• 
• 

Three Part Question

In [a child with Escherichia coli O157:H7 infection] does [a high white blood cell count] predict [progression to haemolytic uraemic syndrome]?

Clinical Scenario

You are the specialist registrar in paediatrics doing the ward rounds. There has been a recent outbreak of Escherichia coli (E coli) O157:H7 gastroenteritis in the community. A 5-year old girl was admitted 3 days ago with bloody diarrhoea, abdominal cramps and vomiting. Her stool has grown E coli O157:H7. She is now stable and her parents are keen to take her home but are understandably anxious and ask you about possible complications. You are aware of the risk for progression to haemolytic uraemic syndrome (HUS). However, you are not sure if all such patients should be closely monitored. You note that the patient now has a normal platelet count and renal function and wonder if there are any simple parameters to predict the risk of HUS, which may take up to 2 weeks to develop.
You talk to the consultant in public health medicine, who kindly directs to you to the national guidelines by the Health Protection Agency (HPA) on the management of E coli O157:H7 infections. Unfortunately, the guidelines do not answer your question. Your consultant has come across anecdotal evidence that leukocytosis may be a predictor for HUS in such children.
You decide to do a literature search and critically appraise the evidence.

Search Strategy

MEDLINE (1966 to date) and OLDMEDLINE (1950–1965) searched by the PubMed interface on 29/10/2006.
"Hemolytic uremic syndrome" AND "Escherichia coli O157" with limits (English, Human, All child: 0–18 years) revealed 267 results.
Addition of the key words (leukocytosis OR "white cell count") to the above yielded only seven results. This was felt to be an inappropriately low number, possibly due to variations in the terms used by authors ("elevated WBC", "high leukocytes", "increased white cell count", etc).

Search Outcome

A wider final search strategy adding suitable new key words to the original search was attempted as follows: ("Hemolytic uremic syndrome" AND "Escherichia coli O157") AND (risk OR predictors).
This resulted in 71 hits. Among these, no systematic reviews relevant to the clinical question could be identified. Ten articles were found to be of relevance

