Assessment of pesticide exposure risks among cocoa farmers in Western region of Ghana

Abstract

A survey was conducted to investigate farmers’ knowledge, attitudes towards pesticide use, storage/disposal, exposure risks and health symptoms in one of the eight cocoa growing regions in Ghana. A considerable proportion of the farmers (32%) used the bush as a storage facility for pesticides, 17% of the farmers stored chemicals in their living rooms, 3% of the farmers stored chemicals in their kitchen, 15% in their food storeroom, and 4% in the animal house. Personal protective equipment (PPE) use was positively associated with advice obtained from agrochemical shops (OR = 1.735, p < 0.01) and extension services (OR = 1.643, p < 0.01) as sources of information for PPE use. Female farmers (OR = 0.481, p < 0.01) were less likely to use PPE. With respect to location, farmers in Suaman district were less likely to use PPE (OR = 0.56, p < 0.010) compared with farmers in Wassa Amenfi. It is recommended that these factors should be considered for policy intervention. Reinforcement of appropriate pesticide storage and PPE education are necessary for securing safety in pesticide use.

Keywords:

• Smallholder
• cocoa
• pesticides
• risk
• exposure
• PPE

1. Introduction

Cocoa is an international crop supplied by Ghana and Ivory Coast providing over 60% of the cocoa beans to the global chocolate industry (Wessel and Quist-Wessel 2015). The cocoa sector represents more than half (from 70 up to 100%) of the income for roughly 800,000 smallholder farmers families in Ghana, providing food, employment, tax revenue and foreign exchange earnings for Ghana (Anim-Kwapong and Frimpong 2004; Dormon et al. 2004). Despite the economic importance of cocoa, its production in Ghana is threatened by pests and diseases, a situation that reduced cocoa production, with adverse impact on the Ghanaian economy (Dormon et al. 2004). The low productivity phenomenon is aggravated by the climate change impacts. Läderach et al. (2013) predicted low productivity in spatially differentiated cocoa growing areas of Ghana and Ivory Coast. These predicted potential effects are already observed in Ghana, where higher productive region for cocoa shifted from Ashanti Region to Western Region (Ghana COCOBOD 2019).

In order to boost cocoa productivity, farmers adopted pesticide use to control pests and diseases, thus increasing yield and maintaining quality. However, the use of pesticides in agriculture, and for that matter the cocoa industry in Ghana, has raised serious concerns about the safety of pesticide residues in cocoa beans, soils and water, as well as factors causing potential exposure to humans (Adeogun and Agbongiarhuoyi 2006; Adejumo et al. 2014). In most developing countries like Ghana, these consequences have often been severe because farmers do not use approved pesticides and do not follow recommended application practices as stipulated by governmental agencies. Indeed, farmers often misuse, apply pesticides indiscriminately (Konradsen 2007; Damalas 2009; Hashemi et al. 2012; Antwi-Agyakwa et al. 2016) with disregard to safety measures and regulations on chemical use. A major global public health hazard about pesticides is causing death (Bertolote et al. 2006). In developing countries, many people die annually from pesticide effects through pesticide mishandling (Konradsen et al. 2003). An estimated number of 220,000 pesticide related deaths and 3 million poisoning cases were reported by WHO in 1990 (Jeyaratnam 1990; Khan and Damalas 2015). Based on the period it takes for toxicity symptoms to manifest, pesticide health effect may be classified as being acute or chronic. A situation where symptoms are observed within a short period of pesticide exposure is termed acute toxicity, while long-term symptoms are reported as chronic toxicity (Damalas and Koutroubas 2016). Whatever the case, occupational health is a topical issue in developing countries and remains unaddressed (Nuwayhid 2004; Khan and Damalas 2015).

The Western Region (WR) produces about 450,000 metric tonnes (avg. of the last 10 years)of cocoa and is currently the leading producing area in Ghana (Ghana COCOBOD 2019). In general, farmers use pesticides extensively to control pests and diseases and maximize crop yields. Recently, Okoffo et al. (2016), and Paintsil (2017) studied pesticide application among cocoa farmers, but their studies were limited to one cocoa district within a region. In this study, we broadened the scope by focusing on three different cocoa districts because information regarding pesticide documentation and safety practices by cocoa farmers in the WR is limited. Unsafe pesticide practices can lead to predictable health impacts on farmers during pesticide application. This information is necessary for understanding the factors influencing farmers’ behaviour, pesticides exposure levels of farmers and eventually, the mobility of pesticides in the environment. Information on pesticides application is important, so that policy interventions to reduce environmental risks and human health impacts can be developed. Also, such information is important for analytical and environmental scientists to gain insights into the socio-environmental factors driving pesticides in the environment. The objectives of this work were: i) to assess farmers’ knowledge of pesticides use; ii) to evaluate farmers’ attitudes in storage of pesticides and disposal practice after pesticides usage; and iii) to identify health risk from pesticide exposure.

2. Methods

2.1. Study area and sampling procedure

The study was conducted in the WR of Ghana with an approximate land cover of 23,921 km², constituting about 10% of Ghana’s total surface land mass and 10% of its population. The WR is the leading producing area of cocoa since 1984 (Ghana COCOBOD 2019). The region receives the highest amount of precipitation nationwide and almost 75% of its vegetation interspersed with the high forest zone of Ghana (Figure 1).

Figure 1. A map showing the study area.

For the purposes of assessing pesticides use among farmers, data were collected through a questionnaire in February 2018. The questionnaire covered demographic characteristics of the farmers, pesticide use practices, attitudes towards pesticide use, wearing of personal protective equipment (PPE) by cocoa farmers, and self-reported pesticide health symptoms of farmers during pesticide applications. The questionnaire was designed in English and then translated into the local language of the area in case that some farmers were uncertain of some technical terms. Farmers with prior knowledge in pesticide use application in the cocoa-growing communities were sampled within the districts.

Multi-stage sampling was used to select respondents for the study (Daniel 2012; Okoffo et al. 2016). One main advantage of multi-stage sampling is that it creates a more representative sample of the population than a single sampling technique and can reduce costs of large-scale survey research (Green et al. 2006). The multi-stage sampling in this study entailed four stages. In the first stage, the WR of Ghana was purposively selected due to the high production of cocoa in the region. In the second stage, Aowin, Suaman and Wassa Amenfi West districts known to be some of the major cocoa growing areas in the WR were randomly selected out of other cocoa-producing districts in the region. In the third stage, three major cocoa growing communities were randomly selected. In the final stage, 25 cocoa farmers were randomly selected from each of the three selected cocoa growing communities. Totally, 225 cocoa farmers were randomly sampled for the study. That is, three districts × three communities × 25 farmers = 225 farmers. In this study, the participants were informed that the data provided would contribute to the overall knowledge about the effects of pesticides on human health. In addition, participants were neither coerced nor financially induced to take part in the research.

2.2. Data analyses

Analysis was conducted using Statistical Package for the Social Sciences (SPSS) Version 21 (IBM, Chicago, IL, USA), STATA 13 (Stata Corp, College Station, TX, USA), and Microsoft Office Excel 2010 (Microsoft Corporation, Redmond, WA, USA). Descriptive statistics (frequencies and percentages), inferential statistics, analysis of variance (ANOVA), and Pearson correlation/chi-square tests were conducted on the data from the respondents to examine significant differences among the identified categorical groups. An alpha (α) level of 0.05 was used as the criterion for statistical significance. The relationship between response variable and explanatory variables was modelled using logistic regression.

