Pesticide Safety Behavior among Cocoa Farmers in Nigeria: Current Trends and Determinants

ABSTRACT Background There is a strong link between pesticide exposure risk and farmers’ safety behavior. However, there is a lack of understanding of farmers’ safety behavior in pesticide use and the psychological factors that influence it, especially in Nigeria. Objective This study aimed to identify safety behaviors in pesticide use and their determinants among cocoa farmers in Nigeria using the Health Belief Model (HBM). Methods We used a suitable questionnaire to retrieve data from 391 cocoa farmers across three major cocoa-producing belts in Nigeria. The data were analyzed using both descriptive and inferential statistics. Results Findings showed that cocoa farmers in this study sprayed insecticides, herbicides, and fungicides on their cocoa farms. The study revealed that most pesticides were not approved for cocoa production by the authority, but the farmers were still using them. The study participants overused surveyed pesticides as the quantities applied were beyond the recommended dosage by the manufacturers. Correlation analysis showed a relationship between education status, pesticide training, perceived susceptibility and cue to action, and the farmers’ pesticide overuse. In all, 42.7% of the farmers showed unsafe behaviors, 49.1% showed intermediate behaviors, and just 8.2% reported safe behaviors in pesticide use. Multiple linear regression analysis showed that perceived barriers, susceptibility, self-efficacy, and cue to action were related to age, education status, and farm experience and accounted for 59.3% of the total variance in the farmers’ pesticide safety behaviors. Perceived barriers had the most significant negative influence on the farmers’ behaviors, while perceived susceptibility, self-efficacy, and cue to action positively influenced the farmers’ behavior. Conclusion Future intervention programs in Nigeria incorporating significant variables, especially perceived barriers, in their designs with proper implementation could be more effective.