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Tserenpuntsag et al, 2005, USA	Hospitalised patients with confirmed E Coli O157 infection, 238 with full case notes included 1998-99	Case control study (level 3b)	Risk factor: WBC >13x10 to the power of 9/l	Confirmed or probable HUS p<0.001. Adjusted OR 4.4 (1.6 to 12.6)	Sound methodology. Clear definitions for confirmed and probable cases Confounding variables accounted for However, study also included adults
Wood et al, 2001, Scotland	170 confirmed or probable E coli O157:H7 infections from Central Scotland outbreak of 1996 monitored in community	Prospective cohort study (level 2b)	Risk factor: WBC> upper limit for age (reference ranges from local laboratory)	Confirmed HUS: Sensitivity 88.9% (51.8 to 99.7). Specificity 79.2% (74.1 to 84.4) Negative predictive value 99.5% (97.1 to 100)	Strict exclusion criteria, well validated data. Patients with severe initial presentation needing hospitalisation excluded. Proportion of children in sample size not clear. Concludes absence of leukocytosis indicates very low risk for HUS. Relative risk for absolute WBC counts not provided.
Dundas et al, 2001, Scotland	120 hospitalised patients with E coli O157:H7 from Central Scotland outbreak of 1996	Case control study (level 3b)	Risk factor: WBC >20x10 to the power of 9/l	Confirmed HUS OR 8.3 (1.1 to 60.3)	Strict exclusion criteria. Confounding variables not fully explained. Study also included adults. Possible overlap with Wood et al's paper on the same outbreak. Gives separate data for neutrophils. Wide confidence intervals for risk
Ikeda et al, 2000, Japan	288 inpatient school children in the 1996 Sakai outbreak in Japan	Case-control study (level 3b)	Risk factor WBC >11x10 to the power of 9/l	Confirmed HUS OR 5.03 (2.13 to 11.87)	Exclusively paediatric study. Confounding factors not clear. High CRP also noted to be significant. "Incomplete" HUS excluded from final analysis. Postulates a "predictor index scoring system" based on WBC count, CRP and temperature.
Buteau et al, 2000, Canada	All children < 18 years with culture-proved sporadic E coli O157:H7 infections seen at Sainte-Justine Hospital 1987–1997	Retrospective cohort study (level 2b)	Risk factor WBC >11x10 to the power of 9/l	Confirmed HUS RR 3.8 (1.5 to 9.7) p<0.002	Excellent methodology Large focussed study looking at leukocytosis Rigid validation criteria. Clear progressive association of HUS with rising WBC count
Wong et al, 2000, USA	Cohort of 71 children <10 years with E Colie O157 positive stool cultures	Prospective cohort study (level 2b)	Risk factor WBC >13x10 to the power of 9/l	Confirmed HUS:Adjusted RR 6 (1.2 to 29.8) p<0.03	Exclusive paediatric study with sound methodology, strict inclusion criteria and clear definitions. Main aim of the study was to evaluate antibiotic-associated risk
Kawamura et al, 1999, Japan	126 junior high school children with proven E Coli O157:H7 infection in the 1996 Sakai outbreak	Prospective cohort study (level 4)	Risk factor: WBC >10x10 to the power of 9/l	Confirmed HUS:34% of patients with WBC above the cut-off level of 10x10 to the power of 9/l developed HUS Significant difference in median, minimum and maximum WBC counts amoung the two groups	Conclusions are based on imperfect statistical analysis. Risk not quantified and confidence limits not provided. Both high WBC and elevated CRP identified as independent risk factors.
Bell et al, 1997, USA	278 Washington State children <16 years with symptomatic confirmed E Coli O157:H7 infection in the 1990 Saitama epidemic	Prospective study (level 4)	Absolute WBC cut-off count not defined. 'Rising' WBC count implicated as a risk factor	'Confirmed' and 'incomplete' HUS: Significant differences in WBC count (p<0.05) on day 4 of illness	Asymptomatic children were also included by screening. Final analysis also included 'incomplete' HUS Study grossly handicapped by lack of clear data and insufficient statistical analysis
Pavia et al, 1990, USA	20 residents with E Coli O157:H7 from two institutions for mentally retarded children in Utah; 13 culture confirmed	Case control study (level 4)	Risk factor: Elevated WBC count (median count among cases 23.7x10 to the power of 9/l)	Confirmed HUS:Median WBC count cases vs controls 23.7x10 to the power of 9/l vs 9.1x10 to the power of 9/l, p=0.018	Also included unconfirmed cases of possible E coli infection. Limited size of 'case' and 'control' population Odds ration and confidence limits are not provided

Comment(s)

