SCOPE 49 - Methods to Assess Adverse Effects of Pesticides on Non-target Organisms

	2.1 MODELLING TECHNIQUES TO ESTIMATE PESTICIDE EXPOSURES
		2.1.1 MODELS AND MODELLINGWHY USE THEM?
		2.1.2 VALIDATION OF ENVIRONMENTAL FATE AND DISTRIBUTION MODELS
		2.1.3 USEFULNESS OF MODELS TO PREDICT EXPOSURE PROFILES
		2.1.4 VALUE OF MODELS TO DESIGN MONITORING PROGRAMMES AND TO EVALUATE DATA
	2.2 METHODS TO MEASURE EXPOSURE TO PESTICIDES
		2.2.1 USES OF MONITORING DATA
		2.2.2 PURPOSES OF MONITORING DATA
		2.2.3 FATE AND TRANSFORMATION: HUMAN
			2.2.3.1 The concept of exposure monitoring
			2.2.3.2 Patterns of exposure
			2.2.3.3 Specimen selection
			2.3.3.4 Specific media
			2.2.3.5 Occupational exposure
			2.2.3.6 Non-occupational exposure
			2.2.3.7 Dietary exposure
		2.2.4 ECOSYSTEM MONITORING
	2.3 RECOMMENDATIONS

The distribution and eventual fate of a chemical released into the environment is governed by a variety of processes, the rates of which are determined by many factors. These processes consist of those that remove the chemical from the environment (e.g., biodegradation or photodegradation) and those that govern the distribution of the chemical (e.g., chemical sorption by soil or sediment). The processes are competitive and may be complex. Their relative importance is dependent upon the physico-chemical properties of substances, the environment in which they are released, and the time lapsed since their release.

Actual exposure subsequently encountered is dependent not only on the application rate and method and on the time since application, but also on the nature of the environment in which a substance is introduced. Depending on the relative influence of each process in a given situation, the behaviour of a pesticide varies drastically from one location to another, and the duration and magnitude of exposure are highly variable. Since considerable information is available on the processes which may attenuate the concentration of a pesticide after its release, pesticide exposure can be examined in terms of the particular environment in which the substance is introduced.

Because pesticides must be registered in many countries, information is generally available on the transformation of the pesticidal agents in mammals and soils, including their degradation and persistence under assorted environmental conditions. A detailed review of the fate and distribution of pesticides is beyond the scope of this report, and exhaustive reviews of the behaviour of pesticides in the environment already exist. Furthermore, the International Union of Pure and Applied Chemistry (IUPAC) has released a critical examination of this testing procedure, and the reader is referred to this material for specific details.

Within different environments, exposure to a pesticide can be determined either directly by quantitative measurement or indirectly by predicting the nature and magnitude of its presence from an understanding of its behaviour under alternative environmental conditions. The usefulness of large collections of data and of current monitoring approaches to characterize exposures has been raised to a high degree of importance. Since an understanding of exposure patterns is a prerequisite to any riskbenefit analysis, this issue needs to be resolved before risk-benefit analysis can be accomplished successfully.

* This section  was prepared by G. Becking, P. Bourdeau, G. Butler, D. Calamari, V. Morley, and R. Roberts.

2.1 MODELLING TECHNIQUES TO ESTIMATE PESTICIDE EXPOSURES†

For many years, information on the distribution and fate of chemical substances was acquired largely through retrospective studies involving large-scale monitoring that incorporated an enormous number of chemical analyses. The retrospective approach leaves wide margins for error in the environmental management of chemical substances, for it provides a weak base from which to anticipate problems.†

One approach to reduce this difficulty is to forecast the environmental behaviour of a pesticide through the use of computerized models. Over the past decade, some investigators have advanced the idea that one can predict the behaviour of a chemical on the basis of comparisons of measured properties with those of compounds for which greater amounts of environmental data are available. For example, the concept was introduced for the use of hypothetical `evaluative models' based on properties of stylized environments and pollutants to provide quantitative approaches to exposure estimation.

2.1.1 MODELS AND MODELLINGWHY USE THEM? 

Environmental scientists must assess impacts of pesticides in a multiplicity of situations. It is not practical to establish the associated exposure profiles through direct measurement except in rare cases. Thus, environmental scientists and pesticide regulators require tools to estimate exposures to pesticides from one situation to another.

Tools are needed first to optimize the design of monitoring programs, and secondly to test the validity of (a) study designs used in existing monitoring programs, and (b) compilations of data to estimate environmental pollutant levels. A valid study design must include the sampling of at least the matrices in which the pesticide is most likely to be found at any given time. In complex systems, which matrix, stratum, or sampling site is most likely to be contaminated is not necessarily obvious. Models provide the analyst with the tools to optimize the design of monitoring schemes and hence to increase the chance of detecting a compound, should it be present.

†This section was prepared by P. Bourdeau, D. Calamari, G. Lacy, R. Roberts, and T. Soldan.

Such tools are predictive models that depict how a pesticide may behave in various environments. These models may vary in complexity from structured conceptual models of a pesticide's behaviour in a specified situation to complex mathematical integrations that attempt to describe various processes of translocation and removal. The descriptions of processes can be simple, empirically determined relationships; or they can be complex, theoretically-based relationships, reflecting a chemical's properties and their influence on its translocation and degradation. The exposure models currently used are composed of (a) simple ones that characterize a single input, transport, and removal process controlling the fate of a chemical, and (b) complex models of numerous, simultaneous environmental influences on the ultimate distribution of a chemical. The usefulness of a model is dependent on its ability to provide correlations between the nature of a chemical and its mobility and persistence in a given matrix.

A hypothetical model of the environment, sometimes called the `environmental system' is first constructed by conceptualizing several environmental `compartments' considered to have distinct, uniform characteristics, such as pH, percentage of organic matter, or temperature. The discharge, loading, or emission rates of a chemical are then described by discrete models. Finally, mathematical relationships are developed to describe the intercompartmental transfer and removal processes that influence the dynamics of a chemical.

