1.2.1. Trends in worldwide use of insecticides

A wide range of insecticides, fungicides, molluscicides, bactericides and herbicides, including fumigants, are used, mainly in developed countries but also (and increasingly so) in developing countries. Organochlorine insecticides are still used in the latter but are being replaced gradually by organophosphorus, carbamate and pyrethroid insecticides. Another important use for insecticides is in the control of ectoparasites.

The pests responsible for the greatest losses are locusts, but effective, inexpensive insecticides to control the massive locust infestations that plague some parts of the world have yet to be developed. The crop on which most insecticides are used is cotton.

The most common formulations are emulsifiable concentrates and ultra-low volume concentrates. In urban areas, organochlorine insecticides are now little used; they have been replaced by pyrethrins, pyrethroids and organophosphorus insecticides, such as chlorpyrifos, dichlorvos, fenitrothion, fenthion, malathion and temephos. The worldwide requirements for pesticides in urban public health programmes worldwide are substantial, the annual cost being over US $ 100 million. In 1980, about 50 000 tonnes of pesticides were used in public health programmes in developing countries. It was estimated that such programmes account for about 10% of total pesticide use, the remainder being used mainly in agriculture.

Insecticides are used on a number of crops of different relative importance for world agricultural production (Table 1). Herbicides are used mainly on corn and soya beans, insecticides mainly on cotton and horticultural crops and fungicides mainly on horticultural crops and wheat. Worldwide use of insecticides in 1985 was approximately 29% in Japan and the Far East, 23% in the USA, 12% in western Europe, and 36% in the rest of the world; the corresponding values for herbicides were: 10, 46, 21 and 22%, respectively.

Table 1.

The insecticide market by crop in 1985 (million US $ in 1984).

1.2.2. Application principles and techniques

Methods of application of pesticides, including insecticides, have been reviewed (Haskell, 1985). The aim of insecticide application is to distribute a small amount of active ingredient to the appropriate insect with minimal contamination of non-target organisms. The diversity of possible targets—e.g., insects, plants, soil, walls of dwellings—necessitates a variety of application techniques, which can be summarized in five groups:

1. release or propulsion through the air to the target either
   • – in the solid state as dusts or granules, or
   • – in the liquid state as sprays;
2. application directly to or injection into the plant;
3. injection into the soil;
4. release into irrigation water; or
5. release into the air with diffusion to the target (fumigation).

Hazards due to drift and inhalation of particles less than 30 µm in diameter have resulted in a decline in the use of dusts, except for treating seeds and small seedlings at the time of transplanting, for which specialized equipment is available. Seed treatment is ideal for protecting young plants with minimal quantities of toxicant, but as phytotoxicity can be a problem the use of granules accurately placed alongside seeds at sowing has increased. Equipment is also available for spot treatment of individual plants, and granules are often broadcast, sometimes by hand; but this requires a higher dose than other application techniques (Matthews, 1985).

Special equipment is needed to meter granules, so the majority of pesticides are applied as sprays. The volume of spray liquid applied varies with the size of target and on whether discrete droplets or a complete film of spray is to be distributed on the target. While 50 to more than 1000 litres/ha may be applied to field crops, bushes and trees, as little as 5 litres/ha of pesticides may be applied using newer ultra-low volume spray techniques (Matthews, 1985)

Less than 0.1% of the applied dose may reach insect pests in a field crop treated with a foliar spray, whereas up to 30% of an applied herbicide penetrated experimental plants sprayed ma greenhouse (Matthews, 1985). The low efficiency of sprays has been due largely to the wide range of droplet sizes emitted by traditional spraying equipment; control of droplet size is designed to suit the intended target and the method of application (Table 2)

Table 2.

Optimal droplet sizes.

The five basic classes of ground application equipment include hydraulic sprayers, air sprayers, foggers and aerosol generators, power dusters and hand-held equipment (Anon 1981). In hydraulic sprayers, the pesticide is delivered under pressure by a pump to one or more nozzles. The type of nozzle regulates droplet size and spray pattern. Hydraulic sprayers are of four basic types:

1. Multiple-purpose sprayers provide versatility for a variety of problems. Spray pressure is adjustable; tank size ranges from 190 to 750 litres; sprayers are skid- or wheel-mounted and powered by auxiliary engines or a power take-off; spray is dispensed through a hand-gun or field boom.
2. Small general-purpose sprayers are useful for small jobs, for instance in greenhouses, large gardens and golf courses. Tank size ranges up to 100 litres; power is provided by a small engine that furnishes a wide range of pressures (50–500 psi [3.5–35.2 kg/cm²])- spray is dispensed through a hand-gun or short boom, and the sprayers are usually mounted on a hand-operated cart or attached to a garden tractor.
3. Low-pressure, low-volume sprayers are commonly used on crops. They can be mounted directly on equipment or are equipped with wheels; tank size ranges up to 950 litres; power is usually provided by a power take-off, but may be supplied by an auxiliary engine; operating pressure is up to 100 psi [7.0 kg/cm²] and spray is dispensed through a field boom.
4. High-pressure, high-volume sprayers are used by fruit growers and truck farmers in order to obtain good penetration and coverage in tall growing trees and dense crop growths. These sprayers are essentially the same as multiple purpose sprayers except that larger engines provide up to 1000 psi [70.3 kg/cm²] and tank size ranges up to 2300 litres (Anon., 1981)

Air sprayers (also known as ultra-low-volume, concentrate blower, airblast and airmist sprayers) are used for spraying orchards, large shade trees and field crops. Pesticides are applied in concentrated form, using relatively small volumes of water, in contrast to hydraulic sprayers. A low-volume pump delivers the liquid spray under low pressure to the fan, where it is discharged into an air stream in small droplets by a group of nozzles or shear plates. Pump pressures range from 50 to 400 psi [3.5–28.1 kg/cm²], and fans deliver 5000–25 000 ft³/mn [2.4–11.8 m³-Vs] or air velocities of 100–150 mile/h [160–240 km/h] (Anon., 1981; Joyce, 1985).

Foggers or aerosol generators are designed primarily for control of mosquitoes and flies in large buildings, parks, resorts and communities. These machines disperse fine particles of pesticides into the air, as fogs or mists, where they remain for a considerable time. Fogs and aerosols are produced by either thermal or mechanical methods or a combination. Aerosol equipment is not practical for application of most agricultural pesticides because of its tendency to create drifts (Anon., 1981).

Power dusters are run by engine or power take-offs. Like airblast sprayers, dusters also utilize air streams from a centrifugal fan to carry the pesticide to the target area. They may have single or multiple outlets. Dusters may be impractical for application of some pesticides because of drift hazard (Anon., 1981).

