You are here

Article Text

Article menu

Download PDFPDF

Original articles
Can urgency classification of the Manchester triage system predict serious bacterial infections in febrile children?
  1. Ruud G Nijman1,
  2. Rob LJ Zwinkels1,
  3. Mirjam van Veen1,
  4. Ewout W Steyerberg2,
  5. Johan van der Lei3,
  6. Henriëtte A Moll1,
  7. Rianne Oostenbrink1
  1. 1Department of General Paediatrics, Erasmus MC – Sophia Children's Hospital, Rotterdam, The Netherlands
  2. 2Department of Public Health, Centre for Medical Decision Making, Erasmus MC, Rotterdam, The Netherlands
  3. 3Department of Medical Informatics, Erasmus MC, Rotterdam, The Netherlands
  1. Correspondence to Rianne Oostenbrink, Department of General Paediatrics, Erasmus MC – Sophia Children's Hospital, Dr Molewaterplein 60, 3015 GJ Rotterdam, The Netherlands; r.oostenbrink{at}erasmusmc.nl

Abstract

Objective To evaluate the discriminative ability of the Manchester triage system (MTS) to identify serious bacterial infections (SBIs) in children with fever in the emergency department (ED) and to study the association between predictors of SBI and discriminators of MTS urgency of care.

Methods This prospective observational study included 1255 children with fever (1 month–16 years) attending the ED of the Erasmus MC – Sophia Children's Hospital, Rotterdam, The Netherlands in 2008–9. Triage urgency was determined with the MTS (urgency (U) level 1–5). The relationship between triage urgency and SBI was assessed with multivariable logistic regression, including effects of age, sex and temperature. Discriminative ability was assessed by receiver operating characteristic curve analysis.

Results SBI prevalence was 11% (n=131, 95% CI 9% to 12%). The discriminative value of the MTS for predicting SBI was 0.57 (95% CI 0.52 to 0.62), and the MTS did not contribute to a model including age, sex and temperature. The sensitivity of the MTS (U1–2 vs U3–5) to detect SBI was 0.42 (95% CI 0.33 to 0.51) and specificity was 0.69 (95% CI 0.66 to 0.72). MTS high urgency discriminators include several known predictors of SBI, such as fever, work of breathing, meningism and oxygen saturation, but apply to non-SBI children as well.

Conclusion The MTS has poor discriminative ability to predict the presence of SBIs in children presenting with fever to the paediatric ED. Important predictors of SBI are represented within the MTS, but are used in a different way to classify urgency.

https://doi.org/10.1136/adc.2010.207845

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Children with fever are at risk for serious bacterial infections (SBIs) and present with a wide range of clinical signs and symptoms.1,–,3 Early identification of SBIs, especially invasive bacterial infections such as meningitis and septicaemia, allows for immediate medical care, thereby improving patient outcome.4 Prediction rules have been developed to identify SBIs and several studies have identified predictors of SBI that are readily recognised at first assessment.5,–,11 As patient triage is used to assess urgency of care at first presentation, it might be a useful tool for identifying SBIs early in the diagnostic pathway.

    Several triage systems have been developed in recent years.12 13 The Royal College of Nursing Accident and Emergency Association and the British Association for Accident and Emergency Medicine developed the Manchester triage system (MTS) in 1997.14 15 With the MTS, patients are classified into one of five urgency categories, determined by a positive discriminator in a flowchart which best describes the patient's presenting problem.14 Study results have proven the reliability, reproducibility and validity of the MTS and have led to its widespread implementation in emergency departments (EDs).12 16 17

    What is already known on this topic

    • ▶ Triage systems are useful for assessing the ill child's appearance and defining the urgency of care at initial presentation.

    • ▶ Prediction rules for serious bacterial infections (SBIs) have identified predictors that are easily recognised at first assessment.

    What this study adds

    • ▶ Although important predictors of SBIs are represented within the Manchester triage system, a different role is assigned to them in the urgency classification.

