Findings and Potential Preventive Control Strategies
Willette Crawford, Ph.D., MPH, FDA Center for Food Safety and Applied Nutrition
Mansoor Baloch, Ph.D., CDC Environmental Health Services Branch - National Center for Environmental Health
Kevin Gerrity, National Food Expert, FDA Division of Domestic Field Investigations (DDFI) - Los Angeles District/San Diego Resident Post
December 29, 2010
Also available in (PDF - 191 KB).
- Initial On-Farm Investigation
- Hypothesis Development
- The Environmental Assessment Approach
- Hydrologic Analysis
- Potential Preventive Control Strategies
An environmental assessment1 is an investigation that is triggered by an outbreak of foodborne illness or a food contamination incident. The purpose of the assessment is to determine how the environment may have contributed to the introduction and transmission of pathogens or other hazards that caused illness or contamination. FDA works with the involved firm and the appropriate government agencies in conducting such an investigation. FDA analyzes the evidence and makes conclusions about relevant potential preventive controls. FDA also determines if any regulatory action is needed with regard to the involved firm, or if information was learned during the assessment that warrants other measures. This report summarizes the investigation that was conducted into the factors that FDA considers to have been the most likely contributors to a 2010 foodborne illness outbreak, including possible contributing factors, possible root causes and identification of potential preventive controls to avoid future outbreaks.
This environmental assessment was conducted in response to a multi-state foodborne disease outbreak investigation involving 33 cases of E. coli O145 infection in five states in the spring of 2010. An epidemiologic investigation found that the illnesses were associated with the consumption of shredded romaine lettuce processed at one firm in Ohio. FDA’s investigation at the processor did not identify a likely source of contamination at the firm. FDA conducted a traceback investigation from the processor that led to the farm that is the subject of this report.
The FDA and the authors wish to thank the following organizations for their assistance in conducting this Environmental Assessment:
California and Arizona Leafy Green Marketing Agreements
Arizona Department of Agriculture
Arizona Department of Environmental Quality
Arizona Department of Water Resources
Arizona Department of Health Services
Wellton-Mohawk Irrigation and Drainage District
Dr. Willette M. Crawford is a consumer safety officer with FDA’s Center for Food Safety and Applied Nutrition (CFSAN), Office of Food Safety/Produce Safety Staff. Before coming to FDA, Dr. Crawford was a food microbiologist in the fresh produce industry where she developed, implemented, and evaluated good agricultural practices and food safety and defense programs. Dr. Crawford has also worked in various research disciplines addressing public and environmental health issues, and has presented her food safety and food defense research internationally as a guest lecturer. She received a doctorate in food science from Purdue University; a masters in public health degree from East Tennessee State University, and Bachelor of Science degrees in both chemistry and biology from South Carolina State University.
Dr. Mansoor Baloch is an ORISE fellow at the Environmental Health Services Branch in the National Center for Environmental Health at the U.S. Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia. Dr. Baloch is a member of the Environment Health Specialist Network-Water program, concentrating on the role of hydrology and hydrologic processes on water-related disease outbreak prevention and investigation. His current research is focused on the role of watershed-based systems approach as a strategy for the investigation and prevention of water-related disease outbreaks. His formal training is in Civil Engineering (Bachelor of Science) and Environmental Sciences and Engineering (Ph.D).
Kevin Gerrity is a National Food Expert for FDA’s Division of Domestic Field Investigations (DDFI) at the Los Angeles District/San Diego Resident Post. Gerrity holds a Bachelor of Science degree in Chemistry from San Diego State University. His FDA experience includes laboratory work as a chemist and sensory analyst, specializing in detecting natural toxins and fraud in seafood; field work as an investigator specializing in food safety including: seafood, low acid canned foods, acidified food, juice, ready-to-eat produce, general foods, and farm investigations. Gerrity’s farm investigation experience includes seafood, shell eggs, sprouts, and leafy greens.
