MEAT SCIENCE

Meat Science 74 (2006) 7688 www.elsevier.com/locate/meatsci

A review of quantitative microbial risk assessment in the management of Escherichia coli O157:H7 on beef
Geraldine Duy
b

a,*

draig Nally b, Stephen O Brien a, Francis Butler , Enda Cummins b, Pa

a Department of Food Safety, Ashtown Food Research Centre, Teagasc, Ashtown, Dublin 15, Ireland Biosystems Engineering, School of Agriculture, Food Science and Veterinary Medicine, University College Dublin, Earlsfort Terrace, Dublin 2, Ireland

Received 15 March 2006; received in revised form 24 April 2006; accepted 24 April 2006

Abstract Since Escherichia coli O157:H7 rst emerged as a food borne pathogen in the mid 1980s, it has been linked to many cases of food poisoning across the world. While multiple sources and routes of transmission for this pathogen are now recognised, beef and beef products remain an important vehicle of the pathogen and continue to be linked to outbreaks across the developed world. Much research has been directed at E. coli O157:H7 transmission, survival and control in the beef chain and this paper presents an overview of current knowledge on this pathogen in the beef chain from primary production through slaughter, processing, distribution, nal preparation and cooking. In order to strategically manage E. coli O157:H7 and to devise approaches to reduce the public health risk posed, many national and international groups have applied quantitative risk assessment techniques to model the risk posed by E. coli O157:H7 in beef, particularly in ground/minced beef which is most often linked with infection. This paper reviews these quantitative risk assessments and their application in managing the risk posed by E. coli O157:H7 in beef. 2006 Elsevier Ltd. All rights reserved.
Keywords: E. coli O157:H7; Quantitative risk assessment; Beef

1. Introduction to quantitative risk assessment Risk analysis is a valuable tool in the management of microbial food safety issues and can provide a systematic approach for the regulatory authorities and the food industry to control the risk posed by a pathogen in a particular food commodity. Risk analysis consists of three elements: risk assessment, risk management and risk communication. Risk assessment is the scientic part of the process in which the hazards are identied and the risk posed by that particular hazard (i.e. pathogen) is calculated. The principles of risk assessment including the four stages involved (hazard identication, exposure assessment, hazard characterisation and risk characterisation) are outlined by the Codex Alimentarius Commission (Codex, 1999).

Each of the stages is summarised below. 1.1. Hazard identication A hazard is dened as an agent having an adverse eect on the public health of the human population and may pose a short term, chronic, or fatal risk to a person. The identication of microbial hazard associated with a particular food is generally based on information generated from routine microbial analysis of the commodity or from an epidemiological linkage of a particular pathogen with a case of food borne infection. 1.2. Exposure assessment Exposure assessment is a quantitative estimation of the presence of a contaminant in a serving of food at the time of consumption, or as close to this stage as is scientically possible and practical. However, the nal estimation of the

Corresponding author. Tel.: +353 18059500; fax: +353 18059550. E-mail address: Geraldine.Duy@teagasc.ie (G. Duy).

0309-1740/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.meatsci.2006.04.011

G. Duy et al. / Meat Science 74 (2006) 7688

77

numbers and prevalence of a pathogen in the food is often based on an accumulation of data on the prevalence and numbers of pathogen at key points in the food chain with data included on how particular stages in the food chain aect the numbers/prevalence of the pathogen. The nal step in the process estimates the amount of contaminant in a single serving, with information on the typical amount of food consumed in a serving procured from nutritional databases. The exposure assessment model can be deterministic, i.e. derived using single data points along the food chain. However, this approach may result in outlier values being ignored and thus under or overestimating the risk. A more common approach is to use a probablistic or stochastic analysis, in which a distribution curve representing all data is used as opposed to a single point estimate. Typically a Monte Carlo analysis is used to include data from all the distributions along the chain and is done using software such as @Risk (Palisade, NY, USA). In these analyses, a single data point is chosen at random from each distribution curve and used to calculate an outcome. The process is repeated several thousand times (multiple iterations) with a dierent data point in each distribution chosen each time and with the nal output being based on all the iterations. The error in the predicted risk may be due to variability or uncertainty, and there is increasing emphasis being placed on quantifying and separating the impact of both uncertainty and variability in risk assessments (Cohen, Lampson, & Bowers, 1996; Pouillot, Beaudeau, Denis, & Derouin, 2004). 1.3. Hazard characterisation Hazard characterisation relates exposure to a hazard with the probable public health outcome (illness/death). A doseresponse relationship can be used to estimate the amount (number) of pathogens which causes illness. The data used in generating doseresponse models are derived from a variety of sources including human clinical trials, epidemiological studies based on food poisoning outbreaks, animal clinical trials, in vitro studies using cell lines, biomarkers or expert opinion. In some cases, the dose responses will describe the susceptibility of dierent populations, i.e. general population and immunocompromised. 1.4. Risk characterisation The nal stage in the process estimates the adverse public health eect, or risk as a consequence of exposure to the hazard. This may be a prediction of illness per typical serving or calculated as an annual risk of illness. Depending on the hazard characterisation data available, the risk estimates may be broken down into age categories, based on dierences in immune status in order to identify groups which may be at higher risk following exposure to the contaminant. The risk characterisation model is generally developed using commercial software

such as @Risk or Crystal Ball (Decisioneering Inc., Denver, USA). These programs can separate the distribution for the overall risk prediction into uncertainty and variability to allow more complex risk estimation and analyses of the data. The generated model can be used to assess which parts of the chain signicantly aect risk or to assess the changes in predicted illness by incorporation of a new hypothetical risk mitigation strategy at a particular point in the chain. This paper reviews Escherichia coli O157:H7 in the farm to fork beef chain and examines how quantitative risk assessment models have been applied to establish and manage the risk posed. While other serovars of verocytotoxigenic E. coli (including E. coli O26, O111, O103, O145) are now emerging as a cause of similar illness to E. coli O157:H7 they are not addressed in this paper as there is still limited information on their transmission thorough the beef chain and they have not been included in any published quantitative risk assessment models. 2. E. coli O157:H7: human clinical aspects E. coli O157 is a member of the Enterhaemorrhagic group of E. coli (EHEC) and was rst implicated in infectious disease in the early 1980s (Riley et al., 1983). The symptoms of infection include bloody diarrhoea and severe abdominal pain. Haemolytic uraemic syndrome (HUS), a cause of acute renal failure, may be a complication of the illness, and neurological problems in the form of thrombotic thrombocytopaenic purpura (TTP) may also occur. Immuno-compromised patients, including young children and the elderly, are at particular risk of developing HUS. The time from exposure to onset of symptoms ranges from 1 to 14 days (Coia, 1998). However, with complications the illness may last many months and lead to permanent damage or even death. Pathogenicity is related to the ability of the organism to adhere to and colonise the human large intestinal epithelial tissue, forming attachment and eacing (AE) lesions and the production of verocytotoxins. The E. coli verocytotoxins are closely related to the Shiga toxin of Shigella dysenteriae and are typically bacteriophage encoded. There are two main classes of verotoxin: VT1, a homogeneous group of toxins, virtually identical to the Shiga toxin of Shigella and VT2, a heterogeneous group of toxins, more distantly related to the Shiga toxin. E. coli O157 with the eae gene and VT2 are most often associated with HUS in patients (Werber et al., 2003). Outbreaks of VTEC infections involving serovar O157 have now been reported from United States and Canada Bell et al. (1994) (Lisbea), Asia (Michino et al., 1998), Australia (Desmarchelier, 1996), Europe (Tozzi, Gorietti, & Caprioli, 2001), and Africa (Germani, Soro, Vohito, Morel, & Morvan, 1997). However, the majority of cases are sporadic and contribute signicantly to overall cases of infection. There is considerable variation in infection rates between dierent geographical regions. In Europe, the highest rates of infection are in Scotland with approximately

