Hostname: page-component-7c8c6479df-r7xzm Total loading time: 0 Render date: 2024-03-28T20:14:34.187Z Has data issue: false hasContentIssue false

Estimating true incidence of O157 and non-O157 Shiga toxin-producing Escherichia coli illness in Germany based on notification data of haemolytic uraemic syndrome

Published online by Cambridge University Press:  29 July 2016

A. KUEHNE*
Affiliation:
Robert Koch Institute (RKI), Department for Infectious Disease Epidemiology, Berlin, Germany
M. BOUWKNEGT
Affiliation:
Centre for Infectious Disease Control, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands
A. HAVELAAR
Affiliation:
Centre for Infectious Disease Control, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands Emerging Pathogens Institute and Animal Sciences Department, University of Florida, Gainesville, FL, USA
A. GILSDORF
Affiliation:
Robert Koch Institute (RKI), Department for Infectious Disease Epidemiology, Berlin, Germany
P. HOYER
Affiliation:
Clinic for Pediatrics II, Essen University Hospital, University Duisburg-Essen, Essen, Germany
K. STARK
Affiliation:
Robert Koch Institute (RKI), Department for Infectious Disease Epidemiology, Berlin, Germany
D. WERBER
Affiliation:
Robert Koch Institute (RKI), Department for Infectious Disease Epidemiology, Berlin, Germany State Office for Health and Social Affairs, Berlin, Germany
*
*Author for correspondence: Dr Anna Kuehne, Robert Koch Institute, Department for Infectious Disease Epidemiology, Seestr. 10, 13353 Berlin, Germany. (Email: KuehneAn@rki.de)
Rights & Permissions [Opens in a new window]

Summary

Shiga toxin-producing Escherichia coli (STEC) is an important cause of gastroenteritis (GE) and haemolytic uraemic syndrome (HUS). Incidence of STEC illness is largely underestimated in notification data, particularly of serogroups other than O157 (‘non-O157’). Using HUS national notification data (2008–2012, excluding 2011), we modelled true annual incidence of STEC illness in Germany separately for O157 and non-O157 STEC, taking into account the groups’ different probabilities of causing bloody diarrhoea and HUS, and the resulting difference in their under-ascertainment. Uncertainty of input parameters was evaluated by stochastic Monte Carlo simulations. Median annual incidence (per 100 000 population) of STEC-associated HUS and STEC-GE was estimated at 0·11 [95% credible interval (CrI) 0·08-0·20], and 35 (95% CrI 12-145), respectively. German notification data underestimated STEC-associated HUS and STEC-GE incidences by factors of 1·8 and 32·3, respectively. Non-O157 STEC accounted for 81% of all STEC-GE, 51% of all bloody STEC-GE and 32% of all STEC-associated HUS cases. Non-O157 serogroups dominate incidence of STEC-GE and contribute significantly to STEC-associated HUS in Germany. This might apply to many other countries considering European surveillance data on HUS. Non-O157 STEC should be considered in parallel with STEC O157 when searching aetiology in patients with GE or HUS, and accounted for in modern surveillance systems.

Type
Original Papers
Copyright
Copyright © Cambridge University Press 2016 

INTRODUCTION

Shiga toxin-producing Escherichia coli (STEC) is an important cause of gastroenteritis (GE) and life-threatening haemolytic uraemic syndrome (HUS) in many countries. STEC has a zoonotic reservoir (mainly ruminants) and is transmitted by inadvertent ingestion of small amounts of faecal matter. The serotype is an indicator of the genomic strain content and incidence of human illness and disease severity varies by serotype [Reference Boerlin1, Reference Preussel2]. Evidence from observational studies suggests that STEC of serogroup O157 with serotypes H7 or H- (O157 STEC) are, on average, substantially more virulent than other (‘non-O157’) STEC implicated with human illness [Reference Preussel2Reference Hedican4]. O157 STEC is the leading cause of paediatric HUS [Reference Tarr, Gordon and Chandler5] and the most frequently isolated aetiological agent in STEC outbreaks worldwide [Reference Donnenberg and Whittam6]. These organisms can be easily identified by culture on selective and differential agar [Reference Gould7], with the exception of rarely identified sorbitol-fermenting (sf) clones [Reference Bielaszewska8, Reference Werber9].

