Pesticides in the Diets of Infants and Children, by NRC


Postby admin » Sun Mar 13, 2016 3:45 am

Part 1 of 2

4. Methods for Toxicity Testing

THE PURPOSE OF THIS CHAPTER is to familiarize the reader with the testing that is currently conducted by a manufacturer prior to and during the process of submitting a petition to register a pesticide. Codified toxicologic evaluation of potential pesticides has been a requirement in the United States for approximately 50 years. The testing requirements and guidelines continue to evolve based on new science. This chapter identifies the current testing that is pertinent to the young animal and young human as well as aspects of testing that are needed to fill the data gaps to better ensure the protection of infants and children. The current testing guidelines can be found in Pesticide Assessment Guidelines issued by the Environmental Protection Agency (EPA, 1991a,b).

Data, including those derived from toxicity testing, crop residue analyses, environmental fate testing, and ecotoxicology testing, are generated by the manufacturer of a pesticide to meet the mandatory requirements of the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) for pesticide registration. Although these data are essential to the EPA's registration process, other data generated by EPA itself, as well as by other government institutions and academia, are considered in the registration decision-making process.

EPA has issued 194 registration standards on 350 chemicals used as active and inert ingredients in pesticide products. These standards are published by EPA and are intended to upgrade and update the data base on a previously registered pesticide or class of pesticide products. They call for additional studies in the areas of toxicity testing, crop residue analyses, environmental fate, and ecotoxicology testing. This testing must be conducted within an EPA-mandated time frame to allow for the continued registration of given product. The list of pesticides for which registration standards have been issued is referred to as List A and can be found in Appendix I of the Federal Register notice of February 22, 1989. Under the FIFRA Amendments of 1988, the data bases on the remaining registered pesticide products are being upgraded in five phases over a 9-year period.

TABLE 4-1 End Points for Various Toxicity Studies

Study / End Points

Developmental toxicity / Fetus: mortality, growth retardation, skeletal variations, gross external malformations, soft tissue/internal organ defects
Female parent: general toxicity

Reproductive Toxicity / Male parent: general toxicity, effects on fertility, reproductive organ changes
Offspring: effects on viability, sex ratio, growth, behavior

Carcinogenicity / Tumor development and general toxicity

Neurotoxicity / Behavior, function, and motor activity deficits; microscopic nervous tissue changes

Mutagenicity / Heritable lesions leading to altered phenotypes


The first sections of this chapter describe in detail the present toxicity testing procedures for pesticides in relation to their registered use patterns and EPA's proposed changes or additions to these procedures. The conclusions and recommendation of the committee for further changes and additions to the toxicity testing battery to allow for more adequate consideration of the special testing needs for infants and children are presented.


Toxicity studies are required to assess potential hazards to humans through the acute, subchronic, and chronic exposure of laboratory animals to pesticides. The more specific types of toxicity that are determined include carcinogenicity; developmental (including teratogenicity in offspring) and reproductive toxicity; mutagenicity; and neurotoxicity (Table 4-1). Detailed information on the metabolism or biotransformation of the pesticide is also obtained. Consideration is given to testing individual metabolites in animals, and in or on pesticide-treated plants to which humans could exposed through their diet. The extent of metabolite testing required depends on the level of potential toxicity and environmental persistence of the metabolite. With the exception of the acute toxicity tests, most tests are conducted to determine the nature of any toxicity that can be produced by repeatedly dosing animals over an extended period. The results enable toxicologists to estimate the safety of a material of humans (Loomis, 1978).

Weil (1972) published the following set of guidelines, which reflected a consensus among toxicologists. These should be considered before initiating a toxicity test:

1. Use, wherever practical or possible, one or more species that biologically handle the material qualitatively and/or quantitatively as similarly as possible to man. For this, metabolism, absorption, excretion, storage and other physiological effects might be considered.

2. Where practical, use several dose levels on the principle that all types of toxicologic and pharmacologic actions in man and animals are dose-related. The only exception to this should be the use of a single, maximum dosage level if the material is relatively nontoxic; this level should be a sufficiently large multiple of that which is attainable by the applicable hazard exposure route, and should not be physiologically impractical.

3. Effects produced at higher dose levels (within the practical limits discussed in 2) are useful for delineating mechanism of action, but for any material effect, some dose level exists for man or animal below which this adverse effect will not appear. This biologically insignificant level can and should be set by use of a proper uncertainty factor and competent scientific judgment.…

4. Statistical tests for significance are valid only on the experimental units (e.g., either litters or individuals) that have been mathematically randomized among the dosed and concurrent control groups….

5. Effects obtained by one route of administration to test animals are not a priori applicable to effects by another route of administration to man. The routes chosen for administration to test animals should, therefore, be the same as those to which man will be exposed. Thus, for example, food additives for man should be tested by admixture of the material in the diet of animals.

In general, Weil's guidelines are considered by EPA in its toxicity testing requirements and subsequent evaluation of results for pesticides. One exception to Weil's points is found in his guideline 3. EPA does not recognize the existence of a dose level at which a carcinogen will not exert its effect. For carcinogens, EPA generally accepts a risk of 10-6, as extrapolated from bioassays using the nonthreshold modification of the linearized multistage model of Armitage and Doll (1954), as adequate for the protection of humans.

The selection of animal species for toxicity tests depends on life span, behavior, availability, and overall costs. EPA recommends using rats for subchronic, chronic, carcinogenicity, and reproduction studies; mice for carcinogenicity studies; and dogs for subchronic and chronic studies. Rats are routinely used for acute oral and inhalation studies and rabbits for eye and skin irritation studies and acute dermal studies. One exception to this is the use of guinea pigs for dermal sensitization testing. The rat and rabbit are recommended for developmental toxicity (teratogenicity) testing. Justification must be provided for the use of species other than those outlined above.

The number of animals to be tested in each dose group depends on a number of factors, including the purpose of the experiment, the required sensitivity of the study, the reproductive capacity and the fertility of the species, economic aspects, and the availability of animals (IPCS, 1990). Table 4-2 lists the minimum number of animals required by EPA for some toxicity studies. For the most part, these numbers are consistent with those recommended by the International Program Chemical Safety (IPCS).

The selection of dose levels for subchronic studies should be based on the results of acute toxicity testing, on range-finding studies, and on pharmacokinetic (metabolism, including rate in various tissues) data. For subchronic studies, four dose groups of animals should be included: a control group; a low-dose group (a dose that produces no compound related toxicity); a mid-dose group (a dose that elicits some minimal signs of toxicity); and a high-dose group (a dose that results in toxic effects but not in an incidence of fatalities that would prevent a meaningful evaluation; for nonrodents, there should be no fatalities) (EPA, 1984). This same guidance is relevant to chronic toxicity and reproduction studies. For teratology studies, the highest dose tested should elicit some signs of maternal toxicity, but the toxicity should not obscure the results.

The one notable exception to this guidance pertains to carcinogenicity studies. The highest dose levels for these studies should be at a maximum tolerated dose (MTD), as determined in 90-day toxicity studies in the appropriate test species and from pharmacokinetic information on the material being tested. The Committee on Risk Assessment Methodology of the National Research Council (NRC) recently examined the criteria for the MTD and other doses used in carcinogenicity studies (NRC, 1993). The EPA has issued its own guidance for the selection of this dose level. Some of the factors to consider in selecting an MTD are: 10% decrement in body weight gain in 90-day study; observation of potential life-threatening lesions during microscopic examination of organs, e.g., liver necrosis; significant inhibition of cholinesterase activity in two biological compartments, such as brain and plasma; and significant signs of anemia or other biologically relevant effects on blood.


TABLE 4-2 Animal Model Requirements in Toxicity Studies

a Either males or females may be used in this test.

b The number of animals used depends on the method used. Several different experimental methods are acceptable.
c EPA prefers that one male rat be housed with one female during mating.

d Number of pregnant females required.

e 50 rats and 50 mice of each sex.

SOURCE: EPA, 1984.

In general, EPA has set a cap on dosing of 1.0 g/kg/day for toxicity tests other than acute studies. This dose level is referred to as the limit dose and corresponds to approximately 20,000 ppm in the diet of rats, 7,000 ppm in the diet of mice, and 40,000 ppm in the diet of dogs.

The duration of exposure for toxicity testing of a pesticide depends on the expected duration of human exposure to the pesticide in practice. The typical length of various toxicity tests and the number of doses administered are shown in Table 4-2. Repeated dosing refers to dosing once per day for the designated number of days. When the material is given to the test animals in their diet, dosing is usually continuous for 7 days a week. If the material is administered by gavage (oral bolus dose), by dermal application, or by inhalation, doses are frequently given 5 days a week, which is acceptable to EPA because of practical considerations (EPA, 1984).

The type of statistical analysis performed on the toxicity data resulting from these studies depends on the type of data under consideration (see, for example, Gad and Weil, 1982, for review). Interpreting the meaning of statistical significance for any particular parameter depends on the dose level at which it was achieved, the biological significance of the finding, and the normal spontaneous occurrence of this finding in the strain and species being tested.

For regulatory purposes, the no-observed-effect level (NOEL) is defined as a dose level at which no effects attributable to the pesticide under test can be found. A no-observed-adverse-effect level (NOAEL) can also be determined for each study; however, EPA does not routinely use the NOAEL to regulate pesticide usage. To establish a NOAEL, the toxicologist must determine what is and what is not adverse effect, which can be defined differently by different scientists. For example, effects such as hair loss can be considered adverse by some and not by others. Plasma and red blood cell cholinesterase inhibition can be viewed as either an adverse effect or simply as a market of exposure to a pesticide.

EPA uses the NOEL to calculate the acceptable daily intake (ADI) of the pesticide under consideration. More recently, the EPA has replaced the ADI with the reference dose, or RfD. Chronic studies, such as reproduction studies and lasting 1 year or longer in the rat or dog are used for this purpose. EPA does not routinely use the NOEL determined from teratology (developmental toxicity) studies for calculating ADIs because the observed effect are not considered chronic; however, these NOELs can be used to support the calculated ADI. EPA does routinely use developmental toxicity NOELs for other types of risk assessments, such as calculating the risk from acute, daily dietary or occupational exposure or from exposure of homeowners to a developmental toxicant.

EPA's toxicity testing requirements for food and nonfood use pesticides have been published in 40 CFR Part 158. In general, for food use chemical with maximum human exposure, the following toxicity tests are required:

• acute oral toxicity
• acute dermal toxicity
• acute inhalation toxicity
• primary eye irritation
• primary dermal irritation
• chronic feeding toxicity
• dermal sensitization
• acute neurotoxicity
• 90-day toxicity
• 21-day dermal toxicity
• 90-day neurotoxicity study
• reproduction study
• carcinogenicity
• developmental toxicity
• mutagenicity tests
• general metabolism study

More than 30% of the tests for pesticides submitted to EPA in the past have been rejected. Those rejected must be resubmitted until they are in conformance with EPA criteria before registrations can be granted. The criteria for rejection are summarized in Table 4-3. Some of them fall in the category of regulatory policy; others involve scientific concerns. The most commonly cited reason for noncompliance is lack of characterization of the test material. To improve the quality of testing and incorporate new scientific methods in its testing requirements, EPA is currently revising the 40 CFR Part 158 data requirements for food and nonfood use pesticides. The proposed revisions to these requirements can be found in Table 4-4.


General Description

Acute toxicity studies provide information on the potential for health hazards that may arise as result of short-term exposure. Determination of acute oral, dermal, and inhalation toxicity is usually the initial step in evaluating the toxic characteristics of a pesticide. In each of these tests the animal is exposed to the test material only once on 1 day. Together with information derived from primary eye and primary dermal irritation studies (also 1 dose on 1 day), which assess possible hazards resulting from pesticide contact with eyes and skin, these data provide a basis for precautionary labeling and may influence the classification of a pesticide for restricted use. Acute toxicity data also provide information used to determine the need for child-resistant packaging, for protective clothing requirements for applicator, and for calculation of farm worker reentry intervals. A minimum number of animals, usually adults, are used in these studies and only the end points of concern are monitored, i.e., mortality, observable skin or eye effects, dermal sensitization, and observable neurotoxic behavioral changes. One exception is the inclusion of microscopic examination of neural tissues in the newly required acute neurotoxicity study.

EPA's Proposed Changes

Guideline number 81-1 (EPA, 1984), acute oral study in the rat, would be revised to include special visual system testing, which would be required for all organophosphate pesticide and other pesticides known to affect the visual system.

TABLE 4-3 Summary of EPA Rejection Factors

Guideline / Rejection Factor

Acute Oral Toxicity (81-1) / Lack of characterization of the test material
Inadequate dose levels to calculate LD50

Acute Dermal Toxicity (81-2) / Lack of characterization of the test material
Inadequate percentage of body surface area exposed
No quality assurance statement
Improper number of animals tested per dose group
Only one sex tested
Omitted source, age, weight, or strain of test animal

Acute and 90-Day Inhalation (81-3 and 82-4) / Less than 25% of particles were <1 µm; LC50 could not be calculated; highest concentration did not produce toxicity
Inadequate reporting of exposure methodology
Protocol errors
Lack of characterization of the test material
Compound preparation
Chamber concentration not measured

Primary Eye Irritation (81-5) / Lack of characterization of the test material

Primary Dermal Irritation (81-5) / Lack of characterization of the test material
No quality assurance statement and/or no Good Laboratory Practice (GLP) statement
Improper test material application/preparation
Omitted source, age, weight, or strain of test animal
Missing individual/summary animal data

Dermal Sensitization (81-6) / Control problems
Dosing level problems
Lack of characterization of the test material
Unacceptable protocol or other protocol problems
Individual animal scorers or data missing
Scoring method or other scoring problem
Reporting deficiencies or no quality assurance statement

90-Day Feeding—Rodent (82-1(a)) / A NOEL was not established
Lack of characterization of the test material or incorrectly reported
Lack of clinical chemistry and/or lack of histopathology

90-Day Feeding—Nonrodents (82-1(b)) / Reporting deficiencies
Lack of characterization of the test material
A NOEL was not established
An investigation parameter missing

Guideline / Rejection Factor

90-Day Feeding—Nonrodents (cont.) (82-1(b)) / Information on the pilot study and other problems associated with dose level selection
An investigational parameter missing
Information on the pilot study and other problems associated with dose level selection

21-Day Dermal Toxicity (82-2) / Lack of characterization of the test material
Raw data analyses incomplete or missing
A systemic NOEL was not established
Inadequate percentage of body surface area exposed in each dose group
Insufficient number of dose levels tested

90-Day Dermal Toxicity (82-3) / Lack of characterization of the test material
A systemic NOEL was not established
Incomplete/missing raw animal data analyses
Insufficient number of dose levels tested
Poorly controlled test environment

Chronic Feeding/Carcinogenicity—Rats (82-3(a) and (83-2(a)) / Missing histopathology information
Missing information in study reports
MTD was not achieved
Missing historical control data
Lack of characterization of the test material
Deficiencies in reporting the study data

Carcinogenicity—Mice (83-2(b)) / Histopathology information missing
MTD was not achieved
Lack of historical control data
Information missing in study reports
Lack of characterization of the test material
Deficiencies in reporting of study data

Developmental Toxicity—Rodents (83-3(a)) / Missing historical controls
Lack of characterization of the test material
Information missing or requiring clarification of the laboratories' methods
Information missing or requiring clarification of the laboratories' results
A NOEL was not established
Statistical problems
Did not use conventional assessments for skeletal or visceral examinations

Developmental Toxicity—Nonrodents (83-3(b)) / Clarification of laboratory procedures of interpretation of the data
Individual maternal or fetal data missing
Missing historical controls
Lack of characterization of the test material
Excessive maternal toxicity

Guideline / Rejection factor

Developmental Toxicity—Nonrodents (cont.) (83-3(b)) / A NOEL was not established
Statistical problems

Reproduction (83-4) / Information missing from laboratory results
Lack of characterization of the test material
Information missing or requiring clarification of laboratory methods or results
Missing historical controls
A NOEL was not established due to effects at the lowest dose tested
Low fertility and/or inadequate number of animals were used per dose level
A NOEL was not established in the absence of reproductive effects

Metabolism (85-1) / Inadequate or missing data on identification of metabolites
Improper methodology or dosing regimen
Inadequate number of animals were used in the dose groups
No individual animal data
Improper reporting
Inadequate or missing tissue residue analysis data
Testing at only one dose level
Only one sex of animal used
Lack of an intravenous dose group
No collection of 14 CO2

Dermal Penetration (85-2) / Incomplete/missing data evaluation
Improper test material preparation/application
Raw data missing and incomplete summary tables
No signed quality assurance statement
Missing purity or concentration of test material

SOURCE: P. Fenner-Crisp, EPA, personal communication, 1992

The additional acute study proposed in guideline number 81-4 is acute neurotoxicity testing in the rat. This study would be required for all pesticide registrations (food and nonfood) and experimental use permits (EUPs), and it would include assessments of function and activity as well as histopathological (microscopic) examination of selected neural tissue. EPA presently requires that this study be conducted by manufacturers wishing to reregister.


General Description

Subchronic exposures do not elicit effects that have a long latency period (e.g., carcinogenicity). However, they do provide information on health hazards that may result from repeated exposures to a pesticide over a period up to approximately 30% of the lifetime of a rodent. Subchronic tests also provide information necessary to select proper dose levels for chronic studies, especially for carcinogenicity studies for which an MTD must be selected. According to EPA (1984), rats selected for these studies should be started on the test material shortly after weaning, ''ideally before the rats are 6 and, in any case, not more than 8 weeks old." For dogs, dosing should begin when they are 4 to 6 months of age and "not later than 9 months of age."

Most subchronic toxicity studies monitor clinical or behavioral (neurological) signs of toxicity, body weight, food consumption, eye effects, certain plasma or serum and urine parameters, organ weights, and gross and microscopic pathology. Clinical and behavioral signs of toxicity are observed and recorded daily. They can consist of activity, gait, excreta, hair coat, and feeding and drinking patterns. Body weight and food consumption data are routinely recorded throughout the study at intervals (usually weekly) determined by the length of the study. Ophthalmoscopic examinations are conducted at the beginning of the study and, typically, just before it terminates. The laboratory parameters typically examined are summarized in Table 4-5.

The results of hematology testing indicate whether, for example, the chemical affects blood cell formation and survival, clotting factors, and platelets. Clinical chemistry and urinalysis results can indicate possible kidney, liver, pancreas, and cardiac function or toxicity as well as any electrolyte imbalance. Urinalysis results can indicate adequacy of kidney, liver, and pancreas function.

After necropsy, the weights of certain organs are also recorded. These organs generally include brain, gonads, liver, and kidneys, which are the four required according to EPA testing guidelines (EPA, 1984). If toxicity is known to occur in another organ from previous testing, the weight of this organ should also be reported. For thyroid toxicity, for example, the weight of the thyroids should be recorded. Changes from untreated control animals are generally an indication of potential toxicity in this organ.

A complete necropsy is performed after sacrifice or death of the test animal. Generally all tissues are examined, and those saved for microscopic examination are aorta, jejunum, peripheral nerve, eyes, bone marrow, kidneys, cecum, liver, esophagus, colon, lung, ovaries, duodenum, lymph nodes, oviduct, brain, stomach, pancreas, skin, mammary gland, rectum, heart, spleen, spinal cord, testes, musculature, thyroid/parathyroid, pituitary, epididymis, salivary glands, ileum, adrenals, thymus, trachea, urinary bladder, accessory sex organs, and gallbladder.

The data described above [below] are not required for all subchronic studies. For the 21-day dermal study, for example, only limited necropsy data are required.


TABLE 4-4 Toxicity Data Requirements Proposed by EPA for Food and Nonfood Uses of Pesticides

a Food use includes terrestrial food and feed, aquatic food, greenhouse food, and indoor food.

b Nonfood use includes terrestrial nonfood, aquatic nonfood outdoor, aquatic nonfood industrial, aquatic nonfood residential, greenhouse nonfood, forestry, residential outdoor, indoor nonfood, indoor medical, and indoor residential.

c R = required; brackets [] indicate data requirements that apply when an experimental use permit is being sought.

d MP = manufacturing-use product; TGAI = technical grade of the active ingredient.

e EP = end-use product.

f CR = conditionally required.

g PAI = pure active ingredient; PAIRA = pure active ingredient, radiolabeled.

h Choice of several substances, depending on studies required.

Notes for Table 4-4: Specific Conditions, Qualifications, or Exceptions to the Designated Test Procedures

1. Not required if test material is a gas or highly volatile.

2. Not required if test material is corrosive to skin or has pH <2 or >11.5; such a product will be classified as toxicity category 1 on the basis of potential eye and dermal irritation effects.

3. Required when the product consists of, or under conditions of use will result in, an inhalable material (e.g., gas, volatile substances, or aerosol/particulate).

4. Required unless repeated dermal exposure does not occur under conditions of use.

5. Required for uncharged organophosphorus esters, thioesters, or anhydrides of organophosphoric, organophosphonic, or organophosphoramidic acids or of related phosphorothioic, phosphonothioic, or phosphorothioamidic acids, or other substances that may cause the neurotoxicity sometimes seen in this class.

6. Additional measurements such as cholinesterase determinations for certain pesticides (e.g., organophosphates and carbamates) may also be required. The route of exposure should correspond to a primary route of human exposure.

7. Required if intended use of the pesticide is expected to result in human exposure via the oral route.

8. All 90-day subchronic studies can be designed to simultaneously fulfill the requirements of the 90-day neurotoxicity study.

9. Studies must include additional end points so as to provide an immunotoxicity screen in the rodent. An equivalent independent study may fulfill the requirements for an immunotoxicity screen.

10. In most cases, where the theoretical maximum residue contribution (TMRC) exceeds 50 percent of the reference dose (Rfd), a 1-year (or longer) interim report on a chronic (2-year) feeding study is required to support a temporary tolerance. This report is to be in addition to the 90-day feeding studies in rodents and nonrodents.

11. If the pesticide is found to leach into groundwater or may contaminate drinking water, a 90-day drinking water study may be required unless data demonstrate that there are no significant differences in toxicity observed when the test material is administered in feed versus when the test material is administered in drinking water. This study may be requested in addition to any 90-day oral studies that may be required.

12. Required if intended use of the pesticide is expected to result in human exposure via the dermal route and data from a subchronic 90-day dermal toxicity study are not required.

13. For nonfood uses, a 90-day dermal toxicity study is required, since intended use of the pesticide is expected to result in repeated dermal exposure of humans.

14. For food uses, required if: (a) the active ingredient of the product is known or expected to be metabolized differently by the dermal route of exposure than by the oral route, and a metabolite of the active ingredient is the toxic moiety; (b) the active ingredient of the product is classified as toxicity category I or II on the basis of acute dermal toxicity data.

15. Required if the active ingredient is a gas at room temperature or if use of the product results in respirable droplets and use may result in repeated inhalation exposure at a concentration likely to be toxic, regardless of whether the major route of exposure is inhalation

16. Required for substances when statistically or biologically significant effects were seen in the acute study (Guideline 81-7), or if other available data indicate that the substance can cause this type of delayed neurotoxicity.

17. Required if either of the following criteria is met: (a) use of the pesticide is likely to result in repeated human exposure over a significant portion of the human life span (e.g., products intended for use in and around residences, swimming pools, and enclosed working spaces or their immediate vicinity); (b) the use requires a tolerance for the pesticide or an exemption from the requirement to obtain a tolerance for the pesticide or an exemption from the requirement to obtain a tolerance, or requires issuance of a food additive regulation.

18. Based on acute and subchronic neurotoxicity studies, and/or on other available data, a functional observational battery, an assessment of motor activity, and perfusion neuropathology may be required.

19. Studies designed to simultaneously fulfill the requirements of both the chronic feeding and carcinogenicity studies (i.e., a combined study) may be conducted. Minimum acceptable study durations for chronic feeding and carcinogenicity studies are as follows: chronic rodent feeding study (food use pesticide)—24 months; chronic rodent feeding study (nonfood pesticide)—12 months in usually sufficient; chronic nonrodent (i.e. dog) feeding study—12 months; mouse carcinogenicity study-18 months; and rat carcinogenicity study—24 months.

20. Required active ingredients or any of their metabolites, degradation products, or impurities are structurally related to a recognized carcinogen, cause mutagenic effects as demonstrated by in vitro or in vivo testing, or produce a morphologic effect in any organ (e.g., hyperplasia, metaplasia) in subchronic studies that may lead to neoplastic change. The use requires a tolerance for the pesticide or exemption from the requirement to obtain a tolerance or requires the issuance of a food additive regulation. Use of the pesticide product is likely to result in exposure of humans over a portion of the life span that is significant in terms of either the timing or duration of exposure (e.g., pesticides used in treated fabrics for wearing apparel, diapers, or bedding; insect repellents applied directly to the skin; swimming pool additives; or constant-release indoor aerosol pesticides).

21. Range-finding studies of at least 90 days duration in rats and mice are generally required to determine dose levels adequate to demonstrate an MTD in carcinogenicity studies. A subchronic 90-day oral study conducted in accordance with Guideline 82-1 may also be acceptable for this purpose.

22. Testing in two species is required for food uses. For products intended for nonfood uses, testing in two species is required if significant exposure of human females of child-bearing age may reasonably be expected. For other nonfood uses, testing in at least one species is required. A study in one species is required to support a temporary tolerance.

23. Testing in a second species is required if significant developmental toxicity is observed after testing in the first species.

24. The test substance or vehicle is usually administered by oral intubation, unless the chemical or physical characteristics of the test substance or pattern of human exposure suggest a more appropriate route of administration.

25. Under certain conditions where a pesticide is determined to be a developmental toxicant (e.g., after oral dosing), additional testing via other routes (e.g., dermal) may be required.

26. Required to support products intended for food and nonfood uses if the use is likely to result in exposure of humans over a portion of the life span that is significant in terms of the frequency, magnitude, or duration of exposure (e.g., pesticides used in treated fabrics for wearing apparel, diapers, or bedding; insect repellents applied directly to the skin; swimming pool additives; or constant-release indoor pesticides used in aerosol form). Also may be required for nonfood uses if adverse effects on organs of the reproductive system are observed in 90-day or other studies, and/or if developmental toxicity is demonstrated by available data (Guideline 83-3).

27. In most cases, where the TMRC exceeds 50% of the RfD, a first-generation (or longer) interim report on a multigeneration reproduction study is required to support a temporary tolerance.

28. Conditionally required to more fully assess any of the manifestations of developmental toxicity. These studies permit assessment of potential functional deficits that cannot be evaluated in the classical developmental toxicity study (Guideline 83-3). Protocols for these studies are usually designed on a case-by-case basis.

29. On the basis of acute and subchronic neurotoxicity studies, and/or on other available data, a developmental neurotoxicity study may be required. For this type of postnatal study, a guideline is available.

30. An initial battery of mutagenicity tests with possible confirmatory testing is minimally required. Also, results from other mutagenicity tests that may have been performed and as complete a reference list as possible shall be submitted. Subsequent testing may or may not be required based on the evidence available to EPA's Office of Pesticide Programs in accordance with the objective and considerations for mutagenicity testing. Current protocols for tests in the initial battery and other mutagenicity tests are given in the EPA's Office of Pesticides and Toxic Substances Health Effects Testing Guidelines (40 CFR Part 798, Subpart F—Genetic Toxicity). Because of the rapid improvements in the field, applicants are encouraged to discuss test selection, protocol design, and results of preliminary testing with the agency.

31. Choice of assays using either mouse lymphoma L5178Y cells, thymidine kinase (tK) gene locus, maximizing assay conditions for small colony expression and detection; Chinese hamster ovary (CHO) or Chinese hamster lung fibroblast (V79) cells, hypoxanthine-guanine phosphoribosyl transferase (hgprt) gene locus, accompanied by an appropriate in vitro test for clastogenicity; or CHO cells strain AS52, xanthine-guanine phosphoribosyl transferase (xprt) gene locus.

32. Choice of assays; initial consideration usually given to rodent bone marrow, using either metaphase analysis (aberrations) or micronucleus assay.

33. Required for all food uses and when chronic and/or carcinogenicity studies are required. Also may be required if significant adverse effects are observed in toxicology studies (e.g., reproduction and developmental toxicity).

34. May be required, on a case-by-case basis, to support registration of an end-use product if cats, dogs, cattle, pigs, sheep, horses, or other domesticated animals will be exposed to the pesticide product, including, but not limited to, exposure through direct application for pest control and consumption of treated feed.

35. Dermal penetration studies are required for compounds that have serious toxic effects, as identified in oral or inhalation studies, and for which a significant route of human exposure is dermal. Thus, this study is required when any of the following exposure studies are required: passive dosimetry—dermal exposure (Guidelines 133-3, 231 or 233), foliar dislodgeable residue dissipation (Guideline 132-1), soil dislodgeable residue dissipation (Guideline 132-1), and indoor surface residue dissipation, unless the toxicity studies (including Guidelines 82-6, 82-7 83-1, 83-2 83-3, 83-4 and 83-6) that triggered the need for these exposure studies were conducted via the dermal route of dosing. Registrants should work closely with the agency in developing an acceptable protocol for performing dermal penetration studies.

36. Special testing (acute, subchronic, and/or chronic) is required for organophosphates, and may be required for other cholinesterase inhibitors and other pesticides that have demonstrated a potential to adversely affect the visual system. Registrants should consult with the agency for development of protocols and methodology prior to initiation of studies.

37. Testing of the end-use product dilution is required if it can be reasonably anticipated that the results of such testing may meet the criteria for restriction to use by certified applicators specified in 40 CFR 152.170(b) or the criteria for initiation of special review specified in 40 CFR 154.7(a)(1).

SOURCE: Code of Federal Regulations, Title 40, Parts 150 to 189, 11992.


TABLE 4-5 Laboratory Parameters Measured in Various Data Categories

SOURCE: EPA, 1984.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 3:53 am

Part 2 of 2

EPA's Proposed Changes

For the 90-day oral study (Guideline 82-1; EPA, 1984) in the rodent and other test species, three changes are proposed:

• the studies would be required for all pesticide uses that could result in oral exposure of humans and would not depend on frequency, magnitude, or duration of exposure;

• this study could be modified to include additional end points for neurotoxicity and immunotoxicity; and

• a separate drinking water study, in addition to the other dietary studies, could be required if the pesticide were found to leach into groundwater or contaminate drinking water.

Either a 21-day dermal (Guideline 82-2; EPA, 1984) or a 90-day dermal study (Guideline 82-3; EPA, 1984) would be required (not conditionally required as in the past) to support all registrations. A 21-day study would be required for all food uses, except when acute dermal toxicity is observed. A 90-day dermal study would ordinarily be required for all nonfood uses. Special tests for neurotoxicity or immunotoxicity could be added to these studies if these toxicity end points are not studied in other tests required for a particular pesticide.

A 90-day inhalation study (Guideline 82-4; EPA, 1984) would be required more frequently whether or not the major route of exposure is inhalation, especially for a nonfood use pesticide that is a gas or whose use generates respirable droplets. The requirement for a 21-day inhalation study for a tobacco use pesticide would be deleted. Special tests for neurotoxicity or immunotoxicity could be added if those end points are not studied in other EPA-required toxicity studies.

The conditionally required 90-day neurotoxicity study in the hen or mammal (Guideline 82-5; EPA, 1984) would be replaced by two new studies:

• a 28-day delayed neurotoxicity study in the hen (Guideline 82-6; EPA, 1984), which would be conducted under the same conditions as the 90-day study; and

• a 90-day neurotoxicity study in the rat (Guideline 82-7; EPA, 1984) (previously required only conditionally) to support all registrations and food/feed use EUPs (this is now required only for organophosphate and carbamate pesticides).

Testing would include assessments of function (functional observation battery), motor activity, and histopathological examination of the nervous system.


General Description

Information derived from chronic studies is used to assess potential hazards resulting from prolonged and repeated exposure to a pesticide over a large portion of the human life span. These studies usually last 12 to 24 months. Of particular importance are long-term carcinogenicity studies, the purpose of which is to observe the test animals for the development of neoplastic lesions after a lifetime of exposure at dose levels up to and including the MTD determined from subchronic testing.

The emphasis of the carcinogenicity study is the detection of tumors in animals. For these studies, both concurrent and historical control data are used to evaluate the relevance of tumors. Historical control data should be derived from studies in the same species and strain and, preferably, in the same laboratory as used in the study under consideration. Carcinogenicity studies should be 24 months long in rats and 18 months long in mice. The age of test animals in carcinogenicity (rat and mouse) studies and other chronic (rat and dog) studies is determined by the same criteria as for subchronic toxicity studies. The parameters to be examined in carcinogenicity studies are also generally the same as those discussed above for subchronic and chronic studies, except that clinical chemistry and urine parameters are not required and only limited hematology data are required.

EPA's Proposed Changes

Modifications to chronic feeding studies in two species (rodent and nonrodent; Guideline 83-1; EPA, 1984) may be required to include additional end points for neurotoxicity or immunotoxicity or special visual system toxicity (for organophosphates) if these were not tested in other studies.

Range-finding studies of at least 90 days duration in rats and mice will generally be required to determine dose levels that are adequate to test the carcinogenicity (Guideline 83-2; EPA, 1984) of a pesticide. Studies conducted to satisfy the requirement for Guideline 82-1 (EPA, 1984) will also be acceptable to satisfy this 90-day study requirement.


General Description

Developmental toxicity studies are designed to assess the potential of developmental effects in offspring resulting from the mother's exposure to the test substance during pregnancy. These effects include death of the developing organism, structural abnormalities, altered growth, and functional deficiencies. In addition to the classic teratology (now called developmental toxicity) study, a postnatal study is required by the EPA on a case-by-case basis. It is in this study that functional deficiencies are best studied.

The EPA prefers that the rat and the rabbit be used in these studies; however, hamster and mouse are also acceptable. Doses should be administered over the period of major organogenesis (major visceral and skeletal formation) in the fetus. The maternal animals only are dosed in this study and only for specified periods. When day 0 is the day that evidence of mating was observed, the rat and mouse are dosed on days 6 through 15; the rabbit, days 6 through 18; and the hamster, days 6 through 14. Dosing is usually administered by gavage (oral bolus dose). The pregnant animal should be observed daily for signs of toxicity. Maternal body weight should be monitored at least every 2 to 3 days during gestation. At sacrifice, the maternal animals should be examined for any abnormalities or pathological changes that may have influenced the pregnancy. The uterus is then removed and examined. The number of corpora lutea and live and dead fetuses should be recorded. The sex of the fetuses should be determined. Each fetus is then weighed and examined externally and malformations recorded by litter along with weight and sex. A certain percentage (depending on the animal species used) of the fetuses are then prepared for visceral examination and the remainder for examination of skeletal anomalies. Although the litter is considered the most relevant unit for statistical analysis, data should also be presented and assessed for each fetus.

Historical control data are also useful for determining the biological importance of visceral or skeletal anomalies that are elevated to a statistically significant level by treatment. Again, only historical control data from studies on the same species and strain of animal should be used for comparison purposes.

EPA's Proposed Changes

At least one developmental toxicity (formerly teratogenicity) study (Guideline 83-3; EPA, 1984) would now be required for all nonfood uses. In the past it was required only if there was expected exposure of women of childbearing age. A second study could be required if concerns are raised from the results of the first study. For food use EUPs accompanied by a temporary tolerance request, a second study could also be required, depending on the results of the first study.

A postnatal development toxicity study (Guideline 83-6; EPA, 1984) is proposed as a conditional requirement. This study could be required to more fully assess the manifestations of developmental toxicity, especially potential deficits in function or developmental neurotoxicity.

The parameters that need to be studied in a postnatal study depend on the effects seen in the prenatal study. Guidelines are presently being developed by EPA.


General Description

Multigeneration reproduction studies are designed to provide information concerning the general effects of a test substance on overall reproductive capability. Such studies may also provide information about the effects of the test substance on neonatal morbidity and mortality and about the meaning of preliminary data for developmental toxicity. EPA requires that the study include a minimum of two generations and that one litter be produced each generation. Dosing of both parents should begin when they are 8 weeks old and continue for 8 weeks prior to mating. Dosing of parental males should continue at least until mating is completed. Dosing of parental females continues through a 3-week mating period and pregnancy and up to the time of weaning 3 weeks after delivery of the pups. Dosing of pups selected for mating to produce the second generation should begin at weaning and continue as discussed above. The dosing and breeding schedule is clarified in the timeline presented in Table 4-6.

Parental animals should be observed daily for signs of toxicity. This is especially important for females during pregnancy in order to detect signs of difficult or prolonged parturition. Weights of parental animals are recorded weekly. The duration of pregnancy should be determined from the time evidence of mating was first observed. Each litter should be examined for the number of dead and live pups and for gross abnormalities. Live pups should be individually weighed on days 0 (optional), 4, 7 (optional), 14, and 21 after birth. A complete gross necropsy should be performed on all parental animals, all pups found dead prior to day 21 (weaning), and all weanlings not selected as parental animals for a next generation. Pups culled on day 4 do not have to undergo gross necropsy. Histopathology is required for reproductive and target organs (those known from previous studies to be adversely affected by the test material) for all control and high-dose parental animals and should be conducted on weanling animals (except for those selected as parental animals in the next generation) as described for parental animals (EPA, 1988).

EPA's Proposed Changes

The addition of a fertility assessment of parental males is recommended by EPA if fertility or reproductive parameters are found to be affected by the test chemical. The parameters to be examined or reported in this assessment include weight of reproductive organs, spermatid count, total cauda epididymal sperm count, assessment of sperm morphology and motility, examination of epididymal fluid for debris and unexpected cell types, and additional histopathology of the testes. A reproduction study (Guideline 83-4; EPA, 1984) could also be required to support nonfood uses if adverse effects on the reproductive system or developmental toxicity are observed in other studies.


TABLE 4-6 Approximate Dosing and Breeding Schedule for a Rat Two-Generation Reproduction Study

SOURCE: EPA, 1984.


General Description

A battery of mutagenicity tests is required to assess the potential of each test chemical to affect genetic material. The test selection criteria focus on the test's ability to detect, with appropriate assay methods, the capacity of the chemical to alter genetic material in cells. When mutagenic potential is demonstrated, these findings are considered in the assessment of potential heritable effects in humans, in the weight-of-the-evidence evaluation for carcinogenicity, and in the decision to require submission of a carcinogenicity study if otherwise conditionally required. Mutagenicity results per se are not used by themselves for risk assessment purposes, even when results suggest possible heritable genetic effects in humans.

EPA's Proposed Changes

EPA has already published changes to the 40 CFR Part 158 data requirements for mutagenicity (EPA, 1984).

As described in Pesticide Assessment Guidelines: Subdivision F (EPA, 1984), the original mutagenicity test battery consisted of three assays: one for gene mutations, one for structural chromosome aberrations, and one for other genotoxic effects. Other testing included DNA damage and repair. The revised guidelines would require an initial battery of tests consisting of:

• Salmonella typhimurium reverse mutation assay;

• mammalian cells in culture forward gene mutation assay allowing detection of point mutations, large deletions, and chromosome rearrangements; and

• in vivo cytogenetics.

Results derived from these assays could trigger the requirement for further mutagenicity testing. The type of additional required testing would depend on the observed results from the initial battery and other toxicity testing results. For example, testing could involve cytogenetic testing in spermatozoa if other test results suggest that they are targets.


