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.
CHAPTER 5 APPLICATION OF THE MODEL TO CHEMICAL HAZARDS The Committee on the Scientific Basis of the Nation's Meat and Poultry Inspection Program (NRC, 1985 ~ described the various sources of chemical residues in meat and poultry products and the approach used by FSIS to control them. That committee also made several recommendations for improving the FSIS inspection program and urged the adoption of formal risk-assessment procedures to provide maximum protection of public health. Specifically, the committee recommended that risk assessment play a major role in the establishment of limits for chemical residues in meat and poultry products destined for human consumption, in the prevention and characterization of hazards, in the setting of priorities for controlling residues, and in the design of sampling methods. This chapter contains a discussion of risk assess- ment as a guide to the management of chemical hazards in poultry products, criteria for j udging the safety of poultry products con ~ tanning residues, some approaches to ensuring that safety criteria are met, and the types of data and analysis needed to assess the public health impact of chemical residues in poultry products. It also identifies the necessary elements of a risk-management program and describes the risk-assessment methods needed to establish this program . It does not include cons iteration of current FSIS inspection which is evaluated in subsequent chapters of the report. &; ~ GENERAL METHODS FOR ASSESSING THE PUBLIC HEALTH RISKS OF CHEMICALS There is extensive documentation on deaths and injuries from accidental poisonings by household products, pesticides, and thera- peutic agents. Ordinarily there is little difficulty in estimating the relationship between the extent to which these substances are used and the frequency of poisonings and in documenting the association between a given exposure and a given poisoning when the effect is immediately observable (i.e., acute). It is more difficult to assess risks associated with chemical exposures when no immediately observable effects are produced when the fact or degree of exposure is itself highly uncertain. Since most chemical exposures associated with residues in poultry products are uncertain, the risks must be predicted and those predictions used to set health protection standards. 100
101 Although the methods used to predict chemical risks are uncertain (e.g., because of incomplete data, the need to extrapolate beyond data, and the lack of knowledge concerning the extent of future human exposure), they are based on a strong scientific foundation (NRC, 1980b, 1983~. The safe use of products, including food ingredients, pesticides, and drugs, depends upon these methods of risk prediction and their use in the establishment of low risk (or safe) exposures (FSC, 1980~. People are exposed to a large number of naturally occurring and man-made chemicals through poultry products and other environmental media. If they are to be protected from the possible adverse effects of these substances, methods to assess the risk assessment of such exposures must be applied. It is a premise of this report that predictive methods developed for and widely used in many areas of public health protection are appropriate for assessing the risks of exposure to chemical residues, establishing appropriate health protection standards for such residues, and guiding the development of programs to manage the risks presented by the residues. Parts or all of the s premise have been adopted by the Food and Drug Administration, the Environmental Protection Agency, and other government agencies charged with protecting consumers from such residues, especially for risk assessment and the establishment of standards. However, there are important limitations in the methods themselves and in their application to specific problems, including those associated with poultry products. THE COMPONENTS OF RISK ASSESSMENT A National Research Council committee described four basic components of risk assessment in the federal government: hazard identification, dose-response assessment, exposure assesment, and risk characterization (NRC, 1983~. Figure 5-1 shows the relationships between these components of risk assessment, research, and risk management. Hazard Identification Toxicity. All chemical substances, whether natural or man-made, can cause some form of biological injury under some conditions of exposure. The purpose of the first phase of risk assessment is to collect and evaluate information on the inherent toxic properties of chemicals of interest. Identifying these properties is not equivalent to identifying possible risk. Thus, it should not be assumed that a substance displaying toxicity presents a risk to human health. All steps of risk assessment must be completed before any statement can be made about risk.
102 0 o So o ° E ~ ~ o _ It he 5 In In U) In In fir: 0 Ca, 3 _ O ~ Co 3 0= ~ =0 CON ~ 0 ~ S ~ 5: I G CO 2 E ° en Al C AL C O ~ 0 ° a, 2 - ° ~E ~ ~ ~ 0 ~ _ ,,, = ~ ~ O ~ ~ ~ 0 ~ ,1 0 0 ~ 0 0 o C 0 =0 s So 4, ~ ~ _ _ _ Cat ~ ~ C,~3s . . 4, ~ ~ _ 0 a, ~s, ~ ~ 0 ~ ~ o3 0 at oe O ~ ~ 2'2 0 0 0 3 I: a, ~ u, 0 a~ 0 0 0= Ji __ o' 0 0 a~ S0 o ~ C ~ 2 ~ C O O ~ O _ O i~ ~ o s E Bo ~ 0 ° a' cn ~Q V, a' ~Q ;0 - E ~ c ~s ' 0 0 4, C o o o ,,, o o ~ ' c 2 a; - o ~ o ' E 0 0 s o [L ~ C, Q 1 U~ ~Q 00 - o U) - Za) o
103 There are two principal sources of information about the toxic properties of chemical substances: investigations of exposed human populations or individuals Epidemiological or clinical investigations and experimental studies in laboratory animals or other biological systems. Knowledge of the molecular structure of some substances may be helpful in predicting toxic properties, but this aspect of toxicological science is still immature (Asher and Zervos, 1977; Klaassen and Doull, 1980; NRC, 1980b; OSTP, 1985) e Data on Humans. Well-conducted Epidemiological and cl inical investigations provide pertinent data for the evaluation of the hazardous properties of environmental agents. Epidemiological studies have provided convincing evidence about the cancer-causing properties of such agents as cigarette smoke, asbestos, vinyl chloride, and diethylstilbestrol (DES) and about the teratogenic effect of thalidomide. Clinical investigations of exposed persons have provided information on the toxicity of consumer and industrial products (MacMahon and Pugh, 1970; OSTP, 1985~. There are, however, the following limitations in the use of both Epidemiological and clinical data for identifying the toxic properties of chemical substances: · The deliberate, controlled exposure of human beings to identify toxic effects is, with few exceptions , unethical. Exceptions include short-term exposures to substances (e. g., certain drugs) that produce mild, fully reversible effects. Epidemiological and clinical studies cannot be conducted on newly introduced chemicals or chemicals for which there has been little or no previous human exposure. Accurate data on the chemical nature of the substances to which populations or individuals may have been exposed and on the intensity and duration of their exposure are rarely available, especially when exposures have taken place in the distant past. It is difficult to provide proper controls for Epidemiological studies when the cause-and-effect relationships of a chemical cannot be easily established, as is the case for chemical workers who may be exposed to unknown amounts of other substances in addition to the chemical of immediate interest. In investigations of diseases with long latent periods, such as cancer, it is usually difficult to follow exposed persons for periods long enough for the disease to reach a clinically detectabl e state and thus for firm conclusions to be drawn about the presence or absence of an effect. Epidemiological studies cannot generally detect small but possibly important changes unless the study population is very large (rarely practicable) or the resulting disease is rare (e.g., occurrence of vaginal adenocarcinoma during adolesence in daughters of mothers given DES during pregnancy).
104 Because of these limitations, public health officials must frequently turn to experimental data for information about the toxic properties of chemicals in the environment. Experimental Animal Data. Laboratory animal studies have an advantage over epidemiological and clinical investigations (NRC, 1980b). The experiments can be controlled so that causal relationships between exposure to a substance and toxicity can be established, and the relationship between the intensity and duration of exposure and the magnitude of toxicity can be studied (NRC, 1980b). (See section on Dose-Response Assessment below.) Animals can be studied for functional changes or killed at var' ous times during the experiment and examined for the presence of a variety of biological injuries and pathological changes that are not observable clinically. In rats and mice, the effects of lifetime exposures to an agent can be detected in 2 or 3 years - - the normal lifespan of these species (OSTP, 1985 ~ O These advantages of data from animal studies are partially offset by the obvious fact that animals are not biologically identical to humans. To conclude that some agent can cause a certain form of toxicity in humans because it does so in laboratory animals requires inclusion of some untested assumptions about the biological similarity of various mammalian species. There is evidence that results from animal studies are often applicable to humans. For example, most substances known to be carcinogenic in humans are also carcinogenic in animals. Similar examples could be collected for a variety of other toxic effects (NRC, 1983; OSTP 9 1985 ~ . Exceptions are also common, however . Unless human data are adequate to refute a specific finding of toxicity in animals or there is some other biological reason to do so, it is reasonable to infer a potential for toxicity in humans from observations in experimental studies of animals. Animal experiments are the principal source of toxicity data for assessing the human risks and safety of pesticides, food and color additives, and food and drinking water contaminants, and there is no reason not to rely on such data for similar assessments of chemical residues in poultry products. Manifestations of Toxicity and Tests to Identify Them. Systematic investigation of the toxicity of a chemical substance usual] y begins with a determination of its acute toxicity, which includes a determination of the dose of a substance that in a single exposure (lethal dose) will cause the deaths of the exposed animals within a short time after administration. At successively lower levels of exposure, the percentage of animals that respond decreases correspondingly. The relationship between dose and the percentage of the animal population that dies is called the dose - response relationship for the end point in question- - in this case, death. The range of doses over which deaths are observed and the shape of the dose-response relationship vary from one substance to another, and both are critical to an assessment of a substance's capacity to cause death in an exposed population.
