A faecal exposure assessment of farm workers in Accra, Ghana: a cross sectional study

Accra is the capital city of Ghana with an estimated population of 1.9 million [10]. It is estimated that less than 6 % of Accra is connected to a sewerage system with the majority of the city reliant on onsite forms of sanitation like pit latrines and septic tanks [11]. There are over twenty wastewater treatment facilities in Accra, but only seven were reported to be functioning adequately as of 2010 [11]. There are seven major sites where wastewater in agriculture is used, with a total area of 160 ha [12]. Farmers at these sites apply water through watering cans using irrigation water sources that include drain water, channelled rivers and dugouts. The dugouts are man-made ponds used to store the various sources of water used for irrigation. The most commonly cultivated crops are salad vegetables including lettuce, cabbage, and spring onions. Three of the sites (Korle Bu, 12.3 ha, Dzorwulu, 8.2 ha, and Marine Drive, 3.5 ha) were selected for this study because of their size, the cultivation of salad crops normally consumed uncooked and the use of wastewater for irrigation.

Data analysis was done using STATA 12, the WHO QMRA model and @Risk 6 (Palisade Corporation, NY-USA). All E. coli concentrations were Log₁₀ transformed for calculations of medians and inter quartile range (IQR) for irrigation water samples and means, standard deviations and 95 % confidence intervals (CI) for soil samples. Unlike irrigation water, E. coli concentrations in soil were normally distributed after log transformation. The Mann–Whitney and Kruskal-Wallis tests were used to test the association of risk factors with irrigation water quality. One-way anova and two sample t-test were used to assess the effect of risk factors on soil contamination. Only factors that were significant at 10 % in the univariable analysis were included in a multiple regression model that was used to identify risk factors for soil contamination. Statistically significant differences between exposures and outcomes in the multiple regression model were measured at 5 % significance level using likelihood ratio test. Irrigation water was also reclassified as a binary variable (≤3 Log E. coli/100 ml and > 3 Log E. coli/100 ml) representing the water quality guidelines set by the WHO [20]. The proportion of time farmers worked unprotected for each farm activity was determined as a proportion of the time they work unprotected over the total time used to undertake the activity. The observed annual contact time for each type of contact (e.g. feet-to-soil), was calculated as the product of the total daily contact time for all observed farm activities, the number of days farmers work within a month and the months they work in a year. The expected dose of E. coli likely to be ingested due to hand-to-mouth events was estimated from a Poisson distribution [21] and Monte Carlo simulation using the soil quality from this study and the range of soil quantity (10 - 100 mg/d) assumed to be accidentally ingested by farmers. Regarding the norovirus infection risk to farmers, only the median and 95th percentile risk per person per year were reported after the Monte Carlo simulation.

During the survey only 7 % of the 160 irrigation water samples and 9 % of the 163 soil samples were found to be free from E. coli. Overall, the median concentration of irrigation water was 5.6 Log E. coli/100 ml, while the mean concentration of soil was 2.3 Log E. coli/g.

The use of poultry manure, the quality of irrigation water and time since soil was last irrigated were all associated with increased levels of soil contamination in the univariable analysis (Table 2). In the multivariable analysis the effect of irrigation water and seasonality remained strongly associated with the levels of soil contamination after controlling for the use of chicken manure. There was interaction between seasonality and irrigation water quality which was significant in the multivariable analysis (p = 0.02, 95 % CI = -0.36, -0.04) and resulted in higher levels of soil contamination in the dry season than in the rainy season (Table 2). For irrigation water, univariable analysis showed significant differences in the E. coli quality among the different types of irrigation water with drain water being the most contaminated (p < 0.001, Table 2).

Of the 80 farmers all but 4 were male with an average age of 40 years (range 22 – 72). Agriculture formed the main source of income for the large majority of farmers (80 %), while over 70 % of farmers were literate. There were no toilet facilities found on the sites and 73 % of farmers reported that they practiced open defecation when working at their fields (Table 3). The majority (77 %) of farmers ate their food in their fields and mostly consumed it cold after keeping the food over a period of time.

