ~: n ~ ~s AS ~ C: 3 ~ O ~ — :> Z o ~2 ~Q ~S ~ ~ O U~ P ~: O O ~ z ~ ~1 I _ X O O ~ S p~ C: 41) O ~ ~S O Q) c: ~ O AS O C) P' li4 m ~s o o ~U o ~S S~ 3 CS) Q o S2 _ m o2 o2 ~2 c: V 0 as a'

89 TYPES OF ORGANTZATIO~ AWNS The prominent proposals for reforms in organizational structures and procedures for risk assessment have featured three interrel ated principles: . Risk assessment activities should be strictly separated from the analysis of risk management options and selection of regulatory strategies. · Risk assessment act ivities should be centralized in a single body that serves all regulatory agencies. · Expert panels composed of nonagency scientists should be used either to perform risk assessments for an agency or to review assessments developed by agency staff. The Committee outlined four idealized models that reflect various combinations of these three principles. The models are integration, intra-agency separation (with or without centralization), extra-agency separation (with or without centralization), and use of scientific review panels. Examples of agency organizations that roughly approximate each model are identified below and in Table III-I. Most of the exiles chosen have many distinctive characteristics that obscure or at least outweigh the three organizational principles. In addition, they are not the only examples of a particular model; others could have been reviewed.* INT~:GRATION In this type of arrangement, a single organizational unit both performs risk assessments and develops regulations. In general, this arrangement is the most common for Begun latory programs. For example, for the assessment of chron' c hazards involved with chemicals f rom consumer products, the Consumer Product Safety Commission (cpsc) *The Committee considered the possible merits of reviewing risk assessment procedures used by other countries as well and decided not to pursue this line of investigation, because of the great differences in political and institu- tional structures between this country and other coun tries. Such differences would make it very difficult, if not impossible, to extrapolate findings on institutional structures used in other countries to the United States.

so Directorate for Health Sciences is the responsible unit. Before 1977, the Directorate for Health Sciences had few people involved in the risk assessment process, and risk assessments as such were not generally used. Since then, the Directorate has acquired the expertise needed to per- f orm r isk assessments itself . The r isle assessment Is performed within the Directorate, which is distinct from the Commission's politically appointed policy decisions makers. Two different examples of this model examined by the Committee are the OSBA Directorate of Health Standards Programs and the FDA Bureau of Foods (Table III-1). In the former example, risk assessors and those responsible for formulating and recommmending regulatory strategies are in the same organizational unit. FDA's Bureau of Foods has a separate office that performs risk assessment, but this separation stems from a functional division of scientific disciplines; it is not intended to and does not result In formal separation.of the risk assessment staff from the regulatory staff. INTRA-AGENCY SEPARATION In this model, risk assessment is performed by a group that is ostensibly separate from and independent of the office responsible for regulation in the same agency. An into e-agency risk assessment unit could be programs specific or agency-wide. There are examples of program specific, organizationally separate risk assessment units (notably the Environmental Criteria and Assessment Offices In EPA), but the Committee did not examine them; instead, it reviewed activities of the EPA Carcinogen Assessment Group as an example of an internally separate, agency- wide body. EXTRA-AGENCY SEPARATION l In this model, an agency's risk assessment Is developed outside the agency. The examples reviewed demonstrate the wide variety of arrangements included in this model. Full organizational separation can be achieved by having one institution perform risk assessment and a separate institution regulate exposure to hazardous substances. The relation between NIOSE and OSLO was studied as an example of a permanent, statutory arrangement of this kind. A regulatory agency's use of expert panels to

91 perform risk assessments can also result in extra-agency separation of risk assessment and regulation. Committees of the National Research Council and several groups of panels used by FDA to review the safety and effectiveness of drugs provide varied examples of such arrangements. The National Toxicology Program Panel on Formaldehyde is an example of an ad hoc assessment group that consisted of government scientists, was organizationally separate from the regulatory agencies (although not without agency members), and served all four agencies (i.e., it was centralized). Because the Interagency Regulatory Liaison Group did not perform r isk assessments, it has not been examined as an example of an extra-agency assessment body. USE OF SCIENTIFIC REVIEW PANELS . Agencies may use independent scientific panels to perform risk assessments or to review assessments prepared by the agencies. This distinction has been used by the Committee to facilitate separate discussion of panels that perform assessments as examples of full organizational separation (see preceding discussion) and panels that review agency assessments as examples of independent review panels. However, the dichotomy is somewhat artificial, in that there may be difficulty in classifying a particular panel. For example, if a panel responsible for performing risk assessments comes to rely heavily on preliminary analyses prepared by agency staff, it can be thought of as acting in a review capacity. Conversely, panels assembled solely for the purpose of reviewing agency assessments have often displayed remarkable independence, sometimes preparing long critiques of agency documents and suggesting sub- stitute Findings and reasons. In such cases, to specify which group had performed and which had reviewed the agency's final assessment of risk is difficult. The extent to which agencies have used independent scientific panels has varied considerably. For example, OSKA has available two types of advisory committees: standing bodies, such as the National Advisory Committee on Occupational Safety and Health, and ad hoc committees that provide advice on specific standards. Members of both types of committee are expected to be knowledgeable about occupational safety and health and may include persons mainly interested in law or regulatory policies. In addition to their professional expertise, however, members of OSHA committees are intended to be represen-

92 tative of groups interested in occupational health and safety. Several committees have reviewed risk assess- ments prepared by OSLO or NIOSE. Bowever, because men bers were intended to be representatives of interest groups, reviews were usually forums for policy debates, not scientific evaluations of risk assessments. In its initial years, OSHA routinely appointed an advisory committee for each regulatory proceeding. CPSC has had the least experience with expert panels. Before 1981, the Commission was not required to have any assessment of carcinogenic hazard reviewed by an outside panel, although it did make occasional use of such panels (most notably CPSC's request for the National Toxicology Program to form a panel on formaldehyde?. CPSC's ~eautho~ rization in 1961 included a pro:. A ~~---~ regulatory action could be proposed on a substance poten tially presenting a carcinogenic, teratogenic, or muta- genic hazard, a chronic hazard advisory panel (CHAP) must be established, with the cooperation of the National Academy of Sciences, to review the toxicity of the sub- stance. The first CHAP has recently been convened to review the toxicity of asbestos. Thus, CPSC relies on two methods of peer review for any proposed action. First, independent peer review by outside experts, as well as by a scientific review panel, is performed before - Second, the Commission relies on a public rule-mak~ng proceeding in accordance with the Administrative Procedure Act during which comment is invited through a Federal Resister notice on all aspects of the proposed action. Extensive written comments have been received in the past by this procedure, from industry, consumer groups, mothers of the academic ~ __ _ __—~^ ~ nab_ O—TON— t~IlV a notice of proposed rule-mak~ng as issues. and scientific communities, and others. Add~tzona~'y, open, informal public hearings may be held in which inter- ested groups present their views orally; in the past, several such hearings were held during the consideration of a single substance (formaldehyde). EDA has often used independent scientific panels both -^ perform and to review agency assessments. The Bureau of Drugs has used standing committees to review and eval- - ~ or——~—~ uate data on the safety and effectiveness of drug products and to make appropriate recommendations to the Co"~ls- sioner (see preceding discussion). The use of indecent dent panels by the Bureau of Foods, however, has been on an ad hoc basis, usually at the agency's discretion. However, there are exceptions; for example, the Food, Drug, and Cosmetic Act requires that carcinogenicity

93 issues related to color additives be referred to a come settee of experts selected by the National Academy of Sciences. EPA, in contrast, has had less choice in its relations with its advisory committees. Several statutes require I: PA to consult such committees for scientific review of agency risk assessments or regulations. Examples of mandated advisory committees with a primarily scientific role include the Agency-wide Science Advisory Board; the Clean Air Scientific Advisory Committee, a part of this Board, which reviews criteria documents for air-qual~ty standards; and the Scientific Advisory Panel, which focuses on scientific issues in the Agency Office of Pesticide Programs. The Committee has examined this pane' and a subcommittee of the Science Advisory Board as examples of scientific review panels. Agency actions, including risk assessments, have been reviewed in the Executive Office of the President; how- ever, because these reviews have, with a few notable exceptions, focused primarily on risk management con- cerns, the Committee has not examined them. LEVIS OF Al PROS FOR USE AS=SS~ This section describes the practices used for risk assessment in each of the organizational examples reviewed by the Committee. The descriptions that follow reveal some strengths and weaknesses of particular approaches and permit some tentative generalizations to be made. Such generalizations, augmented by the experi- ence and judgment of Committee members, lead in turn to recommendations applicable to organizational arrangements for the performance of risk assessment. The Committee's necessarily retrospective review of agency performance has focused on events and practices of the 1970s, which triggered the current proposals for reform. Changes have been implemented, or at least are contemplated, in the procedures of several of the agen- cies studied, and the Committee recognized that such changes could alter the performance of risk assessment. Some of the descriptions of agency practices presented here may be dated. However, our purpose is not to des- cribe the current organizational structure of agencies, but rather to discern in the historical record any gent eral relationships between organizational design and procedures and the quality of risk assessments. The

94 paucity of experience with recent organizational changes and the tendency of any new administration to disclaim the approaches of predecessors while proclaiming the effectiveness of reforms make very recent history less germane to the Committee Is purpose. OSLO ~ S DI=C=~= OF ~= STEMS PR=~ (DESP) OSEA's health standards were expected by Congress to be based on criteria and recommended standards provided by NIOSE. Bowever, improvements in OSLO s scientific copay bility and a court directive that OSHA itself review all studies included in the risk assessment supporting a proposed standard prompted Me agency to rely less heavily on NIOSE and to begin performing its own risk assessments. Until 1976, OSHA had only a few personnel in the health sciences; however, DESP has since become an organization staffed primarily by health professionals, including industrialhygienists, responsible for performing risk assessments and for preparing standards, relying on economic and technical analyses supplied by the Office of Regulatory Analysis In a separate directorate (Figure III-1). In addition, the Directorate normally has used a number of consultants who assist with risk assessment or other aspects of standard development, contributing considerable specialized expertise to the organization. OSHA tried to achieve organizational separation of risk assessment from the preparation of standards in the case of carcinogens. One office in DSSP was supposed to do risk assessment, another to draft standards. In prac- t~ce, however, such separation was not achieved, largely because personnel shortages required that individual staff mothers perform both functions. Agenda and Procedures DH5P's regulatory and risk assessment agenda has been determined largely by two external forces: petitions by labor unions for action on particular hazards and dr='n=tic discoveries of previously unidentified workplace hazards. Court remands of several OSLO standards, such as the bend zene standard, provided new work for Oslo, but none of the mandated re~examinations has led to a final standard. Criteria documents prepared by NIOSE also contributed to O=A's agenda, in Mat DESP staff always read these docks

95 - - ~ ~ Ic - - o o = -~ - ~ = ~ oar a _ e O ~ ~ = _ - O O ~ = ~ ~ O ._ _ _ ~ O == O ~ = ~ ~ ~ - ~ SIi ~ = = ~ ~ O ._ _ _ ~ 0~= _ .- ;~ ~] C < S ~ ~= !~ ~ ~ _ of HE _~= o _ ~ = ~ 0= == 1 _ [ _ _ L 0 = . ~ o ~ 0 = o ~ o o o c o 1

96 meets when they were received and normally published a Federal Resister notice soliciting further information. DESP's risk assessments usually began with a NIOSH criteria document or other NIOSE input, whatever informal tion was submitted with a labor petition if there was one, the data available from any precipitating discovery, and assessments performed by others, such as the National Academy of Sciences. A literature search and review were conducted by DHSP personnel, often with the help of con- sultants and NIOSE personnel; and sometimes environmental data on the workplace were solicited or obtained by contractors to contribute to the exposure assessment. DESP has not prepared special assessment documents before issuing notices of proposed rule making. Thus, the first indication provided to the public of the results of an OSHA risk assessment and of the conclusions it intended to draw therefrom was In the Federal Resister preamble to its proposed standard. Public comment was invited on all aspects of the proposed standard, includ- ing the risk assessment. Extensive written c ~ rents were usually received from industry, labor, and others, such as members of the academic scientific community. Cus- tomarily a hearing was held at which oral presentations were made and at which questioning of witnesses by OSlIA personnel and other witnesses was permitted. The preamble to the final rule, if one were issued, included OSEA's f inal risk assessment, which incorporated a literature review and 05EA's conclusions on the available scientific data. In 10 years, OSLO produced permanent health standards for 23 substances or processes, 14 of which were regu- lated together In a single rule-making. OKRA has also proposed standards for eight substances for which final standards have never been issued, and assessments were conducted for several substances for which new or updated standards are now being considered (Table III-2). Methods and Use of Guidelines For most of its history, OSHA has not had formal guides lines for carcinogenic risk assessment. Instead, agency staff have conducted their assessments by choosing options for the components of risk assessment on a case-by-case basis. However, the generic guidelines for identified tion and classification of carcinogens proposed in 1977 and revised and promulgated in 1980 were intended to