Logistic regression has three common link functions, that is, the logit, the probit, and the complementary log-log. The logit and the probit are symmetric link functions, while the complementary log-log is applicable when the data come from multiple groups and are not symmetric in the 0–1 interval and increase slowly at small to moderate values, but increase sharply to 1. This response implies that we must not have the response curve of the participants showing 50% in the affirmative and the other 50% in the negative (Collett 2003; Armah et al. 2019). In this study, a complementary log-log regression model was fitted to binary outcomes data at the multivariate level. The complementary log-log transformation is expressed as: (1) $$\text{log}\hspace{0.17em}\left\{-\hspace{0.17em}\text{log}\hspace{0.17em}\left(1-\mathrm{p}\right)\right\}{=\mathrm{\beta }}_{\mathrm{o}}{+\mathrm{\beta }}_1{\mathrm{X}}_1{+\mathrm{\beta }}_2{\mathrm{X}}_2+\cdots \cdots {+\mathrm{\beta }}_{\mathrm{k}}{\mathrm{X}}_{\mathrm{k}}=\mathrm{Z}.$$(1) where, β_o = the constant of the equation, β_1,2,3 = the coefficient of the explanatory variables 1, 2, 3 to be estimated; X₁……. X_k are sets of explanatory variables; p is the predicted probabilities; and log{-log(1-p)} is the link function. (2) $$-\hspace{0.17em}\text{log}\hspace{0.17em}\mathrm{ }\left(1-\mathrm{p}\right)\mathrm{ }=\mathrm{ }\mathrm{e}\left({\mathrm{\beta }}_0{+\mathrm{\beta }}_1{\mathrm{X}}_1{+\mathrm{\beta }}_2{\mathrm{X}}_2+\cdots \cdots {+\mathrm{\beta }}_{\mathrm{k}}{\mathrm{X}}_{\mathrm{k}}\right).$$(2)

As the probability increases, the transformation approaches infinity more slowly than either the probit or the logit. By using equations (1)(1) $$\text{log}\hspace{0.17em}\left\{-\hspace{0.17em}\text{log}\hspace{0.17em}\left(1-\mathrm{p}\right)\right\}{=\mathrm{\beta }}_{\mathrm{o}}{+\mathrm{\beta }}_1{\mathrm{X}}_1{+\mathrm{\beta }}_2{\mathrm{X}}_2+\cdots \cdots {+\mathrm{\beta }}_{\mathrm{k}}{\mathrm{X}}_{\mathrm{k}}=\mathrm{Z}.$$(1) and (2)(2) $$-\hspace{0.17em}\text{log}\hspace{0.17em}\mathrm{ }\left(1-\mathrm{p}\right)\mathrm{ }=\mathrm{ }\mathrm{e}\left({\mathrm{\beta }}_0{+\mathrm{\beta }}_1{\mathrm{X}}_1{+\mathrm{\beta }}_2{\mathrm{X}}_2+\cdots \cdots {+\mathrm{\beta }}_{\mathrm{k}}{\mathrm{X}}_{\mathrm{k}}\right).$$(2) , the relationship between the response variable and the explanatory or predictor variable was modelled. The response variable is the use of PPE and the explanatory or predictor variables are the agrochemical shops, extension services availability, years of farmer’s experience, age, education and the districts. The role of respondents using PPE or not in determining factors that influence famers’ choice to use PPE during pesticide application was estimated using a complementary log-log model and reported as exponentiated coefficients or odds ratios (OR). An OR value of 1 means that the predictor does not affect odds of influencing farmers to use PPE during pesticide application, while OR > 1 means that the predictor is associated with higher odds of influencing farmers to use PPE during pesticide application. Finally, OR < 1 means that the predictor is associated with lower odds influencing farmers to use PPE during pesticide application. The study accounted for clustering of observations in units of communities and robust estimates of variance was used to correct for any statistical outliers in the estimation of standard errors. The study made use of 95% confidence interval (CI) and the level of statistical significance was set at 0.05. The main or key predictors used were agrochemical shop services, extension services and years of farmers’ experience. Some compositional factors [(biosocial variables (age), socio-cultural variables (education)] and contextual factor (the districts where the study was conducted), which were known in the literature to influence farmer’s choice to use PPE during pesticide application were controlled for in the models. Four models: main predictors, agrochemical shops/extension services availability (model 1), biosocial (model 2), socio-cultural (model 3), and contextual factors (model 4) were run. Selection of reference groups for the independent variables in the models was based on theory and literature. “I do not get information about pesticide use from agrochemical shop” (No) was chosen as the reference group for the agrochemical shop services. The reference group for extension services was “I do not get information about pesticide application from extension services” (No). For farmers’ years of experience, “I do not get information about pesticide use from farming years of experience” (No) was used as the reference group. Similarly, no formal education group was chosen as the reference group for education since this has direct influence on farmers’ knowledge of dangers associated with pesticide use. The rest of the reference groups were “Male” for gender, and “Wassa Amenfi West” for district.

In addition, to validate the sample distribution, estimate the accuracy of a given parameter, and strengthen the stability of the statistical model, the bootstrap technique was used to provide support. The method takes a sample with replacement from the original sample and calculates the statistic of interest repeatedly (Islam and Begum 2018). In this study, the proxy sample population was 225 and large sample size of 2000 was estimated in bootstrapping.

2.3. Ethical statement

The study was approved by the Institute Review Board (IRB) of the University of Cape Coast. Agricultural extension officers of COCOBOD offered permissible access to cocoa farmers. Consent to collect and publish data was obtained from the participants. In addition, participants were not coerced through financial means to take part in this research. They were informed that the outcomes of the research would enhance their welfare in terms of pesticide usage.

3. Results

3.1. Farmers knowledge and understanding on pesticides

3.1.1. Socio-demographic description of the respondents

Table 1 presents the demographic background of the respondents from the selected districts.

Table 1. Socio-demographic characteristics of the study population.

	N (%)	N(%)
Respondent characteristics	Male	Female
Sex	154 (68.44)	71 (31.56)
District
Wassa Amenfi West	57 (76)	18 (24)
Aowin	44 (58.67)	31 (41.33)
Suaman	53 (70.67)	22 (29.33)
Age
10 − 20	1 (50)	1 (50)
21 − 30	15 (75)	5 (25)
31 − 40	41 (62.12)	25 (37.88)
41 − 50	35 (68.63)	16 (31.37)
51 − 60	44 (74.50)	15 (25.42)
61 and above	18 (66.67)	9 (33.33)
Level of Education
No formal	47 (63.50)	27 (36.50)
Primary	54 (68.40)	25 (31.60)
Secondary	53 (73.60)	19 (26.40)
Economic Activities
Farming	147 (95.45)	7 (4.55)
Small business	67 (94.37)	4 (5.63)
	Mean	Std. Err.
Farm size (ha)	9.025	0.512

N = number of respondents in each category add up to 225; numbers in parenthesis indicate percentages; (ha) = hectares. Std. Err. = Standard error of the mean.

