National Academies Press: OpenBook

Regulating Pesticides (1980)

Chapter: 4. Risk Assessment

« Previous: 3. Selecting and Scheduling Compounds
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 65
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 66
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 67
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 68
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 69
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 70
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 71
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 72
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 73
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 74
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 75
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 76
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 77
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 78
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 79
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 80
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 81
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 82
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 83
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 84
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 85
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 86
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 87
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 88
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 89
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 90
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 91
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 92
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 93
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 94
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 95
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 96
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 97
Suggested Citation:"4. Risk Assessment." National Research Council. 1980. Regulating Pesticides. Washington, DC: The National Academies Press. doi: 10.17226/54.
×
Page 98

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.

4 Risk INTRODUCTION Assessment The purpose of pesticide regulation in this country is to protect the human population, animals, useful vegetation, natural amenities of all sorts, and property from the "unreasonable adverse ejects" of the use of chemical pesticides (P~ 92-516, 19724. All pesticide regulations promul- gated by EPA are intended to serve this purpose. Accordingly, a key component of the preparation of any regulation is an assessment of the dangers presented by the compound under review. If the assessment indicates there are substantial dangers, estimates are required of the extent to which they will be mitigated by various alternative restrictions and regulations that might be imposed. This chapter reviews the methods now employed by oPP in forming the requisite analyses and recommends a number of changes in those procedures. Although in principle the risk assessment of any pesticide entails consideration of all the affected categories listed above, in practice, dangers to human health are currently EPA'S predominant concern. Indeed, within the area of human health, oPP's attention is generally focused on possible oncological and mutagenic effects of suspect pesticides, since these are the most apparent adverse ejects of the chemical pesticides now in widespread use and currently being intro- duced. The discussion in this chapter will therefore concentrate on the assessment of dangers to human health and particularly on the danger of inducing cancers. This narrow focus is dictated by time and resource 65

66 REGULATING PESTICIDES constraints imposed on the study. It means that a number of important matters, in particular the assessment of environmental risks and eventual indirect elects on humans from long-term environmental effects, have been treated very briefly or not at all. Determination of whether a pesticide poses a serious potential hazard is based on two considerations that are operationally separate. The first is the extent of exposure, that is, the number of people who may be expected to receive dosages of different levels and by different routes if the pesticide is used freely or if it is regulated in various possible ways. The second consideration is the pathological activity or toxicity of the pesticide, including the probability that a person exposed to specified doses by specified routes will super adverse erects of various degrees of severity, sometimes called the dose-response relationship. The analyses of these two aspects employ entirely different data and methods. They are conducted separately and are discussed separately below. Assessment of the dangers to human health caused by the use of a pesticide is treated in the first major section of the chapter. The discussion is divided into three subsections, the first dealing with exposure analysis, the second with pathological activity, and the third with combining the previous two to obtain an overall assessment of risk. In each subsection the procedures currently used are reviewed critically and suggestions for improvement are made. The second major section of the chapter deals, more briefly, with the analysis of risks other than those to human health. HAZARDS TO HUMAN HEALTH EXPOSURE ANALYSIS Current Approach The purpose of oPP's exposure analysis is to determine in as quantitative a manner as possible the number of people exposed to a pesticide by various routes and the doses they receive. The analysis is developed on a use-by-use basis, and a special effort is made to understand how a particular pesticide is used and what human activities are associated with each use. For example, when an analysis is required for a pesticide with multiple uses, estimates are made of exposure by all routes for each use. The analysis includes a brief description of use practices, a summary of available data, and exposure estimates derived from the data. The exposure analysis is used at two stages in the RPAR decision- making process (Severe 1978a):

Risk Assessment 67 (1) the initial decision to issue an RPAR rests in part on the likelihood of he exposure, so that a preliminary assessment of exposure (i.e., preliminary exposure profile) is needed at this stage; and (2) once an RPAR has been issued, the final nsk/benefit decision generally requires a more thorough analysis of exposure (the degree of completeness required depends In part on the toxic potency, extent of use, and magnitude of benefits to be derived from use of the pesticide In question, and is determined by the Project Manager and/or Working Group during the analysis leading to the final nsk/benefit document Position Document #3~. To date, there are no official agency guidelines for preparation of exposure analyses; however, a draft Procedures Manualfor Preparation of Human Exposure Analyses (Severe 1978a) and other agency documents (e.g., internal memoranda) provide guidance until such guidelines become available. Preliminary Exposure Profile A Preliminary Exposure Profile (PEP) is prepared for use in pre-RPAR activities. Essentially, the PEP is a rough estimate of the number of people exposed to different dose rates (for example, in terms of dose per hour of application) (Severe 1978a). Since few data on pesticide use are likely to be available at the pre-RPAR stage, the project manager maintains a core data base consisting of product label files, information from worldwide literature searches, and agency files of existing exposure information. The rough exposure estimates are determined by tabulating each use listed on the labels and comparing it with model exposure (that is, experimental application) situations, taking the compound's chemical and physical properties into consideration. The PEP thus lists each use indicated on the label along with an estimate of exposure from that use. As a compound proceeds through the RPAR process, additional data are sought to make possible a more detailed evaluation of the exposure situations with which the compound is associated. Data for Exposure Analyses Ideally, a detailed exposure analysis for a pesticide would include estimates of exposure by all routes, both for the entire U.~. population and for particular subgroups that may have different levels of exposure, especially applicators and pickers. Therefore, data on numerous aspects of a particular pesticide are needed for precise estimation of the degree of human exposure associated with its use. oPP has identified several factors critical to the assessment of various exposure situations. The factors include group size; dose from each route of exposure; duration of exposure; statistical reliability of exposure

68 REGULATING PESTICIDES estimates; and exposure to metabolites, by-products, impurities, and contaminants (Severe 1977a). The worldwide computer literature survey made for each RPAR candidate pesticide identifies studies relevant to various aspects of the principal compound and its metabolites or degradation products (Severe 1977a). The studies serve as a primary source for many of the data needed to prepare a detailed exposure analysis of a given pesticide. Additional data sources include, among others, agency files, the USDA, the FDA, industry, user groups, the open literature, and universities. Information on patterns and practices associated with the use of a pesticide serves as a basis for ranking use patterns according to potential for human exposure. Each use practice is thoroughly described, including all sites of application; formulations used at each site; application rates and dilutions; representative labels and packaging information; methods of mixing and loading; application techniques and schedules, including a description of apparatus and common practices during application, and the times and numbers of applications per year; number of applicators involved and their identity (farmers, commercial applicators, industrial users, and so on); extent of use (total acres treated and pounds used annually by crop and state); number of associated personnel involved in application (such as mixers, loaders, and flaggers); estimate of total hours of application activity; extent of use and kind of protective clothing; and percentage of each crop treated annually (Severe 1977a). Data regarding patterns of exposure serve as a basis for estimating the amounts of a pesticide received through ingestion, inhalation, and dermal routes. Relevant information for exposure through ingestion includes data on food tolerances, residues, food consumption patterns, food processing and distribution practices, and drinking water surveys (Severe 1977a). The data come primarily from the open literature and Registration Division files of oPP. Estimates of food consumption patterns are based largely on nationwide averages (usually provided by USDA) and allow for variations in both geography and age (Severe 1977a). In addition, background data on food processing and distribu- tion practices allow estimation of the extent to which foods consumed may be contaminated by residues of the pesticide. Estimation of inhalation and dermal exposures is based on data from air monitoring, applicator practices, dynamics of application, and absorption of the compound (Severe 1977a). EPA surveys and the open literature are primary sources of available air monitoring data. Requisite data on applicators include numbers, extent of training, work schedules and practices, and protective clothing used. Information concerning the

Risk Assessment TABLE 4. ~ Sources of Data Used In Exposure Analyses of Selected Pesticides 69 Data EPA Source Treflana Air concentrations Registrant Inhalation rate Bioastronautics Data Book (1964) Duration of exposure Doane Agriculture Station (1975) (applicators); USDA (field workers) Number of field workers USDA Dermal exposure estimates Wolfe et al. (1967); Hayes (1975) Chlorobenzilateb Number of applicators USDA Inhalation/dermal Wolfe et al. (1967) exposure estimates Average adult food USDA consumption rates Residue data Florida; USDA; EPA (limited) LindaneC Duration of exposure EPA Inhalationldermal Wolfe et al. (1967) exposure estimates Food tolerances EPA Food factors (commodity EPA distribution) Extent of pesticide use EPA a Source: Severn (1977b). b Source: Severn(1978b). c Source: Donoso and Collier (1978). dynamics of the application of a pesticide is based primarily on data concerning drift and transport near adjacent populations, Canon, persistence and reentry, and presence of particulates (Severe 1977a). Human monitoring data come primarily from the open literature and EPA projects (e.g., the Human Monitoring Program). Relevant data include surveys of blood, urine, adipose tissue, and mother's milk (Severe 1977a). Also data from household surveys indicate the potential for exposure via pesticide-contaminated dust and home-use practices (Severe 1977a). Data used in selected exposure analyses for several pesticides are summarized with respect to type and source in Table 4.1. Inhalation Exposure Estimates of respiratory exposure (i.e., via inhala- tion) are presented in terms of ambient air concentrations of the pesticide in the breathing zone of exposed persons (Severe 1978a).

