Click for next page ( 104


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 103
103 FEDERAL RISK ASSESSMENTS FOR POlENl:161 CARCINOGENS: AN EMl?IBICAL REVIEW Robert I . Fie Id Lawrence E. McCray INl:ROOlJCT ION There is danger that discussion of federal agency performance in assessing risks are based on anecdotal and inaccurate conceptions of actual agency practices. The evaluation of ways to improve risk assessment is necessarily based on some notion of what a ''typical" agency risk assessment is like, and wit! not be helpful if the evaluator is misinformed about current practice. Me purpose of thin paper is to summarize available empirical information on the nature of the policy problem of chemical carcino- genicity and the federal regulatory response. In general, the results are somewhat disappointing: ~ detailed empirical documentation of risk assessment practices in federal agencies would be a massive undertaking, which perhaps helps to explain why none has yet appeared. Some literature does exist on discrete aspects of the area, however, and a partial picture can be assembled. We have attempted to provide below what objective answers exist to basic questions concern- ing risk assessments. The ques tions include: How big is the overall regulatory problem? One often hears the plaintive remark that "everything seems to cause cancer nowadays, " but the number of chemical regulations that actually reach national head- lines remains relatively small. How many suspected carcinogens are there, and what is the state of scientific knowledge about them? To regulates. and how much? What legislation governs the regulat ion of potential carcinogens, and what agenc ies and programs implement there laws? Has the government, as some have said, rushed to ban all suspect chemicals, or has it, has others have feared, moved only deliberately after it has assembled substantial proof of human hazard? - NOTE: This paper was original l~r prepared for the use of the National Research Council's Committee on the Institutional Means for Assessment of Risks to Public Health. It is not intended to present independent positions or interpretations on scientific or policy matters. It does not necessarily reflect the judgment or position of the Committee or the National Research Council. It has not been subjected to the internal review procedures chat apply to reports prepared by NRC commi ttees .

OCR for page 103
104 What i ~ ~ ~~ Discussions of risk ~ ~ assessment often assume that federal risk assessment as a we. t_ characterized, routinized and homogeneous process. Is the assumption accurate? If not, why not? KNOWLEDGE ABOUT CHEMICAL CARC INOGENS ANl) THE FEI)E RAL REGULATORY RE SPONSE One American in four will contract cancer during his or her lifetime, and the weekly death toll from the disease exceeds 1000. Current scientific understanding of cancer incidence leads to a conclusion that a large faction of these cases could have been prevented if the causative agent could have been identified and public exposures reduced or eliminated. The actual proportion of preventable cancers is still being discussed, with figures as high as 90: suggested; the UOSo Government' s Second Annual Report on Carcinogens more cautiously states that "many scientists now believe that about one-third to two-~hirds of al 1 cancers are agents contained in the air, water, food, or soil'' (NTP9 198L). What We Know About Chemicals and Cancer Me basic problem for public policy, of course, is that current scienti fic theories of cancer do not permit a definitive identi- fication of general classes of chemicals that cause cancer, and, accordingly, this leaver a very large number of chemicals to be assessed individually. The Chemical Abstract Service of the American Chemical Society lists well over 4 million known substances, with the number increasing by an average of 6,000 each week (Maugh, 1978~. As of this time, it is thought that as many as 63 ,000 chemicals were "in common use" (Maugh, 1978) and EPA estimated that as of 1979, 449O00 were used commerc tally (Roderick, 1981) O About 559 Q00 synthetic chemicals were produced and used in significant quantities as of 197S, with 1, 000 new ones being introduced each year (Ames, 1979) . Accord- ing to one estimate, the manufacture of synthetic organic chemicals has doubled every seven to eight years (Davis and Magee, 1979~. The number of chemicals under the jurisdiction of any single federal program may be very large. For example, EPA estimates that there may be as many as 1,500 different active ingredients in pesticides O FDA extirpates that about 4,000 active ingredients are used in drugs, and about 2,000 other compounds are used for purposes such as promoting stability and restricting bacteria growth. In foods, there are thought to be 2 g 500 additives used for nutritional value and flavoring and 3 ,000 used to promote product life (Maugh, 1978~. Another 12,000 chemicals are indirect food additives (Flamm, 1981 ) .