Hand-held equipment is designed primarily for application of pesticides in small areas; this type of equipment includes hand-pump atomizers, aerosol dispensers, compressed air sprayers, knapsack sprayers and dusters. The hand-pump atomizer has a hand-operated pump to force an air stream over the tip of a siphon tube; pesticide is sucked from the tube and atomized in the air stream. These sprayers were commonly used to control flying insects in houses but have been almost completely replaced by aerosol dispensers. Aerosol dispensers are probably the most common type of applicator for household pest sprays. The pesticide and a propellant, usually freon, are forced, under pressure, through an atomizing nozzle Compressed air sprayers are designed to hold 4–12 litres in the tank. A hand pump is used to pressurize the tank and deliver the pesticide, under pressure, to the nozzle. Spray patterns and droplet size can be regulated by nozzle type. Solutions, emulsions and suspensions of pesticides can be used at pressures of 30–50 psi (2.2–3.5 kg/cm²]. Knapsack hand sprayers are carried on the back and usually have a capacity of 20 litres; a hand-operated piston or diaphragm pump provides the pressure (30–100 psi [2.2–7.0 kg/cm²]) to expel the pesticide Duster hand sprayers range from small self-contained units to those mounted on wheelbarrows. Air velocity for dispensing the dust is created by a plunger, hand crank or belt attached to a fan or blower (Anon., 1981).

The subject of aerial spray equipment and accessories is complex; however many aspects of aerial application are similar to ground application. For example, sprayers are basically constructed of the same components. Several classes of aircraft may be used for the application of pesticides, including high-wing monoplanes, low-wing monoplanes, biplanes, multi-engine aircraft (used extensively in forest and rangeland application) and helicopters. Helicopters have some advantages over fixed-wing aircraft: operation at slower speeds; increased safety; improved accuracy of swath, coverage and placement of chemical; and operation without airport facilities. Pesticides are generally released at greater heights than from conventional sprayers (Anon., 1981).

Application equipment can be constructed for dispersing dry or liquid pesticides. Dry chemicals are dispensed from fixed-wing aircraft primarily by ram-air spreaders and spinners. In a ram-air spreader, dry materials are metered from a hopper into the propeller slip stream. The fact that ram-air systems cannot spread materials in a wide swath led to the development of spinners, which consist of spinning vanes mounted under the hopper that throw material outward in a uniform pattern. The use of spreaders and blowers can nearly double the swath width. In helicopters, two types of dispenser are used: a blower driven by the engine forces dry material from two side tanks and out of short booms, but the material may be spread using spinners instead of the boom; or a single hopper can be suspended on a cable and dry material is dispensed using spinners (Anon., 1981).

Two types of liquid spray systems exist for fixed- and rotary-wing aircraft: the pressure type, in which the spray is applied under specific pressures; and the gravity-feed type, in which the flow of spray solution from the tank dispersion unit relies on gravity. Swath widths of 12–18 m, in the application range of 1.5 to 15 litres/ha, are normal when material is released 1.5–2.5 m above the ground. Booms for fixed- and rotary-wing aircraft, although mounted differently, are basically the same in construction. Boom pipes are round or aerodynamic in cross section. In a fixed-wing aircraft, they are mounted on the trailing edge of the wing and are usually three-quarters the wing length. Atomizers have a nozzle and a variety of spinning screen cages, discs and wire brushes; they are usually driven by fans or electric motors. Atomizers produce droplets of more uniform size and are useful in low-volume spraying, such as for grasshopper and mosquito control (Anon., 1981).

The present trend in pesticide use is to apply highly concentrated material at low rates: ultra-low-volume rates for mosquito control are as low as 0.15 litres/ha. The use of such formulations requires the use of special equipment and application procedures, as the systems must deliver fine droplets to be effective. This can be accomplished by using spinning or flat fan nozzles that discharge 0.15 litres/ha at 40–55 psi [2.8–3.9 kg/cm²]. For helicopter operations, a single spinning nozzle may provide adequate output at very low rates such as required for mosquito control. Because the ultra-low-volume systems produce fine droplets, the location of the nozzles is important (Anon., 1981).

1.2.3. Occupational exposures

Occupational exposures may occur during the manufacture and processing of insecticides as well as during their use. In addition, pesticide residues on plants or fruits may cause significant exposure of farm workers picking or handling the products. Among the more specific occupations with potential exposure are: manufacturers (production workers), formulators, vendors, transporters, mixers, loaders, applicators/operators (farmers or professionals) and pickers and growers.

The relative importance of the routes of occupational exposure is usually in the following order: dermal exposure > respiratory exposure > oral exposure. The occupational groups can have long-term exposure of this type, but very few reports of any effects are available, and further studies are needed to describe better the conditions of chronic exposure that do occur. It is relatively easy to identify people in occupations where exposure to pesticides is common and/or heavier than in the general population, but documentation or measurements of exposure to specific pesticides are seldom available. Furthermore, even if monitoring data or other documentation of exposure are available, it is difficult to evaluate risks associated with specific pesticides because most applicators have contact with a variety of pesticides and because variation in work practices can greatly affect the delivered dose.

Different levels of exposure are encountered depending on the type of application equipment used (Table 3). Furthermore, considerable differences in exposure have been found among the different jobs; most frequently, mixer-loaders have been found to receive the highest exposures, as they work with large quantities of concentrated materials.

Table 3.

Estimated exposure of applicators using different application equipment.

Selected studies in which occupational exposures in spraying and application of insecticides have actually been measured are presented in Table 4.

Table 5.

Summary of findings for cancers at selected sites from descriptive and ecological studies on use of insecticides in the USA.

Table 4.

Occupational exposures in spraying and application of insecticides.

In a study of the exposure of agricultural workers to carbaryl, Maitlen et al. (1982) found that the mixer-loader operation was not inherently different for ground, aerial or hand-gun applications. Factors that affected the hourly dermal exposure of mixer-loaders included: the formulation used; the use of gloves; and the method (scooping or pouring) of removing powdered insecticide from its container prior to mixing. Powdered formulations resulted in higher total hourly dermal exposures (43.3 mg/h with gloves; 107 mg/h without gloves) than liquid formulations (3.0 mg/h with gloves; 40 mg/h without gloves). An average of 76% of the total exposure of all mixer-loaders was on the hands. The techniques of scooping and pouring powdered formulations from the container without wearing gloves resulted in total average hourly dermal exposures of 176 and 38 mg/h, respectively.

Dermal exposure to malathion was monitored for mosquito control spray teams treating interior household surfaces in Pakistan. Exposure was found to vary with job category spraymen experienced the highest exposure to the forehead (mean, 39.3 µg/cm²) and chest (mean, 13.6 µg/cm²), while mixers had the highest exposure to the arms (mean, 49 6 µg/cm²) Supervisors had the lowest exposures to the arms (mean, 2.9 µg/cm²). During this study the authors observed improper work practices which increased dermal exposure to malathion Spraymen's clothes were wet at the end of the working day, smelled strongly of pesticide and were worn for several days without washing. Both spraymen and mixers had extensive skin contact with the pesticide while filling and pressurizing the spray tanks. Some mixers mixed the malathion suspension with their hands. Many spray cans leaked pesticide onto the arms hands and chests of the spraymen. When spray nozzles became clogged, the spraymen sometimes blew through them to unclog them (Baker et al., 1978).