97 TABLE 171-2 A Summary or OSHA Standards Standards Standards Standards Proposed, But Being Completed Not Completed Developed Asbestos Vi nyl chlor ~ de Ar Venice Benzene Coke oven emission 14 catch nogens Lead Cotton dust 1 ,2-I)ibromo-3- chloropropane Acrylon~tr ile Arsenic Beryllium Sulfur dioxide Ketones Bearing con- servation (noise) Toluene Ammonia MOCA Tr ichloro~ ethylene Ethylene oxide Asbestos Ethylene dibromide Cotton dust, nontextile sector s ache arsenic standard was remanded to OS" by the Court of Appeals for the Ninth Circuit for purposes of making a significant-risk determination consistent with the Supreme Court's benzene decision. replace criteria used in individual cases with generic guidelines that would be applied consistently to all risk assessments of potential carcinogens. The choices ~ncor- porated in the 1980 cancer policy reflected the policy orientations of incumbent senior agency officials. Changes now cont-—lated in these guidelines reflec t the policy orientation of the current OSEA management. Sow ilarly, although for many years OSEA did not perform quantitative risk estimates for use in setting standards for carcinogens, it now intends to do so where appropri- ate. This change results from policy decisions of senior agency officials, based, at least in part, on their inter- pretation of the Supreme Court's decision on benzene. (Agency officials have interpreted the decision to mean that quantitative dose~response assessments should be

98 performed for individual substances if data are sufficient.) Peer Review l OSHA historically has done a less thorough job than other agencies in obtaining relevant scientif ~ c information and independent peer review of this information before ;=c,~i~n ~ novice 09 Dronosed rule-making. y c. A___ -- =~~r~ Instead/ the agency has relied primarily on the public rule making proceeding to identify new information, much of which i s ;~ the r`^ccmc~ion of interested parties and is unlikely A. -~— .c~ at ~— _~__~ to be brought forward except ~ n the context of rules making. Similarly, although NIOSH's and OSHA's initial assessments often did not provide a critical review of relevant data, critiques of this intormac~on were yet. to the agency during rule-making proceedings, and the agency's final assessment of the risks posed by a chem~- cal often was substantially changed as a result. OS~A's use of rule-making proceedings to provide scientific review stands in sharp contrast with the other agencies' procedures for review. In the Co~mittee's opinion, this Olin an nn~lic Proceedings to strengthen and refine I.—_ _~ & ~~ ^_ =~ it J— —— ___ A _ _ _ the scientific basis for the agency's regulatory actions has not been an adequate substitute for independent peer review. In addition, reliance on public proceedings surely precipitated some of the criticism or agency actions and may have jeopardized the scientific integrity and procedural legitimacy of the agency's risk assess- ments. Although OS"'s standard-setting actions have stony - lated intense controversy, much of it has focused on Issues separate from risk assessment. Questions of costs and technologic feasibility (risk management issues) have stimulated much debate. Discussions of the agency's risk assessments have usually focused on its conclusions and their relationship to the agency's regulatory mandate, rather than on its characterization of risk. When OSLO' s risk assessments were challenged during rule-mak~ng, some key subjects of contention were OS~A's adherence to the assumption that carcinogens have no threshold for causing adverse effects, its tendency to give positive data greater weight than negative data, its use of single ep~- demiologic studies to support regulatory action, the validity of specific experiments and the agency's inter- pretation of the data from them, and the decision as to

99 whether quantitative assessments of r isk should be con- sidered. These issues, of course, have both policy and scientif ic implications. FDA ' S BUREAU OF FOODS The Food and Doug Administration enforces the Federal Food, Drug, and Cosmetic Act and several related statutes. Its jurisdiction ranges from basic foods to the most advanced pharmaceuticals and medical equipment. The agency assesses the risks associated with thousands of new and existing products every year, functioning through product-oriented units whose responsibilities are reflec- ted in their titles: Foods, Drugs and Biologics, Veteri- nary Medicine, and Devices and Radiological Bealth (Figure III-2) . The bureaus ~ agendas are dictated both through internal planning and by external events, particularly applications for approval of new products. Because the _~ ¢ ~ Bureau of Foods has had considerable experience with products that pose potential cancer risks, the Committee has focused on this part of FDA in its review. Agenda and Procedures The Bureau's risk assessment functions fall into three broad categories: review of petitions for marketing of new compounds for which the manufacturer provides sups porting toxicol ogic and exposure (or use) data; planned retrospective or cyclic review of approved compounds, supporting data on which the Bureau generally must take as it finds them; and review of inadvertent contaminants in food, supporting data on which are derived from many sources, including open scientific literature, monographs, reports, manufacturers' data, and agency-generated data. In 1981, the Bureau of Foods evaluated 65 food addi- tives, two color additives, and approximately 45 an~mal- drug petitions. These totals, however, do not reveal the total number of Bureau inquiries that could qualify as risk assessments, albeit perfunctory. Each tome a new contaminant is discovered, for example, the Bureau per- forms some assessment of the risks, although the avail- able data are often limited and little tome Is available to gather data before it must decide whether to initiate control measures. Similarly, every reported change in

Too o c' c) - <^ - .>o~o I' to o L - so == r r CC C) Cat ~ .. _ ~ ~ U' - . An O O = ~ X C, O C: C, n' 3 ~ O ~ O ec — O ~ ' ; O tC . OC' 0~ o Sat S V o - N I: A Sat to 1 Sol lo: V

101 degree of contamination invites a new risk assessment. As one would predict, the tome and effort required vary with the context. The Bureau's procedures for reviewing food additives, color additives, and residues of animal drugs are more routine than those for evaluating food contaminants, whose occurrence is largely unpredictable. On receipt in the Division of Food and Color Additives, a ~ood-additive petition is evaluated to determine whether it is acceptable for filing. This involves not only review of its formal adequacy, but a preliminary assess- ment of the toxicologic data to determine whether all potential health effects have been studied. After official filing of the petition, scientists from the appropriate divisions (ordinarily with the assistance of scientists outside the agency) study the supporting chemical, toxicologic, and exposure data to decide whether the compound is safe. The food-additive law has been construed as requiring, even when the Delaney clause is not applicable, Treasonable certainty that no consumer will be harmed. No effort is made to evaluate the benefit fits that an additive might provide, but the Bureau must be satisf fed that the additive achieves its intended effects. This exercise usually has two parts: first, Division of Toxicology scientists determine a no-observed- effect concentration for the additive on the basis of acute, subchronic, and chronic feeding studies in angels; second, applying a so-called safety factor, they determine a permissible extent of use in human food or an acceptable daily intake. . - This value is then compared with the esti- mated daily human exposure based on the manufacturer's proposed use and predicted hewn consumption of the foods in which the additive is to be used. An acceptable expc- sure to an additive is one at which human exposure is at or below the acceptable daily intake. . . . Under current law, Ants intake value cannot be established for a direct food or color additive that is carcinogenic; such a substance may not be approved for use. m e risk assessment function is performed entirely by Bureau scientists. Bureau staff, including the reviewing scientists, may meet with representatives of the petz- tioner to discuss uncertainties, request additional data, or suggest reduced use. Typically, both the scientif ic and the regulatory aspects of food-add~tive petitions are reviewed and resolved at the division level in the Bureau of Foods. On petitions that raise difficult scientific and policy issues or that pose the question of nicity, the divisions generally seek advice or carcinogen direction

102 from the associate directors, Bureau deputy directors, or the Bureau Director. The Bureau may, in turn, seek advice from the Chief Counsel, from other bureaus, or from the Commissioner's office during the review of petitions that present particular scientific, legal, or policy questions. Once the responsible unit is satisfied that an additive is approvable and thus that a regulation is appropriate, the Division of Food and Color Additives prepares a docu- ment package consisting of an action memorandum, ~ draft Federal Resister document, and supporting material, which is then forwarded through established review channels to the Director's office for final Bureau approval and trans- misszon to the Commiss~oner's office. The action memoran- d~m recommending approval by the Associate Commissioner for Regulatory Affairs, to whom the Commissioner has dele- gated formal approval authority, necessarily incorporates both scientific assessments and regulatory judgments. Because the governing legal standard focuses exclusively on the health effects of the additive, the approval pror ~ _ cess is not influenced by consideration or economic o over benefits. The sequence of analysis in the Bureau for environ: mental contaminants does not differ sharply from that described above for food additives, although different divisions may participate in the process and economic factors are consciously considered. The statutory pro- vision under which FOA regulates food contaminants content plates that it will balance the risk posed by a substance against the effects of reducing consumer exposure, such as loss of food and increases in price. Accordingly, the action memorandum sent to the Bureau Director recommends an exposure 1-;~t based on three criteria: an assessment of the risk posed by the contaminant, an evaluation of available me~ch='s of chemical analysis to monitor its presence, and ~ estimate of Me economic effects of alternative limits. Methods and Use of Guidelines Although the Bureau's approach to the evaluation of acute toxicants has remained stable over a long period, its methods for evaluating potential carcinogens have under- gone substantial change since the early 1970s. In 1918, the Bureau Director formed a Cancer Assessment Committee in the Office of Toxicological Sciences to evaluate the carcinogen~city of substances being considered for

103 approval or regulation and to perform risk assessments. A list of substances reviewed by this Committee in 1981 is given in Table III-3. The 12 members of the Committee are all FDA employees and include toxicologists, pathol- ogists, mathematicians, and chemists. The role of the Committee is to render al1 final decisions on casino genicity for the Bureau of Foods on the basis of scier~- tif~c information available to it. Its primary function is to determine whether, on the basis of a fair evalua- tion of all available data, a chemical is a potential or actual carcinogen. Because the Delaney clause, which forbids exposure of any food or color additive that induces cancer, applies to many substances in the Bureau' s jurisdiction, quantitative (e.g., dose~response) assess- ments are not always performed. For some substances, such as contaminants, the magnitude of the risk is relevant, and scientists from the various divisions collaborate with staff responsible for gathering information on human en sure to perform risk characterizations. The Cancer Assessment Committee does not typically prepare formal written assessments, so there Is no document available that outlines the relevant data and the rationale for the choices of options made in the assessment of risks. The Cancer Assessment Committee apparently does not follow comprehensive written guidelines, although it does follow some general guidelines that were used in previous deci- sions and are set out in the agency's drug-residue proposal. Peer Review . In recent years, the Bureau of Foods has sought indepenr dent scientific review of the data on a number of sub- stances. Often Bureau staff informally solicit the judg- ments of individual outside scientists on major issues. me Bureau routinely uses outside panels established under the auspices of ache Federation of American Societies _ for Experimental Biology nor perloalc review on suck now generally recognized as safe (GRAS). Ad hoc panels were convened to evaluate the data on such substances as cyclamate, saccharin, Red No. 2, and Red No. 40. More recently, the Bureau has turned to a standing panel, the Board of Scientific Counselors of the National Toxicology Program. The Board's review of the data on color additive Green No. ~ illustrates the Bureau's approach to external peer review. The Board reviewed the

104 TABLE lII-3 Substances Evaluated for Carc inogenic Sty by the FOA Cancer Assessment Committee in 1981 Acry}onitr ile Lead acetate Vinyl chloride D ioxane =Toluidine Bydrazine 1,2-Dichloroethane Diethylhexylphthalate D' ethylbexylad ipat e Furazolidone Cinna~,l anthranilate Tr i methylphosphate original data from a study done by ~ commercial labora- tory, which were submitted with a petition for approval of the substance. The Board also considered aspects of the analysis done by Bureau staff and conducted an inde- pendent evaluation of the pathology slides and a statis- tical analysis of the study results. Bureau scientists asked that the Board reach a conclusion concerning the strength of the evidence of carcinogenicity. Thus, the Board was limited to scientific issues and did not cone sider the possible social implications of its finding. After the Board's finding that the evidence was incon- clus~ve and before the Bureau's conclusion that the additive was unlikely to be a human carcinogen, Bureau staff performed a risk characterization to estimate the potential risks if this conclusion were in error. The decision to consult an outside panel for review of risk assessments for potential carcinogens is made by the Chairman of the Cancer Assessment Committee. The Bureau currently is considering establishing a standing commit- tee that could be called on to review agency assessments. It is likely that the impetus to form a standing review committee stern from criticisms of past agency practices, especially those followed in the evaluation of the data on nitrite. In this instance, FDA's contemplated action against nitrite in 1979 was announced before Bureau scien- tists had had an opportunity to evaluate the critical toxicity data and to refer the data to an independent panel. This controversial chapter In FDA's history of regulating food ingredients has often been cited as demonstrating the need for systematic peer review of the agency's risk analyses in order to avoid the problems that can arise when risk management considerations affect the conduct of risk assessments. The existence of a standing panel, although no guarantee, may discourage