78

G. Duy et al. / Meat Science 74 (2006) 7688

4 cases per 100,000 (SCIEH, 2006). In the Republic of Ireland the incidence per 100,000 has ranged from a peak of 2.2 in 2003 to 1.3 in 2004 (HPSC, 2004). In Northern Europe infection rates are very low ranging from 0.04 per 100,000 in Norway and Finland to 1.1 in Denmark in 2000 although Denmark has in 2006, reported its rst general outbreak of E. coli O157 attributed to contaminated milk (Jensen et al., 2006). In 2004, the incidence rate for E. coli O157:H7 in North America was 0.9, a drop from 1.1 cases in 2003. In Asia, Japan has experienced the most problems related to E. coli O157:H7 with an average incidence rate of 2.74 per 100,000 between 1999 and 2004 (Sakuma, Urashima, & Okabe, 2006). A number of sources and reservoirs of E. coli O157 including beef and lamb, lettuce, sprouts, fruit juices, vegetables, raw milk, water have been implicated as vehicles of transmission (Bell et al., 1994; Cowden, Ahmed, Donaghy, & Riley, 2001; Hilborn et al., 2000; Michino et al., 1999). Person-to-person is also an important mode of transmission, particularly in day care centers (ODonnell et al., 2002) and direct contact with animals carrying the organism or with faecally contaminated mud (Anon, 1999; Crampin et al., 1999) are also recognised sources of infection. 3. E. coli O157:H7 in the beef chain Ruminants, in particular cattle and derived beef and beef products, are still considered to be one of the main

vehicles of the pathogen. Many outbreaks of E. coli O157:H7 have been linked to beef (Table 1) and analyses of sporadic cases of E. coli O157:H7 infection have identied undercooked beef as an important risk factor (Kassenborg et al., 2004; Mead et al., 1997). 3.1. Cattle The reported prevalence in cattle faeces varies widely depending on location and study (Table 2). The concentration of E. coli O157 in the cattle faeces is rarely reported but a small study in Australia on 22 positive samples using a most probable number (MPN) method indicated that numbers ranged from <3 to 2.4 104 MPN g1 (Fegan, Vanderlinde, Higgs, & Desmarchelier, 2004). The typical pattern of shedding in a herd is sporadic with epidemic periods of shedding interspersed with periods of non-shedding. In addition, it is usual that only a small number of animals in the herd are shedders. These epidemics occur mainly during warm weather, suggesting that environmental proliferation may play an important role in the epidemiology of E. coli O157:H7 (Hancock et al., 1998). Cattle are generally asymptomatic carriers of E. coli O157 with illness only reported in young calves (DeanNystrom, Bosworth, Moon, & OBrien, 1997). Studies have shown that weaned calves can be clinically healthy but colonised with E. coli O157 (Cray & Moon, 1995). Colonisa-

Table 1 Summary of selected outbreaks of E. coli O157 attributed to contaminated beef Location USA France USA USA USA USA Japan Japan Year 19921993 2005 2004 1995 2002 2003 2002 2005 Beef product and mode of transmission Undercooked beef burgers in fast-food outlets Frozen branded beef burgers at retail outlets Beef tacos at fast-food restaurant Roast beef at catered banquet Ground beef at retail outlets (multi-state) Frozen steak: blade tenderized and injected with marinade Grilled beef Beef oal served in a restaurant Number of cases 501 26 13 61 28 12 28 7 Number of HUS (deaths) 45 13 3 0 5 1 0 0 (3) (0) (0) (0) (0) (0) (0) (0) Reference Bell et al. (1994) Anon (2005) Jay et al. (2004) Rodrigue et al. (1995) Anon (2002) Laine et al. (2005) Tsuji et al. (2002) Maruzumi et al. (2005)

Table 2 Summary of studies on prevalence of E. coli O157:H7 in cattle faeces Country Spain Spain Ireland England/Wales Netherlands Norway Slovenia USA USA Canada Switzerland Japan Sample year 19931994 19981999 19981999 1999 1996 1995 19961997 2002 2001 19921993 2005 19921994 Type of cattle Beef/dairy cattle on farms Feedlot cattle Faeces of slaughtered cattle Dairy, suckler and fattening herds Dairy cattle on farms Heifers and milking cows Slaughtered cattle Cull dairy cattle Feedlot cattle Dairy cattle Dairy cattle Slaughtered cattle Number positive (%) 1/686 (0.1%) 55/471 (12%) 6/250 (2.4%) 219/4663 (4.7%) 75/1152 (7%) 6/1970 (0.3%) 2/250 (0.8%) 21/1026 (2.1%) 1087/10,662 (10.2%) 12/1478 (0.8%) 23/500 (4.6%) 7/387 (2%) Reference Blanco et al. (1997) Blanco et al. (2000) McEvoy et al. (2003) Paiba et al. (2003) Heuvelink et al. (1998) Vold et al. (1998) Andlovic and Marinsek (1997) Dodson and LeJeune (2005) Sargeant et al. (2003) Wilson et al. (1996) Kuhnert et al. (2005) Miyao et al. (1998)

G. Duy et al. / Meat Science 74 (2006) 7688

79

tion of cattle is short term (12 months) and shedding of E. coli O157:H7 within this period appears to be transient. Studies have, however, shown that within a herd there can be small number of persistent high shedders or super shedders with reported concentrations of the pathogen in naturally infected cattle of up 6.7 105 and 1.6 106 CFU g1 (Matthews et al., 2006). Colonisation of cattle occurs predominantly in the large intestine, and may especially target the follicle-associated epithelium (FAE) in the terminal rectum. A study comparing susceptibility of bovine colonic tissue from weaned calves and adult cattle to STEC showed no dierence in colonisation between these animals or indeed between grain-fed or forage-fed animals. However in tissue from well fed animals signicantly less STEC colonised the muscosal surface which may be a contributing factor in the increased shedding of STEC seen in poorly fed or fasted animals (Cobbood & Desmarchelier, 2004). Studies on horizontal transmission of E. coli O157:H7 in cattle housed together have shown that the pathogen can be transmitted from animal to animal following ingestion of the pathogen at relatively low levels and that animal grooming and the hide are important sources of transmission (McGee, Scott, Sheridan, Earley, & Leonard, 2004). 3.2. Slaughter and processing E. coli O157:H7 can potentially be deposited on the surface of beef carcasses during the slaughtering process as a result of cross-contamination from the bovine hide or gut contents. Elder et al. (2000) performed a survey in the USA to estimate the frequency of enterohemorrhagic Escherichia coli O157:H7 in faeces and on hides within groups of cattle from single lots presented for slaughter at meat processing plants and found that faecal and hide prevalence were signicantly correlated with carcass contamination. Carcass dressing operations which may reduce the number of E. coli O157:H7 include trimming of visibly dirty areas of carcasses, carcass washing (hot water), steam pasteurisation or treatment with decontaminants (organic acids). Carcass chilling is not likely to have any signicant eect on E. coli prevalence or counts (McEvoy, Sheridan, Blair, & McDowell, 2004). McEvoy et al. (2003) reported a reduction in prevalence on carcasses after chilling for
Table 3 Prevalence of E. coli O157:H7 in mince beef/beef products Country Ireland Italy U.K. Switzerland Argentina USA Botswana Sample year 20012002 20002001 19961997 2000 2000 20002003 20022003 Beef type Mince beef and beef burgers Minced beef Beef products Minced beef Mince beef Ground beef Beef cubes Minced beef Beef sausages

24 h and hypothesised that chilling may stress the bacterial cells due to the synergistic eect of low aw and temperature. Similarly, Gill, McGinnis, and Badoni (1996) reported a reduction in coliforms and E. coli on carcasses following cooling processes of between 0.5 log10 units and 2 log10 units. When the carcass is boned and trimmed in to smaller cuts, the concentration of E. coli O157:H7 should not increase if chill conditions are well controlled, but crosscontamination may occur to other cuts and surfaces with distribution of the pathogen throughout the ground meat. The survival characteristics of the VTEC organisms are generally similar to most other E. coli strains. Storage temperature, pH, water activity and salt content are the most important factors in relation to the survival and or growth of the pathogen in the food environment. The pathogen survives at food freezing temperatures (20 C). VTEC strains also show acid tolerance at the extreme range for E. coli organisms and are capable of surviving at a pH of 2.5 (Waterman & Small, 1996) and as such may pose problems in ready to eat low pH fermented meats. E. coli O157:H7 does not display any unusual tolerance to salt concentrations and can only grow in NaCl concentrations 66.5% (Glass, Loeelholz, Ford, & Doyle, 1992). 3.3. Distribution and nal preparation/cooking During distribution and storage, retail display etc. failure to maintain chill temperatures may allow growth of the pathogen. Improper handling of unpackaged meat, or leakage from wrapped packages may also lead to crosscontamination. Studies on beef and beef products in a range countries at retail stage have shown E. coli O157:H7 to be present into 0.435.22% beef/beef products (Table 3). Epidemiologic evidence in outbreaks cases of E. coli O157:H7 attributed to beef, continue to be associated with consumers/service sectors who do not understand the risks of handling raw meat and have inadequate hygiene handling practices and undercook meat. A 1996 US survey indicated that 19.7% of the population consumed pink (undercooked) hamburger at some time during the previous 12 months (CDC, 1998). In Ireland, a telephone survey to 500 people indicated that 87% of consumers prepare beef burgers well done, 12% medium and 1% cooked them rare (Mahon, Cowan, & Henchion, 2003).

