Wastewater use for vegetable production is widespread across the cities of many developing countries. Studies on the microbial health risks associated with the practice have largely depended on faecal indicator organisms with potential underestimation or overestimation of the microbial health risks and disease burdens. This study assessed the Escherichia coli O157:H7 infection risk and diarrhoeal disease burden measured in disability-adjusted life years (DALYs) associated with the consumption of wastewater-irrigated lettuce in Kumasi, Ghana using data on E. coli O157:H7 in ready-to-harvest, wastewater-irrigated lettuce. Two exposure scenarios – best case and worst case – associated with a single consumption of wastewater-irrigated lettuce were assessed. The assessment revealed wastewater-irrigated lettuce is contributing to the transmission of E. coli O157:H7 in Kumasi, Ghana. The mean E. coli O157:H7 infection risk and DALYs in the wet and dry seasons, irrespective of the exposure scenario, were above the World Health Organization tolerable daily infection risk of 2.7 × 10−7 per person per day and 10−6 DALYs per person per year. It is recommended that legislation with clear monitoring indicators and penalties is implemented to ensure that farmers and food sellers fully implement risk mitigating measures.

## INTRODUCTION

Globally, diarrhoeal disease and its complications account for billions of illnesses and millions of deaths annually. Developing countries, particularly those in Africa, have the highest incidence as well as burden of diarrhoeal disease (WHO 2013). In Africa, diarrhoeal disease is the third ranked cause of mortality after HIV/AIDS and lower respiratory tract infections, and led to 800,000 deaths and 55 million lost life years (expressed as diarrhoeal disease burden and measured in disability-adjusted life years (DALYs)) on the continent in 2011 (WHO 2013). A plethora of risk factors have been implicated for the occurrence of diarrhoeal disease in developing countries. These risk factors include, but are not limited to, unimproved water and sanitation facilities, poor hygiene practices and unhygienic food handling. In many cities in developing countries with rapidly expanding populations, wastewater use in vegetable farming is contributing to the mix of diarrhoeal disease risk factors. In these cities, a significant proportion of the vegetables consumed are irrigated with raw or partially diluted wastewater. Farmers using wastewater for irrigation rarely adopt risk-mitigating measures and thus expose themselves and their families, as well as consumers of the wastewater-irrigated vegetables, to pathogens causing diarrhoeal disease (Seidu & Drechsel 2010). However, in most developing countries the link between wastewater irrigation and the transmission of diarrhoea-causing pathogenic organisms is poorly understood. This is partly due to inadequate resources to collect and analyse wastewater and wastewater-irrigated vegetables for pathogenic organisms and a general weakness in existing epidemiological surveillance systems to assess and relate the cause(s) of disease outbreaks. Consequently, the actual contribution of diarrhoeal disease attributed to wastewater irrigation is less well characterized.

In urban Ghana, a significant proportion of vegetable salad served by restaurants and street-food sellers originates from wastewater-fed fields (Seidu & Drechsel 2010). It is estimated that daily, about 800,000 people across the cities of the country consume wastewater-irrigated lettuce salad (Seidu & Drechsel 2010). This number is likely to increase in synch with rapid urbanization, which is now 3% in Ghana (Ghana Statistical Service 2010). Studies on potential disease-causing organisms in wastewater and wastewater-irrigated vegetables in Ghana have focused mainly on indicator organisms, particularly faecal coliforms (Keraita et al. 2007; Amoah et al. 2007), and have not addressed the pathogens directly. These faecal indicator organisms have formed the basis in the construction of models to assess the diarrhoeal disease risk and burden associated with wastewater irrigation (Seidu et al. 2008; Seidu & Drechsel 2010). The use of faecal indicator organisms in risk assessment can lead to an underestimation or overestimation of diarrhoeal disease risk and burdens in the absence of data on pathogenic organisms.

In 2011, the Research Council of Norway (RCN) funded a study to, among other things, assess the diarrhoeal disease risk and burden associated with wastewater use in urban Ghana. As part of the RCN study, wastewater-irrigated vegetables were analyzed for Escherichia coli O157:H7 in wastewater-irrigated vegetables. E. coli O157:H7 diarrhoeal disease outbreaks have been associated with the consumption of vegetables (Jay et al. 2004). Following the outbreak of E. coli O157:H7 in the United States in 1982 from fast-food chains (Riley et al. 1983), additional outbreaks have been related to contaminated vegetables (Michino et al. 1999; Hilborn et al. 1999). From 1996 to 1999, an estimated 22 outbreaks of E. coli O157:H7 reported in California were attributed to the contamination of lettuce and spinach in watersheds due to heavy and increased river flow rates (Cooley et al. 2007). Outbreaks associated with the consumption of E. coli O157:H7-contaminated vegetables have also been reported in Japan (Michino et al. 1999) and Sweden (Söderström et al. 2008). In the United States, E. coli O157:H7 causes about 73,000 infections and 60 deaths annually (Mead et al. 1999), and increased morbidity and mortality has raised public concern due to the fact that infection develops into haemorrhagic colitis and haemolytic uremic syndrome (HUS) which is more pronounced in children, the elderly and the immunocompromised (Abu-Ali et al. 2010). E. coli O157:H7 diarrhoeal disease outbreaks have also been reported in African countries such as Kenya (Sang et al. 1996), Cameroon (Germani et al. 1998), Ivory Coast (Dadie et al. 2000) and Nigeria (Olorunshola et al. 2000). The objective of this study was to assess the E. coli O157:H7 infection risk and disease burden associated with the consumption of wastewater-irrigated lettuce in Kumasi, Ghana, where vegetable farmers mainly depend on diluted wastewater from storm water drains and streams as a source of irrigation water (Seidu et al. 2013).

## METHODS

### Farm sites, sampling and analysis of E. coli O157:H7

Kumasi is the second largest city in Ghana with a population of over two million and an annual growth rate of 2.7% (Ghana Statistical Service 2010). The city has a number of hotels, restaurants and street-food vendors that serve salad prepared with wastewater-fed vegetables as a side-dish to consumers. Within the city two cross-sectional surveys were undertaken in all the diluted wastewater-irrigated farms (n = 8) in the wet (August–November, 2012) and dry (December 2012–May 2013) seasons to assess the level of E. coli O157:H7 in ready-to-harvest lettuce.