Non-O157 STEC represents a genomically heterogeneous group of organisms, comprising STEC with little or no virulence to humans but also with high virulence, e.g. STEC O104:H4 that caused the largest outbreak of HUS thus far [Reference Frank10]. Currently, diagnosis of non-O157 STEC is more complex and requires screening for Shiga toxins or their encoding genes. Culture isolation and subsequent serotyping is often conducted only at public health laboratories. Diagnosis of non-O157 STEC is disproportionately underutilized, even in countries where their diagnosis is recommended. Consequently, surveillance for non-O157 STEC in many countries is less inclusive than for O157 STEC and their contribution to incidence of STEC illness has been insufficiently determined.

Notification data, including statutory, capture only a fraction of illnesses that are occurring in the population. In Germany, median annual incidence (per 100 000 population) of notified cases is 0·06 for STEC-associated HUS (and 1·07 for STEC-GE) for 2008–2012, excluding 2011 (https://survstat.rki.de, data version 1 July 2014).

Studies addressing underestimation in notification data and the quantitative relation of non-O157 STEC to O157 STEC are helpful in informing diagnostic and surveillance strategies – as were previous studies for other gastroenteric pathogens [Reference Havelaar11].

The few available studies suggest a true annual incidence of STEC-associated infections between 47 and 100/100 000 population for Europe [Reference Majowicz12] and Northern America [Reference Scallan13, Reference Thomas14] and 0·15 STEC-associated HUS [Reference Majowicz12]. Estimated proportions of non-O157 in STEC-GE were 62% and 64% in Canada [Reference Thomas14] and the United States [Reference Scallan13], respectively. All available studies extrapolated data from different countries or data on other pathogens than STEC for their estimation models [Reference Majowicz12Reference Thomas14], thus introducing a further source adding to the inherent uncertainty of stochastic modelling. Furthermore, estimates of overall STEC-GE and the proportion of O157 STEC are based, at best, on STEC-GE surveillance data [Reference Scallan13] with all of its diagnostic vagaries mentioned above, or on assumptions [Reference Majowicz12, Reference Thomas14] but not on HUS statutory surveillance data.

Our objectives were to estimate annual frequency and incidence of STEC-associated HUS and STEC-GE in Germany based on German national notification data for enteropathic HUS – overall and separately for O157 STEC and non-O157 STEC – to inform diagnostic, and surveillance strategies.

METHODS

Using HUS national notification data as a starting point, we modelled true annual incidence of STEC illness in Germany separately for O157 and non-O157 STEC, taking into account group-specific underestimation caused by underreporting of notification data and under-ascertainment (see Fig. 1).

Fig. 1. Modelling true annual incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS).

Diagnosis and surveillance of STEC-GE and ‘enteropathic’ HUS in Germany

In Germany, diagnosis of STEC in GE and HUS patients is based on detection of Shiga toxins or their encoding genes in stool enrichment culture or isolates. Subsequent culture isolation and serotyping is recommended but not mandatory and is rarely performed in clinical laboratories. In HUS patients, evidence for an STEC infection can also be established by detecting anti-lipopolysacharide IgM antibodies against E. coli serogroups in blood by specialized laboratories (which in the study period included only antibodies against serogroup O157).

According to the German Protection against Infection Act, both laboratory detection of STEC infection in stool and clinically diagnosed ‘enteropathic’ (i.e. GE-associated) HUS are notifiable (see Supplementary material for national surveillance case definitions).

Electronic case reports are sent from the local health department via State Health Departments to the federal-level public health institute, the Robert Koch Institute (RKI), where reports are hosted in a national database. In addition, RKI conducts active surveillance for paediatric HUS since 2008 in collaboration with the German Society for Paediatric Nephrology. This surveillance entails monthly inquiries to all paediatric nephrology centres (PNCs) in Germany about incident HUS cases in children (aged <18 years) during the past month.

Risk model for STEC illness in Germany

We used German notification data on enteropathic HUS, reported to RKI for the years 2008–2012 (excluding 2011 because of a large outbreak of STEC O104:H4 [Reference Frank10]) as the basis to estimate the true annual incidence of STEC-GE in Germany.

We computed estimates separately for O157 and non-O157 STEC groups, taking into account the groups’ different average capability of causing acute bloody diarrhoea [Reference Werber15] and HUS, and the resulting difference in under-ascertainment caused by symptomatic cases not attending health facilities (for differences in clinical severity) or by not being correctly diagnosed as a case (for differences in diagnostics as outlined above). Furthermore, underreporting of cases from health facilities to public health authorities adds to underestimation of STEC-GE incidence.

Table 1. Input parameters for the risk model to estimate true incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS)

CrI, Credible interval; PNC, paediatric nephrology centre.

* Nominator.

Denominator.

§ The unit of measurement is person-years-at-risk for this parameter.