1. qualitative conceptual comparative distribution models, 
2. generic quantitative situation-specific models, and
3. site-specific quantitative models.

The first group of models provides general qualitative descriptions of a chemical in fixed scenarios. They are used as basic screens for ranking pesticides according to their affinity for various ecological compartments, as well as for the evaluation of data available for understanding the distribution, translocation, and transformation processes.

The second group of models has flexible mathematical bases not fixed to specific scenarios. That is, the user can adjust a scenario to develop a comparative base of exposure profiles. Conceivably, they can also be used as a first approach to predict field distribution and planning studies.

The third group, the site-specific models, are the most difficult to develop, for they are directed at providing detailed descriptions of a pesticide's behaviour in a specific complex ecosystem. The number of processes and compartments in such models requires the use of complex approaches such as sensitivity analysis. Development of such models requires that the ecosystem be described in detail and that measures be available to define the transfer and degradation processes operative within various compartments. This modelling is normally achieved through specific empirical measurement of the rates.

The conclusions resulting from the use of such models are being questioned at present, and their usefulness is questioned by some. At least two reasons exist for this scepticism. First, some managers and scientists have extended conclusions beyond the limits of the assumptions and errors inherent in the models. Second, modellers have added complexity to models without fully evaluating either the implications of the limitations of the input data or the assumptions used to construct the models. Both groups of individuals seem to forget that models are evaluative tools that are useful only within the boundaries of their limitations in their output.

The first two types of models are relevant in hazard assessment schemes, while the third is required in the estimation of risks or benefits.

2.1.2 VALIDATION OF ENVIRONMENTAL FATE AND DISTRIBUTION MODELS

The validation of a model and the modelling process must be performed at different levels depending on the type of model and its ultimate use. In levels 1 and 2, qualitative results are sought; hence, the accuracy of trends identified by the models is the focus of attention rather than the numerical values themselves. At level 3, the convergence between the quantitative output of the models and actual field exposure patterns must be examined.

In either case, two types of error must be examined: first, that resulting from the use of incorrect model structure and assumptions; and secondly that resulting from inaccuracies of the input data.

In levels 1 and 2 qualitative models, trend analyses may be generated from correlations of the sorptive and partition properties of the chemicals with their chemical properties. Generally, correlations based on empirically determined characteristics of chemicals are relatively imprecise, particularly when chemicals with widely differing properties are compared. These correlations have been found to be so imprecise as to fail to characterize trends between isomers. This finding may be of particular concern in hazard assessment at level 2, for a chemical's toxicity can be highly dependent on its isomeric structure.

Correlations based on theoretically calculated properties of a molecule (e.g., fragmentation constants) may provide a sound basis for developing qualitative submodels. The use of non-empirical correlations provides improved internal consistency for comparisons, thereby leading to increased accuracy to predict patterns and trends between chemicals of differing structure. In the qualitative models, the most useful trends are generated when sorptive and advective, rather than degradative, processes dominate, because of the paucity of knowledge about relationships among degradative biological processes.

None of the level 3 models are truly validated. Several models can show convergence between field studies and their output; however, arbitrary decisions are generally made about the characteristics of an ecosystem, a chemical's properties, and the models. Hence, the apparent convergence does not constitute validation.

Laboratory studies can be of considerable value in improving the accuracy of correlations between the physical-chemical parameters and partition-sorption, translocation, and kinetic data. The accuracy of the input data and the models should be based on the objective of the modelling. Simulation chambers and small-scale pilot experiments have a role in demonstrating the validity of trends predicted by the levels 1 and 2 models.

An iterative process exists for the validation of models between their refinement and testing in simulation chambers, small-scale experiments, and field trials. Refinement will occur only through an interactive process of testing and adjustment.

2.1.3 USEFULNESS OF MODELS TO PREDICT EXPOSURE PROFILES

The most accurate models address equilibrium soil-sorption and fish bioconcentration relationships. Models of the sorption of pesticides by other components of aquatic systems are, at best, site-specific; and general predictive relationships are limited. The dynamics under non-equilibrium conditions and at high rates of degradation are at best site-specific and empirically determined. Thus, accurate quantitative predictions are unlikely, given the level of sophistication of today's general models. With sufficient study, site-specific models of a given situation can be developed in which empirical descriptions of the processes are utilized. The degree of precision of such models make them useful for projecting trends within the system, but the results generally cannot be extrapolated to dissimilar situations.

A reason for the limitations is the lack of focus on descriptions of models of the processes. Only as these descriptions improve will models provide useful generic qualitative predictions of exposure.

2.1.4 VALUE OF MODELS TO DESIGN MONITORING PROGRAMMES AND TO EVALUATE DATA

The design of an optimal monitoring program requires the ability to predict the distribution and fate of a chemical at various times within the area to be monitored. Conceptual models of expected equilibrium distribution patterns and dynamic models are both of value.

Those models are particularly useful prior to loss or degradation of significant quantities of a chemical, for it is during this period that simple distribution processes are most likely to govern chemical distribution. In such cases, the accuracy of the evaluations is directly related to that of the description of the relationships between a chemical's partitioning in various phases of an ecosystem. The correlations can be relatively precise and simple.

The predictive relationships for pesticide distribution patterns have sufficient accuracy to be generally useful when the descriptions of the ecosystems are accurate. The difficulty is that accurate models are difficult to obtain in large, complex systems such as rivers and ground-water aquifers. For example, the migration rate of a pesticide through a clay matrix is dependent on understanding the degree of fracturing, a parameter that can only be obtained after detailed geological studies. In the case of a river, good hydrological models are required to predict the location of matrices most likely to contain a pesticide at a given time after its introduction. Where such models exist, they can be very useful to optimize the design of sample programs.