105 Short-term exposures (i.e., one or several exposures repeated over several days or a few weeks) to chemical substances in amounts lower than the lethal dose may produce toxicity that ranges from mild (e.g., reversible eye or skin irritation or transitory nervous system disorders) to severe (e.g., irreversible blindness or liver damage). The toxic manifestations of a short-term exposure to a chemical depend on the intensity and duration of the exposure and the characteristics of the chemicals (Doull et al., 1980; Loomis, 1978; NRC, 1983~. Studies of short-term exposures are generally followed by studies of long-term exposures to lower doses (chronic toxicity)(NRC, 1980b). These experiments are designed to detect effects that arise after many repeated, sometimes daily, exposures that occur over various periods - - from approximately 10% of an animal ' s lifespan (subchronic toxicity studies) to its ful 1 lifespan (2-3 years for rodents; several times longer for other commonly used animals such as dogs and monkeys) or effects resulting from short- term exposures that do not become clinically detectable until much later (e. g., for DES) . Chronic effects may range from relatively mild conditions to progressive and lethal lesions such as cancer . The form of inj ury or disease and its dose-response characteristics are specific to the chemical, but both these features of chemical toxicity can be altered by characteristics of the exposed animal (e.g. 9 its genetic background, health status, age, or sex) and its environment (e.g., the nature of its diet or the presence of other environmental agents). Such interspecies and intraspecies differences in toxic response and dose-response characteristics for a given substance have strongly influenced the methods used by public health scientists to assess risks. (See section on Risk Characterization below). Subchronic toxicity experiments can reveal much about the potential of a substance to inj ure various organs and systems of the body, including the developing fetus, but they cannot reveal whether a substance can induce cancer (OSTP, 1985), except when it is unusually potent. Determination of carcinogenicity usually requires that test animals be exposed for most of their lifetimes. There are now many well-validated test systems used worldwide by both public health agencies and private concerns to establish the acute, subchronic, and chronic toxicities of chemical substances (EPA, 1982; FDA, 1982b; NRC, 1977~. The most thoroughly tested substances are those that must, by law, be evaluated before they can be introduced into commerce (e.g., food and color additives, drugs, and pesticides)(NRC 9 1985~. Quality and Extent of Data. The quantity and quality of toxicity data available on different substances vary greatly. For a few substances the data base may be extensive and may include results of all the standard toxicity tests as well as data specific to each substance under evaluation, whereas for other substances, the data base
106 may include, at best, only a determination of acute toxicity or no significant toxicity data at all. For most important industrial chemicals, the quantity and quality of available data fall somewhere between these two extremes, but more toward the lower end of the scale (NRC, 1984~. There is no straightforward way to define the adequacy of a given data base. A data base may be sufficient to determine the safety of a certain use or type of chemical exposure but may be inadequate to determine the risk presented by another use or type of exposure. For example, many agents tested for occupational risk have not been examined for their potential to cause chronic toxicity or birth defects. If such substances show up in the poultry supply because of environmental pollution, the absence of information on chronic effects and their effect on the developing fetus would be of great concern. Similarly, the absence of chronic toxicity data on many chemicals is of concern if the chemicals are found to be present in poultry products to which people could be chronically exposed. The absence of data does not imply that a risk exists, but it does mean that risk (and therefore safety) cannot be ascertained with an adequate degree of confidence. Various methods are used to compensate for such data gaps. These methods are described below in the section on Risk Characterization. Hazard Evaluation. This phase of risk assessment includes a critical review of clinical 9 epidemiological, and experimental toxicity data and identification of the inherent hazardous properties of a substance, the degree to which these hazards are known, and the uncertainties in the data. A critical feature of this process are j udgments about the strength of inferences for human risk from data derived from animal studies . At this stage of risk assessment, no attempt is made to determine the degree of human risk that might be associated with the substance under evaluation. Dose-Response Assessment For an exposure of a given duration, the frequency and severity of toxic effects in an exposed population (the risk) generally increase with increasing dose. Toxic effects may also change as exposure increases. The dose-response relationship is critical to risk assessment and must therefore be well defined. Well-defined dose-response relationships can rarely be obtained from ep~demiological studies because of uncertainty regarding the exposures that produced the toxic responses seen. Therefore, experimental data are the primary sources of dose - response information for risk assessment. The dose of a toxic agent can be expressed in various ways. Most commonly it is presented as the weight (mg) of the agent taken into the body per unit (kg) of body weight (bw) of the human or test animal per unit of time (usually, per day), e.g., mg/kg low/day. Dividing intake by body weight permits comparisons to be made among species with
107 different average body weights. Other measures of dose, such as mg/kg bw over a lifetime, mg/m of body surface area, parts per million (ppm) in air, water, or diet, are used less often. For most toxic effects, a threshold dose is the amount of exposure that must be exceeded before a specific toxic effect is produced. For other effects such as cancer, however, there appears to be a biological basis for rejecting the threshold hypothesis. As currently practiced, in fact, carcinogenic risk assessment is generally based on the assumption that there is no threshold dose. Rather than entering more fully i nto the complex debate on thresholds, the committee has simply adopted the positions taken by the major regulatory and public health agencies and other NRC committees , i. e., the absence of a threshold for carcinogens . A critical part of dose-response assessment is identification of the dose that produces no adverse response, i.e., the no - observed- effect level (NOEL), in the treated animals. The NOEL is generally taken as the starting point for risk assessment of virtually all effects other than cancer. It may approximate a threshold dose for the animal population under study, but for a variety of reasons the experimentally determined NOEL is probably not identical to the true threshold dose. For carcinogens (even those for which an experimental NOEL for other toxic effects has been determined), the dose-response data are treated differently. The size of the increase in toxic effects at various low doses, where the risk per animal (and by extension, per person exposed) is quite small, is generally of greatest interest but not directly observable because of practical li mits on experi- mentation. Carc~nogenicity data from animal studies generally show that increasingly high doses cause a corresponding increase in the incidence of cancers. However, the doses used in animals are exceedingly high in terms of human risk to compensate for the fact that only small numbers of animals can be used in experimentation of this type (OSTP, 1985~. For example, if 50 animals are exposed to a dose of a carcinogen and 5 develop tumors, the risk is 10% (if no control animals develop a tumor). A cancer risk near 10% would be intolerable in any human setting, but this is about the smallest risk that can be reliably detected in animal experiments of practicable size (OSTP, 1985). The experimentally determined relationship between dose and risk at high doses must therefore be used to assess risk for dose levels corresponding to human exposures. This requires the use of certain mathematical models of the dose-response data (FSC, 1980; NRC, 1980b; OSTP, 1985~. These models generally provide unit risk estimates, i.e., estimates of cancer risk per unit of dose (such as the incidence of cancer at a dose of 1 mg/kg low/day over a lifetime). The models most widely used for low dose carcinogenic risk assessment are based on
108 assumptions that there is no threshold and that risk at very low doses increases in direct proportion to dose. Several models meet this criterion. EPA uses the linearized multistage model, which incorporates an upper 95% confidence limit on the estimated linear term. Table 5-l presents risks per unit of low dose exposure predicted by this model for substances that are potential contaminants of poultry products. Models used by FDA and the National Research Council's Safe Drinking Ilater Committee (NEt.C5 1980a) would yield unit cancer risks close to those shown in Table 5-1. It Is not possible to demonstrate that any mathematical models are fully in accord with biological reality. Because this subject has been discussed elsewhere (NRC, 1980a; OSTP, 1985), the committee simply notes in this report that certain models are widely used in risk prediction and that they are generally interpreted on the basis of little direct evidence as providing upper limits on low dose risk, although recent data suggest that this may sometimes be wrong (J. C. Bailar, Harvard School of Public Health, personal communication, 1987~. Policy choices needed in the face of scientific uncertainty have also been discussed in another National Research Council report (NRC, 1980b). The dose-response assessment phase of risk assessment thus generally concludes with a determination of NOELs (for noncarcinogenic effects) and of estimates of risk per unit dose (unit risks) for cancers. In both determinations there are important uncertainties that need to be specified in the report of the risk assessment. Since many of these uncertainties concern the data on which these dose-response estimates are based and are therefore chemical- specific, they must be defined by experts who have studied a specific substance. Other uncertainties are generic (e.g., some are inherent in models for extrapolating from high to low doses) and therefore apply to all chemicals. Exposure Assessment Exposure assessment is a highly complex subject, and is reviewed here only to the extent necessary to prepare for the later discussion of chemical residues in poultry products. In this phase of risk assessment, knowledge of the magnitude and duration of human exposure to environmental agents and, most importantly, the dose that results from this exposure, is essential As used herein, the term exposure describes a person's contact with a medium (e.g., poultry) containing a chemical. The magnitude of the dose that results from the exposure depends on several factors, which are described in the following paragraphs O To estimate dose, the possible routes of chemi cat ~ ntake must first be identified. For residues in poultry, ingestion is the only route of concern. Occupational exposures might include inhalation, derma] contact, or other routes of exposure, but these routes are not germane
109 TABLE 5-1 Unit Cancer Risks and Strength-of-Evidence Categories for 47 Chemicals, as Evaluated by the EPA Carcinogen Assessment Groupa Level of Evidenceb Unit Cancer Risk . . . Compound Humans Animal s per mg/kg bw/dayC Acrylonitrile L S 0. 24(W) Aflatoxin B1 L S 2, 900 Aldrin I L 11.4 Allyl chloride -- -- 1.19 x 10 2 Arsenic S I 15 (H) Benzo ~ a ~ pyrene I S 11 . 5 Benzene S S 2.9 x 10-2(W) Benz idene S S 2 34 (W) Beryllillm L S 2 .6 (W) Cadmium L S 6 . 1 (W) Carbon tetrachloride I S 1. 30 x 10~ Chlordane I L 1.61 Chlorinated ethanes: 1,2-Dichloroethane I S 9.1 x 10-22 Hexachloroethane I L 1.42 x 10- 1,1,2,2-Tetrachloroethane I L 0.20 2 1,1,2-Trichloroethane I L 5.73 x 10- Chloroform I S 8.1 x 10-2 Chromium VI S S 41(W) Dichlorodiphenyltrichloroethane (DDT) I S 0 34 Dichlorobenzidine I S 1.69
110 TABLE 5-1 (continued) Level of Evidenceb Unit Cancer Risk Compound Humans Animals per mg/kg bw/dayC Dieldrin ~ S 30.4 Ep~chlorohydrin I S 9.9 x 10-3 Bis(2°chloroethyl~ether I S 1.14 Bis~chloromethyl~ether S S 9,300(In) Ethylene dibromide I S 41 Ethylene oxide L S 3.5 x lO~l(In) Heptachlor I S 3 . 37 Hexachlorobenzene I S 1. 67 Hexachlorobutadiene I L 7.75 x 10-2 Hexachlorocyclohexane: Technical grade ~ 4.75 Alpha isomer I S 11.12 Beta isomer I L 1.84 Gamma isomer I L 1.33 Nickel refinery dust S S 1.05(W) Nitrosamines: Dimethylnitrosamine I S 25.9(not by q )~ Diethylnitrosamine I S 43.5(not by q Dibutylnitrosamine I S 5.43 N-Nitrosopy~rol~dine I S 2.13 N-Nitroso-N-ethylurea I S 32.9 N-Nitroso-E-methylurea I S 302.6 Polychlorinated biphenyls (PCBs) I S 4.34 Phenols: 2, 4,6-Trichlorophenol I S 1.99 x 10-2 Tetrachlorodibenzo-~-dioxin (TCCD) I S 1.56 x 10+5
111 TABLE 5 - 1 ~ continued) Level of Evidenceb_ Unit Cancer Risk Coumpound Humans Animals per mg/kg bw/dayC Tetrachloroethylene I L 5.1 x 10-2 Toxaphene I S 1.13 Trichloroethylene I L/S 1.1 x 10-2 Vinyl chloride S S 1.75 x 10~2(In) aAdapted from EPA, 1985. bS = Sufficient evidence; L = Limited evidence; I = Inadequate evidence, according to the International Agency for Research on Cancer. CUCRs are 95% upper-limit slopes based on the linearized multistage model. Calculations of these slopes are based on data from oral studies in animals except for those indicated by In (animal inhalation), W (human occupational exposure), and H (human drinking water exposure) Slopes for humans are point estimates based on the linear nonthreshold model. Not all of the carcinogenic potencies presented in this table represent the same degree of certainty. All are subject to change as new evidence becomes available. dq is the 9S~ upper-bound confidence limit of the linear parameter .
