Background
Shiga toxin producing
E. coli (STEC) can cause bloody diarrhoea which in 2–15% of cases, particularly in children, develop into haemolytic uraemic syndrome (HUS) which can lead to renal failure and death [
1]. More than 90% of diarrhoea-associated HUS cases are due to STEC infections. Routine diagnosis and surveillance of STEC-infections was originally developed for serotype O157:H7 of STEC. However, non-O157
E. coli infections are in certain geographic regions considered to be at least equally important, but may in general be underdiagnosed [
2].
Sporadic STEC infections may be transmitted through food, contact with animals or farming environments or by person-to-person, the last two affecting mainly young children [
3]. Outbreaks are mainly foodborne, and have been associated with a wide variety of products, including undercooked minced beef, unpasteurized milk or apple juice, yoghurt, cheese, lettuce, vegetables, cured sausages and drinking water [
1].
In Norway (population 4.6 million), around 10 to 20 cases of sporadic STEC infection are notified annually. Most have only bloody diarrhoea [
4], and approximately half have been acquired in Norway. The only documented foodborne STEC outbreak in Norway occurred in 1999 with four confirmed cases of
E. coli O157:H7 infection, most likely caused by contaminated domestically produced lettuce [
5].
On 20–21 February 2006, a cluster of four diarrhoea-associated HUS cases was reported to the Norwegian Institute of Public Health (NIPH) from an academic hospital in Oslo. Enquiries to other hospitals in Norway identified two additional HUS cases diagnosed since the beginning of 2006. Since hospital episode statistics indicate less than a handful HUS cases in children per year in Norway, we suspected an outbreak and launched an investigation in order to identify the source and stop the outbreak.
Discussion
We have reported an outbreak of haemolytic uremic syndrome caused by a highly virulent strain of Shiga toxin 2 producing E. coli O103:H25. Dry cured sausages were identified as the vehicle of infection, as evidenced by results of a case control study and subsequent isolation of the outbreak strain in the product.
We also found a strong association between illness and consumption of beefburger at a fast food restaurant chain. The burgers were produced in the same facility as the initially suspected minced beef. The first case-control study that implicated minced beef from producer A included only six cases. Due to the emergency of the outbreak, these cases were used for both generating and testing the hypotheses, which might have introduced an information bias. Later in the investigation, after the suspicion about minced beef was announced, there were many reports about the "burger-bug" in the media. We believe that recall bias may have caused the association with beefburgers from chain B in the second larger case-control study.
Long after the product had been withdrawn, it was discovered through immunoblot analysis that one case (no 8, table
2) probably was infected by another
E. coli than the outbreak strain, and thus was proably not related to the outbreak. We have still presented the analysis including this case (table
1) because it fulfilled the case definition used during ther investigation, and it was this analysis that formed the basis for decision making at the time. This illustrates the dillemma of choosing between a sensitive and a specific case definition during an outbreak investigation.
This investigation illustrates that even when it is considered necessary to warn customers about suspected products early in the investigation, it is important to continue investigating alternative sources vigorously until the source can be confirmed and no more cases occurs after the source is removed.
Genotyping of the isolates was pursued using a recently published MLVA assay, which is designed for differentiation within all
E. coli serotypes [
6]. MLVA is generally considered to be a robust and fast typing method with high discriminatory power [
6,
11]. Prospective MLVA subtyping of STEC may be used to detect clusters of related cases and ascertain the source of infection. The high discriminatory power of the MLVA method for O103 as well as for
E. coli in general, has been documented by Lindstedt et al. [
6].
The outbreak was detected by clinicians reporting a cluster of HUS cases, and not by an increase in laboratory verified
E. coli O103 infections. The same has been the case in a few other reported STEC outbreaks; one STEC O111 outbreak in Australia [
12], some German STEC O157 (sorbitol-fermenting) outbreaks [
13,
14] and two Italian HUS outbreaks involving non-O157 strains [
15,
16]. Most of these outbreaks would have been missed by screening faecal samples using sorbitol MacConkey's (SMAC) agar as this medium is intended for the detection of of sorbitol-negative O157 strains. Underreporting of non-O157
E. coli strains is assumed due to the current laboratory methods used in many countries [
2,
16], including Norway. Therefore, surveillance of diarrhoea-associated HUS may be needed for early detection of STEC outbreaks [
17].