A total of 10 farming activities were observed during the 3 h, though the majority (79 %) of time was spent on 5 key activities, with irrigating the most common (33 % of total time) (Fig. 1). During the observation period almost all farmers (97 %) were observed to have hand-to-soil contact and 89 % of farmers were found to work bare-foot in their field for any of the activities (Table 4). In addition over 90 % of farmers involved in irrigation had their hands and feet exposed to irrigation water. The number of hand-to-mouth contacts per farmer was highest during “forking” with an average of 4 events/3 h, but ranged from 1 to 12 events (Fig. 2). For all activities observed within the 3 h, 86 % of farmers experienced hand-to-soil contact for an average time of 100 min while 93 % worked bare-foot for 145 min (81 % of total time). In addition 63 % of farmers had both feet and hands exposed to irrigation water for at least 88 min (49 % of total time).

Farmers reported to work an average of 7.1 h per day, 28.2 days per month and 11.8 months per year on their farm (Table 3). These then translate to an average annual time spent working in the field of 337 days, or 2,393 hours. In terms of farming activities, farmers spend a median of 720 hours and a maximum of 2,880 hours per year (120 days) irrigating, though irrigation was done on average 324 days (27 days per month) per year (Table 5). Of the five major farm activities, irrigation recorded the highest feet-to-soil contact at 88 min/3 h (Fig. 3) which translated into annual median contact of 1,278 h/y (Table 4). The observed median feet-to-soil contact for farm activities was 2,002 h/y (maximum of 2,556 h/y, or 107 d/y). “Forking” had the highest hand-to-soil contact time (53 %, 152 h/y) while no hand-to-soil contact was observed during irrigation. Every day, farmers had a median of 10 hand-to-mouth events (3,181/y).

Although all farmers reported to wash their hands before eating, only one (6 %) of those observed to wash their hands before eating was found to have used soap (Table 3). The use of poultry manure was common, though significantly higher among farmers in the dry season than in the rainy season (Table 3). Only a few farmers (6 %) were seen using gloves, or masks (1 %), though a slightly higher proportion of farmers reported that they used gloves and masks while preparing chemicals for spraying crops (9 % vs. 6 %). Irrigation cessation was not reported to be practiced, though the average irrigation frequency went down to every other day (56 h) during the rainy season (Table 3).

From the Poisson distribution, farmers could ingest a minimum of 2 E. coli/d and a maximum of 18 E. coli/d if they ingested soil of average contamination 2.3 Log E. coli/g (Table 6). For the maximum soil contamination of 4.1 Log E. coli/g, farmers could ingest between 126 E. coli/d and 1,259 E. coli/d. Farmers were likely to spend an average contact time of 103 d/y and a maximum of 132 d/y in direct contact in the fields for all activities, assuming 100 % feet contact to soil. The median norovirus infection risk for farmers exposed to soil of quality 2.3 Log E. coli/g and ingesting between 10 – 100 mg/day soil was estimated to be 8.5 × 10^-3 pppy and 3.4 × 10^-3 pppy for exposures of 337 days and 132 days respectively (Table 7). When exposed to soil of the highest contamination (4.1 Log E. coli/g), the median norovirus infection risk for farmers increased to 0.42 pppy; and 0.19 pppy for reported and observed exposures of 337 days and 132 days respectively (Table 7).