Nye (1986) found that during aerial application of cypermethrin, pilots were exposed primarily via contact with the aircraft when entering and leaving, as well as from the cockpit ventilation system. Dermal exposure was predominantly via the hands and to a lesser degree the trunk. For mixer-loaders, exposure was more uniform, but the trunk, gloves and forearms contained most of the cypermethrin residues. Pilots were, on average, exposed to ten times less cypermethrin than mixer-loaders. The author noted that mixer-loaders worked principally with the formulation concentrate, while the pilots were exposed principally to diluted spray.

1.2.4. Exposure monitoring

Human exposure can either be measured directly or inferred. Examples of methods for direct measurement are collection of pesticides in breathing-zone air or on pads or clothing worn by workers. These techniques provide a direct, calculable measure of human exposure under actual conditions. Most often, however, direct measurement is not possible; for example, in retrospective studies, exposures may only be inferred from the available information. Coupling biochemical measures with traditional exposure evaluation procedures used in epidemiological studies of cancer offers the best opportunity for improving the assessment of historical pesticide exposures (Moseman & Oswald 1980; Blair et al., 1989).

Exposure to pesticides has usually been estimated by monitoring the ambient environment. In some cases, exposure by inhalation has been measured using personal sampler pumps with absorbent filters approximating the breathing zone of the worker (Franklin, 1989). A new technique for estimating exposure by inhalation is to measure the level of the chemical in exhaled breath (Morgan et al., 1989). Exposure by dermal contact is estimated by placing absorbent patches on the worker's body or clothing or by solvent extraction of clothing worn during application. Exposure estimates obtained from ambient monitoring (personal samplers, patches, clothing, tracers) indicate the amount of pesticide that impinges on the surface of the body (contact exposure) and not the absorbed dose (Franklin, 1989).

A number of definitions have been developed for the biological monitoring of exposures in general and for that of insecticides in particular (Foa et al., 1987; Clarkson et al., 1988; Wang et al., 1989). In this monograph, the term 'biological monitoring' is used to cover all those procedures for assessing dose and the biochemical and physiological effects (possibly reversible) in human biological specimens after exposure to insecticides. Industrial hygiene practices and clinical diagnostic procedures are not included in this discussion. Within the limits of this definition, it is obvious that the identification of risks and of groups of subjects at risk is crucial.

Biochemical effects of organophosphorus pesticides can be assessed by measuring the inhibition of certain blood enzymes, as their inhibition usually mirrors that of the corresponding enzyme within the nervous system which represents the molecular target of the organophosphate. Thus, measurement of acetylcholinesterase in red blood cells is widely used to assess cholinergic effects, whereas inhibition of neuropathy target esterase in lymphocytes might be used to assess the delayed neurotoxic effects of some organophosphate insecticides (Hayes et al., 1980; Ames et al., 1989; Lotti, 1989; WHO/UNEP, 1990).

Insecticides can be measured in biological samples by the usual analytical techniques (see the monographs on individual insecticides) or by biological methods. A number of reports are available in which insecticides and/or their metabolites have been measured in body fluids after occupational exposures (Coye et al., 1986; Maroni, 1986; Wang et al., 1989). Examples include the measurement of dialkylphosphates in urine after exposure to organophosphorus insecticides (Coye et al., 1986), of para-nitrophenol after exposure to parathion and methylparathion (Wolfe et al., 1970) and of 1-naphthol after exposure to carbaryl (Comer et al., 1975) (e.g., see Coye et al., 1986). Although these procedures are used for quantitative assessments of dose, interpretation of the results is hampered in the absence of pharmacokinetic data in man. Such data are also essential when extrapolating data on toxicity across species.

A somewhat different approach to biological monitoring is the measurement of adducts to proteins (Shugart et al., 1989). Adducts to haemoglobin have been detected with several pesticides (Sabbioni & Neumann, 1990). The advantages of such measurements include the possibility of assessing dose closer to the target, of assessing individual capacity to form electrophiles and of extrapolating data on toxicity more easily across species. When the mechanism of action of a pesticide is understood, more specific markers can be used.

2.1.1. Mortality statistics

In Central Luzon in the Philippines, the use of organophosphate and organochlorine insecticides increased after adoption of modern rice varieties in the late 1960s. Insecticides were applied by backpack sprayer, usually by men. Protective clothing was not worn. Mortality was studied in three rural municipalities, with a population of 96 000 in 1980 and where more than 80% of the heads of households in typical villages were employed primarily in rice farming. The study covered two periods: 1961–71, with low use of insecticides, and 1972–84, with high use. For men aged 15–54 years, the mean age-standardized mortality rate per 100 000 for cancer (all sites except brain) increased from 21.1 to 25.9. Mortality from leukaemia among men increased from 0.6 to 3.6 per 100 000. Seven of the 11 leukaemia cases recorded since 1961 occurred in 1979–84. The leukaemia rates for women in the same periods were 0.6 and 0.7 per 100 000, respectively (Loevinsohn, 1987).

2.1.2. Proportionate mortality studies

Several studies have been undertaken on the basis of death certificates from Wisconsin, USA (Blair & Watts, 1980; Blair & White, 1981; Cantor, 1982; Cantor & Blair, 1984; Saftlas et al., 1987). In the most recent of these, a study population was selected of 35 972 white men, 18 years and older, who resided in 69 of the 70 Wisconsin counties and died in 1968–76 and whose occupation on the death certificate was farm owner, tenant or labourer. Proportionate mortality ratio (PMR) and proportionate cancer mortality ratio (PCMR) values were calculated using mortality of white, non-farming Wisconsin men for comparison. Data from the agricultural censuses in 1949, 1964 and 1969 and from the population census in 1960 were used for constructing indicators of exposures in agriculture in each county. The PMR for all cancers was significantly lower than expected (5634 observed; PMR, 0.92 [95% confidence interval (CI), 0.90–0.94]). This deficit was due in particular to tobacco-related cancers. When the data were analysed after excluding smoking-related causes of death, statistically significant excess risks were found for cancers of the stomach (PCMR, 1.1 [95% CI, 1.0–1.2]), prostate (PCMR, 1.1 [1.1–1.2]) and eye (PCMR, 3.4 [2.2–5.2]) and for all lymphopoietic cancers (PCMR, 1.1 [1.0–1.2]). Analysis by agricultural exposure level revealed a statistically significant excess risk for cancers of other lymphatic tissue (two-thirds were multiple myelomas) in counties with heavy use of insecticides (28 observed; PCMR, 1.6 [1.0–2.2]). The risk was also increased in counties with heavy use of herbicides and fertilizers and a high proportion of maize production.