105 agency officials from deviating from standard Bureau procedures that are now designed to ensure adequate peer review. EPA ' S CARCINOGEN ASSESSMENT GP~13P - EPA's Carcinogen Assessment Group (CAG) was created ~n 1976 by the EPA Administrator to implement generic and uniform agency guidelines for carcinogenic risk assess- ment. Initially, it was a separate body in the Agency's Office of Research and Development and reported directly to its Assistant Administrator. In 1979, however, the Office of Health and Environmental Assessment was estab- ~ished in the Office of Research and Development, and CAG became one of several assessment groups (Figure III-3). Organizationally, Go<; staff are separate from, and index pendent of, the risk management function; i.e., it is an example of intra-agency separation. It also serves as an example of an internally centralized assessment body, in that it performs assessments for several different regula- tory programs in EPA. Although CAG personnel do meet and talk with regular tory program personnel and are customarily well aware of any programmatic interest in particular substances and of interest-group preferences, this office is insulated from the day-today pressures of program offices. Thus, the organizational arrangement that places CAG In the Office of Research and Development does have the initial effect of freeing risk assessment personnel from specific policy issues that arise when risk management options are consid- ered. Bowever, when a scientific review committee exam ines documents produced by this office later in -he pro- cess, interest groups are able to express their views and CAG personnel are no longer isolated from such influences. Currently, all CAG assessments are done by in-house staff, although in the past some were done by consultants. Usually, contractors are employed only for the t~me- consuming and mechanical task of conducting literature searches. Responsibility for each assessment is assigned to a particular person, but other staff members contribute to various sections according to their particular special- ties and expertness. Its staff has been remarkably stable; since 1976, only one person has left the group. As of October 1982, 11 full-t~me professionals were on its staff, nine of whom had doctorates. Most staff memo bers have an academic background, and their professional work experience averages 10 years. The staff includes

106 . E _ _ C3 l l _ ~o C~ _ _ ~ o c, _ C) G — _ ~ _ _ Ci E _ C} C C S - ~ {D _ o _ _ o _ ~ - E C o - o _ _ _ ~ _ ._ _ C~ ~ x U. 1 l PC--_ -L~ ·0S3 1~ sO 1 _~:'Ec L° C o C) _ _ o c, C C) = _ ~ ~ C: _ _ a'' o o E t-' ._ _, ~ C, o ~ ~ _ ~ 6 C' S: o - N - ~: o 1 z

107 three b~ostatisticians, two biochemists, two ep~demiolo- gists, one biophysicist, one pathologist, one pharmacolo- gist, and one endocrinologist. The fonmer Director, now a consultant, is the only physician associated with the office. Agenda and Procedures CAG does not initiate its own assessments; instead, it responds to requests from regulatory (progress offices in EPA. It does, however, set its own priorities in consul- tation with the program offices, on the basis of the workload of requests and the urgency of the need for the assessments. Although it serves as a risk assessment body for the whole Agency, not all programs in EPA use CAG. The most notable exception is the Office of Toxic Substances. Apparently, one factor cited by program offices as leading to this lack of use is the length of tome CAG requires to complete an assessment. Since 1976, QG has prepared assessments for approxi- mately 150 chemicals. The length and scope of the docu- ments produced vary with the data available, with their purpose, and with the needs of the requesting office. They can range from brief and preliminary literature reviews relevant to hazard identification or tentative estimates of risk as a function of dose to complete and thorough literature reviews leading to a comprehensive risk characterization. In-depth evaluations may or may not include quantitative dose~response assessments. As an example of its work agenda, CAG has covered 41 chemi- cals for the Agency's Office of Air Quality Planning and Standards. Inrdepth evaluations were performed for nine (see Table III-4), and preliminary assessments for 32. Methods and Use of Guidelines The risk assess ents performed by this group are based on Agency guidelines developed initially by QG in 1976 for use by the entire Agency. These guidelines have been revised after initial publication, and some of ache changes have also been published (EPA, 1979, 1980). Normally, individual assessment documents produced do not reexamine or indeed articulate underlying guidelines; rather, the reader Is presumed to know That EPA and CAG rely on guidelines that embody particular choices among several

108 TABLE II-4 Substances Fully Evaluated by the Carcinogen Assessment Group for the EPA Office of Air Quality Planning and Standards Arsenic Benzene Vinyl chlor Ode Acrylonitr ills Coke~oven emissions Ponder review as of October 1982. Methyl chloroform Methylene chlor idea Tetrachloroethylene~ Tr ichloroethylene~ irreverence options available. also, the changes made in the guidelines have not, in many cases, been formally acknowledged; i.e., the current guidelines do not exist in a single publicly accessible written document. CAG's use of guidelines, especially for hazard identif ication, has been regarded by some EPA review panels- - notably, the Subcommittee on Airborne Carcinogens--as too inflexible, possibly misleading, and interfering with critical analy- sis of underlying data. In Fact, bee initial published guidelines {EPA, 1976) did permit different interpreter tions of data and the use of different risk assessment methods; however, the methods embodied in CAG assessments and those related to dose~response assessment and pup fished in EPA's Water Qualitv Methodology for Carcinogens do not ref lect this f legibility . The misunderstandings experienced with the Subcommittee on Airborne Carcinogens (and other review bodies) have stemmed to a great degree from the facts that CAG's guidelines are in flux, remain unwritten, and are not presented in the individual assessment documents provided to the review committees. As a result, reviewers are likely to be unaware of the operational ground rules used In interpreting carcino- genicity data and developing risk estimates. The absence of an frolic it discussion of the application of Agency guidelines and of discussion of the rationale for the choices made in a risk assessment blurs the distinction between science and policy considerations in CAG assessments.

109 Peer Review Drafts are reviewed by all members of the CAG staff and its Director. Drafts are also usually sent for review on an ad hoc basis to knowledgeable persons outside the agency. Bowever, this review process is not part of the public record, and criticism may be accepted or rejected at CAG's discretion. The lack of adequate procedures to ensure that peer review comments are given proper consid- eration may lessen any benefits to be derived from peer review early in the process of developing a risk assess- ment. Draft risk assessments are usually reviewed by the Director of the Office of Bealth and Environmental Assess- ment, directors of other units in this office, and Office of Research and Development staff before being submitted to the requesting program office. CAG assessments are often submitted to committees of EPAts Science Advisory Board or to the Scientific Advisory Panel for peer review. Such reviews take place in public sessions, in accordance with the requirements of the Federal Advisory Committee Act. They provide an opportunity for interested members of the public to review CAG documents and to communicate criticisms to the reviewing co~mnittee and EPA. Reviews of CAG assessments by EPA panels have been mixed, with some panels, such as the Scientific Advisory Panel, often approving the assessments and etchers finding numerous shortcomings related to both substance and format (e.g., the Subcommittees on Arsenic as a Possible Bazardous Air Pollutant and on Airborne Carcinogens of the Agency's Science Advisory Board). This public review process usually leads to revisions. NIOSE-OSBA The Occupational Safety and Bealth Act of 1970 created two new organizations: OS" and NIOSE. OSEA was a new component of the Department of Labor. NIOSE was placed in the Department of Health, Education, Ad Welfare, now the Department of Bealth and Buman Services. Since 1973, NIOSE has been a part of the Centers for Disease Control in the o.S. Public Bealth Service. The colon mission set for both agencies was the protection of the health of American workers. NIOSEts primary functions included the conduct of research and development of criteria for recom mendations to OSLO for occupational health standards. In addition, the Act authorized NIOSE to redevelop and estab-

110 fish recommended occupational safety and health stan- dards. ~ Although it Is not technically correct to refer to NIOSH criteria documents simply as risk assessments, because the doc~ents contain additional information concerning risk management (e.g., engineering cons~dera- tions) as well as recommended standards, ache documents normally included sections that dealt with Me adverse health effects of the substances being considered. The health-effects sections would correspond to the Come mittee's definition of hazard identification. The legislative history of the Act makes it clear that Congress intended a close coupling between NIOSE's recom mendatzons and OSEA's standards. Nevertheless, relatively few NIOSE criteria documents have led to OSLO standards. This disjunction between the two agencies has stemmed from the difficulty of coordinating two organizations that are physically separated and responsible to different depart- ments. As mentioned earlier, the degree to which OS" has relied on NIOSE for its scientific expertise has varied. In the early 1970s, OSEA relied heavily on NIOSE for evaluation of health effects; later, OSHA developed its own staff of health scientists and, with considerable help from consultants and contractors, performed its own risk assessments to support agency standar~setting activities. Because OS" conducts its own assessments of risk, as well as setting standards, and NIOSH does risk assess- ments and recommends standards, the relation of NIOSE and OS" as it has existed since 1976 represents, in some sense, duplication, rather than true extra~agency separate tion. The earlier relation between the two agencies is, however, an example of extravagancy separation. This sec- tion focuses on NIOSE's production of criteria documents during both phases and reflects procedures used throughr out the 197Os. Agenda and Procedures In the past, NIOSE had an elaborate procedure for setting priorities, which included soliciting nominations of cant dilate substances from OSLO and the public. ~ In practice, however, before 1976, NIOSE's criteria document agenda was set by agency personnel and the Director, on the basis of their views of the seriousness of various occupational hazards and the number of workers exposed to such hazards. OSlIA played little or no role in the selection process,

111 and NIOSE's agenda for documents therefore did not reflect or greatly influence OS~A's regulatory agenda. One cause of this lack of correlation between the two schedules was their physical and organizational separation. In the late 1970s, NIOSE did receive communications from osaA that led NIOSE to begin production of process- and industry- or~ented criteria documents. Table 771-5 lists criteria documents transmitted to OSHA. Methods and Use of Guidelines ~- Preparatzon of a criteria doc''=ent involved a preliminary review of literature and the identification of gaps in the relevant knowledge. This gap analysis was fed into N}OSE's research planning and led to research directed at filling the gaps. Brief studies could be completed in time for their results to be incorporated into the docu- ment. Others would continue after the document was come pleted and sometimes resulted in revision or updating. The literature review and preparation of a draft document were commonly performed by an external contractor under the supervision of NIOSE personnel. Because NIOSE does not have written guidelines for risk assessment, whether personnel preparing the documents used similar approaches to evaluate data and reach conclusions regarding risks is unclear. NIOSE's failure to develop risk assessment guidelines has helped to obscure the distinction between scientific and policy judgments in Abe risk assessment process. Although the rationale for separating NIOSE from OSEA has been to allow an independent scientific evaluation without the consideration of economic implz- cations that is necessary in OSElA rule~making activities, the effectiveness of this institutional separation in eliminating the effects of such risk management considera- tions on the conduct of risk assessment by NIOSE is dif f icult to determine. Peer Review The initial review of a draft criteria document was typi- cally performed by NIOSE staff in the same division of the agency that produced the document. The division draft was then submitted to other NIOSE divisions for review. This was followed by a review performed by knowledgeable exerts from industry, labor organizations,

112 TABIE lII-5 NIOSE Criteria Documents Sent to OSEA by May 1982 Substance or Sub Sect Acetylene Ac rylamide Acrylonitrile Alkanes Allyl chloride Ammonia Antimony Arsenic, inorganic Asbestos Asphalt f Ames Benzene Benzoyl peroxide Benzyl chloride Berylli,~- Boron tr if luor ide Cadmium Carbaryl Carbon black Carbon dioxide Carbon disulfide Carbon monoxide Carbon tetrachloride Chlorine Chloroform Chlorophene Chromic acid Chromi''= the ) Coal - Gasification plants Coal-liquefaction (Vole. I and II) Coal-tar products Cobalt Coke~oven mission Confined spaces (as workplaces) Cotton dust Cresol Cyanide, hydrogen, and cyanide salts Decomposition products of fluorocarbon D ibromochloropropane Diisocyanates D initro~o~crenol Dioxane Emergency egress from elevated work stations Epichlorobydrin Ethylene dibromide Fibrous glass Fluorides, inorganic Formaldehyde Furfuryl alcohol Glycidyl ethers }lot environments Bydrazines Transmitted to OSEA 1976 1976 1977 1977 1976 1974 1978 1974, ~ 975 1972, 1976 1977 974, 1977 1977 1978 1972, 1977 1976 1976 1976 1978 1976 1977 1972 1975, 1976 1976 1974, 1976 1977 1973 1975 1978 1981 1977 1981 1973 1980 1974 1978 1976 1977 1977 1978 1978 1977 1975 1976 1977 1977 1975 1976 1979 1978 1972 1978