Number positive (%) 43/1500 (2.8%) 4/931 (0.43%) 36/3126 (1.1%) 5/211 (2.3%) 6/160 (3.8%) 189/26,521 (0.71%) (7/134) 5.22% (5/133) 3.76% (3/133) 2.26%

Reference Cagney et al. (2004) Conedera et al. (2004) Chapman et al. (2000) Fantelli and Stephan (2001) Chinen et al. (2001) Naugle et al. (2005) Magwira et al. (2005)

80

G. Duy et al. / Meat Science 74 (2006) 7688

4. Human doseresponse models for E. coli O157:H7 The severe nature of the illness caused by E. coli O157 means that volunteer human doseresponse studies cannot be carried out. Surrogate dose models for E. coli O157:H7 have been developed from Shigella feeding studies to humans (Crockett, Hass, Fazil, Rose, & Gerba, 1996). A model based on E. coli O157:H7 fed to rabbits has been reported by Haas, Rose, and Gerba (2000) but it predicts a much higher dose (500 times as many organisms) to infect 50% of animals than the Crockett et al. (1996) model. Powell, Ebel, Schlosser, Walderhaug, and Kause (2000) put forward a model based on the infectious dose levels for humans infected with closely related bacteria S. dysenteriae and Enteropathogenic E. coli (EPEC). EPEC was chosen to represent the lower bound of an E. coli O157:H7 doseresponse function on the assumption that E. coli O157:H7 is unlikely to be less pathogenic than EPEC. S. dysenteriae was selected as an upper bound to the E. coli O157:H7 dose response function based on the assumption that E. coli O157:H7 is unlikely to be more pathogenic Shigella. An alternative approach has been to generate and validate dose response model based on quantitative data from actual human outbreaks (Strachan, Doyle, Kasuga, Rotariu, & Ogden, 2005; Teunis, Takumi, & Shinagawa, 2004). 5. Quantitative risk assessments for E. coli O157 in beef A number of quantitative risk assessment have been developed for E. coli O157:H7 in beef and some of these models are reviewed below. 5.1. E. coli O157:H7 in ground beef in Canada/North America (Cassin, Lammerding, Todd, Ross, & McColl, 1998) This model was developed for the production of ground beef burgers prepared at home. The exposure assessment modelled the farm to fork chain; from prevalence and concentration of the pathogen in cattle, thorough production of beef trim in factory, grinding at retail, nal preparation and consumption. 5.1.1. Exposure assessment data and assumptions The start point for the exposure model was the prevalence and concentration of E. coli O157:H7 in cattle faeces. Data on the concentration of E. coli in faeces was based on an experimental study by Zhao, Doyle, Shere, and Garber (1995) and ranged from undetectable by direct plate (<2 log10 CFU/g) to 5.0 log10 CFU/g. The data used to describe prevalence and distribution of the pathogen in cattle faeces were based on a number of studies carried out in Canada and USA between 1986 and 1994 and ranged from 0% to 3.1%. Processing was dened as the operations covered between slaughter of the animal and the packaging of the beef trim into 5 kg vacuum packs. This part of model quan-

titatively described the behaviour of E. coli O157:H7 during the beef carcass dressing operations and subsequent grinding/mincing of the beef. The prevalence and concentration of the pathogen on carcasses was assumed to be proportional to that in the cattle faeces, and the aggregate reduction in counts of E. coli O157:H7 on the carcass was assumed to be 12.5 log following all carcasses dressing procedures (trimming of visibly dirty areas of the carcass, decontamination, etc.). The prevalence in bags of trimmings (5 kg) was assumed to be higher than on carcasses due to mixing of meat from a number of carcasses in each bag and the level of contamination on the carcass surface (CFU/cm2) was correlated with contamination on the beef trim (CFU/g). The bags of trim (5 kg) were assumed to be sent to the retailer where they were ground and displayed in 3001000 g lots. Retail/home storage temperatures distribution were assumed to be a minimum of 4 C, mode 10 C and at a maximum of 15 C. In the nal preparation stage cooking practices based were based on US survey data indicating approximately 20% of consumers prepared rare or medium rare, corresponding to mean internal temperatures of 58.6 C or less. The amount of hamburger ingested in a single serving was assumed to be distributed normally with a mean of 83 g and a standard deviation of 48 g. 5.1.2. Outputs and risk estimate The exposure model calculated a mean prevalence for E. coli O157:H7 of 2.9% in contaminated retail packages (3001000 g) of fresh ground beef with a predicted number in contaminated packages of 87% <10 CFU; 90% <1000 CFU/package. The doseresponse model was a modication of that based on Shigella feeding studies (Hass, 1983) and assumed that the infectivity of E. coli O157:H7:H7 was similar to Shigella. The model considered increased risk in children and the elderly (i.e. HUS and mortality) based on epidemiological data. The predicted illness among the dierent population groups are shown in Table 4. 5.1.3. Risk analysis The model ranked the factors which were the most important predictors of risk (Table 5) and showed that concentration of pathogen in the faeces and host susceptibility had the most impact on predicted illness. Potential targets/interventions for risk management were assessed using hypothetical assumptions (Table 6). The mitigation strategy having the most eect was strict control of retail storage temperature and lowering the maximum temperature to 8 C from 10 C with a worst abuse case of 13 C, which resulted in an 80% predicted reduction of illness. 5.2. Shiga toxin-producing E. coli (STEC) in ground beef in Australia (Lammerding, Fazil, Paoli, & Vanderlinde, 1999) This model described the risk from shiga toxin-producing E. coli in ground beef derived from beef trimmings in

G. Duy et al. / Meat Science 74 (2006) 7688 Table 4 Probability of illness from consumption of a single serving of beef contaminated with E. coli O157:H7 Beef Ground Ground Ground Ground Ground beef beef beef beef beef Location North America North America Australia Australia USA USA Population group Average adult Children Average adult Children Average population Average population JuneSeptember OctoberMay Children Average adult Illness 5.1 10 3.7 105 6.4 104 4.6 104 9.6 107 1.7 106 6.0 107 2.4 106 1.1 106
5

81

HUS 3.7 106 6.44 105 4.60 105 4.2 109

Mortality 1.9 107 7.72 106a 2.3 106 5.9 1010

Model Cassin et al. (1998) Cassin et al. (1998) Lammerding et al. (1999) Lammerding et al. (1999) USDA-FSIS (2001) and Ebel et al. (2004) USDA-FSIS (2001) and Ebel et al. (2004)

Ground beef

Ground beef Beef Burgers

a

USA Ireland

USDA-FSIS (2001) and Ebel et al. (2004) Duy et al. (2006)

Note rate is for elderly people.