STEC O103:H2 is one of the most frequently isolated non-O157 serotypes in many European countries and can cause severe illnesses comparable to those caused by serogroup O157 [
18]. It was first identified as a causative agent of HUS in 1992 [
19], however cases are mainly sporadic and outbreaks of STEC O103:H2 have rarely been reported. One family outbreak was reported in Japan [
20] and one day-care centre outbreak in Argentina [
21], both caused by
stx
1-positive STEC O103. The sources were not identified. The serotype of the present outbreak – O103:H25 – is extremely rare. A few sporadic cases have been reported [
22,
23], and where the toxin profile has been reported, all have been
stx
1-positive. Outbreaks or sporadic cases caused by
stx
2 positive
E. coli O103:H25 have to our knowledge never previously been identified.
More than half the patients developed HUS. In spite of widespread testing of contacts and enhanced laboratory procedures during the outbreak, very few additional mild cases were identified. Only one family member of a HUS case reported symptoms. An increase in non-outbreak related STEC infections in patients with milder symptoms were reported, which could indicate that the sensitivity of testing improved during and after the outbreak. Thus, we believe that the outbreak strain was particularly virulent. Also in some other outbreaks, a similar high proportion of HUS cases have been observed [
13,
14,
24]. STEC strains, independent of serogroup, that contain the
eae gene and produce Stx2 are much more likely to cause severe disease and HUS than are those that produce Stx1 [
25‐
27]. In addition, sequencing of the
stx
2 operon showed that it encoded a Stx2 variant that is more often associated with HUS than other subtypes [
9]. The reason why
stx-genes could be demonstrated in only two of the outbreak isolates, despite the fact that the isolates were closely related genetically, will be the subject further research. Bielaszewska et al studied patients with HUS hospitalized in Austria and Germany, and found that at the time of microbiological analysis, 5% of HUS patients shed no longer the causative STEC, but excreted
stx-negative derivatives that had lost
stx during infection [
28]. Other studies have also shown that STEC may loose the
stx-genes during infection or subcultivation [
29‐
32].
A Danish study found that serogroups O157 and O103 were independent risk factors for bloody diarrhoea [
33]. Earlier reports have suggested that
E. coli O103:H2, can be regarded as an emerging foodborne pathogen [
34] and warned that additional uptake of
stx
2-phages by
E. coli O103 can result in the emergence of a strain with increased virulence as has occurred with
E. coli O26 [
18,
34].
The two cases diagnosed with
E. coli O103:H25 with the same MLVA- and virulence profile in 2003 and 2005 may indicate that this strain has been present over a longer time period in Norway. However, it is not known if these cases were related to cured sausage or other products containing mutton. Earlier reports have shown that Norwegian sheep may carry human-pathogenic STEC [
35], including STEC O103 [
36]. Mutton products were recently associated with an outbreak of STEC O26 in France [
37].
Fecal contamination of carcasses is practically unavoidable, and STEC are able to survive the fermentation, drying, and storage stages of sausage production [
38], as shown in experimental studies [
39‐
41]. Several STEC outbreaks caused by dry fermented sausages have been described [
12,
42‐
45]. These products are often made of combinations of meat mixed with spices and curing materials that then undergo fermentation and drying. In general, it is believed that short ripening time may lead to incomplete fermentation and thus to conditions (e.g. pH) that still allows growth or survival of enteric pathogens. Since these sausages are eaten raw and the infectious dose for STEC is very low, the producer must assure microbiologically safety. The products are believed to constitute a 'medium' risk hazard for STEC infection [
39,
46,
47], and the US Food Safety and Inspection Service has developed guidelines [
48] for sausage manufacturers to validate processes to ensure a 5 log10 unit (5D) reduction in counts of STEC O157:H7, for instance by heating the final product. Still, we believe that the consumers should be made aware of the inherent risk associated with these products [
46].
We do not know why STEC survived the curing process in the incriminated "morrpølse". An evaluation committee suggested that there should be stricter requirement on hygienic quality for raw meat used for such products. In addition, the producer should document the effectiveness of the production process in reducing the level of pathogens if present in raw meat or other ingredients.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PA was responsible for leading the outbreak investigation. BS, KN and HME were principal investigators of the epidemiological investigation and BS drafted the manuscript. JL and GK were responsible for the microbiological investigation of the outbreak. BAL and LTB carried out the genotyping and virulence factor testing of patient and environmental isolates. All authors have read and approved the final manuscript.