In each survey, lettuce heads were aseptically collected from beds in the farms with sterile knives into sterile plastic bags (Stomacher® laboratory system, Seward, UK) and sealed. From each bed, the lettuces were collected from the far ends and in the middle to represent triplicate samples. The samples were transported in cooling boxes to the laboratory for analysis. A total of 107 and 89 ready-to-harvest triplicate lettuce heads were collected from the farms in the wet and dry seasons, respectively.

Lettuce samples were weighed, aseptically cut and homogenized with a blender. One millilitre of the blended lettuce was transferred into 9 mL of 0.85% isotonic saline and poured onto Eosine Methylene Blue agar Petri dish plates and incubated at 44 °C for 24 h. Plates showing growth of E. coli colonies (blue black with metallic sheen appearance) were sub-cultured onto Sorbitol MacConkey agar plates and incubated at 44 °C for 24 h. Colonies that demonstrated a colourless appearance were isolated and re-cultured for pure colonies (Baron et al. 1994). Pure colonies were then serologically confirmed with E. coli prolexTM latex agglutination test kit (Oxoid, UK). One drop of E. coli O157:H7 antisera was dispensed onto a reaction card and a loop full of the pure colony of the test organism was added. This was then carefully emulsified into a smooth suspension. The card was carefully rocked in a circular motion. If agglutination occurred within 1 minute then the test organisms were considered positive for E. coli O157:H7. A similar reaction was carried out as a control using normal saline instead of the E. coli O157:H7 antisera. Quantifications were based on colony forming units (CFU) expressed per 100 g of lettuce.

### Assessment of E. coli O157:H7 infection risk and DALYs

An integrated quantitative microbial risk assessment and DALYs approach was used to assess the E. coli O157:H7 infection risk, diarrhoeal disease morbidity and mortality and burden associated with the consumption of wastewater-irrigated lettuce salad. This was based on the following steps.

#### Hazard identification

The main hazard was E. coli O157:H7 in the ready-to-harvest, wastewater-irrigated lettuce. The E. coli serotype O157:H7 is regarded as a sub-group of enterohaemorrhagic verocytotoxin-producing E. coli associated with bloody diarrhoea (haemorrhagic colitis) and HUS (Su & Brandt 1995).

#### Exposure assessment

Occurrence of E. coli O157:H7 in ready-to-harvest lettuce: To account for variability and uncertainty in the level of E. coli O157:H7 in the lettuce, probability distribution functions (PDFs) were fitted to the wet and dry seasons’ E. coli O157:H7 data. The PDFs were fitted using the maximum likelihood estimation (MLE) method (Haas et al. 1999). The best fit PDF was obtained based on Akaike Information Criteria (AIC). Bootstrapping methods were used to obtain the uncertainties around the parameters of the PDFs.

Amount of lettuce consumed and exposure scenarios: The amount of lettuce consumed per single serving was obtained from a survey involving 270 street-food sellers and restaurants in Kumasi. Lettuce in ready-to-eat salad was weighed (Mettler balance, P-1200 (No. 903316)), PDFs were fitted to the weighed lettuce data, and the best fit PDF and bootstrapped uncertainty parameters were obtained using the MLE method (Haas et al. 1999). Two main exposure scenarios were considered in the risk assessment: (i) a worst-case scenario where there is no reduction in the level of E. coli O157:H7 in the wastewater-irrigated lettuce salad prior to consumption; and (ii) a best-case scenario where there is 0.5–2 log reduction of E. coli O157:H7 in wastewater-irrigated lettuce prior to consumption. The first scenario reflects the situation for most of the street foods where poor food safety practices prevail. The latter scenario is representative for high-class restaurants in the city where treatment of wastewater-irrigated lettuce using disinfectants (e.g., chlorine, salt solution, vinegar) is undertaken during salad preparation (Amoah et al. 2007).

The dose (D) of E. coli O157:H7 was obtained from the equation
1
where C is the concentration of E. coli O157:H7 in wastewater-irrigated lettuce salad in a given season (wet or dry); r is the reduction of E. coli O157:H7 on the salad (i.e., 0 for the worst-case scenario and log 0.5–2 for the best-case scenario); and A is the amount of lettuce salad consumed.

#### Dose–response (DR) assessment

It has been shown from outbreak studies and feeding trials that the beta-Poisson DR model best describes the dose(s) of E. coli O157:H7 ingested and the probability of infection (Haas et al. 2000; Teunis et al. 2008). The approximate form of the beta-Poisson DR model is given as (Furumoto & Mickey 1967)
2
where Pinf is the probability of infection; D is the dose of E. coli O157:H7 ingested via lettuce salad; and and are the DR parameters. In this study, the beta-Poisson DR model parameters derived for E. coli O157:H7 by Teunis et al. (2008) were used. To account for variability and uncertainty, a triangular probability distribution function was fitted to the α and β parameters of the Teunis et al. (2008) DR model.

#### Risk characterization

The E. coli O157:H7 infection resulting from a single consumption event of wastewater-irrigated lettuce was assessed. To account for variability and uncertainty separately in the risk estimates, a second-order Monte Carlo simulation was implemented. A description of the model parameters in the second-order Monte Carlo Simulation are presented in Table 1. The risk estimates accounted for both the wet and dry seasons expressed as the probability of infection per consumer per single consumption of lettuce salad. The second-order Monte Carlo risk models were constructed and run in R statistical software version 2.15.0 (www.r-project.org).