OCR for page 103
105 The evidence on carcinogenicity for any one of these chemicals is likely to be highly limited. The most direct type of evidence is that from epidemilogical studies of the effects of exposure of a chemical to humans. However, epidemiological studies of cancer are expensive, time-consuming, and fraught with difficulties--not the least of which is the problem of establishing the actual existence and level of past human exposures to any particular suspect chemical. As a result, direct human evidence is available for only a few cne~cal~; in fact, the International Agency for Research on Cancer lists fewer than 60 chemicals as having been adequately evaluated as cancer hazards in humans (IARC, 1980~. IARC lists 32 chemicals and 4 industrial processes that have been associated with cancer through analysis of human data (Davi~, 1981) 0 At least fragmentary evidence is available on other compounds; in fact, 82 chemicals are counted as showing "some epidemiological evidences of carcinogenicity (Maugh, 19783. Lee limitations on direct human evidence necessarily throw the spotlight on animal testing. While reliance on bioassays can be used to inform policy decisions on many more chemicals, the overall supply of test data cannot be characterized as rich. It has been estimated, for example, that only about 7000 chemi cals--less than 20: of the number of chemicals said to be in common commercial use--have ever been tested. Of these, only 1500 are said to have been tested under what are presently considered to be scientifically adequate conditions (Toxic Substances Strategy Committee, 1980), although one NCI official estimates that only 3500 of the 7000 are "completely inadequate" (Maugh, l978~. Tore striking is the fact that the volume of test results is not expanding very rapidly. I ~ is estimated that only 100 to 300 chemicals are newly subjected to animal bioassay annually (Maugh, 1918 ~ . This number is limited somewhat by the current expense of lifetime animal exposure studies--~n the range of $300,000 to $500,000. In addition, the total volume is limited by the total available supply of toxicotogise~, pathologists, and lab facilities, which is said to permit no more than 500 new bioassays each year (Maugh, 1978~. How many chemicals have been identified as carcinogens from the tee tiny that has been completed? There is no simple answer to this simple question, and a review of published estimates demonstrates that estimates depend heavily on the assessor's standard of proof. It is estimated that 1500 (a little over 20%) of the 7000 chemicals tested show at least some positive results; if one-half of these are assumed to reflect adequate test methods, 750 chemicals can be counted as animal carcinogens (Maugh, 1978?. This number is consistent with the Toxic Substance Strategy Committee' s ( 1980) estimate that 600-800 compounds show "substantial positive evidences in animal tests. IARC's estimate for the number of "suspect human carcinogens" based on animal studies is about 300; while the number of "carcinogens'' is less than 900 (Tomatis, 1978) . OSHA's controversial classification scheme for carcinogens reflected results from short-term tests as well as longer-term bioas says . OSHA counted 961 "proven" c arc inogens, which

OCR for page 103
106 it said has produced either two positive bioassays or one positive bioassay and two or more positive short-em tests. OSHA listed another 196 ''suspect" chemicals 5 which had produced one positive bioassay or positive short-term test (Maugh, 1978~. What do we know about exposure patterns for these suspect carcinogens? Other than the obvious conclusion that, sincemany are produced commercially, workplace risk of exposure to many must be assumed, we found no empirical Hungary. NIP analyzed where exposures to S~ carcinogens substances are found (National Toxicology Program, 1981) . While not a comprehensive died, the compounds studies are described as those thought to have the strongest positive results based on the findings of IARC, the ~P/NCI Carcinogenesis Bioassay Program, ant various agencies. Two occur naturally (aflatoxin and cyeas~n), and two are sources of major exposure in food and cosme~cics (saccharin and safrole). Where are eight pesticides and 14 pharma- ceuticals among the S8. Me remaining 62 include industrial chemicals, miscellaneous chemicals and analogs, industrial processes, and incus t rial byproduc t s . The Regulatory Response Authority to restrict public exposures to toxic substances is distributed among 24 statutes that are administered by regulatory agencies (Toxic Substances Strategy Committee, 19801. Although the oldest of these is the Federal Food, Drug and Cosmetic Act of 1938, the laws are remarkable for their relative regency; on average, the 24 laws have been on the books for only 16 years in 19830 For potential carcinogens g the ma jor regulatory programs are concentrated in EPA, FBA. and OSHA. The sheer number of chemicals in commerce gives any one regulatory program an enormous queue of chemicals to review for regulatory action. EPA has subjected 3,500 chemicals to some sort of ac tive review as shown in i ts _ ~0 . There are so many suspects in the workplace that ~~ has been e~matec plan it would take OSHA over 100 years to regulate all the known hazards on a substance-by-substance basis (Davis, 1981) . tie discovered no convincing attempt to account for the federal government 'a cumulative disposition of ache chemicals on any of the various lists of suspected care inogens. One account (Roderick, 1981 ) reports that the U. S i has regulated only ten of the approximately 30 agents listed by IARC as carcinogens from evidence in epidem ological studies, and that only ~ have been regulated that appear on an IARC list of lit chemicals for which bioassay results indicate carcinoma genicity. While this record may give credence to a theory that the government moves slowly on potential carcinogens, in fact many more than 18 chemicals Zaire been controlled g and it has proven difficult to compile definitive lists of government actions that are based on cancer hazard.