70 REGULATING PESTICIDES Because air concentrations may vary widely, estimates of the likely range and mean of the concentrations are desirable. The physical state (vapor, aerosol, or particulate) of the pesticide is also noted. If sufficient data are available, time-weighted average concentrations are computed. Esti- mates of inhalation exposure for a given population are a function of estimates of ambient concentrations, duration of exposure, and number of people exposed. The Toxicology Branch of HED determines the rate of inhalation and the extent to which the pesticide penetrates and is absorbed into the lungs. Estimates of individual inhalation exposures are commonly derived either by measuring the concentration of the pesticide in samples of ambient air or by determining the amount of the pesticide actually trapped by the filter system of a respirator worn by a worker for a specified period of time (Hayes 1975~. The first method requires calculation of breathing rates before actual inhalation doses can be determined. However, use of either approach appears to be determined more by the nature of available data (i.e., its quality and quantity) than by predetermined Agency guidelines. Dermal Exposure Estimates of dermal exposure are presented in terms of milligrams of pesticide per hour that come into contact with the skin of exposed persons. The clothing worn by agricultural workers plays a critical role in the determination of dermal exposures (Severe 1978a). The extent to which pesticides that are deposited on skin are absorbed is determined by HED'S Toxicology Branch. An important dermal exposure situation arises from reentry into areas previously treated with pesticides (Severe 1978a). It is difficult to predict quantitatively the actual dermal (and respiratory) exposure of, for example, orchard fruit pickers. Such exposure depends on the amount of residues remaining at the site, which relates directly to persistence and degradation characteristics of the pesticide in question. The Environmental Fate Branch of HED maintains a file of data on dislodgeable residues (mostly organophosphates) and other information on reentry. When an analysis requires an estimate of exposure during reentry, experts in particular geographical areas are usually consulted. Ingestion Exposure The general approach to determining the amount of a pesticide ingested by humans in their diets is to multiply an estimate of the number of micrograms of the pesticide per kilogram of food in the various foodstuffs that may contain it by estimates of the amounts of those foods in a normal daily diet. The estimate of the amount of the pesticide per unit of a food is obtained in either of two ways. If there are actual measurements of pesticide residues in foods, those measurements

Risk Assessment 71 are used. More frequently, the residue concentrations are too small to be measured by available analytical methods. In such cases, it is assumed that the foods contain the maximum amount of pesticide residue permitted by EPA (i.e., the tolerance level). The amounts of the foods contained in normal diets are derived from food consumption surveys conducted primarily by USDA, which are often adjusted for both geographic and age variations in consumption patterns. Assumptions Data on many of the factors that are critical to the preparation of a detailed exposure analysis for a particular pesticide often are unavailable. In such cases, oPP either makes assumptions that it feels are necessary under the circumstances or, alternatively, derives estimates of exposure from data on other compounds that are used in similar patterns. Assumptions made in preparing exposure analyses for the three pesticides displayed in Table 4.1 are summarized in Table 4.2. oPP's approach to estimating exposure of spray applicators to chlorobenzilate, for example, was based largely on the assumption that inhalation and dermal exposures vary the same way under different application conditions. The same assumption was used in the exposure analyses of Treflan (Severe 1977b) and Lindane (Donoso and Collier 1978~. In the absence of data on actual applicator exposure to chlorobenzilate, probable estimates of both dermal and respiratory exposure were based on data for other pesticides used under conditions similar to those associated with chlorobenzilate (Severe 1978b). The data, as reported by Wolfe et al. (1967), consist of measured dermal and respiratory doses received by spray applicators while applying azinphosmethyl, DDT, dieldrin, malathion, and parathion. However, since the data reported by Wolfe et al. are based only on orchard spray conditions, oPP is initiating the development of models for other application situations (D. Severn, OPP, EPA, Washington, D.C., personal communication, 1978~. The assumption that 10 percent of the amount of a pesticide (in solution) that comes into contact with the skin is absorbed plays an important role in evaluation of dermal exposures. Although pesticides may be absorbed through the skin with varying efficiencies (Hayes 1975), the absorption rate of 10 percent has been used in several exposure analyses prepared by oPP (e.g., chlorobenzilate and Treflan; see Table 4.2~. When, for example, information on protective clothing worn by agricultural workers is lacking, it is assumed that exposed workers wore short-sleeved shirts and long trousers but no hats or gloves (Severe 1978a). In this situation, estimates of dermal exposure are derived from existing data on measured skin deposition from a known spray

72 TABLE 4.2 Exposure Analysis Assumptions REGULATING PESTICIDES Treflan Air sampling data follow log-normal distribution All inhaled NDPAa is retained, not exhaled Ten percent of the amount of pesticide that comes into contact with the skin is absorbed Field workers wear no protective clothing Inhalation and derrnal exposures vary the same way under different application conditions Treflan will continue to be used indefinitely at about the current rate(s) of application Chlorobenzilate Occupational exposure of citrus pickers is less than that of spray applicators Ten percent of the amount of pesticide that comes into contact with the skin is absorbed Inhalation and dermal exposures vary the same way under different application conditions Residues in treated commodities approach established tolerance levels Inhalation per applicator-hour is the same as for other pesticides used in similar situations Chlorobenzilate will continue to be used indefinitely at about the current rate(s) of application Lindane Residues in treated commodities approach established tolerance levels Inhalation and dermal exposures vary the same way under different application conditions Lindane will continue to be used indefinitely at about the current rate(s) of application a Nitroso dipropylamine. concentration of another pesticide (Severe 197Ba). The dermal dose of other pesticides with similar spray characteristics can then be calculated from the spray concentration used. Since patterns of pesticide use are difficult to observe and enforce, there is, in many cases, a total absence of data on dermal and inhalation exposures during application. A1- though estimation of dermal ejects attempts to incorporate both chemical and toxicological aspects of a particular compound, the 10 percent skin-absorption rate may be inaccurate by an order of magni- tude. Studies are now under way to evaluate the roles played by skin and protective clothing as physical barriers in determining occupational exposures (D. Severn, oPP' EPA, Washington, D.C., personal communica- tion, 1978~. For dietary exposures worst-case estimates are usually based on the assumption that residues exist in or on commodities at the limit of established tolerances. This assumption was used in both the Lindane and chlorobenzilate exposure analyses (see Table 4.2), but the availabili- ty of actual residue-monitoring data may permit more reliable estimates. When estimates of daily exposure are converted to lifetime equiva- lents, oPP assumes that a pesticide will remain on the market and in use indefinitely. For respiratory and dermal exposures, which are usually

Risk Assessment 73 occupational, exposure is assumed to occur over a typical number of work years for the number of days per year that a pesticide is used. For example, it was estimated that spray applicators were exposed to chlorobenzilate for 10~0 days per year (depending upon the number of applicators), over a 40-year work life (U.S. EPA 1978a). Dietary exposure was assumed to occur daily over a full 70-year lifetime. Occasionally, there may be too few data available to permit a quantitative estimate of exposure. The 25,000 30,000 citrus pickers who may be exposed to chlorobenzilate represent a case in point. Here, oPP assumed that the pickers were less frequently exposed than the spray applicators (Severe 1978b), but no quantitative estimates were made. In considering enforcement, oPP assumes that label restrictions will limit occupational exposure to some extent, and in this context, develops various regulatory options that may result in reduced levels of exposure. For example, the recommended regulatory option for chlorobenzilate includes requirements for specific types of clothing and respirators to be worn during application (U.S. EPA 1978a). A more detailed review of the chlorobenzilate exposure analysis is presented in Chapter 7. Comments and Recommendations In the Committee's judgment, oPP makes sensible and competent use of the often incomplete information available in performing its exposure analyses. The Committee does not recommend any far-reaching changes in oPP's general approach to exposure estimation, but there are a number of important changes that ought to be made in some of the detailed procedures followed and in the methods of presenting results. Data Gathering and Use Exposure to a pesticide is not a simple mechanical matter. It depends on such properties of the pesticide as persistence, solubility, vapor pressure, adsorbability, partition coefficient, and thermodynamic characteristics. These properties influence the extent of vapor contamination, water contamination, biological availability, and persistence of residues. Estimates of exposure require information about all these chemical and physical properties of the pesticide and careful evaluation of their influence on the doses received through various routes by exposed populations. Estimates of exposure should take these considerations into account more extensively than now appears to be the case. In estimating exposures, as in other phases of its work, oPP is constantly hampered by lack of adequate data, and is forced to resort to indirect and inaccurate methods in its effort to make plausible estimates.