Table 1

Summary of the parameters used in the risk assessment and DALYs estimations

Model parameter Unit Probability distribution or assumption Description
Dose of E. coli O157:H7 ingested (D) CFU/g Cwet season = gamma (shape = 0.052; rate = 0.00019)1 U & V
E. coli O157:H7 in irrigated lettuce (C)   Log g Cdry season = gamma (shape = 0.013; rate = 0.00025)1
Reduction in E. coli O157:H7 (r)    Uniform (0.5, 2)2 U & V
Amount of lettuce consumed per serving (A)    Lognormal (3.67; 0.19)1
E. coli O157:H7 DR model  Parameters for the beta-Poisson model 3U & V
α = Triang (2.62 × 10−4; 0.373; 398.9)
β = Triang (0.056; 39.71; 3.96 × 104
Probability of infection ( Beta-Poisson DR model U & V
Probability of illness (Pill) with a certain symptom   =
= × 0.53*
= × 0.47*
= × 0.07*
= × 0.1#
Probability of fatality (  = × 0.0003#
= × 0.17#
Severity weights for illness and fatality   = #
= #
= #
= 1#
Duration of illness Days  = days*
= 12.2 days *
= #
Model parameter Unit Probability distribution or assumption Description
Dose of E. coli O157:H7 ingested (D) CFU/g Cwet season = gamma (shape = 0.052; rate = 0.00019)1 U & V
E. coli O157:H7 in irrigated lettuce (C)   Log g Cdry season = gamma (shape = 0.013; rate = 0.00025)1
Reduction in E. coli O157:H7 (r)    Uniform (0.5, 2)2 U & V
Amount of lettuce consumed per serving (A)    Lognormal (3.67; 0.19)1
E. coli O157:H7 DR model  Parameters for the beta-Poisson model 3U & V
α = Triang (2.62 × 10−4; 0.373; 398.9)
β = Triang (0.056; 39.71; 3.96 × 104
Probability of infection ( Beta-Poisson DR model U & V
Probability of illness (Pill) with a certain symptom   =
= × 0.53*
= × 0.47*
= × 0.07*
= × 0.1#
Probability of fatality (  = × 0.0003#
= × 0.17#
Severity weights for illness and fatality   = #
= #
= #
= 1#
Duration of illness Days  = days*
= 12.2 days *
= #

U, uncertainty; V, variability; C, calculated; gamma, gamma probability distribution function; Uniform, uniform probability distribution function; Lognormal, probability distribution function.

#### Diarrhoeal disease burden

Diarrhoeal illness and mortality: Infections with E. coli O157:H7 can result in a wide range of symptoms and secondary complications. Given a probability of E. coli O157:H7 infection (Pinf), the probability of illness (morbidity) can be derived as
3
where the first part of the equation is Pinf and the second Psymp is the probability of an illness with a certain symptom or secondary complication resulting from the infection given a dose (D) of E. coli O157:H7 with DR parameters and .
Equation 3 can be expanded to obtain the probability of mortality () as follows:
4
where is the probability of fatality given the occurrence of illness () with a given symptom resulting from the consumption of a dose (D) of E. coli O157:H7.

To assess the illness () and mortality () associated with the consumption of E. coli O157:H7-contaminated lettuce salad, the E. coli O157:H7 disease transition model proposed by Havelaar & Melse (2003), with modifications, was followed. It was assumed that 31% of the infected consumers were asymptomatic (Su & Brandt 1995) while the rest were symptomatic. Of the symptomatic cases, 53% had watery diarrhoea () and 47% bloody diarrhoea (haemorrhagic colitis) (). A case–fatality ratio of 0.03% was applied to cases with bloody diarrhoea (Havelaar & Melse 2003) while those cases with mild watery diarrhoea were expected to fully recover. Among the bloody diarrhoea cases, 7% later had HUS (Su & Brandt 1995) and 10% of the HUS cases developed an end-stage renal disease (ESRD) (Havelaar & Melse 2003). The mortality associated with the acute phase of HUS was assumed to be approximately 17% (Fitzpatrick et al. 1991; Havelaar & Melse 2003). The use of this high mortality rate was to account for the relatively poor health services in Kumasi.

Disability-adjusted life years: DALYs estimates were made to assess the burden of E. coli O157:H7 related diarrhoeal disease and mortality estimates derived from the previous section. DALYs were estimated using the equation of Murray (1994) as follows:
5
where YLL is years of life lost due to mortality and YLD is years lived with a disability with weighing factor between 0 and 1 for the severity of the disability or disease. To derive the YLLs and YLDs for the E. coli O157:H7-related diarrhoeal disease and mortality, the following equations were used:
6

7
where K = age weighting modulation factor; C = constant; r = discount rate; a = age of death; ß = parameter from the age weighting function; L = standard expectation of life at age a. In this study, the age cohort 25–29 was used in the DALYs estimation. This cohort represents the highest proportion of consumers of wastewater-fed vegetables in urban Ghana (Seidu & Drechsel 2010). The severity weights for watery and bloody diarrhoea were taken as 0.067 and 0.39, respectively; and that of HUS was 0.93 (Havelaar & Melse 2003). Deaths resulting from diarrhoea, HUS and ESRD have a severity index of 1. A standard life expectancy of 63 years was used (Ghana Statistical Service 2010). Irrespective of the age group, mild diarrhoea and bloody diarrhoea were assumed to last 7 days and 12.2 days, respectively (Su & Brandt 1995). The HUS cases were estimated to last 21 days (Havelaar & Melse 2003). The DALY models were constructed and run in Microsoft Excel®.

## RESULTS

### Occurrence of E. coli O157:H7 in wastewater-fed lettuce and amount of lettuce salad consumed

Figure 1 shows the frequency distribution of E. coli O157:H7 in the wet and dry seasons. The occurrence of E. coli O157:H7 was low in the wastewater-fed lettuce, particularly in the dry season. The likelihood of a lettuce sample being positive for E. coli O15:H7 was 28% (30 out of 107 samples) and 5% (4 out of 89 samples) in the wet and dry seasons, respectively.

Figure 1

Frequency distribution of E. coli O157:H7 in lettuce across the farm sites in the wet (a) and dry (b) seasons.

Figure 1

Frequency distribution of E. coli O157:H7 in lettuce across the farm sites in the wet (a) and dry (b) seasons.

Among the PDFs fitted to the E. coli O157:H7 occurrence data, the gamma PDF was the best fit for both the wet season (shape = 0.052; rate = 0.00019; AIC = 404.67) and dry season (shape = 0.013; rate = 0.00025; AIC = 363). Figure 2 shows the fitted PDFs and the bootstrapped parameters of the gamma PDF for the wet season (shape = 0.052 (95% CI: 2.9 × 10−2–0.128); rate = 0.00019 (95% CI: 8.76 × 10−5–8.9 × 10−4)) and dry season (shape = 0.013 (95% CI: 2.98 × 10−2–1.3 × 10−1); rate = 0.00025 (95% CI: 9.1 × 10−5–8.9 × 10−4)).

