Irrigation of Lettuce With Reclaimed Water: Health Risk Comparison Among United States, Mexico and Brazil

Hazard Identification

Norovirus

Norovirus is the leading cause of acute gastroenteritis in all age groups in the United States. Each year, norovirus causes about 20 million cases of acute gastroenteritis, and contributes to about 70,000 hospitalizations and 800 deaths, mostly among young children and older adults, according to the CDC.

Norovirus affects all ages, with outbreaks peaks in cold weather. Outbreaks often occur in semi-closed environments, such as nursing homes, hospitals and schools, according to Glass et al.. Norovirus is highly contagious. It spreads through close person-to-person contact, contaminated food or water and contaminated fomites. It can also spread through aerosolized vomit that lands on surfaces or enters a person’s mouth and is swallowed, according to the CDC. The virus can withstand a wide range of temperatures (from freezing to 60°C) and persist on environmental surfaces, in recreational and drinking water, and in a variety of food items, according to Glass et al..

The incubation period is 10 to 51 hours and the symptoms include vomiting, which is more common in children, and diarrhea, more common in adults. The illness last typically 28 to 60 hours. The illness is overall less severe than others diarrheal infections, but can lead to dehydration and hospitalization, especially among children <5 yr of age and adults >65 yr of age. Viral shedding occur with peaks at 1-3 days after onset of illness, but may shed for up to 56 days Treatment includes supportive therapy to prevent dehydration. Humans are the reservoir for norovirus, and recent evidence raises possibility of animal reservoir, according to Glass et al..

Norovirus is on top five pathogens contributing to domestically acquired foodborne illnesses resulting in death. The estimated number of deaths is 149, with a 90% credible interval of 84 to 237, according to the CDC. Norovirus was reported in 5 to 31% of patients hospitalized for gastroenteritis and in 5 to 36% of those visiting a clinic, making it the most common cause of diarrhea in adults and the second most common cause in children, according to Patel et al.. In fact, norovirus is the leading cause of severe acute gastroenteritis among U.S. children less than 5 years-old who seek medical care. The virus is responsible for nearly 1 million pediatric medical care visits annually. By 5 years-old, an estimated 1 in 278 children will be hospitalized; 1 in 14 will visit an emergency room, and 1 in 6 will receive outpatient care for norovirus illnesses. Each year, it costs about $2 billion in the United States for healthcare and lost productivity from foodborne illness caused by norovirus, according to the CDC. In Brazil, there were 24 norovirus outbreaks from 2000 to 2013, according to SVS. It is important to take into account that there is an underreporting in Brazil, mainly if the symptoms are self-limiting.

Escherichia coli O157:H7

E. coli O157:H7 can be isolated from approximately 1% of the guts of ruminant animals. Cattle are the largest carriers, according to the CDC and Boyce. It is transmitted through the fecal- oral route. Consumption of contaminated food, unpasteurized milk and non-disinfected water, and contact with ruminant animals or feces can result in exposure. High risk items include unpasteurized milk and apple cider, soft cheeses produced from unpasteurized milk and undercooked ground beef, according to the CDC and Boyce.

E. coli is associated diarrheal illness. Symptoms manifest 1-10 days after exposure, with an average of 3-4 days. Symptoms include severe stomach cramps, diarrhea (including bloody diarrhea), vomiting, and low fever. Symptoms generally last for 5-7 days. Bacterial shedding occurs only when symptoms are present, according to the CDC. There are approximately 73-95 thousand cases of infection each year with a hospitalization rate of 0.2985-0.465, according to Mead, Scallan, and the CDC. The morbidity rate of infection is 0.51, according to Muniesa. E. coli infection results in a case fatality ratio ranging from 0.005-0.0083, according to Mead  and Scallan.

E. coli O157:H7 is a shiga-toxin producing strain of E. coli. Shiga-toxin has been associated with hemolytic uremic syndrome (HUS). HUS occurs in 5-10% of E. coli infections. HUS causes decreased urination, tiredness, loss of color on the face, kidney failure and can lead to death, according to the CDC. This syndrome has a case fatality ratio ranging from 0.03-0.05, according to Boyce. In addition to HUS, infection with E. coli has been associated with the following sequelae: end-stage renal disease, permanent neurologic injury, and hypertension in pregnant women, according to Boyce and Moist.

