ACUTE SYMPTOMS FOLLOWING WORK WITH PESTICIDES

Abstract

Objectives

To explore the frequency, nature and determinants of acute symptoms following work with pesticides.

Methods

As part of a postal survey of men born during 1933-77 and resident in three rural areas of England and Wales, data were obtained on demographic variables; occupational use of five categories of pesticide; whether any of 12 specified symptoms had ever occurred within 48 hours of using such pesticides; and tendency to somatise (graded to four levels). Associations of pesticide-related symptoms with risk factors were assessed by modified Cox regression, and the timing of first recalled symptoms in relation to calendar period and years of work with pesticides by Poisson regression.

Results

Of 10,765 responders (response rate = 31%), 4,108 had at some time used pesticides occupationally, including 935 (23%) who reported symptoms following such work on at least one occasion. In two areas, acute symptoms were most frequent following use of sheep dip (29% and 32% of users), but in the third area, sheep dip was less often associated with symptoms (13% of users) than other insecticides (15% of users). The relative frequency of symptoms was similar for all five categories of pesticide, and there was no evidence that flu-like symptoms clustered unusually among users of sheep dip, as has previously been postulated. Multiple pesticide-related symptoms and symptoms in relation to multiple categories of pesticide were both much more common than would have been expected if individual symptoms were statistically independent. Risk of acute pesticide-related symptoms increased with somatising tendency (prevalence ratio for highest v lowest category 2.4, 95%CI 2.0-3.0), and was higher in men who had used pesticides most often, and who had handled concentrate. First onset of symptoms was more common in the first year of pesticide use (incidence rate ratio 4.5, 95%CI 3.5-6.8 in comparison with first use >15 years earlier), and peaked in the late 1980s. However, this peak of incidence disappeared when men who had worked with sheep dip were excluded from the analysis.

Conclusion

Our findings indicate that acute symptoms are common following work with pesticides, but suggest that in many cases the illness may arise through psychological rather than toxic mechanisms.

Introduction

In Britain, as in many other industrialised countries, pesticides are tightly regulated. Before a product is approved for sale or use, it undergoes an extensive risk assessment, which subsequently is liable to periodic review. One of the aims of the risk assessment is to ensure that use in accordance with the conditions of approval will not lead to toxic effects in operators. This is achieved principally by derivation of acceptable operator exposure levels through the application of assessment (safety) factors to relevant no-effect levels for adverse outcomes in toxicological experiments. Checks are then made that exposures will be below the reference values so defined.^{1, 2}

Despite the large margins of safety that are built into this system, it is important to check that the regulatory controls are effective. Data on hospital admissions in England indicate that accidental poisoning by pesticides of sufficient severity to warrant in-patient care is rare, with approximately 50 cases per year in adults aged 16-69 years.³ However, the frequency of less serious acute illness following work with pesticides is less certain. On average, fewer than 30 of the pesticide-related incidents that are reported each year to the Health and Safety Executive are assessed as confirmed or likely cases of poisoning,⁴ but there may be substantial under-reporting of minor illness through this system. Thus, in a cross-sectional survey of 84 agricultural workers in the south of England, 13 (15%) reported at some time having had an accident or health problem related to the use of an agricultural chemical.⁵

The occurrence of symptoms soon after work with a pesticide does not necessarily imply poisoning. Such illness could be a chance coincidence or perhaps triggered by psychological mechanisms. However, if acute toxicity from pesticides were common among users, even if only mild, there would be a need to review the current approach to risk assessment and the capacity of operators to follow instructions for use.

To explore the frequency and nature of acute symptoms following the use of pesticides, we analysed data from a survey of health and work in men from three rural areas of England and Wales.