For Gamma(r, λ), r = s and λ = 1/N; for Beta(a, b), a = Sum(s) + 1 and b = Sum(N) – Sum(s) + 1.

Our estimations were conducted in the following sequence (see also Fig. 1):

  1. (a) Adjustment for underreporting of HUS. To estimate the true median annual numbers of enteropathic HUS, adjustment for underreporting was conducted separately for cases treated in PNCs and non-PNCs. For PNCs, we used a two-source capture–recapture approach (statutorily passive HUS surveillance and active paediatric HUS surveillance) to estimate the magnitude of underreporting of notification data. We assumed underreporting by non-PNCs to be up to ten times more common than in PNCs as HUS cases are infrequently treated in these institutions. Consequently, knowledge of infectious disease notification requirements, otherwise seldom needed in nephrology units, is likely to be less prevalent in medical personnel in non-PNCs.

  2. (b) Estimating the proportion of STEC-associated HUS. Evidence of STEC infection cannot be established in every case of ‘enteropathic’ HUS. Using literature (described in detail in the Supplementary material) on microbiological evidence of STEC in HUS patients in Germany, we estimated the proportion of enteropathic HUS caused by STEC infection [Reference Gerber16]. This proportion was subsequently multiplied by the estimated number of all HUS cases per year to obtain the number of estimated STEC-associated HUS cases.

  3. (c) Estimating the proportion of O157 and non-O157 in STEC-associated HUS. The proportion of O157 in STEC-associated HUS in Germany was derived from the literature [Reference Gerber16, Reference Mellmann17] and combined as outlined in Table 1. This proportion was multiplied by the annual number of STEC-associated HUS cases to estimate the O157-associated HUS cases (the remaining STEC-HUS cases were thus non-O157 associated). All further calculations were conducted separately for O157- and non-O157-associated HUS cases.

  4. (d) Estimating the number of laboratory-confirmed STEC-GE cases per HUS case. Using literature information on the proportion of HUS cases in laboratory-confirmed STEC-GE cases [Reference Werber18], we multiplied the estimated annual number of STEC-associated HUS cases by the factor for STEC-GE cases per STEC-associated HUS case separately for O157 and non-O157 (beta distribution).

  5. (e) Estimating the proportion of bloody diarrhoea in O157 and non-O157 STEC-GE cases. In addition, we used the literature for estimates on the proportion of bloody diarrhoea in O157 and non-O157 STEC-GE cases [Reference Werber18]. Annual frequencies for STEC-GE with bloody and non-bloody diarrhoea were used to account for under-ascertainment according to severity in a next step (separately for O157 and non-O157).

  6. (f) Estimating the under-ascertainment of bloody and non-bloody diarrhoea. Under-ascertainment was accounted for in a procedure incorporating three steps: using literature information, we first estimated the proportion of symptomatic patients consulting a physician, thereafter the proportion of patients that provided stool specimens for microbiological testing [Reference Haagsma19, Reference Hauri, Uphoff and Gawrich20] and finally the proportion of stool samples tested for STEC [Reference Kist21] based on German laboratory recommendations on test strategies for faecal samples [Reference Kist21].

The estimated annual number of true STEC-GE cases and STEC-associated HUS cases in Germany, differentiated for O157 and non-O157, were converted to annual cumulative incidence/100 000 population, using the mean population size of Germany for 2008–2012 (excluding 2011), obtained from Germany's Federal Statistical Office.

Evaluation of uncertainty

We used Monte Carlo simulation in @RISK v. 6.1.1 (Palisade Corp., USA) with Latin Hypercube sampling and 10 000 iterations to evaluate uncertainty in the outputs. All input data was considered to be subject to uncertainty and parameters were therefore described by probability distributions. Generally, proportions were described by beta distributions and the HUS rate was described by a gamma distribution [Reference Vose22]. Pert distributions were used for multiplication factors where sufficient data to inform beta distributions was unavailable. Distribution parameterization was done as displayed in Table 1. The results are reported as the median and the 95% credible interval.

A sensitivity analysis was conducted to evaluate the contribution of the input parameters to the overall uncertainty in outcome estimates to identify which input parameter shows the biggest influence on the output.

In addition we examined two scenarios using alternative values of particularly uncertain input parameters to investigate their effect on the outcome estimates, keeping all other variables of the model constant. (For details see Supplementary material.) In a conservative scenario (scenario 1) we assumed that degree of underreporting of HUS did not differ between PNCs and non-PNCs and that all stool samples submitted for microbiological testing were investigated for STEC regardless of whether blood was visible. In scenario 2 we re-parameterized the model using input parameters for under-ascertainment based on findings of a survey in the Federal state of Hesse in children aged <16 years [Reference Hauri, Uphoff and Gawrich20], to account for under-ascertainment in our estimates of the higher incidence of STEC illness in children.

