Dermal Exposure Assessment to Pesticides in Farming Systems in Developing Countries: Comparison of Models

In the field of occupational hygiene, researchers have been working on developing appropriate methods to estimate human exposure to pesticides in order to assess the risk and therefore to take the due decisions to improve the pesticide management process and reduce the health risks. This paper evaluates dermal exposure models to find the most appropriate. Eight models (i.e., COSHH, DERM, DREAM, EASE, PHED, RISKOFDERM, STOFFENMANAGER and PFAM) were evaluated according to a multi-criteria analysis and from these results five models (i.e., DERM, DREAM, PHED, RISKOFDERM and PFAM) were selected for the assessment of dermal exposure in the case study of the potato farming system in the Andean highlands of Vereda La Hoya, Colombia. The results show that the models provide different dermal exposure estimations which are not comparable. However, because of the simplicity of the algorithm and the specificity of the determinants, the DERM, DREAM and PFAM models were found to be the most appropriate although their estimations might be more accurate if specific determinants are included for the case studies in developing countries.


The Pesticide Issues
Pesticides are key elements of pest management programs in modern agriculture to increase the levels of production. Their use is stimulated by the commercialization and intensification of agriculture, the difficulty in expanding cropped acreage, the increased demand for agricultural products as the population increases, and the shift to cash crops for domestic and export sales [1]. It is estimated that annually some 2.5 million tons of pesticide are used worldwide and 220,000 people die because of poisoning from these substances. Most of these poisonings occur in developing countries because of weak safety standards, minimal use of protective equipment, absence of washing facilities, poor labeling, and lack of information programs [2][3][4][5][6].
Public health experts have expressed increasing concern about the use of pesticides because epidemiological studies have found that they are associated with different types of cancers [7][8][9][10], neurologic pathologies [11][12][13], respiratory symptoms [14] and hormonal and reproductive abnormalities [15][16][17][18][19]. Regardless of the risks involved in the use of pesticides, they are considered a key input to agriculture allowing intensive production techniques [20]. Therefore, it is crucial to assess the risk due to pesticide use by improving their management, reducing the exposure and protecting human health.
The agricultural sector in Colombia uses 3.8 million hectares of land for permanent and transitory crops. During the last decade, an average of 82,000 tons of pesticides were applied per year (17% insecticides, 47% herbicides and 35% fungicides and bactericides) [21]. This suggests that part of the population and the environment in Colombia are likely to be exposed to negative effects derived from pesticide use. For instance, the potato farming system occupies 128,700 ha with 230,000 production units which had a production of 2.3 million tons in 2012 and used 32.5 kg/ha of pesticide active ingredients [22]. Therefore, the quantification of human exposure to pesticide use in farming systems like potato crops is crucial to provide information about the level of risk faced by farmers and workers and to support the development of proper policy measures.

Risk Assessment of Pesticide Use in Developing Countries
In the agricultural field, there is an increasing concern about the health of farmers, workers and bystanders, since they might be frequently exposed to pesticides for long periods of time. Governments, especially from developed countries, have introduced new environmental policies about the adequate use of pesticides. Meanwhile, in developing countries, like Colombia, a similar attempt has been done but even though the regulation scheme is already defined, this is not efficiently implemented due to the lack of information about exposure assessment and risk characterization [23,24]. The definition and implementation of these environmental policies is a further step after a risk assessment. Therefore, it is crucial to establish a method for the risk assessment of pesticide application in developing countries focusing in the exposure assessment and the risk characterization. The conclusions coming out from this method will be useful for stakeholders not only for the improvement of the risk assessment scheme, identifying the critical factors that influence the level of exposure concentrations, but also for the development of pedagogical programs about the appropriate use of pesticides.
The risk assessment of pesticide application can be divided into two essential parts: exposure assessment (qualitative and quantitative description of the exposure concentrations and related dose for specific pathways) and effects assessment (determination of the intrinsic hazards associated with the agent and quantification of the relationship between the dose with the target tissue and related harmful outcomes) [25][26][27][28]. The first part is known as the initial portion of the environmental health paradigm: from sources, to environmental concentrations, to exposure, to dose. The effects assessment is aiming for the latter portion of the events continuum: from dose to adverse health effects.
In the occupational hygiene field, the attention has shifted to the research of the exposure in the agricultural workplace to improve the pesticide management and to reduce the health risk [28]. This is of special interest in developing countries because pesticide management activities face weak safety standards [3,5,6,29]. Studies in potato farming systems in Vereda La Hoya, Colombia [3,5,23,24,[30][31][32][33], Mojanda, Ecuador [34] and El Angel, Ecuador [35] have shown that pesticide management has no a particular theoretical basis and instead it is performed by trial and error finding out what works out in practice. Furthermore, farmers do not wear adequate personal protective equipment, apply pesticides which are banned in industrialized countries and modify the standard discharge of nozzles to reduce the application time [31]. Because these issues increase the health risk due to human exposure, a risk assessment of pesticide use in these areas is required in order to determine the risk level.

