We present a nationwide outbreak of STEC causing severe HUS in a high percentage of the affected children. The clinical course was characterized by an aggressive disease with significant extrarenal complications. In Norway there are five University Hospitals with paediatric departments, all in close contact with the local paediatric departments. Due to the alert of the outbreak, the University Hospitals were contacted to treat all cases of HUS. The affected children were admitted to four of these departments, the 5th department confirming that no patient with HUS was admitted during the outbreak. Therefore we conclude our material includes all the affected children.
Several clinical and biochemical features at onset of HUS have been proposed to be related to poor prognosis . Among the most often proposed factors are leukocytosis and anuria [9–11]. A case-control study from 2006 dealt with 17 deaths among patients with HUS and concluded that those presenting with oligoanuria, dehydration, WBC > 20 × 109/L and haematocrit > 23% are at substantial risk of fatal HUS . Most of the patients in our material (seven of 10) had white blood count above 20 × 109/L on admission; the highest level (41.3 × 109/L) was registered in the boy who died. He also had the highest haemoglobin-level and thereby haematocrit, corresponding well with the risk-factors pointed out by Oakes et al. . Eight of ten patients were oligoanuric on admission.
Seven of the children needed transient dialysis, with a median duration of 15 days. One patient developed end stage renal failure and received a living related kidney transplant one year later. According to the literature, around half of children with HUS will need dialysis, with a median duration of 5 to 7 days . This epidemic shows a higher proportion of patients developing a very severe disease with extrarenal complications.
CNS involvement is common and is reported in 20-50% of HUS cases [14, 15] and was present in five patients in our material. Common signs of CNS involvement in HUS are seizures, reduced level of consciousness, hemiparesis, visual disturbances and brain stem symptoms. Basal ganglia involvement is a typical MRI-finding in HUS-patients with neurological complications , and was present in two of our patients (Table 2).
The reported incidence of colon necrosis and perforation in case studies varies from 1-8% [16–19]. A review by Siegler in 1994 reported a total incidence of colon necrosis/perforation at 2% . Two of the patients in the present study developed necrosis of colon (20%). Patient 9 underwent subtotal colectomy 27 days after onset of symptoms. In a paper reviewing the occurrence of colonic necrosis in patients with HUS, a mean of 11 days after onset of symptoms was reported . Both our patients were on peritoneal dialysis when the necrosis occurred. To our knowledge peritoneal dialysis being a risk factor for the development of colonic necrosis in patients with HUS has not been reported. However, peritoneal dialysis may mask abdominal symptoms leading to delay in diagnosis and surgical treatment.
Diabetes mellitus is a rare complication of HUS and mainly occurs in severe cases . A systematic review of 21 studies concluded a pooled incidence of 3.2% . Autopsy studies have shown thrombosis of the vessels supplying the islets of Langerhans with preservation of the exocrine pancreas . One girl (patient 9) developed permanent insulin-dependent diabetes mellitus. There was no evidence of autoimmune diabetes as all diabetes related autoantibodies were negative. She was seriously ill on admission and developed necrosis of colon and end stage renal failure, and finally received a kidney transplant. This corresponds to a previous review, stating that children with HUS who develop diabetes mellitus, were more likely to have severe disease with increased mortality risk . Among survivors, 38% were left with permanent diabetes requiring insulin . Even though this patient also needed a kidney transplant, simultaneous pancreas and kidney transplantation was not an option, due to our policy to use living related donors which favourably influence outcome.
All children received blood transfusions. The mean haemoglobin-value at transfusion was 6.9 g/dL. Erythrocyte transfusions in HUS should be avoided if possible, and some suggest it is indicated only when haemoglobin is below 6.0 g/dL . Nevertheless, the usual indications for erythrocyte transfusions apply, i.e. respiratory compromise and cerebral involvement, and 70-80% of patients with HUS will require transfusions [1, 23].
Antibiotics is contraindicated in the treatment of possible STEC infections, due to increased toxin-release from bacterial lysis  or increased production of toxin due to induction of bacteriophages on which stx-genes are located . In our material, six children received intravenous antibiotics. However, the treatment was initiated after the diagnosis of HUS was established in five, and none of the patients had antibiotics started as treatment of HUS, but on the suspicion of secondary bacterial infections.
To our knowledge this is the first outbreak of HUS caused by E. coli O103. The microbiological, serological and epidemiological aspects of the outbreak have previously been published [6, 7]. We found positive faecal samples for O103:H25 in ten children during the outbreak, and four of these developed HUS. This high incidence of HUS among the infected patients contrasts previous reports on E. coli O157:H7 outbreaks, in which 11%-14% developed HUS [26, 27]. During the present outbreak, the attention-level in the population was kept high due to huge interest of the epidemic in the media. National health authorities instructed parents to see a physician if their child had any symptoms of diarrhoea or vomiting. Physicians were informed by the Norwegian Institute of Public Health to collect faecal samples from children with diarrhoea and the number of faecal samples analyzed by the microbiological laboratories increased. On that background it is unlikely that the number of children infected by this specific O103-strain was substantially higher than those diagnosed. The specific diagnosis of O103 was confirmed either through faecal sampling or serology in nine out of ten patients. This corresponds to Lynn et al. who found that 84% of the cases of HUS in UK and Ireland in 1997-2001 were similarly confirmed . In the present report positive faecal samples were found in only four of the patients with HUS. The explanation to this might partially be due to difficulties collecting adequate samples; several of the children did not pass stool for several days after admission to hospital. In a prospective surveillance of Canadian children with HUS from 2000 to 2002, stool cultures showed evidence of bacterial pathogens in 67% of the patients, but only two non-O157-strains were found .