METHODS

Time and setting of the study

This study was conducted in the period from June to September 2000, when pesticides were applied to cotton crops in the field. It was carried out at Menoufiya Governorate, Egypt, where cotton is grown and pesticides are widely used in a large number of fields. A typical workday was from 8 00 am to 12 00 pm, then from 3 00 pm to 7 00 pm, six days per week. The participants were examined during workdays at midday break from about 12 00 pm to 3 00 pm, during which they ate a meal and avoided exposure to the high temperature at that part of the day.

The tests were administered in offices inside buildings of the pesticide application departments (10 buildings in different villages), which belonged to the Egyptian Organisation for Agriculture Credit and Cooperation, Ministry of Agriculture. Sites were selected to conduct examinations and testing in a quiet, comfortable environment with minimal distractions. Each test setting contained office furniture and a table suitable for a clinical examination.

Participants

Pesticide exposed

An official permission letter was obtained from the Ministry of Agriculture to ensure management cooperation. The purpose of the study was explained to the managers and employees at each of the 10 sites. All participants in this study were volunteers who signed a statement of formal consent before testing. The consent form was developed according to the international ethical guidelines for biomedical research involving human subjects prepared by the Council for International Organisations for Medical Sciences in collaboration with the World Health Organisation (Geneva, 1993). The consent form was reviewed and approved by the Ministry of Agriculture, Egypt, which employed and insured the workers recruited into this study. Each participant was asked to do his best on the tests.

Fifty two males were recruited from those who are working in the pesticide application departments at Berket El-Sabe district, Menoufiya Governorate, Egypt (n = 86). Sixty four (74%) employees were eligible for inclusion in the study, while 22 (26%) were ineligible because they were either females or service workers who had never been involved in pesticide application. Among the eligible group, 12 (19%) were excluded or declined to participate, so the overall response rate was 81%. Agriculture engineers constituted 37% of the exposed group. They worked in the fields and were responsible for the entire application procedure, including documentation to register seasonal workers, and types and amounts of pesticides used. Also in the exposed group were assistant agriculture engineers (38%), mixers (10%), and mechanics to repair the spraying devices (15%). The tasks in these job categories required workers to remain in the field throughout the spraying period. Each exposed participant performed these activities virtually every working day, six days per week, during the period between June and September. Table 1 summarises demographic characteristics of the exposed and control participants.

Most participants (88%) reported that they had never used protective clothing; three (6%) of those remaining occasionally, and three (6%) frequently used cloth cone-like masks worn over the nose and mouth to prevent particulate exposure.

All exposed participants had been involved in pesticide application on almost all crops and orchards found in Egypt (including cotton crops). During the past three years they had participated extensively in pesticide applications on cotton crops only. No acute poisoning incidents that led to hospitalisation were reported either during the study or in the past.

Unexposed controls

A control group consisting of 50 male clerks and administrators who had never been exposed to pesticides in their work was recruited from different departments at the Ministry of Agriculture. The response rate among controls invited to participate was 79%. The control participants were from the same socioeconomic class (based on the 1983 criteria of El-Sherbiney and colleagues¹⁴), lived in the same district, and worked in buildings near the cotton fields where the exposed participants worked. The number of clerks and administrators exceeded the number needed to carry out the work of the department. Therefore, most of them did not spend the entire work time (officially between 8 00 am and 2 00 pm) inside offices. Thus, the time spent by the exposed and control participants working with papers, figures, or numbers was roughly equivalent.

Criteria for exclusion

Based on personal data and medical histories, 25 potential exposed and control study participants were excluded because they reported that they were seasonal workers, had less than 12 years of education, or had any of several medical diagnoses (diabetes mellitus, liver or kidney disease, peripheral neuropathy, vitamin deficiency, anaemia, addiction), long term treatment with psychotropic drugs, prior history of head injury resulting in loss of consciousness, or recent exposure to other neurotoxic agents.

Spraying schedule

Based on Egyptian Ministry of Agriculture guidelines, cotton crops were regularly treated with pesticides from the first days of June to the first days of September in 2000. Table 2 lists types and trade names of pesticides used. Typically, pesticides were applied once every 15 days. Insect growth regulators were applied in the early stages of cotton growth, then, according to the severity of infestation, either one or a combination of two or three of the OP, carbamate, or pyrethroid insecticides were applied. During this season, ambient temperatures were 30–37°C and the humidity was typically over 75%. The pesticides were applied by knapsack sprayers (carrying 120 litres of pesticides) or motorised sprayers such as tractors (400–600 litres).