OCR for page 103
107 ~ review of federal carcinogen regulation reported 43 subs tances that had been regulated as of 1978 as recognized carcinogens (Roderick, 1981~. Six different statutes provide authority for these actions, arid 43 rules were issued, although not in a one-to-one cor- respondence to the chemicals . ~ cons iderab le amount of interagency overlap is evident, with asbestos being regulated under three programs, viny! chloride under five, ant DOT under two. Another, more comprehensive study found a total of 102 substances that have either been regulated, been proposed or cons idered for regulation under several ~ but not all) sta~u~ce~ (OTA, 1981) . A summary of the status of these chemicals, prepared in 1980, is presented as Tab le 1. I t reveals two dominant trends in the distribution of federal efforts: I. EPA has had the widest experience, FDA and OSHA somewhat less, and CPSC the least: 55 were addressed under the clean water program at EPA 29 were addressed under the clean air program at EPA tS were addressed under the pesticides program at EPA 2 were addressed under the drinking water program an EPA 24 -were addressed under the food program at FDA 18 were addressed by OSHA - 5 were addres sed by CPSC 2. There is a fair amount of interprogram overlap and some interagency overlap: The Art of Risk Assessment there were 152 agency actions on the 102 chemicals; about 40% of all chemicals are subject of action under two or more statutes 39 are addressed by two or more programs 14 are addressed by two or more distinct agencies the FDA food program overlaps very little (only twice) with other federal programs CPSC almost always addresses substances of interest to other agencies, (four out of five instances), and OSHA often does, (11 of 18 instances) . FE1)ERAL RISK ASSESSMENT PRACTICES - For all of the chemicals that are subjected to regulatory review of one kind or another, some sort of risk assessment must--by definition-- have taken place. In some cases this may cons titute a full-blown analysis of the subs tance including the generation of a quantitative conclusion and a formal report, while in others it may merely involve an informal, qualitative judgment that the presence of carc~nogenicity is or is not established.

OCR for page 103
108 TILE 1 ~st~e ~ulat" se Bend US Ve~e ~ C8 URC C - & - A ~A m" C^& _I~ a i ~'c ~ Me _ S S ~1~ ~ ~ ~" Cal HEM s Cock t" S Came S Cam - m count ~ Cr~oo~le S C^to~o~ (a tn~o~. n. - Arms - - - , Co" on ~~e (I - ~'c I. ~'1 ff Cool. C - ~ - "C~. "C~.tO talc ~ do. 11 D.C ad ~. t2 # "C~ "C Y.~ ~ 1 "C Y - ~ ~ ~ We Y - ~ - . 10 I, cot mC~Oo~I 1 1~ di~th S ,~ at - Byte D~u~n_ero~ {DES) ',~ L D~ L E;r~ - t~ - e ~t~e 1 ~t~ on~ fDsC~o.2 f9eC v~o' - ao. t Fo~.O - ~ - S;~~ n acmo ~1 Source: Assessment of Technologies f or Determining Cancer Risks f rom the Environment, OTA, June 1981.

OCR for page 103
109 TABLE ~ - Continued , ~ - see ~ act " - , ~stJ~e# RMath ~ ~c~no~ane Under Various ~ (Cygnus - 1 ewe Eva C" ~ ail SOWN Flaw OS" aim. (aim ~h" S I,' ~~ - e Q~ - L ~~em~ 2~ - o t 14'cae. If_ Hnwn~ L ~~' S to S Nook Alp S am~t~oso~U~tay~c~ ~~~ S - ~~t n.~,ooylemu'. S Afros - ~ - tnytume (LIEU) 5 Caruso.- ~n~" - ~`4 to 0~~ at cur ~000 l~acr~~Po~~~oo~ ~J) S ~cn~onnet" =~s (PC88; topic In Couth retry t "~o. .~.~een~ _ - ~. ~m l.t~^~ecn ~to, - ~1 t - ~ 1. If_ ~ ~ I. _ _ ~ s _ ~ dS) 1 ~ ~ ^~.~ _ _ ~ 71~ oHm ~ "'le (tle~ ~~g _ C ~ Y~ - - ~ ~ - - - ~o~toce