2.1.3. Case-control studies with ecological information on exposure

Several case-control studies employed similar methods to evaluate cancer risks and potential exposure to agricultural insecticides in the USA (Blair & Thomas, 1979; Burmeister et al., 1982, 1983; Cantor & Blair, 1984). Deceased cases were ascertained from death certificates, and controls were deceased residents from the same state, matched to the cases by date of birth, date of death, race, sex and county of residence. Usual occupation as a farmer was determined from the death certificate. Potential exposure of farmers to insecticides was classified on the basis of their county of residence and insecticide use patterns for that county. Thus, these studies employed both ecological and individual assessments.

Blair and Thomas (1979) found mortality from leukaemia to be elevated among farmers in Nebraska, but the risks were similar among farmers in counties using less insecticides (odds ratio, 1.2; 95% CI, 0.94–1.6) and in counties using more insecticides (odds ratio, 1.3; 1.0–1.7). Similar findings were obtained in Iowa, where the risks for leukaemia among farmers did not appear to be higher among those residing in counties where insecticides were used heavily than in counties where they were used to a lesser extent (Burmeister et al., 1982).

In Iowa, the odds ratio for death due to multiple myeloma for farmers born after 1890 residing in counties in the upper tercile of insecticide use was 2.0 (p < 0.05) (Burmeister et al., 1983). Cantor and Blair (1984) evaluated risks for multiple myeloma in association with ecological assessments of exposure to insecticides in Wisconsin. Using nonfarmers residing in counties where insecticide use was low as unexposed controls, the odds ratios were 0.9 (95% CI, 0.7–1.3) for nonfarmers residing in counties with high insecticide use, 1.2 (0.9–1.7) for farmers residing in counties with low insecticide use and 1.9 (1.1–3.5) for farmers living in counties with high insecticide use. [These results are based on the same data as those of Saftlas et al. (1987).]

Proportionate mortality studies and case-control studies with ecological information on exposure are summarized in Table 5.

2.3.1. Lymphatic and haematopoietic systems and soft-tissue sarcoma

Hoar et al. (1986) conducted a population-based case-control study of white male residents of Kansas, USA, 21 years or older. Information was collected on 139 histologically confirmed cases of soft-tissue sarcoma and 132 of Hodgkin's disease diagnosed in 1976–82 and on a random sample of 172 cases of non-Hodgkin's lymphoma diagnosed in 1979–81. Three population controls were matched to each case on age and vital status: living controls up to 64 years of age were selected by random digit dialling; older subjects were selected from Medicare files; and deceased controls were selected from state mortality files, after excluding index neoplasms, a malignancy at an ill-defined site, homicide and suicide. After exclusions, a total of 1005 controls were selected. Patients and controls, or their next of kin, were interviewed by telephone about farming practices. Interviews were obtained for 96% of patients and 94% of controls. Information on herbicide and insecticide use provided by a sample of farmers among enrolled subjects was validated by interviewing their pesticide suppliers. A small excess of non-Hodgkin's lymphoma (odds ratio, 1.5; 95% CI, 0.9–2.4) was found among patients reporting use of any insecticides. Odds ratios by year of first use of insecticides among farmers were 1.7 for use prior to 1946, 1.5 for 1946–55, 0.7 for 1956–65 and 1.5 for 1966 or later. No association was observed with number of hectares treated. Adjustment for days per year of herbicide use reduced the odds ratio for insecticide use to 1.1 (95% CI, 0.6–2.2). Risks for non-Hodgkin's lymphoma increased slightly, however, with days per year of insecticide use, even after adjusting for exposure to herbicides (odds ratio, 1.2 (95% CI, 0.5–2.8) for one to two days of insecticide use; and 1.4 (0.6–3.1) for more than two days of insecticide use). A further analysis of these data was undertaken by Hoar Zahm et al. (1988). A small excess risk for soft-tissue sarcoma was seen among farmers using insecticides on animals (odds ratio, 1.6; 95% CI, 0.9–2.5) but not among those using insecticides on crops (0.8; 0.4–1.6). Relative risks for use on animals rose with time since first use to 4.9 (0.6–64.1) among farmers who first used them in 1945 or earlier. Potentially greater exposure might be expected during treatment of animals because higher concentrations are sprayed in confined spaces, such as barns. Excess risks were associated with most of the major classes of insecticides. No association was observed between the risk for Hodgkin's disease and use of insecticides.

In a study of similar design to the study in Kansas, all non-Hodgkin's lymphomas occurring among white men 21 years or older between 1983 and 1986 in 66 counties in eastern Nebraska were ascertained through local hospitals (Hoar Zahm et al., 1990). Population-based controls were frequency matched to cases on race, age and vital status. A total of 227 cases of non-Hodgkin's lymphoma and 831 controls were selected. Telephone interviews were conducted to obtain detailed information on specific agricultural chemicals used, and responses were obtained from 201 cases (91%) and 725 controls (87%). There was little evidence of an association between non-Hodgkin's lymphoma among farmers and use of insecticides overall (odds ratio, 1.1; 95% CI, 0.7–1.6); but non-Hodgkin's lymphoma was associated with use of organophosphorus insecticides (odds ratio, 2.4, adjusted for use of 2, 4-D). Risks rose with days per year of use of organophosphorus compounds (odds ratios, 1.7 for 1–5 days/year, 1.8 for 6–20 days/year and 3.1 for 21 or more days/year). [Numbers of exposed subjects and confidence intervals were not provided.]

A population-based case-control study was conducted in Washington State (USA) to investigate the relationship between soft-tissue sarcoma and non-Hodgkin's lymphoma and past exposure to phenoxyacetic acid herbicides and chlorinated phenols (Woods et al., 1987). Between 1981 and 1984, 206 soft-tissue sarcomas and 746 non-Hodgkin's lymphomas were diagnosed in men aged 20–79 years. Of these cases, 13% were excluded due to physicians' refusal; 91% of the remaining patients (or their proxies) were interviewed and information was derived about pesticide exposures. Of the remaining 163 soft-tissue sarcoma cases, 33 were excluded on the basis of a pathology review, and two were excluded for other reasons, leaving 128 soft-tissue sarcoma cases to be included in the analysis. Of the remaining 586 non-Hodgkin's lymphoma cases, 10 were excluded for various reasons, leaving 576 cases to be included in the analysis. Controls for living cases were selected by random-digit telephone dialling (for those aged 20–64 years) and from Health Case Financing Administration files (for those aged 65–79 years); deceased controls were obtained from death certificates and matched on five-year age group. Interviews were obtained for 694 of the 910 controls (76%). Odds ratios for potential exposure to chlordane were 1.6 (95% CI, 0.7–3 8) for non-Hodgkin's lymphoma and 0.96 (0.2–4.8) for soft-tissue sarcoma. Potential exposure to DDT yielded odds ratios of 1.8 (1.0–3.2) for non-Hodgkin's lymphoma and 1.1 (0.4–3.2) for soft-tissue sarcoma. Adjustment for exposure to some other pesticides did not substantially change these risk estimates. Another evaluation of the data from this study restricted analyses to farmers (Woods & Polissar, 1989). The relative risks for non-Hodgkin's lymphoma among farmers potentially exposed were 1.6 (0.5–5.1) for chlordane and 1.7 (0.9–3.3) for DDT.