Table 5 Impact of various parameters along the beef chain on the probability of illness in consumed ground beef servings as determined by dierent risk assessment models Cassin et al. (1998) Sensitivity analysis of impact of factors on probability of illness in ascending order of importance Concentration of pathogen in faeces Host susceptibility Carcass contamination factor Lammerding et al. (1999) Concentration of pathogen in faeces Host susceptibility Dilution factor USDA-FSIS (2001) Ebel et al. (2004) Surface area of carcass contaminated Eectiveness of carcass chilling Max. population of E. coli O157 in ground beef serving Home storage temperature Duy et al. (2006) Initial count on bovine hide Cooking temperature Temperature abuse during transport and storage Hide to carcass contamination factor Hide Prevalence Change in numbers at carcass chilling

Cooking preference Retail storage temperature Decontamination during primary processing Growth during processing Retail storage time Prevalence in faeces Mass ingested

Temperature of cooking Temperature of retail display Mass consumed Washing Prevalence in faeces Trimming Weekend chilling

Table 6 Eect of dierent hypothetical risk mitigation strategies on reducing the probability of illness Intervention Lowering average retail storage temperature to 8 C from 10 C with worst abuse case of 13 C Pre-slaughter treatment/screening of cattle to reduce the concentration of pathogen shed in faeces such that all contamination levels above 4 log CFU/g were eliminated Information campaign targeting consumers to cook burgers resulting in a shift from 18.6% consuming rare or medium rare ground beef to 12% of such consumers Use of hot water decontamination giving expected 14 Log10 reduction in STEC numbers on carcasses Irradiation of de-boned and frozen trimmings with 1 kGy giving an expected reduction of STEC numbers of 1.31.8 Log10 Eliminating or implementing stricter temperature controls for over-weekend chilling such that the maximum proliferation limited to the same as overnight chilling Model Cassin et al. (1998) Lammerding et al. (1999) Cassin et al. (1998) Lammerding et al. (1999) Predicted reduction in illness (%) 80 80 46 25

Cassin et al. (1998) Lammerding et al. (1999)

16 16

Lammerding et al. (1999) Lammerding et al. (1999)

99.7 97

Lammerding et al. (1999)

20

82

G. Duy et al. / Meat Science 74 (2006) 7688

Australia and closely followed the above model of Cassin et al. (1998). 5.2.1. Exposure assessment data and assumptions Data on initial prevalence of STEC in faeces of Australian cattle was from studies in feed lots by Food Science Australia and ranged from 35.4% to 53.4%. The assumptions regarding carcass and trimming contamination were similar to the Cassin et al. (1998) model excepting that weekend chilling of a portion of carcasses was also considered and assumed to allow greater proliferation of the pathogen than overnight chilling. The size of the nal boxed trim was assumed to be 27.2 kg. Retail storage temperatures, cooking preferences and consumption data were as described for Cassin et al. (1998) model. 5.2.2. Risk estimate The doseresponse model employed was as for Cassin et al. (1998) and the expected probability of illness of 6.4 104 per serving for adults and 4.6 104 for a child under the age of 5 years (Table 4) was substantially higher than reported by Cassin et al. (1998) model presumably related to the much higher initial prevalence for STEC inputted to this model. 5.2.3. Risk analysis The eect that various steps in the beef chain had on the probability of illness were ranked (Table 5). As with Cassin et al. (1998), analysis indicated that concentration of STEC in faeces and host susceptibility had the greatest impact on predicted illness. The model was also used to estimate what eect dierent risk mitigation strategies would have on the probability of illness (Table 6). Hypothetical intervention strategies including a hot water decontamination of carcasses and irradiation of boxed beef trimmings had the greatest impact with a 99.7% and 97% reduction, respectively, in illness predicted. 5.3. E. coli O157:H7 in ground beef in North America (Ebel et al., 2004; USDA-FSIS, 2001) This was a baseline risk assessment for ground beef reecting the full range of current practices, behaviours and conditions in the farm to fork continuum (production, slaughter, processing, transportation, storage, preparation and consumption). 5.3.1. Exposure assessment data and assumptions Data on E. coli O157:H7 herd prevalence was taken from data published between 1997 and 2001 with seasonality taken into account. The mean prevalence in cull beef herds was 63%; and within herd prevalence was 3% during low season (OctoberMay), and 4% during the high season (JuneSeptember). In feedlot cattle the herd prevalence was 88%; the within lot prevalence was 9% and 22% during low and high prevalence seasons, respectively. Contamination of positive carcasses ranged

from <0.03 to 3.0 CFU/cm2 and data on prevalence in fresh ground beef predicted by model was adjusted by anchoring to actual survey data. The collective eect of interventions including decontamination (1st wash after dehiding: trim/wash/vacuum; 2nd wash: hot water or steam pasteurization) was estimated to give a cumulative 1.5 log reduction. Cooking practices were based on US survey data (approximately 20% prepared rare or medium rare). 5.3.2. Outputs and risk estimate The model predicted that 0.018% of prepared servings consumed during JuneSeptember (high prevalence season) and 0.007% of servings consumed during the remainder of the year were contaminated with one or more E. coli O157:H7. The doseresponse function was derived from epidemiological data for E. coli O157:H7 illnesses and the doseresponse curve was bounded by upper and lower doseresponse curves from surrogate pathogens S. dysentery and enteropathogenic E. coli (EPEC), respectively (Powell et al., 2000). The predicted risk of illness is shown in Table 4 and was 2.5 times higher in children (05 years) (2.4 106 per serving) than in the average population (9.6 107). Seasonality was estimated to have a very big eect with risk of illness during JuneSeptember predicted to be three times higher during OctoberMay. 5.3.3. Risk analysis Factors were ranked of in order of importance in predicting illness (Table 5). The eect of these factors on the occurrence and extent of contamination varied by season and type of cattle (feedlot herd or breeding herd). Occurrence and extent of E. coli O157:H7 contamination in cooked ground beef was in addition inuenced by the proportion of ground beef that is frozen; the maximum population density of E. coli O157:H7 in ground beef; storage temperatures and cooking. The US population risk was deemed to be inuenced more by the number of contaminated servings than the number of E. coli O157:H7 per serving. Various scenarios were modelled and indicated that the likelihood of nding E. coli O157:H7 through testing of manufacturing trim was substantively higher, by approximately a 5-fold dierence, than through testing of ground beef alone. 5.4. E. coli O157:H7 in steak tartare in The Netherlands (Nauta, Evers, Takumi, & Havelaar, 2001) Risk was estimated for the consumption of steak tartare patties, a lean (<10% fat) ground beef product typically eaten raw or partially raw, in The Netherlands. The model covered the farm to fork chain and dierentiated between size of slaughter operation, industrial and traditional processing, and traditional butcher versus industrial preparation of product.

G. Duy et al. / Meat Science 74 (2006) 7688

83

5.4.1. Exposure assessment data and assumptions The data used to describe prevalence of E. coli O157:H7 at primary production in dairy cattle, veal calves and veal bulls were based on studies conducted in The Netherlands on farm and in slaughterhouses. At farm level prevalence ranged from 0% to 25% and at animal level from 0% to 61%. At slaughterhouse, prevalence ranged from 0% to 33% at herd level and 0% to 15% at animal level. Concentration of pathogen in faeces of shedding animals was based on studies of Zhao et al. (1995). Tartare was assumed to be consumed raw 2.6% of the time and prepared (medium or well done) by remainder of consumers. Expert opinion was used to estimate a number of factors including faecal contamination of carcasses and percentage of patties thoroughly heated (20%). Growth of pathogen in steak tartare at retail and in the home was not assumed to be an important factor. 5.4.2. Outputs and risk estimate The model predicted the prevalence of E. coli O157:H7 to be higher in industrial ground beef contaminated batches than in traditional facilities. However, when made into raw steak tartare patties, prevalence was almost equal in traditional and industrial product. Pathogen numbers in contaminated product were smaller in industrially produced patties, a result of diluting contamination throughout large volumes of uncontaminated meat. Overall the exposure model predicted that 0.3% of raw steak tartare patties were contaminated; with about 64% of positive patties containing only 1 CFU, and only 7% contain more than 10 CFU. The doseresponse model was based on data from a single 1996 outbreak in Japan (Teunis et al., 2004) and resulted in a doseresponse similar to that from feed trials with Shigella which predicts the highest probability of illness per single cell ingested compared with alternate models. The predicted number of illnesses associated with the consumption of steak tartare in The Netherlands is 1284 cases per year, an incidence rate of 8 per 100,000 persons. By comparison, the incidence rate of STEC O157 illnesses from all sources, based on epidemiological data, is estimated to be 13 cases per 100,000. The authors suggested that the model prediction may be overestimating the number of illnesses associated with the product, related to the doseresponse model used which predicts a higher probability of illness than models used by other authors. Using an alternate doseresponse model (Powell et al., 2000), only 17 cases per year were predicted from the consumption of steak tartare. 5.4.3. Risk factors Scenario analysis was used to identify the parameters impacting on risk. The biggest eects on the risk estimate were from in farm prevalence, concentration of the pathogen in faeces and growth/inactivation on carcasses. Based on the model and expert opinion, the factors that were

