On May 19, 2011, the Robert Koch Institute, Germany's national-level public health authority, was informed about a cluster of three cases of the hemolytic–uremic syndrome in children admitted on the same day to the university hospital in the city of Hamburg. On May 20, a team from the Robert Koch Institute arrived in Hamburg to assist with the public health investigation. It quickly became clear that the case numbers were continuing to rise, that there were also cases in adults, and that other areas of Germany, especially northern Germany, were also affected. An investigation of the outbreak involving all levels of public-health and food-safety authorities was initiated to identify the causative agent and the vehicle of infection in order to prevent further cases of disease.

The hemolytic–uremic syndrome, which was first described in children in the 1950s,1 is characterized by the triad of acute renal failure, hemolytic anemia, and thrombocytopenia. Diarrhea-associated hemolytic–uremic syndrome occurs primarily in children, and a precipitating infection with Shiga-toxin–producing Escherichia coli, mainly of serotype O157:H7, is the primary cause.2 The usual reservoir for these bacteria is ruminants, particularly cattle. Human infection with Shiga-toxin–producing E. coli occurs through the inadvertent ingestion of fecal matter — for example, through contaminated food or water or through contact with animals or their farm environment or, secondarily, through contact with infected humans. In contrast, the hemolytic–uremic syndrome with prodromal diarrhea, indicating an infectious cause, is a rare event in adults. For example, from 1989 through 2006, only 21 of the 322 adults (7%) listed in the Oklahoma registry as having thrombotic thrombocytopenic purpura or the hemolytic–uremic syndrome presented with bloody diarrhea.3

This report provides descr iptive epidemiologic, clinical, and microbiologic information on this unusual outbreak. It will be updated after the outbreak has finally ceased, in order to provide a complete picture.


Methods


German Surveillance System

According to the German Protection against Infection Act of 2001, the detection of a Shiga toxin (Stx) in E. coli isolates or of its encoding gene (stx) in stool enrichment culture or isolates must, by law, be reported by diagnosing laboratories to local health departments. This reporting process allows the identification of Shiga-toxin–producing E. coli infection independently of serogroup (serotyping information is requested but not required). The German case definition of Shiga-toxin–producing E. coli gastroenteritis (without the hemolytic–uremic syndrome) requires, besides laboratory confirmation, the presence of at least one of the following symptoms: diarrhea (three or more loose stools in a 24-hour period), abdominal cramps, or vomiting.

In addition, physicians are required to report clinical symptoms compatible with diarrhea-associated hemolytic–uremic syndrome in a patient. The German case definition of the hemolytic–uremic syndrome comprises thrombocytopenia (platelet count of <150,000 per cubic millimeter), hemolytic anemia, and acute renal dysfunction. The third criterion is met if at least one of the following findings is present: an increase in the serum creatinine level (unspecified), oliguria, anuria, proteinuria, or hematuria.

Reported cases of the hemolytic–uremic syndrome or Shiga-toxin–producing E. coli infection are investigated and recorded by the local health department, and the reports are forwarded electronically, without identifying information, through the state to the federal level. To minimize the delay that might result from the local investigation of the details of a case, the Robert Koch Institute, on May 23, asked all health departments to immediately forward all case reports of suspected or confirmed hemolytic–uremic syndrome, relying on the diagnoses of the notifying clinicians. Case details such as clinical and microbiologic information are to be added to the record by local health departments in the future.

Disease onset was defined as the onset of diarrhea, regardless of whether the hemolytic–uremic syndrome developed at a later date. An outbreak case was defined as a reported case of the hemolytic–uremic syndrome or a reported case of gastroenteritis in a patient infected by Shiga-toxin–producing E. coli, of serogroup O104 or unknown serogroup, with a disease onset on or after May 1, 2011, in Germany. We describe here data from the national reporting database on infectious diseases as of June 18, 2011, at 6 p.m. Central European Summer Time. The preliminary descr iptive analysis focuses primarily on reported cases of the hemolytic–uremic syndrome in Germany as an indication of the entire outbreak. To show the outbreak area, a map of the incidence of the disease according to county is provided (Figure 2Figure 2Incidence of the Hemolytic–Uremic Syndrome According to County in Germany.). Since many cases even within Germany are apparently travel-related, a map showing the incidence according to the patients' residence would be misleading. Therefore, for each presumed county of infection, we counted in the numerator both cases among residents of the county who did not have a history of travel and those among case patients who resided elsewhere and had a history of travel to that county; the denominator was the residential population of the county.