112 in this report. Exposure assessment is designed to yield dose estimates for both short- term and long- term studies . Daily doses of chemical residues received by humans through poultry consumption are estimated by determining the chemical concentrations In various poultry products and the average daily intake of each product. FSIS's National Residue Monitoring Program has collected information on 20 chemicals found as residues in young chicken carcass samples examined from 1979 to 1984 (Table 5-2~. Internal absorption also influences dose. Ingested substances must pass through the gastrointestinal wall to produce systemic effects, but there are differences in the rate and completeness with which different substances pass through this barrier. Direct measurements of absorption are rarely available and cannot be obtained without conducting experimental studies in humans. Therefore, it is common either to adopt absorption rates from animal studies of compounds with similar chemical and physical characteristics or to assume that absorption is complete (Calabrese, 1983; Nisbet, 1981~. Precise identification of the population to which the risk assessment will be applied is another important requirement of exposure assessment. This may be the general population or a special subpopulation, such as infants, believed to be at special risk. In this chapter it is assumed that the general population will be exposed to chemical residues in poultry, but the concept and principles described herein apply just as well to any defined subpopulation. The general population contains not only healthy adults but also infants, chic dren, pregnant women, the elderly, the ~mmunosuppressed, and the chronically ill, which represent the full range of susceptibilities to a toxic agent. In such a population, there is likely to be a wider range of responses to a toxic agent than in a group consisting primarily of healthy adult males , e . g . , certain worker groups. Risk assessment based on animal data from highly controlled experiments (e. g., studies using inbred strains, homogenized feed, or uniform holding conditions, or excluding other agents that might act synergistically with the test substances) will reflect an even narrower range of susceptibilities . For risk assessment, therefore, it is important to know the characteristics of the exposed population. Finally, for toxic agents with thresholds and for most nonthreshold carcinogens it is important to know not only the mean dose received by the population but also the distribution of doses, if one is to determine whether even the most highly exposed individuals are exposed to subthreshold doses. Risk Characterization Noncarcinogens. Since there are no methods for estimating risks for low doses of noncarcinogens, it has become the practice to divide the experimentally determined NOEL for these substances by a large
113 Table S-2. Examples of Chemical Residues in Young Chickensa Number of Contaminated Samples Found. by Concentration (ppm) Chemical and 0 0. 01- 0.11- 0. 21- 0 31- 0. 51- 1. 01- 2. 01- 2. 51- Year Reported Tissue 0.10 0.20 0 ~ 30 0. 50 1.00 2.00 2 . 50 5 . 00 PESTICIDES: Benzene ~ hexachloride 1981 Fat 467 S 1982 Fat 432 2 1983 Fat 422 2 1984 Fat 448 6 Chlordane: 1981 Fat 469 3 1982 Fat 433 1 1983 Fat 420 3 Dieldr~ n: DDT: 1 1979 Fat 221 10 1 1980 Fat 582 14b 1981 Fat 458 14 1982 Fat 426 7 1983 Fat 412 12 1984 Fat 428 26 1979 Fat 211 19 1 1 1980 Fat 502 92b 2c 1981 Fat 409 60 2 1 1982 Fat 414 19 1 1983 Fat 375 45 2 2 1984 Fat 391 61 1 Endrin: _, ~ 1980 Fat 595 1 1981 Fat 470 2 1984 Fat 452 2 Heptachlor: 1979 Fat 231 1 1980 Fat 590 6b 1981 Fat 768 4 1983 Fat 422 1 1 1984 Fat 447 7 Lindane: 1979 Fat 231 1 1980 Fat 588 8b- 1981 Fat 470 2 1 1983 Fat 421 2 1 1984 Fat 447 7 MethoxYchlor: 1979 Fat 231 1 1984 Fat 452 1 Polychlorinated biphenyls: 1983 Fat 423 1 1 1 aFrom FSIS, National Residue Monitoring Program, unpublished data, 1979-1984. bA concentration range of 0.01-0.30 ppm was reported.
114 TABLE 5-2 (cont.) Number of Contaminated Samples Found. by Concentration (ppm) Chemical and 0 0.01- 0.11- 0.21- 0 31- 0.51- 1.01 2.01- 2.51- Year Reported Tissue 0.10 0.20 0.30 0.50 1.00 2.00 2.50 5. 00 PESTICIDES (cont.) Hydroxychlorinated biphenyls: 1979 Fat 225 7 1980 Fat 588 8b 1981 Fat 469 3 1984 Fat 450 4 p-Chlorophenol: 1981 Liver 0 6 3 1983 Liver 247 9 1 1 1 ENVIRONMENTAL CONTAMINANTS: Arsenic: 1979 Liver 42 15 15 23 46 81 30 1980 Liver 41 51b 230~ 43d Be 4 1981 Liver 30 23 17 42 103 152 39 2 1982 Liver 30 14 10 22 71 82 16 1 1984 Liver 50 11 12 19 77 130 28 1979 Muscle 1 1 1 1980 Muscle 0 6b 1c 1981 Muscle 41 48 4 1 1982 Muscle 1 1 1984 Muscle 1 1 Mercury: 1979 Liver 1 Selenium: 1979 Liver 4 7 1 1979 Muscle 4 3 3 1 1 1979 Kidney 1 6 5 PHARMACEUTICALS AND ANTIBIOTICS: Chlortetracvcline: 1984 1982 1984 Decoquinate: 1983 OxYtetracycline: Liver 3 Kidney 393 1 Kidney 274 2 Liver 198 1983 Kidney 342 PYrante1 tartrate: 1983 1983 Sulfadimethoxine: 1981 1982 Sulfouimoxaline: Liver 38 24 11 Muscle 2 2 Kidney 313 Kidney 286 1 1981 Liver 4 1 1981 Kidney 312 1 1 1982 Kidney 286 1 17 72 169 63 1 . . . . . _ MA concentration range of 0.31-1.00 ppm was reported. dA concentration range of 1.01-1.50 ppm was reported. eA concentration range of lo51~2~00 pip was reported.
