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.
FORMALDEHYDE The Conquer Product Safety Commission's Risk Assessment for Formaldehyde William M. Stigliani A. BACKGROUNl) AMB CONTEXT 1 ~ Describe the chemical and i ts uses . Formaldehyde is a colorless, flammable gas with a strong, pungent odor. It can form explos ive :aixtures wi th air and oxygen. As an important industrial chemical of ma jor commercial use, formalde- byde is found throughout the environment . In outdoor air, it can originate from many sources such as incinerators, photochemical smog, and engine exhaust. Atmospheric levels of formaldehyde have been reported to range from less than 0.005 ppm to 0.06 ppm near industrial outlets or in areas of heavy smog. Corkers who smoke are exposed to additional levels of formaldehyde, since cigarette smoke contains an mucl, as 40 ppm of formaldehyde by volume. Thus, an individual who smokes a pack of cigarettes ~ day would inhale 0.38 ma, whereas occupational exposure to formaldehyde at 3 ppm could result in a daily intake of 29.0 ma. Production and Uses l Formaldehyde is usually manufactured by reacting methanol vapor and air over a catalyst (chemical initiator). This results in formaldehyde containing trace amounts of methanol and formic acid. Formaldehyde is sold mainly as an aqueous (water-based) solution called formalin, which is 37: to 50% formaldehyde by NOU: Thin case s Judy describes assessment procedures and summarizes issues and interpretations raised by others, but it is not intended to present independent pos itions or interpretations on either scienti fic or policy matters. The case has been reviewed by individuals outside the study project who are directly familiar with the federal analyses ant decisions described; however, responsibility for the paper rests with the author, and it does not necessarily reflect the judgment of the Committee on the Institutional Means for Assessment of Risks to Public Health or the National Research Council. It has not been subs jected to internal review procedures that apply to reports prepared by NRC committees. 1
2 weight. It is also used in its solid form as paraformaldehyde and s-trioxane . The U. S. produced about 6.4 billion pounds of aqueous formaldehyde in 1978. Most of this quantity was used domestically. The U. S. consumption of formaldehyde by the year 1983 will likely exceed 7.5 billion pounds . Half of the formaldehyde produced is used to produce synthetic resins such as urea- and phenol-formaldebyde res ins . These resins are used primarily as adhesives when making particleboard, fibers board, and plywood. Urea-formaidehyde concentrates are also used in various coating processes, in paper products, and in malting foams for thermal insulation. The textile industry uses formal- dehyde for producing creaseproof, crushproof, flame resistant, and shrinkproof fabrics. Acetal resins, made from formaldehyde, are used to mold plastic parts for automobiles, home appliances, hard- ware, and garden and sporting equipment O Formaldehyde is used in some medicines because it modifies and reduces the toxicity of viruses, venoms, and irritating pollens. The use of formaldehyde in embalming fluids is now required by all s Pate laws. The widespread use of formaldehyde is due to its high reactivity, colorless nature, purity in commercial forms and low cost. In making other chemicals, it can link similar asked dis- similar molecules together. In the paper industry, formaldehyde and its derivatives impart wet strength, as well as shrink and grease resistance. Leather and fur can be tanned 'oy formalde- hyde. Formaldehyde is used in the photographic industry because it hardens and insolubilizes the gelatin surfaces of f ilm and papers. Release of Formaldehyde from Urea-Formaldebyde Foam Insulation (UFFI) . UFFI is a cellular plastic product used as a thermal insulation criteria ~ O The product is manufactured at the job ~ ite by feeding, generally, pressurized air along with two 1 iquid chemicals-~a urea- formaldehyde based (or urea-based) res in and a foaming agent-- through ~ foam equipment system. The product that resul, s from this reactive mixture has a starring cream~tike consistency and is usually pumped through relatively smal 1 holes into the wal Is of standing structures. After the llFFI is inside the wall, the insulation becomes firm or self-supporting. Formaldehyde that has not reacted with oilier chemicals after installation can eventually be released as formaldehyde gas. lithe likelihood of release of formaldehyde depends on factors such as temperature, humidity, foam formulation, architectural cons iderat ions, and ins tat let ion techniques .
3 2. Describe how the question of risk was elevated to the agency agenda*. Irritation ant Sens it ization Initial focus of attention to the possible health problems of UFFI began in October 1976, when the Metropolitan Denver District Attorney's Consumer Office filed a petition, requesting the Conquer Product Safety Commission (CPSC) to develop a safety s tandard for certain types of home insulation products, including UFFI. The petition claimed that there is an unreasonable risk of injury or irritation associated with UFFI. Over the next three years, several thousand consumer complaints related to irritation and sensitivity from formaldehyde released from UFFI were filed with CPSC. After considering information compiled by staff, the Commission decided, in March 1919, to defer a decision on the part of the petition relating to UFFI and instructed the staff to evaluate additional information on possible means of addressing the alleged unreasonable risk of injury (44 FR 12080~. In addition, in May 1979, CPSC began the sponsorship of ~ National Academy of Sciences s tudy which "xamined the acute and s ens i- tization effects due to formaldehyde. The final report, issued in March 198D, stated that "an yet there is no evidence of a population threshold for the irritant effects of formaldehyde in humans. " lathe NAS committee also recommended that levels of formaldehyde be held at the "lowest practical concentration. " The latter suggestion may Cadre been influenced to some extent by a preliminary report to CPSC in October 1979 that formaldehyde caused cancer in rats. In November 1919, staff again briefed the Commission on the s talus o f information gathered concerning formaldehyde toxicity and UFFI. Follow~up investigation of some of the reported complaints indicated a range of severity of reported reactions varying from short-term discomfort to alleged chronic impairment, such as loss of visual acuity, reduction in lung function, and sensitization to other sources of formaldehyde. As a results of these briefings, the Commission decided to hold public hearings to obtain additional information on safety ant health problems associated with UFFI (44 FR 69578~. The Commission held public hearings in Portland, Oregon; Atlanta, Georgia; Minneapolis, Minnesota; and Hartford, Connecticut from December 19799 through February 1980. Consumers, industry representatives, state and local government officials, scientists, and others testified a t the hearings as to their experiences with UFFI and formaldehyde. Based on the information obtained in these hearings an well as the consumer complaints and medical and * This case study describes activities only through April 1982.
4 scientific information obtained beforehand, the Commission9 on June TO, 1980, proposed a rule requiring manufacturers of UFFI to give specified performance and technical information to purchasers to alert them about the possible adverse health effects associated with the release of formaldehyde gas from the product after it is installed in residences (45 FR 39434~. Carcinogenic Risk In October 1979, representatives of the Formaldehyde Institute, an ir~dus try bate association, reported to CPSC the preliminary findings from a careinogenicity study sponsored by the Chemical Industry Institute for Toxicology (CTIT). This study showed that rats exposed to 15 ppm of formaldehyde developed squamous cet1 care inoma o f the nose . In January i9809 an Interagency Regulator Liaison Group (IRLG) task force including six pathologists visited CIIT to obtain additional information concerning the longs study and to verify the findings a In a report prepared in February 1980, the group concurred, in general, with the CIIT observations 9 diagnoses, and interpretations. As the ClIT experiment progressed and more data became available, the need for a full review of the potential health risks of cancer and other genetic diseases to toucans from chronic exposure to formaldehyde became evident. In April 1980, CPSC, in conjunction with the National Toxicology Program (NTP) formed the Formaldehyde Parcel, consisting of 16 federal scientists g charged -title reviewing and evaluating the avai late Le published and un- published information including the ongoing CIIT study. In its final report to CPSC in November 1980, the panel concluded that it would be "prudent to regard formaldehyde as posing a carcinogenic risk to 'numans. " Based in part on the findings of the Formaldehyde Panel 9 as well as on its determination of the irritating and sensitizing effects of formaldehyde g CPSC called for a proposed ban of UFFI in February 19810 The rule was made final in April 19632e 3. Under what statutes and agency ; urisdiction does ache chemical fall? What statutory tests governed the decision? . The ban of t3FFI falls under the jurisdiction of two possible CPSC statutes These are: Section ~ of the Consumer Product Safety Act (CPSA) and Section 2( q) ( 1) (B.) of the Federal Hazardous Substances Act (FHSA). The Commission chose to regulate under the CPSA for two reasons. First, under the MESA, the Commission lacked the authority to cover products installed in nonresidential applications such as schoolsO Secondly because of the complex and lengthy nature of the rulemaking proceeding that would be required under the FHSA, the Commission believed it would be in the public interest to regulate this product under the CPSA.