Methods

The study population comprised men born between 1933 and 1977, who were resident in defined areas of north Devon, the Welsh borders and south Lincolnshire. These areas were chosen to be geographically contiguous, with a high proportion of men employed in agriculture at the time of the 1991 national census. Members of the study population were identified from the age-sex lists of local general practitioners, and were sent a postal questionnaire, followed if necessary by a reminder after 10-16 weeks. Because almost everyone in Britain is registered with the National Health Service, general practice lists provide a near complete enumeration of the general population. The mailings were carried out by local health agencies, and the questionnaires, which were identified only by a serial number, were returned directly to the study team.

The questionnaire covered demographic information, smoking habits, alcohol consumption, lifetime occupational history (with details of specified chemical and physical exposures), and various aspects of health. In particular, it asked about occupational use of five different categories of pesticide (sheep dip, other insecticides, herbicides, fungicides and wood preservatives); whether any of 12 listed symptoms had ever been experienced within 48 hours of using such pesticides; and the extent to which the subject had been disturbed or bothered during the past seven days by each of seven somatic symptoms (faintness or dizziness, pains in the heart or chest, nausea or upset stomach, trouble getting breath, hot or cold spells, numbness and tingling in parts of the body, and feeling weak in parts of the body). The last set of questions was taken from the Brief Symptom Inventory,⁶ and the answers to each were graded on an ordinal scale from zero (not at all) to four (extremely). The grades were then averaged over all seven questions to give an overall score for “somatising tendency”, which was partitioned into four levels.

Statistical analysis was carried out using STATA (version 8.2 SE) software. The frequency with which participants reported multiple acute pesticide-related symptoms was compared with the expected distribution calculated with the assumption that each symptom was statistically independent (eg that a report of headache following the use of a pesticide made it no more or less likely that an individual would also report aching limbs after pesticide use). Similarly, the numbers of men who reported acute symptoms in relation to multiple categories of pesticide were compared with the expected frequencies if the occurrence of symptoms with each type of pesticide were statistically independent (e.g. in a man who had used both insecticides and herbicides, it was assumed that a report of symptoms after using insecticides had no impact on the probability of his reporting symptoms following use of herbicides). Possible clustering of specific pesticide-related symptoms was explored by calculation of odds ratios for their pairwise associations. In particular, we looked for evidence of “dippers’ flu”, a cluster of flu-like symptoms anecdotally reported as being associated with dipping sheep.⁷ Associations of pesticide-related symptoms with potential risk factors were assessed by modified Cox regression,⁸ and the results were summarised as prevalence ratios (PRs) with associated 95% confidence intervals (95% CI). The timing of first recalled acute pesticide-related symptoms in relation to age, calendar period, and years of work with the relevant category of pesticide, was examined in a person-years analysis, with incidence rate ratios (IRRs) derived by Poisson regression.

The study was approved by the NHS South West Multi-centre Research Ethics Committee.

Results

After two mailings, usable questionnaires were returned by 10,765 (31%) of the men selected for study, the response rate being similar in all three of the areas studied. The 23,721 non-responders included 1,825 who had died, or who had moved to a new address and could not be contacted.

Among the 10,765 responders, 4,108 (38%) had at some time used pesticides occupationally, mostly in farming, and of these, 935 (23%) reported the occurrence of symptoms within 48 hours of such work on at least one occasion. This prevalence was somewhat higher in those who responded to the first mailing of the questionnaire (24%) than in those who responded only after a reminder (21%). Figure 1 shows the prevalence of individual symptoms by type of pesticide. The most frequently reported symptoms were headache (18% of all pesticide users) and aching limbs (8%), while the least common was blurred vision (1.6%). Acute symptoms were reported most often in relation to use of sheep dip (at least one symptom in 29% of users), followed by other insecticides (12%), but the relative frequency of individual symptoms was similar for all five types of pesticide.

Figure 1.

Lifetime prevalence of symptoms within 48 hours of using pesticides occupationally

When the analysis was broken down by area (Table 1), the high prevalence of symptoms following use of sheep dip was apparent in both Devon (29%) and the Welsh borders (32%). However, in Lincolnshire the prevalence of symptoms in relation to sheep dip was significantly lower (13%, p<0.001), and “other insecticides” was the class of pesticides most commonly associated with complaints.