    • ▶ At first assessment of a febrile child in the emergency department, a triage system and clinical prediction rules should be used in parallel to guide clinical decisions.

    This study aims to evaluate the discriminative ability of the MTS to identify SBIs in children with fever in the ED and to study the association between predictors of SBI and discriminators of MTS urgency of care.

    Methods

    Study population

    This study was conducted as part of a prospective observational study to validate the MTS. Patient characteristics, presenting signs and symptoms and the triage data of all patients visiting the ED of the Erasmus MC – Sophia Children's Hospital are registered routinely in an electronic patient record.18 For this particular study, all children with fever (ie, a temperature ≥38.5oC, a recent high fever or fever as a reason for referral), aged 1 month to 16 years, who attended the ED of the Erasmus MC – Sophia Children's Hospital, Rotterdam, The Netherlands, from January 2008 until July 2009, were eligible. The Erasmus MC – Sophia Children's Hospital is an inner city university hospital and its ED is visited by approximately 9000 patients annually, 90% of whom receive basic paediatric emergency care, with the remaining 10% receiving specialised tertiary centre care.19 Patients with chronic comorbidity, who have an increased risk of acquiring SBIs or developing severe complications and who visited a (subspecialist) paediatrician at least twice in the preceding year, were excluded from the study. Children who visited the ED for the same reason and with the same symptoms within 5 days of their first presentation were only once considered in the analysis; final diagnoses were based on available data from all consecutive visits. The institutional medical ethics committee approved the study; the requirement for informed consent was waived.

    Manchester triage system

    The MTS is used for patient triage in the ED of the Erasmus MC – Sophia Children's Hospital.18 The MTS is a five-level triage system used to allocate clinical priority to adult and children, as assessed by emergency care nurses.14 15 It was introduced in the United Kingdom in 1996 and its translated versions have been adapted around the world. The MTS uses 52 flowcharts which represent the presenting complaints such as ‘worried parent’, ‘limping child’ and ‘shortness of breath in children’. Each flowchart contains general as well as problem-specific signs and symptoms (discriminators) that differentiate between the different urgency categories. The discriminators for urgency level mainly depend on six general discriminators: life threat, pain, haemorrhage, level of consciousness, temperature and acuteness of onset. Selection of a discriminator allocates the patient to one of five urgency levels, each indicating the maximum time a patient should wait before being seen by a physician and the order in which the physician should evaluate the patient.14 15 Urgency level ‘immediate’ (red) demands immediate medical evaluation, ‘very urgent’ (orange) requires evaluation within 10 min, ‘urgent’ (yellow) within 60 min, ‘standard’ (green) within 120 min and ‘non-urgent’ (blue) can wait for up to 240 min before clinical assessment. Trained paediatric emergency care nurses triaged the patients with the official Dutch translation of the first edition of the MTS.14 During the study period we used a modified version of the first edition of the MTS, with several adjustments for the triage of febrile children.20 The MTS has been validated in children recently and showed good inter-rater agreement.18 21

    Outcome measures

    Final diagnoses were classified as either SBI or non-SBI. Within SBI, we distinguished pneumonia, meningitis, septicaemia, urinary tract infections and other less frequently occurring diagnoses such as erysipelas, cellulitis, bacterial gastro-enteritis, cellulitis orbitae, bacterial upper airway infection, ethmoiditis, arthritis and osteomyelitis. Final diagnoses were determined by positive bacteriological cultures of blood, urine, stool and ear, nose or throat, or radiological findings according to a reference standard.10 The reference standard included a follow-up period for all discharged patients to rule out the possibility of missed SBI and to avoid verification bias. Follow-up consisted of checking for consecutive ED visits and hospital admissions in a 1-week period after the first visit. If the final diagnosis was inconclusive, a consensus diagnosis was reached by the investigators (RN, RO, HM).