5 under Section ~ of the CESA, there are two criteria that must be met before the Commission can issue a proposed ban. There are: (1) The consumer product is being, or will be, distributed in commerce and presents an unreasonable risk of injury; and (2) No feasible standard under the Act, including require- ments for warnings and instructions, would adequately protect ache public from the unreasonable risk of injury associated wi th the produc ~ . In CPSC ' s judgment, unreasonable risk was demonstrated by the irritant and sensitizing effects, and carcinogenic potential of formaldehyde . Lee Commiss ion examined the existing information concerning the feasibility of a standard and found that there was no feasible standard, including labeling, that would adequately protec t the pub kc . In its finding that it was not possible to establish a feasibility standard, CPSC denied ~ petition it received from the Formaldehyde Institute on April 16, 1981, which requested that the Commission issue a safety ~ tandard applicable to UFFI. 4. That was the decision schedule? Note any s tatutory or other action deadlines. October 19 76 Action: We Metropolitan Denver District Attorneys Consumer Of Eice filed a petition to CPSC, requesting the Commission to develop a safe ty standard for insulation products, including U.F.F. I. June 10, 1980 Action: CPSC issued a proposed rule to require manufacturers of UFFI to give information to prospective purchasers deco alert them to the possible adverse health effects associated with the release of formaldehyde gas from the product. Deadlines: Written comments to proposed rule to be submitted by . August L1, 1980. (compliance to begin six weeks after publication of final rule. November 20, 1980 Submiss ion of Formaldehyde Panel report to CPSC.
6 February 5, 1981 Action: CESC issues a proposed rule to ban UFFI from residential, commercial, and public buildings in the U.S. CPS(: denies the leerier petition of October 19760 Deadl ines: Written cogent ~ to proposed rule to be submi tted by April 6, 19810 Oral presentation by interested parties concerning proposed rule to be presented Larch 20, 1981. April 10, 1981 Action: The Formaldehyde Institute fi led a petition to CPSC, requesting that the Cotillion issue a safety standard applicable to IJFFI. October 26, 1981 Completion of CPSC 9 S quantitative risk assessment. April 2, 19~82 Action: CPSC issues a rule to ban ~FFI from residences and schoo l ~ in the U ~ S a CPSC withdraws proposed rule of June 10, 19800 CPSO denies Fo~aidehyde Institute petition of April 10, 1981 Deadlines: The bass is effective on or after August to, 1982 or the day after the expiration of the 90 calendar day period of continuing session of Congress following promulgation of the ban, whichever date is later. B IDENTIFICATION OF HAZA R33 O (determining the presence or absence of potential tonic effects) 1. What health endpoints were evaluated? CPSC evaluated a wide range of potential adverse health ef fee ts front fo~aidehyde exposure. Acute effects, such as irrita~eton of the eyes9 nose and respiratory tract were examined9 as well as a wide variety of other effects such as drowsiness, nausea and headaches O
? Sensitization of certain individuals due to fo~=aideh~e exposure was also addressed. CPSC was concerned about a small segment of the population which could be susceptible to severe reactions such as allergic dermititis and asthma. Finally, CPSC reviewed the potential of formaldehyde to ~nduc genetic disease. Ins effects on reproduction ant teratogenicity, mutagenicity and carcinogenicity were reviewed and evaluated, al though the main f ocus was on cancer. 2. What were the key data available for review? (What additional data were sought? ) e Prior to the release of preliminary data from the CITT study in November 1979, CPSC was almost exclusively concerned with potential acute effects and hypersensitivity caused by formal- deh~rde exposure. Over the course of about three years, the Commission had received over 2,000 consumer complaints of adverse health effects from the release of formaldehyde gas from UFFI. These complaints provided the CPSC witl, a readi ly available data base Deco examine the effects of interest. The Commission staff conducted follo~up investigations for over 200 of the complaints. In addition, CPSC asked the National Academy of Sciences (NAS) to conduct a study assessing the health effects of formaldehyde. A final report to the Commission was completed in March 198Q. Two other NAS studies Indoor Pollutants (1981) and Formaldehyde and Other Aldehydes (1981) were available for review and evalua- tion. After November 1979, CPSC was also concerned with the potential carcinogenic effect of formaldehyde. Review of the CIIT rat study began while it was still in progress. In January 1980, an I}lLO tack force was formed to verify the initial findings of CIIT and to ~ eek further informa ~ion. In a report prepared in February 1980, the group concurred in general with the CIIT observations. In April 1980, CPSC, in conjunction with the National Toxicology Program (NTP) formed the Formaldehyde Panel, consisting of 16 federal scientists, to continue Else review of the ongoing CIIT study. In addition, the Pane! was charged with evaluating all available published and nonpublished literature relating to other genotoxic endpoints. In November 1980, the Panel reported its conclusion to CPSC that it is prudent to regard formaldehyde as a human carcinogen. Four other groups of scientists in separate reports endorsed the opinion that formaldehyde should be viewed as if it were a h''m~n carcinogen. The first was a NIOSH Current Intelligence Bulletin. Another was a report based on a rat study conducted by scientists at New York University. In a letter sent to CPSC by Dr. A. Upton, professor and chairman of the New York University Medical Center, the NYU study was cited as providing 'decisive confirmation of the CIIT findings." The third was a report from a
8 working group of the International Agency for Research OR Cancer (IARC) in October 1981. A fourth group, in a report to the American Cancer Society by the Environmental Cancer Information Unit, chaired by I. J. Selikof f and E. C . Hammond (February 1981), found that fo~aidebyde was an animal carcinogen and that human exposures should be reduced or eliminated There were some limited human studies on industry workers and morticians. CESC found these to be inconclusive regarding their ability to predict human carcinogenic effects. The same conclusion was drawn by a working group of IARC. 3. Ah_ ~ background? Available analytic resources? ) Irritation and Sensing ivi ty From the thousands of consumer complaints DISC rece ived, several hundred were singled out for followed investigation by CPSC. Part of the investigation consisted of answering a questionnaires Questions were designed by staff epidemiologists and chemists In addition, when the complainant was treated by a physician, the doctor was contacted to obtain further details. Also, a medical officer on the CPSC: staff reviewed the complaint file. She NAS analysis of health effects of formaldehyde, which was a key document in CPSC ' ~ a~se~:oent o ~ acute symptoms, was pert formed by 13 medical doctors and client. They had expertise in various areas including toxicology, epidemiology, and dermatology . Carcinogenic Effects The GREG tack force that made the preliminary evaluation of the CIIT study in January 1960 consisted of six pathologists representing CPSC9 NIGHS, DOE, EPA and NCI as well as other health scientists. The Formaldehyde Panel which made the final evaluso tion of the CllT study for CPSC consisted of sixteen scientists from various federal agencies including EPA, DOE, NCT, IDA, NIERS, NIOSE, NCTR, and OSHA. All of these had Ph.D.s, Sc.D.s or M.D.s, ant they represented expertise in the areas of carcinogenicity, metabolism, teratology and reproduction, epidemiology, mutagenicity, and risk analysis. The CPSC staff reviewing the Eormaldebyde data in both of these areas consisted of six Ph.D;s or DolteMe8. They hat backgrounds in carcinogenicity, metabolism, mutagenicity, risk analysis and toxicology, biochemistry and pathology a
4. How was uncertainty described in reaching final interpreta- tions ? Were cruc ial as gumptions made exo1 icit? Irritation and Sensitivity lathe identification of the effects of irritation and hyper- sens itivity caused by formaldehyde exposure is quite certain. Controlled Bean studies where small numbers of young healthy adults were exposed to concentrations of for-`ualdebyde varying from 20 ppm to less than ~ pp~n for short periods provide the best evidence . Be NAS s tudy on health ef fects of formaldehyde (March 1980) discusses these and other relevan~c evidence in detail. There is some question as to whether studies on limited populations are applicable to the general population. The NAS study notes this limitation and enumerates the factors that might lead to a wide range of individual susceptibility; e.g., health status, genetic predisposition, age, pregnancy. Notwitlls~anding these factors, the NAS report does caution that "responses reported in controlled studies may occur at an increased rate in the general population, because of the interaction between formaldehyde and other irritants in the environment. " Finally, NAS states: "Some of the factors might decrease susceptibility; mos t may Increase fit. The poss ible exit fence of a threshold to irritation and hypersensitivity raises another point of uncertainty. The NAS study states, "As yet, there is no evidence of a population threshold for the irritant effects of formaldehyde in humans." The NAS leaves this question unresolved but advised maintaining "formaldehyde at the lowest practical concentration to minimize adverse effects on public health. " By the time NAS committee members wrote the final report, the preliminary results of the CIIT study were known. This new information probably affected their final recommendations. Care inogenici ty Since there are no definitive human studies to date for carcinogenic effects of formaldehyde, its potential carcino- genici ty in humans must be inferred from animal data. The lack of direct evidence was clearly stated by the Formaldehyde Panel. The Panel concluded that "definitive expe eiments exis t which demon- strate the mutagenicity and carcinogenicity of formaldehyde under laboratory conditions. " It further stated that "the data presently availab le do not permit a direct assessment 0 f the carcinogenicity of foe =aldehyde to man. " In reaching its final determination it stated: "it is the conclusion of the Panel that it is prudent to regard formalt~hyde as posing a carcinogenic risk to humans . "
10 5. What qualitative factors affected the weighting of data? There mere two particular features of the CIIT study which the Formaldehyde Panel weighed before reaching its conclusions O One was the occurrence of a viral upper respiratory tract infection (sialotacyoadenitis) in the test animals at twelve months of exposure that lasted about one week. The virus event left open the possibility that the infection served as a co-carcinogenic effect contributing to the carcinoma response. Me other feature was that the rats developed severe irri tat ion in their nasal cavities due to formaldehyde exposure. his symptom raised the question of whether irritation effects In and of themselves could cause or contribute to a carcinogenic response. lye Formaldehyde Panel, in considering these two questions, did cot completely discount their importance, but concluded that it was unlikely that either could cause an experimental artifact. With regard Deco the viral infection, the Panel report stated that 'tua~al tumors had probably already formed at the time of the infection," thus casting doubt upon the possibility of ~ · O co-care 1nogen~c response c The Panel addressed the question of irritation by stating they "found no evidence that the induc tion of irritation . O O is a sufficient condition for the carcinogenic activity of an agents" 6. What inference options were used in the Razard Identification step? Were they explicit and in accord with any general guidelines? Inference Options Used in Bazard Identification* (1) Degree of confirmation of positive results) significance . . . Of negative results: The Formaldehyde Panel reached its conch u lion b seed ~~ the positive CIIT rat study as well as other supporting evidenced T.nis evidence included: ~ a) Me NYtJ rat study of Laskin et al . that demonstrated a positive carcinogenic response from a different ~rain. (b) Me demonstrated genotoxicity of formaldehyde * The inference options cited here are drawn from Risk Assessment in the Federal Government.~Managing the Process, NAS ( 1983), ppO 29-30.