deemed mostly likely to decrease the risk associated with the product were lowering prevalence and concentration of E. coli O157 in cattle, improved hygiene at slaughter or increased the frequency of industrial processing. Product control by monitoring at retail did not appear to be practical given the low prevalence and concentrations, and the assumption that growth during storage was unlikely. Intervention at the consumer level, using an information campaign to inuence preparation practices, was not part of the risk assessment model. However, this aspect was addressed through professional opinions from communication experts on potential eectiveness of an information campaign. It was concluded that this strategy was not likely to be favourable to risk reduction in the product. 5.5. E. coli O157:H7 in beef burgers produced in the Republic of Ireland (Duy et al., 2006) A quantitative microbial risk assessment (QMRA) model was developed for E. coli O157:H7 in beef burgers produced in the Republic of Ireland. The risk assessment model covered the slaughter process culminating in the production of boxed beef trimmings; mincing of beef, beef burger formation and retail distribution; and domestic storage, cooking and consumption. 5.5.1. Exposure assessment data and assumptions Initial data inputs to the model on the prevalence and concentration of E. coli O157:H7 were based on microbiological surveys on the pathogen in faeces (McEvoy et al., 2003) (2.4%) and hide (OBrien, 2005) (7.3%, Log100.13 Log104.24 CFU/100 cm2) of animals presented for slaughter at Irish beef abattoirs. The model outputs for prevalence and numbers of E. coli O157:H7 at key points in the model (beef trimmings, beef products at retail) were validated using microbiological surveillance data for E. coli O157:H7 on beef trimmings (Carney et al., 2006) and in beef mince/burgers at retail (Cagney et al., 2004) in the Republic of Ireland (see Table 7). The model assumed that contaminated hide and rumen contents were the vectors for cross-contamination to carcasses at hide removal and eviseration, respectively, and a cross-contamination factor was created based on Irish surveillance data for the pathogen on bovine hide (OBrien, 2005) in rumen contents (McEvoy et al., 2003) and on beef carcasses (Carney et al., 2006; McEvoy et al., 2003). The changes in E. coli O157:H7 numbers on contaminated carcasses during carcass dressing operations including trimming of visibly dirty parts of carcass; carcass washing, evisceration and chilling were estimated based on research studies in the literature on the impact of these operations on pathogen numbers (Gill et al., 1996; McEvoy et al., 2003; McEvoy et al., 2004). The potential increase in numbers of E. coli O157:H7 in the boning hall was assumed to minimal (McEvoy et al., 2004). A factor for estimating the transfer of contamination from carcass to trim was set in the model taking account of the surface area of carcass

84

G. Duy et al. / Meat Science 74 (2006) 7688

Table 7 Prevalence and numbers of E. coli O157:H7 at various sample points along the beef chain in Ireland Sample type Bovine hide Beef carcasses Head meat Beef trimmings Retail minced beef/burgers Sample numbers 1500 132 100 1351 1533 Number positive (%) 109 4 3 32 43 (7.3) (3.0) (3.0) (2.4) (2.8) Numbers present (Log10 CFU) 0.134.24/100 cm2 0.701.41/g 0.701.00/g 0.701.61/g 0.524.03/g Reference OBrien et al. (2005) Carney et al. (2006) OBrien et al. (2005) OBrien et al. (2005) Cagney et al., 2004

which was contaminated, the surface area of the trim, the weight of the trim and the number of trim in a box (27 kg). It was assumed that the beef trimmings were minced into 100 g beef burger patties. Input data for the retail/domestic part of the model was based on two main sources. Information on typical consumer storage and cooking practices in the domestic environment were derived from a questionnaire survey of consumers (Mahon et al., 2003). Data on storage temperatures at retail and in domestic refrigerators were also gathered from temperature studies in both environments (Kennedy et al., 2005; Carney et al., personal communication) and it was assumed that temperatures ranged between 7 and 16 C. Based on the survey of consumer cooking practices it was assumed that 87% of consumers prepare hamburgers well done, 12% medium and 1% cooked them rare. A temperature distribution was set for the cooking temperature based on the assumption that beef burgers are cooked to mean temperatures of 68.3 C (well done), 62.7 C (medium) or 54.4 C (rare). Consumption data gures for beef burgers were derived from an Irish Food Consumption Survey carried out by the Irish Universities Nutrition Alliance (www.iuna.net) (Mahon et al., 2003) and the serving size for a beef burger was set at a mean of 100 g. 5.5.2. Outputs and risk estimate The model indicated a mean simulated prevalence of E. coli O157:H7 on beef trimmings of 2.40% and a mean

count of 2.69 log10 CFU g1. This output was validated against a microbiological survey of E. coli O157:H7 on beef trim in Irish abattoir which indicated a prevalence of 2.36% and counts of 0.71.61 log10 CFU g1 which indicates that the model simulated values and the survey result were similar. The model indicated a mean simulated prevalence in fresh beef burgers of 2.9% and 2.2% in frozen burgers while the mean simulated counts in fresh and frozen burgers were log10 1.96 CFU g1 and log10 0.22 CFU g1, respectively. These predicted values were compared with microbiological survey data on prevalence and numbers of E. coli O157:H7 on these products on retail sale in the Republic of Ireland and shown to be similar (prevalence 2.8%; counts log100.51log104.03 CFU g1). The doseresponse used was based on the model of Powell et al. (2000) as per the USDA-FSIS model described above. The probability of illness caused by exposure to E. coli O157:H7 in fresh beef burgers is reported is for an average individual. It is acknowledged that this dose response relationship may be an underestimate for immune compromised individuals, however to try to create one for individual risk groups was not possible given the lack of a reliable a doseresponse relationship in these categories. The simulated probability of illness from a contaminated serving of fresh beef was 5.94 log (1.1 106) (Table 4). 5.5.3. Risk analysis Analysis of the risk model (by rank order correlation sensitivity analysis) (Table 5) indicated the initial preva-

Table 8 Theoretical performance objectives set for prevalence and concentration of E. coli O157:H7 in raw beef trimming and impact on probability of illness using quantitative risk assessment model of Duy et al. (2006) Performance objectives for Beef trimmings (27 kg lots) Prevalence (27 kg lots) 0.25% 0.25% CFU/g 0.01 0.001 Probability of illness per serving Mean 4.47 108 5th percentile 1.58 108 95th percentile 1 107 2.04 108 1.51 108 2.63 108 0.1% 1 4.67 105 1.0 107 2.45 103 0.1% 0.01 5.12 108 3.80 108 8.71 108 0.05% 1 1.95 106 4.57 108 5.12 105 0.05% 0.01 4.67 109 7.58 1010 1.29 108

Performance objectives for beef burgers (100 g) Prevalence 3% 3% CFU/g 10,000 1000 Probability of illness per servings Mean 1.4 106 5th percentile 2.75 108 95th percentile 7.93 104 4.5 106 2.75 108 7.76 104

2% 100 4.5 106 2.13 108 4.7 104

1% 100 2.3 106 1.02 108 2.4 104

0.5% 1 7.4 107 6.02 109 9.77 105

0.25% 1 5.01 107 2.88 109 5.49 105

0.1% 1 6.17 108 1.02 109 8.17 106

G. Duy et al. / Meat Science 74 (2006) 7688

85

lence and numbers of E. coli O157:H7 on the bovine hide (correlation coecient 0.62) had the greatest impact on overall probability of illness from E. coli O157:H7. A further important factors was the cooking preference (correlation coecient 0.57) and temperature abuse during retail storage temperature, during transport home and home storage was also deemed a signicant parameter inuencing model predictions (correlation coecient 0.48). To assess the impact of changing the prevalence and concentration of E. coli O157:H7 at the stage of boxed trim (27 kg) or formed beef burgers, on the overall predicted illness, hypothetical performance objectives (prevalence and CFU/g at a set point in the beef chain) were set (Table 8) and substituted into the Irish risk model for a fresh refrigerated beef burger. Keeping all other assumptions and parameters as for the original baseline model, the model was re-simulated to generate expected risk of illness. The results (Table 8) show that lowering the concentration of the pathogen in either beef trim or in the beef burger had a big impact on predicted illness. In beef trim lowering the concentration from 1 to 0.01 CFU/g (in a 27 kg lot) while maintaining a prevalence of 0.l% gave a 1000 fold reduction in illness (4.67 105 to 5.12 108) while in beef burgers lowering the prevalence by 0.5% (1.00.5%) and lowering the concentration from 100 to 1 CFU/g lowered the predicted illness by more than 10-fold (2.3 106 to 7.4 107). This type of data could be used to set targets within the beef chain to achieve a target for reducing illness linked to E. coli O157:H7. 6. Conclusions All the models described above vary in their illness predictions. Some of the variation can be accounted for by the dierent inputs for prevalence and concentration of E. coli O157:H7 in the dierent regions. Indeed any of the models could potentially be adapted and run with input data/practices (steps causing an increase or a decrease in number of E. coli O157:H7 in the chain) representative of other countries or regions to assess and manage the risk posed by E. coli O157:H7 in beef in that particular area. The models predictions are also greatly inuenced by the doseresponse models thorough which the exposure assessment is transposed to give a risk estimate. In the ve models described above there are three distinctly dierent doseresponse models used none of which give an ideal representation of the infectivity of E. coli O157:H7 for humans. The Cassin et al. (1998) and Lammerding et al. (1999) employed a Beta binomial model based on that of Hass (1983) which assumes that virulence is similar to that caused by S. dysenteriae. It is now considered that this model overestimates the risk posed as E. coli O157:H7 is likely to be less pathogenic than Shigella dysteteriae. The Nauta model (2001) employed a doseresponse model based on data from a single 1996 outbreak in Japan (Teunis et al., 2004) and resulted in a doseresponse similar to that from feed trials with Shigella which is considered to be an over-