Table 1.

Lifetime prevalence of symptoms within 48 hours of using pesticides occupationally according to area of residence and type of pesticide.

Type of pesticide	Devon		Welsh borders		Lincolnshire		All Areas
	No. of users	No (%) with any symptoms	No. of users	No (%) with any symptoms	No. of users	No (%) with any symptoms	No. of users	No (%) with any symptoms
Sheep dip	803	236 (29%)	1083	350 (32%)	224	30 (13%)	2110	616 (29%)
Other insecticides	789	87 (11%)	662	67 (10%)	585	88 (15%)	2036	242 (12%)
Herbicides	1090	80 (7%)	1107	84 (8%)	649	56 (9%)	2846	220 (8%)
Fungicides	731	46 (6%)	532	26 (5%)	552	34 (6%)	1815	106 (6%)
Wood preservatives	882	49 (6%)	876	53 (6%)	413	19 (5%)	2171	121 (6%)
Any	1557	346 (22%)	1694	447 (26%)	857	142 (23%)	4108	935 (23%)

Where acute symptoms were reported in relation to use of pesticides, they were often multiple (Table 2). In particular, 242 users had experienced four or more symptoms following pesticide use whereas only 8.8 would have been expected had the occurrence of each symptom been statistically independent. However, a search for clustering of particular subsets of symptoms did not reveal any distinctive pattern, either for pesticides overall, or for individual types of pesticide. In particular, there was no indication of an unusual excess of flu-like symptoms following use of sheep dip or insecticides as compared with other pesticides. Among the 616 users of sheep dip who complained of associated symptoms, 81% reported headache, 19% fever or chills, 45% aching limbs and 12% all three of these symptoms (although not necessarily on the same occasion). In symptomatic users of herbicides, the corresponding proportions were 75%, 12%, 23% and 8%.

Table 2.

Numbers of symptoms reported to have occurred within 48 hours of using a pesticide occupationally

Number of symptoms	Number of subjects (O)	aExpected number of subjects (E)	O/E
0	3173	2107.9	1.5
1	332	1477.3	0.2
2	249	439.3	0.6
3	112	74.7	1.5
4+	242	8.8	27.5

^aExpected number calculated with the assumption that the occurrence of each symptom in a pesticide user was statistically independent (see text).

Table 3 shows the number of men who reported acute symptoms in association with multiple categories of pesticide. One hundred and five individuals complained of symptoms following the use of at least three types of pesticide. This compared with an expected number of 1.8 had the reporting of symptoms for each type of pesticide been statistically independent.

Table 3.

Numbers of men reporting symptoms in relation to multiple types of pesticides.

aNumber of pesticide types associated with symptoms	Number of subjects (O)	bExpected number of subjects (E)	O/E
0	3173	3425.5	0.9
1	664	634.7	1.0
2	166	45.9	3.6
3	62	1.8	34.7
4 or 5	43	0.03	1234.1

^aFrom sheep dip, other insecticides, herbicides, fungicides, wood preservatives. Symptoms were “associated” if they were reported to have occurred within 48 hours of using a pesticide.

^bCalculated with the assumption that the occurrence of symptoms with each type of pesticide was statistically independent (see text).

Table 4 summarises the association of acute pesticide-related symptoms with various non-occupational risk factors. Report of such symptoms was significantly less common in men born before 1938 (PR 0.6, 95%CI 0.5-0.9 in comparison with those born during 1968-77), and in ever as compared with never smokers (PR 0.7, 95%CI 0.7-0.9). In addition, there was a positive gradient of risk in relation to somatising tendency score with a PR for the highest as compared with the lowest category of 2.4 (95%CI 2.0-3.0). This association with somatising tendency score was even stronger for report of four or more pesticide-related symptoms (PR 11.1, 95% CI 7.0-17.7), and for report of symptoms in relation to three or more categories of pesticide (PR 6.1, 95%CI 3.3-11.4).