The demographic data of the respondents include sex, age, educational level, family position, and economic activity. Out of 225 farmers, 154 males and 71 females were contacted in the study showing that the males dominated in cocoa farming. The majority of the study participants were males (68%), while the remaining (32%) were females. One third of the respondents (33%) had no formal education, 35% had only primary education and (32%) secondary education. The main economic activity for the sampling group was farming (95%). Most farmers surveyed were between the ages of 31 and 40 years old; however, farmers were within the economically active age range (18–65) (Table 1). Only 9% of the farmers were between 21 and 30 years old. It was observed that the farmers used different types of pesticides. In all, eleven types of pesticides were identified, and the most commonly used were the insecticides followed by the fungicides. About 46% of the pesticides were moderately/slightly hazardous according to WHO classification. In addition, about 8% of pesticide application fell into the non-toxic class (Table 2).

Table 2. Commonly used pesticides by the farmers.

Trade name	Active ingredient	WHO toxicity class
Insecticides
Akate master	Bifenthrin (27 g/l)	II
Acati Power SL	Thiamethroxam (200 g/l)	II
Akate Star 3.5 EC	Bifenthrin (30 g/l)	II
Aceta Star 46 EC	Bifenthrin (30 g/l)	II
Confidor 200 OD	Thiamethroxam (240 g/l)	III
Miricon EC	Pyrethrin (12 g/l) + Deltamethrum (6 g/l)	II
Fungicides
Champion WP	Copper Hydroxide (77%)	III
Ridomil Gold Plus 66WP	Cuprous oxide (60%) + Metaxyl − M (6%)	III
Nordox Super 75 wg	Cuprous oxide (86%)	III
Herbicides
Aduodzi 757 SG	Glyphosate (757 g/l)	III
Tackle360 SL	Glyphosate (360 g/l)	IV

I = extremely hazardous; II = moderately hazardous; III = slightly hazardous IV = non-toxic; (WHO 2005).

3.1.2. Farmers’ knowledge of pesticide toxicity

An analysis of farmers’ knowledge on the routes of entry of pesticides into the human body, on fruits such as cocoa and vegetables, and in the environment brought out the results shown in Table 3. There was a statistically significant difference between educational level of farmers and their knowledge of pesticide entering their bodies through inhalation (χ² = 10.28, p < 0.05), through the skin (χ² = 7.59, p < 0.05), and knowing whether pesticide residue is left on fruits and vegetables after the application of pesticides (χ² = 10.054, p < 0.05). However, the Chi-square (χ²) of variables that were not significant (i.e. whether pesticides can cause negative effects, or pesticides residue can be left on air, soil, etc.) were positively associated with greater percentage of the farmers saying “yes” to the questions, indicating that knowledge is relevant and influential. In general, most farmers had good knowledge of the effects of pesticides on human health by explaining how their bodies react after the spraying of pesticides.

Table 3. Relationship between farmers’ level of education and knowledge on pesticide toxicology.

Question asked	No formal education	Primary education	Secondary education	Inferential statistics
Can pesticides cause negative effects?
No	17 (43.60)	13 (33.30)	9 (23.10)	X^2 = 2.8589, p = 0.239
Yes	57 (30.60)	66 (35.50)	63 (33.90)
Do all pesticides have the same health effects?
No	53 (33.30)	55 (34.60)	51 (32.10)	X^2 = 0.0753, p = 0.963
Yes	21 (31.80)	24 (36.40)	21 (31.80)
Can pesticides be dangerous to use?
No	10 (45.45)	8 (36.40)	4 (18.10)	X^2 = 2.6365, p = 0.268
Yes	64 (31.50)	71 (35.00)	68 (33.50)
Can pesticides enter the body through inhalation?
No	9 (75.00)	2 (16.70)	1 (8.30)	X^2 = 10.2818, p = 0.006*
Yes	65 (30.50)	77 (36.20)	71 (33.30)
Can pesticides enter the body through skin?
No	9 (75.00)	2 (16.70)	1 (8.30)	X^2 = 7.5390, p = 0.023*
Yes	65 (30.50)	77 (36.20)	71 (33.30)
Can pesticide residue be left in the air?
No	6 (46.20)	5 (38.50)	2 (15.30)	X^2 = 1.9726, p = 0.373
Yes	68 (32.10)	74 (34.90)	70 (33.00)
Can pesticide residue be left in the soil?
No	8 (26.70)	14 (46.70)	8 (26.70)	X^2 = 2.0316, p = 0.362
Yes	66 (33.80)	65 (33.40)	64 (32.80)
Can pesticide residue be left in the fruit and vegetables?
No	20(55.60)	9 (25.00)	7 (19.4)	X^2 = 10.0541, p = 0.007*
Yes	54 (28.60)	70 (37.00)	65 (34.40)

N = number of respondents for each category is 225; Numbers in parenthesis indicate percentages; X ²= Pearson chi square; p = probability value; * = Significant result (p < 0.05)

3.1.3. Pesticide acquisition, reason for application, and knowledge of application

Table 4 shows pesticide acquisition, reason for application and knowledge of application using descriptive statistics. Farmers indicated agrochemical shops (27%), local governmental shops (41%), and extension officers (38%) as their main sources of purchasing pesticides. Part of the data (not shown) revealed the following brand names: Akate master, Confidor, Ridomil, and Nodox as the commonly used pesticides. All farmers consented to using motorized sprayers for insecticide application, while the knapsack sprayer was the preferred equipment for fungicide application. When farmers were asked why they use pesticides, 80% of the farmers identified the presence of pests as the driving factor for their decision to apply chemicals. When the respondents were asked where they buy the pesticides from, there was a plethora of sources and some of the sources were not regulated. Less than half of the respondents (41%) were buying pesticides from local governmental shops in villages, while the remaining were distributed among agrochemical shops in towns (27%), and other general shops (7%) while (38%) of the farmers obtained them from extension officers. Regarding timing of pesticides application, 30% followed the recommended calendar spraying schedules, no matter the observations in the field. Application strategies employed by the majority of the farmers involved the application of different chemicals individually (90%), but the remaining group (10%) indulged in the improper farming practice of mixing different chemicals to have rapid knockdown effects of pests. A greater part (88%) of the farmers did not read instruction on labels before using pesticides.

Table 4. Pesticide application information by cocoa farmers.

	Response N (%)	Response N (%)
Questions and predefined answers	Yes	No
Why do you use pesticides?
To protect crops against insects and diseases	180 (80)	45 (20)
To make crops grow better	53 (23.56)	172 (76.44)
Because others use pesticides	5 (2.22)	220 (97.78)
Because I was advice to use it	10 (4.44)	215 (95.56)
Where do you get/buy the pesticides?
Agrochemical shops in town	61 (27.11)	164 (72.89)
Local government shops in the village	93 (41.33)	132 (58.67)
Extension officers	85 (37.78)	140 (62.22)
General shops	15 (6.67)	210 (93.33)
Cooperative societies	0 (0.00)	225 (100)
Timing of pesticides application
Presence of pests	144 (64)	81 (36)
Degree of pest infection	17 (7.56)	208 (92.44)
Date of planning	3 (1.33)	222 (98.67)
On the calendar spray schedules	67 (29.78)	158 (70.22)
Pesticide application strategy
Mix more than one type of chemical	23 (10.22)	202 (89.78)
Depending on the instruction on the label	26 (11.56)	199 (88.44)
Sources of farmers’ knowledge on pesticide application
Agrochemical shops	11 (4.89)	214 (65.11)
Extension officers	156 (69.33)	69 (30.67)
Pesticides labels on packages	23 (10.22)	202 (89.78)
Fellow farmers	40 (17.78)	185 (82.22)
Own experience	23 (10.22)	202 (89.78)

N = number of respondents; Numbers in parenthesis indicate percentages. N (225).