2.2 METHODS TO MEASURE EXPOSURE TO PESTICIDES* 

The sophistication of current analytical and separation techniques makes the acquisition of exposure data relatively simple. Unfortunately in many cases, this has resulted in data collection being an end in itself rather than a means to an end.

Some people view monitoring as simply the collection and storage of data; actually, it is much more. The United Nations defines monitoring as `the collection of analytical data according to an agreed plan of sampling in time and space'. Thus, monitoring data are gathered in response to a defined need or objective, established in consultation between data users and analysts. The definition implies that data should be generated in the most appropriate units to ensure comparability and that a sampling program should be constructed prior to analysis by users of the data.

Consequently, data should be as relevant as possible, and the number of analyses performed should be the minimum compatible with the intended purposes. Because these criteria have not yet been employed, much of the data gathered in the past regrettably cannot be used by modellers, regulators, or others.

*This section was prepared  by G. Becking, P. Bourdeau, G. Butler, D. Calamari, V. Morley, and R. Roberts.

The three principal users of monitoring data are:

1. Regulators wanting measurements of chemicals in various media either to determine regulatory compliance or to predict concentrations and environmental behaviour,
2. Environmental scientists seeking knowledge of concentrations in environmental compartments either to validate environmental transport models or to describe environmental processes, and
3. Risk analysts characterizing risks to NTOs from the use of pesticides. 

2.2.2 PURPOSES OF MONITORING DATA

Data obtained from well-designed monitoring programs can be used for numerous purposes in the fields of public health and environmental protection. For example, these objectives include determinations of trends in environmental concentrations, persistence of bioaccumulation of residues, environmental transport and fate, and estimation of doses to NTOs. Furthermore, monitoring data should be an integral part of validating predictive models.

These data must be collected at appropriate time intervals and in correct environmental compartments to answer authoritatively questions raised by users of this information; otherwise, the measurements of pesticide levels at one point in time in various environmental compartments may be of little value to characterize the dose regimen or to validate the models. Monitoring pesticide concentrations in various environmental compartments (i.e., food, air, biota, water) is required to obtain baseline values and trends over time, thus providing information from which to identify emerging problems.

Most pesticide monitoring programmes analyse only the parent active ingredient, thus failing to take into account any environmental transformation products or other potentially toxic compounds used in pesticide formulations. Either group of substances may be not only persistent but also of toxicological importance. Such information would help identify NTOs at risk, thereby characterizing more comprehensively the overall risk posed by pesticide formulations. Such data would also provide information on environmental persistence and transport, thereby having intrinsic value to validate predictive models.

The mobility, persistence, and distribution of pesticides (e.g., herbicides contaminating ground or surface waters) in the environment can be ascertained through adequate monitoring.

The establishment of causal relationship uncovered in environmental epidemiology studies depend on appropriate definitions of dosage. However, to date, environmental epidemiological studies have suffered from a lack of such information. For example, determining exposure from measurements at one point in time causes much uncertainty in the determination of dose.

Accurate assessment of integrated dosage requires monitoring programs designed from inputs obtained from epidemiologists and chemists. Such measurements need to be frequent and extensive, with samples collected from relevant locations. Such information can, in turn, be used to refine the design of epidemiological studies.

2.2.3 FATE AND TRANSFORMATION: HUMAN*

As pesticidal NTOs of considerable public health concern, humans have come under considerable study to ascertain how and to what extent they are exposed to pesticidal substances. Monitoring human exposures has most often taken the form of measuring the presence of the substance of interest in one or more environmental media: air, food, water, or consumer products. Monitoring has taken place not only at home and at play but also at work. Recently, such monitoring has focused on characterizing human doses most closely associated with injurious consequences, namely, the biological or tissue dose. By using all reasonably accessible tissues, a relatively precise definition of body burden can be obtained, and can aid in deciding appropriate occupational and environmental policies.

2.2.3.1 The concept of exposure monitoring

The term `exposure' refers to the quantity of a foreign chemical entering the organism for a specified duration. Depending on the monitoring methods, other terms may prove useful. `External exposure' may be expressed as concentrations in inhaled air, intake from food and water, or contamination of the skin surface. `Internal dose' is the amount that enters the circulation, and may be computed using the respective transfer coefficients for retention in the lungs, uptake in the gastrointestinal tract, or absorption through the skin.

With a continuous flux of a foreign chemical into the organism, the term `dose rate' represents a satisfactory measure of exposure, irrespective of the route by which the contact occurred; the term is also applicable when two or more routes of entry occur. In real-life situations, however, the information is often insufficient for such computations, and fractional data from each route are used. The `internal dose' or `dose rate' may be measured as concentration(s) of the chemical and its metabolites in tissues and body fluids, including blood and urine. In principle, the procedure, called `biological monitoring', represents the simplest way by which dose rates received by various routes can be integrated, depicting total exposure. A link between the data obtained from biological monitoring and those of external environmental monitoring is often secured by studying respective correlations obtained under field conditions. Such investigations are possible, however, with only one route of entry. For multiple exposure routes, individual transfer coefficients are needed.

Other important factors to interpret biological monitoring data in relation to external exposure are duration of exposure and time interval after termination of exposure. Chemicals in general and pesticides in particular differ dramatically in their toxicokinetic properties, which are manifest as the presence or absence of accumulation in the body during repeated and prolonged exposure, and as slow versus rapid disappearance following cessation of exposure. Biological monitoring may be based on the measurement of either the unchanged substance or selected metabolites in a particular biological medium. For instance, DDT exposure may be measured as `total DDT' in blood or fat tissue (biopsy or autopsy)e.g., during occupational high exposureor as the ultimate metabolite, DDA, in urine. In practice, any such determination may be useful, with the method chosen based on availability.

* This section was prepared by B. Goldstein, J.  Moore, J. Parizek, J. Pietrowski, M. Rostker, and S. Saunders

Contemporary trends in biological monitoring of organic chemicals focus on the products of microsomal metabolism. As a result of activation in the system of cytochrome P₄₅₀-dependent oxygenating enzymes, several organic chemicals yield `active metabolites' such as epoxides or oxidation products of N-hydroxyderivatives of aromatic compounds. These active metabolites may bind covalently with macromolecules, such as nucleic acids or proteins, resulting in relatively stable `adducts' that may be measured selectively.