    Data analysis

    In univariate analysis we compared patient characteristics, referral pattern, height of fever and MTS discriminators for the presence of SBI using the Kruskall–Wallis test (continuous non-parametric variables) and Pearson's χ2 test (dichotomous and categorical variables; Fisher's exact test was used when cells contained less than five cases). Diagnostic performance measures of the MTS to predict SBI included sensitivity, specificity, positive predicted value (PPV) and negative predictive value (NPV), and positive and negative likelihood ratios (LR+, LR−). We dichotomised MTS categories into urgent (U1–2) and non-urgent (U3–5). This cut-off point reflects children needing ‘immediate’ or ‘very urgent’ care within 10 min after presentation (U1–2) and children in less urgent need of medical care who can wait for 60 min or more (U3–5) according to MTS triage.

    We assessed the discriminative ability of the MTS (categorical variable, U1–U5) for the presence of SBI with multivariate logistic regression analysis and receiver operating characteristic (ROC) curve analysis. A ROC area can range from 0.50 (non-informative test) to 1.0 (optimal test).22 As age and temperature are important predictors of SBI and are integrated into the MTS urgency classification,2,–,4 11 14 18 23,–,26 we added these variables to the model and examined the incremental value of discriminative ability. Next, we tested the interaction terms for statistically significant model improvement (p<0.05). Data were analysed using SPSS v 15.0. The diagnostic performance measures (sensitivity, specificity, PPV, NPV, LR+, LR−) of the MTS were calculated using the VassarStats website.27 We tested the linearity of age and temperature with restricted cubic splines (RCS) with R statistical software package v 2.8.1, using the Hmisc and Design library (http://www.r-project.org).28 Both temperature and age showed linearity on the log odds scale and were included as such in the analysis.

    Results

    A total of 1911 children with fever were eligible for inclusion. We excluded consecutive visits of children within 5 days of the first presentation with the same reason for consultation (n=121), children with missing data (n=1) and children with chronic co-morbidity (n=534), leaving 1255 children for inclusion. These children had a median age of 1.8 years (IQR 0.9–3.9) and 743 (59%) were boys. The prevalence of SBI was 11% (95% CI 9% to 12%). Table 1 shows the patient characteristics of children with and without SBI in our study population. Children with SBI are older, more often referred by a general practitioner, more frequently triaged in urgency categories 1–2, and have a higher body temperature at presentation. The most commonly used MTS flowcharts are listed in table 1.

    View this table:
    Table 1

    Characteristics of children with fever diagnosed with SBI (n=131) and without SBI (n=1124)†‡

    Table 2 shows MTS urgency distribution for different SBIs. Four cases were diagnosed with sepsis/meningitis, three of which were classified as ‘very urgent’ and one of which was classified as ‘urgent’. Other SBIs, such as pneumonia and urinary tract infections, were distributed among the urgency classification more heterogeneously, but none in the immediate/red or standard/blue category. In four cases triage data was missing.

    View this table:
    Table 2

    Distribution of different serious bacterial infections among MTS urgency classifications (n=131)*

    Table 3 describes the frequency of positive discriminators as documented in our patients with SBI. The positive discriminators ‘meningeal signs’, ‘increased work of breathing’ and ‘very low SaO2’ classify patients to high urgency (U1–2), while ‘chest infection’, ‘recent problem’ (ie, defined as a problem arising in the last 7 days), ‘vomiting’ and ‘child with fever’ (aged >3 months) indicate lower urgency in children with SBI. ‘Fever’ and ‘pain’, however, occurred among all MTS urgency classes. Table 4 shows the distribution of positive discriminators indicating ‘immediate’ or ‘very urgent’ as observed among children with and without SBI. Discriminators of ‘immediate’ urgency were only documented in children without SBI. The most commonly used positive discriminators to allocate high urgency (U1–2) were ‘child with fever’ and ‘increased work of breathing’. The discriminators ‘increased work of breathing’ and ‘very low PaSO2’ classified children with SBI as ‘very urgent’ significantly more often compared to children without SBI. Two high urgency discriminators were observed in children with SBI only (both n=1): ‘purpura’ and ‘facial oedema’. In our population most discriminators are only used sporadically to define the urgency of the presenting problem.