Table 4.

Association of pesticide-related symptoms with non-occupational risk factors

Risk factor	No. of occupational pesticide-users	aNo. with pesticide related symptoms	bPR (95%)
Birth cohort
1968-77	446	103	1.0	-
1963-67	443	104	1.0	(0.8-1.3)
1958-62	493	115	1.0	(0.8-1.3)
1953-57	527	131	1.0	(0.8-1.3)
1948-52	556	133	1.0	(0.8-1.3)
1943-47	623	152	1.0	(0.8-1.3)
1938-42	577	129	0.9	(0.7-1.2)
1933-37	443	68	0.6	(0.5-0.9)
Smoking habits
Never	2,065	534	1.0	-
Ever	2,018	395	0.7	(0.7-0.9)
Average weekly alcohol consumption (units)
0	792	189	1.0	-
1-7	1440	366	1.1	(0.9-1.3)
8-21	1331	278	0.9	(0.8-1.1)
>21	537	100	0.8	(0.7-1.1)
Somatisation score
0	1317	208	1.0	-
0.1-0.5	1614	374	1.5	(1.3-1.8)
0.6-1.0	705	189	1.8	(1.5-2.2)
>1.0	424	151	2.4	(2.0-3.0)

Totals may differ due to some missing values

^aSymptom(s) within 48 hours of using a pesticide on at least one occasion

^bAll prevalence ratios are derived from a single proportional hazards model, and thus are mutually adjusted

Table 5 shows the prevalence of acute symptoms following use of different types of pesticide according to frequency of use and whether the concentrate was handled. In general, report of symptoms was more common in men who had used pesticides most often and who had worked with concentrates, this pattern being strongest for herbicides (PR 4.3, 95%CI 2.4-7.7 for use on 50+ days v <10 days, and PR 2.0, 95%CI 1.2-3.3 for work with concentrate). The only exception to this was insecticides other than sheep dip, for which there was no association with handling concentrate. Among men who had used sheep dip occupationally, the prevalence of heavy lifetime usage (50+ days) was 34% in Lincolnshire as compared with 30% in Devon and 42% in the Welsh borders. The corresponding prevalence rates for work with concentrate were 65%, 64% and 78%.

Table 5.

Association of pesticide-related symptoms with occupational risk factors

Risk factor	Sheep dip		Other insecticides		Herbicides		Fungicides		Wood preservatives
	an	bPR (95% CI)	an	bPR (95% CI)	an	bPR (95% CI)	an	bPR (95% CI)	an	bPR (95% CI)
Lifetime use of pesticide (days)
<10	69	1.0  -	21	1.0  -	12	1.0  -	10	1.0  -	16	1.0  -
10-49	205	1.5 (1.2-2.0)	38	1.1 (0.6-1.8)	43	1.8 (1.0-3.5)	13	0.8 (0.3-1.7)	39	1.6 (0.9-2.9)
50+	317	2.1 (1.6-2.8)	173	2.7 (1.7-4.2)	155	4.3 (2.4-7.7)	80	2.1 (1.1-4.2)	61	1.6 (0.9-2.9)
Handled concentrate
No	86	1.0  -	53	1.0  -	18	1.0  -	14	1.0  -	57	1.0  -
Yes	524	1.9 (1.5-2.4)	186	1.0 (0.7-1.4)	202	2.0 (1.2-3.3)	90	1.7 (0.9-3.0)	41	1.5 (1.0-2.3)

Totals may differ due to some missing values

^aNumber of users who reported symptoms within 48 hours of using the pesticide type on at least one occasion

^bPrevalence ratios were mutually adjusted and adjusted also for the non-occupational risk factors in Table 4.