Exposure Assessment

This risk assessment can be broken down into three basic parts.

1. Wastewater Effluent: This takes into account pathogen concentrations in effluent water, the time it takes for the effluent to leave the treatment plant and arrive at the field, dilution of the effluent water and the survival of the pathogens to the field.

2. Pre-Harvest: This takes into account the irrigation method, the microbial interaction with the produce (including attachment and internalization) and the survival of the pathogens from the last irrigation event to harvest.

3. Post-Harvest: This takes into account the pathogen survival to the market, country specific household sanitization practices and country specific consumption patterns.

The following table contains a step-by-step presentation of factors considered and assumptions that were made for each individual step.

For input values see Risk Characterization.

Dose Response

A dose response dataset for Norovirus has been published previously by Teunis et al. 2008. These data were experimental human data for a Norwalk (1971) virus isolate (genotype I). Approximately 30% of the studied population was immune to infection (Se-). Data from this paper were used except Se+ and Se- receptor type data were pooled and fit using CAMRA dose response code in R (www.rproject.org). We assumed 1 RT-qPCR = 1 virion and did not convert estimates to focus forming units as has been done previously for viral models using rotavirus as a surrogate. The best fit model for Norovirus using this method was the Beta Poisson with mean alpha and N50 values of 0.0545 (95% CI 0.0359, 0.08353) and 5.238E+06 (95% CI 1.928E+06, 1.26E+10), respectively.

An existing dose response model was used for E. coli O157:H7 strain 86-24 available in the recommended dose response parameter section of this blog. The data were derived from a pig model with an infection endpoint and oral route of exposure. The best fit model was exponential with mean k=2.18E-04 (95% CI 9.40E-05, 5.99E-04).

Risk Characterization

Three risk outputs for each country of interest (US, Brazil, Mexico) were generated using Crystal Ball (R) for a total of 9 scenarios for each microbial hazard. Risks were calculated taking into account surface concentrations only, internalized concentrations only, and a combination. Input distributions were generated using Crystal Ball(R). Distribution types and their sources are shown below.

95% confidence intervals were compared. For EHEC, annual infection risk intervals for the US and Brazil were < 1E-04 for all scenarios. EHEC annual infection risk in Mexico, however, was > 1E-04 for all scenarios. Norovirus annual infection risks for the US were all < 1E-04. Although mean annual infection risks for Brazil were <1E-04, the 95th percentile exceeded this risk for all scenarios. For Mexico, annual infection risks >>1E-04 in all cases.

For EHEC, a sensitivity analysis of surface+ internal contamination with Spearman rank analysis showed that the most important parameters for Mexico and Brazil (in order of effect on calculated annual infection risk) were irrigation water concentration, dose response parameter k, household treatment scheme, volume of irrigated water retained, internalization rate, and surface attachment rate. for the US, uncertainty was dominated by dose response parameter k and household treatment scheme.

For Norovirus, a sensitivity analysis of surface + internal contamination showed that dose response parameters dominated the uncertainty for all three countries while internalization rate, surface attachment rate, decay rate, household treatment, irrigation water concentration, and the volume of water retained on the lettuce plant also affected the analysis (in order of effect).

Risk Management & Communication

A comparison with previous QMRAs performed on irrigated lettuce supports the plausibility of this analysis (range 10^-2 to 10^-9 annual probability of infection in E. coli and enteric viruses).

Accounting for internalization increases annual infection risk ~0.5 log compared to surface contamination alone, however most risks were below a 1E-4 annual risk of infection. Although current US reclaimed water practices appear to be protective for public health, Norovirus research in reclaimed water is a priority for Brazil and Mexico, and EHEC also has a potential for concern in these countries. It should be noted here that Mexico calculations were based on raw wastewater irrigation practices. Given the limited data on irrigation practices in Mexico, quantitative monitoring for these pathogens is recommended as well as the incorporation of a fate and transport model in future risk assessments.

Susceptible populations should be cautious when consuming leafy greens irrigated by wastewater such as farmers and their families, consumers of fresh vegetables, children, the elderly, and those who are Se+ for Norovirus. Communication among government agencies (health, water and food) and research institutions will be important in this effort.