General Description

Data from studies on the absorption, distribution, bioaccumulation, excretion, and metabolism of a pesticide may also allow more meaningful evaluation of test results and more appropriate risk assessment (as a result of more meaningful extrapolation from data on animals to humans). Such data may also aid in designing more relevant toxicology studies. Information on metabolites formed in laboratory animals is also used to determine whether further toxicity testing is required on plant metabolites. If a major metabolite forms in the plant but not in the test animal, separate toxicity testing on the plant metabolite could be necessary. The extent of testing required depends on the level of concern raised by the initial battery of toxicity tests (acute and subchronic studies, one teratology study, and a battery of mutagenicity tests).

As presently designed, the metabolism study consists of four separate parts: a single low, intravenous dose of radiolabeled test material (not required if the test material is insoluble in water or normal saline solution); a single low, oral dose of radiolabeled test material; 14 consecutive daily low, oral doses of unlabeled test material followed by a single low dose of radiolabeled material; and single high, oral dose of radiolabeled test material. Selection of the low dose is based on the NOEL. The high dose should elicit some signs of toxicity but not be so high that it results in mortality. The test species of choice is the rat.

Urine, feces, and expired air are collected for 7 days after administration of the radiolabeled material or until >90% of the radioactivity is recovered. Bone, brain, fat, testes, heart, kidney, liver, lung, blood, muscle, spleen, residual carcass, and tissues showing pathology in this or prior tests should be examined for radioactivity for all animals except those given the intravenous dose. This is done to determine if the test material or radiolabeled metabolite accumulates in any particular organ and to relate this information to the findings in toxicity studies.

In addition, quantities of radiolabeled material in feces, urine, and expired air must be monitored for all dose groups at appropriate intervals up to 7 days after dosing. Furthermore, urinary and fecal metabolites must be identified.

EPA's Proposed Changes

A metabolism study would also be required when significant adverse effects are observed in toxicology studies, including reproduction and developmental studies (Guideline 85-1; EPA, 1984). EPA is currently rewriting to guidelines for conducting metabolism studies and is including a tiered approach for study design and conduct.


General Description

Neurotoxicity studies are required to evaluate the potential of each pesticide to adversely affect the structure or function of the nervous system. The objectives of these studies are to detect and characterize the following:

• effects on the incidence and severity of clinical signs, the alteration of motor activity, and histopathology in the nervous system following acute, subchronic, and chronic exposures;

• the potential of cholinesterase inhibiting pesticides and related substances to cause a specific organophosphate-pesticide-type induced delayed neurotoxicity;

• other neurotoxic effects based on screening studies on certain chemical classes; and

• effects on organisms exposed prior to birth or weaning.

Results from these studies may be used for qualitative and quantitative risk assessment. The guidelines for these studies were published in March 1991 as addendum 10 to the EPA guidelines (EPA, 1991a).

EPA's Proposed Changes

The changes in the requirements for neurotoxicity testing were described above under ''Acute Toxicity" and "Subchronic Toxicity."


EPA intends to develop better definitions of the conditions under which domestic animal safety (Guideline 85-2; EPA, 1984) testing and visual system studies (Guideline 85-4; EPA, 1984) would be required for all organophosphates and other pesticides shown to affect the visual system. These studies could be of acute, subchronic, or chronic duration, whichever is deemed appropriate for the pesticide under study. Since guidelines have not been formulated for these studies, they will be designed in conjunction with EPA scientists.



Current and past studies conducted by registrants are designed primarily to assess pesticide toxicity in sexually mature animals. The protocols for these studies have evolved over several decades and have included some testing paradigms that allow extrapolation to infant and adolescent animals. These studies have produced some valuable information on toxicity and exposure. After reviewing EPA's current and proposed toxicity testing guidelines, however, the committee concluded that current studies do not directly address the following areas:

• toxicity of pesticides in neonates and adolescent animals;

• metabolism of pesticides in neonates and adolescent animals; and

• exposure during early developmental stages (after the second trimester through adolescence) and sequelae in later life.


• Studies should be redesigned and expanded in scope to elucidate the differences in the metabolism and disposition of pesticides in the infant, adolescent, and young adult.

Current metabolism studies are designed to provide information about sex-related differences, metabolic pathways and excretion, bioaccumulation in tissues, and tissue distribution in adult rats. EPA uses the data to determine whether toxicity testing needs to be conducted on individual plant or animal metabolites in addition to the parent compound.

• The metabolism of pesticides in newborn animals needs to be more thoroughly investigated.

Greater knowledge in this area would make it possible to develop computer programs for physiological pharmacokinetic modeling to forecast how information about metabolism in infant animals could be extrapolated to infant humans. The committee realizes that this is a very difficult area of investigation and application. Nevertheless, it urges that such investigations be pursued, since the resulting information could provide more realistic systemic exposure scenarios for risk assessment.

• A study should be conducted to compare the toxicity of several representative classes of pesticides in both adult and immature animals.

Results of such a broad-range study designed to specifically address the infant and young adult animal should indicate whether comparative studies of this nature should routinely be required by EPA. This study should be designed to examine several critical end points in the developing animal, including neural (functional and behavioral), immune, and endocrine systems to cite a few examples. Because the battery of acute toxicity tests now required by EPA is generally performed in adult animals, very little information is available on acute toxicity in immature animals. Such data are important in determining dietary risk to infants and children for acutely toxic pesticides such as organophosphates and carbamates. The committee recognizes that some of these data can be obtained from multigeneration studies if specific observation requirements are added to the current studies.

• Test animals should be exposed to the chemical of concern early in their lives so the risks of exposure of infants and children to the compound can be more adequately assessed.

The committee recognizes the difficulty in dosing animals during lactation and is aware that testing requirements would have to be modified to accomplish these studies. EPA Guideline 83-5 for a chronic toxicity/carcinogenicity study states that exposure of rats to pesticides should begin at approximately 6 to 8 weeks of age, essentially when they are adolescents (EPA, 1984). Because the effects of the pesticide in the rat are not determined early in its lifetime, chronic toxicity/carcinogenicity studies in adolescent animals may not be representative of the responses of younger animals. Current reproduction studies (Guideline 83-4; EPA, 1984) partially address this period in the life of a rat, but the effects of early exposure are not addressed past 21 days of age for second-generation pups or past the death of the second-generation parents (first-generation pups used for mating to produce the second generation). The protocol does not indicate whether exposure early in life has any impact on the adults or whether continuous exposure from birth to young adulthood influences the severity of the toxicity over a lifetime. FDA has used the multigeneration studies to include the F2A or F3A generation of laboratory animals for direct and indirect food additives (Becci et al., 1982).

• To obtain lifelong data on rodents for a given pesticide, the committee recommends that the testing guideline for a rat chronic toxicity/carcinogenicity study be modified to include in utero exposure during the last trimester, exposure through the mother's milk, and after weaning, oral exposure through diet.

This would mean that weanlings from the F1A or F2A generation would be selected from each dose group and tested throughout their lifetimes (see Table 4-6). In addition to this group, another smaller group of rats from the F1 generation would be killed at 6 months and 1 year and necropsied to examine the same parameters normally measured at the end of a lifetime feeding study.

The NTP tested three chemicals using a similar protocol and their standard protocol for an carcinogenicity study. One of the chemicals was ethylenethiourea, which is a breakdown product and metabolite of the ethylenebisdithiocarbamate fungicides and a thyroid toxicant (decreases T3 and T4). In utero exposure did not affect the occurrence of liver tumors in male and female mice, but did result in a sex-dependent increase in the number of malignant thyroid tumors in mice and rats (NTP, 1992).

• Measurement of the serum thyroid hormones T3 and T4 and serum TSH should be routinely added to the EPA chronic/carcinogenicity study protocol or to the subchronic toxicity protocol for the rat so that adverse effects on thyroid function can be determined earlier.

When examining the parameters currently measured in the EPA chronic/ carcinogenicity study, the committee found that endocrine function was adequately covered for all but the thyroid. Although the thyroid is saved in these studies for microscopic examination and its weight is recorded, the committee believes that changes in the functioning capabilities of this organ can occur regardless of whether there are organ weight or histopathologic changes.

• If abnormalities are found during histopathologic examination of the spleen, lymph nodes, thymus, and bone marrow, more detailed and specific studies should be conducted on a case-by-case basis relevant to the types of effects initially seen in immune system tests.

EPA has developed protocols for immunotoxicity testing for some pesticides that affect the immune system, and the agency is considering developing a generic testing protocol. The committee believes that because the human immune system is one of the most robust of systems in terms of resistance to pesticides or other chemical toxicity, initial evaluation using current histopathologic examination of spleen, lymph nodes, thymus, and bone marrow should be sufficient unless abnormalities are noted.

• A modified reproductive/developmental toxicity study in the rat is suggested for registration of all food-use pesticides.

One set of dams in this study would be dosed continuously with the test material from day 6 of gestation through birth of the pups and until weaning of their offspring at 21 days of age. A developmental assessment would be performed on the pups as described in EPA's recently published developmental neurotoxicity testing guidelines (EPA, 1991a). In addition, a set of pups from each dam would undergo gross and histopathologic examination at day 60 post partum. The second set of dams would be dosed from day 6 of gestation to term; however, these animals would not be allowed to deliver but, rather, would be subjected to cesarean section as in a routine teratology study. The fetuses would be subjected to skeletal and visceral examination, as described for a teratology study (Guideline 83-3) designed to examine the prenatal development of pups. This study design allows a determination of the reversibility of postnatal significance of findings seen in fetuses at the time of cesarean section. EPA has indicated in its proposed changes to Part 158 that a similar study be required; however, the committee recommends that this study be made a requirement for registration of all food-use pesticides.

• Because neurotoxicity is such an important consideration for the newborn, EPA should continue to revise its published guidelines on developmental and functional neurotoxicity testing as new information emerges from the actual conduct of preregistration studies and from ongoing research in rodent neurotoxicity.

The committee supports EPA's proposed requirement for acute and subchronic neurotoxicity testing for pesticides and encourages the agency to make this a general requirement for all food-use pesticides—not just for organophosphate and carbamate pesticides. New approaches to neurotoxicity testing are described in the report Environmental Neurotoxicology (NRC, 1992).

• EPA should develop a general guideline for visual system toxicity testing that can be modified and applied on a case-by-case basis.

The eye is exquisitely sensitive to changes in glucose metabolism, blood flow, and neuronal function, and several pesticides have been shown to be visual system toxicants (hexachlorophene, naphthalene, 2,4-DNP, and some organophosphates). In the past, scientists have examined the effects of chemicals that may irritate the eye by accidental contact. More recently, however, researchers have been examining the effects of chemicals on specific sections of the visual system, such as the optic nerve, iris, retina, and lens. The guideline proposed by the committee should be applied to species in which this type of testing appears to be appropriate, e.g., EPA has recently considered protocols for visual system testing in dogs.

Recent studies indicate that visual system damage may be associated with dietary exposure to some cholinesterase inhibiting compounds. Thus the committee supports EPA's proposed testing (the sensory evoked potential test) of such pesticides for visual system toxicity. However, it does not believe that a single protocol would suffice to cover all classes of compounds because different classes would affect different parts of the visual system.


Armitage, P., and R. Doll. 1954. The age distribution of cancer and multi-stage theory of carcinogenesis. Br. J. Cancer 8:1–12.

Becci, P.J., K.A. Voss, F.G. Hess, M.A. Gallo, R.A. Parent, K.R. Stevens, and J.M. Taylor. 1982. Long-term carcinogenicity and toxicity study of zearalenone in the rat. J. Appl. Toxicol. 2(5):247–254.

EPA (U.S. Environmental Protection Agency). 1984. Pesticide Assessment Guidelines, Subdivision F: Hazard Evaluation—Human and Domestic Animals. Revised Ed. November 1984. PB-86-108958. Washington, D.C.: U.S. Environmental Protection Agency.

EPA (U.S. Environmental Protection Agency). 1988. FIFRA Accelerated Reregistration Phase 3 Technical Guidance. US EPA 540/09–0784. Washington, D.C.: U.S. Environmental Protection Agency.

EPA (U.S. Environmental Protection Agency). 1991a. Pesticide Assessment Guidelines, Subdivision F: Hazard Evaluation—Human and Domestic Animals, Carcinogenicity of Ethylene Thiourea [CAS No. 96-45-7] in F/344 Rats and B6C3F1 mice. Addendum 10—Neurotoxicity. Washington, D.C.: U.S. Environmental Protection Agency.

EPA (U.S. Environmental Protection Agency). 1991b. Pesticide Assessment Guidelines, Subdivision F: Hazard Evaluation—Human and Domestic Animals Series 84. Addendum 9—Mutagenicity. PB-158394. 540/09-91-122. Washington, D.C.: U.S. Environmental Protection Agency.

Gad, S.C., and C.S. Weil. 1982. Statistic for toxicologists. Pp. 273–320 in Principles and Methods of Toxicology, A.W. Hayes, ed. New York: Raven Press.

IPCS (International Program on Chemical Safety). 1990. In Environmental Health Criteria 104: Principles for the Toxicological Assessment of Pesticide Residues in Food. Geneva, Switzerland: World Health Organization.

Loomis, T.A. 1978 Essentials of Toxicology. Philadelphia, Pa.: Lea & Febiger.

NRC (National Research Council). 1992 Environmental Neurotoxicology. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1993. Issues in Risk Assessment. Washington, D.C.: National Academy Press.

NTP (National Toxicology Program). 1992. NTP Technical Report on the Perinatal Toxicology and Carcinogenesis Studies of Ethylene Thiourea (CAS No. 96-45-7) in F3441N Rats and B6C3F1 Mice (Feed Studies). NTP TR 388. Research Triangle Park, N.C.: National Toxicology Program.

Weil, C.S. 1972. Guidelines for experiments to predict the degree of safety of a material for man. Toxicol. Appl. Pharmacol. 21:194–199.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 4:23 am

Part 1 of 2

5. Food and Water Consumption

DIETARY EXPOSURE OF CHILDREN to pesticides can be estimated by combining data on levels of pesticide residues in foods with data on food consumption patterns for infants and children. Risks to health can be assessed by combining estimates of dietary exposure with information on the toxic potential of pesticides. Various data bases are available for use in these calculations. In this chapter, the committee reviews the dietary surveys conducted to assess patterns of food consumption by the U.S. population, including infants and children. It examines the methods used in these surveys and evaluates their relative strengths and weaknesses. The committee then describes the age-related differences in food consumption patterns demonstrated by survey data and discusses food consumption estimates. The committee also notes various factors and limitations that must be considered in determining the basis for estimating exposure to pesticides residues in food and assessing the risk to infants and children.


Small-scale studies on food intake and nutrition were first conducted toward the end of the nineteenth century, when processing techniques were leading to rapid changes in the food supply. The variety of foods available to consumers again increased when in-home refrigerators and freezers became generally available, more sophisticated preservation techniques were introduced, and manufactured foods found their way into the retail market. By 1960 approximately 60% of the food items on supermarket shelves had come into existence during the preceding 15 years, that is, since the end of World War II (Hampe and Wittenberg, 1964).

It therefore became of great interest to researchers to determine what the people of this country were actually eating. In 1909 the U.S. Department of Agriculture (USDA) began to identify changes in the foods available to the civilian public by determining the disappearance of foods into the wholesale and retail markets. This is still done annually by subtracting data on exports, year-end inventories, nonfood uses, and military procurement from data on total production, imports, and beginning-of-the-year inventories. Overestimates result from this method, however, because losses that occur during processing, marketing, and home use are not taken into account. Thus, the resultant information is sometimes called availability or use of foods or nutrients (Stamler, 1979).

The USDA estimates national per capita use of foods or food groups by dividing the total available food by the total U.S. population. These data provide information on overall trends in available foods, but they do not indicate how use varies among population subgroups or individuals.

Since 1935 the USDA's Human Nutrition Information Service (HNIS) has conducted a series of Nationwide Food Consumption Surveys (NFCS). The first four (1935, 1942, 1948 [urban only], and 1955) surveyed household food use over a 7-day period. No record was made of waste or difference in use among household members. Surveys conducted in 1965–1966, 1977–1978, and 1987–1988 included information on the kinds and amounts of foods eaten by individuals in the household in addition to household food use. For the reasons given later in this chapter, the 1977–1978 survey served as the major source of consumption data used by the committee in the present study.

The USDA has conducted a planned series of surveys since 1985 solely concerned with individual food intake (USDA, 1985, 1986a,b, 1987a,b,c, 1988). The results of the most recent surveys have not yet been published. In these surveys, called Continuing Surveys of Food intakes of Individuals (CSFII), data are collected on three separates samples: women 19 to 50 years old and their children 1 to 5 years old (the core group); a sample of low-income women and their children; and in 1985 only, men 19 to 50 years old. In the 1989, 1990, and 1991 surveys, data were collected on all individuals.

The National Center for Health Statistics (NCHS), a division of the Department of Health and Human Services, has conducted the National Health and Nutrition Examination Surveys (NHANES) since 1971. The purpose of these surveys is to monitor the overall nutritional status of the population of the United States through comprehensive health and medical histories, dietary interviews, physical examinations, and laboratory measurements. The committee opted not to use the results from these surveys, however, because the number of observations within the age and demographic categories of interest were inadequate for the purposes of the present study.

The only national study of average intakes of pesticides, toxic substances, radionuclides, and industrial chemicals is the Total Diet Study conducted by the Food and Drug Administration (FDA). Four times a year, foods considered to be representative of the average U.S. diet are purchased from grocery stores across the United States and individually analyzed in FDA laboratories for the constituents mentioned above. Until April 1982 the food items used in the Total Diet Study were based on data from the 1965–1966 NFCS. Since then, they have been based on data from the 1977–1978 NFCS and second NHANES, which was conducted during 1976–1980. (The Total Diet Study is discussed further in Chapter 6, which focuses on pesticide residues in food.)

Infants, defined as less than 1 year of age, represent a separate and extremely critical population group with respect to purposes of this study. The number and classification of foods consumed by infants are almost exclusively processed and manufactured by a limited number of companies. Dietary intake studies have been conducted by Gerber Products Company and Ross Laboratories to evaluate intake for these populations, and those dietary studies were used by the committee.


Three basic types of methods are used to gather data in food consumption surveys (Burk and Pao, 1976; Dwyer, 1988): retrospective, prospective, and a combination of retrospective and prospective. Retrospective methods include the 24-hour (or 1-day) food recall and the food frequency questionnaire. Prospective methods require the used of food records or diaries—the weighed food record and the estimated food record (sometimes called the household measure food record). In combination surveys, investigators use both prospective and retrospective methods, e.g., the recall and weighed food record. (For a comprehensive overview, see Dwyer, 1988.)

Retrospective Methods

The 24-Hour (or 1-Day) Recall Method

In surveys conducted with this method, interviewers ask subjects to recall the quantities of particular foods and beverages consumed on the previous day or during the preceding 24 hours. More precise estimates of portions are obtained when the respondents are provided with measurement guides such as food models, abstract shapes, and tableware (DHHS, 1983); measuring cups, spoons, and rulers (USDA, 1987); and pictures (Frank et al., 1977; Posner et al., 1982).

Madden et al. (1976) found no significant differences between mean intake data obtained from the 24-hour recall and actual observed intake over that period. They noted, however, that the data were not representative of an individual intake averaged over different days. Recalls of food consumption over a period longer than 24 hours produce more representative data on general intake patterns, but are less precise than the 24-hour recall because they depend to a greater extent on memory (Block et al., 1986).

In general, the validity of the recall method and the extent of bias depend on a variety of factors related to collection methods (Backstrom and Hursh-Cesar, 1981). In face-to-face interviews it is easier to elicit a response. Such interviews are more efficient when there are several family members or a long questionnaire; however, they are expensive primarily because of travel costs. In contrast, telephone interviews are relatively inexpensive, quick, easy to monitor, unlimited geographically, and can ensure a greater degree of anonymity ( Wilson and Rothschild, 1987). Telephones may be used in combination with, or as a follow up to, other interview techniques. They do have several disadvantages, however: subjects can easily terminate an interview prematurely, long or complicated questionnaires are difficult to administer, and many low-income families lack telephones. In all cases, selection and training of the interviewers are critical to the success of the survey.

The single, short (15- to 30-minute) guided interview produces higher response rates than all other methods. Moreover, accuracy is enhanced in several ways: the recall period is short and precise, measurement aids may be used to increase the accuracy of consumption estimates, and the quantities reported can be easily converted to nutrient equivalents.

Mail interviews are inexpensive, and they may be used in combination with other interview methods (Posner et al., 1982). Furthermore, interviewer bias is avoided. Negative aspects include low response rates, exclusion of illiterate persons, and lack of control over who responds and how they respond.

Food Frequency Questionnaires

In its simplest form, the food frequency questionnaire consists of a checklist of foods or food groups and a set of categories indicating daily, weekly, or monthly frequency of food consumption during a specified period—weeks, months, or a year. The checklists may contain as few as 20 items or more than 100. The questionnaire can be administered in person, over the telephone, or by mail, with the attendant advantages and disadvantages described above (Willett et al., 1985). Subjects may use a visual aid to estimate portion sizes (Chu et al.,1984). When the quantities have been estimated, the nutrient content of the foods consumed may in turn be estimated.

The food frequency questionnaire is a quick, inexpensive, and simple method for obtaining information on food intake from large numbers of subjects (Sampson, 1985), and its administration does not require highly trained personnel. Epidemiologist have found the questionnaire useful in studying the relationship between diet and disease risk. Attempts have been made to validate the accuracy of consumption frequency estimates by comparing them to food records (Chu et al., 1984; Willett et al., 1985; Freudenheim et al., 1987).

The limitations of this method outweigh its strengths. Its accuracy may be compromised by the long recall period, estimates of past intake may be biased by current intake, and respondent burden is heavy if the checklist is long and complex. Moreover, the questionnaire may not be appropriate for people who consume unusual diets or for children. In the future, a combination of methods may be used to overcome these limitations.

Prospective Methods

Food Records or Diaries

These usually self-administered reports of current food intake can cover periods ranging from 1 day to as long as 1 year (Basiotis et al., 1987). Subjects or their surrogates, e.g., parents, record the portions of all foods and beverages ingested immediately after each eating occasion.

The foods are either weighed (the weighed food record) (Acheson et al., 1980; Anderson and Blendis, 1981; Marr and Heady, 1986) or measured with a cup, ruler, or other aid (the estimated food or household measure record) (McGee et al., 1982; Acosta et al., 1983; Elahi et al., 1983). Foods eaten away from home must be estimated. The latter method is used most often, but the weighed food record is widely regarded as the most accurate procedure and is often used to validate other methods. Both methods require trained personnel to demonstrate proper weighing, or measuring, and recording techniques. Written instructions are also provided (Sempos et al., 1984).

Reliance on memory is minimal, the recall period is precise, interviewer bias is avoided, omissions of foods and beverages consumed tend to be minimal, and the accuracy of portion estimates is enhanced by the weighing and measuring techniques. However, respondents must be literate and willing to accept the heavy burden of participation—factors that can bias the sample. In long-term surveys, boredom or fatigue may lead to a decline in accurate reporting.

Combined Retrospective and Prospective Methods

A combination of recall and records methods is sometimes used to obtain multiple-day intake data for individuals in a large survey or study (Schnakenberg et al., 1981; USDA, 1983). Because the limitations of one method offset the limitations of the other, a greater accuracy of mean intake estimates can be expected. Examples of combinations include 1-day recall and 2-day record (Patterson, 1971; USDA, 1983), 3-day recall and 4-day record (Futrell et al., 1971), and 1-day recall and 14- to 17-day record (Schnakenberg et al., 1981). Diet histories may also involve a combination of methods (Dwyer, 1988).

Methods Used in USDA Surveys

Although the USDA has sought similar information in its various dietary intake surveys, it has changed the methods used in an attempt to obtain better measures of average intake. The agency has advanced from a 1-day recall only insuring (1965–1966 Household Food Consumption Survey) to a 3-day combination (1-day recall and 2-day record) during all four seasons in the 1977–1978 and 1987–1988 NFCS. The 3 days of intake data for more than 30,000 people, obtained by using this combination method, produced a better measure of an individual's average intake than did the 1-day measure used in 1965.

In the 1985 and 1986 CSFII, 6 nonconsecutive days of intake data were collected by trained interviewers who administered the 1-day recall method every 2 months over the course of the year. The first interview was conducted in person; the remaining five were accomplished over the telephone. Mothers provided the recall information for their children. This system proved costly, and the drop-out rates over the course of the year were high (Table 5-1). The substantial decrease in participation (approximately 50%) can be attributed to either the 145 respondents (10% of the sample) who moved to another geographical region or to such socioeconomic characteristics as being younger, having a low income, being in poor health, being on a special diet, having one or more children, being a suburban dweller, or working.

As a result of the drawbacks in the 1985 and 1986 CSFII, the combination of 1-day recall and 2-day record used in the 1977–1978 NFCS was reinstituted in the 1987–1988 NFCS and the 1989 and subsequent CSFII to obtain 3 consecutive days of dietary intake data. In both the NFCS and later CSFII, dietary information was collected on all individuals—not just the sex-age groups surveyed in the 1985 and 1986 CSFII.


To determine the validity of a survey sample, it is necessary to consider a variety of factors such as the survey design, sample weighting, and comparison of resulting data to standards. The committee began by examining the design of the CSFII and the NFCS. The 1985 and 1986 CSFII and the 1977–1978 and 1987–1988 NFCS were designed to provide a multistage, stratified, probability sample that was representative of the 48 conterminous states.


TABLE 5-1 Unweighted Counts of Individuals for the 1985 and 1986 CSFII

a The numbers in these columns are not additive since the represent different study groups.

b Men were sampled for only 1 day.

c More responses than expected were received for day 1, and they included a large number of low-income households. To reduce the number of low-income households to the targeted number of 1,200 for interviewing on days 2-5, systematic subsamples were drawn for both women and children. The numbers in parentheses refer to those subsamples.


A multistage sample is drawn by selecting random groups in stages. At each stage, groups of individuals are selected from increasingly smaller segments of the population. The term stratified indicates that the population is divided into mutually exclusive subsets, or strata, before the sample is drawn. Taken together, these subsets represent the total population that is being examined. However, the sampling plan is applied separately within each stratum. Because the strata are defined by geographic location, the sampling within each stratum is called an area sample.

In a probability sample, every group has a known probability of selection. Thus, every element in the population must be enumerated before the sample is drawn to facilitate determination of the likelihood of selecting any group of individuals into the sample. These probabilities may or may not be equal for all groups. The use of uniform criteria for each group helps minimize the extent of enumeration required to determine selection probabilities. Each group sampled, therefore, has a known probability of selection. Although this method is fairly complex, it provides data that are statistically projectable, with known sampling error, to the entire conterminous United States. Nonprobability samples, where groups are not enumerated, are generally easier to obtain but do not provide data that are statistically projectable to the general population.


TABLE 5-2 Distribution of Strata within the Conterminous United States as Defined by the Bureau of Census Geographic Divisions

SOURCE: Adapted from the U.S. Department of Agriculture, 1985.

The 48 states were grouped into the nine census geographic divisions, which in turn were divided into three classifications: central city, suburban, and nonmetropolitan (Table 5-2). From these 27 superstrata, 60 strata (17 central city, 28 suburban, and 15 metropolitan) were obtained. Both the CSFII and the NFCS surveys consisted of four stages. During the first stage, the probability proportion to size (PPS) technique was used to select two primary sampling units (PSUs) from each of the 60 strata in both the CSFII and NFCS. These 120 PSUs included counties, cities, or parts of cities, and were relatively homogenous with regard to demographic characteristic.

In designing food surveys, care must be taken to ensure that the sample of consumers is representative of the general population. Even a scientifically designed probability sample may not be representative due to nonresponse and other practical problems. This could lead to bias in the estimates of mean intake, especially if there is a systematic component to the nonresponse. An accepted method for adjusting the estimates of intake would be to weight the data in such a way that they more closely reflect the general population. Fuller (1991) discusses using regression estimation for adjusting intake when appropriate auxiliary information exists. Fuller et al. (1991) adjusted intake for sociodemographic factors for the 1987–1988 NFCS and found a significant difference between the resulting mean intake as compared with mean intake calculated without such weighting.

Nusser et al. (1991) considered characterizing usual daily intake distributions as opposed to mean daily intake. Usual intake is useful in providing information about nutritional deficiencies occurring over a long period. They show that use of the distribution of mean intake as an estimate of the distribution of usual intake can lead to erroneous inferences regarding nutritional status. In particular, the variance of mean intakes contains intra individual variability and is thus greater than the variance of usual intakes. Other parameters of the two distributions may differ as well.


Although the CSFII and NFCS samples were designed to be self-weighting, adjustments to the samples were required because not all eligible households participated, not all eligible individuals in eligible households were interviewed, and not all interviews yielded complete dietary information. Weighting factors were developed for each individual participating in the survey and were applied to data from completed intake records to adjust for these sources of nonresponse. Other weighting considerations included economic homogeneity, geographic heterogeneity, and age.


Sample size is an important determinant of sample variation and statistical precision. Decisions regarding sample sizes therefore depend on the level of precision desired for the data needed to estimate a population parameter of interest.

The size of the sample required an established goal of precision for an estimate can be determined through calculations that depend on the coefficient of variation (CV) and an assumed probability distribution for the data. The coefficient is the standard error of an estimate expressed as a fraction of the sample mean. As the CV of an estimate increases, its accuracy decreases. In general, NFCS and CSFII intake estimates with CVs exceeding 50% are not published. Estimates with CVs of 15% to 50% are published by USDA with caveats regarding their accuracy.


Intake data are often compared with a widely accepted value such as the Recommended Dietary Allowances (RDAs). The RDAs are intended to meet ''the known nutrient needs of practically all healthy persons" in the United States (NRC, 1989); however, since nutrient needs vary among people, margins of safety are built into the RDAs for many nutrients. For this reason, analysts have selected a fixed cutoff point, such as two-thirds or three-fourths of the RDA, as a point of comparison. One committee convened by the Food and Nutrition Board (FNB) of the National Research Council suggested that distributions of requirements or tolerances would be more appropriate for this purpose (NRC, 1986). This FNB committee recommended that a probability approach be used to estimate the prevalence of inadequate intake, i.e., the probability that a specific intake is inadequate to meet an individual's requirement (NRC, 1986).


Food consumption data must be validated before they are used to calculate risk, in part because the integrity of formulated and combined foods is compromised when those foods are broken down into their constituents. The method of calculation used by the EPA to estimate risk converts consumption data directly into food components, thereby eliminating the need to follow all the separate validation steps. (See "Quantification of Consumption Data," below, for a further discussion of the EPA method.)

Substantial uncertainty is inherent in food consumption data because of a variety of factors. Principal among these are recording errors and biases. It is therefore difficult to extrapolate results to the general U.S. population.

Of the different protocols that have been used in dietary surveys, none is uniformly better than all others. Even the most extensive dietary sampling schemes may be subject to biases or practical limitations. Because of this, a variety of survey methods have evolved to address different objectives, each with its own strengths and weaknesses.

Two primary objectives of food consumption surveys are to assess the mean intake of a group of individuals or the mean intake of a particular individual. Because food consumption varies markedly both within individuals and between individuals (Beaton et al., 1979; Beaton et al., 1983; Todd et al., 1983), methods that are appropriate for one objective may not be appropriate for the other.

The validity of the results relative to the study objectives is of central importance (Block, 1982). Validation of data on food consumption is a difficult task. Ideally, all types and quantities of food consumed by the survey respondents would be recorded in a complete and accurate fashion; this record of actual consumption could then be used as a reference value against which to compare estimates of consumption based on different survey protocols. Kim et al. (1984) studied a method whereby individuals are asked to set aside a duplicate portion of the food they ate for future analysis. The investigators found that the process of setting aside the extra food actually led to smaller quantities consumed than in the usual pattern of intake. To avoid biases associated with self-reported food consumption data, an inconspicuous observer could maintain a diary of foods consumed—a technique used by Madden et al. (1976), Gersovitz et al. (1978), and Krantzler et al. (1982). However, these studies were carried out in controlled environments such as dormitories, dining halls, or other congregate meal sites, thus making it difficult to extrapolate conclusions to the general population.

Because of the difficulties in measuring actual intake, validation to a large extent tended to focus on the comparison of results obtained from different survey methods (Block, 1982). For example, relatively new survey methods such as questionnaires have been compared with established methods such as multiple-day records (Jain et al., 1980; Axelson and Csernus, 1983; Willett et al., 1985; Byers et al., 1987; Krall and Dwyer, 1987; Pietinen et al., 1988). Methods are also sometimes compared to determine which one will produce the most reliable estimates of intake of specific nutrients such as vitamin A (Young et al., 1952; Russell-Briefel et al., 1985; Sorenson et al., 1985).


The 1977-1978 NFCS data have the following major limitations.

• Over the years since that survey was conducted, average overall dietary patterns may have changed in response to advanced food technology, advertising, taste, and health consciousness, among other variables.

• The sample of nursing infants was small (n = 106).

• The 3-day survey period reflects too brief a consumption period, even though it was conducted over all four seasons.

• Water consumption was not considered.

Despite these substantive limitations, the NFCS has provided the only comprehensive data currently available for comparisons of food consumption by all age classes in our population, which is the primary reason the committee chose the 1977–1978 NFCS as the basis for this report. Because the EPA also relies on the same food intake data, the committee's findings can easily be compared with the risks estimated by that agency. The committee compensated for the limited sample size for nursing infants by reviewing survey data reported by Purvis and Bartholmey (1988). Appraisal of these results validated the similarity of results, even though methodology and sample selection were different.

The 1987–1988 NFCS data were not used by the committee because of the low response rate (34%) and resulting procedural problems. Furthermore, the committee found that the sampling units were so small that in some cases they were not representative of certain age or geographical sectors. At a later date, the U.S. Government Accounting Office (GAO, 1991) conducted an independent evaluation of the 1987–1988 NFCS data and came to similar conclusions. Thus, the data obtained in this survey were not adequate for use by the committee in fulfilling its mandate.

The CSFII provides a more recent view of consumption patterns; however, these surveys experienced high drop-out rates, as described above, and are limited to children between the ages of 1 and 5 years (USDA, 1985, 1986a). They provide no data on infants less than 1 year old or on children between the ages of 6 and 18 years.

To gain a clearer view of the consumption patterns of infants, the committee examined the results of an Infant Nutrition Survey conducted by the Gerber Products Company in 1986. This was one in a series of surveys conducted by Gerber since 1969 to monitor infant feeding practices, nutrient intake, and nutrition contribution and changes (Purvis, 1973; Johnson et al., 1981; Purvis and Bartholmey, 1988).

With the exception of minor refinements, the same method for collection and compilation of data has been used in each of the Gerber surveys to facilitate comparisons over time (Purvis, 1973; Johnson et al., 1981). Initial contact was made through mail questionnaire. The consumption data were collected in a 4-day diary maintained by a parent who had been given instructions for recording intake information and obtaining additional assistance. After the diaries had been completed and returned, interviewers used the telephone to clarify information when needed. The data were analyzed to determine usual nutrient intake, portion size, food preferences, and age at which supplemental foods were introduced.

The 1986 Gerber sample initially consisted of 1,000 infants between the ages of 2 and 12 months. Balance of age and geographic distribution was achieved to the extent possible through random sample generation (Table 5-3). There were 637 satisfactorily completed diaries—a return rate of 64% (G. Purvis, Gerber Products, personal commun., November 14, 1990).

The results of the Gerber survey were found to reflect the same consumption patterns for infants and children as did the USDA surveys (Figure 5-1). Furthermore, a comparison of data from the CSFII and the NFCS showed that some of the foods that dominate the diets of children have changed little in the decade between the 1977–1978 NFCS and the 1985 and 1986 CSFII. With this additional confidence in the larger NFCS data base, the committee decided to use the results of the 1977–1978 NFCS as the basis for its consumption estimates. The 1985 and 1986 CSFIIs were used to account more fully for intraindividual variation and differences among ages for children less than 5 years of age.


TABLE 5-3 Sample Selected in the 1986 Gerber Infant Nutrition Survey

SOURCE: Based on data from Gerber Products Company, personal communication, 1992.


FIGURE 5-1 A comparison of infant intake data (on raw agricultural commodities) from Gerber Products, 1988, unpublished, and from USDA, 1983.


The National Cancer Institute (NCI) conducted an extensive analysis of the 1977–1978 NFCS data to develop more accurate estimates of water intake (Ershow and Cantor, 1989). Three types of water were considered:

• water intrinsic to food,

• tapwater added to food during preparation in the home, and

• tapwater consumed by itself.

Since total food moisture was not part of the NFCS data base, the NCI investigators calculated the intake of water from this source by applying the following formula:


Data on tapwater intake were collected as part of the NFCS survey. When the amount of water added to foods was not clear in the USDA Food Code Description Files, the NCI investigators used standard dilutions (e.g., for canned soups, frozen juices) or consulted cookbooks. The investigators acknowledged that their estimates of total water intake by nursing infants were low. In the absence of data on human milk consumption, they derived their estimates from other water sources. Tapwater intake estimates may also be low because ready-to-feed infant formula was assumed when the survey results did not specify the type of formula consumed. In fact, some of the infant formula not clearly identified may have been powder or concentrate to which tapwater had been added.

Dietary Sources of Water

The dietary sources of total water and tapwater reported by NCI are shown in Figure 5-2 (Ershow and Cantor, 1989). Relatively few beverage and food items contributed to total water and tapwater intake for most age groups. For infants less than 1 year old, formula provided 32.7% of total water intake, milk and milk drinks contributed 24.7%, and drinking water, 16.1%. For the 1- to 10-year age group, infant formula was no longer a factor, drinking water increased to 30.3%, and milk and milk products remained at levels similar to those for infant intake (25%). Of the tapwater intake by infants less than 1 year old, 69% was provided by drinking water and 11.9% by formula. For the 1- to 10-year age group, 64.8% was provided by drinking water, and 13.6% by fruit juices, tomato juice, and noncarbonated drinks—up from 4.5% for the infants.

Water Intake Estimates

As shown in Table 5-4, mean total water intake by infants during the first 6 months of life is, 1,014 ± 294g/day. Intake increases to 1,258 ± 322g/day before their first birthday. These estimates correspond to 189 ± 73.5 and 141.7 ± 43.0g/kg body weight (bw)/day, respectively. Mean tap water intakes for these age groups are 272 ± 247g/day (52.4 ± 53.2g/kg bw/day) and 328 ± 265g/day (36.2 ± 29.2g/kw bw/day).

There is a steady increase in both total water and tap water intake into adulthood and a gradual decrease after the age of 65 years. When viewed on a grams-per-kilogram-of-body-weight basis, however, the highest intakes are found for infants during the first 6 months of life (Figures 5-3 and 5-4). Daily total water intake decreases from 189 ± 73.5 g/kg bw during the first 6 months to 41.9 ± 15.6 g/kw bw for the 11- to 14-year age group (Table 5-4).

Males generally had higher mean intakes than females (p = 0.05) for all age groups over 1 year of age adjusting for race, region, season, body weight, urban residence, and age. As shown in Figures 5-5 and 5-6, these differences were relatively small for young children, and consumption figures for female infants were actually higher than those for males of the same age.



FIGURE 5-2 Dietary sources of total water and tap water by age, expressed as percentages.

SOURCE: Ershow and Cantor, 1989, pp. 28–29.


TABLE 5-4 Mean Intake of Total Water and Tap Water by All Ages, Both Sexes, All Regions, and All Seasons

SOURCE: Based on data from Ershow and Cantor, 1989, pp. 42, 51, 65, and 74.


FIGURE 5-3 Mean daily intake of total water per unit of body weight by age group and sex.

SOURCE: Ershow and Cantor, 1989, p. 26.