Literature survey

We searched Medline and Scopus literature for information about STEC in Germany published since inception of the Medline and Scopus bibliographic databases to 31 December 2014 with the objective of identifying, for patients in Germany, the proportion of STEC-associated HUS in enteropathic HUS cases (as input parameter for estimation step b), the proportion of O157 STEC in STEC-associated HUS (step c) and the proportion of HUS and bloody diarrhoea in laboratory-confirmed STEC-GE separately for O157 and non-O157 serogroups (step d). Our second objective was to identify under-ascertainment of bloody and non-bloody diarrhoea (step f), including the proportion of physician consultations in cases of bloody and non-bloody diarrhoea and the proportion of physicians taking stool samples in cases of bloody and non-bloody diarrhoea.

We used the search terms (enterohaemorrhagic Escherichia coli OR STEC OR Escherichia coli O157 OR E. coli O157) AND (Germany) to identify input parameters for steps be. We used search terms (gastroenteritis OR gastrointestinal illness OR gastrointestinal infections) AND Germany AND (healthcare OR medical care) in titles and abstracts for step f.

We required articles for all steps to provide data in sufficient detail for O157 and non-O157 regarding proportion of HUS and bloody diarrhoea and to refer to data that pertained to Germany recognizing that serogroup distribution among GE and HUS cases as well as health-seeking behaviour may vary between countries. In addition, we required information for steps df to be derived from population-based surveys or sentinel surveillance projects to increase accuracy of these estimates. Search results for Medline and Scopus were combined and de-duplicated. Two investigators screened documents independently, in the event of discrepancies consensus by discussion was sought. Documents were first screened by reviewing titles and abstracts were available. Identified documents were screened against inclusion and exclusion criteria outlined above. From the identified documents absolute numbers were extracted and used as input variables in the estimation model as outlined in Table 1.

RESULTS

We identified five relevant publications, three for steps be and two for step f [Reference Gerber16Reference Hauri, Uphoff and Gawrich20] that together provided information for all required input parameters (see Fig. 2 and Supplementary material). These publications, German notification data and German laboratory guidelines formed the backbone of the simulation model and are outlined in Table 1.

Fig. 2. Results of the systematic review to identify input parameters for the estimation of the true incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS).

We estimated a median annual number of 90 cases of STEC-associated HUS in Germany during the study period, corresponding to an incidence (per 100 000 population) of 0·11 [95% credible interval (CrI) 0·08–0·20]; a median of 60 cases due to STEC O157 (incidence 0·07, 95% CrI 0·05–0·13) and a median of 29 cases due to non-O157 STEC (incidence 0·04, 95% CrI 0·03–0·07) (see Table 2). From these, we estimated that a median of 28 347 STEC-GE cases occurred per year in the German population, indicating an incidence of 35 (95% CrI 12–145); a median of 4969 cases due to O157 STEC (incidence 6.1, 95% CrI 2·2–24) and a median of 22 019 cases due to non-O157 STEC (incidence 27, 95% CrI 8·0–133).

Table 2. Results of modelling median annual case numbers and median annual incidence (with 95% credible intervals) of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS)

* Per 100 000 population.

Our estimates correspond to a median annual underestimation of STEC-associated HUS and STEC-GE in the German notification data by a factor of 1·8 (95% CrI 1·3–3·3) and 32 (95% CrI 11–135), respectively.

Non-O157 STEC accounted for 81% (95% CrI 49–96) of all STEC-GE and 51% (95% CrI 16–86) of all bloody STEC-associated diarrhoea.

Sensitivity analysis indicated that the proportion of HUS cases in laboratory-confirmed non-O157 STEC exerted the biggest influence on the outcome of all input parameters, followed by the proportion of stool samples tested for STEC and the completeness of HUS notifications from non-PNCs (see Fig. 3).

Fig. 3. Sensitivity analysis of influence of input parameters on frequency of STEC-GE in Germany based on notification data of haemolytic uraemic syndrome (HUS).

In scenario analysis, the median annual incidence (per 100 000 population) of STEC-GE ranged from 17 (95% CrI 7·6–61) in scenario 1 to 72 (95% CrI 22–339) in scenario 2 and of STEC-associated HUS from 0·08 (95% CrI 0·07–0·09) in scenario 1 to 0·11 (95% CrI 0·08–0·20) in scenario 2 (unchanged to the point estimate).

