Occupational exposure to pesticides and health symptoms among family farmers in Brazil

ABSTRACT OBJECTIVE: To explore the association of occupational pesticide exposure with acute and mental health symptoms. METHODS: Cross-sectional survey carried out with 78 Brazilian family farmers, who were pesticide applicators and helpers conveniently selected. Symptoms and exposure data were collected by interviews, and mental health outcomes by the Self-Reporting Questionnaire. Blood samples were analyzed to assess cholinesterase levels. Exposure indicators and symptoms were compared between applicators and helpers, and Poisson regression was performed to estimate prevalence ratios. RESULTS: Farmers reported exposure to multiple pesticides from early ages; they worked without safety training, technical support, and full protective equipment, and they had a high prevalence of acute and mental health symptoms (e.g., headache, mucosal irritation, tachycardia, and depressive signs). Applicators had more cholinesterase changes than helpers, but less symptoms. Helpers used less personal protection and had significantly higher prevalence ratio of headache, dyspnea, wheezing, cough, poor digestion, tiredness, and feeling worthless, after adjustment. CONCLUSIONS: Acute and mental health symptoms were observed, both among farmers and helpers. Thus, surveillance actions must be reinforced in Brazil, technical support and safety training improved, focused on applicators and helpers, who are occupationally and environmentally exposed to pesticides. Agricultural practices of these groups with less pesticide use should receive incentive.


INTRODUCTION
Excessive and unsafe use of pesticides represents a serious risk to human health, environment, and quality of food. About 25 million people experience unintentional pesticide poisoning yearly worldwide 1 , resulting in 200,000 deaths, mainly affecting low-and middle-income countries (LMIC) 2 . In LMIC, the occupational exposure to pesticides has been associated with gastrointestinal, musculoskeletal, respiratory, allergic, and nervous effects [3][4][5][6] , and common mental disorders (CMD) such as depression, anxiety, and suicide [7][8][9] . However, these adverse effects are not restricted to LMIC, and occupational exposure to pesticides was associated with health outcomes in high-income countries such as the USA, England, South Korea, and Spain 10,11 .
Between 2010 and 2015, more than 600,000 pesticide poisoning cases and 2,074 deaths occurred in Brazil 12 , but the cases are vastly underreported by national information systems. It is estimated that for every registered case there are 50 unregistered ones 13 . Mental illness is a major public health concern in terms of lost health and burden of disease, and its symptoms are often overlooked by health services 14 . Depression and anxiety affect, respectively, 5.8% and 9.3% of the Brazilian population, more than 4.4% and 3.6% affected worldwide 15 .
Farmers from LMIC, mostly located in tropical areas with easy pest proliferation, tend to be more exposed to pesticides due to the lack of safety regulation, surveillance and training, increased use of highly toxic chemicals, low risk awareness, misuse of personal protective equipment (PPE), and careless handling and pulverization 2,4,5 . Studies conducted in Brazil showed that farmers commonly use complex mixtures of pesticides without precautionary measures, which could potentially reduce exposure and protect their health [16][17][18] .
Brazil is a world's leading agricultural producer, and the largest consumer of pesticides since 2008, trading highly toxic chemicals banned in many countries 13 . Cholinesterase-inhibitor pesticides, such as organophosphates (OP) and carbamates (CM), represent an important risk for human health and are considered the main responsible for pesticide poisoning in LMIC 4 .
Family farming is the primary income source for 40% of the active population in Brazil and for 90% of municipalities with fewer than 20,000 inhabitants 19 .
Although studies conducted in Brazil had explored the health effect of pesticide use 8,9,18,20 , many regions and crops are still underrepresented, and mental health and tomato growers were not studied recently. Therefore, this study aims to explore the association between pesticide exposure and the prevalence of self-reported acute and mental health symptoms among family farmers in São José de Ubá, state of Rio de Janeiro.