Variations between regions and smaller seasonal differences were observed for all age groups (Figures 5-7 and 5-8). For infants, total water and tap water consumption were lowest in the northeast, tap water was highest in the west, and total water was highest in the midwest. Intake of both tap water and total water was highest in the summer in all regions for all age groups.

Water consumption data reported by Gerber showed a wide variation among the various age groups. In general, however, the data supported those reported by NCI and described above (Gerber Products, personal commun., 1992).


The 1977–1978 NFCS results were reported as daily consumption of individual foods. Gerber reported consumption data for foods as formulated and processed for feeding. Tolerance levels established by the EPA for pesticide residues in foods, however, are based on crop-based components


FIGURE 5-4 Mean daily intake of tap water per unit of body weight by age group and sex.

SOURCE: Ershow and Cantor, 1989, p. 26.

of the foods reported in the 1977–1978 NFCS. Therefore, to assess pesticide exposures and to compare those exposures with established reference doses, it is necessary to break down the foods consumed into raw agricultural commodities (RACs)—the components used by the regulatory agencies. For example, pizza is broken down into wheat flour, water, yeast, tomato paste, tomato sauce, cheese, and other ingredients and expressed in grams in order to match the form in which pesticide tolerances are reported in the Code of Federal Regulations (1986a,b). (RACs are discussed in more detail in Chapter 6.)

EPA developed and uses the Dietary Residue Evaluation System (DRES) to estimate dietary exposures of humans to pesticides through the diet. This system is based on pesticide residue concentrations found in RACs, which are sampled from harvested agricultural crops and analyzed at the farm gate. The residue estimates are then multiplied by the food consumption estimates to assess the extent of human exposure and to allow the development of new tolerance concentrations. DRES includes recipe files for each food tested. Furthermore, DRES may be used to estimate consumption of various foods in 22 different population groups based on such characteristics as age, nursing or nonnursing status, and ethnic background from survey information available in the NFCS. The DRES method, formerly called the Tolerance Assessment System (TAS), and the use of RACs as derived by EPA from the NFCS to estimate the exposure of humans to pesticides have been described in detail in other publications (e.g., Research Triangle Institute, 1983; Saunders and Petersen, 1987).

Several problems are associated with this method of calculating food intake, however. The processing effects cannot be accurately considered because there is no compensation for the fractionation of food components (e.g., the stripping of soybean oil). Individual components of specialized foods cannot be identified when only the total of the components is presented. For example, the term milk solids applies to fresh milk, milk in formulated foods, and milk in infant formula. This practice fails to compensate for ingredient selection and processing differences for specialized infant foods.

With these caveats in mind, the committee used these data to develop tables showing the predominant foods in the average U.S. diet and in the diets of various subgroups: nursing and nonnursing infants (<1 year old), children from 1 to 6 and 7 to 12 years old, teenagers from 13 to 19 years old, and adults over the age of 19. The result is a clear picture of the most commonly consumed foods in the diets of infants and children and how these differ from those consumed most frequently by adults, as discussed in the next section.


FIGURE 5-5 Mean daily intake of total water by age group and sex.

SOURCE: Ershow and Cantor, 1989, p. 24.


FIGURE 5-6 Mean daily intake of tap water given in grams by age group and sex.

SOURCE: Ershow and Cantor, 1989, p. 24.

Because EPA presented intake data for milk as RACs (that is, as the dry constituents such as fat and nonfat milk solids and lactose), comparisons of intake to other foods consumed could not be reasonably made without making an adjustment for the water content of the milk. To accomplish this, the committee estimated the solid content of milk at 15% and its water content at 85% and adjusted the data for the dry constituents accordingly. In this way, percentages of the total diet for each of the age groups could be calculated to present a more representative approximation of their dietary patterns.

The 85% was derived from the DRES recipe files, which show that milk and infant formula are approximately 85% water, depending on the product. Other estimates are quite close. For example, Petersen and Associates (1992) reported the water content of milk-based baby formula as 87%, and of light cream, 74.1%. Clearly, the water component of milk products consumed by infants and children is an important consideration in developing dietary comparisons.


Table 5-5 lists the 17 foods (expressed as RACs) that comprise more than 1% of the average U.S. diet, as reported in the 1977–1978 NFCS. Individual age categories were compared to the U.S. average and multiples determined. Only two age categories had multiples in excess of 2: nonnursing infants and 1- to 6-year-old children. Table 5-6 shows percentages for the foods comprising more than 1% of the average diets of various age groups. Table 5-7 presents the multiples of the intake of those foods compared with the U.S. average (e.g., 2.00 means twice the U.S. average; 5.25 means 5-and-a-quarter times the U.S. average). The information in these tables is also presented as RACs.

The numbers in these tables were derived from EPA intake data based on milligrams per kilograms of body weight and are presented as RACs consistent with regulatory practice. Water intake was not considered, except for the water component of milk products, as noted above.


FIGURE 5-7 Mean daily total intake of water by source, age group, and region.

SOURCE: Ershow and Cantor, 1989, p. 27.


FIGURE 5-8 Mean daily total intake of water by source, age group, and season.

SOURCE: Ershow and Cantor, 1988 p. 27.

In the process of reviewing the data and developing the comparisons, the committee noted variances in dietary patterns across age categories, within age categories, within individuals, and over time. All these factors must be considered in estimating exposure to pesticide residues and assessing risk for infants and children. Clearly, the marked differences between children's diets and the diets of adults and their relationship to the U.S. average have implications for assessing patterns of dietary exposure to pesticides.

Milk in its entirety constitutes an extremely large portion of the U.S diet—not only for infants and children, but also for adults. This can be attributed to the relatively large quantities of milk contained in many of the foods consumed, e.g., formulated foods and baked goods, and its use as a diluent, e.g., for cereal.

Human milk is a dominant source of nutrition for nursing infants less than 3 months old and a substantial source for those less than 11 months. It is consumed exclusively by more than 60% of nursing infants until they reach 3 to 6 months of age. Data collected by Ross Laboratories in 1992 indicate that 53.3% of all mothers in the United States breastfed their newborn children but only 19.7% of them continued the practice when their infants were 5 to 6 months of age (Table 5-8). The highest percentages were noted for white mothers (59.2% for newborns; 22.6% at 5 to 6 months) and Hispanic mothers (51.8% for newborns; 16.1% at 5 to 6 months), in contrast to black mothers (25.8% for newborns; 7.3% at 5 to 6 months). There were also wide differences that correlated with socioeconomic status across all ethnic groups. For all mothers in families with annual incomes of $25,000 or more, breastfeeding rates were 67.1% for newborns and 27.4% for 5- to 6-month-old infants. At incomes less than $10,000, the rates were 33.3% and 8.5%, respectively. Data from the 1988 National Maternal and Infant Health Survey show similar trends: 52.4% of new mothers in the United States were breastfeeding their newborns and that percentage dropped to 11.8% at 6 months and to 1.5% at 1 year (Table 5-9). They also show that breastfeeding rates among low-income women are lower than the national average.


TABLE 5-5 The 17 Foods Comprising More than 1% of the Average U.S. Diet in 1977–1978 and the Age Class Consuming the Highest Multiple of That Average

a The EPA intake figures for fat and nonfat milk solids were divided by 0.15 to derive a percentage that would more accurately reflect the contribution to the total diet made by milk with an 85% water content. See text for further discussion.

SOURCE: Based on data from the 1977–1978 Nationwide Food Consumption Survey.


TABLE 5-6 Foods Comprising More than 1% of the Average Diet of Age Groups Indicated

NOTE: EPA intake data for fat and nonfat milk solids were divided by 0.15 to derive a percentage that would more accurately reflect the contribution to the total diet made by milk with an 85% water content. No percentages are given for foods comprising less than 1% of the diet.

SOURCE: Based on data from the 1977–1978 Nationwide Food Consumption Survey.


TABLE 5-7 Foods Comprising More Than 1% of the Average Diets of Different Age Groups and Multiple of U.S. Average Consumption

NOTE: No multiples are provided for foods comprising less than 1% of the diet of the age groups indicated. The multiples for fat and nonfat milk solids take into consideration their water content (see NOTE under Table 5-6).

SOURCE: Based on data from the 1977–1978 Nationwide Food Consumption Survey.


TABLE 5-8 Percentage of Mothers Breastfeeding Newborn Infants in the Hospitals and at 5 or 6 Months of Age in the United States in 1992, by Ethnic Background and Selected Demographic Variables

a Hispanic is not exclusive of white or black.

b College includes all women who reported completing at least 1 year of college.

SOURCE: From Fritz Krieger, Ross Laboratories, personal communication, 1992.


TABLE 5-9 Percentages of Breastfeeding Respondents in the 1988 National Maternal and Infant Health Survey

a ''In this table, low income is defined as total income less than or equal to 185% of the federal poverty line.

SOURCE: Unpublished data from the 1988 National Maternal and Infant Health Survey. Provided by J. Tognetti, Office of Analysis and Evaluation, Food and Nutrition Service, U.S. Department of Agriculture.

A number of studies demonstrate that the volume of milk intake among healthy, exclusively breastfeed infants also ranges widely (Figure 5-9).


FIGURE 5-9. Milk intakes during established lactation. The lines represent the smoothed mean ± standard deviation. The points represent average intakes obtained in 16 studies that included test weighing, validated exclusive breastfeeding, three or more subjects, and monthly reports on milk transfer.

SOURCE: Neville et al., 1988, with permission.

After the first 4 to 5 months, the variance is even greater. For infants who were breast-fed for at least 12 months and given solid foods beginning at 4 to 7 months, milk intake averaged 769 g/day (range, 335 to 1,144 g/day) at 6 months (n = 56), 637 g/day (range, 205 to 1,185 g/day) at 9 months (n = 46), and 445 g/day (range, 27 to 1,154 g/day) at 12 months (n = 40) (Dewey et al., 1990; K. Dewey, personal commun., 1992)

Milk intake is most often determined by weighing the infant before and after feeding. This method leads to underestimations of intake ranging from approximately 1 to 5% (Brown et al., 1982; Woolridge et al., 1985) because of water loss through evaporation from the infant between weighings. Newer techniques based on stable isotopes have been developed to measure breast milk intake (Coward et al., 1982; Butte et al., 1988), but few data have been generated by this method to date (e.g., Orr-Ewing, 1986; Butte et al., 1992). For a more detailed review of these data, see Nutrition During Lactation (NRC, 1991).

Because the volume of human milk consumption varies widely, the estimates of supplementary food consumption shown in Table 5-6 are conservative, i.e., they are higher than the amounts likely to be consumed by these infants. The 1988 National Maternal and Infant Health Survey contained a number of questions that promise to shed more light on infant food consumption during the first 6 months of life. Unfortunately, the analysis could not be made available to the committee in time for inclusion in this report.

As demonstrated by the preceding discussion and accompanying tables, use of available intake data to assess dietary exposures of nursing infants is complex. Human milk is the major food and source of essential nutrients consumed by these infants during their first year of life, but a vast range of variables must be considered: the age at which supplementary foods are introduced, the selection of foods given to them, and the volume of human milk consumed. Factors greatly affecting the feeding patterns of this group include economic status, ethnic background, region, and the age, marital status, educational level, parity, and employment of the mother.

Infant formula is the sole source of food for nonnursing infants for the first 3 months of life. Milk or milk-based food remains the predominant source of energy and nutrients for all infants throughout their first year of life. Averaged over the first 12 months, nonfat and fat milk solids provide 44.2% and 10.4%, respectively, of their diet. These figures were derived from DRES conversion of formula into its component parts: fat and nonfat milk solids, milk sugar, coconut oil, and soybean oil. Coconut oil represents only 1.4% of the average diet of nonnursing infants, but the consumption is almost 50 times greater than that of the national average (Table 5-6).

The diet of infants is gradually supplemented with specially prepared, predominantly processed foods produced by a small number of manufacturers. When presented as RACs, fruit and fruit juice constitute the highest proportion of the early supplemental foods, accounting for an estimated 16.1% of the diet, averaged over the first year of life. These are led by apple juice (4.4% as RAC) and fresh apples (3.2% as RAC).

Caloric consumption by infants per unit of body weight is higher than that for adults—approximately 2.5 times higher for the very young infant (NRC, 1989). Therefore, comparison of the consumption data for infants and adults on the basis of grams per kilogram of body weight results in an elevated value for infants.

The diets of infants and children are less diverse than those of the general population. The 1977–1978 NFCS reports intakes of 375 foods (reported as RACs), for the entire sample of more than 30,000 people. In contrast, 148 RACs were reported for the 457 nonnursing infants sampled, indicating a relative lack of diversity in the diets of this subgroup. It is therefore important to monitor both the percentage of total diet and the multiple of the national average consumption for each food and for each age group to identify areas relative to dietary exposure to pesticides.

Several factors must be considered when evaluating the consumption data on infants. For example, caution must be exercised not to overestimate introduced foods to avoid estimates of intake that greatly exceed the consumption capacity of the infant. Furthermore, changes resulting from various processing techniques, especially fractionation (e.g., for soybean oil), must not be overlooked. Other influences include many of those noted earlier for nursing infants, e.g., region, socioeconomic status, and other variables.

Milk also predominates in the diet of 1- to 6-year-ld children, as shown by the values for nonfat and fat milk solids (30.4% and 13.4%, respectively). Orange juice, fresh apples, apple juice, and bananas together constitute 11.1% of their diet. The data also show that the diets of this group have become more diverse to include wheat, beef, sugar, eggs, and chicken, and more varieties of vegetables are consumed. The number of foods eaten above the U.S. average, and the multiples of their excess, have declined. This is most likely attributable not only to the rapid increase in dietary diversity after the age of 1 year, but also to a diminishing effect of the body weight conversion factor as average childhood weights approach average adult weights.

The diets of children from 7 to 12 years old have attained a greater level of diversity and show changes in proportion. Wheat flour, beef, and potatoes have reached higher percentages of the diet, whereas the intake levels of some foods (e.g., milk constituents and orange juice) have declined slightly. Many of the foods that constitute the greatest percentages of the diet are the same as those for 1to 6-year-old children, although there are differences in the rankings of the percentages of the total diet of those age groups. As shown by the multiples of the U.S. average consumption, the intakes by 7- to 12-year-old children are approaching those of the overall population.

Among teenagers from 13 to 19 years old, wheat flour, beef, potatoes, and eggs continue their ascendance in dietary predominance over fruits and vegetables with the exceptions of orange juice, apples, and tomatoes. Orange juice ranks fifth among the foods that constitute the highest percentages of the diet consumed by this age group. The multiples of the U.S. average consumption have declined, and for 10 food items, are below the national average.

There is a dramatic drop in nonfat milk solids for adults 20 years old and older (from 28.6% of the total diet of teenagers to 18.2% for adults). All the foods comprising more than 1% of the diet of teenagers appear again for adults along with lettuce, fin fish, and fat pork. All food items listed except lettuce are consumed in amounts less than the national average, showing the greater diversity of the adult diet.


Differences in Consumption Among Age Groups

As demonstrated by the NFCS data (Tables 5-5, 5-6 and 5-7), infant and early childhood consumption patterns differ greatly from those of adults and the population as a whole. The most dominant factor that emerges from interpretive evaluation of intake data remains as milk in some form. The proportion of the balance of the diet represented by fruits, by liquids, especially fruit juices and by vegetables is much higher for infants than for older children or adults. For several fruit juices, this relationship is maintained even when commodity intake is not adjusted by body weight. Thus, use of the U.S. average intake as a basis for estimating pesticide exposures of infants and children may result in underestimates of pesticide residues in their diet.

As noted earlier in this chapter, water intake is considerably higher for infants than for other age categories. The sources of this high intake include foods such as concentrated juices, cereals, and infant formula that are mixed with water prior to consumption. Therefore, water must be a major consideration in estimating of the risk from dietary exposure to pesticides for infants and children.

The Gerber data show several changes in the diet over the first year of life that have potentially important implications for the exposure of infants to pesticides (Johnson et al., 1981). There is a gradual substitution of cow's milk and apple juice for human milk and infant formula. Soybean oil and coconut oil consumption decreases, reflecting the declining intake of infant formula. Intakes of fruits and vegetables increase steadily, and the diet in general begins to diversify.

Because of the young ages of the study population, the 90th percentile of the Gerber sample often shows consumption of such foods as peaches, pears, carrots, and oats for only a few individuals. As the percentage of eaters increases as the children grow older, the 90th percentile commonly exceeds the mean of eaters only and would therefore provide a higher degree of protection if used as the basis of risk projections and tolerance setting.

Differences in Consumption within Age Groups

Variations in consumption patterns within age groups may be considerable. Consider, for example, the great differences among the diets of vegetarians, certain ethnic and religious groups, individuals with medically restricted diets, and populations in different regions of the country. It is quite probable that subgroups within the various age categories consume both average daily and daily high levels of food considerably above the NFCS values, but an analysis of these variations is not possible with current survey data.

For chronic risk assessment, EPA has traditionally relied on mean consumption estimates for large groups such as the entire U.S. population or children between the ages of 1 and 6 years or 7 and 12 years. The 90th percentile has been avoided, because no individual consumes the 90th percentile of all foods regulated by the FDA and USDA. Some individuals may, however, consistently consume a few foods at the 90th percentile for their age group. Obviously, these clusters would vary among individuals and eventually resolve to the mean for the age group.

To determine the significance of this observation, the committee identified the 25 foods most consumed by 170 1-year-old children in the 1985–1986 CSFII. Results from DRES calculations based on CSFII results illustrate that mean intake data clearly do not reflect the actual intakes of some foods by some subjects and therefore that their use can lead to underestimates of pesticide exposure. For example, 88 of the children (52%) consumed 6 to 10 foods over the mean, and only 5 (2.9%) consumed no foods over the mean. The many foods over the mean. The many foods consumed at levels lower than the mean constituted a minor portion of the diet.

These considerations regarding consumption are used as the basis for the committee's discussions in the remaining chapters of this report.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 4:27 am

Part 2 of 2


The committee reached its conclusions and recommendations for this chapter after an extensive review of information available on food and water consumption. All major sources of data on food consumption by infants and children compiled by government agencies and the food industry were considered and evaluated. Information on water intake was derived from an extensive evaluation of food survey information conducted for the National Cancer Institute. The committee weighed the strengths and the limitations of each source of information in order to identify the most acceptable data to be used in conjunction with pesticide residue information to determine exposure, as discussed in Chapters 6 and 7.

Information on foods as consumed was broken down into constituents of foods by the same process currently used by government to establish regulatory policy. This conversion process—the Dietary Residue Evaluation System (DRES)—expresses foods in terms of individual raw agricultural commodities (RACs).

Potential dietary exposures to pesticides are related to food and water intake. To identify the differences in exposures of infants and children compared with adults, intake data were grouped into various age categories. Consumption of water, human milk, and processed foods were considered separately in order to assess their respective contributions to dietary pesticide exposure for infants and young children.


• Food consumption patterns for infants and children differ markedly from those of adults.

• Children consume more calories relative to body weight than do adults.

• Dietary diversity increases with age: infants and young children consume fewer distinct foods than do adults.

• On a body weight basis, infants and young children consume notably more of certain foods than do adults.

• Water, as drinking water and as a component of food, is not adequately considered in most consumption surveys.

• Examination of intake data by age categories clearly illustrates the differences in consumption patterns that must be considered when estimating exposure of infants and children to pesticides; however, current information on food and water intake by age category is insufficient to produce credible exposure estimates.

• Processed foods are predominant in the diets of younger age groups.


Knowledge of food consumption is an important consideration in assessing the risk to infants and children from dietary exposures to pesticides. Therefore, more focused, direct, comprehensive, and contemporary dietary information is required for infants and children. Because of the myriad and rapid changes in diet that occur during the developmental stages of life, intake data must be precisely divided into age subdivisions and the sample must be large enough to produce meaningful results.

• A simple, uniform method needs to be developed for conversion of a product as consumed to its components in terms of raw agricultural commodities.

• Those foods most frequently consumed by young children and infants need to be identified and quantified more specifically. Reporting should be specific and discrete foods clearly identified.

• Water intake and food intake should both receive full consideration in estimating dietary exposure and assessing risk, especially for infants and young children.

• Because of the changing nature of children's diets during growth, food consumption surveys should include adequate sample sizes of children aged 0 to 12 months, 13 to 24 months, 25 to 36 months, 37 to 48 months, 49 to 60 months, 5 to 10 years, and 11 to 18 years.

• Intake data and survey methodology need to be standardized to make them more useful in a variety of applications, including estimating exposure and assessing risk.

• Food consumption surveys should be coordinated among the involved organizations and performed on a continuing basis in order to examine trends of food and water consumption, especially by infants and children.


Acheson, K.J., I.T. Campbell, O.G. Edholm, D.S. Miller, and M.J. Stock. 1980. The measurement of food and energy intake in man: An evaluation of some techniques. Am. J. Clin. Nutr. 33:1147–1154.

Acosta, P.B., C. Trahms, N.S. Wellman, and M. Williamson. 1983.Phenylalanine intakes of 1- to 6-year-old children with phenylketonuria undergoing therapy. Am. J. Clin. Nutr. 38:694–700.

Anderson, G.H., and L.M. Blendis. 1981. Plasma neutral amino acid ratios in normal man and in patients with hepatic encephalopathy: Correlations with self-selected protein and energy consumption. Am. J. Clin. Nutr. 34:377–385.

Axelson, J.M., and M.M. Csernus. 1983. Reliability and validity of a food frequency checklist. J. Am. Diet. Assoc. 83:152–155.

Backstrom, C.H., and G.D. Hursh-Cesar. 1981. Survey Research, 2nd ed. New York: Macmillan.

Basiotis, P.P., S.O. Welsh, F.J. Cronin, J.L. Kelsay, and W. Mertz. 1987. Number of days of food intake records required to estimate individual and group nutrient intakes with defined confidence. J. Nutr. 117:1638–1641.

Beaton, G.H., J. Milner, P. Corey, V. McGuire, M. Cousins, E. Stewart, M. de Ramos, D. Hewitt, P.V. Grambsch, N. Kassim, and J.A. Little. 1979. Sources of variance in 24-hour dietary recall data: Implications for nutrition study design and interpretation. Am. J. Clin. Nutr. 32:2546–2549.

Beaton, G.H., J. Milner, V. McGuire, T.E. Feather, and J.A. Little. 1983. Source of variance in 24-hour dietary recall data: Implications for nutrition study design and interpretation. Carbohydrate sources, vitamins, and minerals. Am. J. Clin. Nutr. 37:986–995.

Block, G. 1982. A review of validations of dietary assessment methods. Am. J. Epidemiol. 115:492–505.

Block, G., A.M. Hartman, C.M. Dresser, M.D. Carroll, J. Gannon, and L. Gardner. 1986. A data-based approach to diet questionnaire design and testing. Am. J. Epidemiol. 124:453–469.

Brown, K.H., R.E. Black, A.D. Robertson, N.A. Akhtar, G. Ahmed, and S. Becker. 1982. Clinical and field studies of human lactation: Methodological considerations. Am. J. Clin. Nutr. 35:745–756.

Burk, M.C., and E.M. Pao. 1976. Methodology for Large-Scale Surveys of Household and Individual Diets. HERR No. 40. Washington, D.C.: U.S. Department of Agriculture.

Butte, N.F., W.W. Wong, B.W. Patterson, C. Garza, and P.D. Klein. 1988. Human-milk intake measured by administration of deuterium oxide to the mother: A comparison with the test-weighing technique. Am. J. Clin. Nutr. 47:815–821.

Butte, N.F., S. Villalpando, W.W. Wong, S. Flores-Huerta, M.J. Hernandez-Beltran, E.O. Smith, and C. Garza. 1992. Human milk intake and growth faltering of rural Mesoamerindian infants. Am. J. Clin. Nutr. 55:1109–1116.

Byers, T., J. Marshall, E. Antony, R. Fiedler, and M. Zielezny. 1987. The reliability of dietary history from the distant past. Am. J. Epidemiol. 125:999–1011.

Chu, S.Y., L. N. Kolonel, J.H. Hankin, and J. Lee. 1984. A comparison of frequency and quantitative dietary methods for epidemiologic studies of diet and disease. Am. J. Epidemiol. 119:323–334.

Code of Federal Regulations (CFR). 1986a. Tolerances and Exemptions from Tolerances for Pesticide Chemicals in or on Raw Agricultural Commodities. 40 CFR 1, Part 180.

Code of Federal Regulations (CFR). 1986b. Tolerances for Pesticides in Food Administered by EPA. 21 CFR 1, Part 193.

Coward, W.A., T.J. Cole, H. Guber, S.B. Roberts, and I. Fleet. 1982. Water turnover and measurement of milk intake. Pfleugers. Arch. 393:344–347.

Dewey, K.G., M.J. Heinig, L.A. Nommsen, and B. Lönnerdal. 1990. Low energy intakes and growth velocities of breast-fed infants: Are there functional consequences? Pp. 35–44 in Activity, Energy Expenditure, and Energy Requirements of Infants and Children, B. Schürch and N. Scrimshaw, eds., Lausanne, Switzerland: International Dietary Energy Consulting Group-Nestlé Foundation.

DHHS (Department of Health and Human Services). 1983. Dietary Intake Source Data: United States 1976–80. DHHS Pub. No. (PHS)83-1681. National Center for Health Statistics. Washington, D.C.: U.S. Government Printing Office. 483 pp.

Dwyer, J.T. 1988. Assessment of dietary intake. Pp. 887–905 in Modern Nutrition in Health and Disease, 7th ed., M.E. Shils and V.R. Young, eds. Philadelphia: Lea and Febiger.

Elahi, V.K., D. Elahi, R Andres, J.D. Tobin, M.G. Butler, and A.H. Norris. 1983. A longitudinal study of nutritional intake in men. J. Gerontol. 38:162–180.

Ershow, A.B., and K.P Cantor. 1989. Total Water and Tapwater Intake in the United States: Population-Based Estimates of Quantities and Sources . Bethesda, Md.: Life Sciences Research Office, Federation of American Societies for Experimental Biology.

Frank, G.C., A.W. Voors, P.E. Schilling, and G.S. Berenson. 1977. Dietary studies of rural school children in a cardiovascular survey. J. Am. Diet. Assoc. 71:31–35.

Freudenheim, J.L., N.E. Johnson, and R.L. Wardrop. 1987. Misclassification of nutrient intake of individuals and groups using one-, two-, three-, and seven-day food records. Am. J. Epidemiol. 126:703–713.

Fuller, W.A. 1991. I. Regression estimation for sample survey data. In Regression Estimation for the 1987–1988 Nationwide Food Consumption Survey. Washington, D.C.: U.S. Department of Agriculture.

Fuller, W.A., M.M. Loughin, and H. Baker. 1991. II. Application to the Nationwide Food Consumption Survey. In Regression Estimation for the 1987–1988 Nationwide Food Consumption Survey. Washington, D.C.: U.S. Department of Agriculture.

Futrell, M.F., L.T. Kilgore, and F. Windham. 1971. Nutritional status of Negro preschool children in Mississippi: Evaluation of HOP index. J. Am. Diet. Assoc. 59:218–223.

GAO (General Accounting Office). 1991. Nutrition Monitoring: Mismanagement of Nutrition Survey Has Resulted in Questionable Data . (GAO/RCED-91-117) Washington, D.C.: U.S. General Accounting Office.

Gersovitz, M., J.P. Madden, and H. Smiciklas-Wright. 1978. Validity of the 24-hour-dietary recall and seven-day record for group comparisons. J. Am. Diet. Assoc. 73:48–55.

Hampe, E.C., Jr., and M. Wittenberg. 1964. The Lifeline of America: Development of the Food Industry. New York: McGraw-Hill.

Jain, M., G.R. Howe, K.G. Johnson, and A.B. Miller. 1980. Evaluation of a dietary history questionnaire for epidemiologic studies. Am. J. Epidemiol. 111:212–219.

Johnson, G.H., G.A. Purvis, and R.D. Wallace. 1981. What nutrients do our infants really get? Nutr. Today 16:4–10, 23—27.

Kim, W.W., W. Mertz, J.T. Judd, M.W. Marshall, J.L. Kelsay, and E.S. Prather. 1984. Effect of making duplicate food collections on nutrient intakes calculated from diet records. Am. J. Clin. Nutr. 40:1333–1337.

Krall, E.A., and J.T. Dwyer. 1987. Validity of a food frequency questionnaire and a food diary in a short-term recall situation. J. Am. Diet. Assoc. 87:1374–1377.

Krantzler, N.J., B.J. Mullen, H.G. Schutz, L.E. Grivetti, C.A. Holden, and H.L. Meiselman. 1982. Validity of telephoned diet recalls and records for assessment of individual food intake. Am. J. Clin. Nutr. 36:1234–1242.

Madden, J.P., S.J. Goodman, and H.A. Guthrie. 1976. Validity of the 24 hr. recall: Analysis of data obtained from elderly subjects. J. Am. Diet. Assoc. 68:143–147.

Marr, J.W., and J.A. Heady. 1986. Within- and between-person variation in dietary surveys: Number of days needed classify individuals. Hum. Nutr. Appl. Nutr. 40:347–364.

McGee, D., G. Rhoads, J. Hankin, K. Yano, and J. Tillotson. 1982. Within-person variability of nutrient intake in a group of Hawaiian men of Japanese ancestry. Am. J. Clin. Nutr. 36:657–663.

Neville, M.C., R. Keller, J. Seacat, V. Lutes, M. Neifert, C. Casey, J. Allen, and P. Archer. 1988. Studies in human lactation. Milk volumes in lactating women during the onset lactation and full lactation. Am. J. Clin. Nutr. 48:1375–1386.

NRC (National Research Council). 1986. Nutrient Adequacy: Assessment Using Food Consumption Surveys. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1989. Recommended Dietary Allowances: 10th ed. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1991. Nutrition During Lactation. Washington, D.C.: National Academy Press.

Nusser, S.M., A.L. Carriquiry, and W.A. Fuller. 1991. A semiparametric transformation approach to estimating usual daily intake distributions. Unpublished.

Orr-Ewing, A.K., P.F. Heywood, and W.A. Coward. 1986. Longitudinal measurements of breast milk output by a 2H2O tracer technique in rural Papua New Guinean women. Hum. Nutr. Clin. Nutr. 40:451–467.

Patterson, L. 1971. Dietary intake and physical development of Phoenix area children. J. Am. Diet. Assoc. 59:106–110.

Petersen and Associates, Inc. 1992. Background: Derivation of Water Consumption Estimates Used by the Tolerance Assessment System. Bethesda, Md.: Petersen and Associates.

Pietinen, P., A.M. Hartman, E. Haapa. L. Rasanen, J. Haapakoski, J. Palmgren, D. Albanes, J. Virtamo, and J.K. Huttunen. 1988. Reproductibility and validity of dietary assessment instruments. II. A qualitative food frequency questionnaire. Am. J. Epidemiol. 128:667–676.

Posner, B.M., C.L. Borman, J.L. Morgan, W.S. Borden, and J.C. Ohls. 1982. The validity of a telephone-administered 24-hour dietary recall methodology. Am. J. Clin. Nutr. 36:546–553.

Purvis, G.A. 1973. What nutrients do our infants really get? Nutr. Today 8:28–34.

Purvis, G.A., and S.J. Bartholmey. 1988. Infant feeding practices: Commercially prepared baby foods. Pp. 399–417 in Nutrition During Infancy, R.C. Tsang and B.L. Nichols, eds. Philadelphia: Hanley & Belfus .

Research Triangle Institute. 1983. Interim Report Number 1: The Construction of a Raw Agricultural Commodity Consumption Data Base. RTI Project No. 252U-2123-7. Research Triangle Park, N.C.: Research Triangle Institute.

Russell-Briefel, R., A.W. Caggiula, and L.H. Kuller. 1985. A comparison of three dietary methods for estimating vitamin A intake. Am. J. Epidemiol. 122:628–636.

Sampson, L. 1985. Food frequency questionnaire as a research instrument. Clin. Nutr. 4:171–178.

Saunders, D.S., and B.J. Petersen. 1987. An Introduction to the Tolerance Assessment System. Hazard Evaluation Division, Office of Pesticide Programs. Washington, D.C.: U.S. Environmental Protection Agency.

Schnakenberg, D.D., T.M. Hill, M.J. Kretsch, and B.S. Morris. 1981. Dietary-interview technique to assess food consumption patterns of individual military personnel. Pp. 180–197 in Assessing Changing Food Consumption Patterns, Committee on Food Consumption Patterns, National Research Council. Washington, D.C.: National Academy Press.

Sempos, C.T., N.E. Johnson, E.L. Smith, and C. Gilligan. 1984. A two-year dietary survey of middle-aged women: Repeated dietary records as a measure of usual intake. J. Am. Diet. Assoc. 84:1008–1013.

Sorenson, A.W., B.M. Calkins, M.A. Connolly, and E. Diamond. 1985. Comparison of nutrient intake determined by four dietary intake instruments. J. Nutr. Educ. 17:92–99.

Stamler, J. 1979. Population studies. Pp. 25–88 in Nutrition, Lipids, and Coronary Heart Disease, R.I. Levy, B.M. Rifkind, B.H. Dennis, and N. Ernst, eds. New York: Raven Press.

Todd, K.S., M. Hudes, and D.H. Calloway. 1983. Food intake measurement: Problems and approaches. Am. J. Clin. Nutr. 37:139–146.

USDA (U.S. Department of Agriculture). 1983. Nationwide Food Consumption Survey. Food Intakes: Individuals in 48 States, Year 1977–78. Report No. I-1. Hyattsville, Md.: Consumer Nutrition Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1985. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Women 19–50 Years and Their Children 1–5 Years, 1 Day, 1985. Report No. 85-1. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1986a. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Low-Income Women 19–50 Years and Their Children 1–5 Years, 1 Day, 1985. Report No. 85-2. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1986b. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Men 19–50 Years, 1 Day, 1985. Report No. 85-3. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1987a. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Low-Income Women 19–50 Years and Their Children 1–5 Years, 1 Day, 1986. Report No. 86-2. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1987b. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Women 19–50 Years and Their children 1–5 Years, 1 Day, 1986. Report No, 86-1. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1987c. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Women 19–50 Years and Their Children 1–5 Years, 4 Days, 1985. Report No. 85-4. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1988. Nationwide Food Consumption Survey. Continuing Survey of Food Intakes of Individuals: Low-Income Women 19–50 Years and Their Children 1–5 Years, 4 Days, 1985. Report No. 85-5. Hyattsville, Md.: Nutrition Monitoring Division, Human Nutrition Information Service.

Willett, W.C., L. Sampson, M.J. Stampfer, B. Rosner, C. Bain, J. Witschi, C.H. Hennekens, and F.E. Speizer. 1985. Reproducibility and validity of a semiquantified food frequency questionnaire. Am. J. Epidemiol. 122:51–65.

Wilson, L.B., and B.B. Rothschild. 1987. Results of an Exploratory study of longitudinal measures of individual food intake. Pp. 9–19 in Research on Survey methodology: Proceedings of a Symposium. USDA Admin. Rep. No. 382. Washington, D.C.: U.S. Department of Agriculture.

Woolridge, M.W., N. Butte, K.G. Dewey, A.M. Ferris, C. Garza, and R.P. Keller. 1985. Methods for the measurement of milk volume intake of the breastfeed infant. Pp. 5–21 in Human Lactation: Milk Components and Methodologies, R.G. Jensen and M.C. Neville, eds. New York: Plenum.

Young, C.M., G.C. Hagan, R.E. Tucker, and W.D. Foster. 1952. A comparison of a dietary study methods II. Dietary history vs. seven-day record vs. 24-hr. recall. J. Am. Diet. Assoc. 28:218–221.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 4:29 am

Part 1 of 3

6. Pesticide Residues

DATA ON DIETARY LEVELS of pesticide residues combined with food consumption estimates provide the basis for exposure estimates used by the Environmental Protection Agency (EPA) to assess the risks of pesticide exposure in the diet. Thus, sampling and residue testing methods to estimate levels of pesticide residues in the food supply are extremely important components of the risk assessment process.

The committee examined pesticide usage, residue sampling and testing methods, and the data pesticide residues to

• understand the relative quality of data sets available to EPA as a foundation for recommending practical improvements in data collection and testing;

• identify the foods in the diets of infants and children with residues of pesticides that cause the greatest public-health concern;

• assess the need for residue sampling methods and residue testing procedures that can provide the data needed to ensure the protection of infants and children;

• recommend residue monitoring methods that could be incorporated into an exposure assessment methodology that would ensure the protection of infants and children;

• identify steps to improve risk assessment and establish priorities for those steps; and

• determine which, if any, data are sufficient quality to support risk assessment models designed to protect infants and children.


Despite the importance pesticides have attained in agricultural production, data on the amount and distribution of their use are remarkably scanty. There is no single, comprehensive data source, derived from actual sampling, on pesticide usage for all crops and all chemicals.

The U.S. Department of Agriculture's (USDA) Economic Research Service (ERS) conducted national surveys of pesticide use in 1964, 1966, 1971, 1976, and 1982; smaller areas and fewer crops have been included in successive surveys. The 1964, 1966, and 1971 surveys included field crops, fruits, vegetables, and livestock. In 1976 fruits and vegetables were excluded from the survey, and in 1982, only major field crops (e.g., corn, soybeans, cotton, wheat, barley, oats, peanuts, tobacco, alfalfa, and hay) were sampled (Osteen and Szmedra, 1989). The foci of later reports on pesticide usage are even narrower: vegetable, melon, and strawberry crops in Arizona, Florida, Michigan, and Texas (USDA, 1991); fruits and nuts, in 12 states (USDA, 1992a); and eight field crops (corn, cotton, peanuts, potatoes, rice, sorghum, soybeans, and wheat) in different numbers of states, ranging from 47 states for corn down to 2 states for rice and 1 for durham wheat (Osteen and Szmedra, 1989).

Resources for the Future maintains a county-based file of annual pesticide usage estimates by county and by crop for the 184 widely used pesticides that appear on EPA's list for the National Ground Water Survey and the California Priority Pollutant List (Gianessi, 1986). The usage information was derived from the limited ERS surveys and from the annual California survey (State of California, 1981), which included only restricted-use of chemicals until 1991, when the state's reporting system was extended to all pesticides, including unrestricted chemicals. Resources for the Future has also estimated the amounts of pesticides applied to lawns and in nurseries.

The data in Table 6-1 illustrate the variation in the kind and amount of pesticides used on crops in various geographic regions. The corn belt, for example, accounted for 39% of all pesticides used on major crops in 1982. Most of this volume was represented by herbicides; fungicides constituted only 2% of total usage. In contrast, the southeast accounted for only 8% of total pesticide applications but for 66% of fungicides used. There are similar differences in use patterns between other regions.

The implications for residue and exposure estimation are more clearly illustrated in Table 6-2, which focuses on one crop (fall potatoes) and one class of pesticides (fungicides) and their application in the northeast, midwest, and western regions of the United States. In 1991, 96% and 90% of croplands planted with potatoes in the northeast and midwest, respectively, were treated with fungicides, while only 52% of croplands in the west were treated. Fungicides were also applied more times during the growing season in the northeast and midwest. As a result, the northeast, which accounts for 11% of the hectares planted with potatoes, accounts for 30% of all fungicide hectare treatments


TABLE 6-1 Regional Distribution of Pesticide Use on Major Crops in Selected Regions in 1982

NOTE: Major crops included corn, soybeans, cotton, wheat, barley, oats, peanuts, tobacco, alfalfa, and hay.
a Totals do not add up to 100 due to rounding.

SOURCE: Based on data from Osteen and Szmedra, 1989.