Modeling Dermal Exposure to Pesticide Use
Indirect methods to assess human exposure have been used since the early 1990s [36]. Tools for dermal exposure, such as Control of Substances Hazardous to Health (COSHH) regulations [37], Dermal Exposure Assessment Method (DREAM) [38], Estimation and Assessment of Substance Exposure (EASE) [39], European Predictive Operator Exposure Model Database (EUROPOEM) [40], Pesticides Handlers Exposure Database (PHED) [41], Risk Assessment of Occupational Dermal Exposure to Chemicals (RISKOFDERM) [42], Qualitative Assessment of Occupational Health Risks (STOFENMANAGER) [43], and the approaches proposed by the U.S. EPA [44] are targeted at occupational situations encountered in industrial processes in Europe and the USA, but they do not consider agricultural processes such as pesticide management and there might be uncertainties when they are applied in study areas in developing countries. Dermal Exposure Ranking Method (DERM) [45] is a method focused on occupational activities in pesticide management in developing countries; nonetheless, its semi-quantitative estimations still lack reliability and validity [46,47]. Pesticide Flow Analysis Model (PFAM) [48] is a model focused on farming systems in developing countries based on the material flow analysis method, however, it is still not validated. Because of the lack of studies about the application and further evaluation of these models in farming systems in developing countries, there is no consensus about the best method to evaluate dermal exposure and the health risk in those systems. Therefore, existing models for dermal exposure (DERM, DREAM, PHED, RISKOFDERM, COSHH, STOFENMANAGER, EASE and PFAM) were evaluated in order to find out the most appropriate to be applied in case studies in developing countries. Along this evaluation the following research questions were addressed: 1. Which of the existing models for dermal exposure assessment are feasible to be applied in case studies in farming systems in developing countries? 2. According to the parameters and determinants included in the model structure, which model assessment is more complete in terms of the evaluation of dermal exposure? 3. When comparing the model outcomes with the dermal exposure measurements in the study area, which model assesses dermal exposure more accurately?

Multi-Criteria Analysis
After a literature review, eight available models were considered for the analysis: COSHH [37], DERM [45], DREAM [38], EASE [39], PHED [41], PFAM [48], RISKOFDERM [42], and STOFENMANAGER [43]. These models were selected because of their availability, clear model description and their potential applicability for the assessment of pesticide use in farming systems in developing countries. They were analyzed according to a group of criteria such as availability, guidance, knowledge required, reliability, type of outcome, type of substance, target group, dermal exposure descriptor and dermal exposure pathway which are explained in Table 1.

Estimation of Dermal Exposures in the Study Areas
From the results of the multi-criteria analysis and based on the model characteristics five models (i.e., DERM, DREAM, PFAM, PHED, and RISKOFDERM) were selected to be applied in the case study of potato farming systems in Vereda La Hoya in the highlands of Colombia. The data used as input comes from a previous survey made in the study area with 197 smallholder potato growers in four communities [3] and previous studies about dermal exposure in the same study area [24,31]. The input data and the scoring system for each determinant within each model are shown in the annexes. Because PFAM model required a specific pesticide with the total amount applied per hectare, the dermal exposure assessment was estimated for the pesticide methamidophos.