115 safety factor to estimate acceptable human doses ~ i . e ., acceptable daily intake, or ADI). This method was first used during the 1940s and 1950s to regulate food additives and pesticides (Lehman and Fitzhugh, 1954) and has since been extended to other categories, such as drinking water contaminants (NEC 1977, 1980a). The safety factor approach is based on several cons iterations . One is statistical; that is, a small sample of animals is not likely to exhibit a statistically significant number of reactions at some low dose even if the frequency with which such reactions occur is high enough to be of major concern in the human population. Another consideration is the likelihood that a human population will have a wider range of susceptibilities than the test animal group from which the NOEL is obtained. The human population is genetically more diverse than study groups of laboratory animals and contains certain subgroups (e.g., infants and the ill) that are likely to be more susceptible than healthy test animals. The human population is also exposed to a wider range of other environmental agents and lifestyle factors, e.g., it includes people who smoke, drink alcoholic beverages, and take medicines. Because of such additional exposures, the background of risk for human populations is different and often higher than that of test animals and may generally increase the susceptibility of the human population in comparison to test animals. It is also likely that the variation in threshold doses (doses below which no toxic response is observed) for a particular substance will vary more widely in the human population than in a small, genetically and experimentally homogeneous animal group. For these reasons, the threshold dose for the human population is likely to be lower than that approximated by the NOEL derived from studies in animals. To take into account all the uncertainties about the relative susceptibilities of the human and animal populations as well as the wide range of susceptibilities within the human population, the experimental NOEL is divided by a safety factor to derive an ADI for humans. If the NOEL is based on the results of a well-conducted chronic toxicity study and relates to an effect other than cancer, a safety factor of 100 is usually applied (NRC, 1980a) . If NOELs have been estimated for several animal species, the value for the most sensitive species (i.e., the lowest value of NOEL) is used to estimate the safe level for humans. For example, if a substance has been shown in chronic studies to cause liver damage in rats at high doses, and if the experimental NOEL for this effect is 100 mg/kg low/day, the ADI is set at 1 mg/kg low/day. This ADI is considered safe for humans in the sense that there is little likelihood of a toxic response in members of the human population exposed daily to the 1 mg/kg dose (NRC, 198Oa). If humans are exposed to an agent having this ADI, and if it is assumed, for example, that their exposure comes solely through ingestion of drinking water, that their weight is 60 kg, and that they drink 2 liters of water per day, then a 30 mg/liter concentrationlOf this agent in drinking water would be considered acceptably safe. Calculated as follows: 30 mg/liter x 2 l~ters/day = 60 mg intake per day. 60 mg/day . 60 kg bw = 1 mg/kg low/day.
116 A more refined calculation is needed if one is to consider additional sources of exposure. For example, if intake of a substance with an ADI of 1 mg/kg bw derives from both ingestion of 2 liters of water per day and consumption of contaminated poultry products, then 30 mg/liter in water would not be considered safe. To decide what level of contaminants could be permitted in both water and poultry so that the ADI would not be exceeded, it is necessary to know the daily intake of both water and poultry products and to make a policy decision about how to apportion the ADI between these two sources. Safety factors larger than 100 are commonly used when data on humans are not available and experimental data are limited (Calabrese, 1983; EPA9 1980; FDA, 1982b; NRC, 1986~. For example, a factor of 1,000 is used for NOELs derived from subchronic toxicity studies when no chronic data are available (NRC, 1980b). Larger factors may also be used for especially serious effects, such as cancer or birth defects. Teratogens may be effective following just a few exposures during critical times during pregnancy. Teratogenicity might be regarded as a form of acute toxicity, but the effects are permanent. Smaller safety factors are sometimes used for certain populations (e.g., relatively healthy, adult worker populations) believed to be less vulnerabl e than the general population and, rarely, when adequate data on NOELs are available from studies in human populations. In a distinct but related approach, risk characterization for noncarcinogens is accomplished by determining the margin of safety (MOS), that is, the numerical value derived when the experimental NOEL is divided by the actual human dose. A judgment is then needed to determine whether the MOS is sufficiently large to protect most members of the exposed population. Guidance on the adequacy of a given MOS can be obtained by comparing the commonly used safety factors for establishing ADIs to the MOS. For chronic toxicity, an MOS of 100 might be considered adequate, assuming that the data are adequate. There is an element of policy in the selection of safety margins, and any standard-setting activity should be seen to have a risk-management component. Under this system, the smaller the value of the MOS, the larger the risk. If the MOS is close to one, many members of the exposed population might be at high risk of toxicity. This is not true when the effects and dose-response characteristics of a substance are well-established in humans. Some air pollutants (e.g., carbon monoxide) can be protected against by using relatively small MOSs. If the MOS is only a small fraction of the NOEL, the risk may be low or zero, but there is no way to determine with complete certainty whether this is the case. Conversely, occasional exposures exceeding the ADI may not be associated with any risk at all. The limitations and uncertainties of this system are discussed later in this chapter Carcinogens. If a unit cancer risk (UCR), i.e., risk per unit dose, has been determined for a carcinogen, this value is multiplied by
117 the average daily lifetime human dose to derive an estimate of risk. In this case, the risk is estimated, and its value ranges from 0 and 1. Even for well-studied carcinogens, this estimate of risk is uncertain and it can not be claimed to be the true risk. EPA and FDA state that these values are upper bound risks and that the true risk is likely to be less. Despite the uncertainties, numerical estimates of risk are commonly used as decision-making tools for separating significant from negligible risks (NRC, 1980b; Rodr~cks and Taylor, 1983). USES OF RISK ASSESSMENT IN STANDARD SETTING Some standards are based on assessments performed by the methods described above. Following are some examples of the use of risk assessment in the setting of standards, some of which can apply to poultry products. Food and Color Additives FDA establishes ADIs based on data submitted by industry and application of the methods described above. Under law, no ADI can be established for carcinogens directly added to food including poultry (thus, no carcinogen can be directly added to food). For certain classes of unavoidable food constituents that are carcinogenic, FDA has permitted residue levels corresponding to doses presenting a negligible lifetime cancer risk as estimated by the methods described above (FDA, 1982a). FDA not only requires that the total human dose from all uses of an additive not exceed the ADI, but also that limits (tolerances) for additives be set at the lowest level at which the desired technical effect will be produced. Animal Drug Residues on Food FDA establishes ADIs for animal drug residues using the procedures described above for food and color additives. Tolerances for residues in individual tissues are based on the expected distribution and concentrations of residues in poultry products (learned from metabolism and pharmacokinetic studies) and on the criterion that all sources of exposure together do not exceed the ADI. Carcinogenic drug residues are permitted up to a level corresponding to an upper limit lifetime risk of 1 in a million (FDA, 19851. Pesticide Residues on Food EPA uses the same risk-assessment methodology used by FDA for food and color additives and applies it to data submitted by industry. An upper-bound lifetime risk of 1 in a million is used as a guide for determining acceptable residue levels of carcinogenic pesticides that do not meet the legal definition of a food additive, and risk-benefit balancing is a part of EPA's standard- setting procedure. ADIs are used . : -or noncarclnogens.
118 Environmental Contaminants - Food and other media may be contaminated with certain organic and inorganic chemicals of both industrial and natural origin. Limits on such agents in food, including meat and poultry products, are established by FDA. ADIs may be established, but standard safety factors may not always be used. This is especially true for various metals that present a substantial background exposure. For carcinogens, risks higher than those accepted for additives have sometimes been tolerated (see, e.g., FDA limits on PCBs in fish; FDA, 1979~. Tolerances for poultry products should take into account the magnitude of all other exposures to the same and related substances. Industry has not had to supply all the data necessary to establish acceptable limits on exposure, although some relevant data are submitted to EPA or required by the Toxic Substances Control Act. LIMITATIONS AND UNCERTAINTIES IN RISK ASSESSMENT Assessments of risks are based on data that vary in quality and quantity among different environmental chemicals. Some chemicals have been tested for toxicity far more extensively than others, often because of regulatory requirements for such testing in certain classes of chemicals before they can be accepted into commerce (e.g.9 pesticides 9 food and color additives, drugs). The Toxic Substances Control Act stipulates similar testing requirements for other classes of chemicals in commerce, primarily chemicals newly proposed for substantial commercial applications. The Department of Health and Human Services's National Toxicology Program is also producing data on commercial products but is limited largely to carcinogenicity testing. Despite these testing programs, it will probably be many years before a relatively complete and uniform toxicity data base will be available for these commercially important substances (NRC, 1984~. The inadequacies in both quality and quantity of data for many different chemicals introduce uncertainty, and the type and degree of uncertainty vary from one substance to another. For example, one chemical may have been well tested for subchronic toxicity but its possible carcinogenic effects may not have been studied, whereas another substance may have been tested for carcinogenicity but not for teratogenic effects O Test quality is another important consideration. Exposures of many people to high levels of a substance that appears not to be carcinogenic but has not been fully tested in adequately designed animal studies may be of greater concern than low levels of exposures of a few people to a substance found to be weakly carcinogenic in a well - designed study in animals O There are no hard- and-fast rules to