Two separate lines emerge in these trends, of which one centres on adducts with DNA (measured in the white blood cells) and the other on adducts with haemoglobin (measured in the red blood cells). This type of monitoring requires sophisticated methodologies such as ³²P-post-labelling linked with thin layer chromatography or mass spectrometry of the hydrolysis products of the adducts. Therefore, the development of such methods is largely confined to a few advanced and well-equipped laboratories. For pesticides, the best known example of such monitoring has been the fumigant ethylene oxide, which yields DNA adducts.

One important property of macromolecular adducts is their relative stability. For instance, adducts to haemoglobin persist in a red blood cell throughout its life-span (approximately 4 months); therefore, the readings are likely to increase greatly with the duration of exposure. This situation offers a convenient way of integrating individual doses received over time, yielding the average daily exposure level.

2.2.3.2 Patterns of exposure

In practical terms, the collection of information relevant to human exposure to pesticides can be assigned to two broad categories: environmental (or external) measurements such as air, water, and diet; and biological (or internal) measurements. Biological measurements can be conveniently divided into actual measures of the pesticide (or metabolites) or physiological variables. The latter are surrogate measures that must have an established causal relationship with a chemical of interest to be relevant to exposure.

In many situations, pesticide exposure results from multiple sources. In an agricultural setting, for example, workers may receive dermal and inhalation doses from pesticide spray application. In the home, exposure can occur through ingestion of unwashed garden produce that contains pesticide residues and from inhalation of indoor pest control sprays. Dietary exposure to pesticides occurs for almost all members of the population.

Depending upon the use pattern, one or two exposure routes may be of dominant importance and account for a majority of exposure. Nevertheless, secondary or relatively minor routes may also be influential in the summation of total exposure.

Designs for exposure studies must consider statistically correct sampling schemes, and include sufficient numbers of samples to account for loss or attrition, control or baseline measures, and potentially small but significant variations in exposure patterns. To study cholinesterase inhibition, it is frequently important to have baseline or control samples for comparison with those from potentially exposed individuals. Multiple samples are useful to establish ranges of biological variability and to reduce uncertainties surrounding the biological significance of measurements. Many blood variables exhibit cyclic and diurnal patterns; multiple samples taken through time help to distinguish among pesticide exposure-induced changes and normal variability.

Ultimately, total exposure is the relevant measure for assessment. Integration of all routes of exposure, along with characterization of the frequency and duration of exposure, leads to a complete exposure assessment. Knowledge of the endpoint or effects of exposure is important to structure study designs that are statistically strong and relevant to the nature of the exposure. At present, many methodologies are still primarily focused upon single-route exposures. To some extent, an approximation of total exposure is calculated via summation of exposure routes. While this technique is a good start toward total assessment, better understanding of the ways to integrate different routes of exposure is needed. Growing use of carefully designed monitoring and assessment studies which utilize repeated sampling regimes, and which are sensitive to small variations in patterns, is one way to promote improved dosimetry.

2.2.3.3 Specimen selection

A variety of human tissues and excreta can be sampled for determination of exposure to pesticides.

Chemical characteristics

The chemical characteristics of a xenobiotic such as a pesticide are of primary importance to determine how the compound and its metabolites are distributed in and excreted by the body. A major characteristic is lipophilicity (hydrophobicity), which to a large extent governs distribution in adipose tissue. Lipophilic compounds tend to be stored in fatty tissues, and are not excreted into the urine unless metabolized to a relatively water soluble form (e.g., glucuronides). Excretion of fat soluble compounds may occur through the bile duct, but in most instances such compounds are reabsorbed through the enterohepatic circulation. Accordingly, highly lipophilic compounds that are also poorly metabolized, such as DDT, may persist in the body for many years, and may be measurable only through sampling of adipose tissue; however, water soluble compounds or metabolites may be immediately detectable in urine.

Other important chemical characteristics include the presence of specific chemical moieties capable of being acted upon by the human metabolic machinery. In some cases, metabolism can produce a specific detectable product in blood or urine; in others, it can act too rapidly and completely degrade a xenobiotic, so that it cannot be readily detected in biological media. Varying rates of metabolism resulting from differing genetic or environmental backgrounds can produce wide variations in the relationship between measured metabolites and external dose, thereby complicating exposure assessment.

The choice of environmental or personal monitoring approaches often depends on technical feasibility. Protecting the worker by measuring a chemical in exposure media is particularly problematic for pesticide applicators and agricultural workers, because exposures usually occur outdoors. Thus, the standard approach in an enclosed workplace of using air measurements is relatively impractical. Furthermore, worker exposure to pesticides is often through skin contact, which is difficult to measure under any circumstance. These problems, as well as the technical problems related to human monitoring, are detailed elsewhere in this document.

Ethical aspects of human monitoring sampling approaches

Some sampling approaches are far more feasible than others in terms of technical difficulty. However, a major overall principle in selecting a human monitoring approach is the absolute need to do no harm to the person undergoing the procedure. This principle restricts the sampling media to excreta and to those tissues that can be sampled with little or no risk to the subject. In the case of blood, venipuncture is considered to be a small risk, as long as proper sterile techniques are followed and the amount of blood withdrawn is appropriate for the age, size, and health status of the individual. If analytical procedures permit, use of capillary blood obtained by finger puncture should carry less risk. Willingness to undergo human monitoring often depends on perceived risk, discomfort, or inconvenience of the procedure. Implicit in this discussion is that any sampling of an individual must be to that individual's direct benefit. Fully informed consent must always be obtained.