74 REGULATING PESTICIDES A typical example is the use of the dermal and respiratory doses received by spray applicators while applying DDT, dieldrin, and several other compounds to estimate the doses received by chlorobenzilate applicators for whom no data exist. The valid use of such indirect evidence requires close and subtle familiarity with the pesticide under consideration, including its chemical, physical, and pathological properties, and details of the methods by which it is applied. Such familiarity can seldom be gleaned from the literature. The Committee therefore makes the following recommendation. · It should be routine practice for the members of the EPA stay team reviewing a pesticide to visit sites where it is applied, facilities where it is formulated and handled, and laboratories where it is studied, and on these visits to hold informal discussions with the people involved in day-to-day manufacture, handling, and use of the pesticide. Not only is there no substitute for this firsthand contact as a basis for informed judgment, but it has the further advantage of demonstrating to the people who will be affected by any future decision that their knowledge and views have been taken into account in the course of arriving at the decision. Agricultural experiment stations are particularly important sites for these visits and have the added advantage of often directing attention to useful publications of the stations or other sources that the usual literature indexes do not include. Economic Life of a Pesticide For many pesticides, particularly those likely to induce cancers, the likelihood that an effect will eventualize is cumulative, so that estimates of lifetime exposures, rather than of rates of dosage during short periods, are relevant to risk assessments. As noted earlier in this chapter, the usual practice for making such estimates at present is to assume that if a pesticide is reregistered, it will continue to be used indefinitely at about current levels of application. In fact, the economic life of a pesticide or the length of time that it is expected to be bought and used is -limited by (1) the rate of development of resistance or tolerance to it in the target pest, and (2) the introduction of more elective or economical alternative pesticides into the market. Thus, as do most tools, pesticides have a limited useful life. Information on the economic life of pesticides should be included in all risk (and benefit) analyses of pesticides. Exposure, and hence risk, would generally be expected to drop to near zero as soon as a pesticide's economic life is spent and the pesticide is no longer used. Of course, there are always exceptions. For example, an environmentally persistent pesticide such as DDT may continue to present a potential for low-level

Risk Assessment 75 exposure for a period ranging from a few months to more than 10 years beyond its economic life. In order for oPP to make well-founded estimates of lifetime exposures to pesticides (as well as accurate benefit estimates), the Committee makes the following recommendation: · oPP should undertake or sponsor a study of the economic lives of pesticides and the factors that influence them. Estimates of both lifetime exposures and economic benefits should be based on periods of use consistent with thefindings of the study. The Committee's best estimate on the basis of available information is that the use of a pesticide for specific pests has averaged about 10 years in a range of 2 to more than 34 years. (It should be recognized that the total economic lifetime of a pesticide encompasses all uses and therefore may be longer than the lifetime for a particular use.) When regulatory options are considered on a use-by-use basis, as they are in this report and in the oPP evaluations, the 10-year average figure, with its accompanying range, appears appropriate for estimating anticipated economic lifetimes until more reliable estimates become available. This figure, however, is rough and purely provisional and should be quickly superceded. Factors such as increasing testing costs and their eject on innovation in the pesticide industry may substantially alter estimates of the economic lives of pesticides in the future. For pesticides that have already been on the market for a number of years, an educated guess based on expert opinion will have to suffice for the time being for estimating the additional average number of years those pesticides can be expected to remain on the market. For example, in Chapter 7, the Committee estimates that if reregistered, chlorobenzi- late would continue in use for another 10 years beyond the more than 20 years it has already been used on citrus. In cases of this type, it should be assumed that, should registration of the pesticide be continued, addition- al exposure of the population and the resulting biological effects will not, on average, exceed the effects attributable to the additional years of use (unless persistence is known to be a problem). Presenting Probable-Case Estimates and Confidence Limits There is a general tendency when estimates are uncertain, which is almost always the case, to adopt "conservative" estimates. If"conservative" means tending to err on the safe side, it must be pointed out that neither side is safe. On the one side, if a regulator decision is predicated on erroneously low estimates of the number of people who would be exposed to injurious doses of a pesticide whose use is unrestricted, the decision will be biased toward inadequate restriction, with possible

76 REGULATING PESTICIDES harmful consequences for public health. On the other side, if the estimate of exposure is excessively high, the resultant regulation is likely to impose economic costs that are disproportionate to the hazard averted, with a consequent waste of economic resources and of the goodwill on which EPA'S electiveness ultimately depends. The errors in question are by no means trivial; the phenomena leading to exposure and to adverse erects are so complex and so little understood, and the data relating to them are so fragmentary, that even best estimates may be substantially more than one order of magnitude in error. The closest to a safe course in these circumstances is for the analysts to arrive at their best, unbiased (i.e., not intentionally conservative) judgment of the likely consequences of any regulatory option, and to present these estimates together with an indication of the range of uncertainty to the officials responsible for arriving at a decision. It is the responsibility of the officials, not the analysts, to weigh the relative seriousness of making errors on one side or the other. Those officials should be able to rely on the reports prepared for them to present fair, unbiased estimates from the facts and assumptions on which their decisions must rest. Many of the estimates will be incorporated in public documents. Here, the same principle applies. That is, the public, including legislators and groups of interested citizens are all entitled to know the unvarnished truth: the best estimates that informed and thoughtful consideration can arrive at together with the ranges of uncertainty that surround them. Users of the estimates can then be relied upon to introduce whatever "conservative" biases they deem appropriate. At present the position documents and supporting reports almost invariably violate this principle. Indications of ranges of uncertainty are rare. The position documents generally present as estimates of exposure the single, upper-limit, "worst-case" values for each exposed group. The qualifying considerations, probability factors, and ranges of uncertainty are not mentioned, leaving the estimated values unqualified by any assessment of their probability. Thus, the decision maker is not provided with information about how reasonable the worst-case values presented are nor with guidance for judging the levels of exposure that are likely to be the result of alternative regulatory options. The Committee's analysis of the exposure estimates for chlorobenzilate, given in Chapter 7, shows that in that instance the worst-case levels of exposure are highly improbable. The use of tolerance levels for estimating the concentrations of pesticide residues in foods is an extreme instance of the same bias. When tolerance levels are imposed, food producers aim at concentrations that

Risk Assessment 77 are safely below those levels, since if they aim at average concentrations too close to the permissible levels, much of their produce will be in violation. In practice, average concentrations of residues in foods are likely to be one or more orders of magnitude below the tolerance levels, so that basing dietary exposure estimates on tolerance levels generally overestimates such exposures by a factor of 10 or more. It should be mentioned that worst-case estimates are not necessarily "conservative" that is, they do not necessarily increase the likelihood of decisions that will protect health to the greatest extent possible in the circumstances. This perversity can arise when the difference between the worst-case estimates under unregulated use and under a contemplated regulation is smaller than the difference between the most probable results in the two situations. In such a situation the advantages of regulation may appear negligible or not worth the cost if worst-case estimates are used, while less biased estimates would disclose substantial probable reductions in exposure. There is an additional reason for avoiding the practice of presenting only worst-case estimates in a risk appraisal: it interferes with the Administrator's exercise of judgment in choosing among alternatives. If a worst-case estimate of individual exposure is multiplied by a worst-case estimate of the number of people exposed, and the product is multiplied by a worst-case estimate of the carcinogenicity of the pesticide in question, then the result will be an unrealistically high estimate of the health costs of using the pesticide. Ideally, the Administrator would like to base decisions on the analyst's best judgment of the probable erects of adopting any option together with the analyst's judgment of the worst possible consequences consistent with the available data (i.e., the worst case). lion: For these reasons, the Committee makes the following recommenda · oPP should continue to use its current procedures (with the modifications discussed aboveJ for estimating exposures associated with various regulatory options. It should employ those procedures to derive estimated ranges of possible exposures under the different options. Those ranges should always be presented as a pair of numbers, one showing the exposure (or other aspect of risky that is deemed most probable, and the other showing the maximum exposure (or component of risky that is likely to be experienced in the absence of an implausible array of untoward circum- stances, i.e., the worst case. A clear definition of "range" is required in order for this recommenda- tion to be implemented intelligently. The Committee suggests that ranges be interpreted to mean that the probability that the true exposure is