TABLE 6-2 Total Fungicide Use on Fall Potatoes in the United States, 1991

NOTE: Numbers do not add up to 100 due to rounding.

a Northeast: Maine, New York, Pennsylvania; Midwest: Michigan, Minnesota, North Dakota, Wisconsin; West: Colorado, Idaho, Oregon, Washington.

b Hectare treatment: number of hectares treated times number of applications per year.

SOURCE: Derived from USDA, 1992c.

In contrast, the west, which accounts for 59% of all hectares planted with potatoes, accounts for only 28% of total fungicide hectare treatments. This variation of pesticides used on the same crop grown in different regions means that the amount and kind of residues will depend not only on the crop, but also on where it is grown.


Pesticide residues originate when a crop or food animal (commodity) is treated with a chemical or exposed unintentionally by drift, in irrigation water, in feed, or by other routes. The size of the residue depends on the exposure level (treatment rate), its dissipation rate, environmental factors, and its physical and chemical properties. For example, an insecticide sprayed on apples any volatilize into the atmosphere. This is influenced by the insecticide's volatility or vapor pressure and the temperature and wind movement in the orchard. Removal by rainfall or overhead irrigation is governed by the insecticide's water solubility and the amount of rain or irrigation water. The chemical may also degrade (as influenced by the molecular makeup of the insecticide and by such factors as sunlight, moisture, and temperature) or it may dissipate by growth dilution (e.g., as the fruit becomes larger, the residue concentration will decrease even in the absence of physical or chemical dissipation). In farm animals and some plants, metabolism and excretion are the primary mechanisms. The degradation products become the major constituents of the remaining residue. In a few cases, chemical residue concentrations may actually increase over time after exposure ceases. This would result from weight loss by the commodity, e.g., loss resulting from the conversion of grapes to raisins after treatment with a relatively stable, nonvolatile chemical.

The overall dissipation rate is a composite of the rate constants of the individual processes (e.g., volatilization and degradation). Typically, overall residue concentrations (parent plus degradation products) decrease over time after exposure ends. Because most individual dissipation processes follow first-order kinetics, overall dissipation will have the characteristics of first-order kinetics. In first-order decline, the logarithm of concentration is linearly related to time, and a plot of concentration remaining versus time is asymptotic with respect to the time coordinate. Thus, residue concentrations will approach zero over time but in theory will never cease to exist entirely (Zweig, 1970). Stated simply, a commodity treated with or exposed to a pesticide theoretically can never totally be rid of all traces of residue. In time, however, the residue will cease to be detectable because of the limitations of current measuring instrumentation and the continuing asymptotic decline processes. This limit of detection (LOD) will therefore vary according to the sensitivity of the analytical method used. (LODs are described below under ''Detection Limits.")

Conventionally, residues in raw commodities are monitored until they have declined to a concentration approximately 1/10th that of the legal maximum—that is, the tolerance or action level. Very little public monitoring is intended to identify the residues that the consumer may ingest, which may range from the legal maximum to 1/10th, 1/100th, 1/1,000th, or smaller fractions of that level on foods prepared for consumption. One can expect that consumers are exposed to small residues if their food was treated with or exposed to pesticides during production, processing, or preparation; however, we do not always know the quantity of those residues either because they are lower than the LOD or because there are no monitoring data available. For these reasons, it is difficult to estimate actual dietary exposure to pesticides and any associated risk with a high degree of certainty.


Early in the development of a pesticide, the manufacturer must identify the analytical methods used to ascertain the concentrations of chemicals in formulations (formulation methods) and the fate of the material on target crops, in laboratory animals and livestock, and in environmental media (soil, water, air) that might be exposed to the chemical (residue methods). Most companies that develop and register chemicals employ staffs to develop these analytical methods, whereas others hire or fund commercial or university laboratories for this purpose.

Development of analytical methods is a lengthy and technically difficult process because the methods must account for the parent chemical or control agent as well as toxicologically significant formulation impurities, metabolites, and environmental conversion products. The impurities and products may not be known early in the development phase and thus must be included later but before registration is sought from EPA. Before a pesticide can be registered under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), a tolerance level must be established for each food use or an exemption granted, e.g., for a pesticide that is essentially nontoxic. To obtain a formal tolerance level, pesticide manufacturers must submit their analytical methods to EPA, which then verifies that the pesticide can be detected at a certain tolerance level for each proposed food use. It is not unusual for a food tolerance level to include the parent chemical and several breakdown products. In such cases, versatile residue detection methods must be available to detect the various tolerance levels in every food or feed product for which registration is being sought. Typically, the primary method will have several variations extending it to soil, water, air, and nontarget organisms such as fish and wildlife.

The manufacturer applies these methods to determine the rate of dissipation or decline of the pesticide on target crops in field trials. The results are submitted to EPA with the registration data for use in establishing a tolerance level for the raw agricultural commodity and determining the interval required between the last application of the pesticide and harvest to achieve residues below that tolerance.

Field trials are conducted in several geographical regions of the United States that typify areas in which the crop is produced, so that different climatic conditions and soil types are represented. The test plots are treated with pesticides in concentrations high enough to eradicate a large percentage of the target pest(s). If the trials are not complete, if the data are too variable, if conversion products are not adequately included, or if the analytical methods themselves are considered imprecise, inaccurate, not sufficiently sensitive, or otherwise deficient, registration may be denied. EPA may base its judgment on the data submitted by the manufacturer, or it may inspect the company's raw data in accordance with the FIFRA provision for data audits. Field trial data are further evaluated in Chapter 7.

Methods must be provided by the manufacturer when requested by any federal or state regulatory agency and may be included in the Pesticide Analytical Manual, Volume II (PAM II): Methods for Individual Residues, which was first published by the Food and Drug Administration (FDA) in 1968 but has been updated in a series of revisions since then. These methods do not need to fit within the available multiresidue methods (MRMs) used by the FDA to screen food or feed products entering commerce (see section on "Methods for Sampling and Analysis," below, for a further discussion of MRMs). More recently, EPA has asked pesticide manufacturers to determine whether new compounds are detectable by existing MRMs. If they are not, however, the registration process is not impeded.

Interregional Project Number 4

Use of pesticides on some crops (e.g., strawberries, hops, artichokes, cranberries) may be too limited to provide the economic incentive needed for chemical companies to develop the analytical methods and residue data required for registration. In such cases, this work is performed by Interregional Project Number 4 (IR-4), which operates within State Agricultural Experiment Stations (SAES) with funding from USDA's Cooperative State Research Service (CSRS) and Agricultural Research Service (ARS). The nation's four IR-4 leader laboratories are located at Cornell University, the University of Florida, the University of California at Davis, and Michigan State University. Several participating laboratories are situated at other land-grand institutions and within ARS.

The IR-4 laboratories use methods provided by manufacturers to EPA for pesticide residues on the major crops listed on the chemical's label. If the method fails on the minor crop, they modify the company method to make it fit the minor crop situation. Occasionally, they develop new methods for minor crops of interest.

SAES or ARS field scientists establish the plots, sample the commodity at harvest, and provide samples to IR-4 laboratories, which then conduct the analyses. All data are submitted to EPA. If the petition is approved, the minor crop is added to the pesticide label. IR-4 actions annually account for approximately half of the petitions processed by EPA.

Universities and the ARS

Several U.S. universities and the ARS conduct research on pesticides to study their field behavior, formation of breakdown products, persistence during food processing and storage, and analytical behavior, Many advances in food residue chemistry (e.g., detection of previously unrecognized toxic metabolites) and new approaches to residue analysis (e.g., the immunoassay) result from this basic research. In addition, this academic environment provides the training ground for pesticide scientists who eventually enter the industrial, government, and commercial sectors.



Sampling should be conducted

• by persons trained in the practice of sampling;

• randomly, so that all individuals in the population sampled have an equal chance of selection in the final analysis;

• with replication, so that analytical results can be treated statistically;

• in such a manner as to maintain sample integrity by adequate containment, preservation, and prevention of contamination; and

• with care and attention to record keeping, including visual observations, sample preservation, and safeguards against cross-contamination.

Usually omitted from reports are the manner of collecting samples (where, by whom, and how) and information on compositing, subsampling, preservation of samples and subsamples, and other important matters. Lykken (1963) generalizes, however, that all residue monitoring programs operate somewhat as follows:

• Several commodity units (e.g., bunches of grapes, oranges, heads of cabbage) are taken from the field or lot to be sampled.

• These commodity units are composited to form to gross sample.

• The gross sample is reduced in size to produce the composite sample.

• The composite units are then peeled, husked, or further reduced in size by cutting or chopping in accordance with the Code of Federal Regulations, which identifies the portion(s) of the commodity to which the tolerance applies.

• The individual parts of the commodity may then be quartered to reduce bulk and perhaps subdivided to smaller aliquots. These samples are generally frozen or preserved in some other way, transported to the laboratory, and preserved further until analyzed. If a freezer stability test is to be conducted (a recent Good Laboratory Practice [GLP] requirement; 40 CFR Part 160), control samples may be spiked at this point and then handled the same as the treated sample. This is usually done when field plots are sampled to determine residues for registration requirements, but less frequently for monitoring and enforcement of tolerance levels for registered chemicals. Sampling and sample handling for field trials are described by the National Agricultural Chemicals Association (NACA, 1988).

• At the time of analysis, individual subsamples may be more extensively chopped or blended or reduced further in size prior to extraction with a solvent and analysis.(See PAM I or II for more detailed description.)

The absence of uniform training has likely led to haphazard sampling or bias resulting in samples that are neither random nor representative. To rectify this situation, FDA and most state agencies are taking steps to improve their training and written sampling guidelines. Furthermore, true replication, with three or more field composites, appears not to have been common practice, evidenced by the fact that averages and standard deviations are absent from virtually all residue monitoring reports. Sample handling has improved since implementation of GLP protocols; but again, without accompanying quality assurance records, older data must be questioned for reliability. EPA and FDA are now training and certifying field inspectors to ensure proper sampling by all personnel engaged in work to meet FIFRA requirements. In addition, the American Chemical Society's Committee on Environmental Improvement has prepared a comprehensive volume dealing with the basics of environmental sampling (Keith, 1988).


Methods for analyzing pesticides are expensive, time consuming, and difficult, and they require a skilled analyst. Furthermore, methods are tailored to specific purposes (e.g., monitoring, enforcement, or registration). As a result, considerable variability is associated with the methodology. In many cases, descriptions of differences among the specific methods used do not accompany the residue data, thus diminishing public confidence in the data. Furthermore, the committee found no analytical program directed toward water specifically as an ingredient of foods or as a component added to foods. This results in an important gap in the residue data, since water represents such a large part of the diets of infants and children.

There are two general types of analytical methods for determining residues in foods: single residue methods and multiresidue methods. These are described in the following sections.

Single Residue Methods

Single residue methods (SRMs) are used for the quantitative determination of a single pesticide (and its toxicologically important conversion products, e.g., through metabolism or degradation) in all foods for which tolerance levels have been established. This is generally the type of method submitted by the manufacturer to EPA and eventually published in PAM II after registration is secured. It may also be the method used (sometimes in modified form) for IR-4 petitions.

Multiresidue Methods

Multiresidue Methods (MRMs) are capable of detecting and quantifying more than one pesticide in more than one food. These methods are commonly used by government agencies for surveillance and monitoring to determine which pesticides (and how much) are present in a given food sample. FDA's MRMs are published in PAM I; the MRMs of state agencies, foreign governments, private industry, and academia are published in the open literature or in special reports. Some MRMs are rapid; others are more comprehensive and therefore more time consuming. In general, MRMs may be used for screening and quantitation. In screening, MRMs are used to determine rapidly if any pesticide is present near or above the tolerance level. This approach usually precedes a more detailed analysis. Cholinesterase enzyme inhibition tests screen for organophosphorus and carbamate insecticides; insect bioassays screen for any insecticide residue. Immunoassays may be used in the future for targeted chemicals or classes of chemicals. In quantitation, MRMs are used to detect and measure multiple pesticide residues and their metabolites that might be present in a given sample. These MRMs are usually based on gas or liquid chromatography or both. FDA and other agencies often use simplified versions of MRMs in their surveillance program to determine if violations exist in given samples before proceeding to full quantitation with a more elaborate version. Because all MRMs can accommodate only a limited number of chemicals, agencies use SRMs for targeted pesticides that are not included in the MRM. They also use SRMs in special circumstances such as when public health is endangered by a single pesticide or when a single pesticide comes under special review and, thus, special scrutiny is required for its presence in foods.

Most laboratories improvise when using an MRM, and the improvisations are often not subject to peer review or published. Requesting the latest method from an agency is usually the only way to obtain up-to-date information on the method being used, the number of pesticides it can accommodate, and its LOD. MRMs used by regulatory laboratories are frequently modified in response to changing availability of solvents and analytical instrumentation within the laboratory and the need to expand the MRM's coverage or lower its detection limit.

Criteria for Selecting a Method

Single Residue Methods or Multiresidue Methods?

SRMs are selected when the sample is known or believed to contain the residue of a chemical not included in the MRM. MRMs are used when the residue history of the sample is unknown and the presence and quantity of pesticide residues must be determined. MRMs will provide information on a much broader range of pesticides than an SRM for the same investment of time, energy, and resources.

Breadth of Applicability

MRMs most commonly used by the FDA can determine roughly 50% of the approximately 300 pesticides with EPA tolerances and other chemicals for which no tolerances have been established. Some of the MRMs can also detect many metabolites, impurities, and alteration products of pesticides with and without tolerances (FDA, 1991). Typically not included are polar chemicals of high water solubility (e.g., paraquat, glyphosate), very volatile chemicals (e.g., fumigants), and compounds that are unstable to Florisil chromatography (e.g., some carbamates). An aliquot of the sample (or its extract) must be analyzed separately so that these chemicals can be included in the analytical report.

Detection Limits

All analytical methods have a limit below which the chemical could not be detected even if present. This limit of detection (LOD) is the lowest concentration that can be determined to be statistically different from a blank. Elsewhere in this report, the committee refers to the limit of quantification (LOQ), which differs from the LOD in that it refers to the concentration above which quantitative results may be obtained with a specified degree of confidence.

The LOD is influenced by extraneous, background material that is always present in the sample and the sensitivity of the instrumentation used for detection and quantification. Moreover, the LOD may vary according to application. LODs are determined by analyzing background (untreated) samples of the food products of interest and spiked samples, which contain known amounts of the chemicals. LODs for a given method will vary with the type of sample, the chemical, and the extent of sample cleanup provided.

LODs can be as low as twice the background reading. That is, a signal that is twice the background could be measured and result in a calculated residue value. In practice, however, most laboratories set an LOQ that is several time higher than the theoretical LOD. Keith (1983) provides general guidelines for establishing the LOQ, but in practice, the criteria for setting the LOQ varies among laboratories.

To be of regulatory use, detection limits must be below established tolerance levels. The California Department of Food and Agriculture sets LODs at approximately one-tenth the tolerance level; FDA generally sets them at 0.1 to 0.01 ppm, depending on the chemical; and Florida's Department of Agriculture and Consumer Services sets them at or just below tolerance in order to screen large numbers of samples for clear violations. Unfortunately, LODs are not always specified in residue reports so that samples with no detectable residue levels cannot be assigned an upper limit of finite residue content. Furthermore, the reports do not clearly describe the extent to which residues below tolerance but above LOD are quantitated and confirmed, and they may not include a complete list of the pesticides that were not found but could have been detected had they been present. A report of pesticides not found (i.e., below the LOD) is usually not included in descriptions of the results of the overall programs but is done when there is special regulatory interest in specific pesticide residues (McMahon and Burke, 1987). Reporting only positive findings leads to a bias in the residue results.

Accuracy and Precision

Accuracy refers to agreement between a measured value and the true value. In residue methods, accuracy is often defined as the percentage recovery. Acceptable residue methods will give 80% to 120% recovery, indicating that if 1 ppm of a chemical were present, the analytical method would yield results between 0.8 and 1.2 ppm. Precision refers to reproductibility and the variability existing in a set of replicate measurements. Precision errors caused by variable reproducibility in residue methods tend to run high, with relative standard deviations (expressed as a percent of the mean) of 25% or more. The total error (accuracy plus precision) ideally should not exceed 100%. This must be assessed by the analyst running replicate spiked samples through the method.

Speed and Cost

Regulatory agencies require fast-response methods that can produce results in an 8-hour workday or less so that produce does not spoil when awaiting the results of an analysis. These faster methods are less expensive because they require less of an analyst's time. Needless to say, however, the quickest, least expensive analytical method may not be the best one in terms of other criteria. As a result, many of the methods used are compromises of speed for quality.


Most regulatory agencies rely on element-selective gas chromatography determination. The more rigorous methods based on mass spectrometry are not practical, especially for screening, given the cost of the equipment. The expense is becoming less of a barrier, however, as analytical laboratories acquire more sophisticated instruments and as the cost of technician time overtakes capital costs of instruments as the primary budgetary consideration.


Methods should be validated before they are used routinely for regulatory purposes. The most rigorous level of validation is a collaborative study of the method by several different laboratories. The Association of Official Analytical Chemists conducts such studies and publishes the validated method as "official" in the Official Methods of Analysis . Because this is a time-consuming process, most methods are validated less rigorously—perhaps by one cross-check either by investigators in the same laboratory or by one outside laboratory. For example, the MRM used by the California Department of Food and Agriculture was developed in-house and had not been subjected to outside collaborative validation when put into service.


The following discussion of monitoring activities for pesticide residues is based primarily on information that existed for 1988 and earlier. The committee realizes that changes in the design and scope of monitoring programs have occurred after 1988 but, unfortunately, information on more recent developments was not generally available for inclusion in the committee's discussion.

Federal Activities

Four federal agencies have primary jurisdiction over pesticide residues in food—the EPA, the FDA, and the USDA's Food Safety and Inspection Service (FSIS) and Agricultural Marketing Service (AMS). Their efforts are supported by USDA's ARS and CSRS and the U.S. Fish and Wildlife Service (FWS). CSRS has its own programs through land-grant universities. All these agencies have some analytical capability associated with pesticide monitoring in foods (FDA, FSIS) or pesticide research.

Following is a list of the principal responsibilities of each agency:

The Environmental Protection Agency

• registers pesticides under FIFRA;

• sets pesticide tolerance concentrations for individual commodities, including meat and poultry, and for processed foods (the tolerance concentration for individual commodities is established by EPA for raw produce at the farm gate; as produce is processed into finished foods, pesticide concentrations may either decrease, increase, or remain the same);

• serves as lead agency for enforcement;

• reviews manufacturers' registration data, including analytical methods; and

• conducts research on the environmental fate of residues.

The Food and Drug Administration

• enforces compliance with residue tolerance concentrations in food, except meat and poultry, and feed;

• monitors residues in domestic and imported food; and

• develops analytical methods for monitoring.

The Food Safety and Inspection Service (USDA)

• enforces residue tolerance concentrations in meat and poultry;

• monitors residues in meat and poultry;

• develops analytical methods for monitoring; and

• gathers information on the incidence and concentrations of pesticide residues in the food supply.

The Agricultural Marketing Service (USDA)

• enforces compliance with residue tolerance concentrations and monitors residues in raw egg products.

The Agricultural Research Service (USDA)

• conducts research on pesticides, including efficacy and residue fate, and

• conducts ARS segment of IR-4 and National Agricultural Pesticide Impact Assessment Program (NAPIAP) operations.

The Cooperative State Research Service (USDA)

• oversees and funds NAPIAP and IR-4, both of which operate at land-grant universities.

The Fish and Wildlife Service

• monitors pesticides in fish and wildlife.

The accomplishments and shortcomings of all programs in sampling and analyzing pesticide residues in foods were reviewed by the Office of Technology Assessment (OTA, 1988). Much of the following description is based on that report.

EPA does not monitor pesticide residues in food. The agency's residue chemistry section within the Office of Pesticide Program's Registration Division reviews registration data compiled by pesticide manufacturers, and its laboratories in Beltsville, Maryland, and Bay St. Louis, Mississippi, may test the methods with spiked samples. Acceptable methods are submitted for publication in PAM II. EPA has for some time been studying the feasibility of combining the wide array of data bases in existence to maximize their utility for scientific and regulatory purposes. In 1989, for example, the agency contracted with Dynamac Corporation to compile and summarize residue data obtained from Agriculture Canada, FDA's state monitoring program, and the National Food Processors Association (NFPA). Together these sources provided data on 286 pesticides in an estimated 49,857 samples. In a report prepared for the EPA, Dynamac (1989) noted the difficulties encountered in attempts to compare these data bases, especially the differences in information reported and sampling methods. It made three fundamental recommendations intended to improve the utility of the data with minimum cost increases: that standard residue sampling protocols be used by state and federal agencies to facilitate comparison, that certain minimal information be provided with each sample (for example, identification of sample, purpose of sampling, and analytical method used), and that a standard data coding system and data base format be used.

EPA recognizes the need for uniform record keeping, sampling, and analytical methods in order to determine exposure and assess risk, especially for infants and children. It is important to this effort that EPA information be made compatible with FDA data. The utility of the data for estimating exposure and risk varies with the intended purpose of the monitoring programs. At present, EPA is evaluating the feasibility of drawing on the many and varied food intake and dietary exposure data bases to improve assessments of total human exposure. This information-gathering activity is one component of a larger effort to design a national human exposure survey that, among other things, will measure the route, magnitude, duration, and frequency of human exposure to environmental chemicals.

In May 1988, FDA's MRMs included 316 pesticides for which tolerance levels had been set, 74 pesticides with temporary and pending tolerances, 56 pesticides with no EPA tolerance levels (those previously canceled or those used only in foreign countries), and 297 metabolites, impurities, inert ingredients, and other pesticide-associated chemicals (OTA, 1988). Of these, 35% to 40% were covered by five primary FDA MRMs: one liquid chromatographic method used primarily for N-methyl carbamate pesticides and four gas chromatographic methods—one for organophosphorus pesticides and metabolites, one for both polar and nonpolar pesticides involving a variety of selective detectors, and two for nonpolar (primarily organochlorine and organophosphorus) pesticides in fatty and nonfatty foods.


FIGURE 6-1 Structure of the FDA program to monitor pesticides in foods.

SOURCE: McMahon and Burke, 1987. Reprinted from the Journal of the AOAC, Volume 70, Number 6, pages 1072–1081, 1987. Copyright 1987 by AOAC International.

In its monitoring program in 1987, FDA analyzed approximately 15,000 commodity samples in its 16 laboratories (Figure 6-1). Most samples were collected at random; the remainder were taken from targeted food sources after a violation or suspected violation. Approximately 7,000 of the samples are domestic and 8,000 are imported.

Since the early 1960s, FDA has gathered its information on pesticide residues through its Total Diet Study (TDS), also called the Market Basket Study, in which the dietary intakes of pesticide residues (as well as some industrial chemicals, toxic substances, and essential minerals) are estimated for eight age and sex groups from infants to senior citizens. To accomplish this, FDA personnel purchase foods from local supermarkets or grocery stores four or five times per year in three cities in each of four different geographic regions of the United States. The cities are changed each year. Each market basket contains 234 food items intended to be representative of the diet of the U.S. population. The foods are prepared for consumption, e.g., by peeling bananas or making beef and vegetable stew, and then are analyzed for pesticide residues. The results, combined with food consumption data, provide a model of dietary intakes. Because of the limitation of the food intake data and residue monitoring methods, coverage is not complete. Human exposure to all pesticides cannot be estimated because some pesticides cannot be detected by the analytical methods used. It is even more difficult to derive estimates of human exposure for population subgroups.

FDA's program has been criticized as being too slow in terms of analyses, in need of better sampling and enforcement of imported foods (GAO, 1992), and too limited in the numbers and types of pesticides detected (GAO, 1986a,b,c). Despite these criticisms and the age of the data, this program proved to be an important source of information for the committee's purposes. If the TDS were improved, risk from exposure to dietary residues could be better assessed. Increased funding for TDS would almost certainly be required to improve it. Later in this chapter, the committee addresses how the sampling program might be restructured to provide the data necessary to estimate the exposures of infants and children.

FSIS in its National Residue Program annually analyzes approximately 50,000 samples for about 100 residues of pesticides, animal drugs, and environmental contaminants in meat, poultry, and raw egg products. About one-third of the samples are analyzed for pesticides. Most samples are collected at random by FSIS inspectors located at slaughterhouses. The results are reported in the Journal of the Association of Official Analytical Chemists and in a series of annual reports entitled ''Residues in Foods" issued by FSIS.

State Activities

Thirty-eight states monitor pesticide residues in food but vary widely in the number of samples they process and the purposes of their programs. Figure 6-2 shows the range of sampling activity for 10 states in 1987. California has the largest and oldest program, which is designed to monitor the major raw commodities produced in, or imported into, the state. Its purpose is to enforce tolerance levels for residues on both domestic and imported commodities. This program is administered by the California


FIGURE 6-2 The number of food samples analyzed for pesticide residues in monitoring programs conducted by 10 states in 1987. SOURCE: OTA, 1988.

Department of Food and Agriculture, which in 1987 analyzed 13,500 samples of fresh fruits, nuts, and vegetables (State of California, 1987). Although routine postharvest monitoring is the largest component of the California program, efforts also include considerable preharvest monitoring (for early detection and deterrence), focused monitoring (to determine levels of specific chemicals of primary health concern), and foods to be processed (samples destined for processing and taken up to the point of actual processing).

Florida's Department of Agriculture and Consumer Services began monitoring raw agricultural commodities for pesticide residues in 1960. That program is targeted to potential problem areas in contrast to the random sampling conducted in California.

Many states have much more extensive pesticide analysis programs than is apparent in Figure 6-2, since they incorporate nonfood pesticide monitoring such as programs for farm worker health and safety and for groundwater contamination. To coordinate pesticide residue data from states and the FDA, the Mississippi State Chemical Laboratory, with FDA funding, has developed two U.S. data collection and dissemination programs: FeedCon provides information on contaminants in animal feeds; FoodContam provides similar information for human foods. Participation in these programs is voluntary. At present, 21 states participate in FoodContam. There is an effort to enlist all states and agencies as data sources (Minyard et al., 1989).

Food Processing Industry Activities

The food processing industry has a special interest in pesticide residues in produce. If processed and packaged foods are found to contain illegal residues or residues with the potential to cause adverse health effects, entire lots may need to be recalled from distribution centers and even from grocer's shelves. Tolerance levels must be established for any residue that concentrates during food processing (e.g., milling, cooking, and dehydrating). The behavior of residues during processing is evaluated based on processing studies required of the pesticide manufacturer before registration. Nevertheless, continual surveillance is needed to detect any unanticipated behavior, such as the formation of a previously unrecognized metabolite (Elkins, 1989).

The National Food Processors Association (NFPA) represents nearly 600 companies, including most of the major food processing companies in the United States. Approximately 450 of these companies are involved in processing common foodstuffs. Since 1960 the food processing industry has used the NFPA Protective Screen Program—detailed recommendations for preventing illegal and unnecessary residues, published annually in the Almanac of the Canning, Freezing and Preserving Industry (see, e.g., Judge, 1992). The recommendations involve washing, blanching, and processing steps, and emphasize proper use of chemicals at the farm source. NFPA operates National Food Labs in Dublin, California, in Washington, D.C., and in Seattle, Washington, to which food processors may submit fresh and processed commodities. Many food manufacturers and trade associations maintain analytical programs and data bank resources to accomplish the same purpose wherever possible and generally use the same multiresidue procedures as FDA. For example, NFPA uses the Luke method (PAM 212-3) to analyze for chlorinated hydrocarbons, organophosphates, carbamates, and substituted areas.

In 1988 EPA enlisted the help of NFPA and other groups to determine the availability and quality of data on pesticide residues in foods. Data were collected from 16 companies, including 3 major baby food companies, on more than 86,000 samples. Results indicated that 81% of these had no detectable residues (E. Elkins, NFPA, personal commun., 1992). More than half of the samples (47,737) were classified as fruit. Vegetables (14,037), tomatoes (8,424), and meat (8,215) were the next highest categories.

Potatoes (964) were given a separate category. The balance of the samples (6,947) were combined under miscellaneous; these were mostly flour samples in which the pesticide of interest was ethylene dibromide (EDB). In the fruit category, 8,573 of the samples were fresh or juice concentrates that were analyzed for Alar from 1985 to the present. Of these, 8,056 were negative for Alar.

From 1988 to the present, NFPA has been building a new pesticide residue data base that currently contains almost 74,000 samples, 97.5% of which had residues below the limit of quantification (LOQ) of the method used (E. Elkins, NFPA, personal commun., 1992). In the current data base, only 20% of the samples are raw foods. The balance is processed. Data from NFPA members usually pertain to crops on which the pesticide is known to have been used, but other data are random.

The data have been obtained from the food industry, from NFPA, and from other sources. NFPA plans to add data from the FDA/USDA APHIS monitoring and expects to eventually have more than 100,000 samples.

The pesticide residues reported so far include aldicarb on bananas, potatoes, and oranges; benomyl on apples and tomatoes; guthion on tomatoes and peaches; methamidophos in tomatoes; malathion on wheat; and diazenon on apples and tomatoes. The concentrations reported were at or near the LOQ.

Private, for-profit certification programs serve as third parties to verify the retailer's claims regarding residues on their produce. Some of the programs also certify growers who comply with a full disclosure statement of chemical usage and allow sample testing of their produce at random. Some large farming operations and commodity groups are beginning to do their own monitoring to ensure acceptability of their produce.

Private Laboratories

Commercial analytical laboratories have flourished in recent years, especially in analyses for pesticides and other toxicants in foods, water, soil, and waste sites. These laboratories must undergo a rigorous EPA certification procedure if they are to analyze groundwater/drinking water contamination by the pollutants of priority interest to the agency. There has been no similar certification for food residue analyses by either the FDA or any other agency. However, some states (such as California) are now instituting laboratory certification programs that involve inspection, performance standards, and adherence to GLPs. As a result of these certification programs, the data generated by these laboratories are becoming more consistent than in the past and promise greater reliability and more standardized operating procedures in the future.


All analytical laboratories (government, academic, and private) performing work of a regulatory nature must comply with GLPs as set forth in the Federal Register. In practice, this requires written protocols for

• sample preservation;

• sample processing and extraction;

• sample preparation, including cleanup, concentration, and derivatization;

• sample determination, including instrument tune-up and operating parameters, and confirmatory measures; and

• data handling, including maintenance and interpretation of records.

Quality assurance/quality control (QA/QC) procedures specify that the method be validated by spiking-recovery experiments designed to assess each portion of the overall method from sample preservation through tabulating and interpreting numerical results. A QA officer must oversee the analyst and analysis. Records must be kept so that an outsider can reconstruct the analysis, including calculation of final results.

Because the GLPs and the quality assurance and control procedures are relatively recent developments, the quality and completeness of recent data may differ considerably from those produced from analyses done 3 or more years ago (Garner and Barge, 1988). In the past, for example, a zero was entered into the NFPA data base when no residue was detected, and the zero was entered into the NFPA data base when no residue was detected, and the zero was averaged with the positive values. Today, the term None Detected replaces the zero and the method's detection limit is stated. Furthermore, NFPA data were sometimes obtained by nonstandard methods (e.g., methods not in PAM) or without stated recoveries, detection limits, or quality assurance measures. At present, however, NFPA is requiring that each contributor of data complete a residue report form for each reported pesticide-food combination. Among the information required are the analytical method used, detection limits, quantitation limits, and recovery information. In addition, there has been no coordination with federal and state monitoring efforts but, as noted above, NFPA plans to include FDA and USDA in the future. Thus, NFPA's findings have added considerably to the data base on exposure to residues in food because the samples were "as-served," finished products, and the ongoing effort promises to be even more useful because of the new requirements.

MRMs used by individual agencies or laboratories may be modified, usually are not peer reviewed, and usually are not externally validated. Although this does not necessarily make them less reliable or of lesser quality than standard methods, the lack of external quality assurance makes it difficult to assess their reliability. Improvements can be seen as a growing number of analytical laboratories adopt GLP. Compliance data may have the best quality because they require more stringent methodological control.

In general, the lack of quality control procedures limits assessments of data quality. Confidence intervals cannot be assigned to monitoring results because of the general lack of sample replication, and degrees of uncertainty cannot be clearly defined.


Sources of Error

All residue determinations are subject to error caused by limitations in the training or skill of the analyst, in laboratory glassware and other equipment, in the reference standards used as the basis for quantitation, and in the instruments (chromatographs and spectrometers) used in the final determination. Error is also partially attributable to difficulty in identifying a relatively low response above a high (and variable) background noise from nonpesticide materials in the sample. As noted previously, total errors of 50% or more are not uncommon, even for standard procedures in the hands of experienced analysts. For example, a method that gives 80 ± 20% recovery (20% relative error and 20% relative standard deviation) yields a total error of 60%. In most cases, this is quite acceptable for laboratory validation of a new method. Such errors can be dealt with by running a larger number of replicate samples—a costly solution that is not often used.

Another source of error or uncertainty is the large, difficult-to-gauge variability due to sampling. This variability may be small in well-designed field experiments in which a chemical is applied by a calibrated sprayer to a uniform stand of grapes, or it may be unknown (but almost certainly large) when samples are taken from large shipping bins of grapes from several fields treated with different chemicals and applied by a variety of spray equipment. It is difficult to estimate the sampling error in random sampling of a field or lot, especially because of differences in the training of personnel and sampling protocols. A sampling error of 100% is probably conservative. When added to an analytical error of 50 to 100%, the overall uncertainty in a given analysis of a single commodity may be 200% or more. This is in addition to the inherent variability of residues due to uneven field applications of pesticides to the commodity of interest. Thus, a residue analysis that yields a 1-ppm level should be assigned a range of 0.3 to 3 ppm in the absence of any accompanying data relating to the actual error limits (Keith, 1983; Hance, 1989). This error tends to swamp the generally small variation in the data sets in which LOQ, 0, or 0.5 of LOQ are used to compute averages. This can be seen in Table 6-3, which shows the consequences of averaging fictitious residue data when nondetects are assumed to be at the LOQ, zero, or one-half the LOQ.


TABLE 6-3 Consequences of Averaging Fictitious Residue Data Sets in Which Nondetectsa Are Used at LOQ, 0, and 0.5 of LOQ

a Samples for which analyses show no residue detected at or above the LOQ.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 5:05 am

Part 2 of 3

Other Limitations

The limitations of the residue data derive also from the lack of consistency among methods used for sampling, analyzing, and processing residue data. The lack of commonality among the analytical methods used by agencies to monitor food for residues, in the number of chemicals included in the studies, especially metabolites and degradation products, or in the limits of detection of the methods impedes comparability of data and limits the utility of the data for exposure assessment. Thus the basic validity of the sample, including the extent to which it represents the population sampled, is frequently difficult to assess. This contributes to the uncertainty in the final residue report.

Federal, state, and industry groups differ considerably in their processing and reporting of data on residue levels. Residues that exceed established EPA tolerance levels are subject to confirmation and are included in the residue reports. Residues below the tolerance levels are usually reported but are not always confirmed. When a given method includes chemicals that are not present above the LOD, these absences are not always specified in the final residue report.

In calculating averages, some laboratories use only positive data (above the LOD), some include the LOD for nondetectables, and others enter a zero for nondetectables. The rigor with which positives and nondetectables are recorded, and positives are confirmed, varies from laboratory to laboratory. Thus it is difficult to judge the quality associated with individual residue results or averages. These are some of the reasons why it is not always possible to assess a given data set or to include its data in calculating average dietary exposures of the U.S. population.

Data are collected for different reasons and from clearly different populations. Random sampling of consumption data reflects overall estimates of dietary and pesticide intake. Surveillance or compliance samples are directed to problem areas suspected of violating tolerance levels and therefore involve intense appraisal of products to which the compounds have been applied.

The biases that exist in terms of the number (frequency) and types of samples collected are apparent. Negative observations are more frequent in a truly random sampling program. On the other hand, surveillance, targeted, or compliance sampling may provide primarily positive values. Residues are expected in these types of sampling because they are designed primarily to intercept violations of FIFRA and related federal and state codes. Identification, separation, and calculation of the discrete observation categories are essential, because data are drawn from clearly different populations. Yet many residue report compilations do not discriminate between the categories, or reasons, for sampling.

The results of many residue data sets are clearly skewed due to inadequate sampling plans or a heavy emphasis on pesticides or products that have high potential for violation or health risk. Sampling may be biased to seek positive results when application of the pesticides is known. Sampling intensity tends to decrease when the potential for a measurable residue does not exist, for example, when the pesticide is not used. This impedes assessment of residue exposure of children or infants, whose primary food items may not be sampled frequently enough to provide a broad data base of residue information.

A lack of detectable residues may be an important signal that the compound is not used because of a lack of approval, local practice, or lack of a need to control pests. Several factors may lead to uneven, or no, use of a pesticide, even when it has been granted regulatory clearance. Prominent among these are climatic conditions, agricultural practices, and economic exclusion. Because pesticide costs account for a relatively large portion of the total cost of growing crops, unneeded pesticides or excessive quantities of pesticides are rarely used and expensive products are replaced by less costly compounds when available. Furthermore, pesticide usage may be governed by recommendations issuing from a state university extension program or a licensed pest control advisor and by the preference of the grower.

Agricultural practices can also be influenced by regulatory matters, consumer attitude (e.g., toward Alar), and manufacturer decision (e.g., about benomyl). Approvals may have been terminated for a specific use of a compound. Examples include ethylenebisdithiocarbamates (EBDCs) and 1,3-dichloropropene (1,3-D). 1,3-D was suspended by only one state (California); its use was continued in others. Integrated pest management programs also favor some chemicals over others, or result in the increasing substitution of nonchemical control measures for chemical compounds.

When applied according to recommendation and practice, some pesticides are used to maintain soil conditions and are not absorbed by the plants. For example, when applied to soil according to specifications, aldicarb maintains soil activity and is not absorbed systemically by all produce. Some pesticides dissipate so rapidly that residues are usually not detectable at harvest.

Knowledge of the relationship between residue level and the amount actually consumed is incomplete. This stems from a generally inadequate understanding of residue behavior during the storage, processing, and preparation of foods as influenced by biological constituents, pH, temperature, moisture levels, and heat treatment. Data on residue behavior provided by registrants are not adequate. Extrapolation of the limited data to population subgroups may be subject to major sources of error.


Because more water is consumed per kilogram of body weight than any other item in the diet (see Chapter 5), it is an important medium to consider in assessing total dietary exposure. For the pesticides examined in the Nonoccupational Pesticide Exposure Study (Immirman and Schaum, 1990), "exposure [to pesticides] from drinking water appeared to be minimal." (See Chapter 7, section on "Nondietary Exposure to Pesticides.") Unfortunately, however, the contribution of pesticide residues in drinking water is difficult to assess because of the variety of water sources (surface water and groundwater; public and private water systems) and because of the geographic differences and seasonal variations in pesticide use and consumption patterns. Moreover, no single survey of pesticides in food commodities has included both surface and groundwater sources of drinking water. As a consequence, it is not yet possible to estimate with any degree of certainty all the variations that must be considered in assessing dietary exposure to pesticide residues in water used in the processing and preparation of foods. The data that have been produced are discussed in the following paragraphs.


Approximately 53% of the U.S. population (more than 97% in rural areas) draws its drinking water from groundwater sources (USGS, 1988), which supply 40.1% of the water in public systems (USGS, 1990). This source of water has been the subject of several studies.