Multi-media and multi-pesticide considerations

Pesticide exposures do not always occur through one medium. A single field pesticide application may result in a body burden through (a) inhalation during spray application, (b) passage through the skin resulting from direct skin contact or contact with treated soil, (c) ingestion in food or water consumed by field workers, and (d) ingestion of contaminated soil particularly by children. Exposure to multiple pesticides occurs when workers use mixtures of pesticides on different crops, and when the consumer ingests food purchased at the supermarket or uses pesticides while gardening as a hobby. These different, and sometimes unexpected, routes of single or multiple exposures must be considered when selecting specimens for study. Furthermore, human monitoring is necessarily the required approach for significant multiple exposures.

Specimen selection should also consider the various inerts, solvents, and other ingredients in pesticide formulations, since they are almost always a composite of several chemicals. Pesticide formulations may contain the following:

1. One or several chemicals that possess the desired pesticidal property, i.e., chemical `A' for destruction of broad-leafed plants and herbicide `B' for grassy plants;
2. Chemicals that enhance the pesticidal properties, such as surfactants or stabilizers;
3. `Trace contaminants' contained in the base chemicals used in the synthesis of the pesticide or formed during the synthetic process, and

4. Solvents or carriers that facilitate the application of the pesticide; these materials often constitute 9099 per cent of the total volume of pesticide. 

Solvents may possess toxic properties that are of human or ecotoxicity concern. Perchloroethylene is an example of the former, toluene one of the latter. Trace contaminants may include nitrosamines and chlorinated dibenzo- p-dioxins.

Regional differences

A number of cultural aspects determine specimen selection. In certain parts of the world, obtaining a blood specimen is particularly difficult because of local beliefs about the importance of blood. In less-developed countries, there might be regions in which it is not technically feasible to obtain, handle, or store certain types of specimens. There are also areas of the world in which legal and administrative burdens are sufficiently imposing as to hinder certain types of human monitoring.

2.3.3.4 Specific media

The following media are addressed briefly to indicate their respective merits for biological monitoring: urine, hair, blood, adipose tissue, and breast milk.

Urine

Urine is a useful medium for human monitoring of water soluble pesticides or metabolites. An example is chlordimeform, which can be monitored through measurement of its urinary metabolite, chloro-o-toluidine. Several countries require such monitoring for workers using chlordimeform. The profile of urinary metabolites of a single pesticide can sometimes be helpful in distinguishing the pattern or time of exposure.

The advantages of using human urine to determine exposure include its ready availability, the non-invasiveness of sampling, and, in some countries, reduced legal or cultural impediments. A major disadvantage is the need for refrigerated storage, to prevent bacterial contamination. The extent and timing of urine collections will often depend on the exposure pattern and toxicokinetics of a pesticide.

The use of human urine as a means of measuring exposure to pesticides should be encouraged increasingly. Enhanced understanding of the toxicokinetics of environmental chemicals in recent years permits increased confidence in extrapolation of measurement in various body compartments to the initial dose. This knowledge has been accompanied by rapid advances in analytical chemistry to detect ever smaller amounts of parent compounds and their metabolites. 

Hair

Hair is a useful matrix to detect the presence of metals and metaloids such as mercury and arsenic, each of which has been an active ingredient of some pesticides. Hair can be obtained easily and non-invasively. Furthermore, one can take advantage of its relatively slow growth to differentiate from chronic exposure by analysing different hair segments.

Nevertheless, a number of problems exist with the use of hair for the determination of exposure. Foremost is the need to distinguish between substances that have been incorporated into the hair matrix during its formation and those that have been added to the hair as part of grooming preparations or through the deposition of dust. Great care must be taken to clean and remove added substances, a process that can be technically difficult.

Blood

Monitoring humans exposed to pesticides is frequently performed through blood sampling. Both direct measurement of the pesticide or its metabolites, and indirect approaches through the measurement of a biological effect of a pesticide (e.g., acetylcholinesterase (AChE) activity) are used.

Blood is a tissue containing cellular and plasma components that can be assayed separately or together. It contains components that will attract both hydrophilic and hydrophobic substances. Thus, not only will water soluble compounds travel readily in the plasma, but highly lipophilic substances can be accommodated in such hydrophobic matrices as the red blood cell membrane, serum albumin, and lipoproteins.

Recent interest in biomarkers has led to the study of potential macromolecular adducts of environmental chemicals using human white blood cells as a source of DNA and albumin or red cell haemoglobin as a source of protein. Haemoglobin is of particular interest, because it is present in very large quantities in routine blood samples, is readily purified, and has a normal lifetime in the blood of approximately 120 days, thereby allowing it to be used to integrate exposure over 4 months and, by density-red-cell-age-separation techniques, theoretically permitting detection and dating of a brief exposure.

By contrast, the term `biological monitoring' refers to the measurement of an effect in the indicator medium, especially in the case of organophosphate pesticides. So far, the only effect utilized has been the inhibition of the AChE in the red blood cells. This measurement is a special example where the effect of real concern (inhibition of the AChE at nerve synapses) is reflected also in the inhibition of the enzyme contained in the blood, which serves as the indicator medium.

When interpreting the results of such an `effect monitoring', one has to distinguish it clearly from data on `chemical monitoring'. Both phenomena differ in their toxicokinetics; therefore, a simple relation can be found only in the steady state of the AChE activity. The inhibition of AChE activity occurs to varying degrees in response to equal doses of varous OP pesticides. Therefore, this method is used, for practical purposes, as a self standing (i.e., independent) measure of exposure.

Adipose tissue

Analysis of body fat samples is a valuable approach to characterize the body burden of lipophilic compounds with relatively long half-lives in the body, such as chlorinated hydrocarbon pesticides. While occasionally used for diagnostic purposes or as part of a research protocol, fat biopsy is too invasive a procedure to be acceptable for routine monitoring of work populations exposed to pesticides.