The proportion of non-O157 STEC in STEC-GE, bloody diarrhoea and STEC-associated HUS did not vary in the different scenarios (see Supplementary material for detailed results).

DISCUSSION

We estimated the true frequency and incidence of STEC illness in the German population, separately for O157 and non-O157 STEC, based on statutory notification data on HUS. The study yielded the following main findings: The median annual incidence (per 100 000 population) was estimated at 35 (95% CrI 12–145·00) for STEC-GE and 0·11 for STEC-associated HUS (95% CrI 0·08–0·20). German notification data underestimated STEC-associated HUS and STEC-GE incidences by factors of 1·8 and 32·3, respectively. Non-O157 STEC accounted for ~80% of all STEC-GE, half of all bloody STEC-associated diarrhoea and one-third of all STEC-associated HUS cases, hence contributing to STEC illness to an even larger extent than previously estimated [Reference Scallan13, Reference Thomas14].

Our incidence point estimates for STEC-GE and HUS are slightly lower than those published for Europe (47 and 0·15, respectively) [Reference Majowicz12], the United States (59 for STEC-GE) [Reference Scallan13] and Canada (100 for STEC-GE) [Reference Thomas14], but in keeping considering the degree of uncertainty accompanying our estimate. The incidence for O157 STEC-GE is in particular lower than estimated for other European countries such as The Netherlands [Reference Haagsma19, Reference Havelaar23], Denmark or the UK [Reference Haagsma19], and for the United States and Canada [Reference Majowicz12, Reference Scallan13]. In Germany, neither laboratory-based (passive) surveillance of STEC-GE nor (active) HUS surveillance ever identified an outbreak with ‘classical’ non-sf O157 STEC comprising ⩾5 persons, but did so for outbreaks with other serotypes [Reference Alpers24, Reference Werber25]. We are unaware of specific control plans for O157 STEC in animal reservoirs or the food-production chain that would explain this observation. Thus, our estimation of a comparatively low O157 STEC incidence adds additional weight to the view that O157 STEC poses a limited public health problem in Germany.

Of note, according to surveillance data (2008–2012, excluding 2011) reported to the European Centre for Disease Control and Prevention (ECDC) from other countries in the European Union, a slightly higher percentage (40%, 391/659) of all STEC identified in reported HUS patients belonged to non-O157 serogroups (data provided by ECDC extracted from The European Surveillance System; TESSy). This may indicate that non-O157 STEC contribute to STEC-GE incidence in other European countries even more than in Germany (where non-O157 STEC account for 80% of STEC-GE). Yet, only 33% of STEC-GE captured in surveillance systems in Europe were attributed to infection by non-O157 strains during the study period [26, 27], underscoring the large degree of under-ascertainment of these STEC strains in GE patients in Europe. In recent years, the proportion of non-O157 STEC increased, probably indicating a more frequent use of serogroup-independent testing in Europe [26, 27].

In Germany, the contribution of the different non-O157 serogroups to STEC illness has remained fairly constant over the last 10 years (except in 2011) according to German surveillance data with serogroups O26, and O103 being the most frequently isolated non-O157 STEC in children and O91 in adults [Reference Werber18, Reference Werber28]. The numerous different non-O157 STEC vary markedly in their virulence. On average though, they less frequently causes life-threatening HUS (in children) or disease outbreaks, and, importantly, their diagnosis currently is more complex, time-consuming and expensive. Thus, the question about the cost-effectiveness of screening for non-O157 has been raised [Reference Kiska and Riddell29, Reference Marcon30]. Apart from their markedly more frequent occurrence as an aetiological agent in human GE than STEC O157 and their substantial contribution to the burden of bloody diarrhoea and HUS, new STEC strains are likely to evolve, some of which will cause outbreaks (e.g. STEC O104:H4) [Reference Frank10]. For the latter reason alone we believe that modern STEC diagnosis and, consequently, surveillance systems should encompass timely detection of non-O157 STEC (including information on the serotype or other epidemiologically meaningful subtyping information), even in countries where STEC O157 appears to dominate.

Validity of risk model

Our ‘top-down’ approach of estimating STEC incidence based on HUS notification data is new and we believe is advantageous for at least two reasons. First, statutory HUS surveillance is more sensitive than STEC-GE surveillance and in conjunction with active paediatric HUS surveillance in Germany allowed for an accurate estimate of its underreporting. Furthermore, STEC aetiology in (paediatric) HUS patients has been extensively studied in Germany [Reference Gerber16, Reference Mellmann17]. Taken together, HUS incidence and the individual contribution of O157 and non-O157 STEC could be estimated with little uncertainty.