Hallberg (1989) reported that residues of 39 pesticides and their degradation products have been detected in the groundwater of 34 states and Canadian provinces. The sources of these data ranged from controlled field studies to ongoing programs to monitor public water systems. The pesticides most frequently reported were mobile or volatile compounds used in soil treatments, such as aldicarb and its products, which were detected in 24 states from California to Maine. EDB was found in 12 states, 1,2-dichloropropane (1,2-D) in 7 states, and dibromochloropropane (DBCP) in 5 states. Also prominent were herbicides widely used in the humid regions of the corn belt. These included alachlor, atrazine and its products, cyanazine, dicamba, dinoseb, metolachlor, metribuzin, simazine, trifluralin, and 2,4-D. The most frequently detected pesticides were the triazine herbicides atrazine, cyanazine, and simazine.

Most of the herbicides found in groundwater in these studies are still widely applied (USDA, 1992b); however, many of the fumigants and nematicides are no longer in use. According to a study conducted for the EPA by Williams et al. (1988), registrations of 16 pesticides have been canceled or their use severely restricted, including the fumigants and nematicides 1,2-D, DBCP, and EDB.

In a compilation of data from groundwater monitoring studies conducted by pesticide registrants, universities, and government agencies, Williams and colleagues confirmed detections of 46 pesticides in the groundwater of 26 states resulting from normal agricultural use and 32 pesticides in 12 states attributed to point sources or pesticides misuses. Most frequently reported were atrazine (normal use, 13 states; point source, 7 states) and alachlor (normal use, 12 states; point source, 7 states). The median and maximum concentrations reported as a result of normal use were 0.90 ppb and 113 ppb for alachlor and 0.50 ppb and 40 ppb for atrazine.

In mid-1987, the Monsanto Company, manufacturer of alachlor, initiated the National Alachlor Well Water Survey (Holden and Graham, 1990). Its purpose was to estimate the proportion of private rural domestic wells that contained detectable residues of alachlor and other herbicides (atrazine, cyanazine, metolachlor, and simazine). The investigators used a three-stage, stratified, unequal probability selection procedure to obtain samples of 1,430 wells from the estimated 6 million private, rural, domestic wells located in the 89 counties in 45 states where alachlor was sold in 1986. These wells serve 6.5 million households consisting of 20 million people. The wells located in areas of highest alachlor use and in areas of groundwater vulnerability had a higher probability of selection. Most counties were located in the midwest, northeast, and southeast, where pesticides containing alachlor are used primarily to control annual grasses and certain broadleaf weeds in corn, soybeans, and peanuts.

The results indicated that 100,000 people in the sampled area are consuming water from wells with detectable concentrations of the compound. They also suggest that an estimated 36,000 people are exposed to minimum concentrations of 0.2 µg/liter and that 3,000 people are exposed to concentrations at or exceeding 2 µg/liter—the maximum contaminant level (MCL).

The investigators found that 12.95% of the wells contained detectable residue levels of herbicides, the five highest being atrazine in 11.68%, alachlor in 0.78%, metolachlor in 1.02%, cyanazine in 0.28%, and simazine in 1.6%. Not only was atrazine detected in the highest percentage of the wells, it also exceeded the proposed MCL level by the highest percentage (0.09%), compared with alachlor (0.02%) and simazine (0.01%). Metolachlor and cyanazine were not found in levels higher than the MCL (Holden et al., 1990).

In 1990 the EPA completed a 5-year National Survey of Pesticides in Drinking Water Wells, the first survey undertaken to estimate the frequency and occurrence with which pesticides and their degradation products as well as nitrate were detected in drinking water wells (EPA, 1990). The investigators sampled 1,349 drinking water wells for 126 pesticides and products as a statistical representation of the more than 10.5 million rural domestic wells and 94,600 wells operated by the 38,300 community water systems that use groundwater. Their findings indicate that 10.4% [6.8–14.1%, 95% CI (confidence interval)] of the community water system wells and 4.2% (2.3–6.2%, 95% CI) of the rural domestic wells contain more than one pesticide.

The pesticides most frequently detected were the degradation products of the herbicide 2,3,5,6-tetrachloro-1, 4-benzenedicarboxylic acid dimethyl ester (DCPA), which were found in 6.4% of the community wells sampled and in 2.5% of the rural domestic wells. Next highest was atrazine (1.7% of the community wells and 0.7% of the domestic wells). Simazine, prometon, and DBCP were also found in both sources of groundwater, but with considerably lower frequency. Other pesticides found at lower frequencies were hexachlorobenzene and dinoseb in community wells and EDB, γ-hexachlorocyclohexane (γ-HCH), ethylene thiourea (ETU), bentazon, and alachlor in domestic wells in the United States.

Reported levels of DCPA products ranged from 0.35 to 0.89 µg/liter in community systems and 0.22 to 0.63 µg/liter in domestic wells. Atrazine concentrations were 0.20 to 0.81 µg/liter and 0.18 to 1.04 µg/liter in community and domestic wells, respectively. The level of at least one pesticide was found to exceed the MCL or health advisory level (HAL) in 0.6% of the domestic wells and 0.8% of the community wells.

The EPA survey was designed to examine the relationships among contamination, groundwater vulnerability, and intensity of agriculture. Although the survey was stratified by patterns of pesticide use and groundwater vulnerability, the number of positive samples was generally too low to be considered representative of groundwater contamination or that could be used effectively by the committee in the estimation of exposure through groundwater.

Surface Water

Surface water contributes 59.9% of the water in public water systems (USGS, 1990) and supplies drinking water to approximately 47% of the U.S. population (USGS, 1988). The data on this source of the nation's water supply are even sparser than those for groundwater.

In addition to his study of groundwater, Hallberg (1989) compiled data on pesticide detections in raw and finished drinking water drawn from surface water supplies in Illinois, Iowa, Kansas, and Ohio. In most state samples, the detection rate for the herbicides alachlor, atrazine, cyanazine, metolachlor, and 2,4-D exceeded 67% in both raw and finished water. Less frequently detected were the herbicides butylate, dicamba, linuron, metribuzine, simazine, and trifluralin and the insecticides carbofuran and chlorpyrifos. Alachlor was detected in 54% of 334 raw water samples in Illinois, 67% of 15 raw water samples and 52% of 33 treated samples in Iowa, and 100% of 3 raw water and 4 treated samples in Kansas. Atrazine was found in raw water in 77% of the Illinois samples, in 93% of the Iowa samples, in 100% of the Kansas samples, and in lower percentages of the samples obtained in 12 other states (Hallberg, 1989).

Seasonal variations in pesticide concentrations in surface water are striking (Figure 6-3). Baker and Richards (1989) reported that time-weighted concentrations of atrazine, alachlor, and metolachlor peaked during the late spring and early summer months in the Maumee River, the Sandusky River, and Honey Creek—all in Ohio. Atrazine, alachlor, and metolachlor concentrations exceeded maximum contaminant levels and health advisory levels during that period but were well below them during the rest of the year (Figure 6-3). Similar seasonal patterns were noted in statewide observations of atrazine concentrations in Illinois (Good, 1988).

During the reporting period (1983 through 1988) in the Ohio study, the concentrations of the three pesticides also varied extensively from year to year. In the Sandusky River, for example, a time-weighted mean atrazine concentration of 7.81 µg/liter was reported in 1986 compared with 0.90 µg/liter in 1988. In general, the highest mean concentrations for all pesticides in all rivers were highest in 1986 and 1987, dropping in 1988 to levels lower than those recorded at the beginning of the observation period in 1983. The authors attributed these differences to variations in the timing, intensity, and amount of rainfall.


FIGURE 6-3 Monthly time-weighted mean concentrations of atrazine, alachlor, and metolachlor in 1983–1988 reported at the Maumee River, Sandusky River, and Honey Creek stations in Ohio.

SOURCE: Baker and Richards, 1989.

Thurman et al. (1991) reported similar results from a study of a much larger watershed encompassing most of the corn belt region (parts of South Dakota, Nebraska, Kansas, Minnesota, Iowa, Missouri, Wisconsin, Illinois, Michigan, Indiana, Kentucky, and Ohio). During May and June (the planting period), median concentrations of atrazine, alachlor, cyanazine, and metolachlor were 10 times higher than levels reported in March and April (before planting) and in October and November (after harvest).

The occurrence of herbicides (primarily molinate and thiobencarb) in California's Sacramento River has been monitored extensively. Peak concentrations, which occur during May and June, have declined over the past decade because of the initiation of water management programs designed to minimize herbicide runoff from rice fields (Department of Fish and Game, State of California, 1990).

The Importance of Water Data to Infants and Children

As demonstrated in Chapter 5, water is an important component of the diets of infants and children. Water consumed by itself, water added to infant formula, and water used in the preparation of foods may represent a significant source of pesticide exposure by ingestion. However, because of the limited information on pesticide residues in water and the lack of monitoring data on water intake by infants and children, quantitative risk estimates cannot be made at this time. The committee noted, however, that water residues tend to run in the low or sub-ppb levels when present, so that the contribution of waterborne residues to ingested food prepared by using water will generally be expected to be low, except in specific locations where water contamination is far above the U.S. average.


Infant formula is one of the most important processed foods fed to babies not breastfed because it is usually their sole source of nutrition during the first few months of life. Although pesticide application to some components of processed foods is likely to have occurred at some point (e.g., field applications to crops used as infant formula ingredients), measurements have consistently demonstrated that no pesticides are detected in finished infant formulas (Gelardi and Mountford, 1993). These invariably negative analytical findings are attributable to ingredient selection and processing procedures that reduce the potential for pesticide residues to appear in the finished product.

In preparing ingredients for use in infant formula, manufacturers use numerous separation and purification procedures and heat treatments that reduce pesticide residues in raw agricultural commodities (Swern, 1979; Pancoast and Junk, 1980; Considine and Considine, 1982a,b; Snyder and Kwon, 1987). Chemical and physical processes of refinement and purification include washing, solvent extraction, filtration (including carbon filtration), acidification, basic extractions, clarification (centrifugation), crystallization, deodorization, evaporation, spray drying, and heat treatments such as ultra-high temperature (UHT). Because of processing and the relatively low levels of the individual ingredients in finished products (e.g., soybean oil constitutes only 1.7% of some formulas), any likely pesticide residues in or on raw agricultural commodities are reduced below detectable limits.

Water is the principal ingredient (by weight and volume) of all liquid infant formulas. For example, ingredient water comprises approximately 87% of commercial ready-to-feed infant formulas. Ingredient water used in the processing of most infant formula is passed through activated carbon filtration columns. Analysis of influent and effluent for trihalomethanes (THMs) has shown the columns to be highly efficient at removing THMs from the water. THMs are among the most difficult compounds to remove from water by activated carbon filtration (McGuire and Suffer, 1983). Water treated in this manner is therefore considered by the manufacturers to be free of pesticide residues.

Infant formulas are broadly classified into two categories: those based on cow's milk and those based on soy protein. The manufacturing systems for producing the protein and carbohydrate ingredients used in the two formula types are quite different.

Infant Formula Based on Cow's Milk

The principal ingredients of infant formula based on cow's milk include cow's milk solids (after milk fat is removed), lactose (derived from cow's milk), and a combination of fats to provide an optimal lipid source for infants. In some cases, whey proteins derived from cow's milk are also a part of the formulation. The composition of a typical infant formula based on cow's milk is shown in Figure 6-4.

The effects of processing (e.g., fat removal, isolation of lactose, isolation of whey proteins) on any potential residues of pesticides of interest can be logically postulated from the chemistry of the pesticides and the effects of various stages of processing. The production of lactose and whey protein concentrate is illustrated in Figure 6-5. Figure 6-6 illustrates the production of condensed skim milk and nonfat dry milk from raw cow's milk. Temperature extremes and highly effective purification processes, such as crystallization, can be expected to reduce any pesticide residues (Buchel, 1983; Hartley and Kidd, 1983; Hayes, 1975).


FIGURE 6-4 The composition of a typical infant formula based on cow's milk.

SOURCE: Infant Formula Council, Atlanta, Georgia, unpublished.

The likely effect of processing potential pesticide residues in raw cow's milk and the contribution of ingredients derived from cow's milk is illustrated in the following example. Assume that chlorpyrifos is present at its tolerance level of 0.020 ppm in raw cow's milk (FDA, 1989). Processing of milk to lactose could be expected to reduce that residue to 0.0002 ppm in the lactose (R.C. Gelardi, Infant Formula Council, personal commun., 1990). This is based on the assumption that 99% of any residue present would remain in the curd along with the lipids during the processing of milk into lactose (see Figure 6-5) because of the hydrophobic nature of chlorpyrifos (79% solubility in isooctane compared with 0.0002% in water). Similarly, processing is likely to reduce the residue in raw milk to 0.0005 ppm in condensed skim milk, assuming that 99% of any residue present would remain in the cream along with the lipids during the processing of milk into condensed skim milk (see Figure 6-6) because of the hydrophobic nature of chlorpyrifos (R.C. Gelardi, Infant Formula Council, personal commun., 1990). A concentration factor of 2.5 is assumed for finished condensed skim milk, resulting in a concentrating effect of the 1% of the residue assumed to reside in the skim milk fraction. Because infant formula contains 5.5% lactose and 3.5% condensed skim milk, the total possible theoretical contribution of these two milk-derived ingredients to the formula is 0.00003 ppm. This calculation is based on the preceding worst case assumptions and the percentages of lactose and condensed skim milk in a typical cow's milk-based infant formula, as noted in Figure 6-4. Thus, although condensed skim milk and lactose are major ingredients in infant formulas based on cow's milk, they are not expected to contribute greatly to potential pesticide residues.

The primary sources of lipids used in these formulas are soy oil and coconut oil. As with lactose and condensed skim milk, the contribution of these oil ingredients to pesticide residue levels in the finished product can be theoretically predicted. For example, the EPA tolerance concentration for malathion on soybeans is 8 ppm. Processing is expected to reduce the residue level in soybean oil to 4 ppm. And since infant formula based on cow's milk contains about 1.7% soybean oil, the theoretical maximum concentration of this pesticide in the soybean oil contained in the finished product would be 0.068 ppm.


FIGURE 6-5 Steps in the production of lactose and whey protein concentrate from raw cow's milk.

SOURCE: Infant Formula Council, Atlanta, Georgia, unpublished.


FIGURE 6-6 Steps in the production of condensed skim milk and nonfat dry milk from raw cow's milk.

SOURCE: Infant Formula Council, Atlanta, Georgia, unpublished.

Soy-Based Infant Formula

The composition of a typical soy-based infant formula is shown in Figure 6-7. The principal ingredients include soy protein isolate and soy oil (the predominant ingredients derived from soybeans) as well as corn syrup solids, sucrose, and coconut oil. Soy protein isolate is a specific protein fraction derived from soybeans. Following isolation, purification and modification are required to provide a protein source that will be nutritionally beneficial to infants. The soy protein isolation process involves several physical and chemical operations that effectively decrease any pesticide concentrations (Figure 6-8). Flaking and drying (heat treatment) are used to physically alter beans to an extractable form. Hexane extraction is used to separate the oil (lipid) fraction of the soybean flakes from the nonfat portions. Any lipophilic pesticide residues that have survived previous treatment can logically be expected to be carried into the solvent phase in this operation. Residual hexane is removed from the protein-containing solids by evaporation (heat treatment). The protein is then extracted from the solids with alkali (destructive to alkaline-sensitive residues), and the proteins are then precipitated by adjusting the pH to their isoelectric point. This step reduces acid-sensitive residues, and water-soluble survivors are partitioned into the aqueous supernatant.

The acid protein precipitate constitutes the basis for the soy protein isolate. This protein fraction is further processed by UHT treatment, neutralization, and spray drying, which result in the finished soy protein isolate. It is therefore not likely that significant residues on the raw agricultural product (soybean) would still be present in the finished product.

The following example illustrates the extent to which processing can reasonably be expected to diminish residues that have survived harvest and are present at the soybean processing plant. The tolerance concentration for malathion on soybeans is 8 ppm. Processing can be expected to reduce the level of malathion in soy protein isolate to 0.016 ppm, assuming that 30% of any initial malathion residue would be extracted into hexane and an additional fraction (approximately 50% of the original residue) would be decomposed in the alkaline extraction of protein from the defatted meal. Only 1% of the remaining residue (or 0.2% of the original residue of 8 ppm at the tolerance concentration) would be expected to precipitate with the soy protein isolate (R.C. Gelardi, Infant Formula Council, personal commun., 1990). Since soy protein isolate typically constitutes 2% of soy-based infant formulas, the potential maximum possible level of malathion (contributed by soy protein isolate) in the final product would be 0.0003 ppm.


FIGURE 6-7 The composition of a typical soy-based infant formula.

SOURCE: Infant Formula Council, Atlanta, Georgia, unpublished.


FIGURE 6-8 The physical and chemical operations involved in the isolation of soy protein.

SOURCE: Infant Formula Council, Atlanta, Georgia, unpublished.

Thus, the reductions of pesticide residues achieved by processing raw agricultural commodities combined with the relatively low concentrations of these ingredients in infant formulas account for the extremely low residue levels that can theoretically be predicted for infant formula, assuming maximum pesticide residue at EPA tolerances are initially present on the raw agricultural commodity. As stated earlier, extensive analytical testing of infant formula has failed to result in the detection of any pesticide residues.

Theoretical Maximum Residue Contributions

In numerous instances, a pesticide residue will be substantially diminished—or traces of the compound eliminated—as a result of processing steps such as evaporation, distillation, partitioning, isoelectric separation, or oxidation-reduction. Such changes have been observed during the isolation of soy protein and deodorization of oil (Swern, 1979; Snyder and Kwon, 1987). Enriched samples are used to evaluate residue losses that may occur during processing. Residue levels present in unenriched samples are typically below the limit of detection, since compounds are either not used on the food or are removed through processing. Theoretical maximum residue contributions in infant formula and limits of quantification are shown in Table 6-4.


TABLE 6-4 Theoretical Maximum Residue Contributions and Limits of Quantification

a Detection limits vary with the method used and the criteria for definition.

b Predicted potential maximum is below limit of quantification.

SOURCE: Data from Infant Formula Council, 1993.


Tissue concentrations of chlorinated organic pesticides are found in most people in the United States, although generally in declining levels in recent years as chlorinated pesticide use in the United States and worldwide has declined. The highest levels appear in regions where these pesticides are most commonly applied. Many of these lipid-soluble compounds are not cleared rapidly. Human milk is one route of elimination, but unfortunately that route increases the exposure of infants (Jensen, 1983; Wolff, 1983; Jensen and Slorach, 1990). In 1951, Laug et al., were the first to report an analysis of pesticides in human milk, in this case the presence of DDT and its metabolites. Since then, there have been many surveys of pesticides in human milk—some in response to episodes of food or dairy product contamination. The data from these surveys have been used to compare pesticide concentrations in human milk to allowable daily intakes.

In general, the more recent surveys of pesticides in human milk have demonstrated that the concentrations are lower than was observed in previous surveys. Despite this decrease in concentrations, more effort is needed to characterize the potential adverse effects of the low concentrations of chlorinated pesticides found in human milk. Initial attempts at estimating such effects associated with low concentrations have recently been described by Mattison (1990a,b, 1991) and Rogan et al. (1991).

DDT and Metabolites

DDT is an effective pesticide that has fairly low acute toxicity and has a long history of use worldwide (WHO, 1984; Murphy, 1986; Kurtz et al., 1989; Baker and Wilkinson, 1990). Concern over reproductive effects of DDT and its breakdown product DDE in birds and its long biological persistence led to the cessation of DDT use in the United States approximately 20 years ago (Hayes, 1975). Despite the two-decade ban, this pesticide and its metabolites continue to be found in human milk in the United States at decreasing concentrations over time, demonstrating the remarkable biological persistence of these halogenated organic compounds.

Surveys conducted before 1986 (Jensen, 1983; Wolff, 1983; Jensen and Slorach, 1990) demonstrated that p,p'-DDT or its metabolite p,p' -DDE were present in quantifiable concentrations in essentially all human milk samples assayed. Among samples analyzed in Arkansas in 1986, quantifiable concentrations of p,p'-DDT were found in 19%, and the metabolite p,p'-DDE was found in all samples surveyed (Mattison et al., 1992). The range of means of p,p'-DDT and p,p'-DDE reported in surveys of human milk in the United States before 1986 varied from 0.2 to 4.3 ppm and 1.2 to 14.7 ppm in milk fat, respectively. The mean concentrations among quantifiable samples from the survey in Arkansas were at the low end of these means. Among all samples, the mean concentrations were considerably lower than those previously reported.

The mean concentration of p,p'-DDT in all samples assayed in Arkansas was 0.039 ppm; the highest quantified level was 0.203 ppm (Mattison et al., 1992). In earlier surveys conducted in the United States, investigators found the mean concentration of p,p'-DDT to be between 0.2 and 4.3 ppm and p,p'-DDE to be between 1.2 and 14.7 ppm. Among all samples surveyed in Arkansas, the mean concentration of p,p'-DDE was 0.952 ppm. Among those with quantifiable concentrations, the mean was 0.954 ppm. This appears to show a continued decrease in DDT concentrations in human milk over the years.


Dieldrin is an oxygenated metabolite of aldrin that persists in adipose tissue. Because of its persistence and toxicity, the parent compound, aldrin, has generally been removed from use in developed countries (Murphy, 1986; Kurtz et al., 1989; Baker and Wilkinson, 1990). Previous surveys conducted in the United States have identified detectable levels in 0.04% to 100% of the human milk samples analyzed (Jensen, 1983; Jensen and Slorach, 1990). Mean dieldrin concentrations ranged from 0.05 ppm to 0.24 ppm in milk fat (Jensen, 1983; Jensen and Slorach, 1990). The mean concentration in the 2% of Arkansas samples with quantifiable levels was 0.071 ppm. The mean concentration among all samples was 0.001 ppm (Mattison et al., 1992).


Lindane is one component of a mixture of various isomers of hexachlorocyclohexane (HCH). The composition of the commercial pesticide is α-HCH or lindane (53–70%), γ-HCH (3–14%), µ-HCH or lindane (11–18%), δ-HCH (6–10%), and other isomers (3-10%) (Reuber, 1980). Lindane has been used as a substitute for DDT. In previous surveys conducted in the United States, HCH isomers were found in quantifiable concentrations in 4 to 68% of the human milk samples analyzed (Jensen, 1983; Wolff, 1983; Jensen and Slorach, 1990). α-HCH and γ-HCH, because of more rapid clearance, are generally found in fewer samples and in lower concentrations. This was also observed among the samples analyzed in Arkansas. However, because of its persistence, there are typically more samples with quantifiable levels of β-HCH and higher mean concentrations. For example, the β-HCH isomer was found in 27% of the human milk samples tested from Arkansas women (Mattison et al., 1992). Among those with quantifiable levels, the mean concentration was 0.12 ppm and among all samples, the concentration was 0.03 ppm. Lindane's agricultural uses in the United States have been virtually eliminated by changes in regulation over the past 20 years.


Hexachlorobenzene is a persistent chemical with a variety of sources, including its previous use as a pesticide and its presence as an impurity in several other pesticide formulations. This compound can disrupt porphyrin metabolism (Jensen, 1983; Jensen and Slorach, 1990). Fatal cases of infant poisoning have been reported from ingestion of highly contaminated human milk. Because of persistence and fat solubility, hexachlorobenzene has been detected in many surveys of adipose tissue and human milk (Jensen, 1983; Jensen and Slorach, 1990). Despite the fact that hexachlorobenzene has been detected in human adipose tissues in the United States, few studies have explored the presence of this chemical in human milk. Among previous surveys conducted in the United States, the mean concentration was 0.04 ppm in milk fat (range, 0.018 to 0.063). Among the 6% with quantifiable levels in Arkansas, the mean was 0.03 ppm, and among all samples, the mean concentration was 0.002 ppm; both are lower than earlier reports (Mattison et al., 1992). Hexachlorobenzene is no longer registered for agricultural use, and its occurrence as a formulation impurity in other registered products has been greatly curtailed.

Other Cyclodiene Pesticides

Heptachlor, chlordane, and their metabolites (heptachlor epoxide, oxychlordane, trans-nonachlor) are closely related cyclodiene pesticides (WHO, 1984; Murphy, 1986; Kurtz et al., 1989; Baker and Wilkinson, 1990). Surveys conducted in the United States have demonstrated that between 25% and 100% of human milk samples analyzed had quantifiable concentrations of heptachlor or heptachlor epoxide ranging from 0.035 to 0.13 ppm (Jensen, 1983; Jensen and Slorach, 1990). A somewhat greater proportion of samples (46% to 100%) had quantifiable concentrations of chlordane and oxychlordane (range, 0.05 to 0.12 ppm), perhaps reflecting frequent use as a termiticide in houses (WHO, 1984; Murphy, 1986; Kurtz et al., 1989; Baker and Wilkinson, 1990). Among samples surveyed in Arkansas, 5% had quantifiable concentrations of heptachlor (mean, 0.03 ppm) and 74% had quantifiable concentrations of heptachlor epoxide (mean, 0.06 ppm) (Mattison et al., 1992). Two percent of the samples in that study contained quantifiable concentrations of chlordane; 77% and 84% had quantifiable concentrations of trans-nonachlor and oxychlordane, respectively. The mean concentrations among quantifiable samples measured in Arkansas for trans-chlordane, cis-chlordane, and oxychlordane were 0.18, 0.15, and 0.06 ppm in milk fat respectively.

In most studies, heptachlor and/or heptachlor epoxide were detected in human adipose tissue samples. Curley et al. (1973) measured the concentration of chlorinated insecticides in fat samples from 241 Japanese people and found heptachlor epoxide concentrations that ranged from =0.01 ppm to 0.2 ppm (mean, 0.02 ppm). In Finland, Mussalo-Rauhamaa et al. (1984) demonstrated mean human adipose tissue concentrations of heptachlor epoxide as 2.7 µg/kg in males and 1.9 µg/kg in females. The difference in male and female mean concentrations may represent different volumes of adipose tissue or elimination via breastfeeding among the women studied.

In a study of approximately 1,500 women, Savage et al. (1981) explored regional differences in the pesticide content of human milk. In the southeast region of the United States, including Arkansas, 76% of the samples tested had detectable levels of heptachlor epoxide. The distribution of heptachlor epoxide concentrations in samples was also higher in that southeast region. Only 23% of the samples tested contained trace or undetectable concentrations. Half of the samples (52%) had heptachlor epoxide concentrations ranging from 0.001 ppm to 0.1 ppm. The remainder (approximately 25%) contained concentrations above 0.1 ppm. This was the highest of all regions in the United States. The mean concentration of heptachlor epoxide in these samples with detectable levels was 0.128 ± 0.209 ppm, also the highest mean level of all the regions surveyed in the United States.

Similar studies of human milk in Pennsylvania (Kroger, 1972) and Missouri (Jonsson et al., 1977) have demonstrated mean heptachlor epoxide concentrations of 0.16 ppm and 0.0027 ppm, respectively. Studies conducted in Hawaii (Takahashi et al., 1981; Takahashi and Parks, 1982) whose inhabitants have also been exposed to heptachlor and heptachlor epoxide in dairy products have demonstrated levels ranging from 0.001 ppm to 0.067 ppm (range, 0.036 ppm) among women on Oahu, and from 0.015 ppm to 0.052 ppm (mean, 0.031 ppm) among women on neighboring islands.

With the exception of endosulfan, which does not exhibit the persistence and bioconcentration characteristics of other chemicals in the group, virtually all agricultural uses of the cyclodiene pesticides have been eliminated or greatly restricted by regulatory actions over the past 20 years.

Risks to Infants from Pesticides in Human Milk

Breastfeeding has substantial benefit (Lawrence, 1989), including psychological, immunological, and general health promotion. In many countries, breastfeeding confers measurable benefit such as decreased rates of infectious disease and increased rates of growth and development. Despite the concentrations of pesticides found in human milk, no major studies have demonstrated that these pesticide concentrations have led to adverse health outcomes in the children exposed through breastfeeding. Therefore, although there is concern that exposure to pesticides in human milk may carry some potential for adverse health effects for the mother and for the nursing infant, it is important to recognize that the benefits of breastfeeding are clear and have been demonstrated. Furthermore, surveys conducted in the United States over the past four decades have shown that the number of samples with detectable pesticide concentrations have fallen—even when improved analytical methods with increased sensitivity have been used.


Recognizing the wide variation in sampling and analytical methods, size, and the overall design and objectives of existing residue testing programs, the committee requested residue data for its review from a variety of sources, including FDA, state regulatory agencies, the infant formula and baby food industries, the food processing industry, retail distributors, the agricultural chemical industry, and commodity associations. The most comprehensive and best recorded data were those collected through FDA's market basket sampling and analysis.

No single data bank provides ideal residue values. Residue analyses are complex, difficult to perform, and expensive. All data should be judged within this perspective. It is therefore important that samples be carefully identified and described, their processing and application history defined, and the capability and periods of analyses provided. Evaluation of the data bases provided to the committee emphasized the need for uniform sampling, collection, and reporting to reflect adequately the data's quality.

The FDA Surveillance Data

The FDA monitors pesticide residues in all food other than meat, milk, and eggs. The agency's monitoring program is not designed to determine dietary exposure to pesticides. Rather, its objective is to enforce compliance with EPA's tolerance levels. The usual point of sampling is the commercial warehouse, border crossings, points of importation, or some similar point after, but as close as possible to, the harvest. Table 6-5 shows the total number of samples and positive detections in FDA's surveillance program in 1988 and 1989. The ranges in both parameters are so wide that comparisons are very difficult. For example, dicrotophos samples were 100% positive, but the number sampled (15) is too small to be representative. For those pesticides with larger (>1,000) sample numbers, the highest percentages positive were found for benomyl (28.5%), EBDC (11.6%), and acephate (9.6%).

The committee reviewed FDA residue data on more than 100 different pesticides and selected a smaller subset for further analysis. The purpose was to isolate specific compounds for more extensive studies. Criteria for evaluation were


TABLE 6-5 Total of Samples and Positive Detections in FDA Residue Data

SOURCE: Based on FDA Surveillance Data, 1988–1989, unpublished.

• volume of use and potential exposure,

• route of exposure (food, water),

• type of residue (surface, systemic),

• toxic potency,

• toxicity of particular concern with regard to infants and children,

• general toxic end point (cancer, cholinesterase inhibition),

• ability to characterize the potential risks that could arise from metabolic and physiologic differences between infants and adults, and

• utility of the data in evaluating different risk assessment models.

The subset of pesticides considered further by the committee includes

• total organophosphates and carbamates (high volume of use; some are easily detected by commonly used MRMs; and a common acute toxicity end point is cholinesterase inhibition);

• EBDCs and their metabolites, e.g., ETU (EBDCs are used on foods such as apples, bananas, and other fruits and vegetables consumed in large amounts by infants and young children; ETU is a possible carcinogen);

• benomyl (systemic residue; used on fresh fruits, berries, vegetables, and in fungicide problem areas; high percentage of residue occurrence in FDA surveillance program);

• captan (used on foods such as apples, peaches, and pears, which are consumed in large amounts by infants and young children; potential reproductive effects; high percentage of residue occurrence in FDA surveillance program);

• aldicarb (most acutely toxic pesticide registered; systemic residue presents high probability for exposure, potential for water-based exposure because it leaches into groundwater; used on citrus fruits, sugar beets, peanuts, pecans, and a variety of field crops and ornamentals);

• atrazine (widely applied pesticide; used on grasses, grains, corn, macadamia nuts, and guava; detected in ground and surface water; putative hormonal effects); and

• alachlor (widely applied pesticide; used on potatoes, corn, beans, peas, peanuts, and cotton; detected in groundwater; carcinogenic effects in animals).

Aldicarb, benomyl, and organophosphates were ultimately selected for the examples presented in Chapter 7.

The committee compared FDA's sampling frequency and sample sizes with the 18 foods shown to be most eaten by nursing and nonnursing infants in the USDA's 1977–1978 Nationwide Food Consumption Survey (NFCS) (seeChapter 5). As shown in Table 6-6, of the 18 foods most consumed by nursing and nonnursing infants under 1 year old, only four—apples, peaches, peas, and green beans—are among the top 18 foods sampled by the FDA. This emphasizes the importance of certain foods in the diet and, hence, their potential for contributing pesticides to the diet. A more focused and specific monitoring system is indicated for targeting foods in infant diets.

FDA monitoring is focused on crops with a history of residues exceeding EPA tolerance levels, residues with no EPA-established tolerances, and crops harboring the greatest number of different pesticides. The committee could not determine if the high number of pesticides found on some frequently sampled foods was influenced by the larger sample size and, therefore, if larger sample sizes from other crops would lead to a finding of more pesticides on them also.

The committee recognized that the 1988–1989 FDA residue data were collected for compliance purposes and were therefore not appropriate for comparisons with other residue data sets and for risk calculation. Nevertheless, the FDA data were valuable in comparisons with the 1977–1978 NFCS data on food consumption and provided the committee with useful information on residues in the food supply.


TABLE 6-6 Comparison of 18 Foods Most Sampled by FDA and Those Most Consumed by Nursing and Nonnursing Infants

NOTE: Foods are expressed as raw agricultural commodities (RACs)

aSOURCE: Based on data from USDA, (See Tables 5-6 and 5-7 in Chapter 5).

bSOURCE: Based on FDA Surveillance Data, 1988–1989, unpublished.

c Foods both sampled and consumed the most by infants.

The committee then reviewed FDA's sample sizes and positive detections in the 18 foods most consumed by infants for aldicarb, benomyl, captan, chlorpyrifos, dimethoate, the EBDCs and their breakdown product ETU, and parathion-methyl (see Table 6-7). In general, the pesticides most frequently detected were benomyl, captan, chlorpyrifos, dimethoate, and EBDCs. Highest residues were found for EBDC (succulent peas), captan (fresh peaches), ETU (succulent peas), and captan (fresh apples). The pesticide with no positives was aldicarb. Parathion-methyl occurred at a very low frequency at very low levels. In just a few cases did maximum residues exceed tolerance, and in only one case (EBDC in succulent peas) was the maximum residue substantially above tolerance (23 compared with 7 ppm). In a few cases, residues were found for a chemical/ commodity for which no tolerance had been established; an example is dimethoate, which was found at very low frequency in fresh peaches, fresh bananas, and orange juice. The sources of occasional residues of dimethoate in food crops for which it has no tolerance was not known to the committee, although misuse or drift from neighboring fields is a possibility.

Number of FDA Samples

Adequate numbers of randomly selected samples are fundamental to developing good residue estimates for exposure and are essential to a statistically reliable basis for risk assessment. Because the FDA residue monitoring system is not designed to produce statistically reliable survey data on pesticide residues in food, sample sizes for individual foods are often small. This is true for several foods that are prominent in the diets of infants and children and for several pesticides with known toxicity.

FDA has recently to increase its focus on gathering information on pesticide residues in infant foods and other foods eaten by infants and children. The agency has summarized regulatory monitoring conducted in FYs 1985–1991 on eight fruit, fruit juice, and milk foods that are prominent in the diet of infants and children; incidence/level monitoring of approximately 900 prepared infant foods collected and analyzed in FY 1990–1991; and Total Diet Study (TDS) results for 33 different types of infant foods (strained/junior) collected and analyzed in FYs 1985–1991 (Yess et al., 1993). Infant foods analyzed under TDS and through the incidence/level monitoring showed that residue levels were well below EPA tolerances and that residue intakes are well below ADIs set by the Food and Agriculture Organization (FAO) and the World Health Organization (WHO). FDA's increased focus on residues in infant/ children foods, exemplified by information in the cited paper, is a positive development toward improving the quality and quantity of data available to ascertain dietary exposures of infants and children to pesticides.


TABLE 6-7 Seven Pesticides Sampled by FDA in the 18 Foods Most Consumed by Nursing and Nonnursing Infants

NOTE: NT, no tolerance level has been established by the EPA.

a Tolerance for milk is given in the absence of tolerances for specific components.

b EPA tolerance for Maneb, Zineb, and Ziram.

SOURCE: Based on FDA Surveillance Data, 1988–1989, unpublished.

Milk, apple juice, and orange juice rank high in the diets of infants but do not appear in the list of foods most often sampled (Table 6-6). In fact, during the 2-year period (1988–1989) covered by the FDA data , milk constituents had been sampled only for captan, dimethoate, parathion-methyl, and chlorpyrifos in sample numbers ranging from a high of 424 for chlorpyrifos in milk fat solids to 7 for dimethoate in milk nonfat and fat solids.

FDA sampled foods (commodities) as consumed. The agency did not sample milk sugar (lactose), soybean oil, or coconut oil, which are frequently components of infant food but are not consumed as individual commodities by infants and children. The FDA program would require restructuring in order to provide the data required to estimate exposure of infants and children for use in risk assessment.

Aldicarb, ETU, EBDCs, and benomyl cannot be detected by FDA's routine multiresidue methods; they require a single residue method. As shown in Table 6-7, the sample numbers for these pesticides are small (<25) for many of the foods most consumed by infants. These low numbers of samples can be of concern when pesticides are found on a high percentage of the samples and a large percentage of the crop acreage has been treated with the pesticides. Sample numbers higher than 25 are shown in Table 6-8 along with the percentage of positive detections in those samples. Overall, the FDA data appear to be adequate for the Committee's purposes, except for the low number of milk and juice samples. They are useful for comparison and discussion; however, the low number of samples does emphasize the need for careful selection and reporting. More complete and reliable information would be required for a comprehensive risk assessment.

Positive Detections

The ability of a survey to detect pesticide residues depends on at least three factors: the percentage of crop acreage treated with the pesticide, the sampling design of the survey, and the analytical limit of detection for the pesticide in that food. Other factors include the stability of the chemical; the time between pesticide application, harvest, and sampling; and the degree of postharvest processing. FDA laboratory staff that tests for pesticide residues in food have no prior knowledge of the pesticides applied to the crop being treated. The testing and sampling protocol is not designed to provide statistically valid sampling data.


TABLE 6-8 The Percentage of Positive Detections for Six Pesticides in Various Foods for Samples Larger than 25

SOURCE: From FDA Surveillance Data, 1988–1989, unpublished.

Nonetheless, thousands of data points are generated by FDA. For the pesticides in the infant diet examined by the committee, the data show that positive pesticide residue detections are clearly more common in fresh fruits and vegetables, specifically in apples, peaches, pears, bananas, peas, green beans, and carrots (see Table 6-7), than in other commodities. Closer examination reveals that positive detections were less than 10% in most crop-pesticide combinations with 2-year samples larger than 25 (Table 6-8). However, the percentage of positive detections for these fruits and vegetables varies by pesticide and crop. The range extends from 0.3% positive for captan on carrots and peas to 50% positive for benomyl on peaches.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 5:05 am

Part 3 of 3

Residue Levels

Ultimately, it is the levels of residues on foods that are of concern in estimating exposure and risk. FDA data indicate that most of the residues detected in food are well below the established EPA tolerance levels (Table 6-9). In general, values most dramatically below those tolerances are residues of pesticides long used, such as captan and the EBDC fungicides. Rather high tolerances were established for these chemicals in the late 1950s without the benefit of current data on the health effects of pesticide residues.

The variability and small numbers (<25) of many of the 2-year samples of the foods and pesticides of concern make it difficult to calculate residues with any certainty. When the samples were of sufficient size for calculation, the mean residues were always below the EPA tolerance level.