Clinical Information

We analyzed clinical data from two groups of patients at the Hamburg University Medical Center (HUMC): patients at their first presentation to the HUMC between May 19 and June 1 who were positive for stx (data extracted from electronic medical records) and a cohort of adults who were seen between May 25 and June 6 at a Shiga-toxin–producing E. coli unit that was set up during the course of this outbreak. The study protocol was approved by the ethics committee of the Hamburg Chamber of Physicians. Patients were enrolled in the study if they presented with bloody diarrhea or if they had any diarrhea after contact with a patient who had Shiga-toxin–producing E. coli infection. All patients provided written informed consent before enrollment. Patients were followed for at least 14 days and were tested for the outbreak strain according to the protocol of the National Consulting Laboratory on Hemolytic–Uremic Syndrome.4 Only data from patients infected by the outbreak strain were included in the analysis. The proportion of patients with the hemolytic–uremic syndrome among all patients who were positive for Shiga-toxin–producing E. coli was calculated. Platelet counts and creatinine and lactate dehydrogenase levels were monitored daily.


Microbiologic Analysis

Shiga-toxin–producing E. coli infection is diagnosed by private microbiologic laboratories either by screening for Stx with the use of one of several commercially available enzyme immunoassays or by detection of stx with the use of polymerase chain reaction (PCR). The National Reference Centre for Salmonella and Other Bacterial Enteric Pathogens confirms, culturally isolates, and characterizes Shiga-toxin–producing E. coli from samples in local or regional laboratories that are positive for Stx or stx. Chromogenic agar mediums for Enterobacteriaceae that are positive for extended-spectrum beta-lactamase (ESBL) are used for isolation of the strain. Biochemical characterization of the strain was performed with the use of various commercially available tests (VITEK, bioMérieux; MicroPlate GN, BIOLOG; and API, bioMérieux). Shiga-toxin–producing E. coli virulence-factor genes (stx 1, stx 2, eae, and ehx) are detected by established PCR methods.5,6 The presence of virulence-factor genes that are typical of enteroaggregative E. coli, such as aatA, aggR, aap, aggA and aggC, are detected according to established PCR protocols.7 Antimicrobial susceptibility was tested by means of microdilution assays with the use of minimal inhibitory concentrations according to the guidelines of the European Committee on Antimicrobial Susceptibility Testing. Serotyping of Shiga-toxin–producing E. coli followed standard protocols.8 One-enzyme (Xba1) pulsed-field gel electrophoresis was performed on Shiga-toxin–producing E. coli O104:H4 isolates.9 Given the strain's properties, a shortened protocol is recommended by the National Consulting Laboratory on Hemolytic–Uremic Syndrome4 and is currently used by the National Reference Centre and HUMC for confirmation of the outbreak strain.


Statistical Analysis

For statistical comparisons, the z test was used for proportions, and the Mann–Whitney U test for age distribution. The incubation period was estimated on the basis of data from selected patients with the hemolytic–uremic syndrome or Shiga-toxin–producing E. coli gastroenteritis for whom the date of onset of diarrhea was known and who had stayed in northern Germany for no more than 48 hours. The interval between the date of onset of diarrhea and the date of diagnosis of the hemolytic–uremic syndrome was calculated with the use of information from the clinician's notification form, which was sent without identifying information to the Robert Koch Institute.




Results


Outbreak Cases

As of June 18, 2011, a total of 810 cases of the hemolytic–uremic syndrome, including 27 fatal cases (3.3%), and 2412 additional cases of Shiga-toxin–producing E. coli gastroenteritis (all laboratory-confirmed), including 12 fatal cases (0.5%), were reported in Germany, with onset dates of May 1 or later. Thus, the hemolytic–uremic syndrome developed in 25.1% of the patients ascertained in this outbreak. The outbreak grew dramatically starting on May 8; cases of the hemolytic–uremic syndrome appeared to peak on May 21 (median date of onset, May 21), and cases of Shiga-toxin–producing E. coli diarrhea appeared to peak on May 22 and 23 (Figure 1Figure 1Epidemiologic Curve of the Outbreak.), with a median date of hospitalization for hemolytic–uremic syndrome of May 24.

The three case patients with an onset before May 8 include two young children (one 4 years of age and the other 6 years of age), one of whom is not a resident of the outbreak area. Both are stx-positive, but serotypes are not available.