113 TABLE I7I-5 (Cons' nued) Substance or Subject Transmitted to OSEA Hydrogen fluoride Hydrogen sulf ide Bydroquinone Identif ication system for occupationally hazardous mater ials Isopropyl alcohol Kepones Ketones Iliad, inorganic 1,ogging--from felling to first haul Malathion Mercury, inorganic Methyl alcohol Methylene chloride Methyl parathion Nickel, inorganic and ca~ounas Nitric acid Nitriles Nitrogen oxides N i trog lycer ire me thylene glycol dinitrate No ise Organotin Funds Parathion Pesticide mar~ufacturing and formation Phenol Phosgene Polychlorinated biphenyls Refined petroleum solvent S ilica, crystalline Sodium hydroxide Sulfur dioxide Sulfuric acid 1,1,2 ,2-Tetsachloroett2ane Tetrachloroethylene Th ills: n-alkane mono~, cyclohexane, and benzene Toluene Toluene diisocyanate Toluidine 1, 1, l-Tr ichloroethane Tungsten and cemented tungsten carbide Ol tra~riolet radiation Vanadium Vinyl acetate Vinyl chloride Vinyl halides Waste anesthetic gases and vapors Xylene Z inc oxide 1976 197J 1978 1974 1976 1976 1978 1973, 1977 1976 1976 1973 t976 1976 1976 1977 1976 1978 1976 1978 1972 1976 1976 1978 1976 1916 1977 1977 1974 197S 1974, ~ 977 1974 1976 1976 1978 1973 1973, 197 8 }978 1976 1977 1972 1977 1978 1974 1978 1917 1975 1975

114 and universities. In addition, other appropriate goveror ment agencies, professional associations, and trade organizations were invited to review the document. After these various reviews were complete and changes were made as deemed appropriate by division staff, the document was forwarded to the Director of NIOSH. Several shortcomings of NIOSH criteria documents were cited In a recent review of the program funded by the agency: the lack of field experience of criteria document managers, the lack of critical analysis of data, and the alleged disregard of reviewers' comments. The latter claim highlights the importance of procedures that ensure that reviewers' comments are adequately addressed. The lack of critical analysis of data has been attributed at least in part to the facts that the documents were often developed by outside contractors and that NIOS]I had little control over the personnel assigned to the contract staff. COMM¢T~BES OF TEE NATIONAL RESEARCH COUNCIL The Rational Research Council (NRC) is the operating unit for the National Academy of Sciences' advisory Function. As part of this advisory function, NRC has been called on by a nether of regulatory agencies to perform risk assess- ments. Regulatory agencies request assessments by N~C for several reasons, including statutory requirements that particular agencies or programs consult with NRC, inade- quacy of agency staff to perform the assessments (as in the case of the FDA request for a review of pre-1962 prescription drugs), and such political objectives as a desire for outside scientific support of an anticipated agency action or a desire to defuse or postpone contre versy. Agencies remain free to accept or reject the analyses and conclusions included in ARC reports. NRC risk assessment reports are usually not sufficient by themselves to dictate specific regulatory action, and a separate assessment is usually conducted by the agency, even if in only the most perfunctory fashion. NRC has done risk assessments for several agencies with jurisdiction over carcinogenic chemicals. However, NRC is in no real sense a centralized risk assessment body and is a very imperfect model for recent proposals to create such a body. First, most of the evaluative work of the NAP is actually performed by individual connt~ittees created on an ad hoc basis for each study. Thus, Nat is not a single risk assessment body, but

115 rather an umbrella for the work of many diverse, if out- wardly similar, committees. Second, each ad hoc committee generally reports to a single agency and does not perform assessments for several bodies at once. The committees of N~C have been included in our survey as examples of ad hoc risk assessment groups that are entirely separate from government regulators. Table I7I-6 is a partial list of N~C reports (published since 1977) that examined the car- c~nogenic risks associated with exposure to particular chemicals. Agenda and Procedures Committee members are appointed on the strength of their professional casual' f ications; they may come from univer- s~ties, industry, government, or another sector of soc,- ety, but they do not serve as representatives of any agency, group, or institution unless they are specif'- cally so designated on appointment. Occasionally when, by virtue of special expertise or for other reasons, per- sons affiliated with interested parties are placed on committees, every effort is made to achieve a balance of interests. In any case, all committee members are asked to complete a statement, Won Potential Sources of Bias,~ which includes information on sources of personal income, sources of research support, and more subtle forms of personal bias, including values held that may influence a member' s judgment. The membership of every committee that will formulate a position, take an action, or prepare a report is reviewed by NRC staff and must be approved by the Chairman of my. The work of the committees is facil- itated by professional and support staff employed by ARC. The conduct of a study varies with its nature and objective, the time permitted to complete it, its politz- ca1 sensitivity, and the personalities involved. In gen- eral, committees have considerable latitude in carrying out their responsibilities and may hold public meetings and schedule technical conferences to collect pertinent information. Committees typically meet three to six tomes a year. Meetings are concerned with planning, discussions of issues and drafts of reports, and, later, the development of f inal conclusions and recommendations. Although a committee has much freedom in planning and executing its study and reaching its conclusions, several rests lotions include the obvious necessity to respond to the charge stipulated in the contract, tine and budgetary

116 TABLE lII-6 Some NRC Reports Dealing with Carcinogenic Chemicals (1977-1982) Report Parent Anna Year An Assessment of Mercury in the Environment An Evaluation of the Carcino- genicity of Chlordane and Beptachlor Drinking Water and Bealth Arsenic Nitrates Sacchari:: -Technical Assess- ment of Risks and Benefits Po lychlorinated Biphenyls Drinking Water and Healths Vol. IIT The Alkyl Benzenes Formaldebydc An Assessment of Its Bealth Effects Regulating Pesticides Aromatic Amines: An Assess- ment of the Biological and Environmental Effects Formaldehyde and Other Aldehydes The Bealth Effects of Nitrate, Nitrite, and N-Nitroso Compounds Indoor Pollutants Selected Aliphatic Amines and Related Compounds: An Assessment of the Biological and Environmental Effects Alternatives to the Current Use of Nitrite in Foods An Assessment of the Bealth Risks of Seven Pesticides f or Termite Control Diet, Nutrition, and Cancer Drinking Water and Health, Vol. IV Quality Criteria for Water Reuse Possible Long-Term Bealth Effects of Short-Term Exposure to Chemical Agents, Vol. 1--Anticholinesterases and Ant~cholinergics APSE CLS CLS CLS CPS~ CDS CPS~ CBS CONS CLS CPS~ CLS CLS CLS CLS CLS CLS CLS CLS CONS CLS CLS 1977 lg77 1977 1977 1978 1918 ~ 979 1980 1980 1980 1980 1981 1981 1981 1981 1981 1982 1982 1982 1982 1982 1982 LAPSER = Commission on Physical Sciences, Mathematics, and Resources; CDS = Commission on Life Sciences.

117 limitations, and the necessity for a central NBC-monitored review of the final report. In addition to providing scientific analyses on which policy or regulatory decisions can be based, NBC reports sometimes make specific recommendations for changes in government policy. Methods and Use of Guidelines NBC risk assessments are not east ly classified or charac- terized. Because different committees prepare risk- related reports and N~C does not have any guidelines on the conduct of risk assessments for the committees to follow, approaches and final products show pronounced variations. The absence of guidelines, coupled with the occasional practice of not including a clear explanation of how conclusions concerning risk were reached or of the assumptions used in the quantitatzon of risk, has led to a blurring of the distinction between scientific and policy judgments made in the assessment of risks. The lack of guidelines has also led to inconsistencies in approach and final decisions among committees. However , the absence of specific guidance for interpreting data and for choosing methods of dose-response assessment or risk characterization is probably to be expected, ina`'l~uch as ARC committees consist of scientific experts whose independent judgments are being sought. Probably only guidelines that are extremely flexible could be adopted by ARC. A subject of much discussion over the last sev- eral years has been the value of including quantitative assessments (in our terms, dose-response assessments or, if exposure data are incorporated, risk characterizations) in reports. The trend in recent years has been to include some form of a quantitative risk estimate. Peer Review Every report from the SAC is reviewed by a group other than ache authors. The process of reviewing is overseen by the Report Review Committee. The reports likely to receive reviews coordinated by that Committee are those judged to have significant policy implications and likely to be controversial; most reports that address risk- related questions would be in this category. (The Report Review Committee also coordinates Me review of noncontro~

118 verbal reports on an ad hoc basis to monitor the overall quality of NBC reports.) A report not receiving such a review is reviewed under the auspices of its parent come mission, independent office, or board. Report Review Committee review entails submission of a draft report to ~ set of reviewers selected in a cooperative process by the the parent body and the Report Review Committee. These reviewers are invited to comment on technical adequacy and accuracy (the expertness of the authors), on clarity and appropriateness of presentation, on response to charge, on cogency of recommendations with respect to data presented, and on degree of objectivity and freedom from bias. The committee and staff respond to reviewers' criticisms and suggestions, and the responses are exams fined by a monitor, usually a member of the Report Review Committee, to determine their appropriateness. Thus, a person outside the unit that prepared the report decides whether adequate consideration has been given to re- In cases of persistent and severe disagreement between reviewers and authors, the matter may be referred to the NBC chairman for resolution. Like the regulatory agencies, N~C has been the subject of controversy in recent years. viewers' comments. Some NRC committees have been accused of bias related to their judgments on the risks associated with the substances they are studying. The absence of a member from a discipline that is impor- tant for a balanced assessment of risk can also weaken the credibility of an NRC report. For example, an inter- nal NRC study (1981) stated that, In a small sample of risk-related studies completed before 1979, such disci- plines as epidemiology were often not represented on the rosters of committees whose subjects appeared to warrant such knowledge. FO`'S DRUG EVALOATIOW PANELS Under the Federal Food, Drug, and Cosmetic Act, FDA regulates the marketing of all medicines for human use~-prescr' ption pharmaceuticals, over-the~counter drugs, and biologic products, which are also subject to the 1902 Biologics Act. In its efforts to ensure the safety and effectiveness of drugs in these three classes, FDA has relied heavily on advisory panels composed primarily of scientists from academic medicine. Two major programs illustrate the important role of such independent expert panels in agency assessments of human

119 drugs: the Drug Efficacy Study, ~ review of the effec- tiveness of pre-1962 prescription drugs undertaken by NRC in 1966; and the o~rer-the~counter Drug Review, in which advisory panels established directly by FDA have evaluated the effectiveness and safety of ingredients of such drugs. Both the NRC review and the FDA-directed review enabled FDA to undertake systematic studies of product performance that would have overwhelmed the agency's own resources and personnel. The two reviews differed in a number of respects that may shed some light on optimal structures and procedures for scientific panels. NBC Review The 1962 Refauver-Earris Amendments to the Federal Food, Drug, and Cosmetic Act required that all new drugs be proved effective, as well as safe, and obliged FDA, after a 2-year grace period, to require proof of efficacy of all pre-1962 drugs. In discharging this obligation for prescription drugs, the agency turned to Nat to establish some 30 panels of six to eight experts in pharmaceutical therapy; each panel was responsible for a class of drugs. The panels evaluated the data supplied to them by FDA and manufacturers and rated the drugs as effective, prob- ably effective, possibly elf active, ineffective, ineffec- t~ve as a fixed combination, or inferior to other better or safer therapies for Me same indications. Their main function was thus to assess therapeutic efficacy, not risk to patient health (except indirectly); all the drugs reviewed had been judged to be safe before original FDA approval. Nevertheless, the panels included comments on the safety of individual drugs, particularly those whose effectiveness was in doubt. An informal N~C coordinating group attempted to review each panel's ratings before for- warding them to FDA, in the hope of ensuring some consis- tency. In practice, however, the panel's verdicts reached FDA largely unreviewed. The clinical and other data on which the panels relied came from FDA files, the medical and scientific litera- ture, and the manufacturers or the drugs. The panels neither performed nor ordered any new research, although their assessments often identified subjects on which further studies were needed. The panels met and worked privately; apart from being invited to submit supporting data, manufacturers had no opportunity to participate in the panels' deliberations, nor did representatives of consumers or FDA staff.

120 To reconstruct precisely how the panels worked or to determine what criteria for evaluation each followed is difficult. The predetermined categories in which they were to rate drugs produced apparent homogeneity in their results, but did not sharply confine or direct their analyses. Evidently, wide variations occurred among the panels. The panels' assessments were reported to EDA largely as statements of conclusions; many of the reports were only one or two paragraphs long. Explanations for the ratings typically took the form of bare reverences to published studies or invocations of the informed judgment of the panelists. In short, the panels provided verdicts, rather than documented evaluations. The weight to be given the panels' assessments was not squarely addressed when FDA contracted for NRC assistance Apparently, it was understood that FDA remained free to accept or reject a panel's judgment, but it must have expected to accept most of the panels' assessments when it contracted with NRC. The agency's primary goal was to spare its own scientific staff the enormous burden of evaluating the effectiveness of thousands of pro 1962 drugs. In practice' FDA has accorded substantial weight to the assessments provided by the NRC panels, usually accepting the rating provided and initiating appropriate regulatory action. A rating of less than ~effective. led to notification of a drug manufacturer that more data were needed to support a Clara of effectiveness; later (often years later), if data were still considered inade- quate, the agency took steps to remove the drug from the market. Some of the agency's efforts provoked protracted litigation and administrative hearings. Bowever, pharma- ceut~cal manufacturers have acceded to the panels' judg- ments in the majority of instances, occasionally by withdrawing products from the market, more frequently by eliminating claims for which supporting evidence was lacking, and sometimes by sponsoring new clinical re- search. One important determinant of the acceptance.of panel assessments was the commercial importance of the product or claim at issue. When a panel rating and ultimate FDA judgment jeopardized the continued marketing of an Important product, the manufacturer often insisted on its full legal rights in the course of combating FOA's efforts at implementation.