E coli O157:H7 is a relatively uncommon cause of infectious gastroenteritis in children. However, the severity of symptoms, the possibility of outbreaks and the potential for severe complications means it is of considerable public health importance. Infections with this subtype are known to lead to HUS in 2–7% of overall cases, which may rise to 30% in outbreaks (Griffin, Morgan). The disease can be particularly severe in children, with reported case fatality rates of 3–17% and a significant incidence of long-term sequelae in survivors (Fitzpatrick). Since more than 90% of children with E coli O157:H7 infection do not progress to HUS, intensive clinical and/or laboratory monitoring of all cases may not be feasible or cost-effective. Attempts have been made to set up monitoring protocols to identify high-risk patients and help with early interventions (Wood). As the median time for development of HUS is 9 days (range 1–15 days), early identification of risk factors will enable triage into high- and low-risk groups and help target interventions appropriately. Identification of clinical risk factors including the intensity of fever, vomiting, duration of illness and presence of blood in stools has been attempted with limited success (Ikeda, Wong). More recent efforts have focussed on objective laboratory criteria such as raised inflammatory markers, low serum protein, elevated lactate dehydrogenase levels and albuminuria (Wood, Dundas). An elevated white blood cell (WBC) count has been noted to be associated with a higher risk for HUS, with some authors claiming a sensitivity and specificity of around 90% (Wood). Being a rare complication of an uncommon infection, case–control studies may be appropriate to define risk in this clinical situation (Tserenpuntsag). However, clustering of a large number of cases during epidemics may enable cohort studies to be performed with relative ease (Wood, Ikeda, Buteau, Wong, Kawamura, Bell). Most authors in our analysis have identified E coli infection on the basis of positive stool cultures and some have also used seropositivity and/or clinical diagnosis (Wood, Bell). In all the above studies, the use of standardised criteria has made the diagnosis of HUS unambiguous. Attempts by some authors to subclassify HUS into confirmed/complete and probable/incomplete forms has only academic relevance and does not affect the validity of the conclusions(Tserenpuntsag, Bell). The relatively short time period between exposure and outcome as well as the serious nature of the condition requiring hospitalisation have ensured accurate data collection as well as lack of attrition in all studies. The relatively short time period between exposure and outcome as well as the serious nature of the condition requiring hospitalisation have ensured accurate data collection as well as lack of attrition in all studies. Leukocytosis has consistently been noted as significant by several authors in their attempts to identify clinical, laboratory and medication-related risk factors for progression to HUS (Tserenpuntsag, Wood, Dundas, Ikeda, Buteau, Wong, Kawamura, Bell, Akashi, Pavia). A comprehensive and focussed study by Buteau et al has demonstrated a clear progressive association of HUS with a rising WBC count. Two well conducted studies by Tserenpuntsag et al and Bell et al have clearly defined the high risk associated with a WBC count of >13x109/l even though the former study included adults as well. Similar results have also been obtained by Dundas et al, Ikeda et al and Wong et al, albeit at different WBC cut-off levels. The latter two reports were exclusively paediatric studies. A well validated prospective cohort study by Wood et al revealed a sensitivity of 88.9% and a negative predictive value as high as 99.5% for an elevation in WBC count. We tried to look at two interesting practical issues, namely the timing of WBC count and the absolute cut-off levels to identify high risk. Most studies assessed WBC counts within 72 h of onset of symptoms, although counts up to 5 days were included by some. The definition of a significant cut-off WBC count varies among different authors, but most studies have quantified the risk to be very high above a WBC level of 11x109/l–13x109/l (Tserenpuntsag, Ikeda, Wong, Bell). None of the published studies suggest that leukocytosis is not a predictor for HUS (Tserenpuntsag, Wood, Dundas, Ikeda, Buteau, Wong, Kawamura, Bell, Akashi, Pavia). Indeed, there is biological evidence to implicate the role of leukocytes in the pathogenesis of HUS.15 (Fitzpatrick, 1992). Leukocytosis has thus been unequivocally delineated as a reliable predictor for the development of HUS following E coli O157:H7 infection. Combining this with other putative risk factors, such as elevated C-reactive protein and albuminuria, might improve the accuracy of defining at-risk groups. Well children seen in the community or hospitalised children who recover rapidly may not require follow-up or investigations. However, where the diarrhoea has not resolved, a clinical impression of "well being" in the child has been shown to be unreliable as a reassuring factor (Dundas, Wong). In such situations the use of validated risk factors is necessary to improve prognostic accuracy. Most of these children would have already had a full blood count and hence additional tests may not be required. A normal leukocyte count will provide reliable reassurance against progression to HUS in nine out of 10 cases (Wood). Similarly, leukocytosis will successfully predict progression to HUS in 70–90% of children (Wong, Bell). Interventions like early intravenous fluid therapy have already been proven to reduce the risk for progression to HUS, while other potential measures such as the use of toxin binders remain understudied (MacConnachie). Formulation of a well defined risk assessment tool based on leukocytosis and other risk factors could help to identify high risk children and streamline the monitoring process and early use of appropriate interventions.

Editor Comment

Authors note:We wish to express our gratitude to Dr Paul Godwin (Consultant Microbiologist, Airedale General Hospital, Keighley, West Yorkshire) for initiating the discussions which led to this literature review.