Table 6 shows the risk of first reported acute symptoms following use of pesticides according to age, calendar period, and time since first occupational use of such chemicals. When associations with these variables were mutually adjusted, risk tended to be higher at younger ages, and was markedly elevated in the first year of pesticide use (IRR 4.9, 95%CI 3.5-6.8 in comparison with first pesticide use more than 15 years earlier). The risk of first symptoms by calendar period increased progressively from 1947 to 1989, but then declined. Thus, in comparison with 2000-04, the IRR in 1985-89 was 1.8 (95%CI 1.2-2.8), while that for 1947-59 was only 0.4 (95%CI 0.2-0.7). When analysis was restricted to the subset of pesticide users who had never used sheep dip occupationally (n=1,997), the risk in the first year of pesticide use was even higher (IRR 8.9, 95%CI 5.1-15.7), and the peak incidence in the late 1980s disappeared. Relative to 2000-04, risk increased gradually from 0.3 in 1947-59 to 1.1 in 1995-99.

Table 6.

Incidence of first recalled symptoms within 48 hours of using a pesticide by age, calendar period, and time since first occupational use of pesticides.

Risk Factor	No. of cases	Incidence per 1000 person-years	aIRR	(95% CI)
Age (years)
14 – 19	131	16.0	1.0	-
20 – 24	133	11.9	1.1	(0.8-1.4)
25 – 29	84	7.6	0.7	(0.5-1.0)
30 – 34	102	10.0	1.0	(0.7-1.4)
35 – 39	45	5.0	0.5	(0.3-0.8)
40 – 44	69	9.3	0.9	(0.6-1.3)
45 – 49	40	6.8	0.6	(0.4-0.9)
50 – 54	30	6.8	0.6	(0.4-1.0)
55 – 65	21	4.9	0.5	(0.3-0.9)
Calendar period
1947 – 59	16	6.0	0.4	(0.2-0.7)
1960 – 64	26	7.3	0.6	(0.3-1.0)
1965 – 69	35	7.1	0.6	(0.4-1.1)
1970 – 74	36	5.5	0.6	(0.3-0.9)
1975 – 79	80	9.6	1.0	(0.6-1.6)
1980 – 84	101	10.3	1.2	(0.8-1.9)
1985 – 89	149	14.1	1.8	(1.2-2.8)
1990 – 94	108	10.4	1.6	(1.0-2.4)
1995 – 99	77	8.0	1.4	(0.9-2.2)
2000 – 04	27	5.1	1.0	-
Years since first occupational use of pesticides
≤1	104	29.4	4.9	(3.5-6.8)
2 – 3	80	12.1	2.0	(1.4-2.8)
4 – 5	72	11.8	1.8	(1.3-2.5)
6 – 10	115	8.6	1.3	(1.0-1.8)
11 – 15	90	8.0	1.3	(0.9-1.7)
>15	194	6.3	1.0	-

^aMutually adjusted incidence rate ratios derived from a single Poisson regression model

Discussion

In this survey, the lifetime prevalence of acute symptoms following occupational use of pesticides, and particularly sheep dip, was remarkably high, but varied significantly by area. There was a strong tendency for some individuals to report multiple symptoms and symptoms related to multiple types of pesticide. However, no distinctive clusters of symptoms were apparent in relation to specific types of pesticide. The strongest risk factor for report of acute pesticide-related symptoms (particularly if multiple or related to multiple types of pesticide) was a general tendency to be worried by somatic symptoms. The occurrence of symptoms was associated with higher usage and with handling of concentrate for almost all the categories of pesticide examined. Reported first incidence of symptoms was highest in the first year that an individual used pesticides, and during the late 1980s, but the peak incidence in the 1980s disappeared when analysis was restricted to men who had worked only with pesticides other than sheep dip.