The majority of the farmers indicated that they obtained pesticide knowledge from extension officers (69%). Other farmers used their own experience (10%) or, they were taught by fellow farmers (18%).

3.2. Pesticides storage environment

3.2.1. Pesticides storage location and level of education

Figure 2 illustrates the result of storage of pesticides options explored by the farmers. Thirty-two percent (32%) of the farmers used the bush as their main storage facility for the pesticides they used.

Figure 2. Pesticides storage location by cocoa farmers.

Some respondents (17%) stored chemicals within their living rooms, whiles 7%, 3% and 4% of respondents stored them in agrochemical shop, kitchen and animal house respectively. The result of linking storage location to the levels of education is presented in Table 5.

Table 5. Relationship between pesticides storage location and farmers level of education.

	Education
Storage location	No formal	Primary	Secondary	Inferential statistics
Agrochemical store	4 (5.4)	5 (6.3)	6 (8.3)	X^2 = 24.05, p = 0.045* Cramer’s V = 0.231
Animal houses	2 (2.7)	4 (5.1)	3 (4.2)
In the food storeroom	8 (10.8)	13 (16.7)	12 (16.5)
Living house	8 (10.8)	9 (11.4)	22 (30.6)
In the kitchen	2 (2.7)	3 (3.8)	22 (30.6)
In the bush	26 (32.9)	26 (32.9)	19 (26.4)
In the toilet	6 (8.5)	5 (6.3)	5 (6.9)
In the Bathroom	18 (24.3)	14 (17.7)	3 (4.2)

X² = Pearson chi square; P = probability value; Cramer’s V value (0–1) indicates how strong the values are associated with closeness to 1 which implies greater association. * = Significant result (p < 0.05); Numbers in parenthesis indicate percentages.

There was a statistically significant association between farmers’ pesticide storage location and their educational levels (X² = 24.05, p < 0.05). This means that farmers’ knowledge of pesticide storage location is influenced by their level of education. Further, the usual and common way of disposing empty pesticide containers and remnants from spraying equipment was throwing them in the farm (Figure 3). Anecdotal evidence shows that empty pesticide containers and sachets were found disposed of indiscriminately on farms. Five of the respondents (2%) revealed that they put empty pesticide containers to other use once they were emptied of its content. Some farmers (8%) also mentioned digging holes on farm and burying containers as their preferred disposal method.

Figure 3. Disposal of empty pesticide containers.

3.2.2. Farmers’ response towards pesticide use

Farmers’ opinion on the effectiveness of spraying was sought by expecting respondents to agree or disagree with certain statements. As shown in Table 6, 55% of the respondents strongly agreed to the statement that pesticide use was important to secure good crops. The majority of the respondents (98.7%) also admitted that knowledge was needed for the application of pesticides. The majority of the respondents agreed that precaution is necessary in the administration of chemicals (53% strongly agreed, 47% agreed). One third of the respondents (32%) disagreed that minimal health risks is associated with pesticide use, while the majority (68%) agreed. Further, 37.8% recognized and agreed that it is necessary to limit pesticides use. This might apparently be due to health symptoms that some of the respondents experienced during pesticide application.

Table 6. Attitudes towards Pesticide Use.

Farmers’ perception on the effectiveness of spraying	Respondents N	%
Important to secure good crops
Agree	101	44.89
Strongly Agree	124	55.11
Proper knowledge is necessary
Strongly Disagree	1	0.44
Disagree	2	0.89
Agree	96	42.67
Strongly Agree	126	56.00
Precautions should be used
Agree	106	47.11
Strongly Agree	119	52.89
Minimal Health Risks attached to pesticide use
Strongly Disagree	1	0.44
Disagree	71	31.56
Agree	81	36.00
Strongly Agree	72	32.00
Limit pesticide use
Strongly Disagree	16	7.11
Disagree	63	28.00
Agree	85	37.78
Strongly Agree	61	27.11

N = number of respondents; Numbers in parenthesis indicate percentages.

3.3. Health and safety impacts due to pesticide application

3.3.1. The use of personal protective equipment (PPE)

Farmers were asked whether they use a single item, or multiple items of PPE. Multiple PPEs most often involve wearing two or more of the following: hats, gloves, goggles, respirator, protective boots, and coveralls. Farmers were asked whether they used full working gear of multiple PPEs for protection during spraying (Table 7). Fifty-five (55) farmers failed to use any safety equipment (zero PPE), while 68 of the farmers revealed they used the full working gear (six PPE items). Most of the farmers with zero PPE usage in the study were noted to be farmers who had no formal education. In addition, 102 of the farmers partially protected themselves before using chemicals on the farms (Table 7). Farmers with either primary or secondary level of education used all six PPEs items or some form of partial PPEs during spraying.

Table 7. Use of PPEs and educational background of the farmers during the application of pesticides.

Number of PPEs	No formal education (N)	Primary education (N)	Secondary education (N)	Total (N)
0	22	21	12	55
1	9	6	12	27
2	8	4	4	16
3	4	7	10	21
4	3	7	5	15
5	7	9	7	23
6	21	25	22	68
Total (N)	74	79	72	225

N = number of respondents; PPE = personal protective equipment.

3.3.2. Common health symptoms associated with frequent pesticides usage

Table 8 shows common health symptoms self-reported and experienced by the farmers due to pesticide application. Data revealed that more than half of the respondents experienced symptoms of headache and burning eyes at 66% and 52% respectively. The remaining symptoms were skin rashes (32%), itching (48%), and chest pain (42%).

Table 8. Common health symptoms associated with frequent pesticide usage.

Symptoms	Response N (%)
	YES
Have you experienced any of the following symptoms after chemical application?
Headache	118(52.44)
Burning eyes	148(65.80)
Skin rash	73 (32.44)
Itching	109 (48.44)
Chest pain	95 (42.22)

N = number of farmers; Numbers in parenthesis indicate percentages.

3.3.3. Factors influencing PPE use and logistic regression modelling

To investigate factors influencing PPE use, Cramer’s V correlation was determined and presented in Table 9. The result showed a positive and negative correlation, with all the factors showing weak correlation <10%, with the exception of gender (30%). The weak correlation might suggest confounding factors influencing the variables, since the literature predicted otherwise. In Table 9 only gender and district of the farmers showed a statistically significant difference (p < 0.01), although the literature indicated several factors influencing PPE usage.

Table 9. Correlation analysis of PPE use and source of pesticide knowledge with demographic variables.

Variables	Sign	Significance
Agrochemicals	(+)	ns
Extension services	(+)	ns
Farmers experiences	(-)	ns
Gender	(-)	s
Education	(+)	ns
Districts	(+)	s

Significant for coefficient: p < 0.05; s for significant and ns for non-significant; + and – signs indicate positive or negative correlation.