78 REGULATING PESTICIDES greater than the upper limit of the range and less than the lower limit is small but not negligible, perhaps about 5 percent. These confidence limits should, of course, take account of all sources of error and imprecision in the estimates, and should not be merely mechanical applications of statistical formulas. ASSESSMENT OF CARCINOGENIC RISKS The following discussion is devoted to the assessment of carcinogenic activity, partly because cancer appears to be the hazard of primary concern to EPA, and partly to permit an exploration of specific issues. Although the Committee has not explored other health hazards in as much detail, we believe that appraisals of other hazards to health have to overcome many of the same problems and should be approached by much the same methods. Current Practice . The general procedures for assessing the risks to human health posed by suspected carcinogens are described in the Agency's Health Risk and Economic Impact Assessments of Suspected Carcinogens: Interim Proce- dures and Guidelines (U.S. EPA 1976, referred to throughout as the guidelines). The basic evaluative framework established by the guidelines has several important features. The guidelines clearly state the Agency's basic philosophy regarding the regulation of suspected carcinogens. It is noted that " . . . in many areas risks cannot be eliminated completely without unacceptable social and economic consequences" (U.S. EPA 1976:21402-3~. Accordingly, the guidelines establish, as the basic regulatory objective, the elimination or reduction of risks " . . . to the greatest extent possible consistent with the acceptability of the costs involved" (U.S. EPA 1976:21403~. This regulatory philosophy allowing for trade-o~s between risks and benefits is quite different from the one imposed upon the FDA by the Delaney Clause of the Pure Food and Drug Act. The guidelines create a two-step decision-making process for the regulation of potential carcinogens. The first step in the process involves determining whether and to what extent a particular substance consti- tutes a cancer risk. The second step involves selecting the specific regulatory assessments are conducted as part of the first step in this regulatory sequence. The guidelines identify two objectives that are to be addressed in assessing carcinogenic risks of suspect chemicals:

Risk Assessment 79 1. To evaluate the evidence concerning a particular agent and from this to judge whether the agent is a potential human carcinogen. 2. If it is, the next step is to judge the likely extent of its effect on public health, with specific reference to cancer, at current and antici- pated levels of exposure. The responsibility for making these judgments resides with EPA'S CAG. In connection with the first issue-whether a substance constitutes a cancer risk-the guidelines recognize the difficulties attendant on "proving" that an agent is a human carcinogen. Thus a substance is to be considered " . . . a presumptive cancer risk when it causes a statistical- ly significant excess incidence of benign or malignant tumors in humans or animals" (U.S. EPA 1976:21403~. The judgment is to be based upon a "weight-of-evidence" approach that relies upon a wide range of data sources, including human epidemiological studies, animal bioassay studies, and short-term in vitro tests. In general, the determination of whether a substance is a human carcinogen is to be based upon available information; the guidelines impose no requirements for the acquisition of new data (U.S. EPA 1976:214031. The guidelines distinguish among several different types of evidence on the basis of quality and adequacy: The best evidence that an agent is a human carcinogen comes from epidemiolog~- cal studies In conjunction with confirmatory animal tests. Substantial evidence is provided by animal tests that demonstrate the induction of malignant tumors ~ one or more species including benign tumors that are generally recognized as early stages of malignancies. Suggestive evidence includes the induction of only those nonlife shortening benign tumors which are generally accepted as not progressing to malignancy, and indirect tests of tumorigenic activity, such as mutagenicity, ~n-vitro cell transformation, and ~n~tiation-promotion skin tests mice. Ancillary reasons that bear on judgments about carcinogenic potential, e.g., evidence from systematic studies that relate chemical structure to carcinoge- nicity should be included in the assessment. (U.S. EPA 1976: Appendix I, 21404) With regard to the second issue determining the extent of the cancer risks the guidelines commit the Agency to quantitative risk extrapola- tions (U.S. EPA 1976: Appendix I, 21404~. The extrapolations are to be based upon the best available evidence concerning exposure levels and are to be performed with a variety of risk extrapolation models, such as the linear nonthreshold model and the log-probit model (Crump et al. 1976, Hoel et al. 1975, Mantel and Bryan 19614. Moreover, the extrapolations must be done separately for all suitable experimental data

80 REGULATING PESTICIDES and human epidemiological data. The guidelines recommend that the results be presented in terms of excess lifetime incidence, or average excess cancer rates. At the same time, the guidelines recognize that there are considerable uncertainties surrounding these risk analyses (see below), and accordingly emphasize that the extrapolation results should be interpreted only as a "warning signal" rather than as an actual indicator of excess cancer incidence. Comments The CAG typically mounts a well-informed and conscientious effort to meet the expectations of the guidelines. It evaluates the available evidence from bioassays in experimental animals and from epidemiologi- cal observations and weighs biochemical and toxicological information in assessing the carcinogenic activity of the pesticide under study. However, although the guidelines indicate that a "weight-of-evidence" approach is to be taken in judging data, precisely how this is done is unclear. Criteria for determining the weight of evidence are not stated in the guidelines and have not been thoroughly discussed elsewhere; consequently, the judgments appear to be made in an ad hoc manner without formal criteria. It would be desirable for CAG to provide some formal discussion of its criteria for making weight-of-evidence judgments. [These criteria should consider how CAG would proceed to arrive at a judgment where multiple sources and types of evidence are available. For example, how would evidence from a well-performed bioassay, with adequate numbers of experimental and control animals, that showed no carcinogenicity for a compound be compared and weighed against a far smaller study, perhaps with inadequate numbers of animals, that showed a strongly positive effect at comparable dose levels? How would strong evidence of mutagenicity or carcinogenicity in short-term tests in vitro be assessed in comparison to marginal studies showing no cancer excess in carcinogen- esis bioassays in vivo? Would the weight of evidence fall on the side of the positive study, the more thorough study, or the study of animals in viva? When the available, technically adequate evidence is conflicting, the risk assessments reviewed by this Committee indicate a strong tendency for CAG to place most weight and credence on data that show the strongest carcinogenic responses. The tendency to err on the safe side appears to be a general CAG practice. To the extent that this is the policy of CAG it should be formally stated. Along these same lines, as noted earlier in this chapter, oPP also prefers to use exposure estimates that may err by indicating an excess in number and dosage of exposed

Risk Assessment 81 individuals to estimates that may err by understating these variables. Estimates of exposure and evaluation of carcinogenesis should be specifically discussed to clarify how the weight-of-evidence judgment was achieved for each from the existing data. Finally, CAG also chooses extrapolation methods that, if they are in error, overestimate rather than underestimate risk. In this way, estimates and extrapolations all are permitted to err but, it is hoped, in such a way that the error will always overstate the risk. The intent of this worst-case approach is to allow the assumption that estimated risk always exceeds real risk. Clearly, however, others judging the weight of evidence might arrive at different determinations of risk. Because weight-of-evidence judgments are inherently subjective, the CAG should explain (1) how such judgments were achieved in each specific case, and (2) its criteria for making such judgments in general. The Committee found a more major problem having to do with EPA'S use of CAG'S numerical estimates of human cancer incidences. The Committee's difficulties stem from the belief that current understanding of carcinogenesis and related pathologies is not adequate to permit reliable extrapolations from animal experimentation and simpler assay systems to actual quantified hazards to human health (see Appendix A for more detailed support of this statement). Adequately controlled and documented epidemiological data for relevant populations are rarely available, and human experimentation with suspected carcinogens in viva is unthinkable. Consequently, experi- mental data from bioassays in animals must be relied upon in most cases. Uncertainty and error from at least three sources infiltrate quantitative estimates of human cancer risk from animal data (see Appendix A). First, pathological evaluations upon which an estimate of excess tumors in test groups are based are to a certain extent subjective, and are often controversial (for example, see Pesticide & Toxic Chemical News 1978 and 1979~. Moreover, such evaluations are not presented with confidence intervals. Second, at least two extrapolations of inadequately tested reliability must generally be applied to bioassay data to derive estimates of human cancer incidence. Extrapolations must be made between species- i.e., from experimental animals to humans and extrapolations must be made from experimentally used dose levels, which are generally quite high, to actual dose levels encountered by humans. (The determinations of heptachlor's pathological activity were based upon experiments in which mice were given doses some 13,000 times as great as the doses to which humans were being subjected.) In some cases, a further extrapolation must be made to compensate for differences in the route of exposure between experimental animals in

82 REGULATING PESTICIDES bioassay studies and actual routes of human exposure. Finally, present methods do not include means for evaluating the influences of several critical determinants of human cancer, such as differences in human susceptibility and additive or synergistic elects operative in the human population. Consequently, the Committee believes that inferences drawn by means of current extrapolation methods lack scientific justification, particularly when they are used without the support of epidemiological data and when they are used as a general method for all suspected carcinogens. Moreover, at a practical level, they have been insufficiently validated by past experience. Although there is obvious value to developing and verifying methods for extrapolating from observed experimental tumor responses in animals to erects at low doses in humans, current methods present significant and controversial scientific problems (see I=G 1979~. ~ti- mately, when there is greater insight into the mechanism of carcinogene- sis and methods are available for integrating variations in individual human susceptibility and the erects of exposures to other carcinogens and co-carcinogens, extrapolations based on this insight may provide reliable estimates of human cancer incidences from the use of a compound. With the present state of knowledge, the results of bioassays and short-term tests may be useful as comparative measures (to compare one carcinogen to another) of anticipated erects of various carcinogens at human exposure levels, but extrapolation techniques are not yet sufficiently precise or well-founded to allow us to make credible quantitative extrapolations about the anticipated frequency with which cancers will be induced in the human population by exposure to the estimated doses of the pesticide. Most important is the possibility that because of unconsidered factors, the worst-case extrapolation could actually be an underestimate. The members of CAG are in a difficult position. They interpret their responsibilities as requiring them to present numerical estimates of the erect of using the pesticide under review on the incidence of cancer in the exposed population. They are aware that current scientific knowledge does not justify such estimates per se, and they consistently qualify their reports accordingly (see Appendix A). Nevertheless, they are required to present numerical estimates and they produce a product buttressed by impressive statistical and mathematical analyses. Whereas CAG arrives at these estimates with appropriate constraints and reservations placed on their final result, the provision of a sophisticated quantitative estimate of human cancers provides a high potential for misinterpretation because the estimates may be used without the required attention to the inherent constraints. In fact, in this Committee's opinion, the convenience of