119 guide assessment of such exposures. As a result, many of the uncertainties in risk assessment are chemical-specific (e.g., data on the magnitude of human exposure) and cannot be reduced to general statements. An adequate risk assessment should contain a description of all the uncertainties and how they were considered in arriving at conclusions about risk. Risk assessments are also based on certain biologically plausible, as yet untested assumptions and inferences. Among the more important inferences are those concerning similarities between humans and test animals and the nature of dose-response functions at dose levels well below the observable range of experimental dose - response relationships see above section on Dose-Response Assessment). In the absence of information about toxicity in humans, it is biologically plausible to divide NOELs for animals by certa' n safety factors (see above section on Risk Characterization), but there is little empirical information for use in determining the most appropriate factor. In fact, without quantitative data on dose-response relationships for specific substances in humans, it is not possible to determine the quantitative difference in toxic response, if any, between human populations and test animal groups. Safety factors have come to be used to compensate for these gaps in fundamental knowledge. The methods of risk assessment discussed above are widely used by regulatory agencies and committees of NRC. Although these methods cannot remove scientific uncertainty, they ensure that the risk assessments properly reflect the best available scientific knowledge. There are no known superior methods for assessing the risk to public health . Application of these methods and determination of acceptable risks for carcinogens or acceptable intakes for other substances based on decisions about acceptable MOSs are designed to ensure that no significant risk is imposed on populations. In most cases, this goal is untested (and may be largely untestable because of the extreme difficulty in acquiring quantitative toxicity data in humans). Nonetheless, it serves as the basis of various health protection standards. CHEMICAL RESIDUES IN POULTRY PRODUCTS AND THEIR PUBLIC HEALTH RISKS Sources and Types The 1985 NRC report on meat and poultry inspection included a survey of the various types of chemicals that are used in meat and poultry production or that may occur as inadvertent contaminants (NRC, 1985~. For present purposes it is not necessary to duplicate that survey, but only to organize the information in a form more convenient
120 for application to poultry products and to develop a strategy for risk assessment. A useful approach Is to examine the various ways in which chemicals come to be present in poultry products. These can be divided into four maj or classes: Class 1. Legally approved (e.g., by FDA or USDA) chemicals intentionally administered or applied to poultry or to feeds used for poultry. These include, for example, the pesticides , pharmaceuticals , and antibiotics listed in Table S - 2 as well as chemicals used in processing, such as sodium benzoate, potassium sorbate, and monosodium glutamate; feed additives, such as butylated hydroxyanisole, butylated hydroxytoluene, nitrates, and nitrites; and certain growth promoters. Class 2. Widespread environmental contaminants to which poultry may be exposed because of their presence in poultry feed or drinking water. These include organic and inorganic chemicals of industrial or natural origin, including pesticides that are present in media for which there is no registered use. Among the possible organic contaminants are polychlorinated biphenyls, polychlorinated dioxins and benzofurans, chlordane, dieldrin, heptachlor, dichlorodiphenyl- trichloroethane, and lindane. Among the inorganic compounds are mercury and arsenic. Mycotoxins such as aflatoxins are also included in this class. Class 3. Chemicals to which poultry may be exposed as a result of accidental contamination. These may be any of the chemicals in Classes 1 and 2 and many other substances in commercial production. Class 4. Chemicals formed when poultry is processed (chlorine, polyvinyl chloride, acrylonitrile), stored (fatty acid hydroperoxides, hydroperoxyl radicals), and cooked at high temperatures (amino-carboline congeners, polycyclic aromatic hydrocarbons). This classification is especially useful both for risk assessment and risk management. Perhaps different risk-management strategies should be used for each of the four classes, which are distinguishable in the following ways: All substances in Class 1 can legally be used only in compliance with preestablished regulations based on the types of risk assessment and standard-setting procedures described above. There may or may not be such regulations for substances in Classes 2 9 3, or 4e The degree to which residues of the four classes can be predicted to occur in poultry varies greatly. Generally, Class 1 substances are highly predictable, Cal ass 3 substances are highly unpredictable, and Class 2 and 4 substances fall between these extremes O Class 1 substances should not be expected to present high risks unless the preestablisned tolerances or limits are repeatedly
121 violated because of intentional or accidental misuse. The risks of Class 2, 3, and 4 substances without preestablished exposure limits cannot be readily judged without a great deal more information than currently exists about their toxic properties and their occurrence in poultry products and other environmental media. For substances in Classes 1, 2, and 4, the principal concern is the risk of chronic exposure. The risk of acute exposure is the major concern for Class 3 substances. The quality and completeness of the toxicological data base used to establish acceptable intake levels are likely to be good for Class 1 substances and generally inadequate for all but a few substances in Classes 2, 3, and 4. Knowledge of the number and identities of chemicals in Class 1 is very good. It is significantly less complete for Class 2 chemicals and highly imperfect for Class 3 and 4 substances. The availability of reliable, sensitive, and practical analytical methods for measuring the chemicals in the four classes varies greatly. Methods for Class 1 chemicals are probably superior to those in the other classes. For many substances in Classes 2, 3, and 4, there are no methods available . Each of these factors plays a role in risk assessment and risk management . Type and Magnitude of Risk For some purposes, it is useful to organize chemicals according to type and magnitude of risk, rather than by source, especially when one must estimate overall public health risk From chemical residues from all sources. To accomplish this, it would be necessary to identify the toxic properties of chemicals known to occupy one or more of the four classes and to regroup them according to their types of toxicity. Traditional categories of toxicants, such as carcinogens, teratogens, liver toxicants, kidney toxicants, and reproductive toxicants, could be used for this purpose. The methods of risk assessment described above could then be applied to residue concentration data and information on human intake of poultry products to estimate the public health risk posed by chemical residues in poultry products. This approach is discussed further in the last section of this chapter. Although categorization of risk by type and magnitude is an important component of a risk-assessment program for poultry, there are important reasons to use source of residue classification as the principal system, which serves as the basis of the following discussion.
122 USING RISK ASSESSMENT TO ESTABLISH RISK-MANAGEMENT PROGRAMS Risk assessments are generally undertaken as a basis for risk management, i.e., the process of controlling risks so they are at acceptable levels. This is especially true for FSIS's assessment of risks presented by chemical residues in poultry. Effective prevention of such risks to public health requires the following types of activities: 1. Identifying substances that could appear as residues in poultry. 2. Setting ADIs or other levels of tolerable daily intake using the tools of risk assessment. 3. Establishing tolerances for residue levels in edible poultry products, taking into account other sources of human exposure so that the ADI (or tolerable intake level) is not exceeded. Setting levels of chemical intake by poultry (through their diets, drinking water, or other sources) that will ensure that tolerance levels in poultry products are not exceeded. 5. Establishing quality control programs to ensure that poultry feeds, drinking water, or other sources do not contain the chemicals of concern at levels exceeding those identif fed in Activity 4. 6. Establishing monitoring progress to ensure that poultry products reaching consumers do not contain residues above the limits established in Activity 3. 7. Establishing enforcement procedures to provide efficient deterrence and, when needed, to ensure removal of contaminated poultry products from commerce. 8. Identifying priorities for the different steps in each of these activities by using the concepts and tools of risk assessment. Given the present state of knowledge, these tasks cannot be accomplished for all classes of residues with equal rigor and certainty. However, it is important to begin them to assess the extent to which each is currently being undertaken and to manage known potential risks in the most efficent way. The risks of Class 4 substances are poorly understood, and the efforts needed to identify and deal with them are substantially different from those needed for the other classes. Thus, chemicals in that class, i.e., those formed during processing, storing, and heating, are omitted from this discussion and are included in a separate section on Special Problems later in the chapter.
123 In the following discussion of the eight activities described above, no attempt is made to identify responsibility unless it is clear that such responsibility already exists (e.g., EPA's responsibility to assign tolerances for pesticide residues). The options for carrying out the various risk assessment and management tasks are discussed in 1 ater chapters of this report. Activity 1. Identifying Chemicals of Potential Concern Class 1 Chemicals. The use of such substances is based on information about their toxic properties and the levels of residues expected to occur in poultry products as well as demonstration to the satisfaction of EPA, FDA, or PSIS that no significant risk will result under specific conditions of use. This information, including approved use conditions, is specified in formal regulations (for a typical example, see CFR, 1986, which concerns the use of a sulfonamide drug in swine). There is no reason to believe that any chemicals in this class have escaped identification. Class 1 chemicals introduced in the future are expected to be similarly well identified. Unidentified metabolites or degradation products of Class 1 chemicals may occur as residues, however, and are not taken into account in established tolerances. This potential applies to all classes and is discussed later in this chapter along with Class 4 substances under Special Problems. Class 2 Chemicals. EPA has collected information on many potential inorganic and organic contaminants of drinking water and has established maximum contaminant levels for many but not all of them. EPA, FDA, and USDA have collected information on some of the known contaminants of feed ingredients, and FSIS has gathered some data on residues in poultry products. It will be necessary to conduct an extensive literature review to identify regular contaminants of drinking water used in the production and processing of poultry and contaminants of all ingredients used in poultry feed. In this way it should be possible to ensure that all important chemicals in Class 2 have been identified. Some Class 2 substances can reasonably be treated as chemical classes. Polychlorinated biphenyls (PCBs), for example, are mixtures of closely related substances that share toxic properties. PCBs have been subjected to toxicity testing as mixtures, and data from these tests have been used to estimate the risks presented by these mixtures. For other chemicals (e.g., chlorinated ethanes), there are data on individual, closely related chemicals. If these chemicals produce similar forms of toxicity, they may be treated as a class for risk assessment.