Study Area and Sample
This cross-sectional study was conducted in July and August 2014, at the end of the crop season, in São José de Ubá (SJU), located in a mountainous region in the Northwest of the state of Rio de Janeiro, Brazil (Figure). SJU is a small municipality of 7,000 inhabitants, where 55% live in the rural area, 16% has formal employment, and 40% have a monthly per capita income equivalent to $100 US dollars or less than half the national minimum wage 19 . SJU is one of the largest tomato producers in Brazil, and the municipality income depends on smallholder family farming 19 . Tomato cultivation demands an intensive care for pest control, commonly based on the use of large amounts of pesticides 13 . Between 2007 and 2017, the yearly tomato production in SJU ranged between 21,000 and 32,000 tons 19 . Previous studies revealed soil degradation and water contamination as a consequence of intensive farming and livestock practices 21 .
The study sample was composed of 78 family farmers, older than 18 years, recruited sequentially by convenience in agricultural areas of SJU with word-of-mouth of residents and stakeholders (snowball sampling). Almost all (95%) individuals recruited agreed to participate in the study, and participants represented about 11% of tomato growers in SJU. The final sample size was delimited by the time and budget constraints of the project.
This study is part of a broader project assessing human health risk by pesticides and metals exposure in SJU, and the study methodology 21 and pulmonary function impairment 22 were discussed elsewhere. The study protocol was approved by the Ethical Board of the Hospital Universitário Clementino Fraga Filho of the Universidade Federal do Rio de Janeiro, and all participants provided written informed consent. Test results were delivered individually, and guidance on health protection provided. Individuals with significant outcomes were referred to the municipal healthcare service. Based on observational visits and self-declared cultivation tasks, participants (n=78) were divided into two groups as a proxy of exposure: a) applicators (n=42), who were daily involved in all crop activities, including pesticide handling and spraying; and b) current or former helpers (n=36), who were farmer's relatives performing additional crop-related tasks, but not pesticide spraying.
Individual interviews based on questionnaire were carried out at participant's households, neighborhood schools and healthcare units to ease community engagement and assimilate the local culture. Socioeconomic and demographic data such as age, gender, marital status, body mass index (BMI), educational level, smoking habits, and alcohol consumption were obtained. Monthly family income was estimated based on the Brazilian minimum wage (R$ 998 in 2019), which was equivalent to approximately US$ 240 and now is around US$ 200.

Exposure Assessment
Participants were asked about their current and previous pesticide exposure, including: duration of work with pesticides, age of first exposure working or helping at crops, home distance from crop areas, use of recommended PPE (cloth mask, visor, hat, gloves, boots, and overall), types of pesticide frequently used, previous training and technical support received, residential exposure to pesticides (either by use for pest control or by contact with contaminated clothes and equipment), and poisoning history. Safety practices (e.g. eating at the field, washing hands, and showering after crop activities) were also investigated.  Some farmers refused to participate in blood tests or had insufficient samples. Samples from 70 participants were collected by qualified health personnel to evaluate acetylcholinesterase (AChE) and butyrylcholinesterase (BChE) inhibition, which are respectively biomarkers of chronic and acute exposure to OP and CM pesticides 13 . The AChE and BChE biomarkers are useful for an initial screening of pesticide poisoning, although they have low specificity and sensitivity, and they are not effective to assess other chemical classes of pesticides 17 .
Measurements were performed by the Centro de Estudos da Saúde do Trabalhador e Ecologia Humana (CESTEH/FIOCRUZ) by the modified Ellman method 23 , in which reference values were defined as 0.56 mmol/min/mg for AChE for both genders, and 2.29 for BChE for men and 1.61 mmol/min/mg for women. Values above these were considered normal.

Self-Reported Symptoms and Mental Health Assessment
Twenty-three acute symptoms previously associated with pesticide poisoning were presented to the participants, who were asked to confirm whether they regularly had them or not. Prevalence of probable CMD was assessed by the Self-Reporting Questionnaire (SRQ-20), proposed by WHO as a low-cost and easy tool for psychiatric screening, and it is recommended for community studies and basic care. The questionnaire has 20 binaries questions (yes/no) about depressive and anxiety signs, reduced vital energy, and somatic symptoms. It was validated in Brazil with good internal consistency 24 , high sensitivity (83%) and specificity (80%) 25 , and the standard cutoff was set as six or more positive answers for men and eight or more for women 14 .