11 (c ~ Suggestive evidence from species other than the rat and tissues other than the nasal epithelium; e.g., lesions interpreted to be carcinomas _ situ induced in the oral mucosa of rabbits. Me Panel reviewed the known data for hamsters, which apparently did not exhibit a carcinogenic response to formaldehyde. This negative result did not appear to influence the Panel's final conclusion, based on the positive animal s tudies . ** (2) Evidence of different metabolic pathways between animals and humans: The Formaldehyde Panel reviewed extensively the known ~- data about for~aldehyde metabolism but did not reach any conclusions about interspecies difference. lathe Panel did recommend as future research, that "the pharmacokinetics of formaldehyde and its interaction with target tissue should be studied in several animal species. " (3) Findings of tissue damage and other effects in the inter- ~ _ preta~ion of tumor data: The Formaldehyde Panel considered both the irritation effects and the cytotoxic of feces of formaldehyde. For irritation the Panel concluded that it was Inconceivable that the 'irritant' effect of formaldebyde...may contribute to some extent to the expression of its carcinogenic activity through a mechanism enhancing the promoting or tumor growth s t age of carcinogenesis. However, it must be added that our knowledge of this type of effect is still quite inadequate and not directly applicable to the reported carcinogenic ef fects of formaldehyde on the nasal mucosa. " Regarding cytotoxicity, the Panel noted that "most care inogens have significant cytotoxic effects. Therefore, formaldehyde is not an unusual case....In short, it is conceivable that the cytotoxic effects of fo'-=aldebyde may play a part of its overall carcinogenicity. " (4) Use of short-term test data: As noted above, the Formal- dehyde Panel used short-term tests demonstrating the genotoxic effects of formaldehyde an supportive evidence for the carcino- genic effects of formaldehyde. - ** 'rhe Panel did cite a (as yet unpublished) report indicating that formaldehyde might act an a tumor Enhancing agents' in the trachea of hams ters ~ but acknowledged the s tudy was 1 imited ~
12 Explicitness and Accordance with General Guidelines All of the inference options cited above were quite explicit in the Formaldehyde Panel report. Accordance of these inference options with general guidelines can be described as follows. ( 1) Degree of confirmation of positive results, significance of negative results: The degree of confirmation used by the Formaldehyde Panel in reaching its conclusions appears to be in agreement with past guidelines of federal agencies: CPSC ( 1978) I=G (1979), 0~ (198Q).* These generally state that positive evidence in one or more animal species is sufficient to treat a chemical as if it were a Truman carcinogen. Most of these also state that negative results in animals should not supersede positive results and that short-term tests should serve as supportive, but not conclusive evidence. The guidelines for the International Agency for Research on Cancer (lARC) deal exclusively with hazard identification and are currently in use. It appears that the Formaldehyde Panel ~ ~ use of positive animal and shortoterm teats is In accord with the I.ARC guidel ine.s as well. Under the IARC classifica~c~an avstem" sufficient evidence of carcinogenicity In animals** exis ts if there is an increased incidence of malignant tumors: (a) in multiple species or strains, or (b) in multiple experiment s; or (c) to an unusual degree with regard to incidence, site or type of tumor, or age at onset. Additional evidence may be provided by data on dose-respon~e effects, as well as informa- tion from short-ee'm Bests or on chemical structure. * Some reviewers o f a d ra f t o f this case s tudy questioned the relevance and appropriateness of citing past agency guidelines. While it is true that most of these are no longer in use (and no new ones have come forth recently) due to a variety of complex policy and scientific reasons, the author of thss case study feels it is nevertheless instructive to indicate accordance with previous guidelines. This does not imply that CPSC or the Formaldehyde Panel utilized the guidelines cited ~ in fact no reference to guidelines is lisped in their reports). It toes imply that there is some histor- ;cal precedent within the federal agencies for the choices of inference options which the Formaldehyde Panel and CPSC applied to their risk as sessmer~ts . ** A determination of 'isufficient evidence" in animals categorizes a chemical as as a "probable" human carcinogen in the IARC scheme.
13 (2) Evidence of different metabolic athwaYs between annals P and humans: The,Formaidehyde Panel made no conclusive tatement on this component due to lack of evidence . Pas t agency guidelines (e. g., OSHA,; IRLG) have acknowledged the importance of metabolic differences between species, but also have noted that normally evidence is not adequate to incorporate such inf ormat ion. (3) Findings of tissue damage and other effects in the inter- pretat ion of tumor data: The author of this case study has found no specific reference in guidelines dealing with tissue damage and other toxic ef facts. (4) Use of short-term test data: The Formaldehyde Panel's use of ~hort-~erm tes ~ data an supportive evidence for carcinogenicity appears to be in accord with most guidelines which have addresses this component. 7. Describe an internal inte~al-advisory (e.g., SPAT SAB) and Y ~ - external (e.~. NAS) scientific review of the initial analysis. ~ , What if anv or iticism was incurred? , it, There was no scientific review of the Formaldehyde Panel's report. 8 Row were issues raised in the review(~) accommodated' · ~ This question is not applicab Le. 9. What other issues arose concerning scientific data and their use? Brief by describe di s sent ins hazard identi fication only) . .__ In response to the proposal to ban {JFFI, CPSC received hundreds of written comments and 21 oral presentations. Included among the commenters were industry representatives, consumers, consumer organizations, con~uttants, scientists, and federal, state and local government agencies. The major points raised are summarized as follows: Irritation and Sensitization Lee relationship of reported symptoms to DEFT was challenged. Counters noted that complaints to CPSC from consumers living in UFFI homes did not constitute a causal relationship between symptoms and formaldehyde exposure. Some commenters s tated that there was a failure of the complaint data to demonstrate a dose-response relationship. Others argued that the absence o f formaldehyde measurement data in some of the complaint reports weakened the al leged causal re lationship.
14 Some commenters contended that levels of formaldehyde found in UFFI homes were too low to cause the reported effects. Exposure studies on human volunteers, which have demonstrated adverse health effects to at leas t as Low as O. .2 ppm, were attacked O Criticism of experimental conditions (alleged crowding tempera- ture, and tumidity) s the lack of a dose-response relationship, and the lack of appropriate control groups in these studies were cited. Some condensers stated that there was evidence avowing there to be no difference in formaldehyde levels between homes with and without UFFI. This contention wan based on a recent University of Iowa study, although apparently no detailed data in support of this finding was presented at the time. In the opinion of another commenter, studies showed that there was no difference in formaldehyde levels between "properly foamed" homes and homes without UFFI. Care inogenic i ty Some conmenters stated that the CIIT rat study is flawed because formaldehyde exposure at L5 ppm caused ulceration and acute exudative inflammations which are in themselves cancer inducing. Others asserted that the viral epidemic in the rats of the CIIT study inval idated the resul ts. These ques t ions had been considered earlier by the Formaldehyde Panel (see Question Bow A number of cementers argued that there were s tudies showing that persons occupationally exposed to formaldehyde did not develop cancer at higher than normal rates ~ Epidemiologic s tudies on formaldehyde industry workers and morticians were cited. CPSC considered the results to be inconclusive rather tears demons trat- ing that formaldehyde is not carcinogenic in humans. Other commenters noted that there appears to be a wide interspecies variation of response to formaldehyde. Hams ters do not appear to exhibit ~ carcinogenic response, mice exhibit a relatively tow responses while rats stow a large response (at 15 ppm). In the view of these co`~..enters, judgment about the car- cinogenic effect of formaldehyde in humans should be tempered by the fact of wide interspecies variation of response in animals. 10. 13 the substance subject to past or possible future re`2ulatorv actions__ - ordinate with other agencies or 2rogra~ns ~ Formaldehyde is subject to possible future regulatory actions by EPA and OSHA. IN addition, HUD is cons idering rulemaking for UFFI in mobile homes DOE has stated that its proposed fJFFI regula- tions would conform with any legal actions taken by CPSC. FI'A has the mandate to regulate hut no serious action has been taken thus far based on the agency decision that no uses for which it 'nas jurisdiction lead to direct nasal exposure or inhalation exposure in humans.