The response rate in the study was substantially lower than we have achieved in earlier postal surveys of the general population. As described elsewhere, this appeared to result, at least in part, from a requirement of the ethics committee which reviewed our protocol, that names and addresses of potential participants should not be released to the research team, and that the mailing should be carried out by organisations that already held the necessary contact details.⁹ Ethical restrictions of this sort are of dubious validity, and inevitably impair the interpretation of epidemiological research. In particular, there may have been a greater willingness to respond to our questionnaire among men who had experienced acute pesticide-related symptoms. We attempted to minimise such bias by embedding the questions about pesticides in a much broader survey of work and health, and the small difference in symptom prevalence between first-time responders (29%) and those who only responded after a reminder (24%) argues against major response bias. Furthermore, even if response rates were selectively higher for men with pesticide-related symptoms, it seems unlikely that this effect would have been differential with respect to individual symptoms, area of residence, or the presence of associated risk factors such as somatising tendency.

Another possible source of bias was incomplete recall of pesticide use and associated symptoms, especially from many years ago. This may have contributed to the apparently lower risk of symptoms in men born before 1937 (Table 4) and the low incidence of first symptoms in the earliest calendar periods covered by the study (Table 6), but again it is unlikely to have led to spurious differences in prevalence by area of residence.

The pesticide-related symptoms about which we asked were chosen because they were all potential acute toxic effects of one or more types of pesticide. For example, all of them could arise from poisoning by cholinesterase inhibitors such as organophosphate sheep dips. They included several flu-like symptoms such as headache, aching limbs and fever or chills that have previously been suggested often to occur after dipping sheep, a phenomenon known as “dippers’ flu”.⁷ However, we found no evidence that this specific combination of symptoms clustered more in users of sheep dip than of other types of pesticide such as herbicides.

We did, however, find a high overall prevalence of acute symptoms in relation to the use of pesticides, and especially sheep dip. This corresponds to a much higher incidence of pesticide-related illness than is reported to the regulatory authorities in England and Wales,⁴ and also to the National Poisons Control Centre of the Netherlands,¹⁰ but is consistent with findings from our earlier survey of agricultural workers in southern England,⁵ and with a study of farm residents in Colorado, in which 69 of 761 participants reported at some time having become ill from exposure to pesticides.¹¹

The occurrence of a symptom within 48 hours of using a pesticide does not necessarily imply that it results from acute toxicity. In some cases, the timing may be a chance coincidence, although chance is unlikely to explain the marked differentials in risk that we observed by area, calendar period and type of pesticide. In addition, symptoms could arise through psychological mechanisms. The higher frequency of symptoms in men who had worked with pesticide concentrates (which tends to give higher exposures¹²) would be compatible with a toxic effect, but several other findings in our study point to an important role of psychological factors.

First, we found a strong tendency for some individuals to report symptoms in relation to multiple classes of pesticides. This could occur if the men concerned were unusually careless in their handling of pesticides, and as a consequence were more heavily exposed than the average to a range of products. However, it is unlikely to reflect differences in individual sensitivity. Like medicines, pesticides are toxicologically diverse, and although there is some overlap between the categories that we examined (for example, pyrethroids and organophosphate cholinesterase inhibitors are used both in sheep dip and as insecticides on arable and horticultural crops), there is no established metabolic abnormality that would render substantial numbers of people unusually susceptible to three or more of the pesticide types.

Second, the patterns of symptoms reported were not specific to particular classes of pesticide. If they were a manifestation of toxicity, they would be expected to differ according to the toxicological properties of the pesticide responsible.

Third, there was marked geographical variation in the prevalence of pesticide-related symptoms, with a much higher rate of complaints among users of sheep dip in Devon and the Welsh borders than in Lincolnshire. This difference was not explained by differences in the proportion of heavy users (50+ times in total) or of those handling concentrate. It may be relevant, however, that in Devon there had been a lot of publicity in the local media about possible adverse effects of sheep dip, whereas in south Lincolnshire, the use of insecticides on brassicas was a local concern. Among farm residents in Colorado,¹¹ pesticide-related illness was reported by only two of 266 participants who had applied organophosphate products to livestock, and the occurrence of such illness was associated principally with applying organophosphates to crops.