Cases of the hemolytic–uremic syndrome have been reported from all 16 states in Germany. The highest incidences are being reported from the northern states of Hamburg (10.1 cases per 100,000 population), followed by Schleswig–Holstein (6.7 cases per 100,000), Bremen (2.7 cases per 100,000), Mecklenburg–Vorpommern (2.2 cases per 100,000), and Lower Saxony (1.7 per 100,000) — the “northern Germany outbreak area.” The outbreak has been almost simultaneous in these areas. Most of the cases from other states can be linked to travel-related exposures in the northern Germany outbreak area. Figure 2 shows the incidence of the disease according to county of infection. Aside from two satellite clusters linked to restaurants in eastern North Rhine–Westphalia and southern Hesse, the area with high incidences (4 to 30 reported cases per 100,000 population) is centered around the city of Hamburg.

A total of 89% of the case patients in this outbreak in whom the hemolytic–uremic syndrome has developed are adults (i.e., persons older than 17 years of age). Among case patients 17 years of age or younger, the median age is 11.5 years. Only 1% of the case patients with the hemolytic–uremic syndrome are younger than 5 years of age, which was the median age among case patients with the hemolytic–uremic syndrome reported in Germany from 2001 through 2010.10 The median age of all patients reported to have the hemolytic–uremic syndrome in the outbreak is 43 years, and the median age is higher for female case patients than for male case patients (44 vs. 41 years, a nonsignificant difference). Among the case patients with the hemolytic–uremic syndrome in the states in the northern Germany outbreak area, only those in Hamburg have a median age (34 years) that is significantly lower than the median age of the case patients in the four surrounding affected states (P<0.001). The median age of case patients with the hemolytic–uremic syndrome who have died is 74 years (range, 24 to 91, except for one 2-year old boy), and the median age of patients with Shiga-toxin–producing E. coli gastroenteritis who have died is 83 years (range, 38 to 89). The incidence of the hemolytic–uremic syndrome peaks in the age group of 30 to 34 years in the case of women and in the age group of 25 to 29 years in the case of men (Figure 3Figure 3Incidence of the Hemolytic–Uremic Syndrome (HUS) in the Outbreak According to Age Group and Sex.).

A total of 68.0% of the case patients with the hemolytic–uremic syndrome and 58.8% of the case patients with Shiga-toxin–producing E. coli gastroenteritis are female; among the patients in the two groups who have died, 77.8% and 58.3%, respectively, were female. The proportion of male case patients with the hemolytic–uremic syndrome rises among cases with dates of onset in June; 47.8% of the patients with an onset of disease in June are male, as compared with 31.9% of the patients with a disease onset in May (P=0.04). Among the case patients with the hemolytic–uremic syndrome in the five most northern and most affected states, 74.3% in Hamburg and 65.5% in the surrounding states are female (nonsignificant difference).

On the basis of data from 43 case patients, we estimated that the median incubation period for this pathogen in this outbreak was 8 days (interquartile range, 7 to 9), without an apparent difference between cases of Shiga-toxin–producing E. coli gastroenteritis and cases of the hemolytic–uremic syndrome. Among 79 case patients for whom data on both the date of onset of diarrhea and the date of onset of the hemolytic–uremic syndrome are known, the interval from the onset of diarrhea to the diagnosis of the hemolytic–uremic syndrome was 5 days (interquartile range, 4 to 6).


Clinical Information

Data on 141 patients, obtained at their first presentation to HUMC, were analyzed; 124 of the patients (88%) were adults. Among the adults, 66% were women, and the percentage was consistent across age groups; among the children, 50% were girls. No patient had a fever (defined as a temperature of at least 37.5°C) at the first presentation. Bloody diarrhea was reported less often in children than in adults (59% [10 of 17 children] vs. 96% [114 of 119 adults], P<0.001), whereas abdominal pain was a very common symptom in both children and adults, occurring in 92% of the children (12 of 13) and in 89% of the adults (105 of 118). Vomiting occurred more often in children than in adults (69% [9 of 13 children] vs. 20% [20 of 98 adults], P<0.001). Most patients did not have significantly elevated leukocyte levels (most were within the normal range; in some cases, counts were approximately 13,000 per cubic millimeter) or C-reactive protein levels (typically about 15 to 35 mg per liter [normal level, <5 mg per liter]). A total of 22 patients (16%) already met the criteria for the hemolytic–uremic syndrome at the time of presentation. Clinical and laboratory values in adults and children, stratified according to the presence or absence of the hemolytic–uremic syndrome, are summarized in Table 1Table 1Demographic and Clinical Characteristics and Laboratory Test Values of Patients Positive for Shiga-Toxin–Producing E. coli at First Presentation..