Risk Assessment 83 comparing risks as estimated numbers of human cancers to benefits in dollars makes it highly unlikely that misuse of extrapolation results can be avoided. In the view of this Committee, CAG should be more aware that users are so hungry for numbers that quantitative estimates, once presented, take on a life and authority of their own, despite all the reservations that CAG may attach to them. In the context of EPA'S procedures, CAG estimates are incorporated into benefit-risk estimates upon which choices among regulatory options are based. To be sure, the estimates themselves are accompanied routinely by warnings that they are subject to error. The Committee, however, did not encounter any document that conveyed the impression that the risk estimates could well be in error by as much as 1,00~10,000 percent. Yet the methods of extrapolation used by CAG, which reflect the best current scientific understanding, are subject to uncertainties of that order of magnitude or greater (see Appendix A). Such risk estimates are quite unsuited to comparisons with benefit estimates that, crude as they are, can be trusted well within one order of magnitude. Recommendations For the preceding reasons the Committee recommends that EPA not require CAG to estimate the numbers of people who would be expected to contract cancer or other diseases as a consequence of pesticide (or other chemical) use under various regulatory options. At the same time, the Committee recognizes that the Administrator needs some quantitative indication of the danger posed by a pesticide that he or she is called upon to regulate. In these circumstances the Committee believes that the best assistance scientific advisors can over the Administrator is to provide intelligible information concerning the experimental and epide- miological evidence upon which a judgment of risk is to be based. With respect to judging whether a compound is a carcinogen, the conservative position appears to be to accept the induction of tumors under laboratory conditions as presumptive evidence that a compound has the potential to act as a carcinogen or mutagen in humans. The Committee accepts that position. Beyond that, since compounds vary enormously in their degrees of carcinogenicity or mutagenicity, the Administrator is responsible for judging whether a specific pesticide imposes a risk great enough that its use might have an unreasonable adverse erect on the health of the exposed population. The judgment usually must be based on experimental indications, primarily from

84 REGULATING PESTICIDES bioassays using small rodents, of the pathological activity of the compound. Although the Committee believes experimental findings do not permit sound numerical estimates of potential human cancer incidences, experimental findings often do permit placing a compound on a scale of probable relative human carcinogenic activity, along with other com- pounds for which comparable experimental results are available. One approach to estimating the relative carcinogenicity of compounds to humans is recommended in the next subsection. The Administrator can thus be informed of how the carcinogenic activity of the compound under study is likely to compare with that of other compounds with which he or she may be more familiar and which have been judged to be suitable or unsuitable for use. (Such comparisons are useful to decision makers only in circumstances where the compound under review yields benefits comparable to those of the compound to which it is being compared. This concept is discussed further in Chapter 6.) The Carcinogenic Activity Indicator The remainder of this subsection is devoted to a procedure by which the Committee feels compounds can be placed on a scale of relative human carcinogenic activity based on animal bioassay data. Expressions of the relative carcinogenic potentials of compounds have been described previously by means of a number of "potency indexes" (for example, see Meselson and Russell 1977~. The following presentation is based on a potency index type concept, recommending the use of a Carcinogenic Activity Indicator (CA~) developed by the Committee and defined as: Excess percentage of subjects In which tumors are observed CAl = Lifetime dose (m moles/kg of body weight) . It is critical to understand at the outset that CAIN will be calculated for animals only, on the basis of bioassay data; estimates of the carcinogenic activity of a single compound in humans are never made. The procedure depends, rather, on comparing cat values derived for different com- pounds in animals under proper conditions to provide indications of the relative carcinogenic potential of these same compounds in humans. (The assumptions that allow this use of CAN'S are discussed in Appendix B.) Table 4.3 presents experimentally derived CAP'S for a number of compounds. The calculation of a CAI and the conditions under which the relative carcinogenicities of two or more compounds can be compared in experimental systems are described more fully in Appendixes B and C. A

Risk Assessment 85 number of difficulties in the interpretation of CAYS should be made explicit here: 1. CAPS for a given compound derived from experiments with different species are likely to diner substantially. For example, the CAPS for chloroform ranged from 23.7 to 41.9 when mice were used, but measured only 1.59 in rats (see Table 4.34. Such differences in CAN'T observed in two species of rodent make it clear that inferences about the CAT of the same compound in humans must rest upon a substantial ingredient of judgment. 2. For any single species, the observed CAT may be different for different routes of administration. 3. A CA' is derived from a single point on a dose-response curve by dividing the response by the dose. The CAT may therefore be expected to diner with different experimental doses. A dose-c curve for vinyl chloride, plotted from data in Table 4.3, is shown in Figure 4.1. Note that log-log scales are used. The observed CAT ranges from about 7.8 for a dose of 0.74 m moles/kg to 1,985 for a dose of 0.0037 m moles/kg. 4. The sample sizes used in many bioassays are so small that the observed values are subject to substantial statistical error. For instance, in a bioassay using 50 animals each in the test and control groups, a 15 percent excess incidence of tumors could be observed in the test group in about 1 experiment in 15 even if the chemical tested were innocuous, while no excess incidence might be observed for a dose of a tumorigenic chemical that on the average increased incidence by 18 percent. All these limitations must be taken into account when CAN'T are used to compare the carcinogenic potentials of different compounds. The comparisons will be most tenable when the experimentally observed CAPS being compared are derived from experiments conducted with the same species, using the same routes of administration, and administered in doses yielding approximately equivalent excess tumor incidences. Even then, due consideration must be given to experimental error. Note again that only animal CA~'S derived from bioassay data are being compared; no extrapolations to humans have been made. Table 4.3 is intended to be more suggestive than definitive, since the Committee's resources did not permit an adequate review of the literature. The CAYS shown were all computed from published experi- mental data. They show estimates of the percent of animals exposed to the compound listed that developed tumors as a result of the exposure in relation to the lifetime dose (in millimoles per kilogram of body weight) received by the animals in the experimental group. Table 4.3 suggests

86 U. O V, o .O a) Cat Can o U. o 4_ Cd Cat ·= ._ ._ C) .e o ·,~, Ct o ~ . - o _ ~ _ 4_ .~l - , o a, 4~ ~ o a, A ~ ~ as ~ Q ·_ ~ ~ C ~ ~ V, c: ~ It o is - \0 ~) c ~D - Cd o ~- ~ ~ - ·- ~- u~ CJ~ - - - - ~tO ~ - ~ z ~ - z ~z z z z ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 ~ r~ ~ _ ~ ~ . . . . . . . . . . . . . . . . . . . . O r~ r~ O ~D ~ ~ ~ ~ ~ ~ ~X ~ ~ \0 X ~ O ~ ~ - O O - X ~ ~D ~ ~ ~ ~ ~t ~ ~ - r~ cr~ oo (~\ ~ t~ _ ~ ~ ~ oo ~ ~ ~ ~ ~ ~ ~ - v~ ~ ~ o ~ ~ - ~ ~ ~ oo - ~ o o ~ - ~ - ~ - oo oo oo ~- ~ ~ ~ oo ~ ~ ~ - ~ O - ~ - ~ _ ~ ~i ~D ~ oo ~ ~ ~ c~ _ ~ _ ~ ~ 0\ 0 _. _ _ - - l- 00 ~ ~ ~ ~ _ _ _ oO ~D- - ~ ~ - ~ § ~; ~ §~ [ ~ g m ~; ~ ~ E ~ oo ~ ~ ~ o ~ o ~ ~ o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ o - ~o o oo ~ ~ ~ ~ ~ ~ ~ ~ ~ o o ~ ~ ~ ~ ~ c~ ~ oo ~ ~ - ~ ~ ~ ~ v) o ~ ~ oo ~ ~C~ m. ~. O ' ~ 0 i ~ d ~ ~ _ ~ ~ _ O _ x - - ~ - ~- (`l ~- '~' ~ ~(~ o o -o ~ ~ ~ ~ v) ( - o l - oo r4 ~ ~ - o o. o. o. v~ ~ ~ - ~ ~ ~ ~ o o o o - v) ~ ~ o ~ o ~ o o o o o o u~ - o o o o o - ~ ~ ~ - cr~ _ ~ ~ ~ 0 oo ~ ~ O ~ ~ ~ ~ ~ ~ ~ - ~o ~ ~ o v~ o - - - - ~D ~ - - o ~ ~ o o o o - ct~ ~ ~- oo w) t-) (~} o o o o ~ - o ~ o o ~ ~ ~ ~ d ~_ _ ~ ; ~ ~o-c ~o ~