Figure 2

Level of E. coli O157:H7 in wastewater-irrigated lettuce in the wet (a) and dry (b) seasons fitted with the gamma probability distribution function and the bootstrapped parameters of the distribution.

Figure 2

Level of E. coli O157:H7 in wastewater-irrigated lettuce in the wet (a) and dry (b) seasons fitted with the gamma probability distribution function and the bootstrapped parameters of the distribution.

Figure 3 shows the amount of lettuce salad consumed per serving as fitted with the lognormal PDF (Inμ = 3.67; Inσ = 0.19) and bootstrapped parameters (Inμ = 3.67 (95% CI: 3.64–3.69); Inσ = 0.19 (95% CI: 0.18–0.22)) of the PDF.

Figure 3

Lognormal probability distribution fitted to the amount of lettuce salad consumed and the bootstrapped parameters of the distribution.

Figure 3

Lognormal probability distribution fitted to the amount of lettuce salad consumed and the bootstrapped parameters of the distribution.

### E. coli O157:H7 infection risk, diarrhoeal disease and DALYs

Table 2 presents a summary of the estimated E. coli O157:H7 infection risk, diarrhoeal disease morbidity and mortality and disability adjusted life years (DALYs) associated with the consumption of wastewater-fed lettuce in Kumasi, Ghana. Consumers of wastewater-irrigated lettuce salad were more likely to be infected with E. coli O157:H7 in the wet season than in the dry season. In the wet season, the one-time E. coli O157:H7 infection risk was 1.7 × 10−1 (95% CI: 0.02–0.39) for the worst-case scenario and 4.2 × 10−2 (95% CI: 0.0013–0.177) for the best-case scenario. In the dry season, the E. coli O157:H7 infection risks were 1.8 × 10−2 (95% CI: 0.0002–0.0810) for the worst-case scenario and 2.05 × 10−3 (95% CI: 7.63 × 10−6–1.36 × 10−2) for the best-case scenario. The variability and uncertainty of the risk estimates derived from the second-order Monte Carlo simulation under the different seasons and scenarios are presented in Figure 4.

Table 2

Predicted infection risk, diarrhoeal disease cases and disability adjusted life years (DALYs) associated with a single consumption of E. coli O157:H7-contaminated lettuce salad in Kumasi, Ghana

Wet season Dry season
Scenario Infection risk (Pinf) (mean (95% CI)) Diarrhoea morbidity (mean (95% CI)) Diarrhoea mortality (mean (95% CI)) DALYs (mean (95% CI)) Infection risk (Pinf) (mean (95% CI)) Diarrhoea mortality (mean (95% CI)) Diarrhoea mortality (mean (95% CI)) DALYs (mean (95% CI))
Worst-case scenario 1.7 × 10−1 (2 × 10−2−3.9 × 10−11.2 × 10−1 (1.4 × 10−2−2.7 × 10−18.25 × 10−5 (9.6 × 10−6−1.9 × 10−42.8 × 10−3 (3 × 10−4−6.5 × 10−31.8 × 10−2 (2 × 10−4−8.1 × 10−11.2 × 10−2 (1.4 × 10−4−5.6 × 10−28.69 × 10−6 (9.6 × 10−8−3.9 × 10−53 × 10−4 (3 × 10−6−1.3 × 10−3
Best-case scenario 4.2 × 10−2 (1.3 × 10−3−1.8 × 10−12.8 × 10−2 (8.9 × 10−4−1.2 × 10−12 × 10−5 (6.28 × 10−7−8.6 × 10−57 × 10−4 (2.2 × 10−5−2.96 × 10−32.05 × 10−3 (7.63 × 10−6−1.36 × 10−21.4 × 10−3 (5.2 × 10−6−9.38 × 10−39.9 × 10−7 (3.68 × 10−9−6.6 × 10−63.4 × 10−5 (1.2 × 10−7−2.3 × 10−4
Wet season Dry season
Scenario Infection risk (Pinf) (mean (95% CI)) Diarrhoea morbidity (mean (95% CI)) Diarrhoea mortality (mean (95% CI)) DALYs (mean (95% CI)) Infection risk (Pinf) (mean (95% CI)) Diarrhoea mortality (mean (95% CI)) Diarrhoea mortality (mean (95% CI)) DALYs (mean (95% CI))
Worst-case scenario 1.7 × 10−1 (2 × 10−2−3.9 × 10−11.2 × 10−1 (1.4 × 10−2−2.7 × 10−18.25 × 10−5 (9.6 × 10−6−1.9 × 10−42.8 × 10−3 (3 × 10−4−6.5 × 10−31.8 × 10−2 (2 × 10−4−8.1 × 10−11.2 × 10−2 (1.4 × 10−4−5.6 × 10−28.69 × 10−6 (9.6 × 10−8−3.9 × 10−53 × 10−4 (3 × 10−6−1.3 × 10−3
Best-case scenario 4.2 × 10−2 (1.3 × 10−3−1.8 × 10−12.8 × 10−2 (8.9 × 10−4−1.2 × 10−12 × 10−5 (6.28 × 10−7−8.6 × 10−57 × 10−4 (2.2 × 10−5−2.96 × 10−32.05 × 10−3 (7.63 × 10−6−1.36 × 10−21.4 × 10−3 (5.2 × 10−6−9.38 × 10−39.9 × 10−7 (3.68 × 10−9−6.6 × 10−63.4 × 10−5 (1.2 × 10−7−2.3 × 10−4
Figure 4

E. coli O157:H7 infection risk associated with the consumption of wastewater-irrigated lettuce in the wet season, worst-case (a) and best-case (b), and dry season, worst-case (c) and best-case (d) scenarios. Grey bands correspond to the 95% uncertainty range on each quartile of variability and dark grey band corresponds to the 50% uncertainty range on each quartile of variability.