Spraying was carried out by a team of 13–15 persons. This team consisted of 6–7 applicators (seasonal workers who were not included in the study), 5–6 agricultural engineers and assistants, a mixer, and a mechanic. The agriculture engineers, assistants, and seasonal applicators worked closely in the field and their main mode of exposure was by inhalation. Mixers were exposed by both inhalation and skin contact. Mechanics were exposed by inhalation, skin contact, and occasionally ingestion during the repair of old machines.

Tests administered

All participants were subjected to a personal, occupational, and medical history questionnaire. A clinical examination consisting of general, chest, heart, and abdominal examination was administered, along with specific neurological tests for sensory and motor functions including: (1) cranial nerves (for example, sense of smell, visual acuity, visual fields, pupillary reactions, extraocular movements, corneal reflexes, hearing and gag reflex); (2) motor system (for example, involuntary movements, muscle status, muscle tone, muscle power, and coordination); (3) reflexes (for example, knee and ankle reflexes and plantar response); (4) sensory system (superficial sensations—for example, pain and touch sensation; deep sensations—for example, vibration sense; position and discriminate sensations—for example, stereognosis, number identification, and two point discrimination).¹⁵ A series of neurobehavioural tests were also administered: (1) Similarities (test of verbal abstraction); (2) Digit Symbol and Trailmaking part A and B (visuomotor speed); (3) Block Design (problem solving); (4) Paced Auditory Serial Addition Test (PASAT), Letter Cancellation (attention); (5) Digit Span, Benton Visual Retention Test (BVRT), Story Recall parts A and B (memory); and (6) Eysenck Personality Questionnaire (EPQ) (personality).

Better performance is evaluated by higher scores obtained on tests of Similarities, Digit Symbol, Block Design, PASAT, Digit Span, BVRT, and Story Recall tests; by contrast, lower latencies or time to complete Trailmaking part A and B tests and a lower score on the Letter Cancellation test indicated better performance.¹⁶

A 5 ml sample of venous blood was taken under complete aseptic precautions from 45 of 52 exposed and from 37 of 50 unexposed participants. Seven and 13 of the exposed and control participants, respectively, either refused to give blood or the quantity of blood collected was not sufficient to complete the analysis. Blood samples were sent for analysis of serum acetylcholinesterase (AChE) and liver and kidney function tests.

AChE was measured by HPLC using the colorometric endpoint method according to Rappaport and colleagues¹⁷ (Sigma-Aldrich Corporation, 1997). Liver function (for example, AST, ALT, alkaline phosphatase, serum bilirubin, albumin, globulin, and total proteins) and kidney function (for example, blood urea and serum creatinine) tests were done to exclude participants with any renal or kidney disease who would be at risk of encephalopathy.¹⁵

Blood samples were collected form all participants after completion of the questionnaire, clinical examination, and neurobehavioural test battery. In the only exception to this, questionnaires were administered to six participants in the fields and blood samples were collected at the same place; clinical examination and the neurobehavioural test battery were carried out later in the same testing room used for other participants.

RESULTS

Demographics

The mean age of the exposed participants (43.63 years) was slightly older than that of controls (42.48 years), but the difference was not significant (p = 0.29) (see table 1). The mean duration of Ministry of Agriculture employment of the exposed participants was 18.04 years, and the mean of the control group was 17.08 years. This difference was also not significant (p = 0.47). Most of the exposed and control participants had completed secondary school education (77% and 68%, respectively), while the remainder had completed university education (23% and 32%, respectively). The difference was not significant (p = 0.37).

Measures reflecting lifestyle choices did not differ between the groups. Body mass index (BMI) was measured because OP pesticides have been reported to reduce food intake,¹⁸ and the concentration of certain pesticides may remain higher in obese people for a longer time.^19,20 However, the mean BMI of the exposed and control participants was virtually the same (p = 0.57). Self report of smoking habits of the two groups was not significantly different, none of the participants reported a history of drinking alcohol, and the mean number of cups of tea and/or coffee consumed per day was virtually the same in both groups (table 1).

The mean level of serum AChE, an indicator of exposure to OP pesticides, was significantly lower in the exposed (87.34 U/ml) than in control (108.25 U/ml) participants (p = 0.0001). Multiple regression analysis showed that serum AChE was significantly associated with exposure (β = −21.34, SE 3.94, 95% CI −29.18 to −13.50), while age, education, and BMI had no significant effect on AChE (F = 0.73, df = 3, 77, p > 0.05).