This report describes an environmental assessment that was performed to determine how romaine lettuce implicated in a 2010 E. coli O145 outbreak (See Investigation Update: Multistate Outbreak of Human E. coli O145 Infections Linked to Shredded Romaine Lettuce from a Single Processing Facility, CDC, May 21, 2010 available at: http://www.cdc.gov/ecoli/2010/ecoli_o145/index.html) may have become contaminated with the pathogen. This environmental assessment was preceded by an epidemiological and traceback investigation and regulatory inspection of the processor of the produce implicated in the investigation, the results of which identified the farm upon which the lettuce was grown as the most likely source of the contamination. Nonetheless, FDA has made no definitive determination regarding how or at what point in the supply chain E. coli O145 contaminated the lettuce associated with the outbreak.
The suspect romaine lettuce was grown in four fields of a farm in Wellton, AZ.2 An initial investigation of the growing fields resulted in three soil samples from two growing fields testing positive for non-O157 Shiga toxin- producing strains of E. coli (STEC); but, the outbreak strain, an O145 shiga toxin 2-producing E. coli, was not detected. Significant potential sources of STEC were not identified in the initial on-farm investigation of the growing fields.
However, isolation of these microorganisms at the farm, coupled with the fact that this was the first produce-associated O145 outbreak identified in the U.S. and the first outbreak associated with produce from this growing region, led to the determination that a comprehensive environmental assessment should be conducted to identify potential contributing environmental risk factors and environmental antecedents.
The subsequent environmental assessment initially identified six potential sources of STEC in the Wellton, AZ area; three Concentrated Animal Feeding Operations (CAFOs), one housing development with a co-located sewage treatment facility, one recreational vehicle (R.V.) park with multiple septic leach systems, and the seasonal grazing of sheep on harvested wheat and alfalfa fields.
The three CAFOs are located within approximately eight to eleven miles from the suspect growing fields. With the exception of birds and wind, no route of pathogen transmission from the CAFOs to the growing fields was found. Based upon the fact that the outbreak was not associated with produce from any other farm, we determined that birds and wind are not reasonably likely to be the transportation mechanism of the outbreak pathogen. It was determined that the CAFOs are not reasonably likely to be the source of the outbreak pathogen.
No information on the location and duration of seasonal sheep grazing was available other than the grower’s statement that no sheep were observed in the immediate vicinity of the Grower’s farm this year. According to the grower, the closest sighting of sheep this year was approximately five miles west of the farm. We found no route of pathogen transmission from westerly growing fields to the suspect fields. However, due to the lack of information available to us, seasonal sheep grazing cannot be completely ruled out as a potential source of the outbreak pathogen.
A housing development is located along an elevated mesa on both sides of the main Wellton irrigation canal. The canal is separated from the development on both sides by canal maintenance roads and berms. The development is serviced by an on-site wastewater treatment plant utilizing subterranean activated sludge tanks. Treated wastewater is held in two plastic-lined ponds approximately 300 yards from the main Wellton canal. Water samples collected from one of these ponds tested negative for STEC. Although the new housing development cannot be completely ruled out as a pathogen source for the outbreak, no evidence of drainage from the development to the irrigation canal was observed.
An R.V. park is located on a knoll directly above the lateral irrigation canal that supplies water to the suspect fields. The R.V. park is serviced by eight on-site septic leach systems. During this investigation we found evidence of drainage from the R.V. park property directly into the lateral irrigation canal. Of particular concern was an area that exhibited evidence of drainage into the irrigation canal in which the soil was moist; no surface source of the moisture was observed and there had not been any recent rains.
Soil samples from these moist drainage areas tested negative for STEC. Non--O157 STEC Shiga toxin 2 (Stx2)-producing E. coli was detected in two drag swabs and one mud sample collected from the irrigation canal adjacent to the R.V. park; but none were the O145 shiga toxin 2-producing outbreak strain. We determined that the R.V. park is a reasonably likely potential source of the outbreak pathogen based upon the evidence of direct drainage into the lateral irrigation canal; the moist soil in this drainage area; the multiple sewage leach systems on the property; the presence of other STEC found in the lateral irrigation canal and in the growing fields of the suspect farm; and the fact that the section of the lateral canal downstream from the R.V. park supplies water to only one other farm in addition to the suspect farm.