Figure 4

E. coli O157:H7 infection risk associated with the consumption of wastewater-irrigated lettuce in the wet season, worst-case (a) and best-case (b), and dry season, worst-case (c) and best-case (d) scenarios. Grey bands correspond to the 95% uncertainty range on each quartile of variability and dark grey band corresponds to the 50% uncertainty range on each quartile of variability.

### E. coli O157:H7-related diarrhoeal morbidity and mortality

In the wet season, the diarrhoeal disease morbidity per consumer per single serving was 0.12 (95% CI: 0.014–0.269) for the worst-case scenario and 0.028 (95% CI: 0.00089–0.12213) for the best-case scenario (Table 2). The estimated mortalities resulting from the morbidity cases per serving were 8.25 × 10−5 (95% CI: 9.6 × 10−6–1.9 × 10−4) and 2 × 10−5 (6.28 × 10−7–8.6 × 10−5) for the worst-case and best-case scenarios, respectively, in the wet season. In the dry season, the estimated morbidities per consumer per single serving for the worst-case and best-case scenarios were respectively 0.012 (95% CI: 0.000138–0.05589) and 0.0014 (95% CI: 5.26 × 10−6 –9.4 × 10−3), respectively. In the same season, the corresponding diarrhoeal mortalities for the worst-case and best-case scenarios were 8.69 × 10−6 (95% CI: 9.6 × 10−8–3.9 × 10−5) and 9.9 × 10−7 (95% CI: 3.68 × 10−9–6.6 × 10−6), respectively.

### E. coli O157:H7 DALYs

The estimated disability adjusted life years (DALYs) for the wet season were 2.8 × 10−3 (95% CI: 3 × 10−4–6.5 × 10−3) and 7 × 10−4 (95% CI: 2.2 × 10−5–2.96 × 10−3) for the worst-case and best-case scenarios, respectively (Table 2). In the dry season the DALYs were 3 × 10−4 (3 × 10−6–1.3 × 10−3) and 3.4 × 10−5 (95% CI: 1.2 × 10−7–2.3 × 10−4) for the worst-case and best-case scenarios, respectively (Table 2).

## DISCUSSION

The study shows that the occurrence of E. coli O157:H7 in wastewater-irrigated lettuce was seasonally dependent, with potential determinants including precipitation, temperature and humidity. On average, wastewater-fed lettuce collected from the farms was nearly six times (28% in wet season compared to 5% in the dry season) as likely to be contaminated with E. coli O157:H7 in the wet season as in the dry season. In the wet season, run-offs from animal farms and grazing sites adjacent to the farms are a potential source of E. coli O157:H7 contamination of the irrigation water. Also, almost all positive farms used chicken manure, a source of E. coli O157:H7, as a soil ameliorant (Griffin & Tauxe 1991; Griffin 1995). E. coli O157:H7 can survive longer in manure-amended soils (Kudva et al. 1998; Gagliardi & Karns 2000; Islam et al. 2004) and even longer in the wet season when the soil is either moist or wet (Solomon et al. 2002). Islam et al. (2004) demonstrated that E. coli O157:H7 in compost could survive for more than 5 months after application to the soil; and can contaminate the leaves of low-growing vegetables such as lettuce in the event of rainfall through soil splashes. The survival of E. coli O157:H7 on the leaves of lettuce is also likely to be longer in the wet season due to favourable environmental conditions (e.g., temperature and humidity) compared to the dry season. Temperature is an important determinant of the persistence of pathogens in the environment. The average temperature across the study sites was lower in the wet season (25.2–31.5 °C) than in the dry season (29.5–37.6 °C). The effect of temperature and humidity on the survival of E. coli O157:H7 as a function of season was recently demonstrated (Oliveira et al. 2012). Oliveira et al. (2012) found that E. coli O157:H7 was more persistent on lettuce leaves in the fall when the humidity and temperature were 10 °C and 82%, respectively, and less persistent in the spring when humidity and temperature were 17 °C and 62%.

The study has shown that wastewater-irrigated farms are contributing to the transmission of E. coli O157:H7 with associated diarrhoeal disease and secondary complications (i.e., HUS and ESRD) in the city of Kumasi. Under the worst-case scenario in the wet season, the E. coli O157:H7 infection risk was 1.7 × 10−1 per single consumption of wastewater-irrigated lettuce. This is about 6 log greater than the World Health Organization (WHO) tolerable diarrhoeal disease infection risk of 2.7 × 10−7 per consumer per day (∼10−4 per consumer per year) (WHO 2006). The wet season worst-case scenario E. coli O157:H7 infection risk resulted in 0.12 diarrhoeal disease cases per consumer per exposure, which is comparable with the diarrhoeal disease prevalence of 0.8–1.3 per person per year for developing countries with a potentially dire economic consequence (Mathers et al. 2002). The relatively high E. coli O157:H7 infection risk and accompanying DALYs has implications for the local economy and health systems of the city in particular. Currently, there are no epidemiological data on the incidence and cost of E. coli O157:H7-related diseases in Ghana in general, and Kumasi in particular. However, studies from the United States suggest that E. coli O157:H7-related diseases could have a devastating impact on the local economy. Mead et al. (1999) estimated that US$405 million was lost annually through E. coli O157 (STEC) diseases related to food or other sources in the United States. Another estimate from the United States revealed that the average cost of E. coli O157:H7-related disease can range from US$26 for individuals who did not obtain medical care to US\$5.2 million for cases that were hospitalized after developing HUS and ESRD (Frenzen et al. 2005). Based on Frenzen et al.’s (2005) study, the social and economic costs will be too prohibitive in the event that consumers of wastewater-irrigated lettuce salad in Kumasi develop HUS and ESRD. For those consumers suffering from kidney failure, the treatment options are kidney transplant or dialysis. The costs of paying for thrice-weekly dialysis would be unaffordable for the vast majority of consumers in Kumasi if they needed to pay out of their own pocket (Callegari et al. 2013). This seriously questions the continuous use of wastewater for vegetable irrigation given that the existing irrigation and farm practices lead to the contamination of vegetables with E. coli O157:H7.