A population-based, multicentre case-control study of multiple myeloma in people under 80 years of age was carried out between 1977 and 1981 in four US areas covered by cancer registries (Morris et al., 1986). Interviews were obtained with 698 cases (89% of eligible subjects) and 1683 controls (83%); 32% of case interviews and 1% of those for controls were with next-of-kin. Controls were obtained by household sampling in one area and by random digit dialling in the remaining three areas; they were matched to cases on age, sex and race. Self-reported exposures to chemicals were grouped into 20 categories, some of which were further subdivided. The odds ratio for exposure to pesticides was 2.6 (95% CI, 1.5–4.6); an odds ratio of similar magnitude was obtained when reports from surrogate respondents were excluded (2.9; 1.5–5.5). The numbers of cases and controls who reported exposure to various classes of insecticides were two cases and five controls for organophosphorus compounds [odds ratio, 1.0; 0.2–5.1], nine cases and eight controls for organochlonnes [2.8; 1.1–7.0] and three cases and five controls for arsenicals [1.5; 0.4–6.2]. [Subjects exposed to other pesticides were excluded when calculating these odds ratios.]

A case-control study on multiple myeloma was conducted within a prospective study of 1.2 million American Cancer Society volunteers who in 1982 filled in a self-administered questionnaire on diseases and several cancer risk factors, including occupation and exposure to pesticides and herbicides. They were followed for mortality up to 1984 or 1986 and 282 deceased subjects with mention of multiple myeloma on their death certificate were identified. Prevalent cases of multiple myeloma and subjects with related symptoms identified at the time of entry into the cohort were excluded, leaving 128 incident cases, who were matched on sex, age, residence and race to four randomly selected controls. A logistic regression analysis adjusting for potential confounders was presented. The odds ratio for occupation as a farmer was 2.7 (95% CI, 1.3–5.7; based on 16 cases and 28 controls exposed); the odds ratio for exposure to pesticides and herbicides was 1.6 (0.7–3.7; 12 cases and 25 controls). When these variables were combined, nonfarmers exposed to pesticides and herbicides experienced no excess risk (crude odds ratio, 1.0; 95% CI, 0.3–3.1), whereas exposed farmers had an elevated risk (crude odds ratio, 4.3; 95% CI, 1.7–10.9), which was greater than that of unexposed farmers (crude odds ratio, 1.7; 95% CI, 0.8–4.0) (Boffetta et al., 1989).

A population-based case-control study of leukaemia among white men aged 30 years or older in Iowa (1981–83) and in Minnesota (1980–82) covered a total of 669 eligible cases (Brown et al., 1990). Living controls were selected by random digit dialling (< 65 years) and from Medicare records (> 65 years); deceased controls were selected from death certificates. Interviews were completed with 86% of the cases or close relatives and with 77–79% of the controls; the study thus included 578 cases and 1245 controls. A standardized questionnaire was used to obtain detailed information on residential history, drinking-water source, nonfarm occupational history, smoking and alcohol use, use of unpasteurized dairy products, medical conditions, family history of cancer and farm activities, including information on 24 animal insecticides and 34 crop insecticides. Relative risks were calculated by comparison to subjects who were nonfarmers, adjusting for age, vital status, state of residence, tobacco use, family history of lymphopoietic cancer, high-risk occupations and high-risk exposures. The risk for leukaemia overall was not significantly associated with reported use of any insecticide (odds ratio, 1.1; 95% CI, 0.9–1.3) among farmers. An excess risk was observed for chronic lymphocytic leukaemia (1.3; 1.0–1.8), but not for other histological types: acute nonlymphocytic leukaemia, 1.0 (0.7–1.6), chronic myelogenous, 1.0 (0.5–1.8) and acute lymphocytic, 0.8 (0.2–2.5). The risk for leukaemia tended to be greater with use of insecticides on animals than on crops. Significant excesses occurred with use on animals of natural insecticides (odds ratio, 1.5; 95% CI, 1.0–2.2) and organophosphates (1.5; 1.0–2.1). Time since first use of insecticides also influenced risk: after a 20-year latency, significant excesses of leukaemia were noted with use on animals of several insecticides, including DDT (1.4; 1.0–2.0), dichlorvos (2.4; 1.1–5.4), famphur (11.6; 1.2–107.0), nicotine (2.0; 1.2–3.4) and pyrethrins (3.8; 1.0–14.8). Nonsignificant odds ratios of 2.0 or more were observed with use of carbaryl (3.0; 0.7–3.1), coumaphos (2.3; 0.6–8.8), methoxychlor (2.1; 0.7–6.6) and toxaphene (2.6; 0.8–8.8). The relative risk for leukaemia associated with frequent use on animals (> 10 days/year) was statistically significant for dichlorvos (3.8; 1.0–14.8) and malathion (3.2; 1.0–10.0); the risk increased with frequency of use of DDT on animals, from 0.6 (0.3–1.4) for 1–4 days per year to 1.1 (0.4–2.7) for 5–9 days per year, to 2.1 (1.1–3.9) for 10 or more days per year. These estimates for agricultural exposures were not mutually adjusted.

Patients with chronic lymphatic leukaemia diagnosed in five hospitals in the middle and south-east of Sweden between 1964 and 1984 who were still alive after 1981 were compared with population controls living in the catchment areas of the hospitals (Flodin et al., 1988). Subjects over 80 years or mentally disabled were excluded, leaving 111 cases (response rate, 91%) and 431 controls (response rate, 83%; replacements were sought for the 17% who did not respond). Information on exposure to ionizing radiation, DDT, solvents and engine exhausts as well as a h, story of previous diseases, smoking and occupation were collected using a self-administered questionnaire. Results of a stratified analysis based on a confounder score including age, sex, exposure to fresh wood, solvents, engine exhausts DDT and horses and employment as farmer were presented. Exposure to DDT was reported by six cases and four controls; the odds ratio was 6.0 (95% CI, 1.5–23.0). [The Working Group noted the limitation of inclusion of prevalent cases because of the potential influence on recall of exposure.]