The data was modelled with the complementary log-log regression and was assessed with goodness of fit tests using Deviance, Pearson and Akaike Information Criterion (AIC). The goodness fit test showed Deviance to be 1.132, Pearson = 1.028, and AIC = 1.167. Since Deviance, Pearson and AIC values are almost similar to each other and with very small values, it suggested that the model fit was considered satisfactory. The actual result of the complementary log-log regression modelling is presented in Table 10 showing the odds ratios, standard errors, p-values, and confidence intervals (CI) associated with the use of PPE, as well as compositional and contextual factors. Model 1 shows that agrochemical shops and extension services were statistically significant, indicating that farmers who depended on agrochemical shops (OR = 1.735, p < 0.005) as a source of information with regard to pesticide application were more likely to use PPE during pesticide application compared to their counterparts. Similarly, the probability that farmers who depended on the extension services (OR = 1.643, p < 0.007) as a source of knowledge with respect to pesticide application increased as compared to their counterparts. This situation could arise as farmers meet the extension officers and thus they receive educational information regarding the use of PPE in pesticide application. Surprisingly, farmers’ years of experience in pesticide application was not statistically significant.

Table 10. Multivariate complementary log-log regression model predicting PPEs usage during pesticide application.

	Source of pesticide knowledge (Model 1)				Bio-social factors (Model 2)				Socio-cultural factors (Model 3)				Contextual factors (Model 4)
Predictors	OR	SE	p	95% CI	OR	SE	p	95% CI	OR	SE	p	95%CI	OR	SE	p	95%CI
Agrochemicals shops (Ref: No)
Yes	1.735	0.343	0.005	1.178–2.554	1.773	0.358	0.005	1.192–2.636	1.777	0.357	0.004	1.200–2.633	1.646	0.334	0.014	1.106–2.452
Extension services (Ref: No)
Yes	1.643	0.303	0.007	1.144–2.359	1.507	0.287	0.032	1.037–2.189	1.497	0.288	0.036	1.027–2.183	1.374	0.268	0.104	0.937–2.015
Farmers’ experiences (Ref: No)
Yes	0.752	0.752	0.362	0.408–1.388	0.833	0.259	0.557	0.452–1.533	0.838	0.260	0.569	0.456–1.539	0.837	0.262	0.57	0.453–1.546
Gender (Ref: Male)
Female					0.481	0.097	0.000	0.325–0.713	0.483	0.097	0.000	0.325–0.716	0.466	0.094	0.000	0.313–0.693
Education (Ref: No formal education)
Primary education									1.010	0.217	0.961	0.663–1.540	1.095	0.240	0.679	0.713–1.681
Secondary education									1.065	0.229	0.770	0.699–1.624	1.090	0.238	0.692	0.711–1.672
Districts (Ref: Wassa Amenfi West)
Aowin													0.882	0.187	0.553	0.581–1.337
Suaman													0.568	0.125	0.010	0.368–0.875

OR-odds ratio; SE-standard error; p-probability value; CI-confidence intervals.

The result for model 2, where biosocial factors were controlled for, showed that farmers who relied on agrochemical shops (OR = 1.773, p < 0.005) and extension services (OR = 1.507, p < 0.032) as a source of information about pesticide application were more likely to use PPE than those who responded negatively. Gender (p < 0.000) also influenced PPE use during pesticide application. It was also revealed that female farmers were 48.1% less likely to use PPE during pesticide application compared to male farmers. This response may be because although female farmers may be owners of the farm, they may consult their male counterparts to do the spraying and may utilize PPE during pesticide application. When the socio-cultural factors (e.g. education) was controlled for in model 3, agrochemical shops (OR = 1.777, p < 0.004) and extension services (OR = 1.497, p < 0.036) continued to predict the use of PPE. However, the probability of PPE use during pesticide spraying was higher for farmers who depended on agrochemical shops for information about of pesticide application compared to their counterparts. Farmers who depended on extension services on the other hand were 100% likely to use PPE than their counterparts. Interestingly, there was no significant relationship between educational levels and the use of PPE during pesticide application.

In model 4, contextual factors influencing famers’ choice to use PPE were considered by controlling districts from which the respondents were drawn. Observations under farmers’ experience were not statistically significant for models 1 to 3, while the extension services variable, which was statistically significant in models 1 and 2 ceased to be significant in model 4, when the contextual factor was added. In addition, agrochemical shop was significant under models 1 to 4. Farmers who depended on agrochemical shop services (OR = 1.646, p < 0.014) for information about pesticide application were still more likely to use PPE compared to their counterparts. The result also showed that female farmers (OR = 0.466, p < 0.000) were still less likely to use PPE during pesticide application than the male farmers. Concerning districts, only Suaman district showed relationship with PPE use. Farmers in Suaman (OR = 0.568, p < 0.010) were less likely to use PPE compared to Wassa Amenfi West district. It is obvious from the results that agrochemical shop services and extension services influence farmers’ choice to use PPE during pesticide application.

3.3.4. Bootstrapping model statistics

In order to validate the model statistics and avoid doubts of sample distribution, bootstrapping was used to measure the accuracy of the logistic regression model (multivariate complementary log-log) parameter estimates. Table 11 shows the result of the bootstrapping used. The bootstrap method attempted to estimate the sampling distribution empirically with the given sample size of 225 and estimate the parameters on large scale of 2000 sample size. There were no differences between the OR, SE, and CI for logistic complementary log-log regression model in Tables 10 and 11. In addition, the p-values were identical as predicted by both the logistic regression and the bootstrap method. The implication of the result is that the sample size distribution and parameter estimates are accurate and correctly predicted by the complementary log-log model.

Table 11. Bootstrap results generated after 2000 samples to validate parameters of multivariate complementary log-log regression model predicting PPEs usage during pesticide application.

	Source of pesticide knowledge (Model 1)				Bio-social factors (Model 2)				Socio-cultural factors (Model 3)				Contextual factors (Model 4)
Predictors	OR	SE	p	95%CI	OR	SE	p	95%CI	OR	SE	p	95%CI	OR	SE	p	95%CI
Agrochemicals shops (Ref: No)
Yes	1.735	0.365	0.009	1.149–2.619	1.773	0.387	0.009	1.156–2.718	1.777	0.370	0.006	1.182–2.673	1.646	0.334	0.014	1.106–2.452
Extension services (Ref: No)
Yes	1.643	0.317	0.010	1.125–2.399	1.507	0.290	0.033	1.033–2.198	1.497	0.299	0.043	1.012–2.215	1.374	0.268	0.104	0.937–2.015
Farmers’ experiences (Ref: No)
Yes	0.752	0.250	0.392	0.392–1.443	0.833	0.284	0.591	0.427–1.623	0.838	0.290	0.610	0.425–1.653	0.837	0.262	0.570	0.453–1.546
Gender (Ref: Male)
Female					0.481	0.103	0.001	0.316–0.734	0.483	0.103	0.001	0.318–0.732	0.466	0.094	0.000	0.313–0.693
Education (Ref: No formal education)
Primary education									1.010	0.227	0.963	0.651–1.569	1.095	0.240	0.679	0.713–1.681
Secondary education									1.065	0.246	0.785	0.677–1.676	1.090	0.238	0.692	0.711–1.672
Districts (Ref: Wassa Amenfi West)
Aowin													0.882	0.187	0.553	0.581–1.337
Suaman													0.568	0.125	0.010	0.368–0.875

OR-odds ratio; SE-standard error; p-probability value; CI-confidence intervals.