Second, our estimations were purposively based solely on information on STEC in Germany, preventing the need of extrapolating from data gathered in other countries as another source of uncertainty.

By far the greatest source of uncertainty was the proportion of HUS in patients infected by a non-O157 STEC because it was based on small numbers. However, our estimate is in agreement with data from other countries [Reference Gould31]. Likewise, other findings are corroborated by data sources not used in our estimation. For example, the estimated proportion of non-O157 STEC-associated HUS (33%) is consistent with that observed in national HUS notification data during the study period (34%). Furthermore, the proportion of non-O157 serogroups in STEC-GE and STEC-associated bloody diarrhoea in Germany is consistent with both national notification data on STEC-GE and with a nationwide laboratory sentinel conducted at the beginning of the century in Germany [Reference Werber18].

Limitations

As with previously published risk models, ours did not account for the effect of age because age-specific data were unavailable for many estimation steps. Yet, the serogroup-specific incidence for STEC-GE and HUS incidence vary with age. Most available studies focused exclusively or primarily on children (who should have the highest true incidence of STEC-GE and HUS in Germany), which is why uncertainty of estimates is likely highest for adults.

In addition, non-O157 STEC consist of different pathogens with a variety of virulence genes, and estimates for non-O157 relate to the fairly stable serogroup distribution (and assumed average genomic content within serogroups) in Germany, which can be different in other countries. Models based on virulence-genes might be preferable but the necessary input data (e.g. stx-type, presence of the eae gene) are currently not available in sufficient detail.

Furthermore, some input data of our risk model lack an evidence base as no study was available to support our assumptions, such as underreporting from non-PNCs and the adherence to laboratory guidelines for testing stool samples of GE cases. These two parameters were in the top-3 influential parameters in the sensitivity analysis, warranting further data collection to decrease this uncertainty. Furthermore, not all literature sources used for our risk model distinguished between (rare) sf-O157 STEC and (‘classic’) non-sf-O157 STEC. Because sf-O157 STEC infection progresses with a higher probability from diarrhoea to HUS [Reference Nielsen32], we slightly overestimated STEC-GE incidence of serogroup O157.

Completeness of HUS notification is probably overestimated in our study because concurrently conducted active paediatric surveillance included reminders of notification obligations when continuously monitoring HUS cases ascertained in the active system.

CONCLUSIONS

Statutory notification data largely underestimate STEC-GE in Germany, where STEC diagnosis is based on serogroup-independent testing for Shiga toxins or their encoding genes.

The contribution of non-O157 serogroups to STEC-GE incidence appears to be higher than previously estimated [Reference Scallan13, Reference Thomas14], not only including a large number of mild illnesses but also half of all STEC-associated bloody diarrhoea cases. Considering European surveillance data on HUS, this finding is probably true for many other countries in Europe. Surveillance of HUS complements that of STEC-GE, not only by allowing for the detection of outbreaks that otherwise go unrecognized [Reference Werber33] and reliably monitoring trends of STEC infection [Reference Mahon34], but also by aiding in estimating STEC incidence, thereby helping to validate notification data.

Non-O157 STEC should be considered in parallel with STEC O157 when searching aetiology in patients with GE or HUS, and accounted for in modern surveillance systems for STEC illness.

APPENDIX. Collaborators of the HUS active surveillance network Germany

Oliver Amon (University Hospital Tuebingen, Tuebingen), Rainer Büscher (University Hospital Essen, Essen), Tobias Hampel (Children's Hospital Memmingen, Memmingen), Henry Fehrenbach (Children's Hospital Memmingen, Memmingen), Sandra Habbig (University Hospital of Cologne, Cologne), Martin Pohl (University Hospital Freiburg, Freiburg), Karsten Häffner (University Hospital Freiburg, Freiburg), Bernd Hoppe (University Hospital Bonn, Bonn), Günter Klaus (University Children's Hospital, Marburg), Martin Konrad (University Children's Hospital, Münster), Kay Latta (Clementine Children's Hospital, Frankfurt), Heinz Leichter (Olgahospital, Stuttgart), Sebastian Loos (University Medical Center Hamburg-Eppendorf, Hamburg), Carmen Montoya (Children's Hospital Schwabing, München), Dominik Müller (Berlin Medical University Centre ‘Charité’), Matthias Galiano (Medical University, Erlangen), Evelin Muschiol (Medical University, Erlangen), Lars Pape (Hannover Medical School, Hannover), Hagen Staude (University Children's Hospital Rostock, Rostock), Elke Wühl (Center for Pediatrics and Adolescent Medicine, Heidelberg), Michael Henn (St Georg Hospital, Leipzig), Simone Wygoda (St Georg Hospital, Leipzig), Michael Pohl (Children's Hospital Friedrich Schiller University Jena, Jena).