15 Initially, there was considerable coordination of agency programs under the Interagency Regulatory liaison Group (IRLG). In the spring of 1980, the IRLG had chosen formaldehyde as one of 6 Category 1 substances. The specific intent of such ~ classi- fication was to ensure that the selected chemical would receive focused attention for the purpose of integrating regulatory programs due to the imminency of regulation by one or more of the agencies. A Formaldehyde Task Group was assigned to accomplish this goal, and a detailed Regulatory Development Workplan, including a timetable for action, was prepared. One activity of the Task Group was to examine the feasibility of developing an integrated, interagency risk assessment for formaldehyde. The ultimate goal of the coordinated risk assess- ment was to develop a single document which could 'he ut it iced by the IRLG agencies as baseline information for regulatory action. This effort '~as only the second attempt of the federal government at a unified approach to risk assessment. (A similar effort for cl~lorofluorocarbons was aixeady in progress at the time. ~ CPSC took the lead in developing a quantitative risk assess- ment which -~as reviewed and approved by other IRLG agencies. RAG hired a contractor to prepare and compile all the exposure data that was available for all possible areas of interest; i.e., occupational, drug related, environmental, and consumer. The idea was to apply the risk assessment to that information in order to estimate where the greatest number at risk might be, so that priorities court be set across agencies. The contract was not completed before IRLG was disbanded in September 1981. Al though CPSC, EPA, OSlIA, and HUD s igned a S tatement of Pol icy Coordination regarding potential regulatory control of formal- dehyde in the spring of 1981, there has been an apparent reversal o f the ef fort to coordinate regulation even previous to the official expiration late of IRLG. This change in trend is highlighted by the different and, in some ways, contradictory approaches toward regulating formaldehyde taken by CPSC and EPA in 1981-82. (See also Question D.2. for detailed analysis.) C QUANTIFICATION ~1) CHARACTER ZA'rlON OF RISK TO HUMANS . (This section deals with Crests quantitative risk assessment of October 26, 1981. Acute effects, mutagens and other pass ible hazards will not be dealt Smith since these have not been quantified. 1. What health_ pi 1 ''~-d? Malignant cancer was the only health endpo int evaluated. Me risk assessment did not specify any particular target organs in humans.
16 2. What were the key data available for review? ('~at additional - data were sou'2ht? Animal Data The risk assessment utilized data exclusively from the CIIT rat srudyO Human Exposure There were two types o f human exposure data a One type was obtained from actual on-site measurement of UFFI and non-UFFI residences. The other type was obtained from controlled laboratory studies using IIFFI panels, formed under conditions of best available technology by manufacturerts representatives. The on-site measurements consisted of 827 data points from UFFI residences In which an adverse health ef feet was reported ("complaint" homes) and 337 data points from UFFI residences in which no c omplaints Revere reported . S ince formaldehyde leve 1 s in Complaint arid "noncomplaintt9 DEFT residences did not signifi- cantly differ from one another, al 1 1164 data points were used to estimate an average age-of-foa~n related formaldehyde leve 1. One hundred three data points were also obtained from non-UFFI residences. This data was used to subtract out a background level of formaldehyde in residences. Sources of data came from "in-depth" CPSC studies 9 numerous reports from ten states and Canada g as well an studies sponsored by various universities and the Formaldehyde Ins titute. Original ly, CPSC utilized ones ite data for "complaint" horses only. lathe laboratory studies of UFFI panels were used by CPSC to assess the exposure for the total population, including UFFI residences where no complaints were reported. Men further on-site data analysis showed no significant difference in exposure levels for "complaint'' and "noncompLaint~' homes 9 the Lab studies were continued so that risk could be calculated using another data base a The lab studies consisted of estimating the average projected formaldehyde level irt a hypothetical corner room for each of 24 panels foamed by manufacturers' representatives under optimum conditions. Me Frankl in Institute Research Laboratories (FIRL ~ performed the initial study under contract from CPSC. Oak Ridge National Laboratory (ORNL) did a follow~up study also sponsored by CPSC. CPSC engineering staff worked closely with ORNL to develop a recommended analytic methodology.
17 3. Who erfo'~ned the initial anal sis? Utahan was their P Y background? Available analytic resources? ) A CPSC staff member with a B.S. in analytic chemistry and 20 years of experience was the main person responsible for verifying the data for on-site exposure measurements obtained from the various sources mentioned in Question C.2. Verification consisted of reviewing the testing methodology to determine its acceptability. Engineers from FIRL, ORNL, and CESC designed the laboratory studies on exposure from the formaldehyde panels. The quantitative assessment of the dose-response curve was originally determined by a Ph. 0. risk expert from the National Center for Toxicological Research as part of the Formaldehyde Panel report . A CPSC staff person with a Ph. D. in biochemis try and two years of experience in risk assessment did an independent analysis for CPSC. (Section C deals entirely with the latter analysis. ~ A complete set of computer programs for dose-response and statistical analysis were available to CPSC staff. 4. To what extent were results presented quantitatively? What factors influenced the degree of quantification? The results were presented in a precisely quantitative canner, as chances per mill ion o f developing malignant cancer during a lifetime. Three different es titrates were given. Two of the estimates were based on the linearized multistage model of Crump, et al. (1979~. They were obtained by applying the model to the two different exposure data bases described in Question C.2. The third estimate, which predicts essentially zero risk, is based on the "best-fit-of-data" mode! which is a purely statistical fit of a pa lynomial curve to the data. The factors that influenced the degree of quantification were: (1) a well conducted rat study, showing a pronounced dose-response effect, which incorporated three different doses, plus a control and consisted of 120 animals per sex per species per dose level and (2) exposure data from two independent data bases; one of which contained over 1000 data points. 5. How was uncertainty described in reaching final inter- pretations? Were crucial assumptions made explicit? Generally, in predicting human risk from animal data, there are two areas of great uncertainty. These are: ( 1) the extrapolation to low dose of risk observable at much larger doses and (2) transferring the risk from animals to humans (interspecies conversion) .
18 The CPSC risk assessment desit with uncertainty in extrapolation by presenting the risk from two different dose- response curves which. most plausibly, define the upper and lower limits of risk. The upper estimate was based on a linerized multistage model (95: upper confidence limit (UCL)), which is usually considered to be a conservative estimate. Me lower estimate, based on a "best-fit-of-data'' model g rainim~ze~ the riske Assumpt ions for us it these models are c [early de lineated . (See Questions Co6 and Co7 for details. ~ The uncertainty in choosing an interspecies conversion factor was not accounted for in the CPSC risk assessment. Only one conversion factor was utilized. The risk to rats for a given concentration in air was assumed to cause the same risk to humans at that concentration. The rat ionale for this premise was a paper by Mantel and Schneiderman (1975~. Furt'nermore, the duration of exposure in rats was corrected for means by calculating duration as a proportion of species lifetime. Although these assumptions may be reasonable, Mantel and Schneiderman describe them as first order approximations. It is not clear whether the selected conversion factor is conservat Eve or liberal with respect to other convers ion fac tore that could have been used . In addition, There was no statement about the pharmacokinetics of formaldehyde in rat ~ and ~nuu~ans and wile ther or not there were any features that might cause d ifferences in response between the two species. It is quite likely that such information was lacking or unavailable 9 but CPSC did not s tate so. 6. Row were qualitative factors dealt with? Mechanisms of action, associated thresholds EEfec ts on population subgroups Other confounding f ac tors Mechanisms o f Action The risk assessment states that the most likely mechanism of action of formaldehyde is its ability to interact "with ongoing processes in the human body which can lead to cancer, and potential interaction with other carcinogens that humans may be exposed tOo t' CPSC supports this assertion by c iting formalde- hyde' s ability to interact with the genetic material (DNA) in a wide variety of short-term tests. Given there mechanisms, CPSC asserts That linearity of the dose-response curve at low exposure Petrels is mos ~ applicable. CPSC considers the "unlikely'' "vent that formaldehyde might not interact with an ongoing chemical process. In that case, the "bestofit-of-data" estimate (with essentially zero risk at low dose) is asserted to be the most applicable.