OCR for page 103
110 A brief examination of two draft documents describing risk assessment at EPA reveal on the one hand the extensiveness of the assessment, and on the other hand the infrequency with which detailed formal analyses are perf armed. A review of 155 chemicals under current examination by two or more EPA offices shows that all but four have undergone risk assessments. The great majority, moreover, have been assessed more than once, with 53 barring been studied from two to seven times. The assessments vary from formal analyses to shorter reports by the Carcinogen Assessment Group (CAG) to brief summary risk assessments. In all 9 only 18 of the 155 substances were sub jected for formal risk assessments, while 71 went through preliminary or summary risk assessments. A listing of the activities of CAG itself, reveals that it has prepared over 200 reports on about 110 chemicals or classes of chemicals. A second s tuty examined the analyses and documental ion prepared by the agency in the regulation of three carcinogenic chemicals: cadmium, trichloroethylene, and arsenic . For cadmium, a total of 21 agency reports were created, of which nine dealt primarily with exposure assessments and seven with health and environmental effects. For trichloroethylene, eight reports were involved, with fire dealing wi th exposure and one wi th heal th and environmental consequences . For arsenic, there were 16 reports, of which 12 analyzed exposure and seven health and environmental ef fee ts . If the EPA experience is typical, federal risk assessment activity is neither uniformly rigorous or uniformly cursory. Examples of the two extremes may be seen in EPA's pre-market notification (P~N) pro- cedures for pesticides and FDA's new drug application (NDA) process. EPA receives large numbers of notifications each year, and mus t make its decisions within 90 days of receipt (although this statutory deadline is not invariably met). A typical notification involves the submission of very little toxicological data, and the agency is often left to infer risk levels from a chemical 's physical structure, chemical propertied and from the notifying corapany's estimates of future production, vol~e, and uses. EPA rarely demands further information. Hazard assessments are done for all substances as notifications come in, but detailed quantitative risk assessments are only performed when strong positive results are indicated. At FDA, on the other hand, large stacks of toxicological information are ~ub- mi~cted with all NDA's. Exposures for drugs are, in comparison with PMN chemicals, easy to characterize, malcing risk assessments much simpler. FDA generally takes about 20 months to analyze an NDA, and in the vast majority of cases, it requests additional data from the applicant (GAO, 1980~. The Diverse Functions of Risk Assessment Risk assessment plays dif ferent roles in the regulatory proces s . These roles fall into two general categories: priority setting and analysis of regulatory controls. The two roles imply distinct demands on risk as sessors .

OCR for page 103
111 Priority setting. Regulatory agencies typically have potential Jurisdiction over ~ large number of substances. Circumstances force them to allocate their resources to a few at a tome. Common sense and public opinion--if not their own policies--induce agencies to try to devote their attention mainly to the largest hazards. This allocation decision requires some sort of de facto risk assessment. Some notion of relative hazard -implicit or explicit, internally generated or imposed by outside groups--is necessary for this function. A part (and, for some critics, a major part) of the general criticism of federal regulation is that the agencies are not setting their priorities sensibly or systematically. In general, it appears that agency risk assessments for priority setting have been informal, and less systematic and visible, than for assessing regulatory controls. Agencies set priorities in two areas: regulatory screening and testing. Regulatory screening involves decisionmaking about which substances should be selected--ant often in what order--for serious formal regulatory review. Virtually all programs have this problem, although there is one important variation. Some programs cover a finite ant known set of chemicals that must be reviewed. Here, the order of the regulatory reviews is the key question, and the job of the risk assessor may be to help the agency implement a 'iworst-first" or another reasoned approach. Some such thinking, for example, is relevant to OSHA's decisions about which occupational standards (cotton dust, benzene5 etc.) it will review first. Similarly, EPA pesticides program has long had lists of "suspect" pesticide in- gredients, and it has had to decide which ones to formally consider for cancellation or for new controls. Other programs, most promi- nently those that must in effect grant official licenses for the production or use of new products or substances, are forced to categorize relative risk on a case-by case basis so they can decide which new applications to concentrate on. Examples include FBA review of new food additives and applications to EPA for federal registration of new pesticides. In both variations, however, the screening function is the same: some sort of relative rating or ranking of risk muse be accomplished, however imprecisely. Variations among screening efforts--even within a single regula- tory agency-are illustrated by three EPA programs. These cases dramas ize the range of agency control over the qual ity and quant i ty of data available for risk assessments used in screening.

OCR for page 103
112 Case 1: Premanufact~ring Reviews: Screening Based on Sketchy Biological Data EPA.'s Office of Toxic Substances is charged to oversee the manufacturing of new chemicals under Section 5 of the Toxic Substances Control Acto Section 5 regulations require pre- man~facturing notification (PMN) but do not require the manufacturer to perform toxicity tests. Consequently, assessors usually must screen in order to isolate the few chemicals needing detailed regulatory review from scarce data. EPA has had to rely heavily on analysis of structure activity relations (SAR) and mutagenicity data, when available, to do this screening. The agency reported it had considered drawing up risk assessment criteria for screening, but found that the limited amount and variety of information to be weighed in such decisions precluded their developing explicit criteria. Each chemical is considered on a case-by-case, necessarily judgmental basin. Case 2: Pesticide Regulation: Qualitative Screening Based on . . Agency Da ~ a Requ i remeet s EPA's Office of Pesticide Programs oversees the federal registration and reregistration of pesticides. In contrast to the PAN process, pesticide regulations require the manufacturer to perform a number of tests dealing with acute and chronic toxicity. In the screening process, each active pesticide ingredient, (and its metabolites or degradation products) is measured against a set of qualitative risk criteria, or "triggers.'' Specific criteria are detailed for carcinogenic, mutagenic, and teratogenic responses. If a pesticide reaches or exceeds these risk criteria, OPP shifts to a higher regulatory gear; it issues a "rebuttable presumption against regulation" and undertakes a more elaborate process to weigh benefits against risks . Case 3: Airborne Carcinogens: Screening Based on Quantitative - Data In some cases, agencies have fairly extensive quantitative data for a list of chemicals ~ but limitations on agency resources preclude regulating the entire liste Setting regulatory priors Sties may require a cl~emical-by-chemical quantitative comparison of the health risk. EPA's Carcinogen Assessment Group (CAG) reports that it has performed such a qualitative analysis for the Office of Air Programs to help it determine which air pollutants should be regulated fits to OSHA' s 1980 Cancer Policy has suggested a similar use for quantitative risk assessment to determine which chemicals in a particular category should be regulated first.