Description of the Study Area
The study area is located in Vereda La Hoya near Tunja, the capital city of the province of Boyacá, Colombia. This is a rural region devoted mainly to the cultivation of potato in production units of around 3 hectares in size. The crop depends on rainfall, therefore, the production is generally organized into two periods, one from March to September and another from October to February, which corresponds to the two rainy seasons. Average annual productivity is 18.3 ton/ha [22]. Potato crops in this region are vulnerable to three major pests: the soil-dwelling larvae of the Andean weevil (Premnotrypes vorax), the late blight fungus (Phytophthora infestans) and the Guatemalan potato moth (Tecia solanivora) [22]. These pests, together with the weeds present in the early phases of the crop, are controlled by the application of chlorothalonil, chlorpyrifos, cymoxanil, glyphosate, mancozeb, methamidophos and paraquat [5,32]. In the study area the pesticide management is performed along three main activities: the preparation of the pesticide, the application itself, and the cleaning of the spraying equipment. During the whole pesticide management, farmers use work clothing consisting of trousers, short-sleeve shirts and plastic boots. These three activities consist of the following series of characteristics: (a) Preparation: This activity includes opening the bottle containing the pure pesticide substance, mixing the solution of (different) pesticides and water, and loading the tank of the knapsack sprayer. Farmers in Vereda La Hoya prepare the pesticides in a 100-L or 200-L capacity container. The pesticide and the water (normally 80 L to obtain four applications of 20 L each) are mixed in this container with the aid of a wooden stick. During the mixing and the filling of the tank there are usually spills out of the container affecting different parts of the body including hands, arms, chest and legs; (b) Application: Once the knapsack sprayer is carried on the back, the pesticide application starts with the spraying process on the field. During this activity the farmers' body is exposed to the droplets emitted by the nozzles. In the study area the spraying is performed with hand pressure sprayers which are, on average, 9 years old [3,24]. They consist of a tank with a 20-L capacity, an injection and pressure system with an external piston pump and a pressure chamber with a capacity of 21 bar, a spraying pressure of 3 ± 0.3 bar and a pressure range between 1 and 14 bar. Farmers use two types of nozzles for pesticide application which differ in the amount of pesticide discharged: a high-discharge (HD) nozzle used during the first crop phases (sowing and emergence) and a low-discharge (LD) nozzle used during the rest of the crop phases (growth, flowering and pre-harvest). The discharges of the HD and LD nozzles measured in the study area were 1.88 ± 0.12 L/min (n = 24) measurements, and 1.26 ± 0.08 L/min (n = 24) respectively. Farmers purchase standard discharge nozzles of 1.05 ± 0.02 L/min (n = 8) and then modify the plastic and metal structures of the nozzles in order to obtain these discharges; (c) Cleaning: Once the application is finished, farmers clean the sprayer and the container by pouring clean water on all the accessories in a procedure repeated three times. This procedure is included in the booklet "Good Agricultural Practices" [49] which farmers use as a reference for the pesticide management. During this activity, there are numerous spills from the equipment and the accessories reaching the farmer's body. Previous studies have measured the dermal exposure and made an attempt to assess the health risk. These results are shown in Table 2.

Multi-Criteria Analysis
The multi-criteria analysis found that only DERM, DREAM, PHED, RISKOFDERM and PFAM can feasibly be applied in case studies in developing countries ( Figure 1, Table 3). COSHH was excluded from the evaluation as it does not consider important criteria relevant for case studies in developing countries such as target group, as it is focused on guidance for small and medium enterprises (SMEs), as it is only available in a website with a user's manual for only some specific industries; concerning outcome, its assessment is qualitative; regarding evaluated substances, it does not evaluate pesticides in farming systems; its dermal exposure descriptor only assesses the potential exposure; and concerning evaluated body parts, it does make a distinction between any body part. EASE was also excluded from the evaluation as it does not consider criteria such as target group, it is focused on industrialized processes, for guidance there is no user's manual with the model description; it provides a qualitative, its dermal exposure descriptor only evaluates the potential exposure and as to evaluated body parts, it only considers arms and forearms. STOFENMANAGER was also excluded from the evaluation as it does not comply with criteria such as target group, it is focused on industrial processes, the website does not show the algorithms or model calculations for guidance, its outcome assessment is qualitative and there is no information available regarding evaluated body parts.