The study found that irrigation water sources used for vegetable cultivation were highly contaminated, with 84 % of water samples exceeding the WHO water quality standard of 3 log E. coli/100 ml for unrestricted irrigation [20]. The high concentrations of E. coli found in irrigation water were similar to those found previously in Ghana [22, 23], though significantly lower than those found in India and Pakistan where farmers were found to use untreated wastewater [24]. Unlike in Pakistan where farmers used raw sewage from a wastewater treatment plant, wastewater used by farmers in Accra was diluted by rainwater, or other sources of storm water. Although water quality was the main factor affecting the presence and concentrations of E. coli in soil, the use of poultry manure further contributed to increased levels of E. coli in soil. The concentrations of E. coli found in soil were lower (2.3 Log E. coli/g vs. 3.0 Log faecal coliform/g) than those found previously in Ghana [22], and this could be due to differences in the microbial quality of irrigation water, or the frequency of manure application to soil between the two studies. E. coli is an indicator organism for faecal contamination and the high concentrations found in irrigation water and farm soil are likely to indicate the presence of a variety of pathogens. Nevertheless, only few studies have enumerated the actual concentrations of pathogens, including viruses in wastewater used for irrigation due to high cost and poor viral detection efficiencies [25].

The WHO QMRA model calculates permissible disease, or infection risk for farmers using wastewater on the accidental ingestion of wastewater contaminated soil during agricultural activities. However there is little evidence to support the assumption that the key risk to farmers is through the soil route, though one study has reported that agriculturalist and archaeologists have higher soil interaction than other workers [26] and that all members of an exposed population will involuntarily ingest at least small quantities of soil adhering to the skin of fingers because of hand-to-mouth activity [27]. This study found the highest concentrations of E. coli in irrigation water and significantly lower concentrations in soil. Farmers however, spent a higher proportion of their time in contact with soil (>80 %) than with irrigation water (49 %). In this study farmers were observed to have direct hand to soil and hand to irrigation water contacts, though hand to mouth events were only observed during soil related activities and not during irrigation. These findings therefore support the WHO QMRA model approach that is based on the accidental ingestion of soil.

The use of watering cans could possibly prevent, or limit farmers’ direct hand to mouth contact of irrigation water since farmers rarely put the watering cans down during irrigation. On the other hand, farmers could ingest some wastewater when engaged in other forms of irrigation application such as spray or sprinkler irrigation. Farmers’ prolonged exposure to wastewater could be more significant when investigating pathogens, or chemical risks that occur via dermal contact rather than through ingestion, especially when exposure to wastewater has been identified as a major risk factor for skin disease in Vietnam [28].

In terms of soil ingestion, the study found that farmers were likely to ingest between 2 E. coli and 1200 E. coli/d. This study did not isolate specific strains of E. coli or pathogens and was therefore unable to determine whether the exposure rates observed are likely to result in an adverse health effect since big differences exist in the infective dose for different E. coli strains. Again, the presence of E. coli only indicates the presence of faecal pollution and does not necessarily guarantee the presence of pathogens that can cause diseases to humans.

All major farm activities were found to expose farmers to faecal contamination, though irrigation, forking and weeding were regarded as the key risk activities. The large amount of time spent by farmers on irrigation comes as a result of the manual method of irrigation application and the long distances farmers walk to access irrigation water. In this study, farmers spent about 80 % of their total working time accessing irrigation water or irrigating, and was higher than previous estimates (40 to 70 %) in Accra and Kumasi [12, 29]. Only 7 % of farmers in this study were seen to wear boots, and often only for short periods while irrigating, as was shown by studies in Kenya, Pakistan, and Côte d’Ivoire where between 5 and 19 % of farmers reported to wear boots, often citing discomfort, heat and the muddy fields as reasons why they did not wear footwear [8, 9, 30]. In India and Pakistan hookworm infection was found to be the main infection associated with the use of wastewater by farmers and the lack of use of footwear was cited as one of the main risk factors [8, 31]. In Ghana, stool surveys among wastewater farmers have not been conducted and as a result it is unknown whether irrigation practices and the lack of footwear affect hookworm prevalence.