Two pumps are located on the main Wellton canal near the lateral canal split that supplies water to fields of the suspect farm; one gasoline powered pump on a trailer and one permanent electric pump with an attached hose. The electric pump supplies canal water to an attached open-end hose. The site is not secured from vehicles and the hose pump is also unsecured. At the time of this investigation there were people living in recreational vehicles on undeveloped land within one mile of the hose pump. The fact that this area is open to vehicles and the pump and hose are unsecured make it possible for an R.V. owner to dump and rinse out their R.V. septic system into the main Wellton canal at the lateral canal split that supplies the farm. The ground near the hose pump shows erosion evidence of drainage into the Wellton canal. Soil collected from this erosion site tested positive for other Stx2-producing STEC, but did not match the outbreak strain.
The farm of interest is divided into multiple growing fields, all of which receive water from the same lateral irrigation canal. Romaine lettuce suspected in the outbreak was grown in four fields. Three of these fields are along the eastern border of the farm; the remaining field is near the western edge of the farm.
During the initial farm investigation soil samples were collected from the suspect fields and water was collected from the irrigation canal, the saline waste water canal, and from a saline waste water well. All water samples tested negative for STEC. Three soil samples from two suspect fields tested positive for Stx1 and Stx2-producing non-O157 STEC strains, but none of the soil sample isolates matched the O145 Stx2-producing E. coli outbreak strain.
The initial on-farm investigation did not identify any significant potential sources of E. coli on the farm. Limited evidence of wildlife on and around the farm included a lone set of coyote tracks and two coyote feces on a farm road, rabbit and rodent feces in an area of the farm not contiguous to the suspect fields that was not used for growing produce, and bobcat feces near the Gila River. No evidence of wildlife forage in the growing fields was observed. Coyote, bobcat, rabbit, and rodent feces were collected; no STEC was detected in any of the feces. Information provided by State and local wildlife experts verified that movement of wildlife was minimal during this growing season since water and vegetation were more abundant as a result of increased rainfall during the growing season.
Based upon the above stated evidence we determined:
- The farm had been contaminated with STEC strains.
- No other lettuce growing fields in the area were known to be associated with the outbreak.
- Shiga toxin-producing E. coli was found in multiple locations on the farm.
- There is no significant source of STEC on or near the farm.
- Water used for diluting pesticides and fertilizers, and for irrigation, is the most likely vehicle of pathogen transmission onto/on the farm.
The following evidence was considered in performing the environmental assessment:
- The outbreak appears to involve romaine lettuce solely from the farm of interest; no other growers / sources have been associated with the outbreak.
- The suspect lettuce was grown in four fields on the identified farm.
- Initial sampling of the suspect fields did not detect the outbreak STEC strain; however, two other STEC strains were found in a total of three soil samples from two of the suspect growing fields. It should be pointed out that there are several hundred known STEC serotypes, many of which are found in the environment and many of which are not known to be pathogenic, hence the significance of this finding should be interpreted with caution.
- No significant local source of STEC was identified in the initial farm investigation.
- The initial farm investigation found no evidence of wildlife foraging in the growing fields, and wildlife feces collected during that investigation tested negative for STEC. Reports from wildlife experts verified that wildlife activity was minimal during the growing season.
- Flies abundant in the growing fields during the initial on-farm investigation were determined to be an unlikely STEC vector in that the species of fly is not known to visit feces.
- The growing area received unusually high quantities of rain during the 2009-2010 lettuce-growing season.
- The initial on-farm investigation found no evidence of flooding on the suspect farm, and the irrigation district reported that no irrigation canals were compromised by flooding or storm runoff.
- Irrigation water used on the romaine lettuce crops came from an open canal system that draws water from the Colorado River; this same water source is used by the majority of growers in the Yuma, Wellton, and Mohawk valley growing areas.
- Liquid pesticides and fertilizers used on the lettuce crops were diluted with both municipal and local irrigation canal waters. Municipal water is treated and periodically monitored. Based on these factors, the municipal water was not considered a reasonably likely source of contamination.
In conducting the environmental assessment, FDA worked collaboratively with the CDC, state and local agencies, and industry to determine what possible underlying factors may have contributed to the outbreak contamination and to draft potential preventive controls to address the issues identified. State and local agencies consulted during this environmental assessment include the Arizona Department of Agriculture, Arizona Department of Environmental Quality, Arizona Department of Health Services, Wellton-Mohawk Irrigation and Drainage District, and the Yuma County Department of Public Works.