Following the pattern of positive detections, the infant-food crops with the highest mean residues are fresh fruit and vegetables—specifically peas, apples, peaches, pears, carrots, and green beans (Table 6-9). Sometimes, residues were found on foods for which no tolerance levels had been set—benomyl on peas, captan on beans, chlorpyrifos on carrots and peas, and dimethoate on peaches. Six of the maximum residues shown in Table 6-9 exceeded the EPA tolerance levels—EBDC on peas, chlorpyrifos on peaches and beans, and dimethoate on peas and beans. In all these cases, however, the mean concentrations were considerably below the established EPA tolerance level and therefore represented a very small number of the samples. The only chemical for which mean residues approached tolerance was EBDC on succulent peas, for which the mean (2.6 ppm) was only slightly less than half the tolerance (7.0 ppm). This demonstrates the inappropriate application of tolerances, which were intended for another purpose—not to provide a margin of safety for infants and children.

Residue Distribution

Understanding the distribution of residues in the food supply is a key factor for estimating exposure accurately. As indicated in the preceding discussions, residues are skewed toward a relatively small portion of the food supply. To better understand this distribution, the committee looked at frequency distributions for pesticide residues throughout the entire food supply and on individual crops.

Effects of Processing

Processing may exert no effect on pesticide residues in foods, or it may increase or decrease concentrations. Therefore, the effects of processing on pesticide residues in food must be considered in a critical evaluation of the dietary exposure of infants and children to pesticides. Earlier in this chapter, the committee discussed the effects of processing on pesticide residues on the ingredients used in infant formula.


TABLE 6-9 Six Foods Among the 18 Most Consumed by Infants and the Tolerances and Residue Levels of Six Pesticides Detected in Them

NOTE: NT, no tolerance level has been established by EPA.

SOURCE: Based on FDA Surveillance Data, 1988–1989, unpublished.

Elkins (1989) reported NFPA data on the effects of food processing operations on the residues of pesticides permitted on raw agricultural commodities. In most instances, washing by itself was shown to reduce residues, blanching reduced them even further, and the canning process led to even further decreases. These data indicated that in tomatoes and green beans subjected to each of these three steps, levels of malathion were reduced 99% and 94%, respectively, and carbaryl concentrations decreased 99% and 73%. Levels of parathion were reduced 66% in spinach and 10% in frozen broccoli. Elkins also pointed out that some processing activities can actually increase levels in certain instances. For example, levels of ETU were increased 94.5% in frozen turnip greens as a result of maneb degradation during cooking in a saucepan.

Elkins noted that the distribution of different pesticides in a product is also an important consideration. In tomatoes, for example, malathion tends to concentrate in the peel or waste, while carbaryl, a fairly polar compound, is easily removed by washing and does not tend to concentrate in waste material. Table 6-10 lists the residues of both pesticides in the washed and unwashed product, in the peeled tomato, and in waste material.

The committee reviewed unpublished data provided by the NFPA on pesticide residues found in foods used in processed baby foods. These foods, along with infant formula, comprise a large proportion of the infant's diet. Of the 6,580 samples tested in 1987, NFPA members found residues in 165 samples (2.5%) distributed among a total of 7 foods (Table 6-11). As shown in the table, the largest number of positive samples and the highest ratio of detected residue to the LOQ were noted for fresh pears. The ratio represents a comparison of the highest concentration of amitraz found in pears to the relatively low (1 ppb) LOQ. The highest concentration found—140 ppb—is considerably lower than the EPA tolerance of 3,000 ppb.


TABLE 6-10 Pesticide Concentrations in Washed and Unwashed Tomatoes, in Peeled Tomatoes, and in Waste Material

SOURCE: Elkins, 1989. Reprinted from Journal of the AOAC, Volume 72, Number 3, pages 533–535, 1989. Copyright 1989 by AOAC International.


TABLE 6-11 Number of Samples and Detections for Foods Used in Baby Food

SOURCE: National Food Processors Association, 1987, data unpublished.

A comprehensive study of the effects of processing on food residues is badly needed. This study should be undertaken by EPA, FDA, or USDA working through grants and contracts to university and private laboratories. The committee believes that the information presently available needs to be brought together in a review document and then supplemented with additional studies. However, this requirement is beyond the time and resources available to the committee. The committee therefore opted to provide some brief examples of the type of information that can be obtained on many of the foods consumed by infants and young children. In these examples, the committee applies basic and deliberately simplified explanations of several processes involved in food preparation and processing.

In addition to infant formulas, which have already been discussed, the most critical and highly consumed foods are apple-based processed products and cereal-based foods. Following are brief overviews of the processing factors that influence pesticide residues in these foods.

Apple-Based Foods

Apple-based foods constitute a substantial portion of foods consumed by infants and young children, as shown in Chapter 5. Knowledge of the form in which these products are consumed is important to the understanding of residue data. Virtually all the foods consumed by infants are processed, and most are manufactured by a limited number of processors, who exercise stringent controls. The processing of apples for applesauce (which forms the basis for many foods) and apple juice is specific, and the controls for the finished products are extensive.

Steps in processing of apples for use in foods for infants and children involve washing, blanching, peeling, pressing (for juice), finishing (removal of fibrous or indigestible material), and heating (sterilization). The washing process removes the exterior (nonsystemic) compounds, and is effective in the removal of many pesticides. Blanching is done with steam or hot water, primarily to inactivate enzyme systems and to prevent discoloration. It involves treatment at high temperature for a relatively short time. The skin is removed by abrasion or by peeling with a knife. A substantial portion of nonsystemic pesticides is concentrated at the surface of the apple or in the peel. The calix (or core) is removed by actually cutting or by removing seeds and fibrous material with a finisher. Physical pressure, usually accompanied by heat or enzyme treatment, is used to separate clear juice from cellulose, fibrous (pectin), and protein material to yield a clear, light-colored juice. Pressing is conducted through a filter press with paper filtration or by ultrafiltration. Substantial pesticide concentrations are removed with the fibrous or protein fractions of apple solids. Thus, estimates at the farm gate are not reflective of the residue content of foods that have undergone such processing steps.

Infant Cereal

Infant cereal is consumed by a large percentage of infants. The lack of large pesticide residues detected in these products probably reflects the extensive and unique processes to which they are subjected.

The basic ingredient of infant cereal is flour, which is the dehulled and fractionated grain (rice, oat, or barley) that has been milled. The flour is formed into a slurry, treated with a hydrolyzing enzyme (a-glucosidase), then heated to cook the cereal, destroy enzyme activity, and sterilize the food. The slurry is then dried on a steam-heated drum, which effectively distills off into steam any volatile material subject to partition.

The resulting product is a sterile, precooked, partially hydrolyzed cereal-based food. Clinical trials of usage have established the acceptance and digestibility of cereal for infant consumption. There has been no relationship established between retention of components in original grain to finished, processed infant cereal.


Data on residues on foods are collected by FDA, state agencies, the food industry, private organizations, and in university programs such as IR-4. Other government, industry, and academic sources were identified by the committee for specific categories such as water, infant formula, and human milk, which are particularly important in the diets of infants and children. The committee also reviewed the complex and varied methods for pesticide residue analysis and sampling performed by these groups.


• There is no comprehensive data source, derived from actual sampling, on pesticide residue levels in the major foods consumed by infants and children.

For example, of the foods (expressed as commodities) most consumed by nursing and nonnursing infants under 1 year of age, data showing the 18 foods most frequently tested for pesticide residues in the FDA Surveillance Program (Table 6-6) include only 4 of the 18 major foods consumed by this age group.

• Data on pesticide residues in foods are extensive, but are difficult to interpret because of variation in sample selection, analytical methods, and quality control procedures. The extensive data available from numerous testing programs for pesticide residues in food would be far more useful in profiling residues in the diet if they were presented in a more complete and coherent form.

Food samples analyzed for pesticide residues may have been selected for surveillance, compliance, or other purposes. Analytical methods, their limits of quantification, and their degrees of precision and accuracy may therefore differ among laboratories. Record-keeping practices in pesticide residue monitoring programs are generally not uniform and not well articulated.

• Many of the existing data on pesticide residues were generated for targeted compliance purposes. Although these data may be appropriate for enforcement, they have limited usefulness in generation- or population-based evaluations of actual exposure. For example, the sampling technique over represents suspected violators and does not adequately represent foods eaten in large quantities by infants and children.

• The limited data available suggest that pesticide residues are generally reduced by processing; however, more research is needed to define the direction and magnitude of the changes for specific pesticide-food combinations.

• Infant formulas and processed baby foods are routinely monitored to ascertain pesticide residue levels. Although sampling and analytical techniques lack the desirable degree of uniformity, residues were not generally detected in these products.

• Human milk is a food whose constituents are subject to wide variation, depending on the diet, medical history, and exposure of the mother. For some infants this may represent the primary route of exposure. Ongoing surveillance indicates that pesticide concentrations in human milk continue to decline over time, especially since organochlorine pesticide use in the United States has been reduced.

• Pesticide residues in water—both drinking water and water used in food preparation—have previously been largely overlooked in assessing dietary exposure of infants and children.


• A computerized data base for pesticide residue data collected by laboratories in the United States should be established.

If standardized reporting procedures were developed and adopted, pesticide residue data could be accumulated in a national data bank in a form accessible for future use.

• In future applications of residue data, consideration should be given to the development of a standardized reporting format for use by all laboratories involved in residue analyses. Since pesticide residue data are collected by a variety of laboratories using different methods for sampling and analyses, it would be desirable to maintain records of sample collection, analytical methods used, the basis of detection, and the precision and accuracy of the results obtained.

Reports of pesticide residue testing should indicate

• food commodity analyzed (and whether it is processed or unprocessed),

• Ranalytical method used,

• compounds tested including metabolites),

• quality assurance-quality control (QA-QC) notation, and

• limit of quantification (LOQ).

These reports should follow a standard format, should be timely and consistent, and should include not only the LOQ but also all negative and positive findings. The methods of reporting must also be consistent (e.g., using similar computer software).

• Food residue monitoring programs should target a special market basket survey designed around the diet of infants and children. The methods to be used in this survey should be validated using fortified samples circulated among the participating laboratories.

• Residue analysis methods need to be standardized in a timely manner through an independent review and validation process conducted by a government or professional organization.

• FDA, working with USDA, EPA, and state and other federal agencies, needs to create:

• a clearly explained sampling strategy that could be used to ascertain the representativeness of the results of food residue analyses;

• guidelines for those generating, processing, and using residue data to ensure that an explanation of LOQs and nondetectables are provided with all reports and are uniformly used in data analyses (e.g., in averaging);

• a residue data management system that will improve the quality, accessibility, and comparability of food residue data, including those generated by the commercial sector; and

• a repository of information on the fate of compounds during food processing and preparation.

• Laboratories performing pesticide residue analysis for regulatory purposes should participate in QA-QC programs, including regular quality control checks by an independent, external organization.


Baker, D.B., and R.P. Richards, 1989. Herbicide concentration patterns in rivers draining intensively cultivated farmlands of northwestern Ohio. Pp. 103–119 in Pesticides in Terrestrial and Aquatic Environments, D. Weigmann, ed. Blacksburg, Va.: Virginia Water Resources Research Center, Virginia Polytechnic Institute and State University.

Baker, S.R., and C.F. Wilkinson, eds. 1990. The Effect of Pesticides on Human Health. Advances in Modern Environmental Toxicology Ser.: Vol. XVIII. Princeton,N.J.: Princeton Scientific Publishing Co. 438 pp.

Buchel, K.H., ed. 1983. Chemistry of Pesticides. New York: John Wiley & Sons. 618 pp.

Considine, D.M., and G.D. Considine, eds. 1982a. Maize (corn) processing. P. 1132–1139 in Foods and Food Production Encyclopedia. New York: Van Nostrand Reinhold.

Considine, D.M., and G.D. Considine, eds. 1982b. Soybean processing. Pp. 1870–1883 in Foods and Food Production Encyclopedia. New York: Van Nostrand Reinhold.

Curley, A.V., W. Burse, R.W. Jennings, E.C. Villanueva, L. Tomatia, and K. Akazaki. 1973. Chlorinated hydrocarbon pesticides and related compounds in adipose tissue from people of Japan. Nature 242:338–340.

Department of Fish and Game, State of California. 1990. Hazard Assessment of the Rice Herbicides Molinated and Thiobencarb to Aquatic Organisms in the Sacramento River System. Environmental Services Division, Administrative Report 90-1. Sacramento: California Department of Fish and Game .

Dynamac. 1989. Anticipated Pesticide Residues in Food. Draft Final Report, Volume 1. Rockville, MD.: Dynamac Corp.

Elkins, E.R. 1989. Effect of commercial processing on pesticide residues in selected fruits and vegetables. J. Assoc. Off. Anal. Chem. 72(3):533–535.

EPA (U.S. Environmental Protection Agency). 1990. National Survey of Pesticides in Drinking Water Wells. Phase I. NTIS Doc. No. PB-91-125765. Springfield, Va.: National Technical Information Service.

FDA (Food and Drug Administration). 1989. Food and Drug Administration Pesticide Program. Residues in foods—1987. J. Assoc. Off. Anal. Chem. 72:133A–152A.

FDA (Food and Drug Administration). 1991. Food and Drug Administration Pesticide Program; Residue Monitoring 1991. Washington, D.C.: Food and Drug Administration.

GAO (General Accounting Office). 1986a. Pesticides: Better Sampling and Enforcement Needed on Imported Food. (GAO/RCED-86-219) Washington, D.C.: U.S. Government Printing Office.

GAO (General Accounting Office). 1986b. Pesticides: EPA's Formidable Task to Assess and Regulate their Risks. (GAO/RCED-86-125, Apr. 18) Washington, D.C.: U.S. Government Printing Office.

GAO (General Accounting Office). 1986c. Pesticides: Need to Enhance FDA's Ability to Protect the Public from Illegal Residues. GAO/RCED-87-7. Washington, D.C.: U.S. Government Printing Office.

GAO (General Accounting Office). 1992. Pesticides: Adulterated Imported Foods Are Reaching U.S. Grocery Shelves. Washington, D.C.: U.S. Government Printing Office.

Garnet, W.Y., and M.S. Barge. 1988. Good Laboratory Practices: An Agrochemical Perspective. ACS Symposium Series 369. Washington, D.C.: American Chemical Society.

Gelardi, R.C., and M.K. Mountford. 1993. Infant formulas: Evidence of the absence of pesticide residues. Regul. Toxicol. Pharmacol. 17:181–192.

Gianessi, L.P. 1986. A National Pesticide Usage Data Base. Washington, D.C.: Resources for the Future.

Good, G., 1988. IEPA surface water monitoring support programs for pesticides. Pp. 97–107 in Pesticides and Pest Management, D. Cavanaugh, ed. Springfield, Ill.: Illinois Department of Energy and Natural Resources.

Hallberg, G.R. 1989. Pesticide pollution of groundwater in the humid United States. Agric. Ecosyst. Environ. 26:299–368.

Hance, R.J. 1989. Accuracy and precision in pesticide analysis with reference to the EC ''water quality" directive . Pest. Outlook 1:23–25.

Hartley, D., and H. Kidd, eds. 1983. The Agrochemicals Handbook. Nottingham, England: Royal Society of Chemistry.

Hayes, W.J., ed. 1975. Toxicology of Pesticides. Baltimore, Md.: Williams & Wilkins.

Holden, L.R., and J.A. Graham. 1990. Project Summary for the National Alachlor Well Water Survey. Monsanto Final Report NSL-9633. St. Louis, Mo.: Monsanto.

Holden, L.R., J.A. Graham, and D.I. Gustafson. 1990. The National Alachlor Well Water Survey. Part C, Statistical Analysis. Monsanto Final Report MSL-9629. St. Louis, Mo.: Monsanto.

Immerman, F.W., and J.L. Schaum. 1990. Nonoccupational Pesticide Exposure Study (NOPES), U.S. Environmental Protection Agency Order No. PB-90-152 224/AS. Springfield, Va.: National Technical Information Service.

Infant Formula Council. 1993. Infant formula: Evidence of the absence of pesticide residues. Unpublished paper prepared for the Infant Formula Council, Atlanta, Ga.

Jensen, A.A. 1983. Chemical contaminants in human milk. Pp. 1–128 in Residue Reviews: Residue of Pesticides and Other Contaminants in the Total Environment, F.A. Gunther and J.D. Gunther, eds. New York: Springer-Verlag.

Jensen, A.A., and S.A. Slorach. 1990. Contaminants in Human Milk. Boca Raton, Fla.: CRC Press. 288 pp.

Jonsson, V., G.J.K. Liu, J. Armbruster, L.L. Kettelhut, and B. Drucker. 1977. Chlorohydrocarbon pesticide residues in human milk in greater St. Louis, Missouri, 1977. Am. J. Clin. Nutr. 30:1106–1109.

Judge, D.P., ed. 1992. The Almanac of the Canning, Freezing, Preserving Industries. Vol. 2. Westminster, Md.: Judge & Sons.

Keith, L.H. 1983. Principles of Environmental Analysis. Washington, D.C.: American Chemical Society.

Keith, L.H., ed. 1988. Principles of Environmental Sampling. Washington, D.C.: American Chemical Society.

Kroger, M. 1972. Insecticide residues in human milk. J. Pediat. 80:401–405.

Kurtz, P.J., R. Deskin, and R.M. Harrington. 1989. Pesticides. In Principles and Methods of Toxicology, 2nd ed., A.W. Hayes, ed. New York: Raven.

Laug, E.P., F.M. Kunze, and C.S. Prickett. 1951. Occurrence of DDT in human fat and milk. A.M.A. Arch. Indust. Hyg. 3:245–246.

Lawrence, R.A. 1989. Breastfeeding: A Guide for the Medical Profession, 3rd ed. St. Louis: C.V. Mosby. 652 pp.

Lykken, L. 1963. Important considerations in collecting and preparing crop samples for residue analysis. Residue Rev. 3:19–34.

Mattison, D.R. 1990a. Estimate of Cancer Risk from Halogenated Pesticides in Human Breast Milk: Arkansas 1986. Society for Risk Analysis Annual Meeting, October 7–10, New Orleans, La.

Mattison, D.R. 1990b. Pesticides in Human Breast Milk: Population Based Estimates of Excess Lifetime Cancer Risk. Society for Risk Analysis Annual Meeting, October 7–10, New Orleans, La.

Mattison, D.R. 1991. Pesticides in Human Breast Milk: Changing Patterns and Estimates of Cancer Risk. Washington, D.C.: Society for Gynecologic Investigation.

Mattison, D.R., J. Wohlleb, T. To, Y. Lamb, S. Faitak, M.A. Brewster, R.C. Walls, and S.G. Selevan. 1992. Pesticide concentrations in Arkansas breast milk. J. Ark. Med. Soc. 88(11):553–557

McGuire, M.J., and I.H. Suffer, eds. 1983. Treatment of water by granular activated carbon. In Advances in Chemistry—Series 202. Washington, D.C.: American Chemical Society.

McMahon, B.M., and J.A. Burke. 1987. Expanding and tracking the capabilities of pesticide multiresidue methodology used in the Food and Drug Administration's pesticide monitoring programs. J. Assoc. Off. Anal. Chem. 70:1072–1081.

Minyard, J.P., Jr., W.E. Roberts, and W.Y. Cobb. 1989. State programs for pesticide residues in foods. J. Assoc. Off. Anal. Chem. 72:525–533.

Murphy, S.D. 1986. Toxic effects of pesticides. Pp. 519–581 in Casarett and Doull's Toxicology: The Basic Science of Poisons, 3rd ed., C.D. Klaassen, M.O. Amdur, and J. Doull, eds. New York: Macmillan.

Mussalo-Rauhamaa, H., H. Pyysalo, and R. Moilanen. 1984. Influence of diet and other factors on the levels of organochlorine compounds in human adipose tissue in Finland . J. Toxicol. Environ. Health 13:689–704.

NACA (National Agricultural Chemicals Association). 1988. Guidelines for Conducting Agricultural Chemical Residue Field Trials in the USA. Washington, D.C.: National Agricultural Chemicals Association.

Osteen, C.D., and P.I. Szmedra. 1989. Agricultural use trends and policy issues. In Agricultural Economic Report No. 622. Resources and Technology Division, Economic Research Service. Washington, D.C.: U.S. Department of Agriculture.

OTA (Office of Technology Assessment). 1988. Pesticides in Food: Technologies for Detection. Washington, D.C.: U.S. Government Printing Office.

Pancoast, H.M., and W.R. Junk. 1980. Handbook of Sugars, 2nd Ed. Westport, Conn.: AVI Publishing.

Reuber, M.D. 1980. Carcinogenicity of benzene hexachloride and its isomers. J. Environ. Pathol. Toxicol. 4:355–372.

Rogan, W.J., P.J. Blanton, C.J. Portier, and E. Stallard. 1991. Should the presence of carcinogens in breast milk discourage breast feeding? Regul. Toxicol. Pharmacol. 13:228–240.

Savage, E.P., T.J. Keefe, J.D. Tessari, H.W. Wheeler, F.M. Applehans, E.A. Goes, and S.A. Ford 1981. National study of chlorinated hydrocarbon insecticide residues in human milk. Am. J. Epidemiol. 113:413–422.

Snyder, H.E., and T.W. Kwon. 1987. Soybean Utilization. New York: AVI Publishing.

State of California. 1981. Pesticide Use Report, Annual. Sacramento: California Department of Food and Agriculture.

State of California, 1987. Perticide Residue Annual Reports. Sacramento: State of California.

Swern, D., ed. 1979. Bailey's Industrial Oil and Fat Products. New York: John Wiley & Sons.

Takahashi, W., and L.H. Parks. 1982. Organochlorine pesticide residues in human tissues, Hawaii, 1968–1980. Hawaii Med. J. 41:P250–251.

Takahashi, W., D. Saidin, G. Takei, and L. Wong. 1981. Organochlorine pesticide residues in human milk in Hawaii, 1979–80. Bull. Environ. Contam. Toxicol. 27:506–511.

Thurman, E.M., D.A. Goolsby, M.T. Meyer, and D.W. Kolpin. 1991. Herbicides in surface waters of the midwestern United States: The effect of spring flush. Environ. Sci. Technol. 25:1794–1796.

USDA (U.S. Department of Agriculture). 1983. Nationwide Food Consumption Survey. Nutrient Intakes: Individuals in 48 States, Year 1977–78. Report No. I-1. Hyattsville, Md.: Consumer Nutrition Division, Human Nutrition Information Service.

USDA (U.S. Department of Agriculture). 1991. Agricultural Chemical Usage 1990. Vegetables Summary. Economic Research Service Ag Ch 1(91). Washington, D.C.: U.S. Department of Agriculture.

USDA (U.S. Department of Agriculture). 1992a. Agricultural Chemical Usage 1991. Fruits and Nuts Summary. Economic Research Service, Ag Ch 1(92). Washington, D.C.: U.S. Department of Agriculture.

USDA (U.S. Department of Agriculture). 1992b. Agricultural Chemical Usage 1991. Field Crops Summary Economic Research Service, Ag Ch 1(92). Washington, D.C.: U.S. Department of Agriculture.

USDA (U.S. Department of Agriculture). 1992c. Pesticide Use on 1991 Fall Potatoes. Pp. 26–28 in Agricultural Resources: Inputs Situation and Outlook Report. Economic Research Service, AR-25. Washington, D.C.: U.S. Department of Agriculture.

USGS (U.S. Geological Survey). 1988. National Water Summary 1986: Hydrologic Events and Ground-water Quality. U.S. Geological Survey Water Supply Paper 2325. Denver, Colo.: U.S. Government Printing Office.

USGS (U.S. Geological Survey). 1990. National Water Summary 1987: Hydrologic Events and Water Supply and Use. U.S. Geological Survey Water Supply Paper 2350. Denver, Colo.: U.S. Government Printing Office.

WHO (World Health Organization). 1984. Environmental Health Criteria 38, Heptachlor. Geneva: World Health Organization.

Williams, W.M., P.W. Holden, and D.W. Parsons. 1988. Pesticides in Ground Water Data Base, 1988. Interim Report. NTIS No. PB-89-164230. Springfield, Va.: National Technical Information Service.

Wolff, M.S. 1983. Occupationally derived chemicals in breast milk. Am. J. Ind. Med. 4:259–282.

Yess, N.J., E.L. Gunderson, and R.R. Roy. 1993. Food and Drug Administration monitoring of pesticide residues in infant foods and adult foods eaten by infants/children. J. Assoc. Off. Anal. Chem. Int. 76:492–507.

Zweig, G. 1970. The vanishing zero. The evolution of pesticide analyses. In the Essays of Toxicology, Vol. 2. New York: Academic.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 5:15 am

Part 1 of 3

7. Estimating Exposures

THE TWO PRECEDING CHAPTERS have reviewed data on the diets of infants and children (Chapter 5) and on pesticide residues in food (Chapter 6). This chapter addresses methods for estimating ingestion of pesticides by infants and children using the data from the preceding two chapters. Although nondietary sources of pesticide exposures such as air, soil, and consumer products are also considered, emphasis is placed on the ingestion of pesticide residues present on foods consumed by infants and children.

Dietary exposure to pesticides depends both on food consumption patterns (Chapter 5) and on residue levels on food (Chapter 6). Multiplying the average consumption of a particular food by the average residue of a particular pesticide on that food yields the average level of ingestion of that pesticide from that one food commodity:

Consumption x Residue = Dietary Exposure.

In reality, however, estimation of dietary exposure to pesticides is more complex than this simplified equation. Since many pesticides are used on a number of food crops, determination of the total exposure to a pesticide must be based on consumption data for all such foods. Also, it may be of interest to consider the total ingestion of different pesticides such as organophosphates and carbamates that fall within related classes and may pose similar risks to health.

The data presented in Chapter 5 indicate that food consumption levels vary both among and within individuals. This variation can be represented in terms of a distribution of food consumption, reflecting both high and low consumption levels, as well as the average level of consumption. Pesticide residue levels present in food will also vary, depending on several variables including application practices in different regions, time that has elapsed since application, degradation during transportation and storage of food, and the manner in which food is prepared by the consumer. Thus, both food consumption and pesticide residue data are characterized not by a single value but, rather, by a broad distribution reflecting high, low, and average values.

The variation in food consumption and residue data produces considerable variation in dietary exposure of pesticides by infants and children. This can be represented by a distribution of exposures across individuals within a particular age group. The distribution of dietary exposures is determined by the distribution of food consumption levels and the distribution of pesticide residues in food. If both the distribution of food consumption and the distribution of residue levels are known, statistical methods can be used to infer the distribution of dietary exposures. The process for combining different distributions into one distribution is termed convolution. The statistical convolution methods that can be used for this purpose are discussed later in this chapter.

Since ingestion of pesticides is dependent upon both food consumption and pesticide residue levels in food, it follows that the quality of dietary exposure data is determined by the quality of consumption and residue data. Although food consumption surveys such as the Nationwide Food Consumption Survey (NFCS) provide data on consumption patterns in the population at large, these surveys have generally not targeted infants and children. Hence, they included relatively small sample sizes within the age groups of primary interest for this report. One exception is the 1985–1986 Continuing Surveys of Food Intakes of Individuals (CSFII), which did focus on food consumption patterns of children.

Determination of the distribution of pesticide residues in foods consumed by infants and children is also difficult: only a fraction of all food consumed can be tested for the presence of pesticide residues. Many of the available residue data are based on surveillance studies that because of their focus on potential problem areas may overstate residue levels in the general food supply. The detection limit of residue monitoring methods can also impart uncertainty as to the residue levels actually present on food, especially when many residues are below the limit of detection and the detection limit is relatively high.

Recognizing these data limitations, the committee has included in this chapter several examples to illustrate possible approaches to estimating the distribution of dietary exposure to pesticides for infants and children. Each of these examples is designed to illustrate different aspects of exposure estimation, including the estimation of average daily exposures for use in chronic toxicity risk assessment and the estimation of peak exposures for evaluating acute toxic effects. Examples are included to illustrate how total exposure to pesticides used on more than one food crop can be estimated, and how exposures from different pesticides falling within the same toxicological class can be combined based on their relative toxicity.

Because of the limitations in the available consumption and residue data, it must be stressed that the purpose of the examples is to identify methods for estimating exposure and not to produce representative estimates of actual exposure. The particular compounds chosen for study were selected data were available to illustrate the approaches to exposure estimation considered by the committee. All results should be taken in the context of the limitations of the data as described in this and the previous two chapters. Application of these methods in a regulatory context will be possible only if adequate data on the distribution of both food consumption and pesticide residues in food can be obtained.

The first example deals with benomyl, a systematic fungicide that has not been permitted for postharvest use in the United States since 1989. Because of the chronic toxic effects of this compound (benomyl has been shown to cause malignant liver tumors in mice), the average daily ingestion of benomyl was considered to be most relevant for estimating long-term exposure. Note that although the focus is on the average daily ingestion by individuals over an extended period, the daily ingestion will vary from person to person, depending on their food consumption habits and the residues of benomyl in the foods consumed by each person. Since residue data were available for apples, grapes, oranges, peaches, and tomatoes, this example was used to illustrate the estimation of total exposure to a single pesticide from multiple food commodities.

Data on benomyl from different residue monitoring programs were available to the committee, permitting a comparison of exposure estimates based on different residue data. For example, field trial data derived from pesticide analysis in the manufacturer's laboratory (using a special method not adapted to multiresidue screening) usually show higher detection rates than those found by government agencies in random sampling of food shipments. Field trial data are useful only as estimates of maximum residue concentrations from field test plot trials at treatment levels proposed for registration purposes. Because field tests are generally conducted at the maximum pesticide use allowed in its registration, the residue concentrations are often higher than those found in random sampling. The results of field trials are generally used to establish farm tolerances and analytical methodology for purposes of registration. Further evaluation of field trial data is required in order to evaluate pesticide degradation following application.

The impact of residue data below the limit of quantification (LOQ), a concentration below which residues cannot be accurately measured, was also investigated in this example. For nondetectable residues, it is possible that the actual (unknown) residue could be as low as zero or as high as the LOQ itself. The limitation of data on actual residue concentrations below the LOQ imparts additional uncertainty about the level of exposure to infants and children.

Aldicarb is the subject of the second example. This acutely toxic pesticide exerts its effects by inhibiting cholinesterase enzymes in the nervous system. The example focuses on dietary exposure to aldicarb first from potatoes and bananas separately and then from potatoes and bananas combined. It serves to illustrate how estimates of exposure to a single pesticide found on more than one food can be derived. In contrast to benomyl, where average daily exposures are of interest, individual daily intakes are examined in this example because of the acute toxicity of aldicarb.

Part of the aldicarb residue data is derived from composite sampling, which may underestimate peak residues found in individual potatoes or bananas as a consequence of compositing prior to residue analysis. Composite samples are not very satisfactory in acute risk assessment for raw food commodities like potatoes and bananas. However, residue levels in processed foods can be estimated by using composite samples.

The third example addresses methods for estimating exposure to a class of pesticides inducing a common toxic effect. Specifically, the committee considered five organophosphate compounds used on different fruits and vegetables. All these compounds can inhibit plasma cholinesterase. A measure of total exposure to all five organophosphates is proposed based on their relative potencies.

Before these short examples are presented, there is a discussion of statistical methods for combining the distribution of food consumption with the distribution of residue levels in food to arrive at a distribution of dietary exposures based on the method of convolution. The chapter concludes with a brief summary of nondietary sources of exposure to pesticides.


Food Consumption Data

The most appropriate dietary exposure data for risk assessment depends on the nature of the adverse health effects of concern. In the absence of specific dose-response effects, the average level of exposure of an individual over a certain period provides a reasonable measure on which to base estimates of chronic toxic effects such as cancer. For acute toxic effects, peak exposures over shorter periods are more appropriate for risk assessment.

Average Levels of Consumption

The development of food consumption data for evaluating chronic toxicity requires careful consideration. In general, food consumption surveys yield data on the consumption of that food over all days for which data are available. The average daily consumption for children within a given age class is then obtained by averaging across all the individuals in the age class.

Estimating the average daily consumption of a particular food within a given class warrants some discussion. Since some foods will not be consumed at all by some individuals, estimates of average daily consumption based on all individuals in the sample will underestimate average consumption for the subpopulation of individuals who consume the food in question. For this reason, separate estimates of average daily consumption for "all children" and for "eaters only" are considered when estimating exposure. Average consumption levels for "eaters only'' are typically 2 to 3 times higher than those for "all children."

Because food consumption data are available for only a few days each year, the proportion of children falling into the eaters-only group is underestimated. This problem is accentuated if only a 24-hour recall or 24-hour food record is used. If food consumption data were available for every day of the year, more children who consume the food of interest on an infrequent basis would be included in the eaters-only group. Thus, since the eaters-only group omits some individuals whose consumption levels are low, the average food consumption for "eaters only" calculated in this way actually overestimates the average consumption for this group. This bias does not occur when information on food consumption is obtained through food frequency questionnaires rather than 24-hour recalls or 24-hour food diaries, since food frequency tables in principle accurately identify those individuals who consume the food at any time during a given year.

Scientists working with food consumption data have long recognized that consumption by a "typical" individual will not be representative of consumption by people who eat large amounts of a particular food. This has stimulated interest in examining the distribution of average daily consumption levels across individuals in order to estimate consumption by individuals who consistently consume greater quantities of the food of interest than the average. This distribution of average daily consumption across individuals can be used to estimate upper quantiles of consumption, such as the 90th, 95th, or 99th percentile. Reliable estimates of extreme percentiles can, however, be obtained only with relatively large sample sizes. Because the distribution of average daily intakes based on a sample of food consumption records for several days includes variability both between children and among days within children, this distribution will be subject to greater dispersion than would be the case if day-to-day variability were eliminated. (In the ideal case, this could be achieved by monitoring food consumption data over a full year or by using food frequency questionnaires.) The implication of such overdispersion is that upper percentiles of consumption will be overestimated.

Peak Levels of Consumption

Although the average level of individual exposure to pesticide residues in food is an important determinant of chronic toxicity, peak levels of exposure are more relevant for evaluating acute toxicity. Episodes of relatively high exposure occurring in a single day or even during a single meal may be more pertinent for acute risk assessment, depending on the toxic effect of interest.

The 1977–1978 NFCS provides information about food consumption during individual eating occasions for 3 different days. These data permit estimation of the total ingestion of a particular pesticide for each individual in the survey on each day. Using data for different individuals in the survey, one can estimate the distribution of person-days of consumption of specific foods. By combining this information with data on the distribution of pesticide residues in the food product or products of interest, it is then possible to estimate the number of person-days each year during which exposure to pesticides in the diet will exceed a critical level such as the reference dose (RfD), as defined in Chapter 8.

Although average levels of consumption and exposure will be reasonably well estimated with this approach, upper percentiles will be underestimated since food consumption data are available for only 3 of the 365 days in a year that are of interest. This is in contrast to the case for chronic risk assessment, where upper percentiles of exposure are likely to be overestimated.

Residue Monitoring

The point at which food samples are taken will influence the residue levels found. The highest residue levels generally occur immediately following application, and are reflected in field trial data. In samples taken for surveillance or compliance purposes, the residues will generally be higher than those in samples randomly drawn from the entire stock of a particular food commodity available for sale in a particular region of the country. Market basket surveys are based on a composite sample of a limited number of commonly consumed foods after they have been cooked or prepared for consumption in the usual manner. Although market basket surveys provide residue data under conditions designed to emulate foods as consumed, they are limited because they provide only composite sampling results on a few foods included in a typical meal.

Most analytical methods for measuring pesticide residues in food are subject to an LOQ below which residue levels cannot be accurately determined. Although improved analytical methods for testing for pesticide residues in food have made it possible to detect lower and lower residue levels, even the most sensitive techniques are subject to an LOQ. When residue levels below the LOQ are reported, it is not possible to determine whether the food contains no residue of the pesticide of interest or whether there is a residue present but at a lower level than can be detected with the analytical methods used.

This uncertainty about the actual residue level with residues below the LOQ confers uncertainty on the distribution of pesticide residues in food products and, subsequently, on the distribution of dietary exposure to pesticide residues by individuals consuming those foods. For example, consider a hypothetical distribution of residue levels based on the analysis of a number of food samples that may have been treated with a particular pesticide, as shown in Figure 7-1. The residues above the LOQ will generally follow a log-normal distribution. However, an appreciable proportion of the samples will produce results below the LOQ.

What can be inferred about residue levels in samples below the LOQ? The only certain inference is that the actual residue level lies between a lower limit of zero and an upper limit equal to the LOQ. (Even this upper bound may not be entirely correct, since analytical results near the LOQ will be subject to some degree of measurement error.) Because not all crops grown in the United States are treated with pesticides approved for use on those crops, it is possible that results below the LOQ may be entirely pesticide free.

Consider, for example, the data on the use of different pesticides approved for use on apples shown in Figure 7-2. The percentages of the U.S. apple crop treated with specific pesticides varies widely, ranging from a low of 1% for malathion to a high of 90% for azinphos-methyl. Thus, most apples will not contain residues of malathion and would produce residue levels below the LOQ when tested. It is also possible tests for azinophos could yield results generally below the LOQ if residues of this widely used pesticide were present at low but nondetectable levels.

The data on pesticide use in Figure 7-2 also reveal marked regional differences in pesticide usage patterns in different regions of the country.


FIGURE 7-1 Hypothetical distribution of residue levels with a log-normal distribution for residues greater than zero.

Captan, for example, is widely used on apples grown in the central, northeast, and northwest regions of the United States but is virtually unused in the western regions of the country. Variation between pesticide usage patterns in different countries also warrants consideration with regard to imported food products.

In the past, results below the LOQ have been handled in different ways. A simple resolution of this uncertainty is to assume that all the results below the LOQ contain no residue and to assign them a residue level of zero. This is an optimistic approach, since the possibility of small but undetectable residues in some or all such samples cannot be excluded. A conservative approach is to assume that all residue levels are present at the LOQ. Although this may provide an upper bound on undetectable residues, it is unlikely that all the samples for which no residue was detected actually contain residues equal to the LOQ. An intermediate approach is to assume all nondetectable residues are present at one-half the LOQ. Clearly, the lower the LOQ, the less difference there will be between these different approaches, and the less uncertainty the LOQ will confer on estimates of potential human exposures.

Combining Residue and Exposure Data

Variation in food consumption patterns and in levels of pesticide residues in food leads to variation in dietary exposure to pesticides among infants and children. This variation in the ingestion of pesticide residues is characterized by a distribution of exposures, reflecting high, low, and average exposure concentrations. Statistically, the distribution of exposures can be obtained by convoluting (i.e., combining) the distribution of food consumption with the distribution of pesticide residues in food (Feldman and Fox, 1991). Thus, once the food consumption and residue distributions have been determined, the distribution of dietary exposures can be calculated (Figure 7-3).


FIGURE 7-2 In the 1990 apple crop, percent of apple production treated with the following chemicals: azinphos-methyl, benomyl, captan, cabaryl, malathion, and EBDCs.


FIGURE 7-3 Convolution of food consumption distributions and residue distributions to produce dietary exposure distributions.

The technical basis of convoluting two distributions can be described briefly as follows. Let C denote the consumption of a particular food by an individual, R the residue level in that food, and e the corresponding dietary intake or exposure level. The level of consumption will vary from person to person in accordance with the cumulative distribution FC(c) with corresponding density ƒC=F1. Note that FC(c) denotes the proportionof the people in a given age group whose consumption C is less than a particular value c; the densities ƒC and ƒR reflect the relative frequency of different levels of consumption and residue, respectively, within the group. Letting FR denote the residue distribution with density ƒR=FR1, the distribution FE(e) of dietary intakes is defined by


assuming that consumption C and residues R statistically independent (Feldman and Fox, 1991, p. 349). This relationship provides the technical basis for combining the consumption distribution FC with the residue distribution FR to obtain the exposure distribution FE.