19 Effects of population subgroups: Mere is no discussion of thin topic in the risk assessment. Other confounding factors: CPSC depended entirely on CIIT for the release of tumor data. CPSC staff had to make certain assumptions regarding the data that probably would not have been necessary had the experiment been under their control. Some of these assumptions were as fol lows: (~) Rats developing tumors early on were assumed to have received a 24~month exposure to formaldehyde; (2) data on the rats was available only through 24 months of age; the analysis does not compensate for the fact that, had they been allowed to live, more rats would likely have developed cancer; (33 CIIT made scheduled serial sacrifices of rats at 5, 12, and IB months. CPSC omitted all data on animals sacrificed early since these would not have had the same opportunity to develop a rumor as they would had they been allowed to survive for longer periods. Eight animals (out of 40) sacrificed at 18 months had malignant tumors at the high dose. This data alas omit ted; and (4) Although there was a significant incidence of benign tumors in the tent animals, ClIT declined to release the data on the ground that no dose-response relationship was - observed. CPSC, therefore, did not consider benign tumors in its assessment. 7 0 What qualitative factors af fected the weighting of data? The major qualitative factor that affected the weighting of data was the cone iteration of biologic plans ibility in the choice 0 f a dose-response model. Both the Formaldehyde Panel and CPSC strongly endorsed the use of the linearized multistage model. This preference was based on their belief that it is the most Connie tent with prevalent theories on chemical carcinogenesis. (See Question C.~.) Another qualitative factor which may have affected the final results was the omission of data on benign tumors. Lois omission -~as not decided on the basis of policy but, rather, because the data was not available from CIIT. Since there was no detailed discussion of the significance of benign tumor data in the risk assessment, it is difficult to determine how the data would have been interpreted and incorporated i nto the final results.
20 Mat inference options were used in the quantification of risk? Were they explicit and in accord with any general guidelines' Inference Options Used in Do-~e-Response Assessment* ~ ~ ~ Choice o f extraco ration mode 1~: CP SC chose a mod i f fed . multistage model which incorporates linearity at low exposure levels (the linearized multistage model). CPSC also cal- culated risk based on a purely statistical fit of the multistage model, making no provision to take linearity into account at low dose. CPSC's risk assessment explicitly endorse s the ~ inear is ed mode ~ ore r the let ter mode 1 . (2) Choice of confidence limits or 'oest estimates: The l linearized multistage model utilized the 953 upper confidence limits Me purely statistical model utilized "maximum ike ~ ihood es time tes . " ~ 3 ~ Choice of interSPecies conversion factor: C8sc assumed that the rink to rats at a particular concentration was equal to the risk to humans a~ the same concentrations and that the duration of risk should be based on the proportion of lifetime ~ ., for a given species. ('~) Use of information co arinz differences in metabolic mp processes and_ra ~ _ _: CPSC did not discuss this component, finding that information regarding it was inadequate or lacking. OPSC implicitly. assumed similarity of processes and rates between rats and humans O (5 ~ Treatment of data when data from more than one species or strain is available: Data from CIIT were a~raitable for two , species, the rat and the mouse. The CPSC quantitative risk assessment utilized data from the rat study only. The rat is considerably more sensitive to formaldehyde than any other species bested to date. Explicitness ant Accordance with General Guide tines The first three inference options were explicitly stated in the CPSC risk assessment, while the latter two were not explicitly stated. Accordance of CPSC' ~ selected inference options W~31 general guidelines can be described as follows: *lithe inference options cited here are drawn from Risk Assessment in the Federal Government, Managing the Process, NAS ( 1983), pp. 31-320
21 (~) Choice of extrapolation Ideas: The use of a linearized mode! was in accord with most guidelines developed by federal agencies in the past. The linearized multistage model, a sophisticated type of linear model, had been adopted by EPA' ~ Carcinogen Assessment Groups (CAG) in the surer of 1980. The use of a bees sophisticated linear model had been advocated earlier by senior EPA officials (EPA, ~L977), although they also stated "the use of several extrapolation models is appropriate to convey the range of uncertainty in these elements. " F1'A endorsed the use of the linear model in its proposed guidelines regarding chemical compounds in food- producing animals (FDA, 19793. The IRLG guidelines (IREG, 1979 ~ stated that linear extrapolation "should always be included among any methods. " The use of a linearized model for dose-response assessment has not been universal ly accepted as a general guidel ine . Some have argued that for particular chemicals, other models may provide ~ better fit, or that linear extrapolation may only apply when The chemical in question is geno~oxic. For example, the Food Safety Council, a trade association, has tated: 1. Where the linear (one-hit) model fits the dose-response ant biochemical data in the observed range as well as other models, it should be used. 2. If the toxicity in question is a genotoxic carcinogen, Allen the one-hit ~ low~dose linear) model is appro- pria~ce. 3. If the two cares above do not apply ant the one-hit model clearly does not fit the data, then the better- fitting motet should be chosen. It should be applied with whatever conservatism seems appropriate from the data. Such conservatism is particularly appropriate when the subs tance is genotoxic . (2) Choice of confidence limits or best estimates: The use of the 95% upper confidence limit was in accord with the procedure of EPA's Carcinogen Assessment Group. The use of "maximum likelihood estimate" has not been explicitly endorsed in any federal agency guidelines to the knowledge of the author of this report. An OSIP report in 1979,, did however, endorse the use of "most likely value" in estimating risk quantita~ci~re ly (OSTP, 1979 ~ .
22 (3 ~ Choice of interspecies conversion factor. lye choice of . . . . conversion factor was based on a paper by Mantel and Schneiderman (1975), and does not appear to be in accord with any specific agency guidelines. Generally, there seems to be no consensus for use of a particular conversion factor among the agenc ies . The Careinogea As sessment Group advocated the use of mg/surface area/day. F13A, in 1979 9 re jected the use of con~rersior~ factors based on surf ace area ratios and advocated mg/weight of total diet. The IRLG document stated "several species-conver~ion factors should be considered." (4) Use of information comparing differences in metabolic . ~ .. ~ . . . . .. processes and rates in experimental animals and humans: Most agency guidelines written in the pas t have advocated the use of metabolic and pharmacokinetic data to demonstrate interspecies differences, when such data is availableO Typically, however, data pertaining to these areas is inadequate or lacking. In the view of CPSC, there was insufficient evidence available to consider interspaces differences in the present risk assess- mentO The lack of sufficient data and the need for more informal ion was acknowledged in the Formaldehyde Panel report recommendation rCor future research. "the pharreacokinet~cs of formaldehyde and i ts interaction with target tissue should be studied in several animal species. " (5 ~ Treatment of data when data from more than one species or straits is available o Most agency guidelines written in the past have not explicitly addressed this questions The IRL& document, however, does False an explicit s tatement: "If data on animals are used as the basis for estimating human risk, data obtained from the most sensitive animal species or strain tested are commonly recommended as the starting point for extrapolation. Of the available data, these are clearly the least likely to underestimate human risks Use of data from less sensitive species or strains is justifiable only if there are strong reasons to believe that the moat sensitive animal model is completely irrelevant to any segment of the exposed human popula tion. " 90 Describe any internal, internal-adv~sorg (eon., EPA's SAB) and external (cog., HAS) scientific review of the initial analysis ~ '~hat, if any, criticism was incurred? CPSC has no formal internal review group. Within the Directorate of Health Sciences,, there is an informal risk assessment work group consisting of staff persons knowledgeable in risk analysisO This group reviewed the document. In additions all staff involved with or interested in formaldehyde reviewed the risk as sessment.
23 In addition, the analysis underwent an external review. Other IRLG agencies were asked to comment (EPA, FDA, and OSHA) . EPA' s review was particularly intensive. Ten people in three different groups were involved in the review process. Outside experts who were involved in the review process were: Dr. Kenny Crump, a private consultant, Or. David Gaylor of NCT8, an MIT group headed by Dr. Nick Ashford, and Dr. Charles Brown of NCI . One criticism dealt -with a refinement of the application of the linearized multistage model. In the initial rick assessment, the value of the upper 95: confidence limit of the lowest observed dose (2. 1 ppm) was calculated, and a linear extrapolation from that value was arbitrarily assumed. Dr. Crump suggested that a rove accurate assessment would be obtained if the upper 95: con- fidence limit of risk is followed to exposure levels below 2. ~ ppm. In this refinement, linearity appears only at doses below 1 ppm. Other criticisms centered around the choice of interspecies conversion factors. The assumption that the risk to rats at a part icular concentration was equal to the risk to humans at the same concentration was challenged. Also, the calculation of duration of risk based on proportion of lifes ice for a given species was questioned. No suggestions, however, were made on how to better incorporate such factors. 10. How were issues raised in the reviews accommodated? Following the suggestion of Dr. Crump (see Question C.9), the reprised risk assessment refined the linearized multistage cal- culation used in the initial assessment. Instead of arbitrarily assuming linearity from the upper 95: conf idence limit value of the lowest observed dose (2.1 ppm), the new analysis calculated values below 2.1 ppm. This recalculation defined a more precise linear funct ion which led to a 2. 4-fold reduction in the estimated risk. 11. What other issues arose concerning scien~tiEic data and their use? Describe dissenting opinion. Interested parties were able to comment on the risk assessment after CPSC published an announcement in the Federal Register. The Formaldehyde Institute, CIIT, and others raised many objections. A major criticism focused on the choice of a linearized dose response curve at low levels of exposure. Critics questioned the bases for assuming a linear model and charged that it was exces- sively conservative and that there were reasons to believe that it was not scientif ically plausible.
