Hemolytic-uremic syndrome (HUS) is a clinical syndrome characterized by microangiopathic hemolytic anemia, impaired renal function and excessive platelet consumption leading to thrombocytopenia [1]. HUS is considered to be the most common cause of acute kidney injury (AKI) in European children, mainly affecting pre-school-aged children [2–4]. In the long term, HUS is associated with complications such as hypertension and end-stage renal disease with a death rate of approximately 3–5% [3]. Based on clinical presentation and probable etiology, HUS is commonly divided into diarrhea-associated HUS (D⁺HUS) and non-diarrhea-associated HUS (D⁻HUS), also called atypical HUS. Recent classifications have suggested a more specific approach based on causality and clinical associations [2]. Usually, around 90% of cases are D⁺HUS, while 10% are atypical HUS [5–9].

The most common cause of D⁺HUS in children is infection with Shiga toxin-producing Escherichia coli (STEC) [1], although not all STEC-associated HUS (STEC-HUS) cases present with diarrhea [7, 10–12]. Shiga toxins are produced and released by STEC bacteria and are the main cause of STEC-HUS [1]. There are two main types of Shiga toxins, Shiga toxin 1 and Shiga toxin 2 (Stx 1 and 2). The latter is more frequently found in bacteria causing HUS than Shiga toxin 1. Ruminants are the main reservoir for STEC and infections are mainly food or waterborne. In D⁻HUS, the most common causes are infection caused by Streptococcus pneumoniae (SP-HUS) and genetic forms of HUS [9].

HUS and STEC infections are under epidemiological surveillance in the EU. In 2012, the European Commission updated the European case definition for STEC, requiring laboratory confirmation of Stx or stx gene(s), except when STEC O157 is directly isolated [13]. Surveillance of HUS and STEC in many countries is based on this case definition, requiring laboratory confirmation prior to notification. There are certain challenges in the surveillance of STEC. Far from all STEC infections are treated by a physician, identification of the infectious agent in mild infections might be judged unnecessary by clinicians and until more recent years, verification of non-O157 serogroups was difficult. Because of this, some countries, including France, have previously used surveillance of HUS to follow trends and identify outbreaks of STEC infection [14]. This is based on STEC being the causal factor in most HUS cases. In 2009, 3573 cases of STEC were reported in the EU, of which half were caused by serogroup O157 [4]. In Europe and America, this serogroup is most frequently isolated from HUS patients [2, 4, 6, 12, 15], while O111 is dominant in Australia [16]. Serotype O157:H7 is the most easily diagnosed serotype, based on its failure to ferment sorbitol within 24 hours of incubation [17, 18]. However, in recent years, changes in diagnostic procedures have led to the isolation of an increasing variety of non-O157 STEC serogroups from HUS patients in Europe, including O26, O91, O103, O111, O113, O121, O128, O145 and sorbitol-fermenting O157 (SF O157) [4, 7, 8, 14, 19]. Recent outbreaks of STEC infection in Europe, with varying degrees of HUS development, have resulted in increased awareness. In 2011, E. coli O104 caused a large outbreak with many HUS cases in Germany. While the majority of cases occurred among adults, it also affected more children than previously seen in any other European outbreak to date [20]. In Norway, a national outbreak of STEC O103:H25 occurred in 2006, in which nine children developed HUS, including one with fatal outcome [21, 22], raising awareness among physicians as well as microbiological laboratories.

Notification of EHEC infections to the Norwegian Surveillance System for Communicable Diseases (MSIS) has been mandatory since 1989 [18]. In Norway, all clinicians and microbiological laboratories analyzing human specimens are required by law to notify cases of group A infectious diseases, such as STEC infection, to MSIS at the Norwegian Institute of Public Health. Prior to the Norwegian outbreak in 2006, notification was only required for HUS cases with laboratory-confirmed STEC infection. In December 2006, after the outbreak, notification criteria were changed to include all D⁺HUS cases, based solely on clinical presentation. Since microbiological confirmation of STEC-HUS takes on average 14 days (measured from the day the stool sample is taken to the day the case is reported), the change in notification criteria aimed to improve the timeliness of reporting. In addition, clinically-based notification criteria were expected to increase the sensitivity of the surveillance system, as detection of STEC in a stool sample can be difficult and the bacteria may not be present at the time a patient develops HUS. Measures to improve diagnostic procedures were also implemented after the 2006 outbreak, and the laboratories methodology was gradually expanded to include PCR screening for presence of stx at all the microbiological laboratories. After implementation, ring tests sent out to all laboratories in the following years showed improved diagnostic capabilities for non-O157 serogroups.

The main aim of this study was to determine the annual incidence of D⁺HUS among children <16 years of age diagnosed in Norway from 1999 up to and including 2008 through a review of medical records, compare these numbers with cases reported to MSIS, and consequently assess the sensitivity of the D⁺HUS surveillance. The secondary aim was to describe annual incidence and etiology of all types of HUS in the same age group and for the same time period.

In the period 1^st of January, 1999, to the 31^st of December 2008, we identified a total of 47 cases of HUS in children <16 years of age, with an estimated average annual incidence rate of 0.5 cases per 100,000 children. Of these cases, 81% were diarrhea-associated HUS (D⁺HUS), though only 61% of these were laboratory verified with a Shiga toxin-producing Escherichia coli (STEC) infection. We also found that before mandatory notification criteria were changed from D⁺HUS with laboratory verified STEC infection to clinical D⁺ HUS in December 2006 [18], only 61% of D⁺HUS cases were notified. After the case definition was amended, the number of cases notified to MSIS corresponds with the number of cases we found when systematically reviewing all patient medical records with relevant ICD-10 codes.