Adipose tissue banks have been found to be particularly useful for trend analysis, such as following the body burden of DDT, DDE, and PCBs over time when regulatory controls were imposed. These banks, for the most part, depend on obtaining a fat sample during an operative procedure or autopsy. Archiving tissues is also valuable for future use as new analytical techniques become available. The sampling strategy must be responsive to the specific aims of the fat bank; i.e., for trend analysis, one needs a sampling approach that considers age, sex, health status, and occupation, and one that does not change from year to year. As with any tissue bank, archival integrity is as crucial as the storage process. To be used successfully, adipose tissue banks need to combine archival expertise with the ability to develop and test relevant hypotheses.

Breast Milk

Various lipophilic substances have been shown to be eliminated from the body in breast milk, thereby posing a potential health risk to the nursing infant. Breast milk concentrates body stores of lipophilic substances, acting to displace these stores from the mother to the infant who, with much less total body fat, may well have a higher fat concentration of the unwanted substance. In essence, monitoring of breast milk permits estimation of pesticide exposure to both mother and infant. Water soluble pesticides might also appear in breast milk, but the mobilization and concentration phenomena for lipophilic substances would not be anticipated.

Nursing mothers may be exposed to pesticides in the food they eat or through occupational exposures that occur before or during pregnancy and lactation. In many of the lesser developed countries and in migratory farm workers in developed countries, mothers with babies may actively participate in agricultural efforts including pesticide use; thus, the infant in the field may be at risk both from direct exposure and, if the pesticide is lipid soluble, from breast milk as well. Presently, techniques to obtain breast milk are relatively straightforward, although care must be taken in storage of samples. The usual approach is to relate the concentration of the chemical to milk fat content rather than to total volume. Unfortunately, analytical techniques to measure fat soluble substances, including pesticides, in mother's milk are now technically difficult and costly. A high priority should be given to development of monitoring techniques useful for the detection of pesticides in the breast milk of nursing mothers engaged in agriculture.

The value of assessing breast milk in exposure monitoring for pesticides should be considered in the context of the overall benefit of breast feeding for infant development.

2.2.3.5 Occupational exposure

Occupational exposure to agricultural chemicals is generally recognized as having the potential for high doses to the individual, because of the magnitude and frequency of contact. Thus, the number of acute illnesses among agricultural workers reported to national health agencies is a significant public health matter.

The basic approach to estimate occupational exposure is passive dosimetry, biological monitoring, or a combination of the two approaches. Passive techniques are generally used to satisfy regulatory requirements rather than as a means of personal exposure monitoring. As such, these techniques are useful to establish a likely dose under a defined set of field conditions; however, they are not commonly used to monitor exposures of workers on a continuing basis.

Biological monitoring approaches have been discussed fully in previous sections. These techniques may be applied to evaluate occupational exposures as a means of establishing doses of a compound; however, the manner or route of exposure cannot be established by this method alone. The newer immunoassay methodologies hold promise for use as personal monitoring devices. One approach utilizes a small card that, using Enzyme-Linked ImmunoSorbent Assay (ELISA) techniques, can measure pesticide concentrations in virtually any medium. The general approach includes having an individual carry a card containing appropriate antisera and reagents; after exposure to the pesticide, the card changes colour; the colour may then be compared to that on a chart to estimate the concentration of pesticide in the sample. Such cards have already been developed for parathion and paraquat, two chemicals which account for a significant number of acute poisonings in the field. This approach could be used to quickly monitor residues on skin or in urine, and hence to estimate doses to individuals in the field. This approach may prove especially useful in developing nations where access to analytical equipment may be limited.

Occupational exposures to pesticides can be reduced through protective clothing, engineering controls, and modification/refinement of application techniques. Absorption through the skin, in particular the hands, is the major route of exposure for most pesticides. Depending on the specific pesticide and occupational activity, estimates of exposure through the skin of an unprotected hand range from 25 to 98 per cent of total dermal exposure. The use of protective gloves would greatly reduce occupational pesticide exposure; however, care must be taken to select a glove that is impermeable to the pesticide. For pesticides that are volatile or form aerosols, the use of respirators may be necessary to reduce exposure. However, these devices are limited by the requirement for periodic maintenance and a lack of compliance among workers because of discomfort and heat.

In tropical climates, full protective gear may be impractical because of the potential for heat stress in the worker. In such instances, a reduction in exposure can be achieved only through engineering controls and improved application techniques. The use of pre-weighed packages reduces exposure by eliminating the need to handle large volumes of pesticide while weighing out or measuring the amounts necessary for application. In addition, the potential for spills is reduced. The use of emulsifiable concentrates rather than wettable powders reduces inhalation exposure, as does pumping rather than pouring the pesticide from the mixing tank into an application apparatus.

The use of closed-cab ground application systems, rather than airplane or airblast, can also reduce exposure to the applicator and reduce drift onto non-target areas. When airplane applications are used, mechanical rather than human flaggers should be employed to direct the spraying.

Since plants carry a negative electrical charge, the use of an electrostatic device to create an airblast mist that carries a positive electrical charge can cause an electrical attraction between the pesticide mist and plants. This attraction minimizes drift, improves application efficiency and coverage, and ultimately reduces the volume of pesticide required to achieve an effect.

Reducing worker exposures through engineering controls and improved application techniques also leads to a reduction in the overall environmental impact of the pesticide application specifically, less drift, fewer spills, and a reduced potential for accidental exposure of other unprotected individuals. Improved techniques for personal monitoring, such as the currently evolving immunoassay methodologies, are desirable to ensure the efficacy and safety of engineering controls and application techniques.

2.2.3.6 Non-occupational exposure

In some instances, pesticides used in and around the house can result in meaningful exposure. The practice of treating houses for the control of structural pests has resulted in persistent elevated exposure to the cyclodiene pesticides aldrin/dieldrin and to chlordane/heptachlor.

Periodic application of insecticides for control of vermin is a potential source of exposure to persistent chlorinated hydrocarbon and other products such as chlorpyrifos.