In practice, estimates of consumption and residue distributions are based on survey data and are represented as histograms based on the observe sample.(If different weights are attached to the survey observations, a weighted distribution should be used.) Computationally, these two distributions can then be convoluted simply by taking the result of each point from the consumption distribution and multiplying it by each point in the residue distribution; the distribution of dietary intakes is then defined by the distribution of these products.

This empirical approach to convolution will work well, provided that the number of observations used to obtain the consumption and residue distributions is not large. With large distributions, the computation burden can be reduced by working with a random sample of both the consumption and residue data. The Monte Carlo approach (i.e., random sampling) to convolution was used by the committee in those examples where the computational effort required to convolute the two distributions was found to be excessive. The form of Monte Carlo sampling used by the committee was simply a means of reducing the amount of computational time required for convolution by using the original consumption and exposure distributions; no artificial distributional assumptions were required to implement this technique. The Monte Carlo distribution of dietary exposures will converge to that based on the entire exposure distribution as the number of Monte Carlo samples increases. As the number of samples converge, the distributions become identical.

The convolution method can be extended to more complex situations such as the estimation of total exposure to a pesticide that may be present on more than one food commodity. In this case, a single point on the exposure distribution is estimated by randomly combining points from the consumption distributions for all foods of interest with points from the corresponding residue distributions (one for each food), and then summing the total exposure across all foods. This process is repeated to generate a distribution of total exposures from all foods combined.

Total exposure to pesticides within the same class can be estimated in a similar fashion using the relative potency values for those pesticides to express the intake in toxicity equivalence factors. This is illustrated in the example of organophosphate pesticides later in this chapter.


The Compound

Benomyl, or Benlate, is a systemic fungicide that was used in the United States from the time of its registration in 1972 until registration was voluntarily withdrawn for postharvest use by the manufacturer in 1989. Before then, it was the most widely used of the fungicides in the family of benzimidazole pesticides.

Benomyl is effective in preventing more than 190 fungal diseases, and it acts as a protective surface barrier while also penetrating the plant tissue to arrest infections. It was applied as a seed treatment, a transplant dip, and a foliage spray and was registered for use on more than 70 crops in 50 countries, including imported foods such as bananas and pineapples. In the United States, more than 100 EPA tolerances were established for benomyl in a variety of foods and feeds.

Benomyl has been shown to induce hepatocellular carcinomas, and combined hepatocellular neoplasms occurred in male and female mice treated with benomyl at all doses. In tests that included methyl-2-benzimidazole carbamate (MBC)—a metabolite of benomyl—investigators observed combined hepatocellular neoplasms in male mice and hepatocellular adenomas, carcinomas, and combined hepatocellular neoplasms in female mice (NRC, 1987). Because of its carcinogenic potential, exposure assessment for benomyl is based on the distribution of average ingestion levels for different individuals.

The Consumption Data

Data from USDA's 1985–1986 Continuing Survey of Food Intake of Individuals (CSFII) were found to be more suitable for use in this example than the data from the 1977–1978 NFCS: the CSFII is more current than the NFCS, and includes consumption data collected over 6 days at 2- month intervals over an entire year, as compared to the 3-consecutive-day sample in the 1977–1978 NFCS. The CSFII included 170 1-year-olds, 195 2-year-olds, 225 3-year-olds, 191 4-year-olds, and 209 5-year-olds.

Intake data were divided into person-day food intake, and consumption was then averaged for each individual in each age class across the days of the reporting period. For example, 170 average daily intake values were recorded for each food for each of the 1-year-old children surveyed. A probability distribution of exposures could then be constructed for each yearly age class.

The Residue Data

Five different sets of residue data on benomyl residues were reviewed by the committee for this example. They were

• results of field trials conducted by the manufacturer,

• results of a market basket survey conducted by the manufacturer,

• 1988–1989 compliance and surveillance data collected by the Food and Drug Administration (FDA),

• data provided by the food industry, and

• data from tests of raw food by a certification business operating in California.

These data were collected using different sample designs, sample sizes, and analytical methods. Table 7-1 compares the number of benomyl samples and the number of detections for the FDA surveillance data, the manufacturer's field trials and market basket surveys, the food industry's data, and data supplied by the certification business. Data on apples, grapes, oranges, peaches, and tomatoes are shown in Figure 7-4 for all but the certification business.

Estimation of Exposure

Exposure was estimated using each of the five sets of residue data reviewed by the committee separately. An individual child's exposure to benomyl from a particular food was estimated by multiplying the mean residue for that food by the average daily intake of that food. The exposures were then summed across up to 26 foods that 1-year-old children consume most to produce an average daily exposure estimate for each child. Note that different foods would be included with different residue data sets, depending on the availability of residue data for those foods. Finally, the distribution of average daily exposure from all foods combined across individuals was calculated.


TABLE 7-1 Number of Benomyl Samples and Detections for Selected Foods Based on Data from the FDA, a Pesticide Manufacturer, the Food industry, and a Certification Business


FIGURE 7-4 Number of benomyl samples and detections in apples, grapes, oranges, peaches, and tomatoes.

The committee did not adjust these estimates for the percentage of the crop acreage treated. That adjustment is customarily applied by the EPA to residue data from field trials, thereby substantially reducing estimated exposures. EPA also multiplies the number of samples with no detected residues by the percentage of crop treated and assumes that residues in those samples are at the LOQ while the remainder of the undetected residues are at zero, i.e., anticipated residues.

In the procedure described in this chapter it is assumed that all crops are treated with benomyl. The committee notes that this is an unlikely scenario; however, the purpose of this analysis is to illustrate the probability distribution approach to estimating exposure.

The Manufacturer's Field Trials

In 1989 the manufacturer submitted to the EPA a substantial amount of residue data obtained from field trials in support of the continued registration of benomyl. These data are useful because they included information on the application rate (i.e., frequency of application [%] and amount used) and on the residue levels detected in each sample of raw agricultural commodity. Furthermore, sample sizes for single raw commodities were often large enough to permit statistical analysis. Unfortunately, many processed foods were not sampled for residues, thus forcing the EPA to rely on assumptions about the fate of residues during processing.

Data generated by the manufacturer for benomyl residues on fruit products following processing are shown in table 7-2. Detected levels shown in those data are likely to be far higher than those in actual market basket data, due to the uneven use of the compound throughout the United States. In a nationwide market basket survey, fewer samples treated with lower amounts of benomyl would actually be found than in the manufacturer's field trials, which focused on crops known to be treated with benomyl.

As shown in Table 7-1, a total of 412 samples of eight unprocessed foods (apples, beans, grapes, nectarines, oranges, peaches, pears, and tomatoes) were tasted. Of these, 343 (83%) contained residues above the detection limit. Since neither apple juice nor orange juice were sampled, EPA must rely on the results of processing studies, such as those shown in Table 7-2, to determine the fate of residues in juices most consumed by young children.

TABLE 7-2 Changes in Benomyl Concentrations During Washing and Processing

Food / Food Form / Reduction, %

Apples / Washed / 13

-- / Juice / 69

-- / Applesause / 82

Peaches / Washed / 73

-- / Canned / 99

Bananas / Pulp / No detectable residue

SOURCE: Based on data from the pesticide manufacturer.

The committee conducted two separate analyses of these data based on two different assumptions: that all reported nondetections were actually zero (Figure 7-5) and that nondetections were really residues at the LOQs, which were provided for each sample (Figure 7-6). (The actual exposure is somewhere in between those shown in the two figures.) The exposure estimates were not greatly affected by either assumption, principally due to the relatively high number of samples containing detectable residues. Estimates of young children's benomyl exposure based on the manufacturer's field trial residue data are almost identical, regardless of whether a value of zero or the LOQ is used in exposure calculations when no residue is detected. Estimates based on the manufacturer's market basket data are also comparable, regardless of the value assigned to nondetectable residues.

That portion of Figures 7-5 and 7-6 displaying the manufacturer's field trial data was constructed by combining the individual consumption reports with the mean of field trial residues for each of 10 foods: apples, apple juice, oranges, orange juice, grapes, grape juice, peaches, pears, green beans, and tomatoes. The committee could therefore produce separate exposure estimates for each food for each child from 1 to 5 years of age. Each analysis was conducted under the assumptions that only those 10 foods were consumed in a child's diet and that the juices lost no benomyl during processing.

The Manufacturer's Market Basket Survey

The sample design of a market basket survey is important, since the results can be dramatically affected by regional patterns of pesticide use and food distribution. A good design can obviate the need to make complex assumptions regarding processing, percentage of crop treated, and food distribution effects.

A limited number of foods was surveyed in the manufacturer's market basket survey. A total of 143 samples of 7 foods, and no juices, were analyzed for a benomyl. Thirty-two (22%) of the samples contained residues at levels above the LOQ. This percentage is similar to that found by the food industry but approximately 50% lower than that detected by the more focused sampling design used by the certification business.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 5:15 am

Part 2 of 3

FDA Surveillance Data

The committee used only FDA surveillance data in assessing chronic exposures, and it did not estimate exposures for any food for which there were fewer than 20 samples. Although benomyl was registered for use on many foods, sample size exceeded 20 for only 10 of the 26 foods listed in Table 7-1. Of the total of 448 samples tested, 112 (25%) had residues that exceeded the LOQ.

FDA monitoring is focused on fresh rather than processed foods. Therefore, many of the processed foods often consumed by young children are never or seldom sampled, and the utility of small samples is limited in estimating exposures. As shown in Table 7-1, a number of foods sampled by other groups were not sampled at all by FDA in 1988 and 1989. Other weaknesses of FDA surveillance data are noted in Chapter 6 on pesticide residues.

The Food Industry

A food industry association provided a large amount of data collected from its member organizations. These data identified the food, the pesticide used, the residue level, and the LOQ of the analytical method used. Since the food industry used a variety of sampling and analytical methods, the representativeness of the data for the nation's food supply is uncertain.

Despite this uncertainty, these data are useful in illustrating the method proposed here for exposure estimation. The majority of the positive findings in this data set relate to apples (with 30 residues above the detection limit observed in 68 samples) and apple juice (with 16 of 30 samples showing positive).

A Certification Business

The committee obtained residue data from a certification business operating in California, a commercial organization that guarantees to grocery store owners and consumers that any residues in produce will be below a detectable level. These data are subject to certain limitations such as nonrepresentative localized sampling, unidentified analytical methods, and analysis of selected produce. Nonetheless, they are used in this example to provide a range of data for comparison purposes. As shown in Table 7-1, this organization tested 447 samples of which 193 (43%) had benomyl residues above the detection limit. This compares to 448 FDA samples with 112 (25%) positive detections and to 203 food industry samples with 42 (20.6%) detections over the LOQ. The difference could be explained by the certification business's focus on produce with a history of residues of concern. Moreover, although the certification business, FDA, and the food industry tested a similar number of foods (16, 17, and 18, respectively), the selection of foods varied. For example, the food industry detected residues in 16 (53.3%) of 30 samples of apple juice, whereas the certification business tested no apple juice samples. Similarly, the certification business detected residues in 18 (64.3%) of 28 plum samples, but plums were not sampled by the food industry.


FIGURE 7-5 Daily exposure of 1-to-5-year old children to benomyl in different combinations of foods, as shown by residue data from a certification business, FDA, the food industry, and the manufacturer (field trials and market baskets). Based on the assumption that the nondetects were equal to zero.


FIGURE 7-6 Daily exposure of 1- to 5-year-old children to benomyl in different combinations of foods, as shown by residue data from a certification business, FDA, the food industry, and the manufacturer (fields trials and market baskets). Based on the assumption that the nondetects were equal to the LOQ.


In this example, the committee examined multiple data sets reflecting concentrations of benomyl in several different foods. These data sets varied in different ways (Table 7-1). The manufacturer's field trial survey showed a much higher frequency of benomyl detections than found in the market basket survey, mainly because the study focused on crops known to be treated with benomyl. In the market place, fewer foods would actually have been treated with benomyl than in the field survey.

The effect of processing on benomyl concentrations in foods is shown in Table 7-2. These data indicate that substantial reductions in residues may occur because of processing. Figures 7-5 and 7-6 show the estimated exposure distributions for children 1 to 5 years of age to total benomyl residues for several common foods where residues below the LOQ were set at zero and at the LOQ, respectively.

The most important result of this analysis is that most children are exposed to relatively small concentrations of benomyl in their diets (less than 0.012 mg/kg bw/day), if they are exposed to any at all. Current exposures could be even less since registration of benomyl for postharvest use has been suspended. In any case, their exposures would be much below the current reference dose (RfD). In the committee's analysis, it is only the manufacturer's field trial data that suggest some children have larger exposures, and as noted above, these data would be inappropriate for this type of analysis. Both the manufacturer's market basket and the certification business's residue data show relatively small total residue exposures. Finally, Figures 7-5 and 7-6 show that assigning a value of zero or the LOQ to nondetectable residues had little effect on the overall outcome of the committee's calculations for benomyl. A principal reason for this finding is that a relatively high number of benomyl samples contained detectable residues. Certain impacts on exposure estimates would be seen for pesticides where a relatively low number of samples contained detectable residues.


The Compound

The evaluation of short-term peak exposures is illustrated using data on aldicarb residues in potatoes and bananas. Aldicarb is an acutely toxic pesticide whose use on potatoes and bananas was voluntarily suspended by the manufacturer in 1990 and 1992, respectively. It is an N-methyl carbamate that exerts its toxic effects by inhibiting the enzyme cholinesterase in the central and peripheral nervous system and at the neuromuscular junctions. Single oral doses of 25 µg/kg bw in humans produces approximately 50% inhibition of blood cholinesterase (NRC, 1986). Inhibition above 30% is usually of concern in humans.

Aldicarb is a systemic toxicant that is used primarily as an insecticide and nematocide. It is absorbed by the roots, stems, leaves, and fruits of plants. Aldicarb sulfoxide is a toxic metabolite that is distributed throughout the plant and degrades relatively slowly. Aldicarb-treated crops commonly eaten by children include potatoes, bananas, and citrus fruits. As demonstrated in Chapter 5, infants and children consume proportionately more of these foods than do adults with the exception that infants do not eat potatoes.

Acute Effects of Dietary Aldicarb Exposure

In 1970 Union Carbide gave three groups of four healthy male adult volunteers doses of aldicarb at concentrations of 25, 50, or 100 µg/kg bw. Subjects given the highest dose became acutely ill; one of those who received the lowest dose developed severe mood symptoms (i.e., anxiety reaction). Whole blood cholinesterase depression was observed in all the subjects. After reviewing the results of this study, the Safe Drinking Water Committee of the National Research Council estimated a no-observed-adverse-effect level (NOAEL) of 10 µg/kg bw/day (NRC, 1986). Applying a 10-fold uncertainty factor, the EPA established an RfD, formerly an acceptable daily intake (ADI), for aldicarb at 1.0 µg/kg bw/day.

Studies in the dog demonstrate depression of plasma cholinesterase at doses as low as 1 ppm (20 µg/kg bw/day). At doses of 50 µg/kg bw/day, there were statistically significant increases in diarrheal stools in both sexes, along with statistically significant increases in both plasma and RBC cholinesterase inhibition in males. Applying an uncertainty factor of 100 on the lowest dose level (20 µg/kg bw/day) to account for interspecies and intraspecies variation, EPA identified an RfD of 0.2 µg/kg bw/day. EPA's Scientific Advisory Panel recommended on November 6, 1992, that the RfD for aldicarb be reestablished at 1.0 µg/kg bw/day, consistent with the 1986 NRC recommendation (EPA), unpublished data, 1993).

In July 1985, severe acute illness was observed in more than 1,000 people in the western United States a few hours after they had eaten watermelons treated with aldicarb—a nonregistered (illegal) use. The symptoms included nausea, vomiting, diarrhea, muscle fasciculations, mood changes, and other symptoms of cholinergic poisoning. The most seriously ill person was a 62-year-old woman who had eaten approximately one-fourth of a watermelon later found to contain a 2.7-ppm concentration of aldicarb sulfoxide, which presented an estimated dose of 57 µg/kg bw. She required emergency room treatment and atropine to reverse the symptoms.

The Consumption Data

Data on consumption of bananas and potatoes by children between 12 and 24 months of age were obtained from the 1977–1978 NFCS. The mean, median, and 90th and 95th percentiles of the average daily consumption of bananas and potatoes are shown in Table 7-3. Of the 529 children surveyed, 157 did not eat either bananas or potatoes on any of the days during which the survey was conducted; of those that did, fewer children ate potatoes than bananas. Of the 1,831 person-days included in the survey, there were 1,077 days on which neither bananas nor potatoes were consumed. The distribution of daily consumption of bananas and potatoes by children between 12 and 24 months of age is shown in Table 7-4.


TABLE 7-3 Average Daily Consumption by Children Between 12 and 24 Months of Age

SOURCE: Based on data from the from the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977–1978.


TABLE 7-4 Daily Consumption by Children Between 12 and 24 Months of Age

SOURCE: Based on data from the from the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977–1978.

The Residue Data

Until the early 1980s, residue sampling focused on aldicarb rather than on aldicarb sulfoxide—its persistent and more toxic metabolite. Furthermore, only composite samples were used. That is, many individual samples of a single commodity were blended and the resulting mixture was analyzed. Because aldicarb is an acute toxicant, and the foods it contaminates are often eaten individually, EPA required a survey in which individual foods were examined separately. In this survey, concentrations higher than the EPA tolerance level were found in individual bananas and potatoes, although they had not been detected previously in blended samples.

Sampling revealed that the distribution of residues in individual potatoes from treated fields followed a log-normal distribution pattern, that is, most individual sample results were clustered at low concentrations. The highest single concentration, 8.7 ppm, was found in one potato. In the event that a 20-kg child consumed that one 200-g potato, cooked by itself in a microwave oven, the child could receive an exposure of 87 µg/kg bw—an acutely toxic level. This illustrates the potential problems associated with use of composite samples for evaluation of exposure to acute toxicants.

Banana trees have been treated with aldicarb since 1977. In 1991 composite and individual samples of bananas were analyzed in five strictly controlled field trials. Half the bananas from one field were found to contain aldicarb residues higher than the tolerance level of 0.3 ppm. If a 20-kg child were to eat the 170-g edible portion of a single banana at the highest level found, 3.14 ppm, the resulting dose would be 26 µg/kg bw—again a potentially toxic dose. Even at the 0.3-ppm tolerance level for aldicarb in bananas, that child would be exposed to approximately 3 µg/kg bw—a level well above the RfD. This does not take into account exposure to other cholinesterase inhibitors in the diet, including possible aldicarb residues in citrus fruit or potatoes.

Since pesticides are usually approved for use on more than one food, it is important to consider the total exposure to a particular pesticide from all dietary sources. The methods for assessing exposure to aldicarb in bananas can be extended to cover multiple foods. Although the committee used only two foods (bananas and potatoes) in its example, extension of the method to more than two foods is straightforward. In fact, the FDA data on aldicarb discussed in Chapter 6 failed to identify residue levels above the LOQ in the 350 samples tested. Consider the data on residues of aldicarb in bananas examined earlier, along with data on residues of aldicarb in potatoes given in Table 7-5. The former data are from the National Aldicarb Food Survey. The latter data are from a special survey conducted by the manufacturer of the compound in 1989. In the manufacturer's survey, representative samples were taken from 26 locations in the states of Washington, Oregon, California, Michigan, and Maine. Only three of the locations, Washington, Oregon, and Maine, had composite samples with residues above the LOQ. Residue data from these three states were also selected because they gave data for individual potatoes. Of the 294 reported residue values, 6 were below the LOQ. The mean aldicarb residue of 0.239 ppm in potatoes is much higher than the mean residue of 0.008 ppm in bananas. This difference is due largely to the use of field trial data for potatoes, which were obtained from crops known to have been recently treated with aldicarb.


TABLE 7-5 Residues of Aldicarb

SOURCE: on data from the 1987 National Aldicarb Food Survey and survey data from the manufacturer.

Effects of Assumptions Regarding Residues Below the LOQ

The implications of results below the LOQ for exposure estimation can be illustrated using data on the levels of aldicarb in bananas obtained from the 1987 National Aldicarb Food Survey. The composite samples tested in this survey were obtained from 225 groups averaging 12 bananas each—a total of 2,700 bananas. These samples were initially tested for the presence of aldicarb with an analytical method that had an LOQ of 0.01 ppm. If any composite sample was found to have a residue greater than 0.01 ppm, each banana in that group was analyzed individually. In this survey, residues over 0.01 ppm were detected in 27 of the 225 composite samples. The investigators then conducted separate tests on the 299 bananas that were available for testing out of the 302 bananas in the 27 samples. They found aldicarb concentrations above the LOQ in 255 of those bananas.


TABLE 7-6 Residues of Aldicarb in Bananas

SOURCE: Based on data from the 1987 National Aldicarb Food Survey.

For risk assessment purposes, let us assume that the remaining 2,442 bananas in the sample had residue levels below the LOQ. This underestimates actual residue levels because compositing masks any unusually high residue levels on individual bananas in a given batch. Let us also assume that individual bananas testing negative do not contain residues above the LOQ. Despite these approximations, it is instructive to examine the impact of assumptions regarding residues lower than the LOQ on estimation of dietary exposures to aldicarb from bananas.

Table 7-6 presents the mean, median, and upper 90th and 95th percentiles of aldicarb on the 2,697 bananas in the survey sample. The mean residue level obtained by assigning a value of 0.01 ppm to all residues below the LOQ is 0.017 ppm—slightly more than twice the value of 0.008 ppm obtained by assigning a value of 0 to nondetectable residues. The median value of 0.01 ppm obtained by substituting the LOQ for nondetectable residues is close to the corresponding mean residue. Assigning a value of 0 to the bananas with no detectable residues leads to a median residue of 0. Since less than 10% of the detections were above the LOQ, both the 90th and 95th percentiles of the residue distribution are unaffected by the value chosen for observations below the LOQ.

Estimating Dietary Exposure

Of the 529 children between 12 and 24 months of age in the 1977–1978 NFCS, only 321 reported eating bananas on any of the 3 days during which food consumption data were recorded. The mean daily consumption of bananas among all the children surveyed was 0.90 g/kg bw/day (Table 7-7). The mean consumption by the 321 children who ate bananas on at least one occasion during the survey was 1.47 g/kg bw/day. Since 61% of the children consumed bananas at least once, the upper 90th and 95th percentiles for the subgroup of eaters are only slightly higher than the corresponding consumption percentiles for the entire sample. The 90th and 95th percentiles will be overestimated since the distribution of average daily consumption contains variability between children and among days.


TABLE 7-7 Daily Consumption of Bananas by Children Between 12 and 24 Months of Age

SOURCE: Based on data from the 1987 National Aldicarb Food Survey and USDA's Nationwide Food Consumption Survey, 1977–1978.

The aldicarb residue distribution shown in Table 7-6 may be combined with the distribution of mean daily intake of bananas shown in Table 7-7 to estimate the distribution of mean daily intakes of aldicarb residues on bananas. Statistically, this is accomplished by convoluting the two distributions by pointwise multiplication of the residue and consumption distributions to obtain an estimate of the distribution of intakes.

The results of this calculation are summarized in Table 7-8. Separate estimates of intake are presented for the entire sample and for the subsample of children who ate bananas during the survey period. Separate estimates of intake are given for residue levels below the LOQ using the assumptions that the residues were either 0 or at the LOQ.

These estimates of mean intake in Table 7-8 are identical to those obtained simply by multiplying the mean consumption of bananas by the mean residue concentration (Table 7-9). As indicated in Table 7-9, however, multiplication of the upper 90th percentile of the residue and consumption distribution in this fashion does not yield the 90th percentile of intake based on the method of convolution (Table 7-8). The discrepancy between these two values is particularly large for the subgroup of banana eaters only with nondetectable residues assigned a value of 0. Thus, estimates of upper percentiles of ingestion should be based on the more accurate method of convolution. At the upper percentiles, estimates are higher than the true percentiles since the average consumption distribution incorporates day-to-day variability.


TABLE 7-8 Daily Intake of Aldicarb from Bananas for Children Between 12 and 24 Months of Age

SOURCE: Based on data from the 1987 National Aldicarb Food Survey and USDA's Nationwide Food Consumption Survey, 1977–1978.


TABLE 7-9 Methods for Estimating the Mean and 90th Percentiles of Aldicarb Intake

SOURCE: Based on data from the 1987 National Aldicarb Food Survey and the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977–1978.

This method is most appropriate for estimating the average daily ingestion of pesticide residues over an extended period. Although average daily ingestion is an appropriate measure of exposure for chronic risk assessment, a different approach is required for acute toxic effects caused by short-term exposure to relatively high levels of substances.

Assume that the total intake of a particular pesticide in a single day represents a good indicator of whether an acute toxic response will occur. In this event, we may examine the distribution of individual daily intakes in Table 7-10 rather than the distribution of average daily intakes shown in Table 7-7. In the present context,this corresponds to the distribution of individual daily intake of bananas for all days for which observations were recorded for all children in the survey. Convolution of this distribution with the aldicarb residue distribution provides an estimate of the distribution of the number of person-days in the sample associated with a given daily intake of aldicarb (Table 7-11). This distribution thus provides a basis for estimating the percentage of person-days during which exposure would exceed a health-based exposure standard, such as a reference dose based on toxicity studies. The upper percentiles will be underestimated since the food consumption data are available for only 3 to 4 days of the 365 days in the year. Although several methods for dealing with results below the detection limit of the analytical method were discussed previously, all nondetectable residues were assumed to be zero in this analysis for simplicity.


TABLE 7-10 Daily Consumption of Bananas by Children Between 12 and 24 Months of Age

SOURCE: Based on data from the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977–1978.


TABLE 7-11 Individual Daily Intake of Aldicarb from Bananas for Children Between 12 and 24 Months of Age

SOURCE: Based on data from the 1987 National Aldicarb Food Survey and the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977–1978.

The mean, median, and 90th and 95th percentiles of average daily intake and individual daily intake of aldicarb from bananas and potatoes alone and for bananas and potatoes combined are shown in Table 7-12 and 7-13 for children between 12 and 24 months of age. The distribution is dominated by the intake of potatoes. Figures 7-7 and 7-8 show the distribution of individual and average intakes of aldicarb from potatoes and bananas, separately and combined. Intake values greater than 0.8 g/kg bw/day represented a very small proportion and were therefore omitted from the figures.

The distribution of aldicarb intake from both bananas and potatoes resembles that of aldicarb intake from bananas or potatoes alone except for a slight shift toward nonzero values; the number of zeros has decreased from 66% of the values to 63%. This is to be expected since some children ate bananas but not potatoes or potatoes but not bananas.


TABLE 7-12 Average Daily Intake of Aldicarb for Children Between 12 and 24 Months of Age

SOURCE: Based on data from the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977–1978, the 1987 National Aldicarb Food Survey, and survey data from the pesticide manufacturer.


TABLE 7-13 Individual Daily Intake of Aldicarb for Children Between 12 and 24 Months of Age

SOURCE: Based on data from the U.S. Department of Agriculture's Nationwide Food Consumption Survey, 1977—1978, the 1987 National Aldicarb Food Survey, and survey data from the pesticide manufacturer.


FIGURE 7-7 Distribution of the average daily intake of aldicarb from bananas and potatoes, separately and combined. SOURCE: Based on data derived from USDA, 1983, the National Aldicarb Food Survey, and survey data from the manufacturer.


FIGURE 7-8 Distribution of individual daily intake of aldicarb from bananas and potatoes, separately and combined. SOURCE: Based on data derived from USDA, 1983, the National Aldicarb Food Survey, and survey data from the manufacturer.


Aldicarb was examined by the committee because it is an acutely toxic chemical that may potentially be found in several foods consumed by children and because there were also good sampling data for aldicarb residues in several commodities. This example could also be used to illustrate approaches to estimating residue concentrations in both individual foods and in multiple foods combined in a child's diet to provide total estimated residue.

Composite samples were used until the early 1980s for measuring residues in foods. Samples of a single commodity were blended, and the resulting mixture was analyzed. Because aldicarb is an acute toxicant, new approaches do not use blended samples. Rather, single commodities are analyzed. These new residue surveys have shown log-normal distribution patterns of aldicarb residues in commodities such as potatoes. As would be expected, most individual samples show residues clustered at lower concentrations or approaching zero.

The committee estimated exposure to aldicarb from both bananas and potatoes. The estimated distribution of aldicarb residues either from the single commodity potatoes or the combination of the two commodities shows exposure above the RfD of 1.0 µg/kg bw/day. In general, when these foods are eaten in the absence of other cholinesterase inhibitors, exposures would be much lower than those that would produce toxic effects. However, in the unlikely event that a single exposure occurred at the highest residue concentrations found in either bananas or potatoes, toxic effects could occur in a child.

Since the use of aldicarb on potatoes and bananas was voluntarily withdrawn by the manufacturer in 1990 and 1991, respectively, children are not presently at risk of cholinesterase depression from residues of aldicarb on these two foods (Debra Edwards, Chief, EPA Chemistry Branch, personal commun., 1993). Nonetheless, this case study illustrates the use of food consumption and residue distributions in estimating the number of person-days on which ingestion of an acutely toxic pesticide might exceed the RfD. As in the benomyl case, total ingestion from more than one food on which residues might be present was taken into account.

Multiple Exposure Assessment: Organophosphate Insecticides

Pesticide regulation in the United States has been focused on single chemicals rather than on combinations of compounds likely to appear as mixtures in the human diet. This practice can be attributed not only to the absence of data on the residues of multiple compounds that coexist on foods but also to the lack of methods for estimating simultaneous exposures to multiple chemicals, which cannot be accomplished merely by combining mean values (or other statistical summaries) of food intake and residue data. The regulatory process has therefore progressed on a chemical-by-chemical basis without consideration of possible additive and synergistic effects that could result from exposures to mixtures.

The committee developed a method for estimating exposure to multiple pesticides with a common toxic effect: in this case, inhibition of plasma cholinesterase (ChE). This method was used to determine how many children are likely to be exposed to unsafe levels of multiple pesticides with that common effect and to express the exposures in the most desirable form—person-day exposures—using actual individual daily consumption data and actual residue data.

More than 25 compounds that inhibit cholinesterase are permitted to exist as residues in foods. Although N-methyl carbamates inhibit cholinesterase, their mechanism of action is reversible and duration of action is shorter than for organophosphates. For purposes of simplicity, therefore, the committee selected five commonly used organophosphates (acephate, chlorpyrifos, dimethoate, disulfoton, and ethion) and used actual data on their presence on eight foods (apples, oranges, grapes, beans, tomatoes, lettuce, peaches, and peas) and three juices (apple, orange, and grape) to explore the development of methods for assessing exposure to multiple chemicals.

Criteria for choosing the five chemicals included the following:

• They must each exert the same adverse effect, in this case, blood plasma ChE inhibition.

• Credible estimates of the no-observed-effect level (NOEL) for ChE inhibition must exist for each chemical.

• The chemicals must be permitted as residues on several of the eight foods analyzed.

• FDA residue data must exist for the chemical-food groups selected.

The selection of foods for analysis was driven by the availability of data on residues and on the amount of each food consumed by 2-year-old children sampled in the USDA's 1977–1978 NFCS. An attempt was made to include foods that children consume most; however, it became apparent that residues were most common on other foods, such as peaches, which were therefore included in the analysis. The frequency distributions for the foods analyzed by the committee are presented in Figure 7-9A-K.

Estimating pesticide exposure in this way was considerably constrained by the absence of residue data for certain foods and compounds, especially processed foods such as juices whose type of processing could greatly influence pesticide residue levels. There are few available residue data for processed foods; thus, little is known about the effects of processing on pesticide residues.

For this analysis, the committee assumed that exposures to ChE-inhibiting compounds should be summed across foods and compounds that induce a similar type of ChE inhibition. Although exposure to a single compound may not exceed the RfD, concurrent exposures to numerous compounds could exceed a safe level because of the increased ChE inhibition. It was also assumed that the toxic potencies of diverse compounds can be standardized by developing estimates of relative potency in the manner described below.


FIGURE 7-9A-K Consumption distributions for (A) apples, (B) apple juice, (C) grapes, (D) grape juice, (E) green beans, (F) lettuce, (G) oranges, (H) orange juice, (I) peaches, (J) peas, and (K) tomatoes. SOURCE: USDA, 1983.


Cholinesterase Inhibition

Among pesticides, organophosphate and carbamate insecticides are the ChE-inhibiting pesticides of primary concern. These chemicals bind with cholinesterases and block their action in the hydrolysis of the acetylcholine (ACh) neurotransmitter. ACh is the principal neurotransmitter at neuromuscular junctions in the parasympathetic nervous system and in many regions of the central nervous system. High concentrations of ACh are also found in areas of the brain linked to higher cognitive functions such as learning and memory.

Organophosphate compounds, such as acephate, chlorpyrifos, dimethoate, disulfoton, and ethion, bind and phosphorylate the active site of ChE, thereby inactivating the enzyme. Carbamates, including aldicarb, lannate, methomyl, propoxur, and carbaryl, also interact with the acetylcholinesterase (AChE) receptor by reversible carbamylation of the seryl hydroxyl moiety at the active site of the enzyme (Murphy, 1986).

Some organophosphate and carbamate insecticides are acutely toxic and are frequently implicated in poisonings of humans. Exposures to high levels of AChE-inhibiting compounds may lead to severe cholinergic toxicity, with symptoms of headache, nausea and vomiting, cramps, weakness, blurred vision, pinhole pupils, chest tightness, muscle spasms, and coma (Ecobichon, 1991). Delayed neuropathy has also been associated with exposure to some organophosphorus esters (i.e., phosphate, phosphorate, and phosphoramidate esters), some of which have been used as insecticides (Amdur et al., 1991). Symptoms of acute organophosphate toxicity are difficult to recognize in the clinical setting for two major reasons: the complaints are nonspecific, and most physicians have limited familiarity with the signs and symptoms of pesticide poisoning.

Most ChE inhibitors degrade relatively rapidly in the environment and do not appear to accumulate or concentrate in the food chain (in contrast to organochlorine pesticides). In addition, these pesticides do not accumulate in the body, since they are rapidly biotransformed and excreted. Nevertheless, ChE inhibition can occur, producing signs and symptoms of poisoning after exposure to small repeated doses. Long-term effects of acute and subchronic exposures to pesticides have been reported. Some investigators have reported chronic, subtle neurologic sequelae to acute organophosphate poisoning (Savage et al., 1988). Epidemiological literature reported by the Office of Technology Assessment provides some evidence of delayed, persistent, or latent effects in humans. The literature includes case reports and studies of agricultural workers with and without histories of acute poisoning (OTA, 1990).

Organophosphates and carbamates may be converted in the environment or in vivo to form metabolites with toxicity potentially greater than that of the parent compounds. Synergism among organophosphate compounds, such as that demonstrated among malathion, O-ethyl O-p-nitrophenyl phenylphosphonothioate (EPN), and other organophosphates, may be an important variable to consider in assessing exposure to compound mixtures (NRC, 1977).

ChE inhibition is widely regarded as a good, general indicator of exposure to organophosphate pesticides except among people occupationally exposed over long periods who have developed persistently low active ChE levels. However, knowledge about this inhibitory effect is still incomplete. For example, the relationship between ChE inhibition and neurotoxicity has not been adequately demonstrated (EPA, 1988).

Neurotoxicity is defined by EPA as an adverse change in the structure or function of the nervous system following exposure to a chemical agent. The level at which ChE inhibition is associated with such changes is unclear (NRC, 1986). Furthermore, some investigators question the validity of measuring ChE inhibition in peripheral tissues, i.e., in plasma and blood, as a surrogate for measuring ChE inhibition of the central nervous system.

Further study is required to correlate ChE inhibition with identifiable changes in the central and peripheral nervous systems. Techniques for measuring neurotoxic effects include nerve conduction studies, sensory studies, evoked brain responses, electrocardiograms, and biochemical assays (NRC, 1992).
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Postby admin » Sun Mar 13, 2016 5:16 am

Part 3 of 3

Relative Potency of Organophosphates

The EPA guidelines for the study of mixtures containing dioxins and dibenzofurans consider the relative potency of different components of the mixture (EPA, 1987). This method permits the estimation of toxicity equivalence factors (TEFs) by comparing the toxicity of the compounds of interest to a standard defined as the most thoroughly tested compound. In the present study, the committee selected as a standard chlorpyrifos—a commonly used organophosphate insecticide. A TEF may be derived by comparing the no-observed-effect level (NOEL), or lowest-effect level (LEL), for any other chemical shown to produce the same type of ChE-inhibiting effect, to the NOEL (or LEL) for chlorpyrifos. The ratio of the chlorpyrifos NOEL to the NOEL of a different chemical (chemical X) provides an estimate of relative potency for chemical X and was used to adjust the laboratory-detected residue levels of the five chemicals of concern (Tables 7-14 and 7-15). The new values based on relative potency may be added to estimate cumulative exposure to chemicals believed to induce similar adverse effects, in this case ChE inhibition.

The committee assumed for this example that ChE recovers to a normal level in a 24-hour period. This may not be an appropriate assumption, however, because intake data are summed over all eating occasions for each day. Primary exposure on day 1 may occur during dinner, whereas primary exposure for day 2 may come at breakfast. Although both meals fall within a 24-hour window, they are presumed to be different 24-hour windows for the purposes of the present example.


TABLE 7-14 Estimating Toxicity Equivalence Using the LOAEL for Chlorpyrifos as Reference Standard

NOTE: NA, not applicable. NOAEL = no-observed-adverse-effect level.

aFrom Integrated Risk Information System (IRIS): EPA, July 1992.

Food Consumption Data

Data from USDA's 1977–1978 NFCS were used in the analysis. As mentioned above, consumption rates of eight foods (apples, oranges, grapes, beans, tomatoes, lettuce, peaches, and peas) and three juices (apple, orange, and grape) for 2-year-old children were selected for the analysis. There was a total of 1,831 person-days of data.

Residue Data

The committee used pooled FDA residue data from 1988 and 1989 compliance, surveillance, import, and domestic sampling. Residues for each chemical were converted to chlorpyrifos equivalents by multiplying each value by an equivalence ratio (chlorpyrifos LOAEL or NOEL/chemical X NOEL or LOAEL). For example, three chemicals allowed on a particular food item each have separate residue data sets consisting of individual sample results for each chemical detected on that food. Each of these data sets were summarized with regard to frequency and then for residue distribution.

To estimate cumulative exposure to the five organophosphate compounds, the committee adopted the assumption that the residue distributions for each compound are independent of one another. This approach may result in an overestimation of actual exposure, since there is likely to be some correlation among residue levels of the different compounds. In particular, substitution among chemicals would lead to scenarios in which all five compounds are never detected on the same sample.

The distribution of the cumulative exposure can be constructed by taking all possible combinations of chlorpyrifos equivalent values for the five chemicals and summing the values for each combination. This procedure will, however, yield large numbers of combinations and is thus impractical even on relatively large computers. A more practical alternative is to use strategic simulation in which the original shape of each component distribution is preserved. This is called a strategic simulation, since it is designed to reproduce the sample proportions for each subset of the original chemical distributions. The following procedure is used to create the distribution of cumulative exposure: a value is extracted randomly from each of the five residue distributions by using this strategic sampling technique; the resulting values are summed and the result recorded; and the procedure is repeated 5,000 times, creating 5,000 possible combinations across the five chemicals. The 5,000 summed residues form the distribution for cumulative exposure, which is expressed in chlorpyrifos equivalents.

This computerized simulation was conducted for each food, creating a single residue distribution that is the random summation of residue values strategically extracted from each of the distinct residue distributions. This final distribution of summed residue values is expressed in chlorpyrifos equivalents. Eight such distributions were created, one each for apples, oranges, grapes, beans, lettuce, tomatoes, peaches, and peas.