    View this table:
    Table 3

    Positive discriminators for children with serious bacterial infection (n=131)

    View this table:
    Table 4

    Positive discriminators for children with MTS high urgency (U1–2) (n=390)

    Table 5 describes the diagnostic performance of a dichotomised MTS (U1–2 vs U3–5). In our population with an SBI prevalence of 11%, a positive likelihood ratio of 1.34 increased the risk of SBI to 16% in children triaged with high urgency levels (U1–2), while a negative likelihood ratio of 0.85 decreased the risk of SBI to 10%. The discriminative value of the MTS (ROC area) to identify SBI was 0.57 (95% CI 0.52 to 0.62). The MTS, however, hardly contributed to a model including age, sex and temperature (ROC area 0.62; 95% CI 0.57 to 0.67). Tests for interaction between MTS urgency and temperature and MTS urgency and age were not significant.

    View this table:
    Table 5

    Diagnostic performance measures of MTS urgency (U1–2 vs U3–5) to identify serious bacterial infections

    To better understand the limited diagnostic value of the MTS to discriminate children with SBI, we matched known predictors of SBI, as identified in a systematic review by Van den Bruel et al11, with discriminators of urgency of care as used in the MTS (table 6). Important predictors of SBI, such as cyanosis, petechial rash, meningeal irritation and unconsciousness, are represented in the MTS by the discriminators ‘(very) low PaSO2’ (U2–3), ‘non-blanching rash’ (U2), ‘signs of meningism’ (U2) and ‘altered level of consciousness’ (U2) or ‘unresponsive child’ (U1), respectively. We could not match positive discriminators for a number of predictors but they match with flowcharts that include all levels of urgency. Parental concern, for example, matches the flowchart ‘worried parent’, changed crying pattern matches ‘crying child’, child appears ill matches ‘unwell child’ and child is irritable matches ‘irritable child’. Although the predictor ‘age’ is not a specific feature of the MTS, it is incorporated in a modified MTS to differentiate urgency classification in febrile children in several commonly used flowcharts.18

    View this table:
    Table 6

    Identified predictors of serious bacterial infections and characteristics of the MTS (first edition)*11 14

    Discussion

    The MTS has poor discriminative ability to identify SBI in children with fever. Comparison of high urgency (U1–2) with lower urgency (U3–5) categories shows poor sensitivity and moderate specificity and only a marginal increase in predicted risk with a low positive likelihood ratio. MTS urgency, adjusted for age, sex and body temperature, has no value in discriminating children with SBI from those without SBI. Some important predictors of SBI11 are included in the MTS by urgency discriminators, as shown in table 6. Other important predictors of SBI, however, are only represented with flowcharts that do not indicate a specific urgency level.

    Triage systems may be a promising tool to identify children with SBIs, as they asses the child's ill appearance at initial presentation in the ED. However, based on our observations, we conclude that the MTS cannot be used to identify SBIs in children with fever. This has not been proven before. The limited diagnostic value of the MTS for SBI in general can be explained by the fact that triage systems predict urgency of disease, which differs substantially from predicting severity of disease or assessment of diagnosis.29 Both SBIs and viral and self-limiting diseases may disturb vital functions requiring immediate intervention and are therefore classified as highly urgent (eg, bronchiolitis with respiratory insufficiency, febrile convulsion, shock from viraemia or dehydration complicating viral gastroenteritis).30,–,34 Some SBIs, in contrast, may have a more benign clinical course and do not require immediate medical care.35 Next, signs of bacterial sepsis and meningitis often occur late in the course of disease35 36 and may not be present at first evaluation, as illustrated by one case of septicaemia not being classified as high urgency. The poor discriminative ability of the MTS illustrates the need for parallel use of clinical prediction rules for SBI at first evaluation.