~D _ ~act Ox - ~- ~ (~\ _ t_ 1- Ce ~t- ~ ~- Ce ~1 _ ~ ~_ _ _ ~ ~_ ~ l ~ ~ cry ~ ~ ~ ~ cry ~ ( V ~- ~ ~ ~ _ ~ _ Z O Z Z Z O Z ;~ d. - ~ ~ ~ ~ ~ ~- oo oo - ~ ~ ~ ~ ~ or ~ no _ ~ oo ~ ~ oo oo x ~ rat ~ ~ oo ~ x ID ° ~ O O O O O ~ ~O O ~ cr ~ ~ ID _ _ ~ cat ~ ~ ~ - 0 - cry . . . .. . . . . . r~ ~0~ 00 0 __ ~ ~ ~' ~ ~ S.~ ~ ~ ~ ~ £ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ C1. ~ ~ ~ ~ ~ CL ~Cl. ~ ~ ~ ~ ~ O. CL CL ~ CL CL ~ ~ ~ ~ ~ ~ ~ o oo ~ ~-~ _ ~o ~ o o o o o o _ ~ _ ~ _ _a~ - - ~ ~ ~ ~ O O O ~ C~ ~ ~ ~ ~ 0 ~ ~D '= ~C~ l- ~C~ ~ °~ ~ ° ~ ~ ~ 0 00 cr ~ ~ doo { - , ~ - ~ O - ~ ~ ~ 0 ~0 ~ ~ ~ ~ ~ ~4 ~ ,~ ,,, 00 , -, ,~, ~ ~_ _ ~ (,,, (~\ ~o ~ ~ o, o, o, o, ~ 0 ~ oo O ~ oo O ~ O, ~ ~ r~ ~ - ~ 0 0 0 0 0 0 0 ~ oo 0 0 ~ 0 0 0 0 0 0 ~ ~ _ o ~oo _ _ ~ 00 ~- ~ ~ _ oo - ~ _ ~ ~ r~ _ 00 0 0 0 0 ~ ~0 ~0 ~ 0 c4 ~ - ~0 0 0 0 ~ _ _ ~_ _ 0 0 0 ~ 0 ~00 _ ~_ 00 0 _ ~o 0 ~ 0 00 ~ _ _ _ 4 ~ 8 ~y ~ C 8 ~8 ~ m a) =~ E ~ Q s:: ~1 ~_, ~_' ~ X ~ ~ a.) ~ ~; I ~ ~ 11 8 Q ° E ~ =^ ° o ° ~ o E 8 ~ ~ ' E ~ ~ ~ E .= ~ ~ ~ ~ ~ i,9 ~= ~ , E o ~ .O C s ~ ~ ~ C .= ~ ~ - ._ 3 o ° o Ct ~ ~ ° ~ ~ =._ ~ ,= U, ~ v ~ C 2 ~ c o 8 ~ ~ o C Es C O C-E E s ~ ~ ~ = ~ o E~ 6~ ~ ~5= Q~ 87 . Ct ~ ~ o D ~ ._ ~ ~ O O ~ CL (L) - 1~ Cd ;> O O Cq ,~ O 8 D O ~ _ CO ·m 3 - ._ ~ s C~ ~ SX D 4) ·- ~ O ~ .m ~_ U, ·- ._ . .. ~ U, ._ .- ~ ~ ~0 ~ t-! 3= o C) C) Cd C) a.) c~o, 3 ~> :3 ce ao c~o 3 (,, D ._ C~ O ~ K t, D S CO ~ - . · - '= S 0~ c~ ~ a~ .m ~ S , - 8.~: a) _ a~ Z° O

88 2,000 1 ,000 <: 1 00 REGULATING PESTICIDES 10 0.001 \ 0.01 0.1 LIFETIME DOSE (m moles/kg body weight) 1.0 FIGURE 4.1 Relationship between CAT and dose for vinyl chloride. Source: Denved from Table 4.3. that on the average for mice administered test compounds orally and demonstrating comparable tumor response levels, heptachlor is approxi- mately 30 times as active a carcinogen as dicofol. This is because under similar experimental conditions a given number of moles of heptachlor per unit of body weight will have approximately 30 times as great an ~ . , .. . . ~ .. . . . i` .. . . .... ~ . . . enect (on the nasls of the data reviewed) on the pronaulllty ot developing excess tumors as the same number of moles of dicofol. Similarly, using data from Innes et al. (1969), chlorobenzilate, when administered orally, is about one third as active as dicofol in inducing tumors in certain laboratory mice. The table therefore can serve as a scale against which the pathological activity of any compound under review can be measured if experimental conditions are comparable (see Appendix B). If CAPS are to be useful for policy purposes, however, they must provide information on the dangers to humans of exposures to potential carcinogens such as certain pesticides. More precisely, the CAPS would have to allow for assertions and comparisons such as, "ingestion of x m

Risk Assessment 89 moles of endrin has about the same probability of inducing a cancer as ingestion of 10x m moles of chloroform." A number of assumptions must be made before such assertions based on experimentally observed CAYS can be justified. One set of assumptions that permits useful inferences to be drawn about the erect of specific pesticides on human health is suggested and discussed in Appendix B. The reader is urged to read and consider those assumptions. One will see that they are not innocuous and that, though intuitively appealing, they have little experimental support. The reason for preferring the evaluation of risks by means of CAN'S to the current procedures is that the current procedures require substantially stronger and less plausible assumptions and produce an end product that is much more liable to misinterpretation. In spite of the limitations that have been noted, the Committee feels that potency indexes, such as the CAN'S, are the best indicators available of the relative danger of different pesticides. Responsible officials and the general public should be informed of such indicators (together with the ranges of experimental error and uncertainty to which they are subject), and regulatory decisions should take them into account. The Committee recognizes that it would be more convenient if regulatory decisions could be based on reliable estimates of the probable elects of different regulatory options on human morbidity and mortality. But such estimates cannot be justified given the current state of scientific knowledge. Accordingly: · The Committee recommends that when laboratory data are used to estimate pathological activity, potency indexes, such as the cods defined above, be used to indicate the pathological virulence of the pesticide under consideration and that no numerical estimates of elects on human morbidity or mortality be extrapolated from laboratory data. The estimated potency indexes should be presented as most probable values accompanied by indications of ranges of uncertainty. How the CAPS can be taken into account will be discussed further below and again in Chapter 6, and illustrated in Chapter 7. COMBINING EXPOSURE AND PATHOLOGICAL ACTIVITY Estimates of exposure and pathological activity must be combined in appraising the hazard to human health posed by the use of a pesticide. The current procedure, to be discussed more fully below, is to make the combination by calculating, for each relevant segment of the population, an estimate of the probability that an individual will contract a disease

go REGULATING PESTICIDES (such as cancer) as a consequence of the use of the pesticide. The preceding discussion indicated that available estimates of the effects of pesticide use on incidence of disease in humans do not merit scientific credence. Therefore, the Committee recommends that the practice of making such estimates be abandoned. At the same time, the procedures for appraising pathological activity recommended by the Committee do not, in principle, lend themselves to similar, quantitative estimates of ejects on human morbidity or mortality. Thus, different methods must be used to combine exposure and pathological information. The current methods and a recommended alternative are discussed in the following two subsections. Current Practice The risks incurred by the use of any pesticide vary. They include carcinogenicity, mutagenicity, other chronic health impairments, and acute reactions. There are also risks to natural biota and to agriculture and livestock. The overall assessment of risks must take all these possibilities into account. For this reason, and perhaps others, the risk assessments in available oPP position documents have not followed a standard format. The risks associated with a pesticide have been appraised by various methods, talking account of the nature of the predominant risks of concern as well as characteristics of the available data. The appraisals share certain fundamental features, however. For example, as noted previously, the risks associated with cancer are estimated by the CAG primarily on the basis of animal bioassay data and evaluations of the metabolic and toxicological characteristics of the compound. Other hazards to human health are appraised by oPP's HED, using similar types of data and epidemiological evidence when available. Hazards to wildlife or biota and potential crop or livestock damage are evaluated by HED also, through searches of the relevant literature. The USDA/EPA benefit assessment teams play an important role in acquiring information about the use of pesticides that may generate such hazards. Potential and actual exposures are estimated, as described above, by well-standardized methods. In the end, these diverse kinds of informa- tion must be pulled together, and it is at this point that standardization ceases. Two examples will suffice. In the appraisal of chlorobenzilate (U.S. EPA 1978a), the induction of cancers was judged to be the primary type of risk with which to be concerned. Accordingly, factors provided by CAG were used to infer the increase in the lifetime probability of contracting cancer that would

