INTRODUCTION
Verocytotoxin-producing Escherichia coli (VTEC) serotype O157 is an important cause of gastrointestinal infection in humans and has the potential to cause haemolytic uraemic syndrome and death [Reference Hawker1]. The incubation period generally ranges from 6 h to 10 days, with a median of 3–4 days [Reference Hawker1, Reference Heymann2]. The major reservoirs of E. coli O157 are cattle and other farm ruminants that shed the pathogen transiently in their faeces. The infection is mainly acquired by ingestion of contaminated food derived from infected animals or cross-contaminated during the preparation process [Reference Heymann2]. It can be also transmitted by direct contact with infected animals and their environment, or by contact with untreated and recreational water [Reference Heymann2–Reference Ihekweazu4]. Secondary spread by the faecal–oral route from infected individuals is common.
Most E. coli O157 infections are sporadic. In England and Wales the Health Protection Agency (HPA) electronic Foodborne and Non-foodborne Gastrointestinal Outbreak Surveillance System (eFOSS) reported from 10 to 17 VTEC O157 outbreaks annually between 2006 and 2010 [5]. National enhanced VTEC surveillance, introduced in England in 2008, collects a standard minimum dataset for each VTEC case in order to identify linkages and improve outbreak recognition [6]. The HPA recommends that local microbiologists send all presumptive VTEC O157 isolates for confirmation and typing to the HPA Gastrointestinal, Emerging and Zoonotic Infections Reference Laboratory (GEZI-RL) in Colindale where isolates are distinguished mainly by phage typing (PT), selected virulence typing [genes for verocytotoxins (VT)] and multi-locus variable-number tandem-repeat (VNTR) Analysis (MLVA).
In this paper we present a small outbreak of E. coli O157 that occurred in Plymouth in summer 2011. Plymouth, situated on the south west coast of England, is mostly rural and the incidence rates of E. coli O157 are higher than the national English average. From 2007 to 2011 the incidence rates ranged from 2·23 to 3·05/100000 inhabitants of the South West (South) Health Protection Unit [SW(S) HPU] area compared to 1·50 to 2·15/100000 inhabitants of England. Identification of clusters and outbreaks of E. coli may be difficult and relies on regular systematic reviews of enhanced VTEC surveillance and laboratory data to identify possible links or common exposures. When we were alerted to four reports of E. coli O157 in one week we investigated with the aims of determining whether it was an outbreak, to identify and control the source of the infection, and to prevent further spread.
METHODS
Detection of the outbreak
Between 5 August and 11 August 2011 laboratory staff at a hospital in Plymouth reported four females with E. coli O157 infection to the SW(S) HPU; the women were aged between 57 and 69 years and lived locally. In completed surveillance questionnaires we noted that three of the reported individuals ate shellfish (two mentioned crab) outside the home at different food venues in Plymouth. All four isolates were E. coli O157 PT21/28 VT2 with an indistinguishable MLVA pattern, confirmed at GEZI-RL. On 15 August 2011 we received reports of two adult males who visited Plymouth during their holidays and developed diarrhoea after consumption of a shared crab sandwich purchased at a food venue in Plymouth. They were subsequently confirmed to have E. coli O157 with the same phage type and MLVA pattern.
On 16 August 2011 we convened an outbreak control team (OCT) to identify and control the source of the infection and to prevent further spread. The OCT was composed of multi-agency representatives, both local and regional. We agreed on active case-finding, an epidemiological analytical study, environmental and microbiological investigations, communication strategy, and institute-relevant control measures.
Case definition and case-finding
We defined a case as an individual with onset of diarrhoea on or after 20 July 2011, residing in or visiting Plymouth in the 10 days before illness, who had either laboratory diagnosis of E. coli O157 PT21/28 VT2 with the same MLVA pattern (confirmed case) or had an epidemiological link to a laboratory-confirmed case (epidemiologically linked case).
We searched for further cases by (a) alerting general practitioners (GPs) and microbiologists in Plymouth and adjacent areas and asking them to report any suspected cases to us; (b) requesting the GPs to search retrospectively in clinical records for any patients with an illness compatible with E. coli infection; (c) reviewing information from VTEC surveillance questionnaires and the laboratory database for the South West region since mid-July 2012 to look for common exposures and possible links; (d) reviewing, in collaboration with colleagues at GEZI-RL and the Gastrointestinal Division at HPA Colindale, the national VTEC database, to identify possibly linked cases reported from other regions in England and Wales. As part of the wider communication we alerted health protection units and environmental health departments across England.