We assume that all D⁺HUS cases reported via MSIS are caused by STEC, since this is internationally recognized to be the most common etiological agent in HUS cases [12, 18]. All D⁺HUS cases in our study are therefore coded as STEC cases. However, only 61% of the D⁺HUS cases found through our review of medical records had laboratory-verified STEC. This may reflect that the HUS cases were caused by other etiological agents causing D⁺HUS that we were not able to recognize, but it is more likely due to problems with diagnosing STEC in stool samples from HUS patients. HUS typically develops as a complication of STEC-related diarrhea, but patients often no longer have diarrhea when they develop HUS and may have stopped shedding the bacteria [1]. In addition, if the patient has non-O157:H7 STEC, the diagnostic methods are more complicated than if STEC is caused by O157:H7, allowing the etiological agent to be overlooked. This is especially true for cases occurring before the outbreak in 2006, as during that period many laboratories still based their diagnostics on cultivation and could easily miss the diagnosis [18]. In our study, we found that 64% of verified STEC cases were non-O157 when the outbreak-related STEC-HUS cases are excluded. However, O157 was the most frequently isolated serogroup causing sporadic HUS in Norway, as has been found in other European countries [6, 7, 10, 11, 14], South America [24, 25] and North America [15, 26].

Based on surveillance data from MSIS, an estimated 23% of children with STEC infections developed HUS, which is high in comparison to other countries; certain studies have shown that 10%-15% of children infected with STEC O157 develop HUS [6, 12, 26, 27]. According to the European Center for Disease Prevention and Control, this proportion is about 8% [3]. However, these studies are based predominantly on O157 STEC cases. In the 2012 Germany outbreak, where a particularly aggressive strain of E. coli O104 was the cause, 22% of adults and a slightly higher proportion of children (approximately 24%) developed HUS [20]. The high proportion of HUS cases reported via MSIS may be explained by either an overestimation of HUS cases, an underreporting of STEC cases, or that STEC in Norway may be more likely to cause HUS than what is described for O157 in the literature. The first explanation is unlikely; in addition to the 23 confirmed STEC-HUS cases, we also found 15 probable STEC-HUS, all with classical clinical presentation of STEC-HUS. When considering the difficulties described for laboratory identification of non-O157:H7 STEC strains especially, it is probable that several of these are actually STEC-HUS cases. It is more likely that this high STEC-HUS/STEC-ratio is due to underreporting of STEC cases. Mild cases of STEC infection may only present with diarrhea, not requiring medical attention or submission of stool samples. Only severe cases, which are more likely to be complicated by HUS, are investigated thoroughly for a source. Isolation of STEC in D⁺HUS cases is also often dependent on stool samples being examined early in the disease progression [18]. Additionally, the difficulties in identifying non-O157:H7 STEC strains may result in several cases being missed, despite samples being taken and analyzed. In Norway, most of the cases are caused by non-O157:H7 STEC, and the virulence of non-O157:H7 strains is probably variable. Some strains, like sorbitol-fermenting O157, are now considered more likely to cause HUS than O157:H7 [28], whereas others might be less likely to cause HUS.

Although the proportion of STEC cases developing HUS was high, the overall incidence rate of HUS reported in our study is low compared to similar studies from other European countries, [6–8, 11, 14], likely due to a low incidence of diarrhea-associated HUS, since this accounts for the majority of HUS in children (81%). This again points to the low amount of STEC cases identified in this group and the study as a whole. A possible explanation for this may be the low prevalence of STEC among ruminants in Norway. In particular, surveys in sheep and cattle have found a low prevalence of O157 [18, 29–31].

Our study also describes the burden and etiology of atypical HUS in children. As there is no mandatory notification of D⁻HUS cases in Norway, these cases were only identified and described after our search through medical records. Only nine D⁻HUS cases were identified in the ten year period, accounting for less than 20% of the total HUS burden in children. Of these, two were related to pneumococcal infection (SP-HUS), three were of genetic origin, one had a suspected associated with campylobacter infection and three were of unknown etiology. It is noteworthy that there were only two SP-HUS cases during the ten year period, as certain studies have indicated that this is an increasing problem globally [32, 33].

There are some limitations to our study. As it is retrospective, it reflects the judgments made by clinicians several years ago, when HUS was relatively uncommon. A lack of awareness could have led to misdiagnosed cases. However, to minimize this we also searched for HUS cases in medical records where patients were diagnosed as acute kidney injury. We thereby identified three HUS cases without a HUS ICD-10 code added by the clinicians, instead coded only as acute kidney injury. Failure to detect STEC in HUS patients due to late sampling and diagnostic problems in the laboratory are also limiting factors in the study, as the etiology was not found in a notable proportion of cases.

Our findings indicate that the occurrence of HUS, although low compared to other European countries, and STEC in Norway is higher than previously assumed. While we have no apparent explanation as to why the incidence of HUS is low, a possible contributing factor might be that the prevalence of STEC is low among ruminants, a known source of infection. The diagnostics have improved after the outbreak in 2006. Despite this, the results reinforce that clinicians should perform early stool sampling in HUS cases where STEC infection is suspected. Therefore, our recommendation is to reinforce the mandatory notification and surveillance of both D⁺HUS and laboratory verified STEC-infections and to further develop laboratory verification techniques of emerging non-O157 STEC serotypes.

Our study also illustrates that the proportion of laboratory verified HUS cases might be low. This highlights the need for surveillance based on clinical HUS without the need for laboratory confirmation.