SUPPLEMENTARY MATERIAL

For supplementary material accompanying this paper visit http://dx.doi.org/10.1017/S0950268816001436.

ACKNOWLEDGEMENTS

The authors thank Anja Hauri for providing data on a survey in the German Federal state of Hesse. The authors are grateful for data provided by ECDC extracted from The European Surveillance System (TESSy).

DECLARATION OF INTEREST

None.

References

REFERENCES

1. Boerlin, P, et al. Associations between virulence factors of shiga toxin-producing Escherichia coli and disease in humans. Journal of Clinical Microbiology 1999; 37: 497503.Google Scholar
2. Preussel, K, et al. Shiga toxin-producing Escherichia coli O157 is more likely to lead to hospitalization and death than non-O157 serogroups – except O104. PLoS ONE 2013; 8: e78180.CrossRefGoogle ScholarPubMed
3. Hadler, JL, et al. Ten-year trends and risk factors for non-o157 shiga toxin-producing Escherichia coli found through shiga toxin testing, Connecticut, 2000–2009. Journal of Clinical Infectious Diseases 2011; 53: 269276.Google Scholar
4. Hedican, EB, et al. Characteristics of O157 versus non-O157 shiga toxin-producing Escherichia coli infections in Minnesota, 2000–2006. Journal of Clinical Infectious Diseases 2009; 49: 358364.Google Scholar
5. Tarr, PI, Gordon, CA, Chandler, WL. Shiga-toxin-producing Escherichia coli and haemolytic uraemic syndrome. Lancet 2005; 365: 10731086.Google Scholar
6. Donnenberg, MS, Whittam, TS. Pathogenesis and evolution of virulence in enteropathogenic and enterohemorrhagic Escherichia coli . Journal of Clinical Investigation 2001; 107: 539548.Google Scholar
7. Gould, LH, et al. Recommendations for diagnosis of shiga toxin-producing Escherichia coli infections by clinical laboratories. Morbidity and Mortality Weekly Reports 2009; 58: 114.Google ScholarPubMed
8. Bielaszewska, M, et al. Phenotypic and molecular analysis of tellurite resistance among enterohemorrhagic Escherichia coli O157:H7 and sorbitol-fermenting O157:Nm clinical isolates. Journal of Clinical Microbiology 2005; 43: 452454.CrossRefGoogle ScholarPubMed
9. Werber, D, et al. Watch out for the even eviler cousin – sorbitol-fermenting E. coli O157. Lancet 2011; 377: 298299.Google Scholar
10. Frank, C, et al. Epidemic profile of shiga-toxin-producing Escherichia coli O104:H4 outbreak in germany. New England Journal of Medicine 2011; 365: 17711780.CrossRefGoogle Scholar
11. Havelaar, AH, et al. Disease burden of foodborne pathogens in the Netherlands, 2009. International Journal of Food Microbiology 2012; 156: 231238.Google Scholar
12. Majowicz, SE, et al. Global incidence of human shiga toxin-producing Escherichia coli infections and deaths: a systematic review and knowledge synthesis. Foodborne Pathogens and Disease 2014; 11: 447455.CrossRefGoogle ScholarPubMed
13. Scallan, E, et al. Foodborne illness acquired in the united states – major pathogens. Emerging Infectious Diseases 2011; 17: 715.CrossRefGoogle ScholarPubMed
14. Thomas, MK, et al. Estimates of the burden of foodborne illness in Canada for 30 specified pathogens and unspecified agents, circa 2006. Foodborne Pathogens and Disease 2013; 10: 639648.Google Scholar
15. Werber, D, et al. Shiga toxin-producing Escherichia coli O157 more frequently cause bloody diarrhea than do non-O157 strains. Journal of Infectious Diseases 2004; 189: 13351336; author reply 13361337.CrossRefGoogle ScholarPubMed
16. Gerber, A, et al. Clinical course and the role of shiga toxin-producing Escherichia coli infection in the hemolytic-uremic syndrome in pediatric patients, 1997–2000, in Germany and Austria: a prospective study. Journal of Infectious Diseases 2002; 186: 493500.Google Scholar
17. Mellmann, A, et al. Analysis of collection of hemolytic uremic syndrome-associated enterohemorrhagic Escherichia coli . Emerging Infectious Diseases 2008; 14: 12871290.Google Scholar
18. Werber, D, et al. Shiga toxin-producing Escherichia coli infection in Germany: Different risk factors for different age groups. American Journal of Epidemiology 2007; 165: 425434.Google Scholar