Case-control study
We conducted a case-control study to test the hypothesis that infection with E. coli O157 PT21/28 VT2 was associated with eating crab obtained from food venues (restaurants, cafes, pubs, etc.) in Plymouth in the 10 days prior to developing diarrhoea.
We used the case definition stated above and defined controls as individuals aged ⩾18 years, residing in Plymouth. We excluded cases and controls who travelled outside the UK or who had a close contact with other individuals suffering from diarrhoea or vomiting in the 10 days before symptom onset of cases, and before the interview day of controls. We aimed to recruit three healthy controls for each case from people who had had the opportunity to eat out in Plymouth, in order to detect odds ratios (ORs) above 15 as significant at the 5% level. We asked each case to nominate controls from adult family members, friends or acquaintances, not sharing the same household. Where case nomination was not appropriate (e.g. cases who visited Plymouth) or possible, controls were selected through voluntary staff at the local government and health commissioning office.
We designed a questionnaire asking about consumption of seafood, meat and other food items, such as salad leaves and sprouts that might be consumed with fish or meat. We separated food history into two sections, first asking about food eaten out/away from home including takeaways, and second, about food prepared and eaten at home. We also collected information on some other known risk factors for E. coli O157, e.g. animal exposure. Questionnaires were completed over the telephone or self-administered by email following instructions given by members of the OCT.
Data from questionnaires were entered into an EpiData file and analysed. We summarized personal characteristics of cases and controls, such as age and sex, and compared them using the Wilcoxon rank-sum test. We analysed exposures between cases and controls to the various food items and calculated ORs and 95% confidence intervals (CIs) for all exposures to describe the magnitude of any possible association between the food item and illness. Fisher's exact test was conducted because of small cell sizes. Analysis was performed in Stata v. 10.0 (StataCorp LP, USA). Exposures with an estimated OR > 1 and P < 0·1 were deemed eligible for inclusion in the multivariable analysis. Further, we analysed exposures between cases and controls, separately for controls that were case-nominated and controls that were recruited from local staff.
Environmental and microbiological investigations
In conjunction with Plymouth City Council we undertook hygiene inspections of the food venues mentioned by cases. We also visited other Plymouth food venues with crab on the menu and a major shellfish processor in Plymouth that were not associated with our cases.
As there was no leftover crab meat available from the dishes eaten by cases at the time of inspections, we sampled supplies of fresh crab meat at two venues mentioned by cases where supplied meat was available. Subsequently we sampled supplies of crab meat, both frozen and fresh, at another three Plymouth venues that were not implicated in the outbreak but had reported a similar origin of crab meat as the implicated places. Further, we took environmental swabs and crab meat samples from the Plymouth shellfish processor. All samples and swabs were examined at Bristol HPA Laboratory.
Cases had obtained crab from a number of different food venues. We interviewed the owners of these venues in order to identify and trace the source of crab meat and other shellfish.
RESULTS
Case-finding
A total of nine cases of E. coli O157 were identified in this outbreak. Eight cases, aged between 35 and 69 years (seven females), were confirmed with infection by E. coli PT21/28 VT2, with an undistinguishable MLVA pattern. The ninth case, without microbiological confirmation, was epidemiologically linked by time and place to one of the confirmed cases. Six were residents of and three visitors to Plymouth. Symptom onset of confirmed cases ranged from 30 July to 15 August 2011 (Fig. 1), seven reported bloody diarrhoea and two required hospitalization.
Onset dates of confirmed cases of E. coli O157 PT21/28 VT2, according to resident/visitor status and sharing of meals, Plymouth, UK, July–August 2011 (n = 8). OCT, Outbreak control team.
Cases had eaten shellfish purchased from food venues in Plymouth in the 10 days prior to developing diarrhoea. Eight ate crab and one had prawns. As shown in (Fig. 2), cases obtained the shellfish at five different food venues. Some cases were linked in time and place, attending the same food venue at the same time.
Traceback of crab meat, from cases (n = 9) to the food venues where the shellfish were obtained, and from the venues to the source supplier X, E. coli O157 outbreak, Plymouth, UK, July–August 2011.
Case-control study
We included only eight confirmed cases as the epidemiologically linked case was not able to be contacted. Cases nominated a total of six controls. Twenty-two (79%) additional controls were recruited from staff at the local government and health commissioning office in Plymouth.
Age and sex distribution of cases: median age 61 years [interquartile range (IQR) 47–68], 75% females, compared to controls: median age 55 years (IQR 43–62), 61% females, was not statistically significantly different (P = 0·32 and 0·46, respectively).
Cases were more likely to have eaten crab outside the home than controls (88% vs. 11%; unmatched OR 58, 95% CI 4–2725). Multivariable analysis was not undertaken as there were no other exposures with an OR > 1 and P < 0·1 (Table 1).