Among the 135 patients who were followed prospectively, the Shiga-toxin–producing E. coli outbreak strain was detected in 59 (44%), and the hemolytic–uremic syndrome developed in 12 of these patients (20%; 95% confidence interval, 11 to 33). Demographic and clinical characteristics at presentation did not differ significantly between patients with diarrhea in whom the hemolytic–uremic syndrome developed and those in whom it did not develop (Table 2Table 2Demographic and Clinical Characteristics of Patients Positive for Shiga-Toxin–Producing E. coli (STEC) Who Were Followed Prospectively.). An examination of the platelet counts and creatinine and lactate dehydrogenase levels 5 days before through 2 days after the onset of the syndrome in 10 patients with the hemolytic–uremic syndrome (Figure 4Figure 4Selected Laboratory Values.) indicates that the development of the hemolytic–uremic syndrome was sudden.


Microbiologic Features

The serotype of the Shiga-toxin–producing E. coli outbreak strain is O104:H4. The strain ferments sorbitol within 24 hours and is lactose-positive, beta-glucuronidase–positive, and subtilase-negative. The strain carries the gene for a Shiga-toxin 2 variant (stx 2a). Other typical Shiga-toxin–producing E. coli genes such as stx 1, eae, and ehx are missing. In addition, the pathogen possesses genes typical of enteroaggregative E. coli, such as attA, aggR, aap, aggA, aggC, located on a virulence plasmid (heat-stable enterotoxin EAST-1 negative). All outbreak strains are resistant to beta-lactam antibiotics (e.g., ampicillin) and third-generation cephalosporins and are partially resistant to fluoroquinolones (nalidixic acid). The strain is sensitive to carbapenems and ciprofloxacin. The outbreak strains produce an ESBL complex (CTX-M15) and beta-lactamase TEM-1. The National Reference Center has so far typed 439 isolates of this outbreak clone from 650 samples screened in local or regional laboratories for Stx production or presence of stx. Of the 60 isolates that have been analyzed thus far with the use of pulsed-field gel electrophoresis (4 from patients with the hemolytic–uremic syndrome and 56 from patients with Shiga-toxin–producing E. coli gastroenteritis), all have had indistinguishable patterns on pulsed-field gel electrophoresis.




Discussion

In this preliminary report, we describe the epidemiologic characteristics of a large outbreak of the hemolytic–uremic syndrome. There have been more than 800 incident cases of the hemolytic–uremic syndrome, predominantly affecting adults, in this outbreak since May 1, 2011. In addition, as of June 17, 2011, as many as 15 other countries, including the United States, have reported cases occurring among people who had traveled to northern Germany: 41 cases of the hemolytic–uremic syndrome (including one death) and 68 cases of Shiga-toxin–producing E. coli gastroenteritis.11 The current outbreak probably began on May 8 or 9; the three cases of the hemolytic–uremic syndrome in patients with an onset of gastroenteritis before these dates are atypical for the outbreak. In addition, the patient who had the first case of Shiga-toxin–producing E. coli gastroenteritis with confirmed serotype O104 infection had a disease onset on May 8.

To date, there are important differences between this outbreak and previous large outbreaks of Shiga-toxin–producing E. coli infection,12-17 such as the one that occurred in Japan in 1996, in which there were 121 cases of the hemolytic–uremic syndrome — all in children.12 First, the hemolytic–uremic syndrome represents a quarter of the ascertained cases, which is a much larger percentage than in other outbreaks. Second, the majority of the cases of the hemolytic–uremic syndrome (approximately 89%) have occurred in adults rather than in children, with the majority occurring in women. Third, the causative agent was a non–O157 Shiga-toxin–producing E. coli strain (O104:H4).

The outbreak strain combines the virulence properties of two different diarrhea-causing E. coli pathotypes: typical enteroaggregative E. coli and Shiga-toxin–producing E. coli. It has been speculated that the outbreak strain is a typical enteroaggregative E. coli strain that has acquired the bacteriophage encoding stx 2a.18 Enteroaggregative Shiga-toxin–producing E. coli isolated from patients with the hemolytic–uremic syndrome have been described previously,19 albeit rarely. A similar Shiga-toxin–producing E. coli O104:H4 strain, with a different set of fimbriae, was isolated in 2001 from two siblings in Germany in whom the hemolytic–uremic syndrome had developed.20 Since typical enteroaggregative E. coli are isolated primarily from humans,21 the origin of this outbreak may not be zoonotic.