4. Discussion

4.1. Demographic characteristics of farmers

This study investigated smallholder cocoa farmers’ knowledge of pesticide use, their attitudes towards storage of pesticides, and evaluated their exposure to health symptoms. The study revealed that the common reason for using pesticides was to protect cocoa plants from insects, pests and diseases. This finding is in line with Khan and Damalas (2015), who reported that in order to control pests and prevent loss of crop yields in their crops, farmers use synthetic pesticides extensively. The present study also showed that the dominant gender involved in cocoa farming in the study area is the male. The large male to female ratio in this study is in line with the findings of Bosompem et al. (2012), Boateng et al. (2014 ), Antwi-Agyakwa et al. (2016), Zhu (2015), and Tijani (2006). The educational background of respondents showed that a good number of farmers had received basic and secondary level education while the majority of the farmers did not have further studies beyond the secondary school level. Nonetheless, the proportion of illiterates was equally low. This case is similar to Paintsil (2017) and Zhu (2015) findings, where farmers had the view that a high level of education is not necessary for carrying out farming. A study from Nigeria (Oluwole and Cheke 2009) also confirmed this trend with a similar finding where the majority of farmers surveyed had no formal education. Hence, it can be deduced that the inability of a farmer to undertake a good agricultural practice is because of poor educational background in rural Africa. The results further show that three-fourths of the respondents had either no or just primary level of education. Thus, farmer’s level of education may be a contributing factor to their inability to read the labels on the chemical containers and in understanding the hazardous nature of pesticides chemistry. In a related study, about 94% of the farmers stated that pesticide labels were difficult to read and understand which was attributed to low educational levels, poor literacy skills and difficulty in following the language used in the wording of the label (Damalas et al. 2006). The cocoa sector has been the mainstay of the Ghanaian economy since 1957 (Vigneri and Kolavalli 2018). Surprisingly, however, it is pathetic to note that this cash crop for Ghana is not attracting those with tertiary level education. It is no wonder, the cocoa farming production is still in the hands of smallholder farmers. Ghana Statistical Service (2013) data indicated that Ghana’s agricultural sector is dominated by 90% smallholder farmers with less than 2 hectares of land. These farmers still use traditional production methods and farm inputs. The current global state of agricultural practice is driven mainly by science and technology, but without adequate level of education, applying science and technology to transform agriculture to a knowledge-based enterprise may become quite problematic. Recently, Kwadzo (2015) examined smallholder cocoa farmers in Mpohor-Wassa East district, Ghana and found that 73% of the farmers had shifted from cocoa to rubber cultivation. The result buttressed the fact that investment outcomes of cocoa have a significant effect on their enterprise-shift behaviour and decisions. If this trend is drifted to other districts and regions, it may not auger well for the sustainability of the cocoa sector.

4.2. Relationship between farmers’ level of education and knowledge of pesticide toxicity

In general, most farmers had good knowledge of the effects of pesticides on human health as they indicated how their bodies react after the spraying of pesticides. It was also discovered that farmers’ knowledge in health effects of pesticides was influenced by their educational background. According to Bagheri et al. (2018), well-educated farmers had more safe behaviours than less-educated farmers. Similarly, well-educated farmers were more likely to pay a high premium for safe pesticides (Khan and Damalas 2015). Thus, training and education positively influence environmentally sound behaviour in pest management (Damalas and Khan 2017). Interventions such as education and training of farmers which enhance safety behaviour should be intensified to minimize pesticides exposure among farmers, (Damalas and Koutroubas 2017). It is clear that having a clear insight regarding farmers’ level of knowledge and farmers’ practices on safe use of pesticides is necessary to augment the current scenario for a new policy change (education). The policy change should involve the farmers, the extension agents, the agrochemical retailers, and regulatory agencies. This new policy change is needed to protect the public health. Contrary to expectations, although, the majority of the farmers in this study were aware of the harmful effects of pesticides, this did not significantly change their practices or attitudes towards safe pesticides use. This finding is consistent with that of Sharifzadeh et al. (2019) and showing that even though good knowledge about pesticide safety is imperative for farmers, this alone is not enough to encourage them to indulge in safety behaviours. Perhaps cultural and social driving forces are strong determinants limiting behavioural change necessary to evoke a collective safety responsibility (Feola and Binder 2010).

4.3. Relationship between pesticides storage location and farmers level of education

The present study also highlighted some unsafe practices regarding storage of pesticides, i.e. in the kitchen, living room and in the food storerooms. Thus, it was clearly shown that farmers were lacking knowledge regarding appropriate places for storing pesticides. Storing pesticides in the living rooms, kitchen, and food storerooms increases the potential for pesticide exposure risks. The majority of the farmers kept chemicals in their farmlands. However, a good number of them kept chemicals within the living room, kitchen, in the food storeroom, and in the animal house. An additional fascinating situation in these rural communities setting was the storage of chemicals in the toilet and bathroom. A very small proportion of the farmers (7%) kept pesticides in the agrochemical stores. This finding is similar to that of Paintsil (2017) because the selected study area is in the same region but in different districts. Oluwole and Cheke (2009) also gave similar support to this assertion with data from rice farmers, while Zhu (2015) recognized a similar trend in vegetable farmers within the cocoa growing belts. Additionally, Bagheri et al. (2018) found out that about 60% of the farmers in Iran stored their pesticides in stalls and warehouses, while about 40% threw empty pesticide containers at the orchard and in the canal. On the other hand, Tijani (2006) uncovered a different pattern, that is, the majority of the farmers were storing pesticides in designated stores and a minority was keeping them in their bedrooms. The attitudinal behaviour exhibited by the farmers in understanding the hazardous nature of pesticides storage location is linked to the educational levels of the farmers. Based on these findings, farmers need to be trained on proper and safe storage of pesticides. Damalas and Koutroubas (2017) have shown that training of farmers was associated with increased levels of knowledge of pesticides and beliefs of pesticides hazard control, which was accompanied by high safety attitude in farmers resulting in lower occupational pesticide exposure. Undoubtedly, farmers who undertake educational programs experience fewer poisoning symptoms by pesticides (Bagheri et al. 2018). Apart from education, farmers can also be motivated to store and dispose of pesticides in a safe manner through the constitution of credit bonuses at the purchase of pesticides. For example, part of the money paid for the pesticide by the farmers can be given back to them when they return the pesticide containers to the manufacturers, retailers or packaging companies (Bagheri et al. 2018).