OCR for page 103
113 A second distinc t type of priority selecting involves the establishment of testing priorities for substances that lack adequate data to permit risk decisions. Risk asses~ment--formal or informal--is an inherent e lement in decis ions about such research/~esting priorities at such organizations as the National Toxicology Program, the National Center for Toxicological Research, and the research efforts of agencies themselves. Establishing regulatory controls. The other broad category of use for risk assessment is to help determination of what appropriate policy measures, if any, are required to protect public health. This application has received the most attention in public discussions of regulation and its deficiencies, and, in general, is the use for which the most formal versions of the art are found. Here, too, there are wide variations in what is expected of risk aseessors. One source of variation is the nature of the statutory direction to the agency on 'ROW it should sleigh various factors in reaching control decis ions. Tab le 2 summarizes the ten regulatory statutes administered by the four ma jar federal regulators: EPA, Fl)A, OSaA, and CPSC. Mere are shades of difference--soraetimes in different sections of the same s tatute--in the degree of protection required, and, more salient, in the relative weights that agencies are instructed to place on risk, control costs, and technical feasi- bility. This latter factor may be divided into three approaches. First, several laws require a balanc ing of costs and benefits. Such statutes generally call on the agency administering a regulatory program to weigh the benefits to be achieved through an action, including the reduction in public health risk against such coots as the ec anomie hardship imposed on those be ing regulated. On some instances, Congress has explicitly listed factors to be balanced in decisions, while in others, this approach has been read into risk legislation by courts in response to vaguer mandates specifying the reduct ion -of "unreasonab le" r i slcs . Examp les o f expl ic i t ba lane ing provisions are found in the pesticide law and the safe drinking water law, while examples of implicit balanc ing are found in the toxic substances law. (Details of these and other examples can be fount in Table 2. ~ The role of risk assessments under these schemes Is usually quite clear. It provides an explicit way for measuring, either quantitatively or qualitatively, the benefits that regulatory actions will provide. Second, some laws call for mandatory control techniques whenever hazard is affirmed. These include the outright ban of products under the De laney clauses in the food law, and the parts of the c lean air law that specify "an ample margin of safety" in emissions standards. This type of statute provides a need for the hazard identification phase of risk assessment, but since the control action is specified once the hazard is affirmed, the contribution of the dose-response information is less clear.

OCR for page 103
114 TABLE 2 Wok ~ - mono tat ~ - uIstb~ ot ~p~ to C~l~e 0_~ ~_~.n~ __ ~~ ma_ ~ ~ _ _*e 0~ _~4 c_ ~ ~ ~4 ~e. tl2 1CI_ ~ ~ ~ _o~ . . .l~ e - _. of moo to. an hCN. ~ ~ 1~ or an Is_ 1~ a __,er~ ~0 __, .~ Bet. lta~tt, c_ -_d~ a_ a To. A. . .ea_. ~\ - 0 "ucr I_ "C. Da~o "' ~_~ t_C ~ ~C _e oo~ ~U m4m ~' ~eo~u. s___ _~_ _~0 t~ d ~s o" __~ ___t ~o~o' ac~ ~_ u ~n~ _,~to~ _~ aaC-~~ as~~~ 1~ ~ WUtl~ 0' _~. ~e ~ee~ _wu~ ~~ _~ t~ ~ ~ a Source: Assessment of Te~hnolog~es f or Determ~ning Car~cer Risks from the En~rironment, OTA, June 1981 "C ~! 44' 4~; _e.~ ~' ~a ~.on eo.~rO; _~ce. o,stem or ~, - ,... ~UO~ 1. t ~s an ur, _ao~o. ~ - ~o ~h o' " - l, co - ~l~t ~o - ~con o' ~~-C to. ~ 'ce~otoo - . "U_n~e~ 80_ - ~eCt' ~0 ~ re" 10 I t" t 2~. h. mto c:~' mo eco~'c. "CICI. a" n ~ts no .