Finally, it is notable that the report of acute pesticide-related symptoms, and especially symptoms following the use of multiple classes of pesticide, was strongly associated with a tendency to report distress from somatic symptoms (Table 4). This could occur if acute pesticide poisoning rendered large numbers of people chronically unwell and prone to a range of physical complaints, but it is also the pattern that would be expected if individuals with greater sensitivity to distress arising from somatic symptoms were more aware of symptoms and more likely to attribute them to perceived hazardous exposures. In support of the latter, a prospective community-based study in New Zealand found that baseline report of higher numbers of symptoms significantly predicted the number of symptoms that individuals attributed to a subsequent aerial pesticide spraying campaign.¹³

A higher risk of pesticide-related symptoms in the first year that an individual used such chemicals (Table 6) could occur if there was a subset of the population who were more susceptible to the toxic effects of certain pesticides, perhaps because of differences in metabolic capacity. For example, research has suggested that a polymorphism of the paraoxonase gene is associated with illness following exposure to sheep dips containing the organophosphate, diazinon,^{14, 15} although another study looking at the relation of the polymorphism to activity of the paraoxonase enzyme has cast doubt on this.¹⁶ Another explanation might be that less experienced workers have poorer working techniques, and as a consequence incur higher exposures. However, it is also possible that individuals who are more aware of somatic symptoms and more concerned about potential external causes for such complaints, are particularly liable to experience symptoms when they are first exposed to a new chemical hazard.

Findings on the first incidence of pesticide-related symptoms by calendar period are likely to have been influenced in part by better recall of more recent events, but bias of this sort would not explain the peak of incidence during the late 1980s. The latter appears to have been specific to sheep dip, since it was not apparent when analysis was restricted to men who had only been exposed to other types of pesticide. Moreover, it accords broadly with trends in the frequency of suspected acute adverse reactions to sheep dip reported to the regulatory authorities, which reached a maximum in 1991.⁷ The decline in first incidence of sheep dip-related illness since 1990 could reflect improvements in safety. For example, a new leaflet giving detailed advice on the safe handling of sheep dips was issued by the Health and Safety Executive in 1991.⁷ Another explanation might be changes in the sheep dips used. Around 1990, sheep dip products based on a number of organophosphate compounds (bromophos, carbophenothion, coumaphos, crotoxyphos and iodophenphos) lost their registration in the UK, and overall sales of organophosphate sheep dips declined fairly steadily from 1986 to 1994.⁷ Again, however, it is also possible that non-toxic mechanisms have contributed to the trend. Thus, in the interests of animal welfare, the dipping of sheep was compulsory in the UK during 1976-92. The fact that farmers were obliged to carry out the operation, whether or not they perceived a benefit for themselves, may in some cases have heightened their awareness of symptoms, with subsequent publicity in the media reinforcing concerns.

Other investigators have also reported findings that would be compatible with a psychogenic component to pesticide-related symptoms. In a study of farm-workers in North Carolina, reported exposure to pesticides in the past month was strongly associated with a wide range of symptoms during the past week, including two that would not be expected to result from such exposure (nail friability and altered taste perception).¹⁷ However, the relation of symptoms to low levels of acetyl cholinesterase was generally much weaker, and was only statistically significant for diarrhoea. And in an Egyptian study of men working with insecticides, neurological symptoms were substantially more common than in controls, with odds ratios for dizziness, numbness and tremors all in excess of eight.¹⁸ However, no significant associations were found between symptoms and serum cholinesterase levels. It is possible that the biomarkers measured in these investigations did not adequately represent the pesticide exposures that were responsible for symptoms, but an alternative explanation is that workers who know that they are exposed to pesticides are more aware of symptoms and report them more readily.

Overall, our study indicates that acute symptoms are common following work with pesticides, but suggests that in many cases the illness may arise through psychological rather than toxic mechanisms. One way of testing this further would be through a longitudinal investigation looking at health beliefs and levels of distress related to somatic perceptions as predictors of subsequent pesticide-related symptoms.