    Presenting signs and symptoms are diverse in febrile children: positive discriminators as documented in children with SBI are distributed equally among the three intermediate urgency classifications (U2–4), as shown in table 3. The presence of some frequently encountered discriminators for high urgency (such as (high) fever in young children and increased work of breathing) are rather aspecific and do not discriminate between children with and without SBI (table 4). Therefore, a different role is assigned to predictors of SBI as regards urgency classification compared to severity of disease classification. Our results do show, however, that most children with invasive SBIs such as meningitis or pneumonia are classified to high urgency by the discriminators associated with ‘meningeal signs’ and ‘severe dyspnoea’.

    Only one previous study by Thompson et al examined the predictive ability of triage urgency to identify SBI in febrile children by dichotomising the MTS into an urgent (U1–3) and a non-urgent group (U4–5). The authors observed a sensitivity of 84% and a specificity of 38%.37 Using similar cut-offs, we observed a sensitivity of 81% (95% CI 0.73% to 0.87%) with a low specificity of 25% (95% CI 0.22% to 0.28%), corresponding with the findings of Thompson et al. However, they did not examine different MTS cut-off points and did not adjust for other relevant predictors.

    In our and previous studies, temperature and age are important predictors of SBI in children with fever.2,–,4 11 23,–,26 In an earlier study we observed that performance of the MTS improved with age-specific adjustments, which led to the implementation of modifications for children with fever.18 These modifications proved to be safe with an increase in specificity and unchanged sensitivity20.

    The MTS was recently validated for children and was shown to have good inter-rater agreement, which enhances the validity of our results.18 21 In addition, compliance with the MTS at our ED is high, with data missing for only four children with SBI (3%) and 34 children in our total study population (3%). Our study population included both self-referrals and patients referred by general practitioner (n=341, 27%), with a higher prevalence of SBIs in the latter. The reported association between referral status and the severity of illness and triage level38 reflects a difference in population selection (self-referred patients present with other clinical characteristics as they are in a different phase of illness) rather than a different approach during triage. We did not observe effect modification between the MTS and referral. Referral status is not included as a determinant in the MTS.14 A limitation of our study is that the data are obtained from a single centre. Also, we observe a number of discriminators that are used to determine urgency of care only sporadically: more data are needed to assess the diagnostic value of these specific discriminators to predict SBI in febrile children. Next, results may need confirmation in a population with more SBIs. Although invasive bacterial infections are only a very small proportion of all SBIs, they are of great clinical importance, and a high predictive value for these SBIs would be useful. Finally, children with comorbidity, who may be at higher risk of SBIs or their complicated course, are excluded from this study. Comorbidity is included as a discriminator in the MTS so confirming our results in a population with comorbidity may be worthwhile.

    Conclusion

    The MTS has poor discriminative ability for predicting the presence of SBIs in children presenting with fever at the paediatric ED. Adjusting for age and temperature does not improve the diagnostic performance of the MTS to identify children with SBI. Although important predictors of SBI are represented within the MTS, a different role is assigned to them in the urgency classification. In the first assessment of a febrile child at the ED, a triage system and clinical prediction rules should be used in parallel to guide clinical decisions.