A study on Hodgkin's disease and non-Hodgkin's B-cell lymphomas was conducted in one of the areas included in the study described above. The same criteria were applied for selection of cases and the same controls were used. There were 54 cases of Hodgkin's disease and 106 of non-Hodgkin's lymphoma (overall response rate, 97%) and 275 controls Logistic regression analysis was carried out including sex, age, farming, exposure to fresh wood and all exposures that gave a crude odds ratio greater than 2.0. Exposure to DDT was reported by three patients with Hodgkin's disease, none with non-Hodgkin's lymphoma and three controls. The odds ratio for Hodgkin's disease was 7.5 (90% CI, 0.8–70.0) (Persson et al., 1989 [The limitation of the study by Flodin et al. (1988) noted above also applies to this suudy]

A case-control study on malignant lymphomas in northern Sweden considered primarily exposure to phenoxyacetic acid herbicides and chlorophenols (Hardell et al., 1981) Cases were all men aged 25–85 years with histologically verified malignant lymphoma admitted to the control hospital in the area in 1974–78. A total of 60 cases of Hodgkin's disease and 109 of non-Hodgkin s lymphoma were matched to 338 population controls (responses available from 335) by age, sex, place of residence, vital status and year of death for deceased cases and controls. Information from self-administered questionnaires was supplemented by telephone interviews when the data were incomplete. A total of 22 cases and 26 controls reported exposure to DDT [odds ratio, 1.8; 95% CI, 1.0–3.2]. Seven cases and 11 controls reported exposure to DDT and not to phenoxyacetic acid herbicides [odds ratio, 1.6; 95% CI 0.6–4.1] Information on use of DDT was not presented separately for patients with Hodgkin s disease and those with non-Hodgkin's lymphoma.

Four case-control studies in Sweden assessed the risk of soft-tissue sarcoma alone primarily in association with exposure to phenoxyacetic acid herbicides and chlorophenols (Hardell & Sandström 1979; Eriksson et al., 1981; Hardell & Eriksson, 1988; Eriksson et al., 1990). Exposure was assessed by methods similar to those described above using questionnaires mailed to subjects or next-of-kin. For subgroups of the subjects, information was supplemented with telephone interviews.

A case-control study in northern Sweden included 52 male cases of histologically reviewed soft-tissue sarcoma (100% response) and 206 population controls (99% response) matched for age sex place of residence and year of the death for deceased cases and controls (Hardell & Sandström, 1979) Four cases and 14 controls reported exposure to DDT(crude odds ratio, 1.2 [95% CI, 0.4–3.7]

A population-based case-control study in southern Sweden included 110 cases of histologically verified soft-tissue sarcoma and 219 controls (responses obtained from all but one), matched for age, place of residence and year of death for dead cases and controls (Eriksson et al., 1981). Seven cases and 11 controls reported exposure to DDT [crude odds ratio, 1.3; 95% CI, 0.5–3.4].

A population-based case-control study in northern Sweden included 54 male cases of soft-tissue sarcoma (responses obtained from all but one) (Hardell & Eriksson, 1988). Two control groups were used: one was population-based, with 311 subjects (94% response) matched for age and place of residence, and the second consisted of 179 cases of malignant disease (94% response) diagnosed in the same period as the cases. Six cases, 19 population-based controls and eight cancer controls reported exposure to DDT [crude odds ratio, 1.9; 95% CI, 0.7–5.0 (population controls); 2.7; 0.9–7.8 (cancer controls)]. Crude odds ratios for exposure to DDT without exposure to phenoxyacetic acid herbicides were [0.6 (0.1–5.0)] for population controls and [1.2 (0.1–11.6)] for cancer controls.

All male patients diagnosed with histologically confirmed soft-tissue sarcoma between 1978 and 1986 were identified from a regional cancer registry in central Sweden. Interviews were completed with 218 (92%) of the cases identified. One control per case was drawn from the National Population Registry matched on age, sex, vital status and county of residence; 212 controls (89%) of those selected were interviewed. Twelve controls who were contacted could not complete the questionnaire and were replaced by the next person on the population register. Subjects or next-of-kin completed a 12-page questionnaire regarding exposures of interest; these data were supplemented by telephone interviews for subjects employed in agriculture, forestry, horticulture, carpentry and sawmills. The odds ratios for soft-tissue sarcoma were 0.61 (95% CI, 0.34–1.1) for potential exposure to DDT and 0.52 (0.19–1.4) for potential exposure to pesticides other than DDT or mercury seed dressings (Eriksson et al., 1990).

2.3.2. Other cancers

A proportionate analysis of occupational mortality in Washington State, USA, identified a 30% increased risk for respiratory cancer among orchardists (Milham, 1983), and a case-control study was undertaken on mortality from respiratory cancer in Washington State in 1968–80 (Wicklund et al., 1988). Death certificates were selected that contained the occupational codes for orchardists, orchard labourers and brush pickers, and farm owners and tenants. Cases were men who had died from respiratory cancer, and potential controls were men who had died from other causes, matched by county of residence, year of death, age at death and occupational code. An attempt was made to contact surviving next-of-kin or other informants. Interviews were obtained for 87.1, 79.6 and 60.5%, respectively, of the three occupational groups. On the basis of the information obtained, orchardists were limited to men who had been involved in orchard work for at least 10 years or 25% of their working history or had been associated with at least one orchard of five or more acres [> 2 ha]. Information on occupational exposures and smoking was obtained from a structured questionnaire. If the informant was unable to recall whether a specific pesticide had been sprayed by the deceased but was able to recall the precise year of his orchard work, a 'presumed' history of spraying a particular pesticide was obtained. Lead arsenate was used by orchardists before 1945 and was replaced by DDT subsequently; thus, if a deceased man had worked prior to 1945, he was presumed to have a positive history of lead arsenate spraying, and if he had worked after 1945 he was presumed to have a positive history of DDT spraying. Information was obtained for all but 7% of the 155 cases and 4.9% of the 155 matched controls. A total of 89 cases and 89 controls were assumed to have had exposure to DDT. When men exposed to DDT but not to lead arsenate were considered, there were 33 cases and 29 controls, and the odds ratio (adjusted for smoking) was 0.91 (95% CI, 0.40–2.1). [The Working Group noted that the unexposed group included men for whom details on exposure to DDT were not available, which may have biased the odds ratio towards the null.]

Two case-control studies in Sweden examined the risks for colon cancer (Hardell 1981) and nasal and nasopharyngeal cancer (Hardell et al., 1982), primarily in relation to exposure to phenoxyacetic acid herbicides and chlorophenols. The same control group, consisting of 541 people, was used in these two studies and in two previous studies conducted by the same group (Hardell & Sandstrom, 1979 [see above]; Hardell et al., 1981). There were 154casesof colon cancer and 71 cases of nasal and nasopharyngeal cancer. Odds ratios for exposure to DDT, without controlling for other agricultural exposures, were [0.8; 0.4–1.7] for colon cancer and [1.2; 95% CI, 0.5–2.9] for nasal and nasopharyngeal cancer. In the study of colon cancer, exposure to DDT was also analysed after excluding subjects who had been exposed to phenoxyacetic acids and chlorophenols; the odds ratio was [0.5; 0.2–1.6].