The committee used two assumptions for nondetected residues: (1) that they were zero and (2) that they were present at the LOQ. The LOQ data used by the committee in this exercise were provided by the FDA. Since the FDA does not record the LOQ for each sample tested, it estimated an average LOQ of 0.01 ppm for all chemicals and foods analyzed in this study and proposed that this value be used in the committee's analysis.

Exposure Analysis

The objective of this exposure analysis is to produce a distribution of possible person-day exposures based on the food consumption data for 2-year-old children, including 1,831 person-day intake values for eight foods and eight separate residue distributions representing cumulative exposure—one for each of the eight foods. Person-day exposures are estimated by applying the following method.

1. Intake of food 1 by person 1 on day 1 is multiplied by some randomly extracted value from the residue data set (specific to food 1). The result is stored as an exposure value.

2. The process is repeated for n foods, still for person 1 on day 1.

3. Steps 1 through 3 are repeated 5,000 times by using the strategic simulation to extract the residue data points from each summarized food-specific residue distribution of the residue data.

4. The 5,000 exposure values for person 1 on day 1 are stored and summarized as counts within exposure intervals.

5. Steps 1 through 5 are repeated for 1,831 person-days, producing 9,155,000 person-day exposure values, all expressed in chlorpyrifos equivalents.

6. The counts within exposure intervals are plotted as a frequency distribution.

7. The proportion of the sample falling above the RfD is estimated. The RfD for chlorpyrifos is 0.003 mg/kg bw/day.

A second exposure analysis was conducted to determine the sensitivity of exposure estimates to assumptions regarding the transfer of residues from raw to processed foods. Exposure was estimated for the same eight foods as above, and then for three juices (i.e., apple, orange, and grape).

The committee assumed that all residues in apples, grapes, and oranges were transferred unchanged to their juices, consistent with EPA practice (Peterson and Associates, Inc., 1992). This provides a maximum exposure estimate that is useful in the absence of statistically reliable data on the effects of processing.


The results of this analysis using five pesticides and eight selected unprocessed foods, excluding potent ChE inhibitors, and assuming that nondetectable residues are actually equal to zero are shown in Figure 7-10. (Results with nondetects set equal to the detection limit of 0.01 ppm are similar and are excluded for simplicity of presentation.) On the basis of these results, the RfD of 0.003 mg/kg bw/day (3 µg/kg bw/day) would be exceeded on approximately 1.3% of the person-days for children considered, representing approximately 120,000 of the 9.1 million person-days simulated (Figure 7-10).

Conclusions regarding the at-risk population are more difficult to reach. If one assumes that these simulated person-day exposure values are an accurate estimation of daily exposure for this population, then one must also assume that the consumption and residue data are also accurate representations. The committee does not believe that the data used do accurately represent the current status of those pesticide residues on foods because of the age of the consumption data, sample sizes, and the methods used by the FDA in interpreting residue values beneath an ''action" or legal tolerance level. However, these data sets are the best now available for a variety of chemicals and foods. (See Chapters 5 and 6 for an extensive examination of the limitations of the data.)


Figure 7-10 Exposure of 2-year-old children to organophosphate pesticides—1,831 2-year-old person-day food intake values. Foods: apples, oranges, grapes, beans, tomatoes, lettuce, peaches, and peas. Chemicals: acephate, chlorpyrifos, dimethoate, disulfoton, and ethion. Strategic simulation: 5,000 exposure values generated per person-day; cumulative distribution is summary of 9,115,000 simulated exposures.


FIGURE 7-11 Exposure of 2-year-old children to organophosphate pesticides, including fruit juices. Strategic simulation: 5,000 exposure values generated per person-day; cumulative distribution is summary of 9,115,000 simulated exposures.

Although only 1.3% of the estimated person-day exposures were above the RfD, 3,584 person-days were more than ten times above the RfD, and the maximum exposure was ten times the RfD. These are clearly low probability events within a population of 3.5 million 2-year-old children; even a one-on-a-million event would occur 3.5 times per day. Even though these estimates are limited by the poor quality of the residue sampling data, they identify both a potential concern and an appropriate methodology for estimating exposure in large populations.

The results of the analysis are shown in Figure 7-11. The primary finding of this analysis is a shift in the distribution to higher residue levels. The percentage of the sample above the RfD rose from 1.3% to 4.1%, primarily because of the large intakes of apple juice and orange juice among 2-year-old children.

The committee concluded that this method is a workable and useful mechanism for assessing exposure, i.e., standardizing residue values by toxicity equivalents, combining these values based on either allowable or actual detected combinations of residues, and simulating exposures by combining residue values with actual reported intake values and summing exposures across foods within person-days. As with the other examples presented in this report, however, this discussion should be regarded as an assessment of methodology rather than a specific attempt to characterize the proportion of children at risk. The method could also be used to study possible combinations of residues for any class of chemicals believed to have a common adverse effect, including cancer, where the end point of concern is not a site-specific tumor but, rather, the probability of a tumor occurring.

Nondietary Exposure to Pesticides

Although it was not generally within the committee's charge to examine exposures to pesticides by routes other than dietary, the committee wishes to point out that infants and children are subject to such exposures from a variety of sources. These sources should not be overlooked when attempting to estimate the total exposure of infants and children to pesticides and are therefore briefly summarized in this section.

In January 1990 EPA published the Nonoccupational Pesticide Exposure Study (NOPES). One of the study's primary objectives was to assess the relative contribution of each source to overall exposure to certain pesticides. Among their findings, the NOPES researches concluded that (1) "house dust may be a source of exposure to pesticides via dermal contact, ingestion, and inhalation of suspended particulates, especially for infants and toddlers"; (2) "acute dermal exposures that occur during application events may contribute substantially to total exposure"; and (3) that, for the pesticides they examined, exposure from drinking water appeared to be minimal (EPA, 1990). Thus, exposure from all sources—not just ingestion—must be considered when estimating total exposure and risk to children.

Exposure via Parents

The child's first exposure to pesticides begins in utero, where chemicals may cross over the mother's placenta. Several studies have suggested an association between parental exposure (occupational and otherwise) to pesticides and childhood cancers. Researchers from the Children's Cancer Study Group (a cooperative clinical trials group with approximately 100 members and affiliate institutions in the United States and Canada) conducted a case-control study of occupational and household exposures of parents of 204 children with acute nonlymphoblastic leukemia (ANLL) (Buckley et al., 1989). Their most consistent finding was an association of ANLL risk when both mother and father had been exposed to pesticides.

Human birth defects possibly associated with prenatal occupational exposure to the organophosphate oxydemeton-methyl were published in 1989 by Romero et al. (1989). Gordon and Shy (1981) used ecologic data to explore simultaneous maternal exposure to multiple agricultural chemicals in Iowa and Michigan and found an increased risk for facial clefts among offspring.

Lowengart and associates (1987) conducted a case-control study of leukemia patients 10 years and younger and the occupational and home exposures of their parents. With 123 matched pairs, a statistically significant association was seen between leukemia and the use of home and garden pesticides by either parent. In a cross-sectional study of 2,463 parents employed in the Central Valley area of California, limb reduction defects were observed to occur more frequently among the offspring of agricultural workers (relative risk of 2.3) at rates that were in excess of nationally established norms (Schwartz et al., 1986).

As is seen with lead and asbestos, children are at risk from toxicants that their parents unwittingly bring home on their clothes and other apparel. Children of parents employed in agricultural settings may be exposed to these "take-home" pesticides when the work clothing of their parents is washed with other family clothes (Formoli, 1990).

Exposure Through Air

Outdoor Air

In April 1990 the State of California canceled the permits for all uses of the soil fumigant 1,3-dichloropropene (commonly referred to as Telone). This action was prompted by an air monitoring program in the Central Valley that measured levels at a school in the area as high as 160 µg/m3 (L. Baker and J. Behrmann, Air Resources Board, Sacramento, Calif., personal commun., 1990). Concentrations of 0.2 µg/m3 of Telone in air are associated with cancer risks of 1 in 100,000 over 70 years, according to the EPA and the California Department of Health Services (CDHS).

Telone is usually applied undiluted to the soil around vegetable and tobacco crops to control nematodes and insects. Exposure to the vapor causes irritation of the mucosa and respiratory tract. The chemical is absorbed through intact skin, and systemic toxicity may follow cutaneous exposure as well as inhalation or ingestion of the compound (Flessel et al., 1978). Telone has been classified as a probable carcinogen in humans by the EPA (EPA, 1986) and is on the State of California's "Proposition 65" list of chemicals known to cause cancer.

Because pesticides are usually finely dispersed as droplets or particles at the time of application, aerial drift may cause them to be carried away from the target area where they were applied (Matsumura and Madhukar, 1984). Air sampled in areas where pesticides are used may contain residues either as vapors or bound to particles. Recent studies have found pesticide residues suspended in fog. Respiratory absorption of chemicals tends to be more rapid than absorption through other routes of exposure, because of the abundant blood supply in and the thinness of the alveolar membrane (Matsumura and Madhukar, 1984).

Exposure to pesticide residues from ambient air sources is generally higher in areas close to agricultural lands and in communities surrounding pesticides manufacturing factories. Where urban and suburban developments are interspersed within agricultural lands, movement of the more volatile chemicals present potentially significant human exposure. Episodes of illnesses in communities near agricultural areas have been reported simultaneously with applications of the fumigants methyl bromide and chloropicrin (Murray et al., 1974; CDHS, 1980; Goldman et al., 1987).

Cotton defoliants have also been associated with a higher incidence of acute health symptoms among residents living near cotton fields. Tributyl phosphorotrithioate, trade name DEF, is a defoliant of primary concern from a public health perspective (Scarborough et al., 1989). DEF is quite stable and has been detected at very low concentrations in the ambient air of residential and urban areas of cotton-growing counties, even at times of the year when cotton is not being sprayed. However, formulation impurities of butyl mercaptan or dibutyl disulfite may be the causative agents rather than the parent defoliant. In a small northern California community, an increase in health symptoms such as headaches, runny nose, and asthma attacks was reported by residents living adjacent to a potato field that had been treated with the organophosphate pesticide ethoprop. Ethoprop, like DEF, releases a strong-smelling mercaptan gas (Ames and Stratton, 1991). Organophosphates and other commonly used pesticides have been detected in ambient air in California's Central Valley, but generally in such low concentrations that they are unlikely to contribute significantly to the total exposure of humans.

Arthur et al. (1976) measured the levels of several organophosphate and organochlorine pesticides in the Mississippi delta and found that the residue levels were the highest in August and September. In rural areas, where residue levels were higher than in urban areas, the highest concentrations were found where spraying was reported (Arthur et al., 1976).

Indoor Air

The widespread use of pesticides such as flea bombs and insecticide sprays and foggers in the home exposes children to pesticides in their indoor environment. Most home-use products contain either organophosphates or carbamates as their active ingredients, both of which are cholinesterase-inhibiting compounds. These compounds affect the nervous system and at low doses may cause a variety of cholinergic symptoms such as drooling, excessive urination, or diarrhea (Berteau et al., 1989). For six pesticides (chlordane, heptachlor, aldrin, chlorpyrifos, diazinon, and gamma-BHC) analyzed in EPA's NOPES study, the mean air exposures were always or often higher than the estimated dietary exposure for the same compounds (EPA, 1990).

Fenske et al. (1990) measured chlorpyrifos (Dursban) concentrations following its application for flea treatment in a carpeted apartment. They found that the chlorpyrifos vapors measured in the infant breathing zone (25 cm above the carpet) were substantially higher than those measured in the sitting adult's breathing zone. Time-weighted averages for the 24-hour postapplication period in the infant breathing zone were 41.2 and 66.8 µg/m3 for ventilated and nonventilated rooms, respectively; this is substantially higher than the interim guideline of 10 µg/m3 proposed by the National Research Council's Committee on Toxicology for chlorpyrifos in indoor air following termiticide treatments (NRC, 1982). In addition, air concentrations increased from the time of application up to 5 to 7 hours later. The authors suggested that the treated carpet served as a source of volatilized chlorpyrifos and that although open windows provided mixing and dilution of air 1 m above the carpet, concentrations near the floor were affected much less (Fenske et al., 1990).

The short- and long-term health effects to exposure to commonly used home-use pesticide products are largely unknown. In assessing risk for infants in chlorpyrifos-treated homes, based on several conservative assumptions, Berteau et al. (1989) calculated an absorbed dose of 2.68 mg/kg. Fenske et al. (1990) found that the total estimated absorbed chlorpyrifos dose for infants exceeded the EPA's no-observed-effect level (NOEL) of 0.03 mg/kg/day in each case. The NOEL for chlorpyrifos is based on measurable changes in plasma acetylcholinesterase.

The indoor use of pesticides in public buildings such as schools and day-care centers leads to an additional source of exposure for children. In one episode, employees of a school for mentally handicapped children became ill within hours of entering a building that had been treated for roaches 3 days earlier and had not been ventilated. No students were admitted into the building until 14 days after the incident, when air levels of the pesticides used (dichlorvos and propoxur) had decreased to an acceptably safe level (White et al., 1987). An air analysis indicated that the levels of dichlorvos in the air were decreasing over time, but at a much slower rate than was expected from the data provided by the manufacturer.

Chlordane was the leading compound for controlling termites in homes for several years. Although it has since been canceled by the EPA for use as a termiticide, research demonstrates that chlordane air levels decline very slowly with time (Menconi et al., 1988). In a cross-sectional epidemiological investigation of 85 chlordane-treated households containing a total of 261 people, investigators found a dose-response relationship between chlordane levels in home indoor air and incidence of migraines, sinusitis, and bronchitis (Menconi et al., 1988). Cases of more serious health effects such as neuroblastoma, acute leukemia, and aplastic anemia have been associated with exposure to chlordane (Infante et al., 1978).

Pentachlorophenol (PCP) is commonly used as a wood preservative and has become the second most heavily used pesticide in the United States (Cline et al., 1989). It is used in wood homes and on playground equipment. Studies indicate that PCP is virtually ubiquitous in the environment, and measurable residues of PCP are found in most people (Hill et al., 1989). Hill et al. (1989) found PCP in 100% of urine samples taken from 197 Arkansas children. Researchers at the Centers for Disease Control (CDC) found that mean serum PCP levels were 10 times higher in residents of PCP-treated log homes than in the controls (40 ppb compared to 420 ppb) and that serum levels for children in the log homes were "significantly higher" than those for their parents (Cline et al., 1989).

Exposure via Contaminated Surfaces

Home-Use Products

Indoor insecticide sprays and foggers may persist on carpets, floors, and other surfaces in the home. Young children, particularly those wearing only diapers, may be exposed playing on previously sprayed surfaces. In 1980 an 11-day-old infant suffered respiratory arrest in a hospital waiting room. Pesticide poisoning was suspected because tests showed his red blood cell cholinesterase levels to be depressed to 50% of normal low baseline levels. Around the time of his birth, the child's home had been treated with chlorpyrifos; the chemical was subsequently found on dish towels, food preparation surfaces, and the infant's clothing (Dunphy et al., 1980).

Pet Products

Flea control products may persist on the pet's fur and be transferred to children during contact with the animal. Flea control products commonly used in veterinary clinics, pet stores, and other commercial establishments include carbaryl, chlorfenvinphos, chlorpyrifos, dimethyldichlorovinyl phosphate (DDVP), fenthion, malathion, phosmet, and propoxur (Ames et al., 1989).

Playground Equipment

Wooden playground equipment is another source of pesticide exposure because of the various kinds of wood preservatives used to prevent microbial and insect attacks. A 1987 California survey estimated that approximately 20% of all wooden structures in parks were treated with chemical preservatives. Some wood preservatives—PCP, chromium, boric acid, creosote, and arsenic—can induce adverse skin reactions such as contact dermatitis, hyperkeratosis, and, in the extreme case, skin cancer (CDHS, 1987).

Exposure via Medications and Personal Products

Another important route of exposure involves the direct application of insect repellents and pediculocides to children's skin. These include such compounds as N,N-diethyl-m-toluamide, lindane, and malathion. Lanolin, used by some breastfeeding mothers on their nipples, is also a concern because of the pesticides it can contain.


N,N-Diethyl-m-toluamide, commonly called Deet, is the active ingredient in numerous commercially available insect repellents. Although insect repellents can provide great personal benefit, rare adverse reactions can occur. Since 1961, at least six cases of systemic toxic reactions from repeated cutaneous exposure to Deet have been reported. Six girls, ranging in age from 17 months to 8 years, developed behavioral changes, ataxia, encephalopathy, seizures, and/or coma after repeated cutaneous exposure to Deet ; three died (Oransky et al., 1989). Neurobehavioral analysis showed strong correlation between Deet exposure and affective symptoms, insomnia, muscle cramps, and urinary hesitation (McConnell et al., 1987).

In August 1989 the New York State Department of Health investigated five reports of generalized seizures temporally associated with topical use of Deet. Four of the patients were boys from 3 to 7 years old (Oransky et al., 1989).

Lindane and Malathion

For almost 30 years the pesticide lindane (a chlorinated hydrocarbon) has been used in a shampoo for the treatment of head lice (Taplin and Meinking, 1988). Concern has been raised about potential central nervous system damage from exposure to lindane. In particular, cases of central nervous system toxicity have been reported from accidental ingestion as well as from single percutaneous exposures (Lee and Groth, 1977). One author reported two instances in which lindane lotion was given orally to children with scabies because of a lack of communication in one case and a language barrier in the other (Taplin and Meinking, 1988). Malathion has been recommended as a preferable treatment over lindane (Taplin et al., 1982; Fine, 1983).


Lanolin, a derivative of sheep's wool, is commonly used as an ointment to treat sore, cracked skin. Mothers who breastfeed frequently use it on their nipples, and it is sometimes applied directly to children's skin. The organophosphate pesticides diazinon and chlorpyrifos and several organochlorine pesticides such as dieldrin have been found at measurable levels in lanolin. The U.S. Food and Drug Administration identified 16 pesticides in lanolin it sampled in 1988. The principal source of these residues is the wool from sheep treated with a pesticide dip to control parasite infestations in the fleece (Cade, 1989). The fat-soluble organophosphate pesticide diazinon presented the greatest concern because of its frequent occurrence (21 of 25 samples) and the high levels identified (up to 29.2 ppm). (T. Levine, EPA, personal commun., 1988).

Occupational Exposures

In agricultural communities, children are often directly exposed to pesticides when they accompany their parents in the field or work there themselves (Pollack et al., 1990). In 1980, some 19 farm workers suffered organophosphate poisoning after working in a cauliflower field (Whorton and Obrinsky, 1983). Five of the workers were 18 years old or younger; three of those were between the ages of 9 and 15 years.

Exposure via Accidental Ingestion

Accidental poisonings are all too common among children. In one study of 37 children who had been hospitalized at Children's Medical Center in Dallas as a result of organophosphate or carbamate pesticide poisoning, ingestion of a liquid was the most common (73%) mechanisms of exposure. Zwiener and Ginsburg (1988) reported that most poisonings took place in the home and here the result of careless storage of the original container or placement in unmarked or uncovered containers.

Conclusions and Recommendations

Like other members of the general population, infants and children are exposed to pesticide residues in their diets. Estimation of dietary intakes requires information on both food consumption patterns and residue levels in food. The purpose of this chapter has been to demonstrate methods for estimating exposure to pesticides in the diet. The committee was guided by previous work on exposure estimation by the National Research Council (NRC, 1988, 1991a,b). Infants and children are also exposed to pesticides by nondietary routes, including air and contaminated surfaces such as rugs and playground equipment. Although a detailed analysis of nondietary routes of exposure to pesticides is outside the scope of this report, it is important for risk assessment purposes to consider the total exposure from all media. The following are the conclusions of the committee.


• Pesticide residues are present in the diets of infants and children. Estimation dietary intakes of pesticides by infants and children requires information on both food consumption patterns and residue levels in food.

• Accurate estimation of dietary intake of pesticides by infants and children is difficult due to the limited amount of data on food consumption patterns of infants and children (Chapter 5) and limitations in the available data on pesticide residues (Chapter 6).

• Dietary exposures to pesticide residues can vary widely. Since most pesticide residues in foods are below the analytical limit of quantification, with comparatively few high residue levels, the distribution of dietary exposure to pesticides includes many low intakes. Some degree of positive skewness may be observed due to the occurrence of high consumption or high residue levels.

• To estimate dietary exposure to pesticides for infants and children, the committee combined probability distributions of food consumption with probability distributions of residue levels in order to obtain a probability distribution of individual exposures. The use of probability distributions for exposure assessment provides a more complete characterization of human exposure to pesticides residues in food than the use of summary statistics such as means or upper percentiles of exposure.

More accurate estimates of upper quantiles of the exposure distribution can be obtained by pointwise multiplication of the residue and consumption distributions than by multiplying the quantiles obtained from the residue and consumption distributions, separately. Moreover, the probability distribution approach based on 1-year age groupings of children provides useful information on differences in exposure patterns for children 1 to 5 years of age.

• Average daily ingestion of pesticide residues is an appropriate measure of exposure for chronic risk assessment, whereas actual individual daily ingestion is more appropriate for acute risk assessment.

Since chronic toxicity is often related to long-term average exposure, the average daily dietary exposure to pesticide residues may be used as the basis for risk assessment with delayed irreversible chronic toxic effects. To take into account different food consumption patterns among individuals, the distribution of average daily dietary intake of pesticides should be examined within the population of interest. Since acute toxicity is more often mediated by peak exposures occurring within a short period (e.g., over the course of a day or even during a single eating occasion), individual daily intakes are of interest for risk assessment for acute toxic effects. Examination of the distribution of individual daily intakes for persons within the population of interest reflects both day-to-day variation in pesticide ingestion for specific individuals as well as variation among individuals. This distribution can be used to estimate the number of person-days in a given period during which intake will exceed a specified level, such as the acceptable daily intake (ADI), or reference dose.

• At present, there is a relatively limited amount of information on food consumption patterns of infants and children. To obtain accurate estimates of the distribution of individual intakes, more elaborate and more intensive consumption monitoring protocols are required.

• Because residue monitoring surveys conducted for compliance purposes are expected to lead to higher residue levels than those present in the general food supply, assessment of human exposure should normally be based on surveillance surveys. In using surveillance data, however, consideration needs to be given to regional differences in pesticide use and resultant residue levels.

The committee acknowledges that pesticide food surveillance data are generated by randomly sampling food items from the distribution system. The purpose of this sampling is to ensure agricultural compliance with acceptable pesticide use practices. This sampling is broad-based and often not focused only on pesticides actually used. Pesticide field trial data are generated under strictly controlled conditions of use. These data better reflect actual levels at the time of harvest when it is known that a specific pesticide has been used. Each data source is used for purposes other than identifying actual dietary exposures, although both are useful in attempting to estimate these exposures.

• Frequently, the levels of pesticide residue in foods are below the analytical limit of quantification (LOQ). Since the actual residue level in such cases may lie anywhere between zero and the LOQ, there is some uncertainty about actual exposures in such cases.

For example, replacing the residue measurements below the LOQ with zero yields lower exposure estimates than substituting the LOQ for the unknown residue level. Notable differences occur when the analytical method is insensitive, the LOQ is high, or a large proportion of residues lies below the LOQ.

• The concentration of pesticide residues in foods may increase or decrease during food processing.

Changes in residue levels that occur during the processing of food are especially important in assessing the exposures of infants and young children, who consume large quantities of single processed foods, such as fruit juices, milk, and infant formula. In addition to the data accumulated by the food industry (Chapter 6), studies by pesticide manufacturers such as those furnished to the committee on the fate of residues during processing need to be conducted for most pesticides that produce detectable residues in food.

• Specific pesticides can be applied to more than one crop and, hence, appear on a number of food commodities. Residues of several pesticides may also appear on a single food commodity.

• Intake of multiple pesticides with a common acute toxic effect can be estimated by converting residues for each chemical to equivalent units of one of the compounds. The standardized residues can then be summed to estimate total residue levels in toxicity equivalence factors, and then combined with consumption data to construct a probability distribution of total exposure to all pesticides having a common mechanism of action.

Certain classes of pesticides such as cholinesterase inhibitors act by a common toxic mechanism. To properly evaluate the potential health effects of exposure to such pesticides, it is important to consider the total exposure to all pesticides in the class.

• Children are exposed to pesticides by nondietary routes.

Occupational exposure of the parent could result in exposure of the child in utero, in the home environment, or in the occupational setting of the parents. Pesticide residues have been detected in outdoor and indoor air, on contaminated surfaces, and in medications and personal products.

• When less than 100% of a given crop is treated with a particular pesticide, consideration might be given to adjusting exposure estimates according to the percentage of crop acreage treated. This adjustment can result in substantial reductions in estimates of exposure. This adjustment will be appropriate when the percentage of the crop treated is similar in different regions of the country, or when the crop is uniformly distributed throughout the country. Such adjustments should not be considered in the case of pesticides inducing acute toxic effects, since peak exposures are of importance in this case.

When these adjustments are used to adjust national data, they may result in averages that do not account for regional differences in pesticide use. It is therefore important that exposure estimates that have not been adjusted for acreage treated be presented and that such adjustments be critically examined.


The following recommendations were developed by the committee.

• Probability distributions based on actual data rather than simple summary statistics such as means or percentiles should be used to characterize human exposure to pesticide residues on food.

The advantage of using probability distributions rather than summary statistics to characterize exposure is that variation in individual food consumption patterns and residue levels in food are taken into account. This will require the collection of more detailed data on food consumption and residue levels as discussed in Chapters 5 and 6, respectively, but will provide more statistically robust estimates than the agency currently develops.

• The distribution of average daily exposure of individuals in the population of interest is recommended for use in chronic toxicity risk assessment; the distribution of individual daily exposures is recommended for evaluating acute toxic effects.

This recommendation is based on the committee's observation that chronic toxicity is typically related to long-term average exposure, whereas acute toxicity is more often mediated by peak exposures occurring within a short period, either over the course of a day or even during a single meal.

• If appropriately designed and conducted, surveillance studies of pesticide residues in food provide unbiased data on residue levels in food products. Field trials are also useful sources of information on pesticide residues in food. Such studies should be continued in order to expand the data base for evaluating dietary exposures to pesticides.

Surveillance studies based on random samples designed to provide a representative picture of residue levels in food are required to obtain unbiased information on dietary exposure to pesticides.

• The committee recommends that research to reduce the uncertainty in estimates of dietary exposure to pesticides be encouraged. Specifically, the development of improved analytical methods for residue analyses and statistical methods for imputing residue levels below the LOQ can lead to improved estimates of pesticide exposure.

All analytical methods for measuring pesticide residue levels in food are subject to an LOQ. Results below the LOQ may be as low as zero or as high as the LOQ itself, thereby imparting uncertainty regarding actual human exposure levels. This uncertainty will be reduced if more sensitive methods with lower LOQs are developed. Such technological improvements should be encouraged even in the absence of other pressures for more sensitive analytical methods.

Statistical methods for use with censored data (i.e., based on specific assumptions) can be used to impute residue levels below the LOQ, provided that the percentage of the residue data lying below the LOQ is not large. The use of such methods can reduce the uncertainty in resulting estimations of human exposure.

• When using multiresidue scans to detect different compounds in one scan of one food sample, all results should be recorded together. This will make possible more accurate evaluation of exposure distributions for multiple chemicals.

• The committee does not recommend the routine application of adjustments for the percentage of the crop treated in estimating dietary exposure to pesticides.

Adjustments for acreage treated are appropriate only under certain conditions. For example, such adjustments may be used when there is little regional variation in acreage treated, or when the crop is uniformly distributed at the national level.

• To determine total dietary exposure to a particular pesticide, intakes from all foods on which residues might be present need to be combined.

Many pesticides are approved for use on more than one crop. In addition, a single crop may be used in the production of a variety of processed foods. To estimate the total dietary exposure to a particular pesticide, it is important to consider the contribution of all foods on which residues might occur.

• To properly evaluate the potential risk from exposure to multiple pesticides with common mechanisms of action, it is necessary to develop measures of total exposure to pesticides within the same class that reflect the overall toxicity of all pesticides combined.

Since the combined effect of pesticides acting by a common mechanism can be greater than the individual effect of any single pesticide, it is important to develop risk assessment methods that address the total risk from exposure to all pesticides within the same class. One possible approach is to establish toxicity equivalence factors based on no-observed-effect levels as was done for organophosphates in this chapter.

• Because infants and children are subject to nondietary sources of exposure to pesticides, it is important to consider total exposure to pesticides from all sources combined.


Amdur, M.O., J. Doull, and C.D. Klaassen, eds. 1991. Caserett and Doull's Toxicology, 4th Ed. New York: Pergamon. 1033 pp.

Ames, R.G., and J.W. Stratton. 1991. Acute health effects from community exposure to N-proply mercaptan from an ethoprop (Mocap R)-treated potato field in Siskiyou County, California. Arch. Environ. Health 46:213–217.

Ames, R.G., S.K. Brown, J. Rosenberg, R.J. Jackson, J.W. Stratton, and S.G. Quenon. 1989. Health symptoms and occupational exposure to flea control products among California pet handlers. Am. Ind. Hyg. Assoc. J. 50:466–472.

Arthur, R.D., J.D. Cain, and B.F. Barrentine. 1976. Atmospheric levels of pesticides in the Mississippi Delta. Bull. Environ. Contam. Toxicol. 15:129–134.

Berteau, P.E., J.B. Knaak, D.C. Mengle, and J.B. Schreider. 1989. Insecticide absorption from indoor surfaces: Hazard assessment and regulatory requirements. Pp. 315–326 in Biological Monitoring for Pesticide Exposure: Measurement, Estimation and Risk Reduction, ACS Symposium Series 382, R.G.M. Wang, C.A. Franklin, R.C. Honeycutt, and J.C. Reinert, eds. Washington, D.C.: American Chemical Society.

Buckley, J.D., L.L. Robison, R. Swotinsky, D.H. Garabrant, M. LeBeau, P. Manchester, M.E. Nesbit, L. Odom, J.M. Peters, W.G. Woods, and G.D. Hammond. 1989. Occupational exposures of parents of children with acute nonlymphocytic leukemia: A report from the children's cancer study group. Cancer Res. 49:4030–4037.

Cade, P.H. 1989. Pesticide in lanolin [letter]. JAMA 262:613.

CDHS (California Department of Health Services). 1980. Ventura County Environmental Health Department, Report to the California Department of Health Services. Sacramento, Calif.: California Department of Health Services.

CDHS (California Department of Health Services). 1987. Report to the Legislature: Evaluation of Hazards Posed by the Use of Wood Preservatives on Playground Equipment . Department of Health Services, Office of Environmental Health Hazard Assessment. Berkeley, Calif.: California Department of Health Services.

Cline, R.E., R.H. Hill, Jr. D.L. Phillips, and L.L. Needham. 1989. Pentachlorophenol measurements in body fluids of people in log homes and workplaces. Arch. Environ. Contam. Toxicol. 18:475–481.

Dunphy, J., M. Kesselbrenner, A. Stevens, B. Vlec, and R.J. Jackson. 1980. Pesticide poisoning in an infant—California. MMWR 29:254–255.

Ecobichon, D.J. 1991. Toxic effects of pesticides. Pp. 565–622 in Casarett and Doull's Toxicology: The Basic Science of Poisons, 4th Ed., M.O. Amdur, J. Doull, and C.D. Klaassen, eds. New York: Pergamon.

EPA (U.S. Environmental Protection Agency). 1986. Guidance for the Reregistration of Pesticide Products Containing 1,3-Dichloropropene (Telone II) as the Active Ingredient. NTIS No. PB-87-1117-87. Springfield, Va.: National Technical Information Service.

EPA (U.S. Environmental Protection Agency). 1987. Risk Assessment Forum: Interim Procedures for Estimating Risks Associated with Exposures to Mixtures of Chlorinated Dibenzo-p-dioxins and Dibenzofurans. Washington, D.C.: U.S. Environmental Protection Agency.

EPA (U.S. Environmental Protection Agency), 1988. Cholinesterase Inhibition as an Indicator of Adverse Toxicological Effect. Washington, D.C.: U.S. Environmental Protection Agency.

EPA (U.S. Environmental Protection Agency). 1990. Review of Cholinesterase Inhibition and Its effects. Report of the EPA SAB/SAP Joint Study Group on Cholinesterase, Washington, D.C.: U.S. Environmental Protection Agency.

EPA (U.S. Environmental Protection Agency). 1992. Integrated Risk Information System (IRIS). Online. Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office. Cincinnati, Ohio: U.S. Environmental Protection Agency.

Feldman, D., and M. Fox. 1991. Probability, The Mathematics of Uncertainty. New York: Marcel Dekker.

Fenske, R.A., K.G. Black, K.P. Elkner, C.-L. Lee, M.M. Methner, and R. Soto. 1990. Potential exposure and health risks of infants following indoor residential pesticide applications. Am. J. Public Health 80:689–693.

Fine, B.C. 1983. Pediculosis capitis [letter]. New Engl. J. Med. 309:1461.

Flessel, P., J.R. Goldsmith, E. Kahn, J.J. Wesolowski, K.T. Maddy, and S.A. Peoples. 1978. Acute and possible long-term effects of 1,3-dichloropropene—California. MMWR 27:50, 55.

Formoli, T. 1990. Pesticide Safety Information Series A-7, California Department of Food and Agriculture Worker Health and Safety Branch. HS-1228. Sacramento, Calif.

Goldman, L.R., D. Mengle, D.M. Epstein, D. Fredson, K. Kelly, R.J. Jackson. 1987. Acute symptoms in persons residing near a field treated with the soil fumigants methyl bromide and chloropicrin. West. J. Med. 147:95–98.

Gordon, J.E., and C.M. Shy. 1981. Agricultural chemical use and congenital cleft lip and/or palate. Arch. Environ. Health 36:213–221.

Hill, R.H., T. To, J.S. Holler, D.M. Fast, S.J. Smith, L.L. Needham, and S. Binder. 1989. Residues of chlorinated phenols and phenoxy acid herbicides in the urine of Arkansas children. Arch. Environ. Contam. Toxicol. 18:469–474.

Infante, P.F., S.S. Epstein, and W.A. Newton, Jr. 1978. Blood dyscrasias and childhood tumors and exposure to chlordane and heptachlor . Scand. J. Work Environ. Health 4:137–150.

Lee, B., and P. Groth. 1977. Scabies: Transcutaneous poisoning during treatment [letter]. Pediatrics 59:643.

Lowengart, R.A., J.M. Peters, C. Cicioni, J. Buckley, L. Bernstein, S. Preston-Martin, and E. Rappaport. 1987. Childhood leukemia and parents' occupational and home exposures. J. Natl. Cancer Inst. 79:39–46.

Matsumura, F., and B.V. Madhukar. 1984. Exposure to insecticides. Pp. 1–25 in Differential Toxicities of Insecticides and Halogenated Aromatics, International Encyclopedia of Pharmacology and Therapeutics Series, F. Matsumura, ed. Oxford, England: Pergamon.

McConnell, R., A.T. Fidler, and D. Chrislip. 1987. Health Hazard Evaluation Report HETA-83-085-1757. Everglades National Park, Everglades, Florida. Govt. Reports Announcements & Index (GRA&I). Issue 17. 1987.

Menconi, S., J.M. Clark, P. Langenberg, and D. Hryhorczuk. 1988. A preliminary study of potential human health effects in private residences following chlordane applications for termite control . Arch. Environ. Health 43:349–352.

Murphy, S.D. 1986. Toxic effects of pesticides. Pp. 519–581 in Casarett and Doull's Toxicology, 3rd Ed. New York: Macmillan.

Murray, R.A., L.E. Mahoney, and R.R. Sachs. 1974. Illness associated with soil fumigation, California. MMWR 23:217–218.

NRC (National Research Council). 1977. Drinking Water and Health. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1982. An Assessment of the Health Risks of Seven Pesticides Used for Termite Control. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1986. Drinking Water and Health, Vol. 6. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1987. Regulating Pesticides in Food: The Delaney Paradox. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1988. Complex Mixtures: Methods for In Vivo Toxicity Testing. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1991a. Human Exposure Assessment for Airborne Pollutants: Advances and Opportunities. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1991b. Frontiers in Assessing Human Exposures to Environmental Toxicants. Washington, D.C.: National Academy Press.

NRC (National Research Council). 1992. Environmental Neurotoxicology. Washington, D.C.: National Academy Press.

Oransky, S., B. Roseman, D. Fish, T. Gentile, J. Melius, M.L. Cartter, and J.L. Hadler. 1989. Seizures temporally associated with use of DEET insect repellant—New York and Connecticut. MMWR 38:678–680.

OTA (Office of Technology Assessment). 1990. Neurotoxicity: Identifying and Controlling Poisons of the Nervous System. OTA-BA-436. Washington, D.C.: U.S. Government Printing Office.

Petersen and Associates, Inc. 1992. Background: Derivation of Water Consumption Estimates Used by the Tolerance Assessment System. Paper prepared for the U.S. Environmental Protection Agency, Washington, D.C.

Pollack, S.H., P.J. Landrigan, and D.L. Mallino, 1990. Child labor in 1990: Prevalence and health hazards. Annu. Rev. Public Health 11:359–375.

Romero, P., P.G. Barnett, J.E. Midtling. 1989. Congenital anomalies associated with maternal exposure to oxydemeton-methyl. Environ. Res. 50:256–261.

Savage, E.P., T.J. Keefe, L.M. Mounce, R.K. Heaton, J.A. Lewis, and P.J. Bucar. 1988. Chronic neurological sequelae of acute organophosphate pesticide poisoning. Arch. Environ. Health 43:38–45.

Scarborough, M.E., R.G. Ames, M.J. Lipsett, and R.J. Jackson. 1989. Acute health effects of community exposure to cotton defoliants. Arch. Environ. Health 44:355–360.

Schwartz, D.A., L.A. Newsum, and R.M. Heifetz. 1986. Parental occupation and birth outcome in an agricultural community. Scand. J. Work Environ. Health 12:51–54.

Taplin, D., and T. Meinking. 1988. Infestations. In Pediatric Dermatology, 2 Vols., L.A. Schachner and R.C. Hansen, eds. New York: Churchill Livingstone. 1674 pp.

Taplin, D., P.M. Castillero, and J. Spiegal. 1982. Malathion for treatment of Pediculus humanus var. capitis infestation. JAMA 247:3103–3105.

USDA (U.S. Department of Agriculture). 1983. Food intakes: Individuals in 48 states, year 1977–78. Nationwide Food Consumption Survey 1977-78, Rep. No. I-1. Hyattsville, Md.: U.S. Department of Agriculture. 617 pp.

USDA (U.S. Department of Agriculture). 1987. CSFII, Nationwide Food Consumption Survey, Continuing Survey of Food Intakes of Individuals, 1985. Human Nutrition Information Service, 5 vols. Hyattsville, Md.: U.S. Department of Agriculture.

White, L.E., J.R. Clarkson, and S.N. Chang. 1987. Health effects from indoor air pollution: Case studies. J. Community Health 12(2-3):147–155.

Whorton, M.D., and D.L. Obrinsky. 1983. Persistence of symptoms after mild to moderate acute organophosphate poisoning among 19 farm field workers. J. Toxicol. Environ. Health 11:347–354.

Zwiener, R.J., and C.M. Ginsburg. 1988. Organophosphate and carbamate poisoning in infants and children. Pediatrics 81:121–126; erratum, 81:683.
Site Admin
Posts: 29985
Joined: Thu Aug 01, 2013 5:21 am


Return to Health

Who is online

Users browsing this forum: Google [Bot] and 1 guest