Small children may receive the more significant levels of exposure as a consequence of repeated contact with room perimeters, carpets, and soil. In addition, hand to mouth manipulations are common at this age, as is the tendency to taste, chew, and ingest objects. Children of all ages also have the potential for more frequent exposure to pesticides in the home yard, particularly treated lawns or treated soil.

The casual use of pesticides by home owners on the lawn, in the flower garden, or in the vegetable plot, must be considered in determining sources of pesticide exposure. Failure to heed instructions for use of protective clothing, preparation of correct concentrations, and proper disposal can result in levels of exposure equivalent to those present in large-scale agricultural applications. Such lax practices may also result in increased exposure residues on foods harvested from the home garden.

2.2.3.7 Dietary exposure

An ideal system to estimate dietary exposures to pesticides should consist of two main components: sufficient data on food consumption that accurately reflect the diverse eating habits and patterns present in a population and its subgroups; and accurate data on the concentration of pesticide residues present in food at the time it is eaten.

Food consumption data used to derive dietary exposure estimates should reflect the diverse eating habits of a population. In some cases, such diversity can be accounted for by using safety factors or exaggerated estimates of residues in food. In such cases, the use of a single population average for food consumption in combination with exaggerated residue estimates will provide adequate protection for all individuals. The result of using this approach is to greatly overestimate dietary exposures.

In other cases, however, it is desirable to obtain the most accurate estimate of exposure possible because of concern for potential toxicity in a particular population subgroup. For example, infants and children are known to consume more food per kilogram of body weight than do adults. Using a population average for food consumption will always under-estimate consumption in children and generally over-estimate consumption in adults. In addition, infants and young children tend to receive a greater portion of their total diet from a small number of foods than do adults. This restriction of the diet to a few foods, mainly milk products and certain vegetables and fruits, can also magnify pesticide exposures in infants and children, since, if these foods have been treated with a pesticide, a greater portion of the infant diet will contain pesticide residues than that of an adult. Since children are known to be uniquely sensitive to some forms of toxicity (e.g., lead neuropathy), it is important to have accurate consumption data for this group to assess the impact of pesticide residues in their diet. Similarly, for reproductive toxicants, accurate data on food consumption must be obtained for women of child-bearing age.

The toxicity produced by a pesticide can be caused by either a single exposure or multiple exposures over time. Aldicarb is a carbamate insecticide that can produce toxicity after a single exposure, as has been demonstrated by incidents involving the misuse of this chemical on watermelons and cucumbers. To protect against this type of occurrence, it is necessary to have data on the amount of a particular food consumed at one time. In addition, within any population subgroup, a distribution of consumption will exist for any particular food. Some approaches to regulating pesticide residues in food use the 90th percentile of consumption to protect the individuals who eat more food than the average for the population. For chemicals that produce toxicity after multiple exposures, data on average food consumption over time would be more relevant.

Pesticide residues occur in food as a result of commercial agricultural practices and, in some cases, as a result of home gardening. Commercial agricultural practices are generally regulated by national governments, and are influenced by international groups such as the WHO/FAO Joint Meeting on Pesticide Residues (JMPR). These groups generally recommend Acceptable Daily Intakes (ADIs) and Maximum Residue Limits (MRLs), which are the upper limits of pesticide residue expected on raw agricultural commodities consistent with good agricultural practice.* In general, these MRLs are mainly for enforcement purposes, that is, to assure that pesticides are not used in excessive amounts. However, since the MRL (a tolerance) is an upper limit, it is not a useful value for calculating actual dietary exposures.

*In the US, this value is legally defined as a 'tolerance' and is established by the Environmental Protection Agency (EPA)

An accurate estimate of exposure should reflect the effects of washing, peeling, processing, and cooking on pesticide residues. In general, these steps will produce a large decrease in the amount of pesticide present in food at the time of consumption, compared to the amount of residue present in the field. In certain other cases, a pesticide may actually concentrate during processing, or it may be converted to toxic metabolites. The ethylene bisdithio-carbamate (EBDC) class of fungicides is known to form the toxicant ethylene thiourea (ETU) during the processing of food treated with these fungicides. Similarly, the growth regulatory daminozide will form unsymmetrical dimethyl hydrazine (UDMH) upon processing (e.g., conversion of apples into apple sauce). In these cases, sampling of field residues alone would not provide an accurate estimate of exposure at the time of consumption. Rather, measurements of field residues combined with studies on the effect of processing and cooking are necessary to accurately predict final residue concentrations at the time of consumption. The residue estimate should also reflect the extent of use of the pesticide. An MRL may be established for a chemical on tomatoes; however, every tomato grown and marketed will not be treated with that chemical.

In addition to residues on fresh and processed produce, pesticide residues may occur in meat, milk, and eggs as a result of feeding items treated with pesticides to domestic cattle and poultry. In certain cases, pesticide residues can occur in meat and milk at significant concentrations. In one instance, dairy cattle were inadvertently fed grain that had been treated with the insecticide heptachlor, resulting in the appearance of heptachlor in milk. In another instance, pineapples treated with heptachlor produced no appreciable residues in the edible fruit; however, when byproducts from a pineapple cannery were fed to cattle, heptachlor residues were detected in milk. In many countries, wastes from food processing are routinely used as cattle feed, resulting in a significant potential for producing secondary residues in domestic food animals. This issue is especially important where residues occur in milk, since milk products form a significant part of the diet for infants and children, who will also receive an exaggerated dose because of their higher rate of food consumption relative to body weight.

In many countries, food has been sampled at the consumer level and analysed for pesticide residues as a means of estimating dietary pesticide exposures. Such techniques include `market basket surveys', which monitor residues in food purchased from stores and markets, and `dual diet studies' in which a portion of prepared food is analysed prior to consumption. These approaches provide the most accurate data on pesticide residues at the time of consumption, but are limited by the cost of obtaining and analysing large numbers of samples necessary to achieve statistical validity.