“Forking” and weeding were found to be the major farming activities associated with farmers’ risk of accidental ingestion of soil. This was due to the high frequency of hand-to-mouth events associated with these two activities, which are often undertaken simultaneously using hand-held weeding knives and the bare hands. Farmers’ hands become contaminated as they remove weeds, stones and other waste materials, and this coupled with frequent wiping of sweat from the face due to the heat and the strenuous activities and the consumption of food make these high-risk activities. The risk of faecal pathogen transmission due to the consumption of food with contaminated hand is, however, not limited to forking and weeding but is common to other farm activities including irrigation. In addition, there is the likelihood of the soil being attached to the farmers’ hands and feet for some time after the soil related activities, and this could also present some health risks to the farmers and those that they come into contact with. The use of chicken manure was reported by between 60 and 99 % of farmers in this study and in other studies by between 70 and 98 % of farmers in Ghana [32]. The high concentrations of E. coli in manure [22, 32] and the fact that the manure is often applied without the use of protective clothing makes this another key health risk that is not included in the QMRA assessment and would not only apply to wastewater farmers but also to regular farmers.

This study found that the use of wastewater in Accra (and potentially other places with similar settings) exceeded the WHO permissible norovirus infection risk (1.4 × 10^-3 pppy) corresponding to a DALY burden of 10^-6 pppy. Similar findings were reported by Mara and Sleigh [33] and Mara et al., [6], where wastewater farmers’ norovirus infection risk exceeded guideline thresholds by at least one order of magnitude if they ingested 1–10 mg, or 10–100 mg of wastewater saturated soil for 100 and 300 days respectively. An earlier study in Accra, also found farmers’ risk of rotavirus infection (7.6 × 10^-2) to exceed guideline value for rotavirus diarrhoea in developing countries (7.7 × 10^-4) by 2 orders of magnitude, after ingesting 10–100 mg of soil for 150 days [34]. The fact that some studies assume a fully-saturated wastewater soil and substitute wastewater quality for soil quality could however, lead to bias results, as other contaminants such as wild animals and birds have been identified to contribute to soil quality [35].

In the current study, farmers’ risk was found to diminish by at least 50 % if an actual observed exposure time to water and soil (132 days), or contact in the field was used. Although the use of self-reported time could lead to overestimation of farmers’ risk; this influence might be more significant when assessing risk transmitted via dermal contact and not through oral ingestion. A better approach to estimate farmers’ risk due to soil ingestion would be the use of actual hand-to-mouth contact (manuscript in preparation) since these events depend more on the type of farm activity performed and not necessarily on how much time farmers spend in the field.

Currently the WHO QMRA model does not consider farmers’ risk via dermal contacts (e.g. hookworm infection). This exposure route could be particularly important in the QMRA model as almost all farmers in this study were found to be working bare-feet for most of the time, though this is less relevant for oral ingestion. For this type of transmission route the use of the direct observed contact time would be more appropriate since the self-reported time does not necessarily reflect the actual time farmers spent in the field or were engaged in risky activities that expose them to faecal pathogens. Further studies in the form of repeated observations or longer observations over the course of the year would be required to confirm farmers actual contact time to faecal pathogens and to better understand their risk behaviours and practices. This is particularly necessary as direct exposure frequency to faecal pathogens in this study was only based on a single 3 h observation per farmer, and also excluded contact to faecal matter during manure application.

The maximum permissible additional disease risk has been under discussion with some arguing that it is too strict for wastewater use in agriculture [19]. This study showed that farmers’ occupational risk was within acceptable limits for a DALY burden of 10^-4 pppy but not for the current guideline of DALY burden of 10^-6 pppy. Only the risk corresponding to the highest soil contamination for an exposure of 337 days exceeded this tolerable risk. A DALY burden of 10^-4 pppy also sets the tolerable number of infections associated with wastewater exposure, though it does not by itself determine the likelihood of pathogen infections.. One of the reasons for the use of a relaxed DALY of either 10^-5 pppy or 10^-4 pppy was that the resulting norovirus/rotavirus disease risk would still be lower than the actual global incidence of diarrhoeal disease of 0.1 – 1 pppy in both developed and low and middle-income countries [36]. In addition, it would result in a reduction in the cost required for wastewater treatment; and hence the extra money saved could be used for other risk reduction interventions.