Introduction
Pesticides are used in global crop production to combat the destruction caused by pests and diseases. 1 This is due to their efficacy against pests and diseases in crop protection, thus improving productivity because no other pest management strategies have been as effective as pesticides. 2 The Food and Agriculture Organization (FAO) data suggest that Asia ranks first in pesticide import (2.04 billion kg), followed by South America (0.84 billion kg), North America (0.73 billion kg), Africa (0.72 billion kg), and Oceania (0.21 billion kg). These data confirm the significance of pesticides to modern-day agriculture; however, the application of these chemicals comes with negative externalities to human health and the environment. 3 There has been a significant increase in pesticide consumption in Nigeria as about 0.0114-0.147 billion kg (11,,000 metric tons) of active ingredients were imported annually from 2008 to 2018. 4 In Nigeria, cocoa remains an important cash crop with an annual production of about 350,146 metric tons on 1.35 million hectares of land. 5 Cocoa contributes roughly 12.5-14% to foreign exchange earnings. 6 There has been about a 30% to 40% decline in cocoa production over ten years (from 2003 to 2013) due to the severity of insect pests (mirids) and fungal diseases (black pod). 7 Pesticide use is necessary to reduce loss and increase production simultaneously. 8 Pesticides used in cocoa production accounted for roughly 31% of the total imported pesticides to Nigeria, 9 implying that around 44,100 metric tons of pesticides are used in cocoa production annually. Some studies show these practices are environmentally unsafe. [10][11][12] Intensive pesticide use to increase agricultural production has resulted in different acute or chronic health problems, such as skin injuries, dizziness, headache, fatigue, and genetic and nervous disorders in agrarian societies. 13 Of 740,000 annual cases of pesticide poisoning in agriculture, there were 7,446 fatalities and 733,921 non-fatal cases recorded globally. 14 There is a link between pesticide use and environmental and societal damage, costing an estimated $9.6 billion, especially in the USA. 15 The FAO reported work-related pesticide symptoms among roughly 23% Indian, 25% Mexican, and 43% Zimbabwean farmers. 4 According to Sowunmi 11 and Sosan 16 pesticide residues in blood samples, farm soils, and cocoa beans in cocoa-producing belts are higher than the permissible limit due to pesticide misuse in Nigeria.
Farmers cannot be blamed entirely for pesticide misuse as some of the determinants are not within their control. According to the Hierarchy of Controls (HoC) model, five pesticide risk mitigation steps exist. 17 These include the physical removal of hazards via organic farming (pesticide elimination), substituting safe pesticides for harmful ones (substitution), and isolating farmers from pesticide exposure (engineering controls). The model also includes administrative controls on using or importing banned pesticides and safety behaviors. Though it is the least preferred risk mitigation step in the HoC model, pesticide safety behavior among Nigerian cocoa farmers needs to be better understood. 18 Individual beliefs or perceptions influence farmers' behavior when handling pesticides. 19,20 Therefore, it is vital to assess determinants of unsafe behaviors in pesticide use as they are one of the prerequisites for effective risk management.
Research on pesticide safety behavior has shown that personal protective equipment (PPE) such as face masks, goggles, gloves, protective clothing, safety boots, and a respirator is an effective risk-mitigation strategy for preventing pesticide-related health problems. 21,22 Several studies, however, found that, despite having good awareness about the adverse impacts of pesticides, [23][24][25] most farmers disregard the use of safety measures that can reduce the risk of poisoning. 22,26 Sociodemographic factors (such as age, farm experience, training, and education status) influence behaviors in pesticide use; however, safety determinants are complex due to the cognitive nature of human behaviors. 19,27 The inability to explain these cognitive factors might result in an insufficient understanding of farmers' behaviors in safe agricultural practices. 28 Theory-based studies are much better at explaining behavioral factors as they provide a theoretical framework for future intervention designs. 29 The Health Belief Model (HBM) offers a theoretical framework for exploring safe pesticide use behaviors among farmers. 30,31 The HBM, a psychosocial model, explains why people do or do not practice preventive health behaviors ( Figure 1). Studies have used the HBM to assess safe pesticide use among farmers in developing countries. [32][33][34] The operationalization of these studies' HBM and safety behavior constructs were inconsistent and incomplete regarding all aspects of pesticide safety. These studies define pesticide safety behaviors and their determinants as PPE use and sociodemographic factors, respectively. Safety behaviors in pesticide use and their influencing factors among farmers are beyond PPE use and sociodemographic factors. 35 Safety behaviors in pesticide use are practices that help reduce human and environmental health risks without undermining crop protection. Such practices include PPE, hygiene, appropriate pesticide use, and health risk avoidance.
Prior studies have not applied the HBM to model safe pesticide use in Nigeria. This study aimed to assess safety behaviors in pesticide use and their cognitive determinants among cocoa farmers in Nigeria. This study surveyed the major pesticides used on cocoa farms as Nigerian cocoa farmers mostly used a cocktail of these pesticides. 36 Understanding safety behavior among cocoa farmers and the factors influencing it is necessary for promoting pro-environmental behavior in pesticide use. The study outcome could also benefit agricultural authorities while providing information for formulating a pesticide safety policy for the country. This study explored a critical question: to what extent can the HBM and its constructs explain pesticide use and safety behavior among cocoa farmers in Nigeria?

Methodology
Overview: We collected quantitative data from cocoa farmers in Nigeria during a cross-sectional study to assess their pesticide safety behaviors and their determinants. Study area: We grouped about 300,000 cocoa farmers across 14 cocoa-producing states in Nigeria into low, medium, and high-production states. 37 The high producer states are Ondo, Osun, and Cross River; the medium producer states are Ekiti, Akwa-Ibom, Edo, Oyo, Ogun, Delta, and Abia; and the low producer states are Kogi, Kwara, Taraba, and Adamawa ( Figure 2). 38 We carried out the study in Ondo (high), Ogun (medium) and Kwara (low) states. Participant selection and recruitment: We selected 403 participants (after adjusting for about 5% attrition) using a sample size formula of Krejcie and Morgan. 39 Due to unequal sizes of cocoa farmers across the three states, we proportionately allocated the sample. A multistage sampling technique was applied to select 200, 133, and 67 cocoa farmers from Ondo, Ogun, and Kwara states. First, we chose the top three cocoa-producing states based on their production capacity in each region. Second, we selected three Local Government Areas (LGAs) in each state and based the selection on the top three LGAs to the total hectares of land area under cocoa production. 40 Third, we randomly selected three villages in each of the nine LGAs from a list obtained from the Cocoa Research Institute of Nigeria (CRIN). With the assistance of the village head/farmers' association, we collected the list of cocoa farmers from the selected villages meeting our inclusion/exclusion criteria as there was no official database for cocoa farmers in Nigeria. The final stage involved randomly selecting the farmers from the compiled lists collated across the 27 villages in Ondo, Ogun and Kwara states.