Statistical Analysis
The prevalence of sociodemographic and pesticide exposure data was compared between occupational groups (pesticide applicators versus helpers). Data with normal distribution were presented as mean and standard deviation (SD), and with non-normal distribution as median and interquartile range (IQR 25-75%). Pearson's chi-square test was used to compare categorical variables, and Fisher's exact test was used for variables with less than five observations. Continuous variables with normal distribution were compared among groups by T-test, and with non-normal distribution by Rank-sum. Cholinesterase activity was compared with the reference values and among groups 23 . The prevalence of each acute and mental health symptom was compared between applicators and helpers, and possible associations tested by Poisson regression with robust variance, adjusted by potential confounders, namely age, BMI, smoking habit, gender, and alcohol consumption. Family income was similar among groups; thus it was not included in the analysis. Statistical analysis was performed using Stata 14 (Stata Corp., College Station, TX, USA) and SPSS 23 (IBM Corp., Chicago, IL, USA). p-values < 0.05 were considered significant.

RESULTS
Most participants were married, had low educational level and family income, they had never smoked or were former smokers, and did not consume alcohol. Most applicators were men (83.3%), had a mean age of 40.3 years, and worked with crops for 27 years. Current or former helpers were mostly women (72.2%), had a mean age of 48.4 years, worked or helped in crop activities for 27 years, and had a significantly higher BMI (Table 1).
Most participants started to help in crop activities at early ages, lived up to 1km from crop areas, never had technical support or safety training to work with pesticides, and they were domestically exposed to pesticides by using them for household pest control or having contact with contaminated clothes or equipment. Most applicators directly handled or sprayed pesticides for 1-3 days (85.7%) or 4-7 days (11.9%) per week during the crop season, while helpers never or rarely handled pesticides. Applicators affirmed to use significantly more PPE than helpers (although not complete), and considered visors hot and uncomfortable, thus they did not use them often ( Table 2).
All participants had normal AChE values (above the reference level), while 11 applicators (28%) and three helpers (10%) presented inhibited BChE (below the reference level), without significant differences between groups. Applicators presented lower mean values for AChE and BChE, but they were not statistically significant ( Table 2). Regarding the products used, 49 pesticides from 31 chemical groups were cited, mainly OP pesticides such as Acephate and Chlorpyrifos, CM pesticides such as Mancozeb and Methomyl, pyrethroids such as Lambda-Cyhalothrin and Deltamethrin, and also nitriles, diamides, neonicotinoids, avermectins, and benzimidazole. Most of these chemicals are classified as extremely and highly toxic to humans 26 . About 30% of applicators did not know which pesticides they were applying because someone else does the mixture and they only apply. Glyphosate and Paraquat were cited by 35% and 17%, respectively, but they are prohibited for tomato cultivation in Brazil 13 . The extremely toxics Chlorpyrifos and 2,4-D are forbidden for tomato crops in Brazil, and the highly toxic Endosulfan was already banned, but it was also mentioned.
Generally, symptoms were more prevalent among helpers than applicators, except for blurred vision, excessive salivation, and tachycardia. Significant differences among occupational . Moreover, a prevalence higher than 30% was observed for dizziness, photophobia, stomach pain, palpitation, and cough among helpers, and for headache, mucosal irritation, and tachycardia in both groups, although without significant differences between groups (Table 3).
Helpers showed higher prevalence of all mental health symptoms in the SRQ-20, except for daily work suffering. Statistically significant differences in PR were observed between occupational groups for poor sleep, feeling easily tired, unable to play a useful part, and feeling worthless, all with higher PR among helpers, while daily work suffering was higher among applicators. After adjustment for age, BMI, smoking habit, gender and alcohol consumption, higher PR among helpers was observed for poor digestion (PR = 7.85; 95%CI 1.17-52.89), feeling easily tired (PR = 3.20; 95%CI 1.33-7.66), and feeling worthless (PR = 7.23; 95%CI 1.69-31.04). Furthermore, some mental health symptoms had a prevalence as high as 50% or 30%, but without significant differences between groups (Table 4).