From the foregoing information, the most critical question that policymakers should address is how to ensure that wastewater use in urban vegetable production does not result in significantly high economic and public health burdens. Several on-farm and post-harvest risk interventions have been piloted in Kumasi and other cities in Ghana to reduce the health risk (including diarrhoeal disease cases) associated with wastewater irrigation. These include on-farm interventions such as drip irrigation (Keraita et al. 2008a), irrigation cessation (Keraita et al. 2007), use of simple filters (Keraita et al. 2008a), sedimentation ponds (Keraita et al. 2008b) and post-farm interventions such as vegetable disinfection (Amoah et al. 2007). On-farm interventions such as drip irrigation, irrigation cessation and simple filtration have been effective in the reduction of faecal coliforms on wastewater-irrigated lettuce, and yet have proven difficult to implement due to economic reasons and farmers' perceptions of health risk. In the short to medium term, these on-farm interventions are not likely to be implemented by farmers without a comprehensive enforcement backed by effective local guidelines and regulations. At the same time, additional efforts should be made towards improving post-harvest handling practices in the kitchens of street-food vendors and restaurants. For example, the study shows that a 0.5–2 log reduction in E. coli O157:H7 in the wastewater-irrigated lettuce would reduce the wet season worst-case diarrhoeal disease cases by a little over 90%.

In Ghana, surface disinfection of lettuce with salt solution is widely practised by street-food sellers and restaurants. Chlorine and vinegar are also used, but sparingly, because food sellers consider them to be expensive (Amoah et al. 2007). While the efficacy of salt solution and vinegar in the removal of E. coli O157:H7 on vegetables has not been documented in the literature, studies have reported a 1.5–2 log reduction of E. coli O157:H7 on vegetables (Beuchat 1999; Keskinen et al. 2009; Al-Nabulsi et al. 2014). Beuchat (1999) showed that the treatment of lettuce with 200 ppm chlorine solution or deionized water was effective in removing E. coli O157:H7 from lettuce. However, chlorine disinfection may prove ineffective if E. coli O157:H7 internalized in the vegetables is not significantly exposed to the active disinfectant agent, as has been demonstrated in several studies. E. coli O157:H7 can be internalized into lettuce through roots (Solomon et al. 2002; Erickson et al. 2010), attach inside stomatal pores (Seo & Frank 1999) and invade interior tissues via cut leaf edges (Kazue & Frank 2000; Erickson et al. 2010). Internalization of E. coli O157:H7 can render the surface treatment of lettuce with disinfectants such as chlorine ineffective (Kazue & Frank 2000). For example, a study showed that treatment of E. coli O157:H7 internalized in iceberg lettuce with 300 and 600 ppm chlorine solutions resulted in reductions of less than 1 log (Niemira 2008). Irradiation has been found to be more effective in the reduction of E. coli O157:H7 internalized in the tissues of lettuce. Niemira (2008) found a 5 log reduction of E. coli O157:H7 in iceberg lettuce treated with 1.5 kGy of irradiation. Nthenge et al. (2007) also showed that E. coli O157:H7 in the leaf tissue of lettuce was effectively removed by gamma irradiation while 200 ppm aqueous chlorine had no effect. Electrolyzed oxidizing water for the removal of E. coli O157:H7 on vegetables has also been demonstrated to be more effective than chlorine (Keskinen et al. 2009). These alternative disinfectants are relatively expensive and are not currently available to the vast majority of food-sellers and restaurants in Kumasi. It is therefore important that food sellers are introduced to less costly but effective disinfection measures. The introduction of any disinfection measure has to be backed by robust legislative instruments with objectively verifiable monitoring indicators to ensure its full implementation.

This study is not without limitations. The culture-based method used to enumerate E. coli O157:H7 could have underestimated the number of E. coli O157:H7 in the wastewater-irrigated lettuce from the study farms. The limitation of culture methods for E. coli O157:H7 enumeration was recently demonstrated by Moyne et al. (2013). The study by Moyne et al. (2013) showed that within 2 h of application, the culturable number of E. coli O157:H7 decreased by 1,000-fold, while molecular methods (polymerase chain reaction) showed the number of cells was equivalent to the inoculum levels. It was concluded that E. coli O157:H7 either died immediately after application or were no longer culturable. According to Moyne et al. (2013), sub-injured cells, which are rarely captured using culture-based methods, can be infectious, and have the ability to regrow under favourable temperature and moisture conditions. Without any further studies to assess the recovery rate of the method, the risk assessment models assumed full recovery (100%) for the method used in this study. This assumption can lead to the underestimation of the infection risk estimates arrived at in this study. It is recommended that molecular methods are used to enumerate E. coli O157:H7 in wastewater-irrigated vegetables.

## CONCLUSION

This study has shown that wastewater-fed agricultural produce in Kumasi is contributing to the transmission of E. coli O157:H7 among consumers of wastewater-irrigated lettuce with potential dire economic and public health burdens. Irrespective of the season, both the infection risk and disability adjusted life years (DALYs) exceed WHO tolerable levels. A reduction of E. coli O157:H7 by 0.5–2 log to reflect the current disinfection used by street-food sellers and restaurants did not result in tolerable infection risk and disease burden. It is recommended that legislation that ensures that both farmers and food-sellers adopt effective risk mitigation measures is implemented. The legislation should have clear, verifiable monitoring indicators with a penalty for those who do not comply.

## ACKNOWLEDGEMENTS

This study was funded by the Research Council of Norway (INTWASTE Project No: 204066). We acknowledge the support of the wastewater farmers in Kumasi and the laboratory technicians at the Department of Theoretical and Applied Biology, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana.