A three-pronged, systems-based approach was taken during the environmental assessment investigation: 1. Identifying possible significant sources of STEC in the growing area and determining the potential for STEC to be transported from these sources to the farm; 2. A comprehensive inspection of the irrigation canal system for any potential sources of STEC contamination of the irrigation water; and 3. A comprehensive hydrologic assessment of the growing area.
The following potential permanent sources of STEC were identified and evaluated:
- A CAFO is located east of the farm in the immediate vicinity of the Gila River. Manure is composted and hauled away. Wastewater collects in three retention ponds. A physical inspection of the facility was performed and the most recent State inspection report for the facility was reviewed. No evidence of uncontrolled drainage from the feedlot, or of any other potential for E. coli to escape the confines of the feedlot other than birds and wind was found. The State inspection report verified that the firm was in compliance with waste handling regulations. Soil samples collected from the Gila River bed both upstream and downstream of the natural potential drainage route from this CAFO tested positive for Stx1 and Stx2-producing STEC, but were not a match for the outbreak strain. The source of the STEC detected was not determined. Due to the lack of a pathogen transportation mechanism it was determined that this CAFO is not reasonably likely to be the source of the outbreak contamination.
- A small dairy that is co-located with a small beef cattle feedlot shares a storm water retention pond with the beef operation. The dairy is located west of the farm. Wastewater from the facility is contained in evaporation ponds. Manure is not composted; it is stored in covered piles and trucked from the property. A physical inspection of the dairy facility was performed and the most recent State inspection report for the facility was reviewed. No evidence of uncontrolled drainage from the dairy or the beef cattle feed lots, or of any other potential for STEC to escape the confines of the facilities was found other than via birds and wind. The State inspection reports verified that the dairy was in compliance with waste handling regulations (no State inspection reports are available for the co-located beef cattle operation; it is too small to require a permit and thus it is not inspected by the State). Water samples collected from the shared storm water retention pond tested positive for Stx1 and Stx2-producing STEC, but were not a match for the outbreak strain. It is likely that the STEC detected in the storm water retention pond were shed by the cattle from one or both operations; however, this was not confirmed. Due to the lack of a pathogen transportation mechanism it was determined that these two co-located CAFOs are not reasonably likely to be the source of the outbreak contamination.
- A housing development is located on both sides of the main Wellton canal. This development is upstream from the lateral canal split that supplies irrigation water to the suspect farm. Wastewater from the entire development is treated on site via the activated sludge method; treated wastewater is used for irrigation within the development and to maintain two ponds in the development. The wastewater treatment plant and ponds are approximately 300 yards south of the main Wellton canal. The main Wellton canal is at a lower elevation than the development; no evidence of drainage from the development into the Wellton canal was observed. There is one pump at a truck filling station between the south side of the main Wellton canal and the development; this pump appears to draw water from a submerged inlet pipe in the Wellton canal. Water samples collected from one of the ponds tested negative for STEC. However, due to its location above the main Wellton canal immediately upstream of the lateral canal split, and due to the submerged inlet plumbing in the Wellton canal, this development cannot be completely ruled out as a potential source of the outbreak contamination.
- An R.V. park is located on a rise above the valley floor. A lateral irrigation canal is adjacent to the property. There is only one other farm in addition to the identified farm that is serviced by the section of lateral canal downstream of the R.V. park. Sewage from the R.V. park goes into multiple on-site subterranean septic tank leach systems. Along one side of the property there are areas that drain directly into the lateral irrigation canal. During this environmental assessment investigation we found moist soil in some of the areas that drain into the lateral canal. No surface source for the soil moisture was observed, indicating that the moisture is from a subterranean source and thus may be overflow from the R.V. park septic systems. Soil samples from these moist areas tested negative for STEC. Drag swabs and mud collected from the lateral irrigation canal around the R.V. park tested positive for Stx1 and Stx2- producing STEC, but were not a match for the outbreak strain. It is likely that the R.V. park is the source of the STEC detected in the drag swabs and mud collected from the adjacent lateral canal. Soil survey findings for the R.V. park site demonstrated that the soil type is not suitable for septic absorption fields due to poor filtration capacity.