DISCUSSION
Our study shows that smallholder family farmers in SJU were occupationally and environmentally exposed to pesticides from an early age, lived near crops, worked without safety training, technical support and full recommended PPE, and had a considerable number of acute and mental health symptoms. First, our hypothesis was that applicators are more exposed to pesticides and have more symptoms than helpers who assist in crop activities; however, most symptoms had higher prevalence among current and former helpers, even after adjusting for possible confounders, including gender. An explanation may be that helpers are also very exposed to pesticides due to their higher residential exposure, lower training, and involvement in re-entry tasks on the same day or day after spraying using less PPE. Selection bias could have occurred when considering the farmers who stopped working because they felt ill as former helpers, though it could also be due to residual uncontrolled confounders. Compared with helpers, participants in the pesticide applicators group were mostly younger men who started to work in agriculture at an earlier age and handled multiple pesticides about three times a week, had a higher duration of exposure at crops, but lower residential exposure.
The symptomatology presented in this study must be interpreted with caution due to the small sample size, although our findings are supported by previous studies conducted in other LMIC, and also high-income countries. In our study, farmworkers performing re-entry activities were less trained, used fewer PPE, performed less hygienic practices, and showed more symptoms than pesticide applicators from Ethiopia 6 and Chile 4 .
Moreover, participants in our study had less safety guidance, used less protection and had higher prevalence of symptoms than both pesticide applicators and non-applicators from Zanzibar 27 . On the other hand, coffee farmers from Dominican Republic exposed to multiple pesticides without using PPE (e.g., only 13% used masks and gloves) had a higher prevalence of all symptoms than participants in our study 5 . These findings emphasize the role of protection equipment and technical support in poisoning prevention.
High-income countries have overcome the acute pesticide poisonings and are more concerned about chronic effects and long-term exposure. But acute and chronic poisonings still persist and are growing problems for LMIC 6 . According to the WHO, an acute pesticide poisoning must present clear signs of exposure, temporal cause-effect relationship, and at least three symptoms compatible with the exposure 28 . In our study, participants were asked to confirm which symptoms they regularly have, so we cannot ensure temporality; however, most individuals were continuously exposed, and 60% of them had more than four acute symptoms suggestive of pesticide poisoning.
Regarding mental health, helpers in SJU had higher prevalence than applicators in all questions about depressive and anxiety symptoms, and nearly twice as high CMD prevalence (44% versus 24%). Other studies assessed the CMD of Brazilian farmers with SRQ-20, and results vary widely. A study found a CMD prevalence of 34%, being significantly higher for women (40%) than men (26%) 7 . Another study with tobacco farmers found a CMD prevalence of 12%. Although this prevalence is much lower than in our study, they observed a higher PR for women (PR = 1.39; 95%CI 1.12-1.72), and individuals performing re-entry tasks (PR = 1.71; 95%CI 1.33-2.20) 8 , which is consistent with our study. Another study with family farmers in Brazil reported a CMD prevalence of 27% for both genders, which is higher than for applicators but lower than for helpers in our study 9 .
A review found that a higher prevalence of mental health outcomes, notably depression, anxiety, and suicide attempts, were found positively associated with exposure to pesticide among farmers in high-income countries as the USA, England, South Korea, and Spain 10 .
Depression was significantly associated with a history of pesticide poisoning, but not low or high cumulative exposure in spouses of pesticide applicators in the USA 29 .
The high variability of SRQ-20 can be explained by the questionnaire high sensitivity and capacity to identify a large spectrum of affections 9 , but also by biological and social factors (e.g. women tend to report more symptoms, and less educated individuals tend to over-report mental health complaints) 14 .