estimation of the risk posed. Indeed when these authors ran the model with a dierent doseresponse (Powell et al., 2000) the risk of predicted illness was substantially lowered. The Ebel et al. (2004) and Duy et al. (2006) model both employed an adapted version of Powell (2000) which assumes that E. coli O157:H7 infectivity for humans is between the infectivity of EPEC and Shigella and thus has a considerably lower estimation of risk for E. coli O157:H7 than the Shigella based model. There is however now some debate about whether the Powell model is underestimating illness as a recent doseresponse model based on outbreak data from across the globe (Strachan et al., 2005) was found to be have an output closer to the Shigella model. Further research is necessary to develop models which more accurately predict human illness and models which can dierentiate between illness in the average population and at risk groups such as young children, the elderly and immunocompromised. Despite the dierent risk predictions there was a marked similarity in all models in terms of the factors having the most impact of risk. Four of the ve models demonstrated that the factor having the largest impact on predicted illness was the concentration of pathogen in the faeces or hide of the animal presented for slaughter. Therefore it is likely that eorts aimed at eliminating animals carrying very high levels of the pathogen from entering the beef chain would yield a good return in terms of risk reduction. Equally temperature abuse at retail also featured highly in most of the model risk factors. Two of the models indicated that if the average storage temperature was lowered to 8 C with the minimum temperature abuse of 13 C it would give a theoretical 80% reduction in illness. It is clear that E. coli O157:H7 remains a problem in the beef chain and while the risk of illness is lower than for other food borne bacterial pathogens such as Campylobacter or Salmonella, the potential severity of the illness makes E. coli O157:H7 a bigger public health threat. The application of risk assessment to model E. coli O157:H7 in the beef chain is most useful in identifying the points in beef chain which have the most impact on predicted illness and can thus be used as a guide to focus resources and eorts. The models can be applied to assess the impact that hypothetical interventions would have on overall predicted illness. On the other hand they may be used to set performance objectives without setting the prescribed manner for this to be achieved. Thus, quantitative risk assessment and risk analysis can be used to guide and direct risk management and to underpin regulatory policy. Current risk assessments for E. coli O157:H7 and indeed for other pathogens are limited by a lack of data in critical areas and going forward, better data is needed along the food chain, including more detailed qualitative and quantitative information on the pathogen along the chain being modelled, epidemiological data on outbreaks and sporadic cases illness and doseresponse models which more accurately describe the illness in humans and in at risk groups.

86

G. Duy et al. / Meat Science 74 (2006) 7688 with drinking unpasteurized apple cider Connecticut and New York, October 1996. Morbidity and Mortality Weekly Report, 46, 48. n Malo, A. T., & Harkin, M. A. Chapman, P. A., Siddons, C. A., Cerda (2000). A one year study of Escherichia coli O157 in raw beef and lamb products. Epidemiology and Infection, 124, 207213. Chinen, I., Tanaro, J. D., Miliwebsky, E., Lound, L. H., Chillemi, G., Ledri, S., et al. (2001). Isolation and characterization of Escherichia coli O157:H7 from retail meats in Argentina. Journal of Food Protection, 64, 13461351. Cobbood, R. N., & Desmarchelier, P. M. (2004). In vitro studies on the colonization of bovine mucosa by Shiga-toxigenic Escherichia coli (STEC). Epidemiology and Infection, 132, 8794. Codex Alimentarius Commission (Codex) (1999). Principles and guidelines for the conduct of microbiological risk assessment. Rome, Food and Agriculture Organization of the United Nations, CAC/GL-30. Cohen, J. T., Lampson, M. A., & Bowers, T. S. (1996). The use of twostage Monte Carlo simulation techniques to characterize uncertainty and variability in risk analysis. Human and Ecological Risk Assessment, 2, 939971. Coia, J. E. (1998). Clinical, microbiological and epidemiological aspects of Escherichia coli O157 infection. FEMS Immunology and Medical Microbiology, 20, 19. Conedera, G., Dalvit, P., Martini, M., Galiero, G., Gramaglia, M., Goredo, E., et al. (2004). Verocytotoxin-producing Escherichia coli O157 in minced beef and dairy products in Italy. International Journal of Food Microbiology, 96, 6773. Cowden, J. M., Ahmed, S., Donaghy, M., & Riley, A. (2001). Epidemiological investigation of the central Scotland outbreak of Escherichia coli O157 infection, NovemberDecember 1996. Epidemiology and Infection, 126, 335341. Crampin, M., Willshaw, G., Hancock, R., Djuretic, T., Elstob, C., Rouse, A., et al. (1999). Outbreak of Escherichia coli O157 infection associated with a music festival. European Journal of Clinical Microbiology and Infectious Diseases, 18, 286288. Cray, W. C., Jr., & Moon, W. H. (1995). Experimental infection of calves and adult cattle with Escherichia coli O157:H7. Applied and Environmental Microbiology, 61, 15861590. Crockett, C. S., Hass, C. N., Fazil, A., Rose, J. B., & Gerba, C. P. (1996). Prevalence of shigellosis in the US: consistency with doseresponse information. International Journal of Food Microbiology, 30, 8799. Dean-Nystrom, E. A., Bosworth, B. T., Moon, H. W., & OBrien, A. D. (1997). Escherichia coli O157:H7 requires intimin for enteropathogenicity in calves. Infection and Immunology, 66, 45604563. Desmarchelier, P. M. (1996). Foodborne disease: emerging problems and solutions. The Medical Journal of Australia, 165, 668671. Dodson, K., & LeJeune, J. (2005). Escherichia coli O157:H7, Campylobacter jejuni, and Salmonella prevalence in cull dairy cows marketed in northeastern Ohio. Journal of Food Protection, 68, 927931. Duy, G., Cummins, E., Nally, P, O Brien, S., Carney, Butler, F., et al. (2006). E. coli O157:H7 in beef burgers produced in the Republic of Ireland: a quantitative microbial risk assessment. Report published by Teagasc, Ashtown Food Research Centre, Ashtown, Dublin 15, Ireland. Ebel, E., Schlosser, W., Kause, J., Orloski, K., Roberts, T., Narrod, C., et al. (2004). Draft risk assessment of the public health impact of Escherichia coli O157:H7 in ground beef. Journal of Food Protection, 67, 19911999. Elder, R. O., Keen, J. E., Siragusa, G. R., Barkocy-Gallagher, G. A., Koohmaraie, M., & Laegreid, W. W. (2000). Correlation of enterohemorrhagic Escherichia coli O157 prevalence in feces, hides and carcasses of beef cattle during processing. Proceedings of the National Academy of Sciences of the United States of America, 97, 29993003. Fantelli, K., & Stephan, R. (2001). Prevalence and characteristics of shigatoxin-producing Escherichia coli and Listeria monocytogenes strains isolated from minced meat in Switzerland. International Journal of Food Microbiolgy, 70, 6369. Fegan, N., Vanderlinde, P., Higgs, G., & Desmarchelier, P. (2004). The prevalence and concentration of Escherichia coli O157 in faeces of