Based upon the R.V. park’s location above the lateral canal; the direct drainage into the lateral canal from the R.V. park property; the multiple septic systems on the property and the subterranean moisture in the area that drains into the lateral canal; soil survey findings; the positive STEC samples from the lateral canal at the R.V. park and at the farm; and the fact that water from this section of canal services the suspect farm and only one other farm, it was determined that this R.V. park is a reasonably likely potential source of the outbreak contamination.
Our investigation also identified a potential seasonal source of STEC: the fall and winter grazing of sheep on harvested alfalfa and hay fields. We were unable to obtain any specific information on the seasonal sheep grazing practices and locations. The grower stated that the nearest sheep to the farm this year were in growing fields approximately five miles west of the suspect farm. Based upon the design of the irrigation canal system we determined that any potential contamination of the irrigation water from westerly growing fields is not a reasonably likely source of the outbreak contamination. However, due to the lack of comprehensive information, seasonal sheep grazing cannot be completely ruled out as a potential source of the outbreak pathogen.
The main irrigation canal from the Colorado River splits into two canals at a location west of the housing development. This is known as the Wellton-Mohawk split; one canal delivers water to the Wellton valley, the second canal delivers water to the Mohawk valley. Water emitting from a float chamber pipe at the Wellton-Mohawk split tested positive for Stx2-producing STEC, but did not match the outbreak strain. The float chamber controls the gates that feed water into the main Wellton canal.
Our inspection of the irrigation canal system found one additional potential contamination point that is not associated with a permanent source of STEC. On the main Wellton canal at the lateral canal split that supplies the farm there is a pump with an attached open end hose. The pump and hose are not secured, and there is open vehicle access to the edge of the canal at the pump. There is erosion evidence of drainage into the main Wellton canal at this point. During this investigation there were at least two recreational vehicles camped on undeveloped land within one mile of the pump and hose. The fact that this area is open to vehicles and the pump and hose are unsecured make it possible for an R.V. owner to dump and rinse out his R.V. septic system into the main Wellton canal at the lateral canal split that supplies the farm. Soil collected from the eroded drainage at this site tested positive for Stx2-producing STEC, but did not match the outbreak strain.
The working hypothesis is that potential runoff due to higher than average cumulative precipitation in the watershed might have increased the land-based microbial loadings into the irrigation canal water, which may have contaminated the crop with the outbreak pathogen. However, no definitive evidence was found to substantiate this hypothesis. The key findings of hydrologic analysis of watershed conditions prevalent during the growing and harvesting of the suspect crop include the following:
- The potential sources of contamination in the vicinity of the Wellton Mohawk canal network were determined based on evidence of runoff and other physiographic conditions, including slope.
- Surface water from the Wellton Mohawk Irrigation and Drainage District was the only source of water for the irrigation of the suspect crop.
- There was higher than average cumulative precipitation in the watershed during the late growing season. Specifically high levels and intensity of rainfall in January may have created higher runoff potential and increased loading of microbial contamination to the canal network.
- No evidence of influence from the Gila River during high flow conditions was found on and around the farms. Similarly no hydrologic connection was found that might have influenced the microbiological quality of the water in the canal network from the Gila River.
The Lower Gila Watershed is an agricultural/rural watershed where canal water is the main source of irrigation water. The surface waters in agricultural watersheds have a background load of microbial contaminants from different sources in the watershed including wild animals, grazing animals, irrigation return flow, rodents and other sources. Characterization of these background microbial contaminant loads from miscellaneous potential sources/land uses will be helpful in understanding the trends of fecal contaminant loading rates in the canal system.
The degree of change or the shift to the watershed as a result of a precipitation event should be evaluated as a potential contributing factor for pathogen contamination. This degree of change or shift is also dependent on the magnitude and duration of the precipitation event and the runoff potential from such an event. Furthermore, any grazing activity in and around the farms just before such a weather event further exacerbates the situation by potentially increasing the microbial contaminant load in the canal as well as in the immediate environ of the farms, thus potentially increasing the risk of contamination exponentially as compared to the background risk level.