Men aged 25–80 who had been diagnosed with liver cancer between 1974 and 1981 and reported to the Department of Oncology, Umea, Sweden, were included in another case-control study (Hardell et al., 1984). Microscope slides were reviewed for the 166 assembled cases, and 103 cases of primary liver cancer were retained for the study; 206 population-based controls were matched to cases on age and residence. Information on exposure was obtained as in previous studies (see Hardell & Sandstrom, 1979); responses were obtained for 102 cases and 200 controls. The analysis was restricted to the 98 cases of hepatocellular or cholangiocellular carcinoma. Odds ratios for exposure to DDT, without controlling for other agricultural exposures, were [0.4; 95% CI, 0.1–1.1] for exposure to DDT in farming and [1.3; 0.4–4.0] for exposure to DDT in forestry.

A total of 240 cases of brain glioma were collected from two hospitals in Milan, Italy, between 1983 and 1984. Patients with non-glioma brain tumours (465) and patients with non-neoplastic neurological diseases (277) recruited from the same hospitals and matched for age and sex to the cases formed two series of controls [response rates not given]. Subjects were asked about their occupational history as well as their use of fertilizers, herbicides and insecticides or fungicides. Exposure of farmers to insecticides or fungicides gave an odds ratio of 2.0 (95% CI, 1.2–3.2) using all controls and of 2.1 (1.3–3.6) using only cancer controls (Musicco et al., 1988).

2.3.3. Childhood cancer

A case-control study of brain tumours in Baltimore, MD, USA, included all cases under 20 years of age diagnosed in 1965–75, and two groups of controls—one selected from birth certificates of the state and one from children with other malignancies. Controls were matched individually to cases on sex, race and date of birth; cancer controls were also matched on date of diagnosis. Interviews were conducted with 84 of 127 (66%) identified cases; interviews with controls yielded 73 matched pairs with population controls and 78 matched pairs with cancer controls. Parents were interviewed with respect to environmental exposures, including insect exterminations in the household, as well as child and family characteristics. A matched-pair analysis was conducted. Insect extermination was more common in houses of cases with respect to population controls (odds ratio, 2.3 [95% CI 0.9–6.6]) but not to cancer controls (1.2 [0.5–2.8]) (Gold et al., 1979).

The results of the case-control studies are summarized by site in Tables 7–9.

Table 7.

Case-control studies on malignant lymphomas containing information on insecticide exposure.

Table 9.

Case-control studies of other cancers containing information on insecticide exposure.

Table 8.

Case-control studies of soft-tissue sarcomas containing information of insecticide exposure.

3.3.1. Cytogenetic studies

Sixteen agricultural workers who had been exposed predominantly to insecticides (mainly the organophosphorus compounds, demeton, ethyl parathion, trichlorfon and naled) in the USA, with a mean exposure time of 12 years, were compared with 16 controls with a variety of occupations not involving pesticides. Blood samples were taken off-season and mid-season from both groups, and 25 metaphase-arrested lymphocytes from each person were scored on each occasion for chromatid breaks and gaps. The frequency of chromatid breaks in mid-season samples was increased from 0.44 ± 0.22 (SD) in the control group to 1.56 ± 0.29 in the exposed group (Yoder et al., 1973). There was no difference in the occurrence of chromatid gaps either off-season or mid-season or of chromatid breaks in off-season samples. [The Working Group noted that no attempt was made to control for confounding factors such as tobacco smoking, that small numbers of cells were scored from each person and that individual results were not reported.]

The effects of low levels of the insecticide fumigant, ethylene dibromide, upon the frequencies of sister chromatid exchange and chromosomal aberrations were examined among forestry workers in mainland USA (Steenland et al., 1985) and papaya workers in Hawaii (Steenland et al., 1986). Blood samples were taken from 14 forestry workers who sprayed pine trees before and after exposure (8-h time-weighted average, 80 ppb [616 µg/m³], with a peak of up to 281 ppb [2164 µg/m³]) and compared with those from six unexposed controls. No effect of exposure to ethylene dibromide was observed, although smoking did increase the frequency of sister chromatid exchange. A group of 60 papaya workers exposed to ethylene dibromide at a geometric mean of 88 ppb [678 µg/m³] (8-h time-weighted average), but with peak exposures of up to 262 ppb [2017 µg/m³], were compared with a control group of 42 workers from a nearby sugar mill. The two groups were matched for age, tobacco, marijuana and coffee use and race. No difference was found in the total frequencies of either chromosomal aberrations or sister chromatid exchange. The frequency of the latter, however, was increased in men who smoked either tobacco or marijuana, and that of chromosomal aberrations showed an increasing trend with age.

Floriculturists (36 men and women) in Argentina involved in spraying various pesticides, including organophophosphates, carbamates and organochlorines, were studied for chromosomal aberrations and sister chromatid exchange. Symptoms of chronic intoxication were observed in 21 workers. The control group consisted of 15 healthy scientists and technicians. A significant difference was seen in sister chromatid exchange frequencies between the symptomatic and non-symptomatic floriculturists as well as between a matched group of controls and sprayers. The frequency of only dicentric and ring type chromosomal aberrations was increased when the whole group of floriculturists and controls was compared (Dulout et al., 1985).

In a study from Hungary, 80 male workers were involved in mixing and spraying pesticides (80 formulations were recorded, including insecticides such as organophosphates, organochlorines, pyrethroids and carbamates). At least 12 weeks' continuous contact with pesticides was recorded during the spraying season. A group of 24 administrative workers and mechanics served as controls. A significant increase in the frequency of chromosomal aberrations was seen in the group of pesticide workers as compared with controls; that of chromosome-type aberrations increased with duration of exposure (Páldy et al., . 1987).

Forty producers of potted plants (17 smokers) working in greenhouses in Argentina and exposed to a mixture of organophosphorous, organochlorine, carbamate and some miscellaneous pesticides, were examined for chromosomal aberrations (Dulout et al., 1987). The control group consisted of 32 healthy hospital blood donors (10 smokers) with no known exposure to pesticides. Since some of the control individuals showed toxic symptoms that are seen after chronic exposure to pesticides, a second control group was selected consisting of 12 blood donors (six smokers) with no symptom of exposure. No difference was observed between the control groups with respect to frequency of chromosomal aberrations or between the producers of potted plants and the controls as a whole.

The frequencies of chromosomal aberrations in 15 vineyard workers in India who were exposed to seven insecticides (DDT, lindane, quinalphos, metasystox, parathion, dichlorvos and dieldrin) and two fungicides (dithane M₄₅ and copper sulfate) were compared with those in 10 controls of similar age and socioeconomic status, but not exposed to pesticides (Rita et al., 1987). The proportion of metaphase cells with chromatid breaks was significantly increased in the exposed group.