In this outbreak, the proportion of patients with outbreak cases in whom the hemolytic–uremic syndrome developed was 25%, despite public advice to patients to seek medical care (and laboratory testing) if they had bloody diarrhea — which probably led to a more complete ascertainment of gastroenteritis cases. Similarly, the hemolytic–uremic syndrome developed in 20% of prospectively observed patients with Shiga-toxin–producing E. coli diarrhea at a hospital in Hamburg. These proportions are higher than those in previous outbreaks12-17 and higher than the proportion (6%) ascertained through active surveillance of Shiga-toxin–producing E. coli O157:H7, the virulent prototype of Shiga-toxin–producing E. coli, in the United States.22 Taken together, these data suggest that the pathogen in the current outbreak is exceptionally virulent. Of note, the Shiga-toxin variant of the outbreak strain has previously been isolated in Germany only from the rare sorbitol-fermenting Shiga-toxin–producing E. coli O157:H−, a hypervirulent pathogen in children that is associated with high mortality.23 Another feature of the pathogen is the estimated median incubation period of 8 days for this outbreak, which is longer than the 3-day to 4-day incubation period reported for Shiga-toxin–producing E. coli O157:H7.13,24

At present, it remains to be elucidated why women are overrepresented among the cases of the hemolytic–uremic syndrome. No sex difference has been observed with respect to the risk of development of the hemolytic–uremic syndrome among a limited sample of patients with diarrhea who were prospectively followed at HUMC. Furthermore, it is unclear whether the atypical age distribution of cases in this outbreak primarily reflects the distribution of exposure or is attributable to the specific properties of this outbreak strain — or both. The outbreak strain lacks the intestinal adherence factor intimin (encoded by the gene eae), which might be of particular importance for virulence in children. The gene eae is found in the majority of Shiga-toxin–producing E. coli isolated from German children who have gastroenteritis (e.g., 85% of children younger than 3 years of age25) and in 97% of these organisms isolated from children with the hemolytic–uremic syndrome in Germany and Austria.26 In contrast, the majority of Shiga-toxin–producing E. coli isolated from adults with sporadic cases of gastroenteritis lack eae,25 and eae-negative strains have previously been isolated from adults with the hemolytic–uremic syndrome.27

The most common clinical sign in adults was bloody diarrhea accompanied by abdominal cramps. The clinical presentation in adults differed from that in children. Bloody diarrhea occurred significantly more often in adults — irrespective of the presence or absence of the hemolytic–uremic syndrome — whereas vomiting was reported more frequently in children. Whether the high proportion of patients with bloody diarrhea reflects the characteristics of the strain or is a consequence of advice to the public to seek medical care in the case of bloody diarrhea is a subject for further investigation. Clinical symptoms such as abdominal pain, bloody diarrhea, and the frequency of loose stools did not differ between patients in whom the hemolytic–uremic syndrome developed and those in whom it did not. Changes in laboratory values, indicating renal failure and hemolysis, occurred quickly, often within 24 hours (Figure 4). Daily laboratory testing of platelet counts and creatinine and lactate dehydrogenase levels appeared to be pivotal for the early diagnosis of the hemolytic–uremic syndrome, and these laboratory tests were more sensitive than were patient-reported symptoms and the physical examination. Indeed, several patients reported that they had begun to recover from bloody diarrhea several days after the initial presentation, at the same time as the onset of the hemolytic–uremic syndrome. For many reported cases, information on exact symptoms (e.g., diarrhea or bloody diarrhea) and additional microbiologic information are not yet available. Consequently, although the clinical picture of the hemolytic–uremic syndrome in adults appears to be very specific for this outbreak, among the cases of Shiga-toxin–producing E. coli diarrhea, cases unrelated to this outbreak cannot be efficiently filtered out with the use of serotype information.

As of the time of this report, the German outbreak is not over, although case numbers are currently declining. Raw produce or salad condiments are suspected as the food vehicle. Investigations are ongoing.



Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.

Drs. Frank and Werber contributed equally to this article.

This article (10.1056/NEJMoa1106483) was published on June 22, 2011 at NEJM.org.

We thank the physicians and laboratory personnel, who are working under intense strain and yet are keeping up their notification requirements; local and state health departments for quickly passing on case data to the Robert Koch Institute; and epidemiologists in other countries for providing detailed travel and disease information regarding patients with travel-associated cases.



Source Information

From the Departments of Infectious Disease Epidemiology (C.F., D.W., M.F., M.H., H.B., M.W., K.S., G.K.) and Infectious Diseases (A.F., R.P.) and the Postgraduate Training for Applied Epidemiology Program (M.A.), Robert Koch Institute, Berlin; and the Department of Internal Medicine, University Medical Center Hamburg-Eppendorf (J.P.C., A.Z., S.J.), and the Health Department of the Hamburg Northern District (A.S.) — both in Hamburg, Germany.

Address reprint requests to Dr. Werber at werberd@rki.de.

The members of the Hemolytic–Uremic Syndrome (HUS) Investigation Team are listed in the Supplementary Appendix