OCR for page 103
115 TABLE 2 - Continued We Len ~~r9 ton ~ 8088 - ~ ~~ t. =~ It d -I 0~ ~ ~~ ea_ ~~ as_ ~ "H_~_~. "h~_~_~ o,~ I. I~ - . a_ a. . . _or ~ c_ _...- 11~ - tam to_ Iota ~ ~ or ~~ a_~ tar a ~_...- ~ ~ - nut he_ TIC - ~ _ _ , __ ~ #I an_ ~ ~ - _ T~_e, ~ 1~_ _ - _ . ~ a_ ~ _a. . . nag ~ ~ _a~ an_ ~~ _as a_ ~ ~ ~ lo c_~. . . ~ ~ _~ ~ o~nc Cow _of _. SAC ~~1 7~_~ 2 In - am_ of mar ma_ ~~ __ - ~_~ ~-'~ al1 ~ ~ ammo ~ot~to~or ~ ~ _ __ _ `~ ~_. _ ~ ECU _ ~ ~ ~ a_ ~~ ~ _ sI~ tle.-- ~__~ d ~ _ I_ - ~ or ~ ~ _-- ~ - "e., - _ ~~ Id ~a" ~ en_ _~. U" ~ _ _t _t ~~ Ha_ . . . - _ ~ of_ ~ l At ~ __ d Cow__ ~ - _ . _ .. a_ no_ -I _~_ ret_ -~__ ~ _ 1 ret ~t~ev_ ___ cow to I_~ _~. __ _ . . -15 - C "C at of _. _ _ . . . ~15 Met mooing ~ _t act ~ Woo not at . . . us c_e- a_ on ups t_ of of, Gem ~ Or__ ha.-- ~ \10C - C. 2861 _ At o of a of l ts ~ ~ a" ad f~ only~ co~-- _ _I COO, ~l'1~" ~ tonic Welts 10 cow. town lurid ~ ~ 10~. Stud ~ to I. em_ t - ~ ~ ~ Bolt

OCR for page 103
116 Similarly, some statutes call for consideration of technological or economic feasibility as the only limiting factors on the control of risks. Examples are the sections of the clean air and clean water laws that specify "best available technology" or ''best practicable technology g t' and the hazardous substances section of the Occupational Safety and Health Act, which mandates the most stringent standards that are "feasible. " In these schemes, the identification of rinks is eke most important phase. Dose-response information can also play a rote, though, in determining the effectiveness of different control methods. Furthermore, under the occupational health law as intern preset by the Supreme Court, the agency has been encouraged to measure ricks to see if their diminution through regulatory actions is in rough accord with the costs of control. Some observers suggest that the importance of these statutory distinctions should not be exaggerated--that, as a practical maker, some sort of informal cost-benefit comparisons are necessary even if the statute seems to discourage formal quantification. The implication is that formal risk assessments could find practical use--either more or les ~ his ibly--in al ~ programs O There is also a practical difference between use of risk assessment in programs that involve pre-maricet approval of substances and in programs thee operate through other post-hoc mechanisms, such as environmental emission limits. A study of federal risk assessment practices prepared by Clement Associates found that this distinction was the greatest single statutory determinant of the way in which risk assessments were conducted (Cleancut Associates, 1981~. The most important effect of this difference may lie in the fact that pre- maricet approval programs, such as "hose for new human drugs and for pesticides generally empower the agency to require the su~omission of data to be used in a risk assessment, while other programs tend to leave agencies to fend for themselves in the acquisition of data. Implications. These varying functions place different strains on risk assessors: the consequence may be that a single risk assessment methodology may not be able to satisfy the different functions. For example, a risk assessment done for the purpose of establishing testing priorities may, appropriately, incorporate many "worst-case" assumptions where there are data gaps, because research should sen- ~ibly be directed at those substances showing the largest and most crucial gaps. However, such simplifying assumptions may be inap- propriate for risk assessments used to analyze regulatory controls, particularly where the regulator's job is partly to ensure that controls do not place 'wasteful strains on the economy. Similarly, for priority setting there is a premium on consistency across assessments, since the main point of the analysis is to make meaningful risk come parisons in order to direc t agency resources rationally. In Contras t,

OCR for page 103
111 consistency may be less crucial for analyzing risk to design regula tory controls; ache same general "rules of thumb' that may be reasonable for priority setting may Cadre to yield to more sophis- ticated and detailed scientific arguments when a substance's commercial life is at shake, and an agency's decision will have to stand up in court . Furthermore, availab le resources--and the resultant analytic care--devoted to a risk assessment for deciding regulation policy for a single substance is likely to be much greater for analyzing control actions than when priorities must toe set across a larger set of substances. Procedura 1 Factors ~- Federal regulatory actions can take two forms: formal and informal. These distinctions are set forth in the Administrative Procedures Act. Formal rulemaking involves an cour~c-like adjudicatory hearing at which competing claims are heard. An administrative law judge presides 9 and subsequently issues a decision based on the evidence presented, both oral and written; this decision is the teas is for the agency's final action. As an example, cancellation of registration for pesticides under FIFRA must follow a formal hearing. In informal rulemaking proceedings, comments from outside groups are gathered and analyzed, but an adjudicatory process is not used. The agency must respond to all substantive comments in the preamble to its final rule. The ultimate arbiter of the adequacy of risk assessments in any proceeding is the court Chat scrutinizes them on judicial review. This means that one primary purpose of assessments is to satisfy judges that the agency has acted reasonably and in accordance with applicable statutory criteria. While the record from a formal adjudicatory hearing can help to convince a court that all relevant factors were thoroughly considered, either type of rulemaking action requires a carefully documented analysis, so that the actual perfor- mance of risk assessment is likely to be similar in both cases. The influence of procedural fac~cors on the time involved in issuing a final rule is illustrated by Table 3. All of the actions listed included hearings The total length of the proceedings from initial action to final standard ranges up to seven years, but the average is in the range of five. It is clear that formal proceedings involve considerable commitments of time. It is likely that judges will expect thorough risl< assessments as part of the support for agency actions under these circumstances. However, informal rulemaking also requires procedures that take time: Figure 1 shows the typical phases in Nlemaking at EPA, which consumed about two years .