Inclusion criteria
a. Both males and females 18 years old or older during data collection b. Both males and females able to communicate in a local language (Yoruba) c. Both males and females residing in the communities where their farms located d. Both males and females who had just one cocoa farmland (to ensure that the farmers used the recommended quantity (or not) for one farmland and not for their other lands to allow individual participants level measurement of pesticide used)

Exclusion criteria
a. Both males and females who had a speech impairment b. Both males and females who planted other crops such as vegetables (to allow proper estimation of pesticide use on cocoa farms only) c. Both males and females who had other jobs (in order to collect unbiased data from fulltime farmers only) d. Both males and females who were not available for face-to-face interviews Questionnaire Development: Figure 3 shows a schematic diagram explaining the development of the survey instrument. We developed a pesticide safety behavior questionnaire based on an in- depth review of previous literature 35, 41 following the FAO guidelines. 42 First, we identified and gave operational definitions of the variables required from the literature based on the study objectives. Second, we developed a draft based on the variables identified using close-ended questions targeting the cocoa farmers. We modified the items according to the local conditions of cocoa farming in Nigeria. Third, four experts assessed the instrument content validity using the item-objective congruence (IOC) method. 43 We modified and corrected the questionnaire draft based on the experts' comments and IOC scores for each variable. Finally, we pretested the final draft on a small sample of cocoa farmers (n = 40) to eliminate potential problems and check its reliability. Table 1 shows the Cronbach's alpha coefficients from the pretest, and all the variables tested were above the recommended threshold value of 0.7.   such as the farmers perceived susceptibility to pesticide risks, the severity of pesticide illness, and barriers to engaging in safety behavior. We also asked questions regarding the benefits of safety behavior, self-efficacy in engaging in safety behavior, and cue to action that facilitates safety behavior. We scored the HBM questions on Likert scales ranging from 1 to 5 (1 = strongly disagree, 5 = strongly agree). Table 3 details information about the HBM measures. Data collection: Data collection for this study occurred from March to June 2020. Due to the researcher's familiarity with the study location, we recruited a research agency staff as a local guide. The recruitment was necessary to gain trust and build rapport among the potential participants. Due to the farmers' low literacy level, we verbally administered the questionnaires in a local language (Yoruba) prevalent in the study locations. The first author and four research assistants collected the data. The assistants were master's degree students studying public/environmental health from a local university and received two-day training from the first author. We asked the farmers to show materials and equipment or demonstrate their applications where necessary to minimize response bias. We collected the data during the spraying season to get accurate or unbiased information. Three hundred and ninety-one (391) cocoa farmers completed the questionnaire yielding a 97.7% response rate. The data collection took place at their cocoa farms or farmhouses (depending on their preference) and lasted from 65 to 90 mins.

Statistical analysis
We analyzed the data using licensed SPSS software (version 23) and an Excel software package. Kolmogorov-Smirnov statistics were used to test for the data normality. The pesticide safety behavior data showed insignificant departure (p > 0.05) from the normality assumption, while pesticide overuse data did not show normal distribution. We used descriptive statistics to describe the responses in the IV. Spearman's correlation was used to assess the relationship between pesticide overuse and HBM measures. Levels of pesticide safety behaviors were also analyzed using descriptive statistics. We used multiple linear regression analysis to evaluate the relationship between the factors influencing pesticide safety behavior among the participants. We also adjusted the model for age, experience, safety training, and education status.

Ethical consideration
The research was approved by the Ethics Review Committee for Research Involving Human Research Participants, Health Sciences Group, Chulalongkorn University, Thailand (COA No. 028/2020). Moreover, we obtained informed consent from all the participants before the interview concerning the study objectives, procedures, risks, and benefits. We also explained the study's significance and data collection methods to the potential participants who met our selection criteria.     Table 6 summarises the Spearman correlation coefficients between pesticide overuse and its determinants. The explanatory variables' correlation coefficients are within the acceptable range. Education status (r = −0.484), pesticide training (r = 0.150), perceived susceptibility (r = −0.578), and cue to action (r = −0.343) all significantly correlated with pesticides overuse.