124 As part of this activity, it is necessary to gather any data that bear on the toxicological properties of these chemicals in poultry. Of particular concern are data suggesting that any of these contaminants can accumulate in edible animal tissues. Information on bioaccumulation need not be obtained only from studies in poultry, but can be derived from studies in other animal species as well. All this information can be used to produce an initial list of chemicals in Class 2, which can be updated as new data emerge, and to develop a scale of the relative probabilities that those chemicals may occur as residues in poultry products. These relative probabilities can be used to set priorities for the remaining risk-assessment and management activities. A surveillance program will be needed to determine the actual occurrence of these residues in poultry products. This program should include poultry sampling and analysis to identify problems. It should not be used to monitor for compliance with established limits. Class 3 Chemicals. A systematic attempt should be made to identify substances that lead to product contamination through occasional misuse or accidents. Information useful in identifying such chemicals include historical and current reports on accidental exposure of poultry and accidental contamination of poultry products, drinking water, and poultry feed ingredients. Also useful are data on industrial activities in the vicinity of poultry production facilities and feed production operations O This type of information, which is available from several sources (e.g., EPA, FDA, USDA, various state agencies, and the scientific literature), can be used to determine whether specific groups of chemicals, including those in Class 1, are especially likely to be accidentally released in ways that may lead to contamination of water and feed ingredients. Activity 2. Risk Assessment to Identify ADTs or Other Tolerable Intake Levels Class 1 Chemicals. Risk assessemnts have been carried out on most if not all substances in this class--by FDA for feed additives and drug residues in poultry and by EPA for pesticide residues--with methods similar to those described above. Use of substances in this class under conditions prescribed in regulations should not lead to residue levels that exceed tolerances. This should in turn ensure that ADIs or other tolerable intake levels are not exceeded. The degree to which the toxicological data on chemicals in Clas s meet currently accepted standards of quality and completeness is unknown to this committee, and can not be determined without a thorough review of data in FDA and EPA files. Information reviewed in the NRC report on meat and poultry inspection suggests that there are
125 substantial inadequacies in the data base (ARC, 1985~. Thus, the data on Class ~ substances should be continually reviewed to determine whether current standards have been adequately satisfied. Only then can FSIS or any other agency determine with confidence whether compliance with current exposure limits ensures adequate public health protection. The committee reviewed the data and found no evidence to sugges t either that current exposure limits for Class 1 substances fail to protect the public health or that the poultry industry does not attempt to comply with these limits. In the absence of firm information about the adequacy of the data bases, however, the committee cannot conclude unequivocally that compliance with current limits is adequate to protect public health. Class 2 Substances. Extensive data on toxicity are available for . . .. _ . some potential members of Class 2 (e.g., certain metals, PCBs, dioxins, aflatoxins), and the data bases have been thoroughly evaluated by groups such as EPA, FDA, WHO, and NRC. For other substances that might be in Class 2, the toxicological data base is quite inadequate. More specific statements about the adequacy of the data cannot be made without ~ thorough literature review, which should make it possible to separate substances for which adequate risk assessments can be performed from those requiring additional toxicological data. Because substances in Class 2 are widespread environmental contaminants, chronic human exposure is possible. It is therefore important to ensure the adequacy of the toxicological data base for estimating limits on chronic exposure. Once the necessary data are available, risk assessments can be performed for this class as they are for Class 1 substances. Because environmental exposure to some Class 2 chemicals is widespread, the degree of acceptable risk may be higher for some Class 1 substances (e . g., noncarcinogens) than for others (e.g., carcinogens). EPA, FDA, and WHO have considered this when setting acceptable levels for PCBs in fish, for metals in certain foods and drinking water, and for aflatoxins in peanut and corn products. The types of risk-management analysis necessary to decide when to depart from the limits usually imposed on Class 1 substances are beyond the scope of the present study but need to be recognized. Class 3 Chemicals. In contrast to Classes 1 and 2, a major concern for Class 3 substances is short-term, occasional exposures, e.g., those that might result from a chemical spill. Short-term dietary exposure limits should be established for substances, including those in Classes 1 and 2, with a high probability of misuse or accidental contamination of diet. Methods for estimating short- term limits are similar to those for estimating ADIs, except that the data requirements are substan- tially reduced. Methods for establishing such limits have been reviewed by NRC committees (NRC, 1983~.
126 Activity Estimating Tolerances in Edible Poultry Products For each substance for which an ADI, short-term exposure limit, or other exposure limit is established, it Is necessary to identify the maximum concentration, or tolerances, to be allowed in edible poultry products. To establish such tolerances, it is necessary to know the amount of each poultry product ~ typically muscle, skin, liver, and kidney) consigned each day by members of the population and other sources of human exposure to the same chemical or clas s of chemicals (diet and drinking water but possibly through air and soil as well). Data on poultry consumption rates have been estimated for different segments of the population, e.g., the average consumer and the 90th percentile consumer. The segment used to define tolerances should be objectively selected and explicitly defined; some precedents for the selection process have already been established by FDA and EPA o Any differences among agencies in consumption data and their uses should be identified and resolved. For substances in Class 1, data on other exposures are usually taken into account by FDA and EPA when tolerances are set, but it is not clear to the committee whether this has been done adequately, for example, whether data on groundwater contamination by pesticides were adequately considered in setting tolerances for the same pesticides in edible animal products. All information on human exposure to Class 1 and 2 substances should be used in setting the maximum tolerable daily intake or ADI for poultry products. The portion of allowable total exposure that should be allocated to poultry products and other media should be determined jointly by all concerned regulatory agencies. The concern is to set tolerances for relatively high exposures of short duration . S ince the chronic background exposure level is ordinarily comparatively insignificant, it is not generally necessary to cons ider other exposure sources . Activity 4. Identifying Acceptable Levels of Chemical Intake for Poultry It is desirable to identify the maximums level of daily chemical intake (through feed or drinking water) that can be tolerated by poultry to ensure that residue levels do not exceed tolerances. With such information, it would be possible to monitor poultry feed and drinking water as an alternative to, or in addition to, monitoring poultry products . Among other benef its, complete information on the association between intake of chemicals and residue levels in poultry products would probably reduce the need for us ing safety factors in the assessment of risks for feed and water. However, a risk-management program can operate without this information and, indeed, should not fail to operate if the information is not available
127 Acceptable exposure levels may of course change over time as a flock matures. The acceptable intake may be set at zero for some period before slaughter (withdrawal period). Class 1 Chemicals. FDA and EPA generally require that metabolic and pharmacokinetic studies be conducted to establish exposure limits for substances in Class 1. These studies are designed to identify the levels in feed or water or levels of drug administration. The data from these studies can then be used to establish tolerance levels for Class 1 substances in drinking water or feed. The adequacy of the toxicological data base for establishing feed or drinking water tolerances for Class 1 substances is discussed under Activity 2. Class 2 Chemicals. Control of feed and water contamination may be the most effective strategy for managing the risks presented by Class 2 substances, but only a few of the common substances in this class, e.g., aflatoxin, PCBs, and certain heavy metals, have been carefully studied to measure the relationship between feed or drinking water levels and levels in edible poultry products. One reason for this is that poultry cannot be held in storage during the slow process of chemical analysis. It is especially important that this information be collected and analyzed immediately. However, without the necessary metabolic and pharmacokinetic data, suitable feed and water tolerances cannot be established. For only a few substances, there are sufficient data for estimating the maximum chemical concentrations in feed, water, or both that will not lead to violation of tolerances in poultry products. Both feed and water tolerances may be necessary for some Class 2 substances (e.g., metals), because both media may be pathways of poultry exposure. Class 3 Chemicals. The data needed for Class 3 substances can be acquired through study of the biological disposition and excretion rates of chemicals in poultry following short-term, high-level exposures. Such data can be used to identify potentially dangerous levels in feed or water. Activity 5. Feed and Water Quality Control A risk-management program based on tolerances in poultry feed and water, coupled with quality control and enforcement that is adequate to ensure that the tolerances are met, might be substantially more effective than a program based on monitoring of residues in poultry products. As long as feed and water tolerances are met, one could reasonably assume that residue levels in poultry products do not exceed established tolerances. This program would correct problems before they occur and seems to be especially well suited to the well-controlled conditions under which most poultry are now raised. If such a program were in effect, however, monitoring of poultry products would still be necessary but would not constitute the principal risk-management tool .
128 Routine monitoring for Class 1 and 2 chemical contamination of feed and water could be required. Priorities for monitoring are discussed below under Activity 8. Illness in flocks, or the detection of an unusual or unexpected substance during routine monitoring for Class 1 and 2 substances a may be the only ways to detect potentially harmful levels of a Class 3 substance in feed and water. Routine monitoring for Class 3 substances is not appropriate 9 however, because of the enormous and ill-defined range of potential contaminants and the rarity with which any one of them will be present in sufficient quantities to pose a risk to public health. If a Class 3 contaminant is found, and if it cannot be determined whether its source was an accident that is not likely to be repeated, potentially affected feed or water should not be used until an adequate ongoing detection program is in place and an ADI or other long- term exposure limit is established. Activity 6. Monitoring of Poultry Products Careful sampling and analysis (monitoring) of feed and water is the most effective line of initial defense against contamination of poultry products but, as noted above, is not sufficient. The risk-management program must also include monitoring of poultry products themselves at a frequency and intensity that is matched to the magnitude of potential risk. Product monitoring is necessary for the following reasons: Monitoring of feed and water cannot be made 100% effective for any substance. The data necessary to establish appropriate feed and water tolerances are not available for many substances. For many chemicals, the relationship between feed and water levels and residue levels is not yet understood. Analytical methods have not been developed for some substances in feed or water. Class 3 substances may be easily overlooked in monitoring feed and water, or they may be encountered at other points in poultry production The intensity and frequency of product monitoring, and priorities assigned to such efforts, should be related to the effectiveness of the feed and water quality control. This is discussed further under Activity 8. Product monitoring should ensure that poultry products containing residues above established tolerance limits for any of the three classes do not leave the processing plant if they ever enter it. However, there are practical limitations to the full realization of this objective, the principal one being that poultry cannot be held in storage during lengthy chemical analysis.