OCR for page 103
118 TABLE 3 so o d o o is_ 1 111 1 'C ~ O la _ e, - O a O C o - C ~ O d a; me - S - o 1: O d ~ d _ Cal l_ ~ 0 0 t_ _ V O ~ O

OCR for page 103
119 TABLE 3 - Continued - to I, o - o c o 1 1 . s d ~ C C O ~ e ~ 0 ~ _ o o a~ e C _ ~ ~ _ __ 64 ~ ~ ~ _ _ ~ ~ _ ~ rat _ 0 ~ ~ _ 0 0 or ~ ~ V e ~ 0 _ C, ~0 DC _ O ~4 ~ _ ~ _ ~ r. 0` _ O _ ~ _ _ IC `0 o - ~ c _ O ~ _ _ C y - - - _ rat er' ~ _ O . 1 Cut - C

OCR for page 103
120 FIGS 1 WEEK O - ~ Administrator Approves Dc~lopma~t and W - ring Group Formation 44 ~ 45 Steering Commin" Clearance ~ Begin Interagency Review 50~ ~ Begin OMB Coordination 53 - End OM8 Ccord;~tion 59 ~ ~ Propose in Federal Register 65 ~ a , ~ End Public Comment Period 91 -a S - Bring Commin" Clearance 94 - - Begin Interagency Review 98 ~ - Begin OM8 Coordination 101 - - End OM8 Coordination 108 - 4 Proton Promulgate in Federal Register J ~ _~' ' Rules Final ' Rules Source: Decision Making in the En~rironmes~tal Protection Agency 9 Vol O IT 9 ARC 9 1977 .

OCR for page 103
121 Secular Trends Many government regulators see risk assessments role in regulatory tecision==t~ing as increasing. At OSHA, for example quantitative risk assessments had been he ld until two years ago to be inherently unreliable as well as unnecessary under the statutory provisions calling for the regulation of hazardous substances (Maugh, 1978~. The Supreme Court 's decision in the benzene case required the agency to account for the benefits of its actions, however, and OSHA responded by indicating it would include quantitative risk asses~men~cs in its regulatory process. Furthermore, NIOSH has recently begun to take more of an ineeres~c in such analyses as part of its research efforts. Another example is FDA' s Bureau of Foods, where risk assessment has gained importance as the policy of banning substances in response to any evidence of a carcinogenic risk has lost factor. Quantitative risk assessment is increasingly being used as a means of deciding where the most s~shstant~al hazards lie. A final example is the control of toxic substances at EPA, where an increasing number of statutory mandates enacted over the past decade have explicitly called for a balancing of risks ant benefits in a way that previous enact- ments have not. Among these are FIFRA, passed in 1972, and TSCA, passed in 1976. In addition to these specific agency examples is the direction of federal regulatory policy in general. Under Executive Order 1229 1, regulatory agencies are required to establish that significant actions will involve benefits to society that outweigh their coeds. This will encourage the use of quantitative risk assessments, so that benefits can be measured. Several factors have emerged over ache past decade that should promote risk assessment, particularly quantitative assessment. Their influence can be particularly seen clearly in Congress's choice to include balancing criteria in more recent rink legislation (Field, 1981 ) . Me first factor is the effect of changing economic conditions. As the financial capabilities of American industry to play a large role in eliminating risks have become more strained, pressures for a precise accounting of risks and bene fits have grown. Second, changes may have occurred over time in the incremental benefits that new risk regulations can achieve. Initial interventions have in many cases produced substan~cial results, so that further efforts may be leading to a point of diminishing ret.. Third, there has been a large expansion in recent years in the amount of data available on specific hazards, making The presence of rinks easier to detect. Test have revealed a greatly increasing number of substances that are dangerous . As a result, the development 0 f echoes for sett ing priorities in risk control and for comparing risks to benefits has become more highly valued. Finally, there have been advances in the methodology of risk assessment, and the public and the scientific community seem to have fewer reservations about it ~ use .