24 On December 15, 1981 Dr. James Gibson, vice president and director of research for ClIT, presented to the CPSC staff an argument which would suggest a much reduced risk at low exposures from that predicted by CPSC's linearized model. Drc Gibson reported that a physical mechanism for clearing fo~ldebyde from the nasal apithelium could substantially reduce the dose delivered to target cell TO He reported that the respiratory epithelium is protected from foreign chemicals in inspired air by a layer of mucus, which can carry chemicals away from the raesal cavity by the propelling action of beating cilia. He stated that at high con- centrations formaldehyde could -saturate the mucus blanket and subsequently contact directly with target cells, but at low concentrations formaldehyde may be completely assimilated within the mucus blanket and removed without reaching the target cells. Ire essence, 0~. Gibson presented an argument for a threshold dose-respon~e model. CPSC' s had stated in its public record that effects due to formaldehyde on the nasal epithelia are observed at low doses. CPSC ci tes the occurrence o f benign tumors a t 2 ppm in the CT IT rat study, and the occurrence of nasal irritation effec ts at concentrations as tow as 00 25 ppm. Some might argue that these data suggest clearance may not be complete at these doses. In addition in a February 15, 1982 letter to Nancy Steorts, chairman, Consumer Product Safety Commission, Dr. Gibson cited studies conducted at CITT which in his judgment demonstrated the i~port- ance of concentration rather than chaste exposure to formal- dehyde. He s dated, for example, that 'ishort exposures to L2 ppm caused severe disease 9 including ulceration and erosion of the lining of the nose after as little as 3 days of exposure. In contrast, There was no ulceration or erosion of the lining of the nose in the rats exposed to 3 ppm. Thus to es timate tumor incidence based on cumula~cive exposure without cons idering the more important issue o f high concentration ef fee ts i s s cien- tifically invalid. Formaldehyde concentrations that are sufficiently high cause acute cell injury9 cell death and ulceration of the nasal mucosa and may be associated with the induction of nasal t''mors. However, in the absence of acute effects nasal tumors are not expected and do not occur.... " In OPSCts view revere ulceration is not a necessary condition for the induction of tumors. Rancher, CPSC believes that the major effect of ulceration is to cause respiratory epithelial cells to be rapidly replaced by squamous cells, which appear to be more resistant to cytotoxic effects. It in the squamous cells, where the induction of tumors has been observed. Thin sequence of events does not imply, however, that tumors would not Cadre occurred without ulceration.
25 Another much debated issue was the question of the geno- toxicity of formaldehyde. It is generally believed that carcinogens which are ge£~otoxic exhibit a greater potency at low doses than carcinogens which are not genotoxic. There are scientific arguments which support the belief that genotoxic agents exhibit a linear nonthreshold response at the target s i te at low doses while nongenotoxic agents exhibit a threshold response . CPSC cons iders formaldehyde to be a geno toxic agent . Dr. Gibson believes that the genotoxicity of formaldehyde may not be expressed at low exposure: It also should be stressed that the results of various generic toxicity studies are mixed; some are positive and some are not. Research by CIIT indicates that the potential genetic toxic) by of formaldehyde occurs only in dividing cells. The increase in cell proliferation brought about by acute tissue damage due to high con- centrations of formaldehyde produce such a circumstance. In the nasal mucosa, CIIT has shown that the normal rate of cell division is very lock, and that the rate is not increased by formaldehyde concentrations o f 0.5 or ~ ppm in rats or by 0.5, 2 or ~ ppm in mice. (Gibson letter to N. Steorts, Feb. 15, 1982. ~ CPSC's position stated that although the results of mutational expression may be mixed, all the data are consistent with formalde- hyde being a weak mutagen; negative resul ts can be explained by the inability of formaldeby~le to reach the target being assayed. Also, CPSC believes that although cell proliferation may be an important factor in cancer development, there are no data which show that it is necessary for formaldehyde to cause this ef feet in order for cancer to occur. CPSC believes that formaldehyde does not have to cause all stages of cancer development; background processes could interact with formaldehyde in cans ing cancer. Another major criticism dealt with exposure assessment. Critics contended that the quality of the CPSC exposure late base was poor. They stated that the data was accumulated from many sources with "no colon denominator and little or no controls, " and the prescribed analytical method lacked sens itivity at the measured exposure levels. OPSC maintained that its method of analysis and the assump- ~ions used caused consistent underestimation of exposure. Finally, critics contended that the external peer review of CPSC's risk assessment involved scientists with a policy orienta- tion similar to CPSC's (i.e., an orientation partial to the use of linear extrapolation through zero models) . ~us, they believe that the peer review did little to challenge major issues of the analys is .
26 12. Is the substance subject to past or possible future regulatory actions in other programs? If so, did the program office coordinate with other agencies or programs? (See Question B. 1Q). D. INTERPRETATION 1. What role did risk assessment have in the final agency document where s tandards were established to control the _ , . chemical? - The carcinogenic risk assess~sent ptayed a ~rital role in se~ting the final standard, which was to ban the use of 13FFI in residences and schools. Previous to the development of this risk assessment, CPSC had considered regulation solely on the basis of irritation effects and hypersensitivity (45 FR 39434, June 10, 1980~. The proposed action was nor a ban, but a rule that would require manufacturers of UFFI to give hazard information to prospective purchasers. The teas is for the di fference between these two regulations depended heavily on CPSC ~ s determination that as high as 50 out of a million increased deaths per lifetime could occur from the use o ~ UFFI. Here there variations--o~rer time or across agency programs-- = _ _ in the assumptions used? Were these variations 3igni ficant to the final risk assessment? Variat ions Over Time In the spring of 1981, EPA staff, from the Office of Toxic Substances (OTS), recommended that priority attention be given to formaldehyde and drafted a notice to appear in the Federal Register (FR) saying that formaldehyde would be considered for regulatory action under Section 4(f) of the Toxic Substances Control Act (TSCA). We EPA draft notice stated, "EPA has determined that there may be reasonable basis to conclude that some exposures to formaldehyde present a s ignif leant risk of -widespread harm to humans. Therefore, the agency is initiating action to investigate those exposures of greatest concern and determine whether they lead ho unreasonable risks . " Short ly after this determination, Anne Gorsuch, the newly appointed EPA Administrator, took office. She did not sign the notice and it did not appear in the Federal Register. On February 10, 1982, Dr. John Todhun~er, the new Assistant Administrator for Pesticides and Toxic Substances sent a formal memorandum to Ms. Gorsuch recom- mending that formaldehyde should not be cons ide red as a priori ty chemical for regulation. In his Judgments the earlier OTS risk assessment was deficient One for point was 'nis charac~eriza- tion of the exposure estimates as poor in quality. He also found
27 fault with the inference options used for hazard identification and dose-response assessment. Specifically, he stated, "The PRL-1 ~ the name given to the 1981 OTS risk assessment document, which formed the basis for the draft notice) concludes that Eo.=aldehyde is an animal carcinogen. It down plays a number of negative bioassays which suggest that its effects may depend highly on species, route and site. No attempt was made, also, to address the question of mechanism of action or other physiological/ biochemical questions relevant to the extrapolation from rat to human even though such information was available from CIIT." He was also critical of the PRL-l treatment of the dose-response data from the CIIT rat ~ tudy . ''The risk ca lculat ion uset by OTS included the 5.6 ppm exposure level as a non-zero incidence data point. CIIT has since determined this point to be statistically no different than the 0 ppm and 2.l ppm exposure results.* If this 5.6 ppm exposure were treated as a zero response points the risk e~timates...would shif~c...to yet lower rate." Furthermore, 0~. Todbunter stated that even if one were to accept the calculated levels of risk determined in PRL-1, they would not be sufficiently high to trigger action under Section 4(f~o As can be seen, nearly all individual risks fall into the range 1 x 10-6 - 1 x 10-~. This places them into a range in which priority action is often not considered...if the human risk were real, the magni- tude of individual risks do not deem to compel a "fas t track" approach. Somewhat parallel developments occurred at the Occupational Health and Safety Administration (OSHA). OS" officials had been planning to release a joint s tatement on formaldehyde Smith the National Institute of Occupational Safety and Health, but in July 1981 this decision was reversed. A petition brought by the United Auto Workers reques tiny an emergency ~ Gangland for formaldehyde was denied by OSHA on October 26, 1981. * CTIT determined that the incidence of 2 malignant tumors versus zero in the study controls was not significant. Others might argue that the incidence is significant if historical control data is con- idered .