121 FDA-Directed Drum Panels The NP:C review of pre~l962 drugs did not address the mar- keting status of most over-~hc-counter drugs. In 1972, FDA launched a second comprehensive review, this time on both the effectiveness and the safety of all active ingre- dients in over-the-counter drugs. At the outset of this review, FDA chartered 17 advisory committees representing therapeutic groupings. These 17 panels met a total of 522 tomes over a 9-year period; they reviewed 722 active ingredients for over 1,400 indications and submitted over 75 reports on different therapeutic categories, e.g., internal analgesics, ant~microbials, and vaginal contraceptives. m e central function of these review panels was to report and explain their assessments of the safety and effectiveness of the ingredients used in over-the-counter drugs. These reports were to set forth not only the panels' judgments rating each ingredient (as generally recognized as safe and effective, as unsafe or ineffec- tive, or as requiring additional study) , but also sup- port~ng documentation and rationale. The panel reports became treatises on the various therapeutic categories, some well over 1,000 pages long. The recommendation segments of the reports were considerably shorter. FDA intended from the outset to rely heavily on the panels' assessments and thus insisted that they produce thoroughly documented findings. In addition, the panels were required to meet in public and to adhere to other requirements of the Federal Advisory Committee Act. Together, these obligations prolonged Me panels' delis Orations. Although the Antacid Panel completed its report in less than a year, more complex categories, containing more ingredients, occupied panels for several years, during which they may have met once a month. The responsibility of producing a fully documented report required the panels to rely on FDA staff to assemble information, handle administrative and steno graphic responsibilities, and often do much of the drafting. Thus, the sharp separation that existed between FOA's Bureau of Drugs and the N~C panels never characterized its relation with the over-the~counter panels. However, because discussions of draft reports were held in public meetings and panel members reached their judgments in these meetings, the fact that the final text and judgments represented the, r views, rather than those of agency staff, was clear. The assessments

122 of the panels generally have commanded considerable accep- tance, because they were reached through public debate and were thoroughly documented. At the outset of the review, FDA forecast that it would implement most of the panels' assessments. The agency has released the panels' recommendations in the form of notices of proposed rule-making, which are published in the Federal Register as the first step in translating them into regulations. The Bureau of Drugs has expressly reserved the privilege of disagreeing with a panel's find- ~ngs either immediately or in a tentative final monograph, and it has sometimes done so. These occasions nave nee': few, but usually controversial; and sometimes the Bureau has retreated from its initial disagreement. No manufac- turer has been successful in overturning, administratively or in court, a panel judgment in which the Bureau of Drugs concurred. Perhaps an even better measure of ache credence given Me panels' assessments is the high degree of voluntary compliance displayed by manufacturers. They have abase doned, albeit often reluctantly, most of the ingredients whose effectiveness the panels have doubted. Almost without exception, they have acceded to the panels' safety judgments. Similarly, they have generally accepted the panels' recommendations for changes in labeling. This remarkable c~erczal deference to scientific judg- ment has several explanations, in addition to the credi- bility of the panels. The slow pace of the review per- mitted manufacturers to make changes in their formulas or labeling without serious market disruption. The proce- dures of the panels themselves afforded opportunities for manufacturers to submit information and mace arguments before a judgment was rendered. Perhaps as important, the panels' assessments, thus far, have not often jeopar- dized the continued marketing of major products or whole classes of drugs. If that occurs, it is likely that the panels' findings will encounter more determined opposition. · . ~ _ _ _ NA~TO~ TOXICOLOGY Phi The National Toxicology Program (=P) was established In 1978 by the Secretary of Me Department of Health and Human Services to coordinate all toxicity testing of chemicals in the Department and to facilitate c03mnunica- tion between the research and regulatory agencies. NT'P

123 embraces the relevant toxicity tests ng activities of the National Cancer Institute, National Institute of Environ mental Health Sciences, FOA (and its National Center for Toxicol ogical Research), and the Centers for Disease Control. OSAKA, EPA, and CPSC also participate In NTP. A major advisory group for NTP is its Executive Committee, when Is mare up of the heads of the agencies listed above, as well as the Director of the National Institutes of Health and the Assistant Secretary for Health. NTP thus serves as a vehicle for cooperation among the four regulatory agencies--FDA, ~PA, OSBA, and CPSC--especially in recommending candidate substances for testing. At least one agency has also called on NTP to review risk assessments: the FDA has on two occasions asked another NTP advisory group-he Board of Scientific Counselors-- to review the carcinogenicity data and the agency's analy- sis of those data on two color additives being considered for agency approval. In addition, NTP has served on one occasion as a structure through which a risk assessment of interest to all four regulatory agencies was performed. In April 1980, CPSC (in cooperation with the }nter- agency Regulatory Liaison Group) requested that the NTP help to form an interagency panel on formaldehyde to review the carcinogen~city data on this chemical. The Panel consisted of 16 government scientists, most of whom were experts in toxicology, pharmacology, and epidemi- ology. Three of the IRVING agencies--EPA, For, and OSEA- also supplied scientists as m- - ers. Although no employee of CPSC was an official Panelmember, a liaison represent tative of the agency attended all meetings and contributed to portions or the final report. In addition, CPSC per- sonnel assisted the Panel by preparing bibliographies and handling arrangements. The Panel on Formaldehyde thus serves as an example of a centralized assessment body that, although placed outside the agencies, maintained some association with the scientific staffs of each. The decision to confine the membership to government scientists was driven, in part, by a desire to avoid delays associated with come pliance with the Federal Advisory Committee Act's require- ments for establishing outside committees. The Panel's creation was viewed as an experiment in interagency coordination. The Panel bet three toes. It generally deliberated in private, and its meetings were not announced. The Panel did consult why Chemical Industry Institute of Toxicology scientists who were responsible for designing

124 and conducting the carcinogenic~ty study being evaluated, and it permitted both oral and written statements from the Formaldehyde Institute, a trade association of users and manufacturers. Although the Panel reported its find- ings somewhat later than initially forecast by CPSC, the tome required was a relatively brief 6-7 months. One unanticipated delay resulted from the necessity for a second review of the pathology slides from the major study being evaluated. The report stated that evaluation of the findings on carcinogenic effect and other related data convinced the Panel members that formaldehyde is an animal carcinogen when inhaled. This finding has been supported by many other scientists, and the Panel's report has since been published in a peer-reviewed scientific journal. The Panel also concluded that none of the avail- able epidemiologic studies negated the inference that formaldehyde posed a cancer risk for humans. It did not att- - t to est~ate the risk of cancer for any exposed segment of the population. It did include, however, a quantitative dose-response assessment. The NTP Panel's formation and performance demonstrate that such ad hoc collaboration is manageable and can function well. Despite the quality of its report and its timely production, however, the NIP Panel's deliberations and report have not yielded any regulatory efficiencies. In early 1982, CPSC banned further use of urearformal- dehyde foam insulation, in part on the basis of the Panel's report, as well as the agency own risk assess- ments of formaldehyde's acute and chronic effects. In contrast, EPA has declined to initiate regulation of formaldehyde in response to the Pane1's assessment. The Agency declined to act under Section 4(f) of the Toxic Substances Control Act, noting that the animal data ava~l- able on carcinogenicity did not constitute a Treasonable basis to conclude that [formaldehyde] presents or will present a significant risk of serious or widespread harm to human beings from cancer. . . .~ However, because the Agency's posture Is equivocal and not clearly documented, the degree to which it relied on the Pane}' s assessment in reaching the conclusion Is unclear. Neither of the other two agencies followed CPSC' s lead. OSLO declined to issue an emergency standard for worker exposure to formaldehyde, concluding that it poses no imminent hazard; and it recently announced that it was unable to proceed to establish a permanent standard, because the evidence of animal c:arcinogenicity did not

125 reveal what, if any, risk exposed workers might confront. These decisions were also based on OSEA's own assessment of risks, but the degree to which O SEA relied on the Panel's assessment for the agency hazard identification step is unclear. Both EPA and OSBA are continuing to collect data on formaldehyde, but no regulatory action appears likely in the near future. FDA has not acted, because the potential formaldehyde exposures from agency- regulated products were judged to be very low. The contrasting regulatory outcomes should not be interpreted as indicative that the Panel on Formaldehyde failed in its mission. Although the four agencies planned to consider its report carefully, the Panel's findings were not expected to be binding. Each agency remained free not only to fashion its own regulatory response on formaldehyde, but to qualify, or to dissent from, the Panel's determination of carcinogenicity and estimate of risk. Factors other than the Panel report's validity and utility are more likely explanations for the divergent agency responses. First, the Panel's report was subs misted shortly before the 1980 national election, whose outcome forecast fundamental shifts in regulatory policy at EPA and OSKA. Second, the agencies confront exposures to formaldehyde that differ widely in character and intend salty, yielding important differences in potential risk. Finally, the statutory criteria governing their decisions could plausibly lead them to accord different weights to the Panelts findings. OSEA, for example, had to decide whether formaldehyde posed a risk sufficient to justify emergency protective measures despite any costs of immediate action. EPA'S USE OF SCIENTIFIC KE:VIEW PANELS The EPA has had considerable experience with independent scientific panels, but they have served the Agency differ- ently from the risk assessment panels discussed in the preceding section. EPA's panels typically have reviewed the work of Agency scientists and analysts, rather than perform their own risk assessments. Also, most panels serving EPA are mandated by Congress and play legally prescribed roles in the Agency's decision-mak~ng process. We examined two such panels: EPA's Scientific Advisory Panel and the Subcommittee on Airborne Carcinogens (a unit of EPA's Science Advisory Board).

126 EPA' s Scientific Advisorv Panel (SAP) The Scientific Advisory Panel was established by Congress in the 1975 Federal Insecticide, Fungicide, and Rodenti- cide Act to review EPA's evaluations of the environmental and health risks posed by specific pesticide uses. Broad- ly speaking, the Panel reviews risk assessments prepared by EPA's Office of Pesticide Programs to support content plated regulatory actions against hazardous pesticides. It also reviews the proposed and final Corms of such actions. Consultation was initially required only when the Agency contemplated suspending or canceling a pesti- cide's registration or issuing general regulations governing pesticide registration. Cancellations and general pesticide regulations must be submitted to the Panel for review before they take effect. Suspensions of registration do not require prior review, but EPA must submit the underlying studies for review promptly after any suspension action. EPA must also submit for peer review the ~design, protocols, and conduct of major scienr tific studies. conducted under the pesticide act. The following description reflects activities undertaken before September 1981.* The Panel normally consists of seven members selected by the EPA Administrator from among six persons nominated by the National Institutes of Bealth and six nominated by the National Science Foundation. Until its last meeting in June 1981, the Panel generally met once a month. Topics covered during 3980 and 1981 are shown in Table III-7. The Panel does not set its own agenda, although the chairman may control the sequence and conduct of indi- vidual sessions. Me risk assessments that the Panel reviews are selected by the two divisions (Hazard Evalua- tion and Special Pesticide Review3 of the Office of Pesticide Programs that use its recommendations. Virtu- al~y all the scientific and exposure information avail- able to the Panel is provided by the division whose assessment is being reviewed, although much of this information comes originally from the registrant of the product in question. Panel members necessarily accept the authenticity of tire information provided, although they sometimes question its quality. *Authorizing legislation expired in September 1981, and new legislation }was not been enacted (as of December 1982) ~

127 TABLE IIT-7 EPA Actions Reviewed by the Scientific Advisory Panel (1980-1981) A . Regular ion Insecticide, Fungicide, and Rodenticide Act 1. Final Rul~m~king for Registering Pesticides in the United States, Subpart E, Hazard Evaluation: Wildlife and Aquatic Organisms 2. Proposed Rulemaking for Registering Pesticides in the United States, Subpart L, Hazard Evaluation: Nontarget Insects 3. Proposed and Final Rulemaking for Registering Pesticides in the united States, Subpart D, Chemistry Requirements: Product Chemistry 4. final Rulemaking for Amendment of 40 CFE 162.31 by Adding Certain Uses of Eight Active Ingredients as Restr icted Pesticides Proposed Rul-~?king for Registering Pesticides in the Onited States, Subpart M, Data Requirements for Biorational Pesticides Final Rul-~king foe Registering Pesticides in the United States, Subpart N. Chemistry Requirements: Environmental Fate 7. Informal Review of Draft Proposed Pesticide Registration Guidelines, Subpart I, Exposure Data Requirements: Reentry Protection 8. Review of Proposed Pesticide Registration Guidelines, Subpart 6, Labeling of Pesticide Products 9. Review of Final Rule on Classification of 11 Active Ingredients for Restricted Use B. Cancellations unde~ Insecticide, Fungicide, and Rodenticiae Act 1. Dime~choate 2. Diallate 3. Lindane 4. Strychnine 5. Ethylene bromide 6. OxyEluorfen (Goal 2E ) 7. Wood preservatives, pentachlorophenol, creosote, arsenicals