129 If poultry feed and water limits are properly set and enforced, there may still be occasional failures to keep residues within tolerance levels. These failures may have only limited health consequences, however, because tolerance limits for chronic exposure include large safety factors. Although there is no precise definition of an occasional failure, it should be assumed to mean rare occurrence in the life of an individual. When such excursions above tolerance levels are detected, it should be determined whether they present any reason for concern about larger, longer, or more frequent violations of standards. Each such finding should trigger an effort to learn the cause of the problem and to find a remedy. In the context of adequate feed and water controls, it is not possible to predict how long such tolerance violations should be allowed to continue before poultry or poultry products are condemned. However, once a tolerance violation occurs, a risk assessment is needed at times to identify the seriousness of the potential risk. A decision could then be made on the need for condemnation as well as the need to change the feed and water tolerances, alter the production process, or intensify inspection. Activity 7. Enforcement Monitoring programs to manage risks are not effective unless they can ensure that excessively contaminated poultry feed or water is not used and that excessively contaminated poultry products do not reach consumers. Regulatory agencies have long had programs of enforcement to ensure that these objectives are met. The need for such programs is obvious, and no additional justification need be given here. Activity 8. Establishing Priorities Any program based on Activities 1 through 7 will require establishment of priorities. Two monitoring efforts (Activity 5, for feed and water, and Activity 6, for poultry products) require the development of sampling plans that can ensure, with some predetermined degree of confidence, that risk-management objectives are being achieved. As stated above, risk-management priorities and the frequency and intensity of monitoring should be based on risk assessment. For much risk-management planning, only the relative risks of various substances are of concern. A methodology for assessing relative risks is proposed in the following paragraphs. Assessing Relative Risks A scheme for assessing relative risks need not include estimation of the absolute risk of any of the substances to be ranked. It is necessary only that it incorporate in a systematic way some measures of both toxicity and exposure that are as accurate as possible; risk assessment cannot proceed without them. The exposure and toxicity data
130 on Class 1, 2, and 3 chemicals vary widely in quality and content. These differences should be taken into account in a systematic way. The primary purpose of a relative risk assessment is to ensure that the two major risk-management activities (monitoring of feed and water and monitoring of poultry products, including whole birds) achieve the intended objectives. That is, the degree of risk-management attention accorded a substance is directly related to the probability that it will be found in food intended for human consumption and to the risk it may pose if it escapes detection Two useful measures for ranking relative toxicity are the ADI ~ for noncarcinogens ~ and the UCR. These measures have the following desirable characteristics. They are derived from toxicity data in ways that are now rather well standardized and accepted. Different types of toxicity data gaps are treated in a relatively uniform way for different substances, e.g., by the application of standardized risk- assessment procedures. These measures are based on chronic exposure. They should be estimated for all substances monitored. (Recall that surveillance i s used to identify substances for which risk assessment and tolerances need to be established, whereas monitoring is restricted to substances for which tolerances have been established. ~ These two measures are adequate to rank the chronic toxicities of Class 1 and 2 substances. The committee knows of no other measures that have all the above characteristics. To provide a systematic way of comparing carcinogens and noncarcinogens, it may be necessary to develop a single toxicity scale that integrates both categories of substances. It is possible to derive an ADI equivalent for a carcinogen from its UCR by using certain assumptions about the level of risk considered to be negligible and the level of risk associated with an ADI for a threshold agent. Under the assumptions used to derive UCRs, carcinogens present a nonzero risk at all exposure levels above zero. Nevertheless, it is commonly accepted that for all carcinogens there is an exposure ra5ge that6presents only a small risk, e.g., lifetime risks less than 10- or 10- , which reflect highly unlikely events O It is not possible to demonstrate that an ADI carries absolutely no risk for the human population. At best, all that can be claimed for an ADI is that any potential risk is not likely to exceed some very small but quantified risk In the absence of evidence to the contrary, and to provide two different toxicity scales for carcinogens and noncarcinogens, it may be assumed that the range of risk associated with an ADI is the same as that as that considered to be very small carcinogens ~ i . e ., 10 ~ 5 to 10 ~ 6 ~ . This assumption is presented here only in the context of this specific risk-ranking obj ective to
131 provide a systematic means for comparing the toxicities of carcinogens and noncarcinogens. It does not imply that the actual risk at an ADI is in the range assumed here. It is further assumed that an ADI for a noncarcinogen will ensure that there is not more than a 10-6 (1 in a million) risk of a toxic effect occurring. An ADI equivalent derived for carcinogens will be taken as the dose estimated to give rise to the same maximum lifetime risk (10-6~. The ADI equivalents for carcinogens can then be directly compared to ADIs for noncarcinogens, because both will have been adjusted for potency and represent the same estimated risk level. Figure 5-2 presents the linear low-dose response of several carcinogens with different UCRs. For each 6f these carcinogens, a dose providing an estimated lifetime risk of 10- can be identified. These ADI equivalents are represented by points for carcinogens A, B. C, And D along the dose scale. Thus, a carcinogen (B) with a OCR of 10- per unit of dose measured in mg/kg low/day would have an ADI equivalent of 0.001 mg/kg/day. A UCR of 10-7 (carcinogen D) would correspond to an ADI equivalent of 10 mg/kg/day. These ADI equivalents can be calculated for the entire range of published UCRs. A representative range of ADIs and ADI equivalents and some possible toxicity ranking scores are presented in Table 5-3. Unfortunately, there appears to be no single, direct measure of potential exposure. Thus , in constructing a ranking system for exposures, many factors must be considered. For example, the following factors all contri bute to potential exposure for Class 1 and 2 substances: 1. The portion of the ADI or other tolerable limit to which people are ordinarily exposed. Frequent exposures to large fractions of the ADI (through poultry products only or through several environmental media) present higher potential risks than occasional exposures to only a small fraction of the ADI. A tolerance violation may have much greater significance for the former than for the latter exposure . 2. The frequency with which the chemical is or is likely to be detected in feed, water, or poultry products. 3. The volume of use for Class 1 substances and the volume of production, industrial use, or natural occurrence for Class 2 substances . 4. The number of birds treated or otherwise exposed. 5. The propensity for bioacc~mulation.
132 l o _ l in o or C: Sit v o 7 o _ I\ to ; ~ lo lo 70 To lo lo lo, _ _ _, _ _ ~ _ o ASIA - ° C,9 a oo , _ o En ~-! ~ o .> o c) o sit H ,,4 ¢ ~ o F=, ~ - o Cal 1 Us
133 Table 5-3. Chronic Toxicity Scoringa ADI Range (mg/kg/day) Toxicity Score <10-7 210-7 <10~6 21o~6 <1o~5 210-5-<10-4 210-4-<10-3 210-3 <10~2 >1o-2-< 210~1 <1 2 1 9 8 7 6 5 4 3 2 1 aFrom Environ Corporation, 1985. For non- carcinogens, use ADI; for carcinogens, use an ADI equivalent, than 10-6. assuming a lifetime risk no greater 6. The frequency and magnitude of consumption of the mayor tissues in which residues occur. (Substances accumulating in skin and muscle are of greater concern than those present only in the lung or kidney.) ~ , FSIS should use these six factors to establish a ranking scheme for chemical exposures. Generally, the range of possible scores for potential exposure should approximately equal the range of toxicity scores (i.e., toxicity and exposure should be given approximately equal weight). One such scheme would be to assign a numeric score to each of the six items: for example, a score of 1 or 2 for factors 1 and 5; a score of 0, 1, or 2 for factor 3; and a score of 0, 2, or 4 for factors 2, 4, and 6. The total of these scores divided by 2 would yield exposure scores ranging from 1 through 9. The overall priority ranking system would then be based on a combination of this exposure score with the toxicity score. for examcle. a score of 1 or 2 for factors 1 and 5; a Use of Relative Risk Scores. ranking procedure, one should By systematic application of this able to establish priorities for
134 monitoring feed, water, and poultry products and to develop sampling plans that are matched to the potential risks . The procedure shou1 d also be useful in establishing ADIs and tolerances for residue levels in poultry products. Use of a single monitoring strategy for all chemicals would be inappropriate, since adequate attention would not be given to potentially high risk substances and too large a share of resources would be devoted to low risk substances. A more appropriate scheme would be based on first ranking the risk (e.g., high, medium, and low risk) and then categorizing the chemicals according to the type of risk they present (e.g., carcinogens, teratogens, liver toxicants). Intensive monitoring programs should be devised for potentially high-risk substances, less-intensive programs for those posing medium risks, and minimal monitoring activity for low-risk substances. Of course, ranking should be continually updated as new data emerge to determine the need for regrouping. Of particular importance for the two - stage monitoring strategy described above is the choice of sampling rate. Statistical sampling strategies can be devised to ensure, with a specified degree of confidence, that products containing excessive levels of chemical res idues are identified for removal from the food supply . The desirable degree of confidence for a potentially high-risk substance should be greater than for other substances. In Chapter 7, the committee recommends criteria for sampling chemical res idues pos ing different levels of potential public health risk. Finall y, all eight essential activities of a risk-management program for chemical residues are based on applying, with varying degrees of rigor, the elements of a risk- assessment scheme based on specific types of data. Although an effective risk-management scheme will require all eight activities, not all of them need be under the direct control of FSIS. Indeed, some activities are already established at FDA and EPAo Nevertheless, to the extent possible, PSIS should ensure that all eight activities are under way and are adequately pursued. SPECIAL PROBLEMS As noted above, two aspects of chemical hazards in poultry require special treatment: Class 4 substances (those formed during processing, storing, and heating) and metabolites and degradation products of chemicals . Class 4 Substances The information needed to perform risk-assessment activities 1 through 3 for Clas s 4 subs tances is the same as that required for Classes 1, 2, and 3, but there are few data on the potential public