It was also found that the most prominent containers and sachet disposal strategies currently employed were throwing in the field, village landfills, burning on farm and burying in a hole (Figure 3). This trend is coherent with data reported by Tijani (2006), Oluwole and Cheke (2009) and Paintsil (2017). Previous studies in Ethiopia and Greece found dumping empty pesticide containers by fields, near, or into irrigation streams and canals and burning them in open fire are well-known practices of pesticide container disposal methods that farmers are often involved in, coupled with using them for storage of fuel, water and food (Damalas et al. 2008; Haylamicheal and Dalvie 2009). It was also observed that some farmers practice rinsing the empty containers by discharging the water into nearby uncultivated lands, throw away empty containers into rivers, lakes or irrigation canals or bury them in the ground. However, it is interesting and alarming to note that some farmers also put the empty containers to other use for storing household items such as salts, palm oil, flour and other products meant for consumption. In addition, the majority of the respondents were found even washing their pesticide containers in rivers, streams or irrigation canals. Similar behaviour and attitudes were observed by Jallow et al. (2017) among farm workers in Kuwait, showing that the practice cut-across various cultural backgrounds. Thus, again this study demonstrated poor knowledge of cocoa farmers about pesticides and their transport in the environment. These poor handling and disposal practices can have devastating effects on soil, water contamination, and the overall impacts on public health. This is because such unsafe practices can release pesticide residues and contaminate the environment (Damalas and Eleftherohorinos 2011, Miyittah et al. 2020).

4.4. Pesticide application information by cocoa farmers

The present study also indicated that most farmers obtained pesticides from local governmental shops as their main source. Anecdotal evidence supported by this reality is that most of these agrochemical retailers themselves need more education on pesticide use and handling. Moreover, if pesticide retailers are well informed they can help by providing accurate source of information regarding environmental and human health impacts of pesticides to farmers who cannot read instructions or labels on the containers. There is, therefore, a need to train and equip pesticides retailers regarding dissemination of agricultural information, since they can play a critical role as a primary information and knowledge source for the farmer. Lekei et al. (2014) reported the impact of retailers as technical advisors on farmers and other end users as a key contributing factor in occupational exposure to pesticides. Additionally, 69% of the respondents obtained information for pesticides from agricultural extension officers. This observation is in line with results obtained by Tijani (2006) and Zhu (2015). However, others prefer to use their own experience or get information from their fellow farmers. There is nothing wrong if the farmer gets information from their peers. However, the difficulty occurs when the said information is not accurate and the source of information cannot be verified by the farmers themselves. The unverified information can be further distorted along the communication channels and such distortion can contribute to propagation of inaccurate information regarding pesticides use among cocoa farmers. Agricultural extension officers in general act as conduits between the Ministry of Food and Agriculture and farmers, or farm workers. Extension aims primarily at improving the knowledge of farmers for rural development. Thus, agricultural extension plays a critical component of technology transfer (Bonye et al. 2012). In general, extension officer-to-farmer ratio in Ghana is about 1:3000. However, the COCOBOD as an agency in charge of cocoa have reduced the ratio gap by having extension services specialized for cocoa affairs.

This gap reduction in extension officer-to-farm ratio may be the reason why 69% of the respondents reported that they obtained information about pesticides through extension officers. A considerable number of the respondents (90%) reported mixing more than one pesticide type and applying to cocoa farms as one of the pesticide application strategy. This practice has further demonstrated that the farmers lack knowledge application doses and the impact it may have on the toxicity of insect pests and on the development of resistance of the insects with respect to the said pesticides, as reported elsewhere (Damalas and Khan 2017). It has been reported that over 600 species of pests have developed some level of resistance to pesticides (Gill and Garg 2014). Therefore, new policy and updated training is urgently required to educate the retailers, extension agents, farm workers and farmers regarding pesticides resistance and the implications it may have on the cocoa sector. There should be a documentation of all pesticides sold and a link of the respective serial numbers on containers with the farmer through the retailer. The farmer should have a pesticide book with a documentation where the pesticides was bought and with the documented location of the retailer. At the end of the pesticide application, a mechanism should be put in place to retrieve all the empty containers. By this approach, the empty containers would no longer be used as alternative storage containers with its concomitant health implications. The present study further showed that most farmers had good knowledge of the effects of pesticides on human health and the environment. Most farmers were positive that pesticide could enter the body via mouth (92%), inhalation (94%), and skin (93%). The majority was also aware that residues of pesticides can be deposited on fruits and vegetables and they can contaminate soils and groundwater.

4.5. Common health symptoms associated with frequent pesticide usage

It was also observed that 66% of the farmers experienced health symptoms such as headaches, burning eyes, skin rashes, itching and chest pain. This may be attributed to the heavy use of pesticides for pest management and the non-use of PPE or the use of inappropriate PPE during the various stages of pesticide usage. Similarly, Atreya (2007), on the other hand found that pesticide operators reported greater signs and symptoms of pesticides exposure such as skin irritations, stomach poisoning, and eye irritations than other farm workers. Similar effect was observed in another study conducted by Neghab et al. (2014) with 268 male farmers in Iran. The result showed that 68% of the participants reported to their general health practitioners, suffering from burning and skin irritations, burning eyes, headaches, vertigo, nausea and vomiting during spraying. Damalas and Koutroubas (2016) found that accurate usage of appropriate PPE in all stages of pesticide handling coupled with less use of pesticides could reduce farmers’ exposure to pesticides. Toxicity symptoms of pesticide may be categorized as mild (skin irritation) and severe (headache, nausea and dizziness) (Damalas and Koutroubas 2016).

4.6. Personal protective equipment (PPE) use by the farmers

About fifty-five farmers failed to wear any personal safety equipment (zero PPE), while 68 of the farmers revealed they used the full working gear (six PPE items). Most of the farmers who used zero PPE in the study were noted to be farmers who had no formal education. In addition, 102 of the farmers partially protected themselves before using chemicals on the farms. Farmers with either primary or secondary level of education used all six PPE items or some form of partial PPE items during spraying. Knowledge, attitudes, and practices (KAP) surveys by Ntow et al. (2006) showed that only 32% of the farmers were wearing full PPE. In addition, measuring the relative toxicity of pesticides used in controlling pest in Akumadan, Ghana, showed that 58% of the farmers did not use any PPE, while only 29% used some form of PPE. Furthermore, high illiteracy rates contribute to farmers’ difficulty in understanding and following instructions and safety advice on pesticide use.