Clinical Bottom Line

Leukocytosis is a valid, independent and reliable early risk factor for predicting progression to HUS in children with E coli O157:H7 infection. (Grade B) The absence of an elevated WBC count during early stages of the illness makes HUS an unlikely event following E coli O157:H7 infection in children. (Grade B) Formulation of a well validated clinical decision rule (CDR) incorporating leukocytosis with other risk factors is likely to help with early identification of children at risk of developing HUS following E coli O157:H7 infection.

References

1. Subcommittee of the PHLS Advisory Committee on Gastrointestinal Infections. Guidelines for the control of infection with verocytotoxin producing Escherichia coli (VTEC). Commun Dis Public Health 2000; 3 (1): 14–23.
2. Tserenpuntsag B, Chang HG, Smith PF, et al. Hemolytic uremic syndrome risk and Escherichia coli O157:H7. Emerg Infect Dis 2005; 11 (12): 1955–7.
3. Wood R, Donaghy M, Dundas S. Monitoring patients in the community with suspected Escherichia coli O157 infection during a large outbreak in Scotland in 1996. Epidemiol Infect 2001; 127 (3): 413–20.
4. Dundas S, Todd WT, Stewart AI, et al. The central Scotland Escherichia coli O157:H7 outbreak: risk factors for the hemolytic uremic syndrome and death among hospitalized patients. Clin Infect Dis 2001; 33 (7): 923–31.
5. Ikeda K, Ida O, Kimoto K, et al. Predictors for the development of haemolytic uraemic syndrome with Escherichia coli O157:H7 infections: with focus on the day of illness. Epidemiol Infect 2000; 124 (3): 343–9.
6. Buteau C, Proulx F, Chaibou M, et al. Leukocytosis in children with Escherichia coli O157:H7 enteritis developing the hemolytic-uremic syndrome. Pediatr Infect Dis J 2000; 19 (7): 642–7.
7. Wong CS, Jelacic S, Habeeb RL, et al. The risk of the hemolytic-uremic syndrome after antibiotic treatment of Escherichia coli O157:H7 infections. N Engl J Med 2000; 342 (26): 1930–6.
8. Kawamura N, Yamazaki T, Tamai H. Risk factors for the development of Escherichia coli O157:H7 associated with hemolytic uremic syndrome. Pediatr Int 1999; 41 (2): 218–22.
9. Bell BP, Griffin PM, Lozano P, et al. Predictors of hemolytic uremic syndrome in children during a large outbreak of Escherichia coli O157:H7 infections. Pediatrics 1997; 100 (1): E12.
10. Akashi S, Joh K, Tsuji A, et al. A severe outbreak of haemorrhagic colitis and haemolytic uraemic syndrome associated with Escherichia coli O157:H7 in Japan. Eur J Pediatr 1994; 153 (9): 650–5.
11. Pavia AT, Nichols CR, Green DP, et al. Hemolytic-uremic syndrome during an outbreak of Escherichia coli O157:H7 infections in institutions for mentally retarded persons: clinical and epidemiologic observations. J Pediatr 1990; 116 (4): 544–51.
12. Griffin PM, Tauxe RV. The epidemiology of infections caused by Escherichia coli O157: H7, other enterohemorrhagic Escherichia coli, and the associated hemolytic uremic syndrome. Epidemiol Rev 1991; 13: 60–98.
13. Morgan D, Newman CP, Hutchinson DN, et al. Verotoxin producing Escherichia coli O157 infections associated with the consumption of yoghurt. Epidemiol Infect 1993; 111: 181–7.
14. Fitzpatrick MM, Shah V, Trompeter RS, et al. Long term outcome of childhood haemolytic syndrome. BMJ 1991; 303: 489–92.
15. Fitzpatrick MM, Shah V, Filler G, et al. Neutrophil activation in the haemolytic uraemic syndrome: free and complexed elastase in plasma. Pediatr Nephrol 1992; 6 (1): 50–3.
16. MacConnachie AA, Todd WT. Potential therapeutic agents for the prevention and treatment of haemolytic uraemic syndrome in shiga toxin producing Escherichia coli infection. Curr Opin Infect Dis 2004; 17 (5): 479–82.