Frequency of exposures in cases (n = 8) and control (n = 28), case-control study, E. coli O157 outbreak, Plymouth, UK, July–August 2011
OR, Odds ratio; CI, 95% confidence interval.
* Exposure: ‘out’ refers to food eaten out/away from home including takeaways; ‘home’ refer to food prepared and eaten at home.
In analyses performed separately for case-nominated controls and controls recruited from local staff, we still found a statistically significant association between infection and eating crab. Cases were more likely to have eaten crab outside the home than case-nominated controls [7/8 (88%) vs. 2/6 (33%), OR 14, 95% CI 1–785], as well as staff controls [7/8 (88%) vs. 1/22 (5%), OR 147, 95% CI 6–6901].
Environmental and microbiological investigations
During the inspection of food venues mentioned by cases we found minor hygiene concerns and poor practices of food handlers at one venue. These concerns were dealt with formally, serving Hygiene Improvements Notices and emphasizing the need for training of food handlers. Following visits to the other Plymouth food venues with crab on the menu and the major shellfish processor in Plymouth we did not identify any hygiene issues.
Following the interviews with the owners of the food venues we discovered that the crab meat in four of the five implicated food venues was supplied by one crab supplier (supplier X) who was not registered with the local authority (LA), i.e. unapproved in legal terms as a supplier, as well as unregulated under LA procedures, e.g. inspections and food safety checks (Fig. 2).
Supplier X was interviewed under formal caution on suspicion of supplying food unfit for human consumption and for supplying food from an unapproved establishment, illegally. It was the legalbasis under which a prosecution could be brought.
Supplier X fished for crabs in the sea locally using his own boat. We identified a number of domestic addresses as sites where the crabs were processed. Crab processing usually includes boiling or pasteurization, followed by hand-picking of the boiled meat and packing it for distribution. The process should be a subject of standard food safety control measures, regulated by hazard analysis and critical control point (HACCP) principles. These domestic addresses were not registered or approved for food production and processing, and were not regulated. We were not allowed to access them for inspection.
The distribution of the processed crab meat to food outlets was based on local acquaintances, agreed by word of mouth. The area of distribution of the supplied crab was limited to Plymouth and the last distribution occurred on 15 August 2011.
The food samples of supplied crab meat and environmental swabs that we took at different food venues and at the major shellfish processor did not show any microbiological evidence of E. coli O157. However, some samples revealed microbiological contamination of E. coli of a different serogroup and other pathogens (Table 2).
Environmental and food samples taken for microbiological testing and the results, E. coli O157 outbreak, Plymouth, UK, July-August 2011
* Major shellfish processor in Plymouth.
† Reportedly these venues received meat from supplier X, no cases were associated with them.
Control measures
During the interview, under caution, the unregistered supplier was formally warned of the consequences of continuing to supply crab processed in unapproved premises thereby committing offences under the Food Safety Act. We have no evidence to suggest that this supplier is still supplying food businesses with crab meat. We also warned the caterers who sourced their crab meat from the unregistered supplier about potential adverse public health consequences of such action.
By 19 August 2011 we had most crab meat sourced from the unregistered supplier removed from the food venues, by agreement with the managers of the food venues or if necessary by seizure using legislative powers. Some additional crab meat that we identified later at another food venue not implicated in the outbreak was seized on 25 August 2011.
We provided written information to all food retail outlets in Plymouth emphasizing the importance of only buying food and ingredients from approved, registered suppliers.
No further cases of E. coli O157 associated with the outbreak have been reported.
DISCUSSION
We identified crab meat as a likely vehicle for infection in a community outbreak of E. coli O157 in South West England. We showed a large, statistically significant association between the infection and eating crab meat (OR 58). In traceback we identified a linkage to a supplier who was not registered with the LA for food processing and distribution. For all cases who ate crab meat we traced it back to the unregistered supplier. The outbreak strain was characterized by MLVA and in each confirmed case the strain was indistinguishable. While PT21/28 VT2 is not that unusual, the MLVA pattern has a high degree of discrimination [Reference Lindstedt7] and provided microbiological evidence of a link between cases. Nevertheless we did not isolate the outbreak strain of E. coli from any food or environmental samples available at the time of investigation and there was no leftover crab meat available from the dishes eaten by cases for microbiological testing.
In our investigation we did not obtain sufficient information on crab fishing, its handling and processing before distribution. Supplier X did not provide a stool sample for microbiological testing. With respect to how the crab meat may have been contaminated it seems unlikely that the crab was cross-contaminated at food venues before serving as a number of venues were implicated and food items were obtained at different times. Contamination or cross-contamination was likely to occur at some point(s) before the crab meat was distributed.