In Brazil, knowledge was not found to influence pesticide application practices because the majority of the farmers admitted receiving information, training and claimed reading labels, adhering to instructions and warnings, but did not take adequate protective measures (Waichman et al. 2007). The level of knowledge and perception of risk were not enough to influence farmers’ self-protective behaviour (Remoundou et al. 2014). It was also reported that farmers were exposed to agrochemicals because of non-use of PPE cover cloths during spraying and leakages from knapsack sprayers. In addition, spraying during windy conditions can cause incidental drifting of the chemicals to unapproved routes. The study also revealed a positive correlation between education, agrochemical services, extension services, and location of farmers with farmers’ PPE use. However, farmers’ years of experience and gender showed a negative correlation with farmers’ PPE use. In addition, the modelling showed that factors such as agrochemical shops, extension services availability as sources of information with respect to pesticide application had positive significant influence on PPE usage. This finding is in line with the observation of Okoffo et al. (2016) that the probability of a farmer wearing PPE increases with the availability of agrochemical shops. This behaviour may be attributed to the fact that agrochemical shop retailers can serve as a conduit in educating the farmers on the dangers in application of pesticides without the use of PPE. Even when the model was controlled for biosocial, socio-cultural, and contextual factors, these parameter estimates were still significant, except for extension services in contextual factor modelling. These services should be used as a medium to educate farmers on the importance of PPE use by providing training and capacity building for extension officers and agrochemical retailers. Previous research has shown that farmers who perceived usefulness of PPE, such as effectiveness, safety, and ease of use were more willing to use PPE in the future (Sharifzadeh et al. 2017). The implication is that proximity of farmers to extension officers and location were crucial in PPE usage. In addition, the sources of information about pesticide application should be supported by governmental interventions regarding pesticide educational activities to encourage the famers to use PPE. Surprisingly, however, farmers’ years of experience in pesticide use was not translated into PPE usage. The parameter estimate for farmers’ years of experience was similar under biosocial, socio-cultural, and contextual factors. The prevailing socio-cultural conditions can serve as confounding factors (Feola and Binder 2010) such that the educational information received maybe masked after massive education regarding pesticide application. The implication is that sociological conditions underpinning the socio-cultural factors hampering the use of PPE must be investigated to shed more light on the situation regarding acceptability of PPE use in the study area. However, other studies have indicated that farmers’ experience with adverse health effects of pesticides significantly influence their safety behaviour and the use of PPE (Feola and Binder 2010; Hashemi et al. 2012; Damalas and Abdollahzadeh 2016; Sharifzadeh et al. 2018, 2019). Thus, the more farmers experience threats and health risks by pesticides, the more they are likely to show safety behaviours (Abdollahzadeh et al. 2015; Damalas and Abdollahzadeh 2016). The inability of the female farmers to use PPE could also be linked to what some of them said during the field survey about the discomfort they go through any time they put on PPE. This phenomenon is worrisome because when female farmers are exposed to pesticides, they can indirectly expose their breast-feeding babies (Lorenz et al. 2012). A study on the analysis of pesticide contamination of farmers in Ghana revealed the presence of residues of organo-chlorine pesticides, including dichlorodiphenyltrichloroethane (DDT), in the breast milk and blood of female farmers (Ntow et al. 2008). The sociological condition might also be the reason why females are less likely to use PPE compared with their male counterparts. Our preliminary interpretation is that women would be more exposed to health risk of pesticide than men. This may be attributed to the fact that female smallholder farmers have limited access to training programs regarding pesticide safety, and hence, they follow just few pesticide safety behaviours when handling pesticides (Naidoo et al. 2010; Damalas et al. 2019). In order to enhance pesticide safety awareness among female farmers, gender-sensitive safety programs should be organized (Wang et al. 2017; Damalas et al. 2019). In this study, about 30% of women were engaged in cocoa farming and therefore, further research is needed in this direction, since pesticide hazards have several debilitating and consequential effects on women as child-life support givers. Educational levels have been observed as having no influence on the use of PPE under socio-cultural and contextual factors. On the contrary, other researchers proposed that educational programs enhance sustainable PPE use among pesticide applicators and smallholder farmers (Sharifzadeh et al. 2019). This means that there is a significant positive effect on PPE use and education, and thus, the educational status of the farmers strongly determines their PPE usage (Al Zadjali et al. 2015; Blanco-Munoz and Lacasana 2011; Sharifzadeh et al. 2019). Educational impacts on individuals occurred in multiple layers with interacting context (Rappaport and Smith 2010; Armah et al. 2019). Moreover, each of these contexts is a domain of social relations and each factor in each domain interacts. Thus, there is a difference between education as a context and education as a process because these two elements have different types of implication on environmental health and can affect pesticide use. This understanding is in line with Feola and Binder (2010) who indicated that socio-cultural factors are usually masked by educational factors; hence, there is a need to disaggregate the various elements and their interacting effects. Under the contextual factor, the location of the district towards the use of PPE is a case in point. For example, farmers in Suaman district were less likely to use PPE compared with those in Wassa Amenfi West. Thus, the sociological mindset occurring in a particular district bounded by language and culture may contribute towards the use of PPE. Elements within the culture and the language that is leading to the influence of PPE use must be studied. Perhaps, more access to extension services within the district may be the contributing factor in using PPE as noted by Danso-Abbeam et al. (2018) who reported agricultural extension plays a critical role in improving the knowledge base of the farmer and in the transfer of technology. The use of PPE is a type of skill that influences productivity and the farmer must have it and understand why it is important. This is because the health of smallholder farmers who has been the backbone of Ghana’s economy for decades is at a risk, and there is the need to protect the human health and the environment in order to sustain the cocoa industry in a sustainable manner. One limitation of this study is that it was based on self-reported data that depends heavily on the sincerity of the participants, which is subject to some extent of biases (Weinstein and Klein 1996; Jallow et al. 2017). Self-reported studies may include some inaccurate data such as respondents trying to be politically correct or report socially desirable behaviours. A second limitation could be the inability to link directly health symptoms with pesticide exposure. It could be that other factors may be responsible for the health symptoms (Jallow et al. 2017). Despite these limitations, the study provided a window of insight into pesticide use, knowledge, and safety practices among smallholder cocoa farmers and it could assist in major policy change to protect public health and the environment. Policy changes are necessary to ensure the overall cocoa beans health for global exports and for the chocolate industry sustainability.

5. Conclusions

The drive to earn foreign exchange through increase cocoa productivity is huge, and so is pesticide usage. In this study, we investigated potential exposure factors that are likely to cause harm to human health and the environment among cocoa farmers. We found that farmer’s method of storing, disposing, and washing of empty pesticides containers after use were inappropriate and potentially detrimental to human health and the environment. Farmers’ level of education had a strong association with the toxicological routes of entry of pesticides into the human body. Common and frequent health symptoms experienced by the farmers were, headache, burning eyes, skin rashes, itching and chest pain. These health symptoms were likely due to inappropriate and inadequate use of PPE. It was also found that farmers’ level of knowledge acquired on the dangers of pesticides was not translated into actual use of PPE. Several factors likely influenced the usage of PPE among farmers. Through modellng, factors affecting the use of PPE were agrochemicals shops, extension services, and farmers’ district location. The obvious implication is that these factors must be brought into the equation for policy interventions that could minimize farmers’ exposure to pesticides as well as health impacts of pesticide use in cocoa farming.

Author contributions

M.K.M conceptualization, study design, methodology, drafting and review. B.A: coordination of the pesticide analysis, software programming, drafting.

M.K software programing, method validation, interpretation of results, review and editing of this article. S.S.L method validation, review and editing of the article and R.K.K software programming, review and editing of the article. All authors M.K.M., M.K., B.A., S.S.L., and R.K.K. revised the manuscript critically for important intellectual content and approved of the version to be published.

Disclaimer

The authors declare that the findings and conclusions in this article are those of the authors and do not represent the views of the organisations of affiliation or agencies.

Ethics consent and permissions

All participants agreed to participate in the research study, and they were free to participate without duress and coercion.

Acknowledgments

We acknowledged the support of Mr Gabriel Addae of Ministry of Food and Agriculture (MoFA), Extension Services Division and the COCOBOD Extension Officers, together with the farmers during the survey.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This study was not funded by any grant.

Data availability statement

The authors confirm that the majority of the data supporting the findings of this study are available within this article.