28 Variations Across Agencies Thus far, CPSG is the only agency to issue a final rule on formaldehyde use. IN based this decision on a risk assessment similar in many aspects to the EPA PRL-1 document which was criticized by Dr. Todhur`~er irt his February 1O5 1982 memo to Ms. Gorsuch. Even though the Todbunter memo does not specifically address the CPSC risk assessment 9 the issues raised therein do serve to highlight differences of interpretation of risk assess- ment data between CPSC and EPA. I ~ should be noted that the degree to which the two agencies differ increased significantly with the advent of the new administration at EPA in the spring of 1981 ~ see Question B. tO) . In specific terms, ])r. Todhunter states that, "Formaldehyde appears, therefore, to exhibit considerable species specificity with the rat, the most sensitive species so far tested. Concern that formaldehyde gas may induce tumors in humans should be tempered by this observation that formaldehyde c arc inogenicity appears to have a high degree of species specificity and a strong dependence on route of exposure." CPSC' s position is that the strong evidence of carcinogenicity in rats provides strong evidence of carcinogenicity in humans, especially in Light of the limitations o f the negative studies . Dr. Todbunter also argues that at low levels of exposure ( 1-2 ppm) formaldehyde may exhibit a toxicity much reduced from that predicted by the 1 ineasized model ut il ized by CPSC O He sites findings by CTIT to support thin argument. These are: (1) Reversibility of byperpLastic and metaplastic effects of formaldehyde at low exposure levels or short exposure t ides; (2) The presence of endogenous levels of formaldehyde in tissue ranging from 3-12 ppm; (3 ~ The absence of cytotoxie Of facts from formaldehyde levels at or below LoO ppm in air; (~) Evidence to sugges t that formaldehyde acts as a promoter. CPSC re jec ts the contention that Cheese arguments sugges t a nonlinear dose-response relationship at low dose, with greatly reduced risk at low levels of exposure. CPSC believes that formaldehyde is both an initiator and a promoter and that the reversibility and promotional aspect ~ points 1 and 4) do not argue against linearity at low doses especially when the property of initiation is CONS idered O CPSC further be lieves that tile absence of cytotoxic effects at loO ppm has not been at all demonstrated;
29 and that, nevertheless, the concept of cy~cotoxicity is not relevant to the shape of the dose-re~ponse curve at low dose where initiation and promotion aspects of formaldehyde are still likely to be operating. Finally, CPSC believes the entogenous levels of 3012 ppm are also not derived from any existing data, and not relevant to the argument. CPSC notes that virtually all formal- dehyde in tissues is not free, but bound in a different chemical fore with entirely different chemical and biological properties. Another point of departure is interpretation of data on benign tu~OtSo Dr. Todbun~er does not directly address the question of benign tumor incidence in the formaldehyde rat study. He does address the topic generically when suggesting a definition for the term "serious. " (Part of the criteria for initiating Section 4( f) is the determination of "serious or widespread harm. " ~ The concept of "serious" harm would have more utility in the consideration of gene mutational events or birth defects than in the case of cancer since malignant neo- plasms in general are serious. Section 4( f) also draws a distinction between benign and malignant growths by use of the term "cancer" rallier than "tumors.'' The CPSC risk assessment did not consider benign tumors, mainly because the information regarding their inc idence was not reported to CPSC by CIIT at the time. CPSC 's risk assessment does state, however, that inclusion of benign tumor data would lead to an increas ed pred ic t ion o f ris k. Another point of difference regarding the interpretation of available epidemiologica1 data for formaldehyde. Dr. Todhunter states, ''There is a limited but suggestive epidemiological base which supports the notion that any human problem with formaldehyde carcinogenicity may be of low incidence or undetectable. In would not appear reasonable to say that a significant risk situation exists from this data." CPSC, after reviewing epidemiological evidence from written statements and oral presentations at public meetings concluded that none of the epidemiological s tudies to date were of ~uf ficient ~ tatistica1 sensitivity or qua L ity to draw any valid inferences regarding the magnitude of carcinogenic risk to humans. Scientists at NIOSH and a working group of [ARC concurred wi th the CPSC evaluation. The two agencies differ sharply on the levels of risk that should trigger concern. Dr. Todhunter states, "In terms of individual lifetime cancer risks, the various federal agencies do not tend tco regulate risks of ~ x 10-5 or lower and tend to be ambivalent about risks between 1 x 10-4 - 1 x 10-5. " He stakes further [hat, "In OPTS, the relative risk range of 1 x t`~~ to ~ x 10 or lower has been a low concern range in general." CESC's estimated risk for inhabitants of UFFI residences is 0.6 x 10 4 - 0.9 x 10-4. Also, in a proposed rule regarding chemical compounds in food-producing animals, F1)A
30 suggested ~ risk of 1 x 10 6 over a lifetime as a trigger for regulatory action: "An increase in the level of risk to l in 10,000 might significantly increase human risk. It is difficult to choose between ~ in ~ million and ~ in 10,000 but the agency chose the more conservative number in the general interest of protecting Furman t~ealth,'t (FDA, 1919~. Finally, Dr. Tothunter states, "lathe results of the CIIT bioassay are, however, sufficient to establish that formaldehyde is a potential animal carcinogen with mode and degree of exposure quite important to the outcome." CPSC's opinion is that forTeal- dehyde is a definite animal carcinogen and should be a presumed human carcinogen. 3. To the extent there were issu~s/concerns about questions of silence 9 would tone outcome have been improved 'oy: - A better system of in-house scientific review? Review by an outside scientific organization? - Coherent federal guidelines on carcinogenic risk assessment? - Better agency guidelines on the performance of risk assessment? - Improved agency dec is ion procedures ? In the view of the author of this case study, many of the issues and concerns about the risk assessment of formaldehyde revolved around a complicated mix of risk assessment policy judgments and scientific judgments. On both s ides of the debate, there was often no clear distinction drawn between scientific and policy judgments O At the time the risk assessment was prepared, there were no uniform federal guidelines for carcinogenic risk assessment irt use. Such guidelines may'~ave been helpful in providing a scientif ic and policy framework to support the CPSC risk assessment, and in drawing sharper distinctions between scientific and policy judgments employed in risk assessments. Critics' comments may have focused on the generic scientific and policy issues raised in the guidelines. Furthermore, adherence to guidelines may have fostered a greater degree of explicitness in the use of inference options applied, and reduced the degree of inconsistency in interpretations among the agencies ~ i O e. ~ CPSC and EPA). The CPSC risk assessment may have !'enefitted from peer review by art independent science advisory panel. In general g this is good practice for all federal risk assessments.
31 E. PERFORMANCE CONSIDERATIONS 1. Ab ility to obtain scientific information. CPSC staff felt that there was a problem obtaining pertinent data from CIIT as it was needed. Time-to-`c~or data and information on benign tumors was withheld until the release of the full final CTIT report in February 1982 (three months after the completion of the final CP5C formaldehyde risk assessments. Consequently, time-to-~or and benign tumor data were not cons idered in the final risk assessment in any quantitative fashion. Both pieces of information would have increased the risk had they been factored into the risk assessment. Other da ta, which could have weighted the evidence toward a reduced risk, was also withheld by CRT, although the results of these studies -were presented in part in November 1980 at a CIIT conference. One study dealt with evidence suggesting that fo.~al- debyde was not genotoxic. Another report was an epidemiology study which yielded negative results. In the view of CPSC, detai led documentation ~ f these resuL ts have, as ye c, not been released. One of the most important pi eces of scientific information is the pathology s [ides from the initial CIIT study. Six patholo- gist~ representing IRI`G there permitted to review the s rides in the middle of the study (January 1980~. No further review has been permitted since then, although CPSC and the IRLG Formaldehyde Task Group had requested such a reviews. CIIT maintains the position that, as a matter of policy, they lo no~c release data to any one party alone (i.e., the government); that the release of data has to be made public, and is nor privy to any one interes t group. 2. credibility of assessments, likelihood that interested parties would accept them as definitive. Hazard Ident if ic at ion The Formaldehyde Panel, IARC, NIOSH, the Sel ikof ~ and Hammond Committee Report to the American Cancer Society, the heads of NCI, NIEHS, and NCTR and three distinguished scientists at New York University Medical Center have all concurred that formaldehyde is a definite animal c arc inogen and should be cons idered to pose a human carcinogenic risk. Some interested parties could dispute these assertions, but it is likely that most would not.
32 Quantification and Characterization of Risk to Humans Representatives of CIIT and the formal'1ebyde industry were highly critical of CPSC'~ quantitative risk assessment. As detailed in Question C.11, two main points of criticism focus on the choice ~ f the dose-response curve, and the quality of CPSC ' s exposure da t a . What was the extent of diversity of policy orientations . .. . represented within the assessment croup i tself . What was the decree to which interes t pressures would be exerted from outside the assessment group? What {:~5 the ·~s~onsivenes s of the assessment to these diverse interests7 _ Both the hazard identification and the quantification of risk to huffs were performed solely by federal scientists. Hence, the extent of diversity of policy orientations was quite limited. It does not appear that there was any degree of integer t pressure exerted from outs ide the group. 4 What were the time ant resources necessary to comr~le to the . . . risk assessment? Hazard Identification Assessment The federal panel consisted o f 16 scientists working part-time actual percentage unknown) for 9 months . Characterization of Risk to Humans Seven CPSC staff persons worked full or part-time over two years collecting and reviewing data and reviewing comments. Me cancer risk assessment document (excluding the effort needed to obtain exposure data) required one person working full time for one year. 5. Responsiveness of assessment agenda to public concerns . . . interest group concerns, professional codeine, an~ e~er~euce of new scientific data. It appears that CPSC responded to public concerns regarding irritation and sensitivity effects with cautious but deliberate action, and it respond to the newly emerging information on formaldehyde as a Potts tial human carcinogen with swift, decisive action.