Previous studies conducted with Brazilian farmers indicated a scenario of exposure similar to our findings: low educational level and income, residential proximity to crop areas, working in crops since childhood, poor technical support and safety training, and lack or misuse of PPE 16,18 . Inadequate safety practices (e.g. drinking and eating at field and showering only at the end of workday) may increase pesticide exposure 3,4 . Moreover, poor understanding of pesticide instruction leaflets, and influence of neighbors on pesticide user's decisions, especially regarding handling and dosage, may compromise farmer's ability to reduce exposure and protect their health 16 .
In Brazil, complex mixtures of multiple pesticides are commonly sprayed by manual pumping or backpack tanks by smallholder family farmers 16,17 , which may result in potential additive and synergic effects, and greater health outcomes 1,13 . Pesticides are authorized for specific target-crops in Brazil, and the use of prohibited chemicals or those not allowed for tomato crops was observed in our study. Some pesticides banned or sale-restricted in other countries due to their high toxicity, such as Abamectin, Acephate, Glyphosate, and Paraquat, are still highly commercialized in Brazil 13 , and their use was mentioned in our study. Most symptoms found in our study were previously associated with OP pesticides exposure 11 , but the effects of pesticide mixtures cannot be ruled out and deserve further attention.
Major challenges on evaluating health effects of pesticide exposure among family farmers in Brazil are their widespread distribution, their continuous exposure to multiple chemicals, large distance of crop areas to health services, and the shortage of laboratories with analytical capacity 13,17 . According to the Brazilian law, all farmers must be periodically subjected to medical exams and cholinesterase tests, albeit this is not provided to millions of smallholder family farmers distributed in 4.4 million properties 19  Another study with Brazilian tomato farmers showed that 62% reported more than one illness after pesticide exposure, but only 21% of poisoned workers sought a health service and 70% self-medicated 20 . This low demand for health services, especially in less severe cases, compromises the visibility and hinders the understanding of the real dimension of the problem 13 .
The lack of an unexposed control group and the cross-sectional design were some limitations of this study. A longitudinal study with more sensible biomarkers would enable the evaluation of health effects over time and precisely associate them with exposure. Moreover, the convenience sampling and sample size of our study limit external validity. Thus, this study assumes an exploratory purpose, which only allows the assessment of associations but not cause-effect relationships, although indicating relevant hypotheses to be further investigated. Possible information and memory bias were minimized by an experienced health professional conducting the interviews, the access to recent symptoms, and trust relationships with participants. Moreover, our questionnaire was based on validated protocols, and pilot interviews ensured that the questionnaire was easily understood.
This investigation contributes to filling the data gap concerning occupational exposure to pesticides and health effects in Brazil by providing complementary evidence about family farmers, responsible for about 70% of food (mostly vegetables) consumed in the country 13 . This study highlights the demand for a more efficient technical support stimulating farmers to use chemicals consciously and better protect themselves. The strengthening of public policies that address the current vulnerability and risk of family farmers is urgent. Moreover, we strongly recommend a careful overhaul of the Brazilian legislation to restrict hazardous pesticides and encourage agricultural practices less dependent on pesticides for family farmers.

CONCLUSIONS
Our study indicates that Brazilian family farmers work without proper technical support and safety behavior that could reduce pesticide exposure and protect their health. Also, it highlights that public policies must target not only the applicators, but also other workers who often assist in crop activities and vulnerable groups both occupationally and environmentally exposed. The high exposure to pesticides must be a major public health concern because it reduces farmers' quality of life, affects the rural workforce, increases morbidity and mortality burden of diseases and health costs. Thus, we strongly recommend the strengthening of surveillance actions, technical support, and educational programs aiming at family farmers exposed to pesticides in Brazil. Promoting sustainable agriculture is the most effective way to protect farmers' health, general population, and the environment.