Estimation of Dermal Exposures in the Study Areas
According to the previous results DERM, DREAM, PHED, RISKOFDERM, and PFAM were selected as the most appropriate models to be applied in the case study of Vereda La Hoya. The determinants included in each model are shown in Table 4 and the input data consider for each model is given in the Appendix Tables A1-A5. Even though the evaluated dermal exposure models provide insights into the level of exposure, their outcomes differ because of the model structure and the determinants included in each model structure (Table 5). Previous direct measurements in Vereda La Hoya found that dermal exposure to pesticides is very high ( Table 2) because of the inadequate work clothing, the modification of nozzles to increase the discharge, the inappropriate cleaning of the application equipment, the pesticide application against the wind direction and the use of pesticide with a high level of toxicity [24,31].
Actual dermal exposure values were also found higher than the reference values for human exposure for some pesticides like metamidophos [24,31]. Therefore, from the comparison of the models estimations and the type of determinants considered by each model, DERM, DREAM, and PFAM were found to be the most appropriate models. However, PHED might give an inaccurate estimation because the model determinants are relevant for farming systems in industrialized countries. Even though the model includes pesticide application scenarios which might be useful for developing countries, the model does not assess processes like pesticide emission and transfer, important processes within the mass transport quantification which should be included in the conceptual model for dermal exposure assessment, according to Schneider [50]. RISKOFDERM estimation might also be inaccurate because the model evaluated the exposure according to a percentage of body exposed and the quantitative estimation cannot be compared with reference values of human exposure as the pesticides have different levels of toxicity and the model only gives a qualitative assessment of "high" based on the quantitative estimation.
DERM is an appropriate model because of the specificity of the determinants for case studies in developing countries; however, the estimation accuracy might be underestimated because important determinants are not consider such as washing the equipment, task duration, wearing gloves, frequency and replacement of gloves, work clothing, personal hygiene and climate conditions. Therefore, this model has the potential to increase the accuracy of its estimations when these determinants are included in the assessment. DREAM was found to be an appropriate model as its estimation corroborates the dermal exposure assessment made in the location [24,31]; however, the estimation accuracy might be improved if there is a differentiation in the protection factor according to the different body parts and other determinants are considered such as climate conditions like wind speed and humidity. If these missing determinants are included the model scope will be wider for not only farming systems in industrialized and developing countries but other industrial processes. Finally, PFAM was found to give a quantitative assessment in terms of potential and actual exposure and how the protection factor influences the actual exposure. In addition it can assess the risk for each pesticide separately. However, it needs to be calibrated with direct measurements before it can be implemented in study areas with the same characteristics. Nevertheless, this model has the advantage of complying with all the required criteria in order to be implemented in case studies in developing countries.
These results are valid for potato farming systems and many other crop systems with similar characteristics in different regions in Latin America and might be also be valid for other regions worldwide with similar pesticide applications in Africa or Asia. However, the results are not valid for other sophisticated pesticide applications in crops in developing countries such as flowers, banana, coffee, sugar cane, rice, etc.
All the models for human exposure such as COSHH [37], DREAM [38], EASE [39], PHED [41], RISKOFDERM [42] and STOFENMANAGER [43] were developed after the conceptual model proposed by Schneider in 1999 [50,51]. Therefore, they were developed with similarities in the structure of the determinants. However, they are built for case studies in industrialized countries and there are uncertainties about their application in developing countries. For instance COSHH is specialized in SMEs in the UK; DREAM, in industrialized countries and farming systems in The Netherlands where tractors and motorized pesticide applications are used; EASE, in industrialized processes in the UK; PHED, in regulatory agencies and the pesticide industry in the USA and Canada; RISKOFDERM, for operational and technical staff in SMEs; and, STOFFENMANAGER, for Dutch companies. Some agricultural case studies in developing countries are characterized by manual pesticide applications with no regulations about the adequate pesticide use and no use of personal protection equipment. Only the DREAM model was applied in study areas in developing countries but the model has not been validated because of some issues regarding the reproducibility and accuracy of dermal exposure estimations [54]. Furthermore, this research found that when this model is applied in case studies in developing countries, most of the determinants do not cover the specific characteristics of these study areas. Based on DREAM, Blanco attempted to develop a model for farming systems in developing countries with DERM; however, this model has faced problems in the validation because of inappropriate procedures in the methodology [47]. However, despite this inaccuracies in the estimations of all the evaluated models, their structure has the potential to redefine and include other determinants which might be the origin to create a brand new model for dermal and human exposure assessment in farming systems in the developing world.