For most individuals, pesticide residues in the diet occur as a result of agricultural practices. However, some individuals may receive significant dietary exposures from other sources. The organochlorine pesticides, which are quite persistent in the environment, will concentrate in fish because of bioaccumulation. Commercial and sport fishermen, who regularly consume such fish from lakes or rivers that have been contaminated with these pesticides, may receive a significant dose of the chemical. Hunters who consume wildfowl may receive similar exposures. In one instance, wild geese were discovered to contain significant residues of the organochlorine endrin as a result of eating grain from fields treated with this pesticide. Some birds died from this exposure; surviving birds that were later eaten caused human illness.

The overall impact of such episodes is difficult to assess. Because few individuals within a population obtain their food from these `wild' sources, statistical population surveys of food consumption will not provide reliable estimates of consumption. However, fish and wild game can be a source of significant hazard to the individual, and should not be overlooked in an assessment of the impact of environmental contamination.

Home gardening, although not a major source of food in most households, may also present a significant hazard to an individual. Generally, the same laws and regulations apply equally to home and commercial agricultural practices. However, some home gardeners, because of a lack of knowledge or through carelessness, have occasionally introduced significant pesticide residues into a food, resulting in illness.

Finally, drinking water is a source of dietary pesticide exposure that has become of concern in many nations. In agricultural areas of high pesticide use, sampling of ground- and surface-water supplies has generally revealed the presence of pesticides. The problem may become especially serious in areas with sandy soil, which is easily penetrated by pesticides, and/or areas with relatively high water tables. For example, in Suffolk County (New York), aldicarb was found in about 25 wells sampled at concentrations of up to 515 p.p.b. In California, a state with pesticide manufacturing facilities and a large number of agricultural operations, more than 50 pesticides were detected in the ground water of 23 counties. One pesticide, DBCP, was found in over 2000 wells (25 per cent of the total number sampled), with the highest concentration of 1.24 p.p.b. near a DBCP manufacturing facility.

Efforts are currently under way in many nations to monitor ground- and surface-water for pesticides and other chemicals. The monitoring of water for pesticides must consider the largely seasonal application of pesticides. For example, agricultural runoff may produce relatively higher concentrations of herbicides early in the growing season, since pesticides are frequently applied to bare ground, which has a greater potential for runoff. The utility of such monitoring efforts is clear to assess hazards to a population resulting from dietary pesticide exposure. For some subgroups with high water consumption, such as infants fed formula prepared with water, such data are vital to an accurate assessment of dietary exposure.

2.2.4 ECOSYSTEM MONITORING

Of the numerous issues confronted in the monitoring of something as diverse, variable, and complex as an ecosystem, the most difficult is sampling. Adequate principles and guidelines exist for sampling human environments for epidemiological studies; nothing comparable exists for non-humans.

Because of the concern for incidental dietary intake by humans of pesticides, considerable attention has been given to developing accepted procedures for sampling foodstuffs for chemical analysis. Guidelines, indicating substantial agreement over procedures to sample foods, exist in international commerce. The Codex Committee on Pesticide Residues has been prominent in this quest for international agreement and standardization. No such agreement has been reached with regard to ecosystem sampling. The OECD program on wildlife sampling and monitoring partially dealt with the matter, but no guidelines or procedures emerged from this study. Environmental sampling has also been the subject of an IUPAC study.

The paucity of guidelines on ecosystem monitoring, the natural complexity and variability of ecosystems, and the diversity of monitoring methodology makes comparisons of result among investigators virtually impossible. Many existing environmental sampling procedures are carryovers from the randomized, unbiased sampling procedures employed in agriculture to measure endpoints such as crop yields; standard textbooks are devoted largely to such sampling techniques. Environmental sampling requirements are similar to those of epidemiology, namely, reliance on biased non-random sampling, which applies to both spacial distribution and time. For instance, the impact of aerial spraying of a forest on NTOs is likely to be maximal at treetop height and just after spraying; similar considerations apply to the use of pesticides to control biting flies in a fast-flowing river. The failure of some laboratory model systems to extrapolate results to the field may be largely attributed to the failure to include sampling criteria.

The basic need for guidelines for environmental sampling would best be met by a workshop organized by an international agency such as UNEP or OECD. The membership of such a workshop should include a broad array of expertise with particular representation among those specialists who make the greatest use of monitoring data.

2.3 RECOMMENDATIONS

The following detailed recommendations supplement those presented in Chapter 1 of this Joint Report.

1. The complexity of the models should be limited to what is required, given the objectives of the study.

2. Research should be oriented mainly toward the development and validation of predictive descriptions of the key processes that influence behaviour, including sorption, photolysis, and biodegradation, rather than on models per se, as well as on the validation and improvement of the quality of the databases.

3. Research to test the convergence between predictions and field results should be focused on the identification of those refinements that are essential for improved forecasting.

4. Methodologies that correctly integrate exposure from all routes into a total dose need to be developed and validated. A generally accepted and validated procedure does not exist. Implementation of this recommendation is expected to lead to a better recognition of the relative importance of individual exposure routes in specific pesticide uses.

5. The majority of basic exposure data has been developed in the temperate climate zones of the world. In many instances, the major applications (and toxic effects) of pesticides occur in the semitropical and tropical zones. Since some physiological and biological properties and rates are dependent upon temperature, humidity, and other environmental characteristics, a substantial need exists to adjust for, and verify, the basic data developed in one climate zone before use in other zones.

6. Simple, reliable, and economic measures to indicate level(s) of exposure in the field are needed to affirm the effectiveness of exposure control techniques in practical situations.

7. Advances are needed in the methodology to assess human exposure to pesticides through biological monitoring. A better understanding of the basic biological mechanisms governing the interaction between pesticides and human tissues would allow the extent of exposure to be inferred confidently from the biological endpoint or understood from toxicokinetic data relating to parent compound or metabolite levels in biological fluids. Improved analytical techniques to determine lower concentrations of a compound, its metabolite, or its macromolecular adduct should be sought.