Neurobehavioural performance

Exposed workers exhibited lower performance than controls on most of the neurobehavioural tests (table 3). Although the groups did not differ significantly with respect to age and education, these variables are so well established as confounders of neurobehavioural performance,^16,21 that multiple linear regression analysis was used to adjust for their influence (with the midpoints of each category as the independent variable). Holm’s modification of the Bonferroni correction equation was applied to adjust for multiple comparisons.²² After adjustment for confounders and multiplicity, performance of the exposed participants was significantly lower on Similarities, Digit Symbol, Trailmaking part A and B, Letter Cancel, Digit Span Forward and Backward, and Benton Visual Retention (table 3).

Effects of serum AChE and duration of work in pesticide application departments on neurobehavioural performance were tested using regression analysis models. Age and education were used in the models as covariates to eliminate their impact, as these factors also co-vary with duration of exposure. Serum AChE did not show any significant effect on performance of the exposed workers. In contrast, there was a trend towards lower performance scores as duration of exposure increased; the difference was significant on Similarities, Digit Symbol, Trailmaking part A and B, Digit Span Forward and Backward, and Benton Visual Retention (table 4). Negative association was reported on the Similarities, Digit Symbol, Digit Span Forward and Backward, and BVRT tests, all of which measured the number of items completed in a fixed time (that is, higher scores reflect better performance). Conversely, there was a significant positive association with Trailmaking part A and B, in which time to completion was the primary measure (that is, lower scores reflect better performance).

Personality and clinical assessment

The Eysenck personality questionnaire revealed that the mean neuroticism score of the exposed participants was significantly higher than that of the controls (p = 0.01), while psychoticism, extraversion, and criminality showed no significant differences between the exposed and control participants (table 5).

Symptoms of dizziness and numbness were significantly higher in the exposed than the control group. While other symptoms and signs were higher in the exposed participants, they were not significant (table 6). Most of the reported neurological symptoms and signs were significantly higher in older workers who, of course, also had the longest work duration, compared to those who were younger and had a shorter duration of work. Serum AChE showed no significant differences (p > 0.05) between those with positive or negative neurological symptoms or signs, although there was a significant trend of lower AChE levels as exposure duration increased (β = −1.18, SE 0.19, 95% CI −1.55 to −0.81 (fig 1).

DISCUSSION

The exposed participants showed a consistent, statistically significant pattern of lower neurobehavioural functioning compared to controls, after adjustment for multiple comparisons.²² This pattern was supported by multiple regression analysis, which adjusted for age and education, the strongest contributors to neurobehavioural performance.^16,21 The deficits were seen in a wide array of neurobehavioural functions, including verbal abstraction (Similarities), visuomotor speed (Digit Symbol and Trailmaking part A and B), visual attention (Letter Cancellation), auditory attention and memory (Digit Span), and visual memory (BVRT) (table 3).

Additional analyses supported the association of the neurobehavioural deficits with work in agriculture. Duration of work in agriculture was inversely correlated with neurobehavioural performance (table 4). The longer the period of agricultural work, and thus exposure to pesticides, the greater the performance deficits of the exposed participants relative to controls. By contrast, there was no evidence that the performance was affected by current OP pesticide exposure. All serum AChE values were within normal limits (40–120 U/ml according to the 1997 Sigma-Aldrich Corporation manual), suggesting that participants were free from acute OP effects at the time of testing. Also, most of the neurological symptoms and signs were markedly prominent among older age participants and among participants with more years of exposure. This is in accordance with Davignon and colleagues²³ and Amr and colleagues²⁴ who reported that annual exposure to pesticides over a period of 10–20 years can result in mild to severe deterioration in neurological function.

Comparison with previous research

The reported neurobehavioural deficits in chronically exposed workers can be compared to other studies of both chronically exposed and acutely poisoned individuals (table 7). In two studies of chronically exposed sheep farmers in the UK, farmers performed significantly worse than controls (non-exposed quarry workers) on tests measuring sustained attention and speed of information processing, including the Digit Symbol test.¹¹ Savage and colleagues,⁴ Rosenstock and colleagues,⁵ and Steenland and colleagues⁶ examined the possible non-reversing effects in people who have been poisoned or seen by a physician following overexposure to OP compounds. Each study has limitations, but together they reveal consistent differences in these populations, a pattern of deficits that is increasingly credible because of the consistency of the findings among the studies. These consistent findings of deficits are seen on the Digit Symbol, Trailmaking, Block Design, Digit Span, and Benton Visual Retention tests (table 7). The findings in the present study parallel those in Stephens and colleagues¹¹ and other studies of OP exposures, but suggest that the neurobehavioural functions affected are far broader than previously reported (table 7, column 2).