Crabs could have been infected at the outset in their sea habitat as the coastal water can be polluted by faeces from domesticated animals, by influx of water from field streams or sewage overflow [Reference Ihekweazu4, Reference Gourmelon8–Reference Fu10]. Use of a personal boat with no toilet and hand-hygiene facilities could have contributed to contamination during transportation. Processing of the crabs, including boiling, hand-picking and handling of the boiled meat, at domestic addresses was not regulated according to HACCP principles, with no records on storage temperatures, boiling lengths and temperatures, personnel hygiene, etc. Supplier X could have been a healthy carrier of E. coli O157. Contamination of fish and shellfish with E. coli due to poor hygiene and sanitary conditions has been documented in the past [Reference Manna, Das and Manna11]. However, due to lack of information on handling crabs prior to distribution, and microbiological evidence, we cannot conclude where or how the crab meat became infected with E. coli O157.
A variety of different food items have been implicated in VTEC outbreaks in the past, such as undercooked meat (especially beef), unpasteurized milk or cheese, apple cider, melons, vegetables (lettuce, cabbage, celery, spinach), and different sprouts (alfalfa, cress, radish sprouts) [Reference Ammon12–Reference Sproston25]. Faecal pollution from domesticated animals in the shellfish culture environment and E. coli O157 contamination of shellfish harvested at the coast have been reported [Reference Gourmelon8–Reference Fu10]. To our knowledge this is the first time that human E. coli O157 infection has been reported in association with the consumption of crab. This emphasizes the need to take into consideration novel, unknown food items as possible sources or vehicles for E. coli infection during outbreak investigations. Further, in this outbreak as in many others, such as the recent large German outbreak of E. coli O104:H4 [Reference Buchholz14], thorough food traceback is essential to identify the source and prevent further spread.
We investigated a relatively small outbreak identified through various enhanced surveillance. As crab dishes tend to be popular in the coastal Plymouth area, particularly in summer, we might have expected more cases if the infected crab meat was distributed on a larger scale. The unregistered supplier was supplying a small number of food venues only in Plymouth and it seems plausible that the population at risk from that distribution chain was also small. However, we identified some food venues that reportedly sourced crab meat from the unregistered supplier and were not associated with any cases. It is possible that some cases may have escaped our case-finding as they had mild symptoms or were asymptomatic.
The small study sample size could potentially under-power the detection of the association with food exposures more prevalent in the general population or with a hidden vehicle such as salad vegetables or other accompaniment. However, there were no other likely exposures identified through the trawling questionnaires and no common accompaniment reported. Five different food venues provided our cases with shellfish dishes, and four of them provided crab sourced from the unregistered supplier. As the implicated dishes were obtained over time ranging from 23 July to 10 August 2011, it appears unlikely that the infection would be associated with any other common food item or garnish.
Some cases were visitors to Plymouth and for these individuals nomination of controls was not appropriate. Therefore we had to choose a different approach. Our aim was to identify a control group that would be representative of the population cases, i.e. having an opportunity to eat out in restaurants and cafes in Plymouth. Taking into consideration the fact that Plymouth is a deprived area [26], people who were in an employment may have been more likely to dine out than the general Plymouth population. We sought controls among employees at the local government and health commissioning office as they may be more similar to cases in socioeconomic terms. If the staff-recruited controls were similar to cases in this regard then the strength of the association we found might have been artificially reduced compared to what we may have found if we had randomly selected controls from across the city.
On the other hand case-nominated controls might have been even more similar to cases in respect of the exposure. We analysed exposures separately, first for cases and case-nominated controls, and second for cases and staff-recruited controls. We found a statistically significant association between the infection and eating crab in both analyses, with a larger effect when the staff controls were considered.
Our investigations identified the likely crab meat vehicle and ensured its swift removal from retail outlets. The legal consequences of committing Food Safety Act offences were highlighted. We warned the catering sector by alerting food venues about potential adverse public health consequences of using unregistered food sources. The actions taken prevented further spread and will hopefully prevent similar future outbreaks.
ACKNOWLEDGEMENTS
We thank the following persons: the colleagues from the SW(S) Health Protection Unit and Public Protection Service at Plymouth City Council for their assistance and involvement in this investigation. Further, we thank the laboratory staff at HPA Colindale, GEZI-RL, and HPA Microbiology Services Bristol for their assistance with sample testing and typing, and we thank EPIET coordinators for their scientific advice.
DECLARATION OF INTEREST
None.