135 health risks presented by these chemicals. For example, it appears that there has been no comprehensive risk assessment for any of these substances found in table-ready broiler chickens. Because Class 4 substances contaminate poultry products by mechanisms different from those for Classes 1, 2, and 3, it is not clear whether the risk- management strategies described for those classes are appropriate for Class 4. The committee believes it would be premature to devise a comprehensive risk-management strategy for them and recommends that FSIS initiate efforts to assess their risks in a comprehensive manner. Metabolites and Degradation Products In the regulation of residues in poultry, most attention has been given to the parent compounds administered to or ingested by the bird rather than to the' r metabolites or degradation products. EPA and FDA have given some consideration to those products, but it is not clear to the committee that the two agencies have treated the subject adequately or consistently. There are no data indicating that metabolites or degradation products pose significant risks that are unregulated or that the risks of these products, to the extent they are considered, are under- or overestimated by the agencies. It nevertheless seems important to examine this issue carefully and to evaluate its present status. ASSESSING PUBLIC HEALTH RISKS OF CHEMICAL RESIDUES IN POULTRY PRODUCTS The magnitude of the public health risk from chemical residues in poultry products has not yet been examined, but the committee believes it important that such risk assessments be undertaken routinely. The chemical residue data developed by FSIS (Table 5-2) are not by themselves adequate for risk assessment, because the following information is lacking: · the quantitative relationship between the levels of residues found in specific tissues examined by FSIS and the levels present in all other edible tissues; the amounts of different poultry products consumed by different segments of the population; the capability of the analytical methods used to detect residues below a certain level of contamination; toxicity data, ADIs, and UCRs for each residue; the level of human exposure resulting from other environmental media in which residues of the same chemicals may be present; and time trends in contamination patterns. The information necessary to perform residue-specific risk assessments is available for many substances, especially those in Class 1. Access to FDA and EPA data files will be necessary to acquire
136 the necessary toxicity and tissue distribution data and to estimate background levels. FSIS residue data should be analyzed statistically to determine the extent to which they are representative of the poultry product supply as a whole. These are tasks that can be completed with varying degrees of thoroughness for different residues. They should nevertheless be undertaken for all commonly found residues, but extreme care must be taken to ensure that the limitations in the data base and in the risk-assessment methodologies used are clearly set forth. Qualitative risk assessments should always be accompanied by descriptions of their limitations. Because compliance with current tolerances for Class 1 chemicals is relatively high, it is likely that risk assessments undertaken for them will result in very low risk estimates. However, compliance with prescribed tolerances does not necessarily ensure low risk 9 since the adequacy of the data base for Class 1 substances has not been reviewed and there may be significant data gaps or limitations. Many substances in Class 1 were approved or registered by FDA and EPA many years ago; however, the committee knows of no routine federal effort to ensure that the data base for these substances meets current standards, except for limited EPA efforts with regard to some pesticides. It would thus be necessary to review the toxicity data base for Class 1 substances before accurate risk assessments can be completed. The data base for Class 2 substances is certain to be less adequate than that for Class 1 substances It is not even clear that all important chemicals in Class 2 have been identified. Nevertheless, risk assessment for Class 2 substances should be undertaken and limitations in data and methodology described to the extent possible with current ;nfo'=ation. It is particularly important to include information on other environmental sources of exposure to these chemicals, such as PCBs and some of the widely dispersed chlorinated hydrocarbon pesticides, so that the contribution of poultry products to total risk can be understood and the information can be used to estate ish special control programs where high risks exist and to reduce or eliminate programs now focusing on trivial problems. Although the committee examined the data and found no evidence of significant public health risks attributable to chemical residues in broilers, risk assessments and data are needed before definitive conclusions can be reached. REFERENCES Asher, I. M., and C. Zervos, eds. 1977. Structural Correlates of Carcinogenesis and Mutagenesis. A Guide to Testing Priorities? Proceedings of the Second FDA Office of Science Sumner Symposium held at the U.SO Naval Academy, August 31-September 2, 1977. DREW Publ. No. (FDA) 78-1046 O The Office of Science, U.S. Food and Drug Administration, Rockville, Md. 241 pp.
137 Calabrese, E. J. 1983. Principles of Animal Extrapolation. Wiley, New York. 603 pp. CFR (Code of Federal Regulations). 1986. Title 21, Food and Drugs; Section 520.2240a, Sulfaethoxypyridazine drinking water. U.S. Government Printing Office, Washington, D.C. Doull, J., C. D. Klaassen, and M. O. Amdur, eds. 1980. Casarett and Doull's Toxicology--The Basic Science of Poisons, 2nd ed. Macmillan, New York. 778 pp. Environ Corporation. 1985. Documentation for the Development of Toxicity and Volume Scores for the Purpose of Scheduling Hazardous Wastes. Prepared for the Office of Solid Waste, U.S. Environmental Protection Agency. EPA Contract No. 68-01-6861, Subcontract No. 38-9. Environ Corporation, Washington, D.C. 130 pp. EPA (Environmental Protection Agency). 1980. Water quality criteria documents; availability. Notice cuff water quality criteria doc~ents . Fed. Regist. 45: 79317-79379 . EPA (Environmental Protection Agency). 1982. Pesticide Assessment Guidelines, Subdivision F , Hazard Evaluation: Human and Domestic Animals . Publ . No . EPA 540/9 - 82 -025 . Office of Pesticide Programs, Office of Pesticides and Toxic Substances, U.S. Environmental Protection Agency, Washington, D.C. 163 pp. Available from the National Technical Information Service in Springfield, Va ., as PB83 -153916 . EPA (Environmental Protection Agency). 1985. Health Assessment Document for Chloroform: Final Report. Publ. No. EPA/600/8-84/004F. Office of Health and Environmental Assessment, U.S. Environmental Protection Agency, Washington, D.C. [415 pp.] Available from the National Technical Information Service in Springfield, Va., as PB86-105004. FDA (Food and Drug Administration). 1979. Polychlorinated biphenyls (PCB's); reduction of tolerances. Final rule. Docket No. 77N-0080. Fed. Regist. 44:38330-38340. FDA (Food and Drug Administrations. 1982a. D&C Green No. 6; li sting as a color additive in externally applied drugs and cosmetics . Final rule . Docket No . BIN- 0301 . Fed. Regist O 47: 14138 - 14147 . FDA (Food and Drug Administration). 1982b. Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food. Bureau of Foods, U. S . Food and Drug Administration, Washington, D. C. ~ 248 pp . ~
138 FDA (Food and Drug Administration). 1985. Sponsored compounds in food producing animals; criteria and procedures for evaluating the safety of carcinogenic residues. Proposed rule. Docket No. 77N-0026O Fed. Regist. 50:45530-45555 FSC (Food Safety Council). 1980. Proposed System for Food Safety Assessment. Final Report of the Scientific Committee of the Food Safety Council. Food Safety Council, Washington, D.C. 160 pp. Klaassen, C. D., and J. Doull. 1980. Evaluation of safety: Toxicologic evaluation. PpO 11-27 in J. Doull, C. D. Klaassen, and M. 0. Amdur, eds. Casarett and Doull's Taxicology--The Basic Science of Poisons, 2nd ed. Macmillan, New York. Lehman, A. J., and 0. G. Fitzhugh. 1954. lOO 'ford margin of safety. Assoc. Food Drug Off., Q. Bull. 18: 33-35. Loomis, T. A. 1978. Essentials of Toxicology, 3rd ed. Lea & Febiger, Philadelphia. 245 pp. MacMahon, B., and T. F. Pugh. 1970. Epidemiology: Principles and Methods. Little, Brown, Boston. 376 pp. Nisbet, I. C. T. 1981. The role of exposure assessment in risk evaluation: Research needs. Pp. 419-429 in C. R. Richmond 9 P. J. Walsh, and E. D. Copenhaver, eds. Health Risk Analysis, Proceedings of the Third Life Sciences Symposium. Franklin Institute Press, Philadelphia. NRC (National Research Council). 1977. Drinking Water and Health. Report of the Safe Drinking Water Committee, Advisory Center on '[oxicology. National Academy of Sciences, Washington, D. C. 939 pp. NRC (National Research Council). 1980a. Drinking Water and Health, Vol. 3. Report of the Safe Drinking Water Committee, Board on Toxicology and Environmental Health Hazards. National Academy Press, Washington, D.C. 415 pp. NRC (National Research Council). 1980b. Risk Assessment/Safety Evaluation of Food Chemicals. Report of the Subcommittee on Food Toxicology, Committee on Food Protection, Food and Nutrition Board. National Academy Press, Washington, D. C. 36 pp. NRC (National Research Council). 1983. Risk Assessement in the Federal Government: Managing the Process. Report of the Committee on the Institutional Means for Assessment of Risks to Public Heat th, Commission on Life Sciences. National Academy Press, Washington, D-C- 203 ppO
139 NRC (National Research Council). 1984. Toxicity Testing, Strategies to Determine Needs and Priorities. Report of the Steering Committee on Identification of Toxic and Potentially Toxic Chemicals for Consideration by the National Toxicology Program, Board on Toxicology and Environmental Health Hazards. National Academy Press, Washington, D.C. 395 pp. NRC (National Research Council). 1985. Meat and Poultry Inspection: The Scientific Basis of the Nation's Program. Report of the Committee on the Scientific Basis of the Nation's Meat and Poultry Inspection Program, Food and Nutrition Board. Press, Washington, D.C. 209 pp. National Academy NRC (National Research Council). 1986. Drinking Water and Health, Vol. 6. Report of the Safe Drinking Water Committee, Board on Toxicology and Environmental Health Hazards. National Academy Press, Washington, D.C. 457 pp. OSTP (Office of Science and Technology Policy). 1985. Chemical carcinogens; a review of the science and its associated principles, February 1985. Final document. Ped. Regist . 50 : 10372 - 10442 . Rodricks, J., and M. R. Taylor. 1983. Application of risk assessment to food safety decision making. Regul. Toxicol. Pharmacol. 3:275-307.