In the area of modelling, improved techniques which separate error related to variability and uncertainty will give a more accurate prediction of illness. The ability to model back thorough the chain from illness to numbers/ prevalence of pathogen at particular points in the chain (food safety objectives and performance objectives) will in the future allow public health objectives to be directly related back to targets for the food industry to meet. On a broader perspective, quantitative risk assessment needs to have better linkages with other management systems including HACCP and economic cost benet analysis. With this multi-disciplinary approach food safety, and E. coli O157:H7 will in the future be managed more strategically leading to overall improvements in public health protection from this pathogen. It should however be recognised that apart from beef, other meats, food, water and indeed indirect sources (animal contact, recreational, person to person etc) also play a role in transmission of this pathogen and they also need to be addressed to reduce the public risk posed by this organism. References
Andlovic, A., & Marinsek, J. (1997). STEC isolated from humans, cattle and minced meat. IVC news 7. Notiziario dellIstituto Superiore di ` , 10, 34. Sanita Anon. (1999). ProMED post, 1999 Oct 16. E. coli, VTEC, cattle, petting zoo Canada (Ontario) (03). Available from <http://www.promedmail.org>, last accessed: March 2006. Anon (2002). From the centers for disease control and prevention. Multistate outbreak of Escherichia coli O157:H7 infections associated with eating ground beef United States, JuneJuly 2002. Journal of the American Medical Association, 288(Aug 14), 690691. Anon (2005). Outbreak of E. coli O157:H7 infections associated with a brand of beef burgers in France. Eurosurveillance Weekly, 10, 051103. Bell, B. P., Goldoft, M., Grin, P. M., Davis, M. A., Gordon, D. C., Tarr, P. I., et al. (1994). A multistate outbreak of Escherichia coli O157:H7-associated bloody diarrhoea and hemolytic uremic syndrome from hamburgers: the Washington experience. Journal of the American Medical Association, 272, 13491353. Blanco, M., Blanco, J. E., Blanco, J., Gonzalez, E. A., Mora, A., Prado, C., et al. (1997). Distribution and characterization of faecal verotoxinproducing Escherichia coli (VTEC) isolated from healthy cattle. Veterinary Microbiology, 54, 309319. Blanco, M., Blanco, J. E., Mora, A., Gonzalez, E. A., & Blanco, J. (2000). Serotypes and virulence genes of verocytotoxigenic E. coli (VTEC) isolated from cattle in Spain. In G. Duy, P. Garvey, J. Coia, Y. Wasteson, & D. A. McDowell (Eds.), Pathogenicity and virulence of verocytotoxigenic E. coli. VTEC in Europe Concerted action (pp. 183). Dublin, Ireland: Teagasc, Ashtown Food Research Centre. Cagney, C., Crowley, H., Duy, G., Sheridan, J. J., O Brien, S., Carney, E., et al. (2004). Prevalence and numbers of Escherichia coli O157:H7 in minced beef and beef burgers from butcher shops and supermarkets in the Republic of Ireland. Food Microbiology, 21, 203212. Carney, E., OBrien, S. B., Sheridan, J. J., McDowell, D. A., Blair, I. S., & Duy, G. (2006). Prevalence and level of Escherichia coli O157 on beef trimmings, carcasses and boned head meat at a beef slaughter plant. Food Microbiology, 23, 5259. Cassin, M. H., Lammerding, A. M., Todd, E. C. D., Ross, W., & McColl, R. S. (1998). Quantitative risk assessment for Escherichia coli O157:H7 in ground beef hamburgers. International Journal of Food Microbiology, 41, 2144. Centers for Disease Control and Prevention (CDC) (1998). Outbreaks of Escherichia coli O157:H7 infection and cryptosporidiosis associated

G. Duy et al. / Meat Science 74 (2006) 7688 cattle from dierent production systems at slaughter. Journal of Applied Microbiology, 97, 362370. Germani, Y., Soro, B., Vohito, M., Morel, O., & Morvan, J. (1997). Enterohaemorrhagic Escherichia coli in Central African Republic. Lancet, 349(9066), 1670. Gill, C. O., McGinnis, J. C., & Badoni, M. (1996). Use of total Escherichia coli counts to assess the hygienic characteristics of a beef carcass during processing. International Journal of Food Microbiology, 31, 181196. Glass, K. A., Loeelholz, J. M., Ford, J. P., & Doyle, M. P. (1992). Fate of Escherichia coli O157:H7 as aected by pH or sodium chloride and in fermented, dry sausage. Applied and Environmental Microbiology, 58, 25132516. Hass, C. N. (1983). Estimation of risk due to low doses of microorganisms: a comparison of alternative methodologies. American Journal of Epidemiology, 118, 573582. Haas, C. N., Rose, J. B., & Gerba, C. P. (Eds.). (2000). Quantitative microbial risk assessment (pp. 449). New York: John Wiley & Sons, Inc. Hancock, D. D., Besser, T. E., Rice, D. H., Ebel, E. D., Herriott, D. E., & Carpenter, L. V. (1998). Multiple sources of Escherichia coli O157 in feedlots and dairy farms in the Northwestern USA. Preventative Veterinary Medicine, 35, 1119. Health Protection Surveillance Centre. (2004). Verotoxigenic E. coli O157 Annual Report. Available from <http://www.ndsc.ie/A-Z/Gastroenteric/VTEC/Factsheet/>, last accessed: March 2006. Heuvelink, A. E., van den Biggelaar, F. L. A. M., de Boer, E., Herbes, W. J. G., Melchers, W. J. G., Huis in t Veld, et al. (1998). Isolation and characterization of verocytotoxin-producing Escherichia coli O157 strains from Dutch cattle and sheep. Journal of Clinical Microbiology, 36, 878882. Hilborn, E. D., Mshar, P. A., Fiorentino, T. R., Dembek, Z. F., Barrett, T. J., Howard, R. T., & Cartter, M. L. (2000). An outbreak of Escherichia coli O157:H7 infections and haemolytic uraemic syndrome associated with consumption of unpasteurized apple cider. Epidemiology and Infection, 124, 3136. Jay, M. T., Garrett, V., Mohle-Boetani, J. C., Barros, M., Farrar, J. A., Rios, R., et al. (2004). A multistate outbreak of Escherichia coli O157:H7 infection linked to consumption of beef tacos at a fast-food restaurant chain. Clinical Infectious Diseases, 39, 17. Jensen, C., Ethelberg, S., Gervelmeyer, A., Nielsen, E. M., Olsen, K. E. P., & Mlbak, K. (2006). First general outbreak of Verocytotoxinproducing Escherichia coli O157 in Denmark. Eurosurveillance Monthly, 11, 2. Kassenborg, H. D., Hedberg, C. W., Hoekstra, M., Evans, M. C., Chin, A. E., Marcus, R., et al. (2004). Farm visits and undercooked hamburgers as major risk factors for sporadic Escherichia coli O157:H7 infection: data from a case-control study in 5 FoodNet sites. Clinical Infectious Diseases, 38, 271278. Kennedy, J., Jackson, V., Blair, I. S., McDowell, D. A., Cowan, C., & Bolton, D. J. (2005). Food safety knowledge of consumers and the microbiological and temperature status of their refrigerators. Journal of Food Protection, 68(7), 14211430. Kuhnert, P., Dubosson, C. R., Roesch, M., Homfeld, E., Doherr, M. G., & Blum, J. W. (2005). Prevalence and risk-factor analysis of Shiga toxigenic Escherichia coli in faecal samples of organically and conventionally farmed dairy cattle. Veterninary Microbiology, 109(1 2), 3745. Laine, E. S., Scheftel, J. M., Boxrud, D. J., Vought, K. J., Danila, R. N., Elfering, K. M., et al. (2005). Outbreak of Escherichia coli O157:H7 infections associated with no intact blade-tenderized frozen steaks sold by door-to-door vendors. Journal of Food Protection, 68, 11982002. Lammerding, A. M., Fazil, A., Paoli, G. Desmarchelier, P. & Vanderlinde, P. (1999). Shiga toxin-producing E. coli in ground beef manufactured from Australian beef: Process improvement. Food Science Australia, Brisbane Laboratory. Magwira, C. A., Gashe, B. A., & Collison, E. K. (2005). Prevalence and antibiotic resistance proles of Escherichia coli O157:H7 in beef