Pesticide safety behavior among the farmers
In all, 42.7% of the farmers showed unsafe behaviors, 49.1% showed intermediate behaviors, and just 8.2% reported safe behaviors in pesticide use, as shown below (Figure 4). Most cocoa farmers had unsafe behavior in PPE use (69.6%) and hygiene practices (45.8%), respectively. Meanwhile, only 5.4% and 17.9% of the farmers showed safe behavior in PPE use and hygiene practices, respectively. For PPE use, the farmers always practice safe behavior in only three of the eleven items on the list. The safety behaviors are wearing long trousers that cover the boots, shirts with long sleeves and rubber boots. Regarding hygiene practices, most farmers did not change work clothes immediately, get first aid when poisoned, wash work clothes used in spraying

The HBM constructs of cocoa farmers regarding the safe use of pesticides
From Table 7 Table 8 shows that perceived susceptibility, barriers, benefits and cues to action significantly predicted the farmers' pesticide safety behaviour (R 2 = 0.593, p < 0.05) after controlling for age, farm experience and education status. The values of standardized regression coefficients showed that

Discussion
In the HoC model, pesticide use behavior is the risk mitigation measure that is least preferred, but it still needs to be better understood in the context of cocoa farmers. We designed the current study to provide insights on safe pesticide use and influencing factors among cocoa farmers using the HBM for the first time in Nigeria. Using the HBM to identify critical points of consideration provides valuable information for a targeted extension that will improve farmers' pesticide use and safety behavior and minimize possible contamination from unsafe practices. Cocoa farmers in this study used insecticides, herbicides, and fungicides on their farms. This finding conforms to the results from other studies. 36,48 These pesticides were not currently approved for cocoa production and were still overused. Data analysis showed a negative correlation between education status and the use of inappropriate pesticides in the inappropriate amounts. This finding implies that educated cocoa farmers are less likely to overuse pesticides. Well-educated farmers (e.g., some college education) preferred motivations for "health maintenance" more than farmers with lower education or no education at all. 49 Pesticide training correlated positively with pesticide overuse. The fact that farmers are only formally authorized to participate in training programs and that these programs have not produced the desired results could be the cause. 50 There was also a negative correlation between perceived susceptibility and pesticide overuse. Previous research showed that farmers who recognized pesticides as harmful were less likely to use inappropriate pesticides in the inappropriate amounts. 50 There was also a negative correlation between cue to action and pesticide overuse. Cocoa farmers with a higher cue to action (e.g., pesticide information from the media, retailers or government agencies) were less likely to overuse pesticides. 51 The farmers' safe practices to mitigate pesticide exposure risks were low (8.2%). Safety behaviors concerning PPE use (5.4%), good hygiene practices (17.2%), appropriate pesticide use (23.3%), and avoidance of health risks (31.5%) were also low among the study participants. Studies have shown that PPE such as safety goggles, face shields, waterproof aprons, coveralls, respirators and masks were uncommon among farmers, especially in developing countries. 22,23 Most farmers did not change their clothes immediately, get first aid when poisoned, wash the clothes used in spraying separately or wash their face before eating. 52,53 The majority of the farmers did not buy the required amount of pesticides, consider toxicity signs on the containers, use pesticides approved by the authorities, and adhere to recommended dose. 54 Our findings also showed that the respondents did not regularly stop working when poisoned on cocoa farms and mixed pesticides with bare hands, which agrees with Damalas et al. 55 Data analysis from 391 cocoa farmers showed that perceived barriers, susceptibility, self-efficacy, and cue to action after adjusting for farmers' age, education status, and pesticide use experience were factors affecting pesticide safety behavior. Farmers with higher education status are more likely to adopt safe behavior in pesticide use. Research showed that well-educated farmers were more likely to pay for less risky pesticides. 56 Cocoa farmers with higher education levels, such as tertiary education, were more likely to adopt safety measures for pesticide use. This finding was in line with the study by Okoffo 57 among cocoa farmers in Ghana. The study results implied that the likelihood of practising pesticide safety behavior decreases as the farmers' age. This result showed that older cocoa farmers perceived safety equipment as unnecessary. This finding conformed to the study of Boadi-Kusi. 58 In a survey among cocoa farmers in Ghana, Okoffo 57 found a negative relationship between PPE use (a component of safety behavior) and farm experience. This study showed that older cocoa farmers being more experienced, preferred traditional methods of crop maintenance on farms; hence, they applied pesticides in a risky manner. Previous research showed that old farmers did not use personal protective equipment because protective measures are unnecessary after years of farming. 59 Linear regression analysis showed that perceived susceptibility, self-efficacy, and cue to action had a positive influence, while perceived barriers negatively affected cocoa farmers' behavior in pesticide safety. The HBM successfully predicted pesticide safety behavior among the studied cocoa farmers (R 2 = 0.593, p < 0.05). This result implied that perceived susceptibility, selfefficacy, cue to action, and barriers accounted for 59.3% of the total variance in the farmers' pesticide safety behavior. The results were congruent with the study of Kien 46 and Moradhaseli. 60 Kien 46 reported that perceived susceptibility and severity explained 53% of the variance in their participants' pesticide safety behavior. Moradhaseli 60 revealed that perceived susceptibility, cue to action, and self-efficacy explained 54.9% of the total variance in their participants' occupational health behavior.
The perceived barrier was the most negative predictor of cocoa farmers' pesticide safety behavior. This finding implied that the more the barriers, the less the likelihood of engagement in safety practices among the farmers and vice versa. Perceived barriers such as the cost of safety measures (such as PPE), lack of time and discomfort from PPE can influence farmers' safety decisions, 35,61,62 especially among the cocoa farmers. Perceived susceptibility also significantly influenced the participants' safe pesticide use. According to Sharifzadeh, 35 farmers with a highrisk perception of pesticide hazards to their health were more likely to engage in safety behaviors as they were apprehensive of the adverse outcome of pesticide exposure. Cocoa farmers who perceive pesticides as hazardous to their health and the environment might fear their harmful impacts, thus adopting pesticide safety measures. According to the HBM, if people believe they are vulnerable to health problems, they would take preventive actions to forestall the undesirable occurrence. Self-efficacy was a positive predictor of pesticide safety behavior. This finding confirmed its significance in limiting the farmers' unsafe behaviors in pesticide use. So, cocoa farmers with high self-efficacy were likelier to utilize and implement safety measures when using pesticides. According to Sharifzadeh, 63 the perceived usefulness of PPE predicted farmers' willingness to use it. Increased motivation and improved cognition, as well as problem-solving capabilities, are elicited by high self-efficacy. 30 Self-efficacy is fundamental to comprehending the complexity of human behavior, which means that one would only practice easier tasks within one's control. 64 Farmers may not adopt sustainable agricultural practices due to low self-efficacy. 65 Lastly, cue to action was another significant positive predictor reported by other studies. 51,66 It enhances the readiness of an individual to perform specific tasks, thereby serving as the motivator to trigger preventive health actions. 67 The findings showed that cocoa farmers exposed to these triggers are more likely to engage in safe behaviors during pesticide use. The HBM suggests that as long as perceived susceptibility or severity is high and perceived barriers are less than benefits, cues to action can trigger or motivate an individual to act. Cues from retailers and government authorities, via suggestions, guidelines, and media adverts, could be essential to improving behavior in pesticide use. 51