19. Haagsma, JA, et al. Community incidence of pathogen-specific gastroenteritis: Reconstructing the surveillance pyramid for seven pathogens in seven European Union member states. Epidemiology and Infection 2013; 141: 16251639.Google Scholar
20. Hauri, AM, Uphoff, H, Gawrich, S. The burden of acute gastrointestinal illness in Hesse – a telephone survey [in German]. Das Gesundheitswesen 2011; 73: 7884.Google Scholar
21. Kist, MM, et al. MIQ 09: Gatrointestinal Infections. Quality Standards in Microbiological Dagnostics, 2nd edn [in German]. Urban & Fischer Verlag/Elsevier, 2013.Google Scholar
22. Vose, D. Risk Analysis: A Quantitative Guide, 3rd edn. West Sussex, England: John Wiley & Sons, 2008.Google Scholar
23. Havelaar, AH, et al. Disease burden in the netherlands due to infections with shiga toxin-producing Escherichia coli O157. Epidemiology and Infection 2004; 132: 467484.Google Scholar
24. Alpers, K, et al. Sorbitol-fermenting enterohaemorrhagic Escherichia coli O157:H- causes another outbreak of haemolytic uraemic syndrome in children. Epidemiology and Infection 2009; 137: 389395.Google Scholar
25. Werber, D, et al. A multistate outbreak of shiga toxin-producing Escherichia coli O26:H11 infections in Germany, detected by molecular subtyping surveillance. Journal of Infectious Diseases 2002; 186: 419422.Google Scholar
26. European Centre for Disease Prevention and Control. Surveillance of six priority food- and waterborne diseases in the eu/eea 2006–2009. Stockholm: ECDC, 2013.Google Scholar
27. European Centre for Disease Prevention and Control. Surveillance of seven priority food- and waterborne diseases in the EU/EEA 2010–2012. Stockholm: ECDC, 2015.Google Scholar
28. Werber, D, et al. Shiga toxin-producing Escherichia coli serogroups in food and patients, germany. Emerging Infectious Diseases 2008; 14: 18031806.Google Scholar
29. Kiska, DL, Riddell, SW. Counterpoint: should all stools be screened for shiga toxin-producing Escherichia coli? Journal of Clinical Microbiology 2011; 49: 23942397.Google ScholarPubMed
30. Marcon, MJ. Point: Should all stools be screened for shiga toxin-producing Escherichia coli? Journal of Clinical Microbiology 2011; 49: 23902394.Google Scholar
31. Gould, LH, et al. Increased recognition of non-O157 shiga toxin-producing Escherichia coli infections in the united states during 2000–2010: epidemiologic features and comparison with E. coli O157 infections. Foodborne Pathogens and Disease 2013; 10: 453460.Google Scholar
32. Nielsen, S, et al. Desperately seeking diarrhoea: outbreak of haemolytic uraemic syndrome caused by emerging sorbitol-fermenting shiga toxin-producing Escherichia coli O157:H, Germany, 2009. Journal of Zoonoses in Public Health 2011; 58: 567572.CrossRefGoogle Scholar
33. Werber, D, et al. Looking for tips to find icebergs – surveillance of haemolytic uraemic syndrome to detect outbreaks of shiga toxin-producing E. coli infection. Eurosurveillance 2008; 13.CrossRefGoogle ScholarPubMed
34. Mahon, BE, et al. Hemolytic uremic syndrome surveillance to monitor trends in infection with Escherichia coli O157:H7 and other shiga toxin-producing E. coli . Emerging Infectious Diseases 1997; 3: 409412.Google Scholar
Figure 0

Fig. 1. Modelling true annual incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS).

Figure 1

Table 1. Input parameters for the risk model to estimate true incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS)

Figure 2

Fig. 2. Results of the systematic review to identify input parameters for the estimation of the true incidence of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS).

Figure 3

Table 2. Results of modelling median annual case numbers and median annual incidence (with 95% credible intervals) of O157 and non-O157 STEC illness in Germany based on notification data of haemolytic uraemic syndrome (HUS)

Figure 4

Fig. 3. Sensitivity analysis of influence of input parameters on frequency of STEC-GE in Germany based on notification data of haemolytic uraemic syndrome (HUS).

Supplementary material: File

Kuehne supplementary material

Kuehne supplementary material 1

Download Kuehne supplementary material(File)
File 175.6 KB