128 Meetings are open to the public, and interested parties are generally encouraged to make presentations. These meetings sometimes focus on risk management issues, rather than on the health and environmental assessments submitted to the Panel, in part because participants making present tations are not confined to addressing scientific aspects of the Agency's risk assessments. Equally important in the consideration of nonscientific issues has been Cow gress's decision not to restrict the Panel to a strictly scientific review of the Agency's risk assessments. (The Panel's mandated review responsibilities extend to contemr plated EPA actions that combine both risk assessment and regulatory policy elements.) Although the rationale for the Panel's creation was to introduce independent sczenr tific review into EPA's deliberations, the mechanism chosen has routinely resulted in the Panel's commenting on the Agency's choice of regulatory options. The Agency has sought to anticipate the Panel's tendency to stray __;A~~ ~ ~~- ~ Chill-= before it and has attempted to -- requested. The participation of the Panel probably has Improved the quality of EPA analyses and added to their credibility among both environmental and industry groups. Bowever, expectations of same EPA critics that it would repudiate the Agency's scientific analyses have not been realized. Over the last ~ years, the Panel has agreed with most Agency risk assessments brought before it. mere have been some notable exceptions, such as the Panel~s dis- agreement with the Agency's handling of 2,4,5-T. Me endorsement of most Agency assessments and Agency actions based on those assessments by the Panel have been ex- tremely helpful in improving Agency credibility and rendered its actions less vulnerable to challenge in administrative or judicial hearings, as with the Panel's support of EPA action on wood preservatives. The Panel's success can be traced to several causes: its public deliberations, which may have made it difficult for EPA to ignore its comments; its continuity (until its author- iz~ng legislation expired), which permitted it to under- stand EPA's approaches and simultaneously strengthened its influence with EPA staff; and the scientific distinc- tzon of individual Panel members. In the case of EPA's decision to suspend use of 2,4,5-T and Silvex (its companion product) for some applications and to hold wide-rang~ng hearings on other applications, the Pane} declined, after 3 days of public meetings, to support the Agency's proposed proceedings. ~ L U1U Why =~ ~ c ~ ~ ~ ~ _ __ _ ~ Frame If Questions on which comments are

129 The Panel believed that additional data, including results of further tests for carcinogen~c~ty and reproductive tox- ~c~ty and of more college monitoring for residues, were required before a hearing could be hel d prof Stably. Because EPA had not asked the Panel to approve the hold- ~ng of a hearing and believed that it would be more efficient to deal with all uses of 2,4,5-T at one time, the Agency persisted and announced a hearing on the risks and benefits of 2,4,5-T, which began in March 1980. This difference, coupled with congressional displeasure wit EPA' s original suspension of 2,4,5-T and Silvex, led ulti— mately to the 3980 statutory amendment mandating that the Scientific Advisory Panel review the studies that underlie suspension decisions. EPA's Subcommittee on Airborne Carcinogens The Subcommittee on Airborne Carcinogens, a part of EPA's Science Advisory Board, was not mandated by statute. It was created in 1980 at the request of the Assistant Administrator for Air, Noise, and Radiation to review the assessments that the Agency is statutorily required to submit for Board review. Members of this Subcommi ttee were appointed by the Administrator; however, it no longer exists, having recently been merged with the Environ- mental Health Committee of the Science Advisory Board. The Subcommittee reviewed six pairs of draft documents that included hazard identification and dose~response assessments produced by the Carcinogen Assessment Group and exposure assessments produced by private contractors for EPA's Office of Air Quality Planning and Standards. The chemicals evaluated in those documents were trichloro- ethylene, perchloroethylene, methylene chloride, methyl chloroform, acrylonitrile, and toluene. Subcommittee members reviewing these documents included a biochemist, a biostatistician, a pathologist, an engineer, an oncolo- gist, a toxicologist, and a meteorologist. Five members were affiliated with universities and one with a research consulting organization; the seventh was a private con- sultant. In accordance with the Federal Advisory Committee Act, the Subcommittee's review was held in public and announced in the Federal Resister, and interested members of the public were invited to make oral and written presenter tions. Several such presentations were made, primarily by representatives of industries that would be affected

130 by EPA regulation of the substances under discussion. EPA and contractor personnel also attended the review and participated actively, briefing the Subcommittee on the contents of documents, answering members' questions, and defending their work against or iticism. The Subcommittee did not write a report after its review, and the absence of a summary report has led to some confusion regarding the nature of its cr~t~cimus. Review of the transcript of its second meeting (September 5, 1980) and discussions with various participants in that review meeting have revealed several general criticisms of the Carcinogen Assessment Group's risk assessments. One was that the documents provided to the Subcommittee were not sufficiently detailed; i.e., they did not prom vice enough scientific information from the various studies cited to permit the Subcommittee to make an inde- pendent assessment of the quality and validity of the studies. Another criticism raised by the Subcommittee was that the conclusions drawn did not reflect the quality of the data on which the risk assessments were based. Some Subcommittee members asserted that such considerations may, in fact, be precluded by rigid adherence to the Agency's guidelines for risk assessment. Other criticisms focused on specific issues, including the validity of basing a conclusion of carcinogenic~ty on an increase in mouse liver tumors, the importance of contaminants in the test chemicals, and the wisdom of using a single model for extrapolating from high to low doses. The Subcommittee viewed these issues as primarily scientific, whereas Agency staff considered them, although resting on scientific principles, as resolvable through the choice of Conservative policy options--a choice embodied in the Agency's guidelines. These differences between the Subcommittee and Agency staff emphasize the conclusion set forth in Chapter ~ that many components of risk assessment lack a fine scientific answer and require a judgment to be made. In same cases, such judgments may be informed by scientific arguments, but may ultimately rest on policy preferences. The difficulties In ca~muni- cation between the Agency and the Subo--uittee also under- score the Importance of explicit risk assessments and written reviews. The differences reported above have not yet been fully resolved. The Agency's experience with the Subcommittee highlights some difficulties in using a review body that has not had sufficient tome to develop a working approach to its task. It also emphasizes the importance of ex-

131 plaining Agency risk assessment procedures, including the adherence to specific guidelines, to review panels. Con cerns similar to those of the Subcommittee have been expressed by me ~ ers of the Environmental Bealth Cow mittee, which replaced it, and Agency staff are currently considering changes in the risk assessment procedures embodied in their guidelines. PROPOSED COWS IN OR~T PATIO ~GE==S FOR RISK ASSESSMENT Proposals to reform the organizational arrangements for risk assessment have been advanced to reduce perceived shortcomings in agency practices. The criticisms to which these proposals respond may be summarized as follows: · Bias. Critics of agency performance suggest that dec~sion~makers approach risk assessment with attitudes about regulation that preclude objectivity. Regulators, for example, may skew their assessment of risks associ- ated with a particular substance to support a preference to regulate or not to regulate that substance. · Exangeration. This criticism Is closely related to the first. The suggestion is that regulatory agencies, accustomed to operating in an adversary mode and expec- t~ng their judgments to be challenged in administrative hearings or in court, typically overstate the risks associated with hazards that they decide to regulate or understate the risks assoc' ated with hazards that they decide not to regulate. The instinct to support a posi- tion with every available argument may distort interpreta- tions of scientific data, choice of extrapolation procedures, and assumptions about human exposure. The critical role of legal staff in preparing agency documents is thought to foster the adversarial style. · Poor Public Understanding. If risks are m~sdes- cribed, it follows that public perception of the risks will be i naccurate . In addition, because agency announc e- ments of regulatory actions typically stress the ultimate risk management strategy, such as the banning of sack chain, and do not explain why a particular action is be' ng taken, the public is led to infer the degree of risk from the action proposed or from the decision not to act. However, an agency's ultimate decision may be dice tated by statutory language or regulatory policies that emphasize considerations other than degree of risk.

132 . Poor-Qua' Sty Personnel. This argument is straightforward, if unflattering. It is that regulatory agencies cannot attract or retain adequate numbers of highly qualified scientists to perform risk assessments. Many of their personnel are removed from active research by time and distance and are unfamiliar with the latest developments in their fields. · Inconsistence. This criticism supports proposals for centralization of risk assessment. To the extent that separation is a prerequisite to centralization, this criticism would also support institutional separation. The suggestion is simply that agencies have applied ~ncon- sistent criteria and reached inconsistent results in assessing the risks posed by the same hazards. Such inconsistency is more likely when each agency is respon- sible for performing its own assessment. · Redundancv. Starting from the assumption that different regulatory agencies have been, and are likely often to be, concerned with the some hazards, the critics argue that current arrangements force government regula- tors, affected industries, and interested scientists to deal with litigation on the risks of a given substance several tomes. Accordingly, a central institution responsible for performing risk assessments for all agencies might yield process efficiencies and reduce costs for all participants. DESCRIPTION OF PROPOSALS The central proposals for changes In institutional arrangements for risk assessments developed by the Office of Science and Technology Policy (OSTP) and the American Industrial Health Council (AIEC) and presented in H.R. 638 have sparked much of the current debate and precipi- tated this study. }for several years before, however, dissatisfaction had been expressed with the procedures by which government bodies used scientific data and resolved what purported to be scientific issues. This dissatisfac- tion led to one of the precursors of the current propo- sals: the idea of a science court for resolving science tif ic issues underlying regulatory decisions. That suggestion and other, more recent proposals for procedural and structural reforms are discussed briefly below. The primary objective of this section, however, is to facili- tate evaluation of the three main proposals that inspired this study.

133 Science Court An important precursor of the OSTP proposal was the science court concept of Rantrowitz (1975). The science court was proposed to assist deciszor~makers with disputed scientific aspects of a decision. Bence, a basic premise of the science count is that it is both possible and desirable to separate the scientific elements of a public- pol~cy decision from social and political considerations. The judges of a court were to be impartial, competent scientists from relevant disciplines who were not involved in the dispute. These Judges would hear testimony from scientific experts on both sides of the issue, who would be allowed to cross-examine each other. The rationale was that scientist advocates are best qualified to present their own cases and to probe the weaknesses of their opposition. In the environment created in such a court, complete objectivity would be neither assumed nor neces- sary. After hearing all witnesses, the judges would issue a spry of their opinion of the meaning of the scientific evidence. Their opinions would deal only with scientific questions and could not include recommenda- tions for public policy. Many details of a science court's procedures and operations are, however, unclear. Even after several years of sometimes heated debate in the scientific and regulatory communities, the overall reactions to the concept can be characterized as at best only lukewarm. Although a genuine science court will probably not be established, the underlying idea of separation of scientific issues from social and political considerations in decision~malcing has since appeared in other proposals. FDA's creation and use of public boards of inquiry is the nearest analogue to the science court that has been put into practice. In 197S, FDA, on its own initiative, adopted regulations describing a public board of inquiry, a new kind of decisional body that could substitute for the traditional trial thy of hearing before an adminis- trative law judge if parties to formal disputes before the agency could agree. A board of inquiry is an ad hoc panel of three independent scientists, qualified in relevant disciplines, who hear evidence and arguments and render a preliminary decision, which may be appealed (like that of an administrative law judge) to the Come missioner of Fry. The procedure assumes that disputes that are primarily scientific can be resolved more accurately, faster, and with greater credibility by an

134 expert tribunal. FOA's novel procedure has been tried only once, to resolve safety issues concerning aspartame, a new artificial sweetener. This experience yielded, at best, equivocal support for the new procedure. Perhaps because of its novelty, the process took over a year to complete. The parties disagreed at length over the makeup of the board, the objectivity of its members, and the procedures it should follow. The FDA Commissioner ulti- mately rejected the board's conclusion that aspartame should not be approved and issued an opinion that both questioned the board's scientific rationale and corrected ins interpretation of the legal criteria for approval of food additives. Other regulatory disputes, including FDA's refusal to approve the injectable contraceptive, Depo,Provera, are scheduled to be heard by boards of . . . ~ napery. OSTP Proposal A 1978 report from OSLO gave impetus to emerging proposals for separation and centralization of scientific aspects of risk assessment. The report recommended several steps to ensure consistency in the identification, characteriza~ tion, and assessment of potential human carcinogens. Two interrelated stages in regulatory decision-making were delineated: Stage I, identification of a substance as a potential human carcinogen, qualitative and quantitative characterization of the risk it poses, and explication of the uncertainties; and Stage II, evaluation of regulatory options and their consequences. This dichotomy closely parallels our own distinction between risk assessment and risk management. The OSTP report recommended that ~ uniform decision-making framework be used in all agencies and that Stage I and Stage II functions be separated within or outside regulatory agencies while sufficient linkages were maintained to ensure relevance and t~meli- ness. Such organizational experiments as the Carcinogen Assessment Group in EPA were highlighted. The report also suggested that the thenrfledgling National Tox~- cology Program might eventually assume an expanded role in coordinating or overseeing some risk assessments for the regulatory agencies.