Risk Assessment 91 result from the continued use of chlorobenzilate. Separate factors were computed for different segments of the population to allow for the different lifetime doses to which people would be exposed. For example, the general U.S. population is exposed to very low doses by eating foods on or in which residues of the chemical remain, while applicators receive much higher doses through dermal and inhalation exposure (see Chapter 7~. The risk analysis data were therefore summarized by displaying the increase in the maximum, or worst-case, lifetime probability of contract- ing cancer for members of each of seven population groups and for seven possible regulatory options (see Table 6.1~. In the analysis of the risk associated with endrin (U.S. EPA 1978b), not cancer but the likelihood of teratogenesis was the primary concern. Three groups of women may be exposed to significant doses of endrin: female pilots of endrin-spraying aircraft (probably a very small number of women), downwind neighbors exposed during the spraying operation, and women who eat fish from water contaminated by runoff and drainage from fields treated by endrin. For each of these groups a plausible daily dose (in milligrams per kilogram) was estimated and a margin of safety was computed according to the formula: Margin of Safety = Largest dose for which no effects were Ibsen n experimental animals Dose to which some (perhaps few) members of the population may be exposed A margin of safety of 300 or less was judged to indicate a significant risk. In general, as suggested by these illustrations, there is no attempt to be uniform in assessing the potentials of different pesticides for harming public health, wildlife, materials, and crops. Each analysis is adapted to particular circumstances. Comments and Recommendations The practices described above, representing current attempts to quantify the risks of using different pesticides, super from at least two serious deficiencies. The first is the noncomparability of the risks estimated for one pesticide with those of another estimated in a different form. The second, which was discussed at length above, is the unreliability inherent in estimates of change in human morbidity or mortality extrapolated from experiments with animals.

92 REGULATING PESTICIDES The lack of comparability is a consequence of the wide variety of ways in which a pesticide may inflict harm. There is no defensible formula for reducing all varieties of damage to human health to a common index of seriousness. Nevertheless, it is important that when similar consequences are at issue, they be estimated and reported in comparable ways. If this can be done, serious inconsistencies among decisions relating to different compounds will be minimized and the accumulation of useful experience in appraising risks, on the part of both the Administrator and the staff, will be facilitated. The appraisal of risks to human health can be systematized by applying the concept of the CAI together with analogous concepts. We have already discussed at length problems of measuring and expressing the potential carcinogenicity of a compound, and we concluded that the best method, generally, is to indicate carcinogenic activity relative to that of other compounds using a CAT. The indicator must then be combined with estimates of the numbers of people exposed to different doses of the compound to yield an overall assessment of the cancer risk that is posed. The question is how to do this. It is not meaningful to combine the CA' with estimates of exposure by multiplying them or by any other simple arithmetic formula. Some people like pesticide applicators are exposed to doses several orders of magnitude greater than others are exposed to. In the absence of a dose-response curve applicable to humans, it is not possible to aggregate the different population segments receiving widely different doses into an overall estimate of the eject of the use of a pesticide on public health. In terms of eject on public health, 1,000 N people each receiving a dose of D is not equivalent to N people each receiving a dose of 1,000 D, nor do we know of any reliable way to compare the effects of the two exposures. The results must be presented as a table or graph that shows the numbers of people exposed to different doses. Furthermore, the dose to which each population segment is exposed may be different under different regulatory options, the ejects of which can be indicated by a compara- tive exposure graph as illustrated in Figure 4.2. The illustration compares the doses to which three population segments are exposed under four regulatory options of increasing stringency. The CAP'S must be used in preparing such a comparison. To illustrate, let us suppose that Option A in Figure 4.2 is the unrestricted use of pesticide X while Option B involves banning its use in certain areas. The farmers in the prohibited areas can then be expected to resort to other expedients: some might use pesticide Y. others pesticide Z. others biological controls, and so on. The eject of these changes on exposure to pesticides X, Y. and Z cannot be foreseen with

Risk Assessment x LU cat o LL In o - 93 AnoIicators (1,000)~ Local Inhabitants (50,000) Consumers (220 million) - - - A R EG U LATO RY OPTI ON FIGURE 4.2 Comparative exposure graph (schematic). D precision, but it can be approximated with the help of the CAN'T. Suppose, for example, that under Option B when pesticide X is banned, Py percent of the crop will be treated with the substitute pesticide Y. Pz with pesticide Z. and so on. Using the assumptions described in Appendix B. it can be shown that if doses Do and Dy are not very dissimilar, the applicator population, for example, is consequently exposed to pesticide Y at a dose equivalent to Dye units of pesticide X where Dy = CA~y CAix Dy. That is, Dyeis the dose of pesticide X that produces the equivalent pathological eject of the dose of pesticide Y that applicators might be expected to receive under regulatory Option B. The same will pertain for

94 REGULATING PESTICIDES pesticide Z. In toto, the average exposure of applicators under Option B in terms of pesticide X equivalents will be PxDx + P'Dy + PzDz This is the number to be plotted on the chart. The approximating assumption used in making the comparison is that pathological response is proportional to dose for moderate ranges of doses, though not for large variations. Where better approximations are known (e.g., linearity instead of proportionality), they should be used. To this point, the risks associated with different options have been expressed in units of exposure to the pesticide under consideration. A final step in the presentation is to note, again using the CAN'T, how the carcinogenic activity of the pesticide compares with the activities of other pesticides currently in use or previously regulated. These compari- sons are discussed in Chapter 6, where benefits of the different regulatory options are compared with their risks, and later illustrated in Chapter 7. Again, the Committee has not studied other risks to public health as carefully as it has studied carcinogenicity. Nevertheless, it believes that many of the problems of appraising risks of mutagenicity, teratogenicity, and acute and chronic toxicity in humans are closely analogous to those encountered in the analysis of cancer risks to the extent that reliance is placed on extrapolations from bioassays using laboratory animals. The same methods of risk assessment should therefore apply. Indicators must be constructed showing the comparative potencies of different com- pounds in inducing mutations, abnormal offspring, and toxic effects. Consequences of different regulatory options can then be compared by the methods just described, using the appropriate activity indicators and, when available, human data. ANALYSIS OF ENVIRONMENTAL RISKS In addition to considering the risks to human health posed by an RPAR compound, the Agency is also obligated under 40 CFR 162.11 to identify and weigh any environmental risks associated with the chemical. Specifically, the environmental risk triggers are (1) acute toxicity to nontarget species, (2) chronic toxicity to members of endangered species, and (3) chronic toxicity to nontarget species (see Note to Chapter 2~. The environmental risk analyses performed by oPP's HED are some- what analogous to the human health risk analyses. In particular, the

Risk Assessment 95 environmental risk analyses attempt to determine the extent to which current use and exposure patterns, and the use-exposure patterns likely to arise under the various regulatory options, may prove lethal to nontarget organisms. oPP's environmental analyses are based upon either theoretical considerations or empirical evidence. For instance, oPP's presumption that endrin is acutely toxic to rabbits and pheasants was based on theoretical calculations of the endrin residues likely to be found on items consumed by these animals (U.S. EPA 1978b: 13~. The theoretical arguments were eventually modified to reflect the findings of actual residue studies submitted by an endrin registrant (U.S. EPA 1978b: 14~. In contrast to the acute toxicity presumption, the presumption that endrin is responsible for significant reductions in nontarget populations was based upon actual data on fish kills derived from the Pesticide Episode Reporting System (U.S. EPA 1978b: 181. Unfortunately, the data available on environmental hazards are often too incomplete to allow for the development of accurate, quantitative risk estimates. Realistically, there is currently no way of developing reliable estimates of, for example, the number of rabbits or pheasants that die each year from ingesting endrin residue on forage or seeds. Even in cases involving significant local population reductions, such as large fish kills, oPP may have little or no quantitative (or even qualitative) evidence. Position Document 2/3 for endrin notes, for example, that the Pesticide Episode Reporting System (which depends on voluntary reporting) is so unreliable that it missed at least 20 endrin-related fish kills over a 5-year period in Mississippi. As a result of these data shortages, the environmental risk analyses tend to rely heavily upon sketchy, perhaps even qualitative, information. The Committee has focused its attention in this report on health effects. This is not to say that it felt the assessment of environmental risks is not significant, but only to confess that the Committee chose not to study it in depth itself. Nevertheless, it is clear to the Committee that an improved data base is necessary. To this end, and on the basis of the Committee's observations and review of selected oPP position docu- ments, we suggest that EPA (1) devote more resources to environmental monitoring and (2) initiate more studies of environmental toxicology of selected pesticides. When quantitative environmental risk analyses are made, we further recommend that estimates be reported as ranges. As for human exposure analyses, the ranges should be presented as a pair of numbers, one showing the most-probable environmental risk and the other showing the maximum-plausible estimate.