Although high, the estimated risk from this study should be interpreted with caution. First, the risk arising from the mean soil quality was found to safe, though soil samples (51 %) with quality just above the mean (2.4 Log E. coli/g) resulted in a risk higher than the guidelines limit. Even with this quality, farmers’ risk would still be within the acceptable limits, or would be marginally safe if exposure (ingestion) to contaminated soil was at most 300 days (9.0 × 10^-3 pppy).

There were limitations of the model that was used to estimate the risk. The model used published ratios between E. coli and norovirus and not necessarily the actual concentrations of norovirus. The use of these ratios often assume a linear relationship between the indicator organism and the pathogen and also ignore other factors such as seasonality, transport characteristics of microbes and other environmental factors which could influence this correlation. There is also inadequate evidence to support the widely used ratio of 1:10⁵ E. coli/faecal coliform to virus relationship, which was based on a study in northeast Brazil [37]. A recent study in Ghana found an average of one norovirus GII to 10^3.2 E. coli or 10^4.8 thermotolerant coliforms, from its quantifiable irrigation water samples, which suggest that the NV-GII to E. coli ratio is much lower than the widely used ratio of 1:10⁵ [23]. In the current study, the ratio of means between norovirus and E. coli was estimated as 1:10^1.7 (1.9 × 10¹ genome copies/mL vs. 8.9 × 10² E. coli/mL, N = 67) from irrigation water samples analysed for both E. coli and norovirus, which is also much lower than the common ratios used in recent publications. The above observation also means that the estimated number of NV-GII in the current study or the previous study in Ghana would be higher than if it were estimated from the higher NV-GII to E. coli ratio of 1:10⁵.

The other limitation is that the study did not assess for helminths and protozoans and hence the model could underestimate farmers’ risk, though the risk associated with viruses is generally considered high enough to adequately protect farmers from bacterial and protozoan infections. The use of soil quality in the risk model instead of water quality as used in some other studies is, however, considered as the “closest” and a better estimate of farmers’ risk due to faecal-oral ingestion. A third limitation is that the study was unable to estimate the actual mass or quantity of soil likely to be ingested by farmers and therefore still relied on the range normally for QMRA estimations from the literature. A further refinement of QMRA input data would be to estimate the total mass of soil ingested but also classify the exposure mass of soil based on the different types of farming activities such as transplanting and weeding.

An updated version of the WHO QMRA model should incorporate actual hand-to-mouth events in the model since these models often deal with ingestion of contaminated products such as soil, irrigation water and produce. In terms of pathogen reductions, farmers’ risk of 0.42 pppy means that reducing the contamination of irrigation water by two to three log units per 100 ml irrigation water or 100 g soil (assuming a fully saturated soil) will keep farmers occupational health risk within acceptable levels for a DALY burden of 10^-6 pppy. A significant part of soil contamination was attributed to the use of chicken manure, and hence adequate treatment of the manure before application is also recommended to reduce farmers’ risk, in particular those caused by zoonotic pathogens, such as Campylobacter; further, manure safety management should form part of the WHO guidelines.

Although these reductions in microbial contamination can be achieved by simple wastewater treatment such as the use of the three-tank or three-pond system which is operated as a sequential batch-fed process [33]; in the short term, wastewater treatment seems an unlikely intervention as farmers are unable to invest in wastewater treatment due to insecure land tenure system, and the high costs. Farmers are also unlikely to allow their irrigation water to settle for 6 days to reduce thermotolerant coliforms levels as recommended by Keraita et al [38] due to the long waiting periods and the fact that farmers seem more concerned about keeping their produce fresh for higher yields and profits. Instead, local authorities and other stakeholders should collaborate with farmers by providing credit or loan schemes and also increase land security to farmers who adhere to agreed and prescribed safe practices. This in turn could motivate farmers to invest more in on-farm risk reduction measures such as on-farm sedimentation ponds, and also adopt other good agriculture practices as well as personal and environmental hygienic practices that could reduce both occupational and consumer risk.