87

products from retail outlets in Gaborone, Botswana. Journal of Food Protection, 68, 403406. Mahon, D., Cowan, C., & Henchion, M. (2003). Mince beef and beef burger consumption and handling practices of Irish consumers. Technical report, Teagasc, Ashtown Food Research centre, Ashtown, Dublin 15, Ireland. Maruzumi, M., Morita, M., Matsouka, Y., Uekawa, A., Nakamura, T., & Fugi, K. (2005). Mass food poisoning caused by beef oal contaminated by Escherichia coli O157. Japanese Journal of Infectious Disease, 58, 397. Matthews, L., McKenrick, I. J., Ternent, H., Gunn, G. J., Synge, B., & Woolhouse, M. E. J. (2006). Super-shedding cattle and the transmission dynamics of Escherichia coli O157. Epidemiology and Infection, 134, 131142. McEvoy, J. M., Doherty, A. M., Sheridan, J. J., Thomson-Carter, F. M., Garvey, P., & McGuire, L. (2003). The prevalence and spread of Escherichia coli O157:H7 at a commercial beef abattoir. Journal of Applied Microbiology, 95, 256266. McEvoy, J., Sheridan, J. J., Blair, I. S., & McDowell, D. A. (2004). Microbial contamination on beef in relation to hygiene assessment based on criteria used in EU Decision 2001/471/EC. International Journal of Food Microbiology, 92, 217225. McGee, P., Scott, L., Sheridan, J. J., Earley, B., & Leonard, N. (2004). Horizontal transmission of Escherichia coli O157:H7 during cattle housing. Journal of Food Protection, 67, 26512656. Mead, P. S., Finelli, L., Lambert-Fair, M. A., Champ, D., Townes, J., Hutwagner, L., et al. (1997). Risk factors for sporadic infection with Escherichia coli O157:H7. Archives of Internal Medicine, 157, 204208. Michino, H., Araki, K., Minami, K., Nakayama, S., Ejima, T., Hiroe, Y., et al. (1998). Recent outbreaks of infections caused by Escherichia coli O157:H7 in Japan. In J. B. Kaper & A. D. OBrien (Eds.), Escherichia coli O157:H7and other Shiga toxin-producing E. coli strains (pp. 7381). Washington, DC: American Society for Microbiology. Michino, H., Araki, K., Minami, S., Nakayama, T., Ejima, Y., Hiroe, K., et al. (1999). Massive outbreak of Escherichia coli O157:H7 infection in school children in Sakai city, Japan, associated with consumption of white radish sprouts. American Journal of Epidemiology, 150, 787796. Miyao, Y., Kataokat, T., Nomoto, T., Kai, A., Itoh, T., & Itoh, K. (1998). Prevalence of verotoxin-producing Escherichia coli harboured in the intestine of cattle in Japan. Veterinary Microbiology, 61, 137143. Naugle, A. L., Holt, K. G., Levine, P., & Eckel, R. (2005). Food safety and inspection service regulatory testing program for Escherichia coli O157:H7 in raw ground beef. Journal of Food Protection, 68, 462468. Nauta, M., Evers, E., Takumi, K. & Havelaar, A. (2001). Risk assessment of Shiga-like producing Escherichia coli O157 in steak tartar in the Netherlands. Report 257851003/ 2001 (p. 169). Bilthoven, The Netherlands: RIVM. OBrien, S. B., Duy, G., Carney, E., Sheridan, J. J., McDowell, D. A., & Blair, I. S. (2005). Prevalence and numbers of Escherichia coli O157 on bovine hide at a beef slaughter plant. Journal of Food Protection, 68, 660665. OBrien, S. B. (2005). An exposure assessment on E. coli in minced beef. D. Phil. Thesis. University of Ulster, Jordanstown, Newtownabbey, Northern Ireland. ODonnell, J. M., Thornton, L., McNamara, E. B., Prendergast, T., Igoe, D., & Cosgrove, C. (2002). Outbreak of Verocytotoxin-producing Escherichia coli O157 in a child day care facility. Communicable Disease and Public Health, 5, 5458. Paiba, G. A., Wilesmith, J. W., Evans, S. J., Pascoe, S. J., Smith, R. P., Kidd, S. A., et al. (2003). Prevalence of faecal excretion of verocytotoxigenic Escherichia coli O157 in cattle in England and Wales. Veterinary Record, 153, 347353. Pouillot, R., Beaudeau, P., Denis, J. B., & Derouin, F. (2004). A quantitative risk assessment of waterborne cryptosporidiosis in France using second-order Monte Carlo simulation. Risk Analysis, 24, 117.

88

G. Duy et al. / Meat Science 74 (2006) 7688 Verocytotoxigenic E. coli (pp. 161180). Connecticut: Food and Nutrition Press Inc. Tsuji, H., Hamada, K., Kawanishi, S., Nakayama, A., & Nakajima, H. (2002). An outbreak of enterohemorrhagic Escherichia coli O157 caused by ingestion of contaminated beef at grilled meat-restaurant chain stores in the Kinki District in Japan: epidemiological analysis by pulsed-eld gel electrophoresis. Japanese Journal of Infectious Disease, 55, 9192. USDA-FSIS. (2001). Draft risk assessment of the public health impact of Escherichia coli O157:H7 in ground beef. Available from <http:// www.fsis.usda.gov/>, last accessed: March 2006. Vold, L., Johansen, B. K., Kruse, H., Skjerve, E., & Wasteson, Y. (1998). Occurrence of shigatoxinogenic Escherichia coli O157 in Norwegian cattle herds. Epidemiology and Infection, 120, 2128. Waterman, S., & Small, P. (1996). Characterization of the acid resistance phenotype and rpos alleles of shiga-like toxin-producing Escherichia coli. Infection and Immunology, 64, 28082811. Werber, D., Fruth, A., Buchholz, U., Prager, R., Kramer, M. H., Ammon, A., et al. (2003). Strong association between shiga toxin-producing Escherichia coli O157 and virulence genes stx2 and eae as possible explanation for predominance of serogroup O157 in patients with haemolytic uraemic syndrome. European Journal of Clinical Microbiology and Infectious Diseases, 22, 726730. Wilson, J. B., Clarke, R. C., Renwick, S. A., Rahn, K., Johnson, R. P., Karmali, M. A., et al. (1996). Vero cytotoxigenic Escherichia coli infection in dairy farm families. Journal of Infectious Diseases, 174, 10211027. Zhao, T., Doyle, M. P., Shere, J., & Garber, L. (1995). Prevalence of enterohemorrhagic Escherichia coli O157:H7 in a survey of dairy herds. Applied and Environmental Microbiology, 61, 12901293.

Powell, M., Ebel, E., Schlosser, W., Walderhaug, M., & Kause, J. (2000). Doseresponse envelope for Escherichia coli O157:H7. Quantitative Microbiology, 2, 141163. Riley, L. W., Remis, R. S., Helgerson, S. D., McGee, H. B., Wells, J. G., Davis, B. R., et al. (1983). Haemorrhagic colitis associated with a rare Escherichia coli serotype. New England Journal of Medicine, 308, 681685. Rodrigue, D. C., Mast, E. E., Greene, K. D., Davis, J. P., Hutchinson, M. A., Wells, J. G., et al. (1995). A university outbreak of Escherichia coli O157:H7 infections associated with roast beef and an unusually benign clinical course. Journal of Infectious Diseases, 172, 11221125. Sakuma, M., Urashima, M., & Okabe, N. (2006). Verocytotoxin-producing Escherichia coli, Japan, 19992004. Emerging Infectious Diseases, 12, 323325. Sargeant, J. M., Sanderson, M. W., Smith, R. A., & Grin, D. D. (2003). Escherichia coli O157 in feedlot cattle feces and water in four major feeder-cattle states in the USA. Preventative Veterinary Medicine, 61, 127135. Scottish Centre for Infection and Environmental Health (SCIEH). (2006). Surveillance data on E. coli O157:H7. Available from <http:// www.show.scot.nhs.uk/scieh/>, last accessed: March 2006. Strachan, N. J., Doyle, M. P., Kasuga, F., Rotariu, O., & Ogden, I. D. (2005). Dose response modelling of Escherichia coli O157 incorporating data from foodborne and environmental outbreaks. International Journal of Food Microbiology, 103, 3547. Teunis, P., Takumi, K., & Shinagawa, K. (2004). Dose response for infection by Escherichia coli O157:H7 from outbreak data. Risk Analysis, 24, 401407. Tozzi, A. E., Gorietti, S., & Caprioli, A. (2001). Epidemiology of human infections by Escherichia coli O157 and other verocytotoxin-producing E. coli. In G. Duy, P. Garvey, & D. A. McDowell (Eds.),