96 REGULATING PESTICIDES RISKS TO STRUCTURES, MATERIALS, AND CROPS Generally speaking, risks to structures and materials entailed by the use of pesticides are negligible. On the other hand, pesticides may be harmful to crops grown In nearby fields, to livestock, or to commercial fisheries. In the latter instances, current practice is to estimate the monetary value of the decreases in yield or increases in cost of maintenance estimated to result from use of the pesticide. In the Comm~ttee's judgment, the methods currently used for making these estimates are straightforward and sound, although we recommend that such estimates be derived and reported for both the most-probable and max~mum-plausible cases. OVERALL ASSESSMENT OF RISKS The use of any pesticide entails a complex bundle of risks: risks to the health of different segments of the population, to wildlife, to vegetation, to crops and livestock, and to buildings and materials. Each of these risks is a result of several factors: the number of vulnerable elements exposed to the pesticide, the dose to which each element is exposed, and the potency or harmfulness of the pesticide. At some stage in the evaluation of regulatory options, appraisals of the different kinds of risks must be combined and compared with the costs of different options. How to consolidate appraisals of the individual types of risks and the extent to which they can be consolidated are among the principal concerns of Chapter 6. The assessments of the several types of risk reviewed in this chapter are necessary ingredients in that final appraisal. REFERENCES Crump, K.S., D.G. Hoel, C.H. Langley, and R. Peto (1976) Fundamental carcinogenic processes and their implications for low dose risk assessment. Cancer Research 36:2973- 2979. Donoso, J. and C.W. Collier (1978) Exposure Analyses for Lindane. Hazard Evaluation Division, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Unpublished) Epstein, S.S. (1976) Carcinogenicity of heptachlor and chlordane. The Science of the Total Environment 6: 103-154. Hayes, J., Jr. (1975) Toxicology of Pesticides. Baltimore: The Williams and Wilkins Company. Hoel, D.G., D.W. Gaylor, R.L. Kirschstein, U. Saffiotti, and M.A. Schneiderman (1975) Estimation of risks of irreversible, delayed toxicity. Journal of Toxicology and Environmental Health 1: 133-151.

Risk Assessment 97 Innes, J.R.M., B.M. Ulland, M.G. Valerio, L. Petrucelli, L. Fishbein, E.R. Hart, AJ. Pallotta, R.R. Bates, H.L. Falk, J.J. Gart, M. Klein, I. Mitchell, and J. Peters (1969) Bioassay of pesticides and industrial chemicals for tumorigenicity in mice: A preliminary note. Journal of the National Cancer Institute 42(6): 1101-1114. Interagency Regulatory Liaison Group (1979) Scientific Bases for Identifying Potential Carcinogens and Estimating Their Risks. Work Group on Risk Assessment, Em, Washington, D.C. (Unpublished. Available from Executive Assistant, HAG, Room 500, 1111 18th St., N.W., 20207) International Research and Development Corporation (1973) Report to the Velsicol Chemical Corporation. (Unpublished. Data with critique are also presented in Epstein 1976) Mantel, N. and W.R. Bryan (1961) 'Safety' testing of carcinogenic agents. Journal of the National Cancer Institute 27:455-470. Meselson, M.S. and K. Russell (1977) Comparison of carcinogenesis and mutagenesis potency. Pages 1473-1481, Book C, Origin of Human Cancer, edited by H.H. Hiatt, J.D. Watson, and J.A. Winsten. Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory. National Cancer Institute (1976) Bioassay of Chloroform for Possible Carcinogenicity. CAS No. 67-66-3. Washington, D.C.: U.S. Government Printing Office; Springfield, Va.: National Technical Information Service. National Cancer Institute (1977a) Bioassay of Chlordane for Possible Carcinogenicity. cats No. 57-749, NCI-CG-TR-8. Washington, D.C.: U.S. Department of Health, Education and Welfare. National Cancer Institute (1977b) Bioassay of Heptachlor for Possible Carcinogenicity. cats No. 76 41 8, NCI-CG-TR-9. DHEW Publication No. (NOSH) 77-809. Washington, D.C.: U.S. Department of Health, Education and Welfare. National Cancer Institute (1977c) Bioassay of Lindane for Possible Carcinogenicity. CAS No. 58-89-9, NCI-CG-TR-14. DHEW Publication No. (NOSH) 77-814. Washington, D.C.: U.S. Department of Health, Education and Welfare. National Cancer Institute (1978a) Bioassay of Aldrin and Dieldrin for Possible Carcinoge- nicity. CAS No. 309~2, NCI-CG-TR-21. DHEW Publication No. (em) 78-821. Washington, D.C.: U.S. Department of Health, Education and Welfare. National Cancer Institute (1978b) Bioassay of Chlorobenzilate for Possible Carcinogenici- ty. CAS No. 510-15-6, NCI-CG-TR-75. DHEW Publication No. (NOSH) 78-1325. Washing- ton, D.C.: U.S. Department of Health, Education and Welfare. National Cancer Institute (1978c) Bioassay of Dieldrin for Possible Carcinogenicity. CAS No. 60-57-1, NCI No. 22. CHEW Publication No. (NOSH) 78-882. Washington, D.C.: U.S. Department of Health, Education and Welfare. National Cancer Institute (1979) Bioassay of Endrin for Possible Carcinogenicity. CAS No. 72-2~8, NCI-CG-TR-12. DHEW Publication No. (NOSH) 79-812. Washington, D.C.: U.S. Department of Health, Education and Welfare. Olson, W.A., R.T. Habermann, E.K. Weisburger, J.M. Ward, and J.H. Weisburger (1973) Induction of stomach cancer in rats and mice by halogenated aliphatic fumigants. Journal of the National Cancer Institute 51: 199~1995. Pesticide and Toxic Chemical News (1978) Dr. Dubin resigns from CAG with stunning charges leveled at Dr. Albert. November 22: 2~26. Pesticide and Toxic Chemical News (1979) 'Weakly positive' evidence shows BAAM a likely carcinogen, CAL says. January 10:7~. Public Law 92-516 (1972) Federal Environmental Pesticide Control Act of 1972. 7 usc 135 (1972).

98 REGULATING PESTICIDES Severn, D.J. (1977a) Data Requirements for Exposure Analyses. (Note: This unpublished memorandum was prepared by D.J. Severn and is available from the Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. 20460) Severn, D.J. (1977b) Estimates of Human Exposure to Nitrosamines from the Use of Trifluralin and Trichlorobenzoic Acid Herbicides. Hazard Evaluation Divison, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Un- published) Severn, D.J. (1978a) Draft Procedures Manual for Preparation of Human Exposure Analyses. Hazard Evaluation Division, Office of Pesticide Programs, U.S. Environn~en- tal Protection Agency, Washington, D.C. (Unpublished) Severn, D.J. (1978b) Exposure Analysis for Chlorobenzilate. Hazard Evaluation Division, Office of Pesticide Programs, U.S. Environmental Protection Agency, Washington, D.C. (Unpublished) U.S. Environmental Protection Agency (1975) Scientific and Technical Assessment Report on Vinyl Chloride and Polyvinyl Chloride. STAR Series, CAS No. 75-014, EPA-600/~75- 004. Washington, D.C.: U.S. Environmental Protection Agency. U.S. Environmental Protection Agency (1976) Health Risk and Economic Impact Assessments of Suspected Carcinogens: Interim Procedures and Guidelines. 41 Federal Register (102)21402-21405. U.S. Environmental Protection Agency (1978a) Chlorobenzilate: Position Document 3. Special Pesticide Review Division, Office of Pesticide Programs, U.S. EPA. (Note: This unpublished report was prepared under the general direction of J.B. Boyd, Project Manager, and is available from oPP' Washington, D.C. 20460) U.S. Environmental Protection Agency (1978b) Endrin: Position Document 2/3. Special Pesticide Review Division, Office of Pesticide Programs, U.S. EPA. Mote: Us unpublished report was prepared under the general direction of K. Barbehenn, Project Manager, and is available from oPP, Washington, D.C. 20460) Wolfe, H.R., W.F. Durham, and J.F. Armstrong (1967) Exposure of workers to pesticides. Archives of Environmental Health 14:622~33.

Next: 5. Benefit Assessment »
Regulating Pesticides Get This Book
×
Buy Paperback | $85.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF
  1. ×

    Welcome to OpenBook!

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

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

    No Thanks Take a Tour »
  2. ×

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

    « Back Next »
  3. ×

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

    « Back Next »
  4. ×

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

    « Back Next »
  5. ×

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

    « Back Next »
  6. ×

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

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

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

    « Back Next »
Stay Connected!