## REFERENCES

REFERENCES
Abu-Ali
G. S.
Ouellette
L. M.
Henderson
S. T.
Whittam
T. S.
Manning
S. D.
2010
Differences in adherence and virulence gene expression between two outbreak strains of enterohaemorrhagic Escherichia coli O157:H7
.
Microbiology
156
,
408
419
.
Al-Nabulsi
A. A.
Osaili
T. M.
Obaidat
H. M.
Shaker
R. R.
Awaisheh
S. S.
Holley
R. A.
2014
Inactivation of stressed Escherichia coli O157:H7 cells on the surfaces of rocket salad leaves by chlorine and peroxyacetic acid
.
J. Food Prot.
1
,
32
39
.
Baron
E. J.
Peterson
L. R.
Finegold
S. M.
1994
Bailey and Scott's Diagnostic Microbiology
,
9th edn.
C.V. Mosby Company
,
St Louis, MO
, pp.
333
351
.
Beuchat
L. R.
1999
Survival of enterohemorrhagic Escherichia coli O157:H7 in bovine feces applied to lettuce and the effectiveness of chlorinated water as a disinfectant
.
J. Food Prot.
8
,
845
849
.
Callegari
J.
Antwi
S.
Wystrychowski
G.
Żukowska-Szczechowska
E.
Levin
N. W.
Carter
M.
2013
Peritoneal dialysis as a mode of treatment for acute kidney injury in Sub-Saharan Africa
.
Blood Purif.
36
(
3–4
),
226
230
.
Cooley
M.
Carychao
D.
Crawford-Miksza
L.
Jay
M. T.
Myers
C.
Rose
C.
Keys
C.
Farrar
J.
Mandrell
R. E.
2007
Incidence and tracking of Escherichia coli O157:H7 in a major produce production region in California
.
PLoS One
11
,
1
59
.
A.
Karou
T.
N.
Kette
A.
Dosso
M.
2000
Isolation of enteric pathogenic agents in Côte d'Ivoire: Escherichia coli O157:H7 and enteroaggregative E. coli
.
Bull. Soc. Pathol. Exot.
93
,
95
96
.
Erickson
M. C.
Webb
C. C.
Diaz-Perez
J. C.
Phatak
S. C.
Silvoy
J. J.
Davey
L.
Payton
A. S.
Liao
J.
Ma
L.
Doyle
M. P.
2010
Infrequent internalization of Escherichia coli O157:H7 into field-grown leafy greens.
J. Food Prot.
73
(
3
),
500
506
.
Fitzpatrick
M. M.
Shah
V.
Trompeter
R. S.
Dillon
M. J.
Barratt
T. M.
1991
Long term outcome of childhood haemolytic uraemic syndrome
.
Br. Med. J.
303
,
489
492
.
Frenzen
P. D.
Drake
A.
Angulo
F. J.
&
The Emerging Infections Programme Foodnet Working Group
.
2005
Economic cost of illness due to Escherichia coli O157 infections in the United States
.
J. Food Prot.
68
(
12
),
2623
2630
.
Gagliardi
J. V.
Karns
J. S.
2000
Leaching of Escherichia coli O157:H7 in diverse soils under various agricultural management practices
.
Appl. Environ. Microbiol.
66
,
877
883
.
Germani
Y.
Cunin
P.
Tedjouka
E.
Ncharre
C.
Morvan
J.
Martin
P.
1998
Enterohaemorrhagic Escherichia coli in Ngoïla (Cameroon) during an outbreak of bloody diarrhoea
.
Lancet
352
(
9128
),
652
656
.
Ghana Statistical Service
.
2010
Population and Housing Census, Summary Report of Final Result
.
Ghana Statistical Service
,
Accra, Ghana
.
Griffin
P. M.
1995
Escherichia coli O157:H7 and other enterohaemorrhagic Escherichia coli
. In:
Infections of the Gastrointestinal Tract
(
Blaser
M. J.
Smith
P. D.
Ravdin
J. I.
Greenberg
H. B.
Guerrant
R. L.
, eds).
Raven Press
,
New York
, pp.
739
761
.
Griffin
P. M.
Tauxe
R. V.
1991
The epidemiology of infections caused by Escherichia coli O157:H7, and other enterohemorrhagic E. coli, and the associated hemolytic uremic syndrome
.
Epidemiol. Rev.
13
,
60
98
.
Haas
C. N.
Rose
J. B.
Gerba
C. P.
1999
Quantitative Microbial Risk Assessment.
John Wiley & Sons
,
New York
.
Haas
C. N.
A.
Rose
J. B.
Gerba
C. P.
2000
Development of a dose–response relationship for Escherichia coli O157:H7
.
Int. J. Food Microbiol
.
56
,
153
159
.
Havelaar
A. H.
Melse
J. M.
2003
Quantifying Public Health Risk in the WHO Guidelines for Drinking Water-Quality: A Burden of Disease Approach, RIVM Report no. 734301022/2003
,
Rijksinstituut voor Volksgezondheid en Milieu
,
Bilthoven, The Netherlands
.
Hilborn
E. D.
Mermin
J. H.
Mshar
P. A.
J. L.
Voetsch
A.
Wojtkunski
C.
Swartz
M.
Mshar
R.
Lambert-Fair
M. A.
Farrar
J. A.
Glynn
M. K.
Slutsker
L.
1999
A multistate outbreak of Escherichia coli O157:H7 infections associated with consumption of mesclun lettuce
.
Arch. Intern. Med.
159
(
15
),
1758
1764
.
Islam
M.
Doyle
M. P.
Phatak
S. C.
Milner
P.
Jiang
X.
2004
Persistence of enterohemorrhagic Escherichia coli O157:H7 in soil and on leaf lettuce and parsley grown in fields treated with contaminated manure composts or irrigation water
.
J. Food Prot.
67
(
7
),
1365
1370
.
Jay
M. T.
Garrett
V.
Mohle-Boetani
J. C.
Garret
V.
Barros
M.
Farrar
J. A.
Rios
R.
Abbott
S.
R.
Komatsu
K.
Mandrell
R.
Sobel
J.
Werner
S. B.
2004
A multistate outbreak of Escherichia coli O157:H7 infection linked to consumption of beef tacos at a fast-food restaurant chain
.
Clin. Infect. Dis.
39
,
1
7
.
Kazue
T.
Frank
J. F.
2000
Penetration of Escherichia coli O157:H7 into lettuce tissues as affected by inoculum size and temperature and the effect of chlorine treatment on cell viability