OCR for page 103
122 Cons is rency in Fedesa 1 Ri sk As sessment One of the main public concerns about risk assessment is that it is not performed consistently over time or across agencies. As out fined, there certainly are subs partial variations in the ~ ice, coverage and form of risk assessment. Nether the assessments vary substantively-. that is, whether the use of inconsistent approaches leats to different conclusions from the same scienti fic data--is less c tear. We found only one systematic attempt to address this question, and white there are interagency guidelines for cancer risk assessment, in practice it is difficult to establish whether they are followed. Those who have looked at specific assessments report that it is often difficult to trace the assumptions that were used in the analysis--so the necessary first step in establishing the degree of consistency across assessments in usually lacking. A report on risk assessment as practiced by EPA and other agencies (Clement Associates, 1981), presents several conclusions on consistency in the techniques uses. Hazard assessments for carcinogens are fairly uniform as compared to such analyses for other health effects. However, many differences in procedural details exist, the most important ones being in the choice of animal data as the teas is for extrapolation, the use of correction factors for partial lifetime dosage, and the use of animal-to-buman scaling factors. These variations can, in some instances, lead to differences in estimates of t`uman risk of ten times or more. There is variability in the ways char the results of risk assessments are presented by different off ices . The Clement Report reached a number of conclusions on ocher agencies, including FDA, CPSC, and OSHA. The see sort of differences elect were found to exist among EPA's various offices were also noted among these agencies. The major aspects of quantitative risk assessment are fairly well standardized, but numerous differences do exist. The IRLG guidelines are reported to be given uneven use, and noncarcinogenic risks tend to be more variable in their assessments than carcinogenic ones. Aside from these guidelines, there is no major mechanism to ensure consistency among the agencies.

OCR for page 103
123 REFERENCE S Ames, B.N. "Identifying Environmental Chemicals Causing Mutations ant Cancer. '' Sc fence, To ~ . 204, p . 587, 1979 . Barna-I"loyd, G. " 'Environmentally' Caused Cancers." Science, Vol. 202, p. 469, t978. Clement Associates, Inc. "Review and Analysis of Hazard, Exposure, and Rick Assessment as Practiced by EPA and other Federal Regulatory Agencies" (mimes, 1981~. Davis, D.L. "Cancer in the Workplace--lihe Case for Prevention." Environment, Vol. 23, No. 6 (July/Augu~t 1981), p. 25 Davis, D.L. ant B.ll. Hagee. "Cancer ant Industrial Chemical Production. " Science, Vol. . 206, p. 1356, 1979. Epstein, S.S. The Politics of Cancer. San Francisco: Sierra Club Backs, 19 7 ~ . Field, R.I. "Statutory ant Institutional Trenda in Governmental Risk Management: The Emergence of a New Structure (Mimeo, 1981~. Flame, W. Gary. Remarks to the Committee on the Institional Means for Assessment of Risks to Public Health, October 1981. General Accounting Office. FBA Drug Approval A Lengthy Process (GAO, Report lIRD-80-64, 1980 ~ . Harrison; D. "Cost-Benefit Analysis and the Regulation of Environmenta l Carcinogens . " In Nicholson, W. J., (Et . ~ "Management of Assessed Risk for (:arc~nc~gens." Annals of the New York Academy of Sciences, \?ol, 363. New York: The New York Academy of , . Sciences, 1981. IARC . "An Evaluation of Chemicals and Indus trial Processes Associated with Cancer in Humans. " Cancer ~search, Vol. 40. No . 1. Maugh, T. H. "Chemicals: How Many Are There?" Science, Vol. 199, p. 162, Jan. 1978. Maugh, T.H. "Chemical Carcinogens: The Scientific Basis for Regulation." Science, Vol. 201, p. 1200, Sepal. 1978. National Toxicology Program. Second Annual Report on Carcinogens. U. S. Department of Health and Human Services, Public Health Service, 1981.

OCR for page 103
124 Of fice of Technology Assessment. Assessment of Technologies for Determining Cancer Risks from the Environment. Washington, DeCo U.S. Go~rernmer~t Printing Office ~ 1981. Roderick, B. "An Alternative to Risk-Benef~t Analysis in Government Becision-Haking About Chemical Carcinogen.' In Nicholson, W. J. Op. cit., 1981. Selikoff, I. J. "Carcinogenic Risk Management in the United States." In Nicholson, W. J. Op. tie., 1981 Tomatis, 1~., et al. "Evaluation of the Carcinogenicity of Chemicals. " Cancer Research, Yol . 3S, 197S, pp. S77 f f . Toxic Substances Strategy Committee. Toxic Chemicals and Public Protection. (U. S .G.P .O., 1980~ . Wo ~ fe , S a Me "S tandards f or Care inogen~: Sc fence Af fronted by Politics." In Hiatt, H. H., J. E). Watson, and J. A. Glisten (Eds.) Origins of Human Cancer. Book C: Human Risk Assessment. Cold Spring Harbor: Cold Spring Harbor Laboratory, 1977 .