Limitations
This study provided crucial information about the determinants of pesticide safety behavior among cocoa farmers in Nigeria; nevertheless, it has some limitations that require additional research. First, this study paid attention to cocoa farmers across major cocoa belts in Nigeria and, thus, should not be generalized to other farmers in the country. Future research should consider farmers cultivating crops other than cocoa as there could be variations in the quantity of pesticide use which may impact their safety behaviors based on the crops being grown. Second, variables of the HBM explained 59.3% of the total variance in pesticide safety of cocoa farmers, implying that more variables may influence their behavior in safe pesticide use. Theoretical frameworks such as the HoC Model, Norm Activation Model (NAM), Health Locus of Control (HLC), and Integrated Action Framework (IAC) could either be applied independently or integrated with the HBM for more explanatory ability. The study findings can be generalized among cocoa farmers in Nigeria but should be interpreted with caution in other locations and populations.

Conclusion
The current study is one of the first attempts in the Nigerian context to assess cocoa farmers' pesticide use and safety behavior and identify their psychological determinants. Pesticide use among the farmers significantly correlated with education status, pesticide training, perceived susceptibility and cue to action. The farmers' pesticide safety behavior was low regarding PPE use, hygiene practices, appropriate use of pesticides, and avoidance of health risks. This result implied that the farmers in this study were highly exposed to pesticides. The most significant predictor of cocoa farmers' safety behavior was perceived barriers to adopting safety measures in pesticide use. The government authority should address perceived barriers such as the high cost of PPE and PPE design as well as pesticide training effectiveness to improve safety behavior in pesticide use. Farmers should receive unformalized and targeted knowledge and skills training in pesticide use. Also, using the HBM to identify critical points of consideration provides valuable information for a targeted extension that will improve farmers' pesticide safety behavior.