In another Hungarian study, 55 male workers were involved in spraying and applying pesticides in greenhouses, plastic tents and open fields. Among the pesticide formulations, various organophosphates, carbamates and pyrethroids were listed. The control group consisted of 60 male blood donors. A slight increase in the frequency of chromosomal aberrations was recorded among the open-field sprayers; however, the values for closed-space workers were at the control level (Nehéz et al., 1988).

Cotton field workers (pesticide mixers and sprayers) in India were studied for the effects of exposure to pesticides (particularly insecticides) on the frequency of chromosomal aberrations (Rupa et al., 1989a) and of sister chromatid exchange (Rupa et al., 1989b). Peripheral lymphocytes were examined from 50 smokers and compared with those from 20 nonsmokers and 27 smokers, none of whom were occupationally exposed to pesticides. The frequency of chromosomal aberrations was significantly increased in the pesticide-exposed group as compared with either of the control groups. Sister chromatid exchange frequencies showed similar responses to both smoking and pesticides exposure; there was a trend towards an increasing number of sister chromatid exchanges with years of exposure to pesticides.

In a study of chromosomal aberrations, 52 nonsmoking cotton-field workers exposed to pesticides (mainly insecticides) were compared with 25 controls. The insecticides mentioned included malathion, methyl parathion, dimethoate, DDT and fenvalerate. The prevalence of chromosomal aberrations was significantly increased in the group of workers, and there was a positive trend with duration of exposure (Rupa et al., 1989c).

Fumigation of phosphine produced from aluminium or magnesium phosphide pellets is used commonly in the grain industry to control beetles. Of 24 professional fumigant applicators, nine were exposed to phosphine alone (mean level, 2.97 mg/m³ in grain bin areas; average duration of exposure, over 20 min daily). This group had a five-fold higher frequency of chromosomal deletions than the control subjects, and the frequency of breaks was also significantly increased. Chromosome banding analysis showed that chromosomal rearrangements were six times more frequent in the exposed workers than in the controls. Sister chromatid exchange frequency was not increased. In lymphocytes in vitro, no sister chromatid exchange was induced, but the frequency of chromosomal aberrations was increased in a dose-related manner (Garry et al., 1989).

Carbonell et al. (1990) studied 27 agricultural workers in Spain who had used insecticides such as fenvalerate, deltamethrin and methomyl. No sister chromatid exchange was induced.

The results of studies on cytogeneticity are summarized in Table 10.

Table 10.

Summary of results from cytogenetic biomonitoring studies on pesticide applicators.

3.3.2. Urine mutagenicity studies

Urine samples were taken from 12 greenhouse owners (11 men and 1 woman) who daily sprayed fungicides and insecticides (dichlorvos, orthene (acephate) and pentac (dienochlor) mentioned) in greenhouses in central New York for 1–40 years. The samples, collected 8 h after spraying, and control samples taken three days later were tested for mutagenicity using Salmonella typhimurium (strains TA100 and TA98) both with and without an exogenous metabolic system. No increase in mutagenic activity was detected in urine from seven subjects. Three subjects who had elevated urine mutagenicity had been working with poor protection. Two of the subjects reported smoking during the days of sample collection (Shane et al., 1988).

San et al. (1989) and See et al. (1990) assayed the urine of nonsmoking orchardists in Canada using chromosomal aberrations in Chinese hamster ovary cells as the endpoint. People were exposed to organophosphates among other pesticides. Chromosome damaging activity was significantly elevated in samples collected during the pesticide spraying season as compared to samples collected during the non-spraying season or to urine samples from nonsmoking controls.

4.2.1. Descriptive and ecological studies

Several death certificate case-control studies in the USA evaluated cancer risks in association with ecological measures of insecticide exposure. The risk for multiple myeloma tended to be greater for farmers residing in counties where insecticides were more heavily used, but that for leukaemia did not.

4.2.2. Cohort studies

A cohort of workers from a large pest control company in the USA had an excess lung cancer risk. Similarly, in a cohort of licensed pest control workers from Florida, there was significantly increased mortality from lung cancer, which was particularly high among workers licensed for 20 years or more; a nonsignificant excess risk for brain cancer was also seen. A follow-up of deaths among plant protection workers and agronomists in eastern Germany showed an increased risk of lung cancer which also increased with length of exposure; survey data indicated that the smoking habits of these pesticide workers were similar to those of the general population.

Among farmers licensed for pesticide use in the Piedmont region of Italy, increased risks for skin cancer and malignant lymphomas were reported; lung cancer incidence was not studied.

A cohort of licensed pesticide applicators in Sweden showed excess risks for cancers of the lip and testis, a slight excess risk for Hodgkin's disease, and risks similar to those of the general population for non-Hodgkin's lymphoma and soft-tissue sarcoma. Overall, there was a deficit of lung cancer risk that was probably related to the lower smoking rates of the applicators.

In a study of a large cohort of grain millers in the USA, flour-mill workers had excess risks for non-Hodgkin's lymphoma and pancreatic cancer; the risk for lung cancer was not increased.

4.2.3. Case-control studies

The risk for non-Hodgkin's lymphoma rose with frequency of use of organo-phosphorus insecticides among farmers in Nebraska, an association that could not be accounted for by use of phenoxyacetic acid herbicides. In Kansas, the risk increased slightly with frequency of use of insecticides as a group. In a study in Washington State, non-Hodgkin's lymphoma was associated with potential contact with chlordane and DDT. DDT use was also associated with non-Hodgkin's lymphoma in one of two studies in Sweden.

Multiple myeloma was associated with use of pesticides (particularly organochlorine insecticides) in a study in the USA. The risk for multiple myeloma was also elevated among farmers in the USA exposed to unspecified herbicides and pesticides.

The results of six studies in Sweden and the USA on soft-tissue sarcoma in association with exposure to insecticides were inconsistent.

Chronic lymphocytic leukaemia has been associated with use of insecticides in the USA and with use of DDT in Sweden.

The risk for brain cancer was associated with exposure to insecticides and fungicides in farmers in Italy.

Overall, the strongest evidence that exposure to nonarsenical insecticides causes cancer in humans comes from the cohort studies of applicators. Two of these studies showed significant excesses of lung cancer. Two showed rising risks with duration of exposure, whereas the third showed an inverse association. These findings were based on small numbers in the subgroups with the longest exposure, and applicators in some of these studies had potential contact with arsenical insecticides. Some case-control studies of multiple myeloma and other tumours of B-cell origin show small excesses among people exposed to insecticides. In most studies, however, potential confounding by other agricultural exposures had not been fully explored.

Overall evaluation

Spraying and application of nonarsenical insecticides² entail exposures that are probably carcinogenic to humans(Group 2A).

Footnotes

1

For definitions of the italicized terms, see Preamble, pp. 26–28.

2

Arsenic and arsenic compounds are carcinogenic to humans (IARC, 1987). This evaluation applies to the group of chemicals as a whole and not necessarily to all individual chemicals within the group.