The scores on the tests in this study are lower when compared to scores reported in many other countries.^25–27 Figure 2 presents data from unexposed controls in these studies. A major concern when performance is low is that control performance may be so near the performance “floor” that true pesticide related differences could not be detected. However, in the present study, while dramatic clinical differences were not seen between exposed and control participants (for example, table 6), many test differences were significant, even after adjustment for multiple comparisons (table 3). The lower scores may arguably be attributed to differences in the educational background and the general inexperience with the testing situation in some countries.^26,27 The participants in the present study were encouraged to apply themselves to the tests and were observed to work diligently on the tests, supporting the validity of the results.

Personality measures

Regarding the assessment of personality, the Lie scale (L) score was not significantly different between exposed (mean 8.02 (SD 1.52)) and control (8.40 (1.39)) participants (p = 0.19). This lack of difference supports the accuracy of the results of the other scales. Further, the Lie scale is designed to indicate the extent to which the individual is being honest or trying to create a false impression (faking bad),^28,29 and thus also supports the validity of the neurobehavioural test performance scores. This study revealed that the mean (SD) neuroticism score of the exposed participants (9.98 (1.16)) was significantly higher than that of the controls (9.24 (1.61)) (p = 0.01), while psychoticism, extraversion, and criminality showed no significant group differences (p > 0.01) (table 5). Neuroticism is the name Eysenck^28,29 gave to a dimension that ranges from normal, fairly calm and collected people to those who tend to be quite “nervous”. Other investigators have also reported psychiatric disorders, particularly neurosis, anxiety, obsessive-compulsive behaviour, phobias, temper outbursts, irritability, and depression in OP exposed participants.^{11,12,25,30–34}

Acetylcholinesterase measure

Serum acetylcholinesterase was significantly lower in the exposed than the control participants (β = −21.34, SE 3.94, 95% CI −29.18 to −13.50). As Namba and colleagues³⁴ noted, OP pesticides do not accumulate in tissue. However, a cumulative effect is produced in the form of reduction of AChE and other tissue esterases since these enzymes serve as buffers, protecting AChE of the synapse.³⁵ This reduction may be related to the phenomenon of aging of AChE as a result of long term enzyme inhibition,³⁶ as supported by the significant negative correlation between AChE and duration of exposure in the present study (fig 1). Clinical experience with workers exposed to OP compounds reveals that some may exhibit mild symptoms of intoxication without any significant change in blood AChE activity.^37–39 It has been suggested that AChE is insensitive to subtle, subclinical alternations in motor and cognitive functions, possibly related to the low level OP exposure.³⁹

Broader range of OP effects suggested

The neurological symptoms of dizziness and numbness were significantly more prevalent in exposed than control workers in the present study (OR = 6.75 and 12.5, respectively, table 6). Odds ratios in table 6 are computed from a logistic regression model. Confidence intervals for odds ratio are based on the likelihood ratio test and are known to be more accurate for small sample sizes.⁴⁰ Amr⁴¹ reported that blurred vision, dizziness, numbness, paraesthesia, headache, vertigo, asthenia, superficial sensory loss, trophic and vasomotor changes, and decreased ankle and deep reflexes were more prevalent among Egyptian pesticide applicators than controls. These findings may suggest that these groups of Egyptian cotton pesticide applicators have sustained higher exposures than seen in the other studies noted here.^9,11,12 The symptoms of dizziness and numbness may suggest the beginning of nervous system effects that could be on the verge of becoming clinically detectable. This may explain why the exposed participants in this study had deficits on so many more neurobehavioural tests than in the other studies of pesticide exposed workers. If this is true, the effects of low to moderate level chronic OP exposure may be more wide ranging than previously realised. Chronic OP exposure has been associated with visuomotor speed as indicated by the Digit Symbol and Trailmaking B tests.^9,11 The results of the present study confirm these findings. However, the present study also suggests that moderate exposure to OP compounds over several years for approximately 120 days per year, may also be associated with deficits on verbal abstraction, attention, and memory. The implication of these findings may be considerable, especially in the developing world, where poisoning is increasingly common and where economic, climatic, social, and cultural conditions are likely to thwart the implementation of preventive approaches to poisoning that have been successfully used in industrialised nations.^42,43