It is therefore recommended that a thorough characterization of how rainfall and other weather events impact the microbial loads of a watershed, the typical background levels of microbial contaminants in the canal system, and all potential sources of contamination to the watershed, be performed in order help identify conditions that may warrant extra protective/preventive measures by the growers and the harvesters who use the canal as the source of their water. It may be difficult to completely restrict the access of wild animals and grazing animals to the areas close to the canal system; however, the quantification (or assessment) of the degree of loads after a rainfall event of certain intensity and duration, would allow for the implementation of appropriate precautionary measures on farms to prevent contamination of crops.
Below are potential preventive control strategies that may help to reduce the possibility of future contamination and outbreaks that relate to the specific circumstances associated with this environmental assessment. FDA is sharing these strategies with the relevant federal, state, and local government agencies and the firm. FDA is sharing what it learned during the environmental assessment to add to the body of information that currently exists on potential risk factors and preventive control strategies. Over time, FDA may use this information, in conjunction with the results of other environmental assessments and other appropriate data, to inform the development of guidance and rulemaking.
Potential preventive control strategies for this specific situation include:
Installing physical barriers to prevent drainage and pet waste from the R.V. park from entering the canal.
- Surveying the R.V. park to determine if the soil moisture observed above the irrigation canal is due to failure of the R.V. Park septic systems.
- Repairing and maintaining the R.V. park septic system to prevent system failure in the future.
- Limiting access of grazing livestock to fields that do not have access to irrigation canals, or installing barriers along canals located in grazing fields to prevent grazing livestock from contaminating irrigation canals.
- Securing the pump and hose along the main Wellton canal to prevent unauthorized use and potential canal contamination.
- Developing and implementing microbiological monitoring protocols for the canal water system, particularly in times of increased precipitation or abnormal weather events.
- Conducting sanitary surveys of the water source and distribution system for water used in produce production operations at the beginning of the production season and after any changes in practices and conditions, particularly unusual weather events.
- Monitoring of water quality by growers and harvesters during and immediately after a precipitation event.
- Conducting a microbiological survey of agricultural waters in the Yuma growing area.
The systems-based approach used for this environmental assessment allowed for discovery of important potential environmental risk factors that would not typically be explored by conventional investigation methods. Learning the possible contributing factors to an outbreak and designing preventive control strategies helps the FDA achieve its long-term goal of preventing outbreaks before they occur.
Important environmental risk factors that were uncovered during the course of this assessment include the impact of unusual weather events and human recreational activity on the microbiological quality of waters used in agricultural production operations. More studies and surveys are needed to determine the effects of precipitation events on runoff potential in the watershed, and resulting changes in the microbial loadings to irrigation canals and irrigation water quality. The findings of this environmental assessment also underscore the importance of using water of adequate quality for its intended use to ensure safety, particularly for applications that will likely contact the edible portion of the plant, and highlights the importance of producers having knowledge about their water source and potential sources of contamination that influence the sanitary quality of that water. (See Guidance for Industry: Guide to Minimize Microbial Food Safety Hazards for Fresh Fruits and Vegetables, FDA, Oct. 26, 1998, section II [Water]. FDA has also issued a draft guidance, Guidance for Industry: Guide to Minimize Microbial Food Safety Hazards of Leafy Greens; Draft Guidance, FDA, July 2009, section V [Production and Harvest] part B [Water].)
1FDA acknowledges that the term "environmental assessment" is also used to describe a specific type of analysis performed in compliance with the National Environmental Policy Act of 1969 (NEPA). FDA's use of the term in this document is intended only as defined herein, and not for any purpose under NEPA.
2FDA has concluded that the supplier/customer relationship between the farm and the distributor is confidential commercial information (CCI). Utilizing the “mosaic effect” approach recognized by the courts, FDA must consider that the distributor has already been publicly identified in the lettuce recall. As a result, if FDA were to disclose the name of the source farm, it would necessarily reveal the supplier/customer business relationship between the farm and the distributor. CCI is exempt from FOIA’s disclosure obligations under 5 U.S.C. 552(b)(4); and the Trade Secrets Act, 18 U.S.C. 1905, prohibits individual government employees from disclosing CCI “to any extent not authorized by law.”