33 As described in detail in Question A.2, the first public concerns voiced in October 1976 regarded irritation effects from formaldehyde release in UFFT residences. Wer the next couple of years, several thousand similar complaints were filed with CPSC. In order to obtain more detailed information on the full extent of human heal th concerns, CPSC asked the National Academy of Sciences, in May 1979, to prepare a comprehensive report. The report -~as completed in March 1980. Public hearings were 'nerd from December 1979 through February 1980 as industry, state and local government officials, scientists, and others testified on the question of UFFI ant formaldehyde release. As a result of all this activity, on June 10, 1980, CPSC issued ~ proposed rule '.~ch would require UFFI manufacturers to label their products giving specified performance and technical information to prospective purchasers. This rule was never made effective because in the view of CPSC the new emerging cancer data required more stringent rutemaking. lathe first preliminary data from the CIIT rat studies was presented to CPSC in October 1979. One month later, a briefing package prepared by s gaff was sent to the commits loners . In January 198~), an IRLG task force visited CIIT to verify the findings. In April 1980, the CPSC-requested Formal`lebyde Panel was formed to assess all current literature and make reco~enda- tions as to the severity of health ef facts. T'ne final report in November 1980 concluded that formaldehyde was a potential human carcinogen. In February 1981, CPSC published a proposed rule to ban UFFI. In October 1981, the final CPSC cancer risk assessment was completed. TO April 1982, CPSC published a final rule to ban IJFF1 . 6. Ability of the risk assessment to identify research needs. Interest in the possible harmful effects of formaldehyde generated an abundant 1 is tiny of research needs . The CPSC-sponsored National Academy of Sciences study on health effects o f Cormal- dehyde (March 1980 ~ identified twelve research needs including the areas of carcinogenicity, reproduction and teratogenic effects, sensitivity of population subgroups, pharmacokinetic studies, sources and fates of formaldehyde, and percutaneous penetration. The Formaldehyde Panel report identified research needs in teratology, animal reproduction, human reproduction, mutagenicity, carcinogenicity (pharmacokinetics, other routes of administration, neoplastic transformation of mammalian cells), epidemiologic research, and the ability of formaldehyde to interac t with other pal lutants .
3/+ A major research effort has already begun with the goal of obtaining more definitive results from epidemiological data. Both NCI and NIOSH are examining data on large numbers of workers with a history of high formaldehyde exposure. '<any of the other needs have been or are being assessed by research programs both within and outside the government. 7. Extent to which risk assessment impeded or facilitated regulat ion. Informed 'nigh level CP5C staff believe that the risk assessment did both. It facilitated regulation by giving the assessors an idea of the range of risk. It impeded regulation because the calculated numbers create a false impress ion of certainty that leaven the assessment vulnerable to attack by critics opposed to regulation. Staff felt there was an urgent need to define the limits of risk assessment more clearly and accurately. 8. Were related risk assessments cons is tent? This question has been covered in Questions B. tO and D.2 . 9. Extent to which there is an explicit distinction between weights accorded to scientific factors and policy factors. As mentioned in Question C. S. ma jar areas of uncertainty affecting the resmelts of quantitative risk assessment are choice of the low dose extrapolation model and choice of the interspecies conversion factor. Choices can be made by adhering to two criteria: (1) how well does the model fit the data, and (2) how biologically plaus- ible is the model. Apart front these science-oriented criteria, choices can also be made solely for policy reasons. For example, an agency may wish, in the face of uncertainty, to Parr on the side of caution. Therefore, it may adopt a policy of choosing a Conservative mode ~ . The CPSC risk as senescent calculates two dose-response curves: the linearized multistage cunre9 and the "best-fit-of-data" cunre. It strongly endorses the linearized multistage curare, which is the more conservative model. However, CPSC ' s stated reason for endorsing this model was purely scientific. It considered it to be more biologically plausible than the "best- fit-of-data" motel. If a policy factor did go into the teci~ionmaking process, it was not explicitly stated in the risk assessment document. (:PSC ~ s choice of an interspecies conversion factor was based on a paper by Mantel and Schneiderman (1975~9 although no concrete scientific reasons were given for the choice . ~ t is not clear 'low Conservative this factor is relative to other conversion factors.
3S 10. Mode and frequency of communication between assessors and regulators . Hazard Identification lye Formaldehyde Panel, in its appraisal of formaldehyde as a potential l`uman carcinogen, appeared to be acting entirely independent ty o f agency regulators . Characterize ion of Rick CPSC appears to 'nave organizational separation between the assessors and the regulators. lathe assessors, consisting of staff from the Directorate of Health Sciences, performed their function wi thou t apparent contac t wi th or guidance by the program team and the commissioners, who are the regulators.
36 SOME: IMPORTANT REFERENCES CITEI) IN FORMALDEHYDE CASE STUDY Crump, K. S . and W. W. Watson. 1979. Global 1979. ~ fortran program to extrapolate dichotomous animal carcinogeraicity data to low doses. Dept. of Mathematics and Statistics, Louisiana Tech Univers ity. EPA (Environmen~at Protection Agency). 1976. Health risk and economic impact assessments of suspected carcinogens. Fed. Reg. 41: 21402. EPA ~ Environmental Protect ion Agency) . Of fice of Research and Developments Carcinogen Assessment Group. August 8, 1980. The Carcinogen Assessment Group's method for determining the unit risk estimate for air pollutants. External review copy prepared for Office of Air Quality Planning and Standards and Office of Air, Noise and Radiation. F})A (Food and Drug Administration). 1979. Chemical compounds in Eood~producing animals 9 criteria and procedures for evaluating assays for carcinogenic residues. Fed. R=go 44 17070. Food Safety Council, Final Report of Board of Trustees. June 1989 0 ~ Proposed Food Safety Evaluation Process. Gibson 9 J o E a December 15, 1981. Le t te r from Chemical Industry Ins t i tute ~ f Toxico logy (CIIT ~ to Consumer Produc ~ .Sa fe ty Commiss ion ~ CPSC ~ . Gibson, J. E. February 1;, 19820 Letter from Chemical Industry Ins t itute o f Toxico logy ~ C IIT ~ to Nancy S teort s, Cha irman, Cons ume r Product Safety ·^ommiss ion (CPSC ~ . IARC (International Agency for Research on Cancer) 9 the Working Group O May 1982. Monograph on the Evaluation of the Carcinogenic Risk of Chemicals to Humans. Some Industrial Chemicals and Dye Stuffs: Formaldehyde, pp. 345-3890 TRLG (Interagency Regulatory Liaison Group), Work Group on Risk Assessment. 19790 Scientific bases for identification of potential carcinogens and estimation of risks. Fed a Reg 0 44:39858. Mantel, N. and M. A. Schneide~=an. 19750 Estimating "Safe" levels, a hazardous undertaking. Cancer Res. 35: 1379-138f'. NAS (National Academy of Sciences). March 1980. Formaldehyde: An Assessment of Its Realth Effects.
37 NIOSH (National Institute for Occupational Safety and Health). Current Intelligence Bulleting #34. April 15, 1981. Formaldehyde: Evidence of Carcinogenicity . NYU (New York University). R. E. Albert, A. R. Sellak~ar, S. Laskin, M. Ruschner, N. Nelson and C. A. Snyder. 1982. Gaseous formaldehyde and hydrogen chloride induction of nasal c ancer in the rat. J. National Cane. Inst. 68 :597-602. OSIP (Office of Sc fence and Technology Po ticy) . D. R. Calkins, R a L a Dixon, c: . R. Gerber, 1) . Zarin, and G. S . Omenn. 1980 . Indentification, characterization, and control of potential human carcinogens: A framework for federal decis ionmaking. J . National Cane. Inst. 64 :169-176. Report of the Federal Panel on Formaldehyde. November 1980. Submitted to S. B. ICing, Chairman, U. S. Consumer Product Safety Commission. Also published February 1982, Environmental 'Realth Perspectives. 43: 139-168. Selikof f, I . ~ . and E. C. Hammond. February 25, 1981. Carcino- genicity of formaldehyde: final report to the American Cancer Society by the Environmental Cancer Info~!uat ion Unit . Environmental Sciences Laboratory, Mount Sinai School or Medicine. Todbunter, J. A. February 10, 1982. Review of data available to the administrator concerning formaldehyde and di(2-ethythexyl~phtha- late (~EHP). Memorandum to Anne Gorsuch, Administrator, En~rironmenta 1 Pro tee ~ ion Agency .