Conclusions
This research evaluated models for dermal exposure assessment focusing on case studies in developing countries. From the multi-criteria analysis and the type of determinants included in the models, DERM, DREAM, PHED, PFAM and RISKOFDERM were found as the most appropriate models to assess the dermal exposure in developing countries. Regarding the specificity to the farming systems in developing countries, DERM, DREAM and PFAM include determinants which are relevant for the system characteristics in the study area. However, all the five selected models are suitable to be modified in their structure in order to include parameters or determinants which might increase the accuracy of the estimations.  The evaluated models have the possibility to assess industrial and agricultural processes in industrialized and developing countries. However, DREAM was found to have a number and type of determinants that not only increase the accuracy of the estimation but they might serve as a basis to develop a new model including more determinants with higher specificity to study areas in farming systems in developing countries.
Previous studies found that because of the inadequate work clothing, the modification of nozzles to increase the discharge, the inappropriate cleaning of the application equipment, the pesticide application against the wind direction and the use of pesticides with a high level of toxicity, the dermal exposure was assessed as very high because both the potential and actual exposure for some pesticides were higher than the reference values for human exposure. Therefore, when comparing these results with the model estimations, it was found that DREAM and PFAM gave the most accurate estimations. However, it is important to take into account that DREAM is a semi-quantitative model easy to apply in the case studies. On the contrary, PFAM gives a quantitative estimation but the transfer coefficients must be determined in the field in order to calibrate the model.

Acknowledgments
This research was funded by the Swiss National Science Foundation. The first phases were developed in cooperation with the University of Zurich, University of Graz, Ludwig Maximilian University of Munich, ETH Zurich, University of Boyaca and National University of Colombia. The final phase concerning the analysis of results and final publicationwas developed in the Saint Thomas University in Tunja, Colombia.

Author Contributions
The results presented in this paper make part of the doctoral thesis "Human Exposure Assessment of Pesticide Use in Developing Countries" developed by Camilo Lesmes Fabian and supervised by Claudia R. Binder within the project "Life Cycle Human Exposure and Risk Assessment of Pesticide Application on Agricultural Products in Developing Countries". The first author had the original idea of comparing the models including the PFAM model. The manuscript was drafted and revised by the authors within the final document of the doctoral thesis.  There is evidence that during the whole pesticide application procedure, there is a leaking in the sprayer and the upper back is exposed. There is a potential exposure in 60% of the body surface. There is a potential exposure in 60% of the body surface as the region has a strong wind.    (a) <1% of task duration = 0 (b) <10% of task duration = 1 (c) 10-50% of task duration = 3 (d) ≥50% of task duration = 10 There is a potential emission during the whole process of the pesticide application. There is evidence that more than 50% of the body surface is exposed (b) ≥50% of body part = 10 The system covers these three processes. The deposition on clothing covers more than 50% of the body surface (b) 10-50% of body part = 3 6 Transfer to clothing and uncovered skin (P T.BP ) (a) <1% of task duration = 0 (b) <10% of task duration = 1 (c) 10-50% of task duration = 3 (d) ≥50% of task duration = 10 There is a transfer to clothing and uncovered skin during some of the pesticide management activities.
(c) 10-50% of task duration = 3      The pesticide solution is mixed with different chemicals in water. In the study area 96% of the farmers sprayed their pesticides (insecticides, fungicides, herbicides) with a backpack sprayer. The sprayers used in in the study area are between 8 and 11 years old. Therefore multiple repairments are made.
(b) Repair = 2   Farmers modify the nozzles and the two types of nozzles were considered. The protection factor given by work clothing and calculated for the application activity is high for legs, thighs, chest, abdomen and lower back (>90%) when both types of nozzles (HD and LD) are used. The protection factor is low in the arms (ranging from 51.8 to 88%) and also in the upper back (ranging from 74.8 to 82.6%).