13S B.R. 638 and the AIEC Proposal The 1978 OSTP report was a broad statement of principles. Two detailed proposals to create new risk assessment institutions have since been advanced. Because these proposals have several features in common, but also present important contrasts, they are summarized together (Table III-8). In February 1980, Representative William Wampler first introduced legislation (U.S. Congress, 198Ic) to establish a National Science Council. H.~. 638 calls for the creak tion of a new panel of scientists, entirely independent of the regulatory agencies, that would decide disputed scientific issues posed by regulatory initiatives. The AIEC had previously (1979) advanced a similar proposal to create an expert science panel that would evaluate the hazards of chemicals considered for regulation. Both proposals stress the importance of uniform, consistent resolution of the scientific questions underlying regu- latory decisions. Both espouse the separation of risk assessment from the design and selection of regulatory responses, and both would use independent scientific experts to perform the assessments. There are some basic differences between the two proposals. Onder B.R. 638, any party could request referral of scientific issues to the National Science Council. The AIEC proposal specifies that, although any party may request a review, only federal agencies or Congress would have the authority to initiate mandatory review of scientific questions by the central science panel. B.R. 638 would apply only in forma} adjudications. The AIEC proposal would apply to any agency proceeding in which risk assessment was at issue. Because rule-making is the primacy mode for regulating hazardous substances, the AIEC proposal would apply to more regulatory actions than would B.R. 638. Under II.R. 638, decisions of the National Science Council would be binding on regulatory agencies. In contrast, assessments of the AIEC's science panel would not bind the agencies, but would carry a presumption of validity, subj eat to rebuttal in later regulatory proceedings. The risk assessment bodies contemplated by the two proposals also differ in composition and procedures. The National Science Council would be a standing body of 15 full-time voting members serving 2-year terns. Individual chemicals would be assessed initially by advisory panels made up only of Council members. Each panel would have

136 TABLE lII-8 Comparison of Major Features of H.~. 638 and the AIEC Proposal H.~. 638 AIEC Proposal Structure: Single continuing panel separate from agencies; centralized Membership: 15 full-time members appointed by chairman of N Sit from NAS nominees; members to be qualified, distinguished scientists scone: Referral by any party of adjudications involving harm to hen health from substances considered by CPSC, FDA, USDA c DEBS,d OSHA, and EPA Functions: Panel could prepare an independent risk assess- ment; its decision would be binding on the agency Public Participation: Parties to adjudication would be involved Implementation: Legislation Single continuing body with rotating members; in the NASH IS part-time members selected according to NAS procedures; members to represent the best scientists Referral by any party or agency (only latter require mandatory cone sideratiOn) concerning proposed rules or agency adjudications; all agencies with regu- latory jurisdiction would be affected Panel could prepare an independent risk assess- ment; its f indings would be advisory, but would be part of record Federal Register notice of referral would solicit submission of data by pub1 ~ c Legislation - ~National Academy of Sciences. "National Science Board. SO. S. Department of Agr iculture. department of Health and Buman Services.

137 at least five voting members. The AIEC science panel would be established under the ·,sohrella of the National Academy of Sciences and consist of 15 part-time members who would serve for terms of 3 years. The panel could establish working groups, which could be composed largely of outside experts. These divergent approaches to places ment and composition of the panels and terms of members reflect different expectations about which status would attract the best scientists and perhaps about the extent to which the results would be binding. For exile, the AIEC proposal assumes that distinguished academic and industry scientists would be unwilling to serve on a full-time basis for any substantial period. Under B.~. 638, the National Science Council would decide scientific questions after conducting a formal shearing on the record,. in which all parties to the agency proceeding could participate. Onder the ALEC proposal, referral of scientific issues to the panel would be announced, and the submission of written evi- dence and arguments would be invited. The less formal procedures visualized by the AIEC are consistent with it objective of obtaining nonbinding expert judgments on scientific issues that underlie decisions. The two proposals embody different expectations as to speed of response. B.R. 638 would require the National Science Council to make a final report to the referring agency within 90 days of receiving a dispute. The ALEC proposal, however, imposes no time ignite on the panel's assessment, except that the panel Operate expeditiously but not precipitously. (Higginson, 1982). Sinqle-Aqencv Proposals E.R. 638 and the AIEC proposal espouse gover~ment-wide reform of the institutional means for risk assessment. Other notable recommendations for institutional restruc- turing have been addressed to individual agencies or agency programs. In 1981, for example, Senator Orrin Batch introduced legislation (AS. Congress, 1981d) to amend the food-safety provisions of the Federal Food, Drug, and Cosmetic Act. Bis bill included a provision permitting FDA to request, or affected third parties to demand, assessment of the risks associated with specific food constituents, with such assessment to be performed by a panel of scientific experts appointed by the National Academy of Sciences. The panel's assessment would be s

138 advisory, rather than binding on the agency. Similar provisions have appeared in other proposals to revise government regulation of food safety, including a proposal developed by the Food Safety Council (1979). These pro- posals appear to share assumptions underlying the DISC proposals: that agency risk assessments cannot be assumed to be objective, thorough, or expert and that an ~ndepen- dent review should be available before a final decision is made. These proposals for independent scientific panels differ from B.R. 638 in three important ways: they would apply to one agency or program; they contem- plate only an advisory role, rather than a resolving function, for the scientific panel; and they would apply to any agency proceeding in which rise; assessments were at issue. The proposals thus can be viewed as agency or _ _ ,= _ _,= ~ program~specific illustrations or tne Ably proposal to create one central scientific panel to serve al' agencies. One such single-agency proposal has been adopted. In 1981, Congress amended the Consumer Product Safety Act (U.S. Congress, Omnibus Budget Reconciliation Act, 1981a) to require CPSC to consult with an ad hoc chronic hazards advisory panel whenever it contemplates rule-mak~ng con- cerning a product believed to pose a risk of cancer, birth defects, or gene mutation. A panel will consist of seven members appointed by the Commission from among 21 scienr fists nominated by the President of the National Academy of Sciences. Nominees may not be employees of the governs ment or have any financial ties to any manufacturer or seller of consumer products. Each nominee must have Demonstrated the ability to critically assess chronic hazards and risk to human health presented by the expo- sure of humans to toxic substances or as demonstrated by the exposure of animals to such substances.. The panel's responsibility is to prepare for the Commission a report on the substance that the agency is considering regular sing. The panel is to review the scientific data and other information related to the substance and Determine if any substance in the product is a carcinogen, mutagen, or teratogen.. The panel will also "include in its report an estimate, if such an estimate is feasible, of the prob- able harm to human health that will result from exposure to the substance. The Act requires that a panel submit its report within 120 days of convening, unless the Com- mission allows it additional tome. A panel's report Shall contain a complete statement of the basis for its determination." The Commission must consider the panel's report and incorporate its evaluation into any advance

139 notice of proposed rule~malcing and any f inal rule. Appar- ently, the agency is not bound by a panel's determination of carcinogenicity or its estimation of the risk associ- ated with exposure. Although it appears that each panel is to perform its own risk assessment, the statute is silent on The role to be played }by agency staff and on the weight that a panel might legitimately accora to analyses prepared by the agency ~ tself. These panels are exempted from the Federal Advisory Committee Act; the exemption pres'?Tnahly means that they are not required to provide advance notice of their meetings or to deliberate in public. A panel may seek information from third parties, but only through CPSC. CRITICISM OF PROPOS=S FOR S=~T7ON ~ ~~I=TION The Four f ederal regulatory agencies have responded skew tically to proposals to separate and centralize the func; talon of assessing the risks of chemicals that are cane didates for regulation (U. S. Congress, 19815) . Other cheers have also found f laws in the proposals. A central crit~ci=. made by those who argue against full organizational separation between risk assessment and regulatory policy-m~king is that simply separating risk assessment from the regulatory agencies would not separate science from policy. , _ _ This argument is based on the fact that the risk assessment process requires analytic choices to be made that rest, at least in part, on the policy con- sideration of whether to be more or less conservative when determining possible publ~c-health risks. A second point is that, although extra-agency separation of risk assess- ment may help to minimize the influence of risk management considerations on this process, the agency responsible for deciding what exposure to permit or what costs to impose must make what is ultimately a political judgment based on the extent of risk determined in the risk assessment and often on the benefits and costs of regulatory action and its feasibility and political acceptability. For its decision to be politically acceptable and the decisions maker accountable, the agency must have responsibility for each or these components of regulatory decisions making. A third argument against institutional senar a- Lion as related to the internal process by which agencies It is claimed that this process is unavoidably an iterative one. Different specialists are called on repeatedly for analysis and advice as an agency reach decisions.

140 identifies and considers new control options in attempting to reach a decision. Although this description may over- state the f luidity of internal agency deliberations, captures something of their ad hoc character. Closely coupled with this argument is Me necessity for agencies to retain scientific capability so that they can under- stand what a risk assessment means and how to use it in developing risk management strategies. Thus, even if risk assessment were performed outside the agency, a scientific staff representing many different disciplines would still be required, to ensure that an assessment would be interpreted and used correctly. Other criticisms of proposals for risk assessment by a centralized panel stress the logistic difficulties of meshing independent risk assessment activities with the internal workings of different agencies. Experience suggests that it will be difficult for any risk assess- ment body to meet even generous time limits. Thus, agency decisions will probably be delayed by ~ requirement to consult, or refer issues to, such a body. A central pane, also might become overburdened and cause additional delays. Critics of I.. 638 and the AIEC proposal chal- lenge the ass, - .tion that the regulatory agencies }rave reached inconsistent conclusions in evaluating various chemicals. The recent differences in Me regulation of formaldehyde constitute a rare example of disparate treatment of the same chemical, and even this disparity may not betray basic disagreement over the interpretation of scientific data, as distinct from the degree of risk that justifies regulation. In the past, the agencies have often selected different control options or imposed different exposure limits for a given chemical, but these disparities have typically reflected differences in expose sure (and thus in risk characterization) or differences in regulatory policy or statutory or administrative requirements; none of the current proposals addresses such differences. CONCLUSIONS The Committee was asked by the Congress to consider tithe merits of an Institutional separation of scientific func- tions of developing objective risk assessment from the regulatory process of making public and social policy decisions and the feasibility of unifying risk assessment functions. " In this chapter, the Committee has addressed

141 these two issues and a third, related issue: the value of independent scientific review of agency risk assessments. In i ts review, the Committee was sensitive to a number of considerations, including the scientific quality and regulatory relevance of the assessments performed. It also tried to ascertain how scientific and policy consid- erations were handled in the performance of risk assess- ment. To reach its conclusions, in the absence of accepted criteria for evaluating agency practices and proposals for change and in view of the sparseness of relevant empirical data, the Committee has relied on discussions with other persons knowledgeable and experi- enced in risk assessment activities, the limited avail- able literature, and especially its own knowledge and experience in regulatory-agency risk assessments, as well as its review and analysis of past agency practices. VALUE OF INSTI'lu$ION~L SEPARATION 1. Although organizational separation mat help to ensure that risk management considerations do not influence the conduct of risk assessment, the degree of organi- zational separation that is optimal for individual agencies cannot be determined on the basis of the Committee' s review. Regulatory programs differ substantially In their degree of organizational separation. In the cases of NIOSE assessments that in the early 1970s were adopted by OSEA and NRC assessments relied on by agencies, the assessment function has been outside the regulatory agencies. At EPA, the risk assessment units in the Office of Health and Environmental Assessment of the Office of Research and Development prepare assessments for regulatory program offices that are organizationally under different assistant administrators. Bowever, ache Office of Toxic Substances does its own assessments, and several other program offices are responsible for their own exposure assessments. The risk assessments for the FDA'S Bureau of Foods are produced within the Bureau, but by an office distinct from offices responsible for fo~mu- lating regulations and enforcement; since 1976, the D~rec- torate of Health Standards Programs in OSLO has both performed risk assessments and formulated all early risk management options. Different agencies also have success-