.
J. Food Prot.
4
,
427
552
.
Keraita
B.
F.
Drechsel
P.
Abaidoo
R. C.
2007
Reducing microbial contamination on wastewater-irrigated lettuce by cessation of irrigation before harvesting
.
Trop. Med. Int. Health
12
(
s2
),
8
14
.
Keraita
B.
Drechsel
P.
F.
Vreugdenhil
R. C.
2008a
Potential of simple filters to improve microbial quality of irrigation water used in urban vegetable farming in Ghana
.
J. Environ. Sci. Health, A
43
,
749
755
.
Keraita
B.
Drechsel
P.
F.
2008b
Using on-farm sedimentation ponds to improve microbial quality of irrigation water in urban vegetable farming
.
Water Sci. Technol.
57
(
4
),
519
525
.
Kudva
I. T.
Blanch
K.
Hovde
C. J.
1998
Analysis of Escherichia coli O157:H7 survival in ovine or bovine manure and manure slurry
.
Appl. Environ. Microbiol.
64
,
3166
3174
.
Mathers
C. D.
Stein
C.
Ma Fat
D.
Rao
C.
Inoue
M.
Tomijima
N.
Bernard
C.
Lopez
A. D.
Murray
C. J. L.
2002
Global Burden of Disease 2000: Version 2 Methods and Results
.
WHO
,
Geneva
.
P. S.
Slustsker
L.
Dietz
V.
McCaig
L. F.
Bresee
J. S.
Shapiro
C.
Tauxe
R. V.
1999
Food-related illness and death in the United States
.
Emerg. Infect. Dis.
5
(
5
),
607
625
.
Michino
H.
Araki
K.
Minami
S.
Takaya
S.
Sakai
N.
Miyazaki
M.
Ono
A.
Yanagawa
H.
1999
Massive outbreak of Escherichia coli O157:H7 infection in schoolchildren in Sakai City, Japan, associated with consumption of white radish sprouts
.
Am. J. Epidemiol.
150
(
8
),
787
796
.
Murray
C. J.
1994
Quantifying the burden of disease: The technical basis for disability-adjusted life years
.
Bull. World Health Organ.
72
(
3
),
429
445
.
Nthenge
A. K.
Weese
J. S.
Carter
M.
Wei
C.-I.
Huang
T.-S.
2007
Efficacy of gamma radiation and aqueous chlorine on Escherichia coli O157:H7 in hydroponically grown lettuce plants
.
J. Food Prot.
70
(
3
),
748
752
.
Oliveira
M.
Viñas
I.
Usall
J.
Anguera
M.
M.
2012
Presence and survival of Escherichia coli O157:H7 on lettuce leaves and in soil treated with contaminated compost and irrigation water
.
Int. J. Food Microbiol.
156
(
2
),
133
140
.
Olorunshola
I. D.
Smith
S. I.
Cker
A. O.
2000
Prevalence of EHEC O157:H7 in patients with diarrhoea in Lagos, Nigeria
.
APMIS
108
,
761
763
.
Riley
L. W.
Remis
R. S.
Helgerson
S. D.
MCGee
H. B.
Wells
J. G.
Davis
B. R.
Hebert
R. J.
Olcott
E. S.
Johnson
L. M.
Hargrett
N. T.
Blake
P. A.
Cohen
M. L.
1983
Hemorrhagic colitis associated with a rare Escherichia coli serotype
.
N. Engl. J. Med.
308
(
12
),
681
685
.
Sang
W. K.
Saidi
S. M.
Yamamoto
H.
Ezaki
T.
Iida
T.
Yoh
M.
Honda
T.
1996
Haemorrhagic colitis due to Escherichia coli O157:H7 in Kenya
.
J. Trop. Pediatr.
42
,
118
119
.
Seidu
R.
Drechsel
P.
2010
Cost-effectiveness analysis of treatment and non-treatment interventions for diarrhoea disease re­duction associated with wastewater irrigation
. In:
Wastewater Irrigation and Health: Assessing and Mitigating Risk in Low-income Countries
(
Drechsel
P.
Scott
C. A.
Raschid-Sally
L.
Redwood
M.
Bahri
A.
, eds).
Earthscan
,
London
.
Seidu
R.
A.
Amoah
P.
Drechsel
P.
Jenssen
P. D.
Stenstrøm
T. A.
2008
Quantification of the health risk associated with wastewater reuse in Accra, Ghana: a contribution toward local guidelines
.
J. Water Health
6
(
4
),
461
471
.
Seidu
R.
Sjølander
I.
Abubakari
A.
Amoah
I. D.
Larbi
J. A.
Stenström
T. A.
2013
Modelling the die-off of E. coli and Ascaris in wastewater irrigated vegetables: implications for microbial health risk reduction associated with irrigation cessation
.
Water Sci. Technol.
68
(
5
),
1013
1021
.
Seo
K. H.
Frank
J. F.
1999
Attachment of Escherichia coli O157:H7 to lettuce leaf surface and bacterial viability in response to chlorine treatment as demonstrated by using confocal scanning laser microscopy
.
J. Food Prot.
62
,
3
9
.
Söderström
A.
Österberg
P.
Lindqvist
A.
Jönsson
B.
Lindberg
A.
Ulander
S. B.
Welinder-Olsson
C.
Löfdahl
S.
Kaijser
B.
De Jong
B.
Kühlmann-Berenzon
S.
Boqvist
S.
Eriksson
E.
Szanto
E.
S.
Allestam
G.
Hedenström
I.
Muller
L. L.
Y.
2008
A large Escherichia coli O157 outbreak in Sweden associated with locally produced lettuce
.
Foodborne Pathog. Dis.
5
(
3
),
339
349
.
Su
C.
Brandt
L. J.
1995
Escherichia coli O157:H7 infection in humans
.
Ann. Intern. Med.
123
,
698
714
.
Teunis
P. F. M.
Ogden
I. D.
Strachan
N. J. C.
2008
Hierarchical dose–response of E. coli O157:H7 from human outbreaks incorporating heterogeneity in exposure
.
Epidemiol. Inf.
136
,
761
770
.
WHO
2006
Guidelines for the Safe Use of Wastewater, Excreta and Greywater: Wastewater Use in Agriculture (Volume 2).
WHO
,
Geneva, Switzerland
.
WHO
2013
World Health Statistics 2013
.
Geneva, Switzerland
.