142 fully used different organizational arrangements for risk assessment. FDA, for example, has often called on NRC and NTP for assessments, but in other cases relied on its own staff. The Committee's review of different agency structures and procedures did not demonstrate that one particular structure produced risk assessments of superior quality and integrity. In addition, the Committee notes that, even if there were a clear finding that a particular arrangement works For a given agency or program, it would be extremely difficult (given the diversity in agency and program mandates, personnel needs, and histories) to justify a suggestion that that arrangement would best serve all agencies or programs. 2. Organizational separation has several important drawbacks that are likely to be intensified with increasing degrees of separation. There are severa} arguments against organizational separation. Separation of the risk assessment function from an agency's regulatory activities is likely to inhibit the interaction between assessors and regulators that is necessary for the proper interpretation of risk estimates and the evaluation of risk management options. Separation cat" lead to disjunction between assessment and regulatory agendas and cause delays In regulatory proceed- ings. Common sense suggests that increased separation would aggravate these drawbacks. In its review, the Com- mittee observed these disadvantages when assessors and regulators were in different organizations (e.g., NIOSE and N~C). Another perceived drawback in extra-agency separation that was neither detected nor likely to --merge in the Committee's review is the erosion of scientific competence within agency staffs if risk assessments are routinely performed outside the agency. Also, any major organizational change may have a disruptive effect on agency performance; thus, such organizational changes are especially questionable when the benefits, if any, are unclear. 3. Organizational arrangements that separate risk assessment from risk management decisionrmakinq will , not necessarily ensure that the Policy basis of - choices made in the risk assessment Process is clearly distinguished from the scientific basis of such choices. _

143 If risk assessment as practiced by the regulatory agencies were pure science, perhaps an organizational separation could effectively sharpen the distinction between science and policy in risk assessment and regu- latory decision-making. However, many of the analytic choices made throughout the risk assessment process require individual judgments that are based on both scien- t,fic and policy considerations. The policy cons~dera- tions In risk assessment are of a different character from those involved in specific risk management decisions and are generally common to all assessments for similar health effects. Thus, even when one has drawn the rela- t~vely obvious distinction between risk assessment and risk management, there remains the more difficult task of distinguishing between the science and policy dimensions of risk assessment itself. We believe that the latter distinction cannot be ensured or maintained through organ- izational arrangements. Given the inherent mixture of science and policy in risk assessment, organizational separation would simply move risk assessment policy into a different organization that would then have to become politically accountable. The Committee believes that other approaches are more likely to maintain the distinc- tion between science and policy in risk assessment, most notably the development of and adherence to guidelines. VALUE OF CENTRALI ZATION 4. Common risk assessments performed primarily by scienr fists from all interested agencies on an ad hoc basis - mav capture the major advantanes of centralization without the drawbacks that accomoanv Permanent, extraragencv centralization. An argument often advanced for centralization is that it might expedite and perhaps reduce the administrative costs of decisionrmaking when two or more agencies contem- plate regulation of the same substance. And if two or more agencies are going to regulate the some substance, there is much to be said for developing a system that facilitates production of a single, common risk assess- ment. This was one rationale for CPSC's decision to empanel a group of scientists to evaluate the carcinoge- nicity data on formaldehyde, and it argues in support of the central panels suggested in E.R. 638 and the American Industrial Health Counczl's proposal. Although the Come

144 Settee endorses government-wide consistency in risk assessment, it is less sanguine concerning the prospects of a permanent arrangement for such centralized risk assessment as contemplated by these proposals, in which the idea of centralized assessment is inextricably linked to extraragency separation. The Committee concluded that extra~agency separation would have disadvantages that would offset any advantages. The Committee did find, however, that agency scientists could collaborate to perform joint risk assessments on an ad hoc basis. Because agency scientists would perform an assessment, such an arrangement would avoid most of the drawbacks of extra-agency separation. The Committee looked at the Panel on Formaldehyde as an example of a centralized assessment group. In the Committee's view, the Panel functioned well and produced an assessment that has been accepted by the scientific community. The Panel's assessment has not produced parallel regulatory action among the agencies, and the Committee observed that similar r isk assessments should not necessarily lead to similar regulatory decisions, which reflect consid- erations that often justify different risk` management responses. USE OF SCIENTIFIC KEVTsW PANELS l 5. Independent scientific review o£~ - assessments improves the scientific ~ual i=.,r of the assessments and strengthens them against later challenge Agencies and programs with mandated peer review panels, such as EPA's Office of Pesticide Programs, which is required to submit to a Scientific Advisory Panel propose sals to cancel or restrict pesticide use, produce final risk assessments in support of regulatory decisions that are generally of high scientific quality and are accepted by the pu}:'lic and the regulated parties. In contrast, the Committee found several cases in which mechanisms for peer review could be markedly improved: OSLO, which uses public comments to ref ine its risk assessments, rather than formal peer review; NIOSH, which has not had a mech- an~sm to ensure that reviewers' comments are given appro- priate consideration; and FDA's Bureau of Foods, which uses ad hoc panels to review its assessments (a procedure that unfortunately can be circumvented).

145 · ~~ d~B . mechanisms to maintain the independence of their members appear to be the most useful review bodies. Continuity and independence of review panels help to ensure that such panels are sensitive to regulatory needs while retaining the necessary scientific objectivity. Examples of standing committees, such as the Scientific Advisory Panel in EPA, support this perception. Con- versely, the Committee observed that short-lived or ad hoc groups, such as the Subcommittee on Airborne Carcinoma gens, often do not have sufficient tome to develop a working relationship among panel mashers and that much of the time allotted to review is actually spent in clar~fy- ing individual versus panel viewpoints and understandings. Similarly, an ad hoc panel may not clearly understand its role in relation to the regulatory process. Thus, stand- ~ng panels appear to be of greater value to the agency than ad hoc committees. Furthermore, the existence of a standing panel might encourage an agency to seek its advice more frequently. Because it is important for review committees to be free to express their scientific judgments without concern for regulatory implications, panels that are formed in a manner that neither compromises nor appears to compromise their independence are more likely to improve ultimate risk assessments. The Committee observed that several review panels used by EPA already have a nomination process Mat places the responsibility for developing a slate of possible panel members outside the agency. Although the EPA Administrator maces the final selections of panel members, Me fact that nominations come from outside the agency emphasizes the intent that EPA panels be independent and as free of agency influence as possible. A related point is that membership on EPA panels, and in fact on most review panels used by the regulatory agencies, rotates; members are usually selected for staggered, f iced terms "generally 3-4 years) . This rotation itself reduces the likelihood that members will develop an institutional b' as. · Review Panels are best Qualified to Dive scien- t~fic advice when they are composed of scientists who are highly knowledgeable in the aPpronr iate disciplines. For carc~nogen~city r Ask assessments, for exhale, some relevant disciplines would be toxicology, pathology, biostatistics, chemistry, and epidemiology. The Com

146 mittee believes that professional or organizational affiliation should not be used as a primacy criterion in the determination of the makeup of a particular panel. That is, in contrast with the advisory panels used by 05aa, which are constituted to reflect balance among different aff il iations and presumed biases, the Committee believes that scientific competence must be the primary factor determining panel membership If review panels are to be asked to give their advice on the scientific aspects of an agency's risk assessments. }however, the Committee notes that panel meters who understand the policy in~pli- cations of their scientific judgments are more likely to be helpful to an agency's assessment process and that an attest to balance viewpoints of scientif ically qualif fed panel members may increase a panel's credibility. · Review unreels will be ~,3= the _ announcement of the agenc~r's intended regulatory actions, except in cases of emergenev. The Committee believes that review panels serving regulatory agencies should serve in an advisory capacity. That is, the judgments of a panel should not be binding on the agency. Nevertheless, the Committee also believes that the authority of agency review panels should be such that agencies must demonstrate that adequate consideration has been given to reviewers' judgments, and prior consul- tation with review panels helps to ensure this. Because announcements of intended actions or proposed regulations must be thoroughly developed and substantiated, review at the time of announcement or later is likely to be too late to influence an agency; although the regulation is only proposed, the decision of whether to act has, for all practical purposes, already been made. In the Comm~- tee's judgment, exceptions to this idea of prior review are appropriate in the case of emergency actions, such as suspension of pesticide registration. Risk assessments supporting such actions could be reviewed after the announced action. . Independent panels with authority to review risk assessments for all agency regulatory decisions, including decisions not _ =~ ~rounds. Panels with the authority to request the review of any agency risk assessment supporting a particular regulatory

147 decision will have a greater impact on agency decision- making. For example, if a panel can review only assess- ments referred to it by an agency, some agency decisions might not benefit from independent review of their screw: tific basis. This is especially likely if an agency has decided not JO regulate. Such a decision may have con- siderable impact and should receive the same careful review as decisions to regulate. In addition, panels with the authority to request reviews can respond to suggestions for review from the public. . Although most requirements of the Federal Advisorv Committee Act are salutary, others mav inhibit agency use of review panels. The Committee believes that most provisions of the Act are beneficial and endorses such provisions as the requirement that advisory committees meet In public and provide advance notice of their meetings. However, the Act does Impose requirements, some burdensome, for agency- created bodies that meet the definition of advisory committee. Notably, the Act requires that an advisory committee be formally chartered by an agency head and approved by the General Services Administration. This procedure has often proved cumbersome. Some agencies, such as FDA, lack independent chartering authority and thus require approval at the departmental level. In addition, procedures used by the General Services Adminis- tration for screening new committees have often imposed long delays, sometimes inspired by political concerns about committee membership or by resistance to the crea- tion of new government "agencies. These legal require- ments of the Act have caused some agencies to seek other ways of obtaining the views of scientific experts, especially when the issues involve single chemicals or tests. In such cases, regulators often confine their consultations to gov~r~ent scientists, who can be. accessible immediately and, if necessary, for extended per iods. . Written reviews help to ensure agency consideration of scientific criticism. A summary of a panells review that is transmitted in written form and made available to the public will help to avoid confusion and to ensure agency consideration of the panel's comments. As mentioned earlier, in the absence of adequate mechanisms to ensure agency cons~der- ation of reviewers' comments, the comments might be

148 ignored, or the public might perceive that they are ignored. Putting its summary in writing should also ensure that the panel states its findings clearly and make it more likely that the agency will interpret its comments correctly. O~;R OBSERVATIONS 6. Preparation of fully documented written risk assess- ments that exPl~citlv define the iudq~ents made and attendant uncertatrti~.~.~y~f]d =~ _!~:Y_: dec~s~onrmakinq process and aids the review Process considerably. When a fully documented written risk assessment is not produced before an agency's decision to regulate or not to regulate, it is difficult to understand the process by which an agency made its assessment. The Committee believes that the creation of such a document encourages public understanding of and respect for agency procedures and provides a basis for review by a scientific advisory panel. Furthermore, a detailed risk assessment document that clearly identifies the inference options chosen in the assessment and explains the rationale for those choices will help to maintain a sharper distinction between science and policy in the assessment of risk and will guard against the inappropriate intrusion of risk management considerations. REFERENCES AIEC (American Industrial Health Council). 1979. Recommended Framework for Identifying Carcinogens and Regulating Thorn in Manufacturing Situations. EPA (Enviror~ental Protection Agency). 1976. Health risis and economic impact assessments of suspected carcinogens: interim procedures and guidelines. Fed. Reg. 41(102):21402-21405. EPA (Environmental Protection Agency). }979. Water Quality Criteria, Request for Comments. Fed. Reg. 44(52):15926. EPA (Envi~ormental Protection Agency). 1980. Water Quality Criteria Documents; Availability. Fed. Reg. 4 5 (231): 79350-79353 ~

149 Food Safety Council. Social and Economic Committee. 1979. Principles and Processes for Making Food Safety Decisions. 54 pp. H~gginson, John. 1982. Report of the Scientif ~ c Workshop on the Critical Evaluation of Proposals by the American Industrial Health Council to Strengthen the Scientific Base For Regulatory Decisions. Rantrowitz, Arthur. 1975. Controlling technology democratically. Am. Sci. 63:505-S09. NRC "National Research Council). 1981. The Bandling of Risk Assessments in ARC Reports. A Report to the Governing Board, National Research Counci}, by the Governing Board Committee on the Assessment of Risk. 19 pp. Office of Science and Technology Policy. 1978. Identification, Characterization, and Control of Potential Human Carcinogens: A Framework for Federal Deciszon-Making. O.S. Congress. 1981a. Omnibus Budget Reconciliation Act. {P.L. 97 - 35, August 13, 1981) 15 U.S.C. 2077. O.S. Congress. 1981b. Bouse of Representatives, Committee on Agriculture, Hearings on the National Science Council Act, B.~. 638. Bearings Before the Subcommittee on Department Operations, Research, and Foreign Agriculture of the Committee on Agriculture, June 23, 1981. U.S. Congress. 1981c. Bouse of Representatives B.R. 638. National Science Council Act 97th Congress, 1st Session, 1981. O.S. Congress. 1981d. Senate. S. 1442. Food Safety Amendments of 1981, 97th Congress, 1st Session, 1981 .

Next: 4 Recommendations »
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!