    References

    1. ↵
      1. Baker MD,
      2. Bell LM,
      3. Avner JR
      . Outpatient management without antibiotics of fever in selected infants. N Engl J Med 1993;329:1437–41.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Baraff LJ,
      2. Lee SI
      . Fever without source: management of children 3 to 36 months of age. Pediatr Infect Dis J 1992;11:146–51.
      OpenUrlPubMedWeb of Science
    3. ↵
      1. Baraff LJ,
      2. Oslund SA,
      3. Schriger DL,
      4. et al
      . Probability of bacterial infections in febrile infants less than three months of age: a meta-analysis. Pediatr Infect Dis J 1992;11:257–64.
      OpenUrlPubMedWeb of Science
    4. ↵
      1. Bachur RG,
      2. Harper MB
      . Predictive model for serious bacterial infections among infants younger than 3 months of age. Pediatrics 2001;108:311–16.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Gorelick MH,
      2. Shaw KN
      . Clinical decision rule to identify febrile young girls at risk for urinary tract infection. Arch Pediatr Adolesc Med 2000;154:386–90.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Margolis P,
      2. Gadomski A
      . The rational clinical examination. Does this infant have pneumonia? JAMA 1998;279:308–13.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. McIsaac WJ,
      2. Kellner JD,
      3. Aufricht P,
      4. et al
      . Empirical validation of guidelines for the management of pharyngitis in children and adults. JAMA 2004;291:1587–95.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Nigrovic LE,
      2. Kuppermann N,
      3. Malley R
      . Development and validation of a multivariable predictive model to distinguish bacterial from aseptic meningitis in children in the post-Haemophilus influenzae era. Pediatrics 2002;110:712–9.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Oostenbrink R,
      2. Moons KG,
      3. Moons CG,
      4. et al
      . A diagnostic decision rule for management of children with meningeal signs. Eur J Epidemiol 2004;19:109–16.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Bleeker SE,
      2. Moons KG,
      3. Derksen-Lubsen G,
      4. et al
      . Predicting serious bacterial infection in young children with fever without apparent source. Acta Paediatr 2001;90:1226–32.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Van den Bruel A,
      2. Haj-Hassan T,
      3. Thompson M,
      4. et al.
      ; European Research Network on Recognising Serious Infection investigators. Diagnostic value of clinical features at presentation to identify serious infection in children in developed countries: a systematic review. Lancet 2010;375:834–45.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. van Veen M,
      2. Moll HA
      . Reliability and validity of triage systems in paediatric emergency care. Scand J Trauma Resusc Emerg Med 2009;17:38.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Hostetler MA,
      2. Mace S,
      3. Brown K,
      4. et al
      . Emergency department overcrowding and children. Pediatr Emerg Care 2007;23:507–15.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Mackway-Jones K
      . Emergency Triage - Manchester Triage Group. London, UK: BMJ Publishing Group 1997. Available at http://books.google.com/books, by using search term: ‘Manchester triage system’
    15. ↵
      1. Mackway-Jones K,
      2. Marsden J,
      3. Windle J
      ,eds. Emergency Triage. Second edition. London, UK: Blackwell Publishing Ltd 2006.
    16. ↵
      1. Maldonado T,
      2. Avner JR
      . Triage of the pediatric patient in the emergency department: are we all in agreement? Pediatrics 2004;114:356–60.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. van der Wulp I,
      2. van Baar ME,
      3. Schrijvers AJ
      . Reliability and validity of the Manchester Triage System in a general emergency department patient population in the Netherlands: results of a simulation study. Emerg Med J 2008;25:431–4.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. van Veen M,
      2. Steyerberg EW,
      3. Ruige M,
      4. et al
      . Manchester triage system in paediatric emergency care: prospective observational study. BMJ 2008;337:a1501.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Bouwhuis CB,
      2. Kromhout MM,
      3. Twijnstra MJ,
      4. et al
      . [Few ethnic differences in acute pediatric problems: 10 years of acute care in the Sophia Children's Hospital in Rotterdam]. Ned Tijdschr Geneeskd 2001;145:1847–51.
      OpenUrlPubMed
    20. ↵
      1. Van Veen M,
      2. Steyenberg EW,
      3. Lettinga L,
      4. et al
      . Safety of the Manchester Triage System to identify less urgent patients in paediatric emergence care: a prospective observational study. Arch Dis Child 2011; 8 March 2011 doi:10.1136/adc.2010.199018.
    21. ↵
      1. van Veen M,
      2. Teunen-van der Walle VF,
      3. Steyerberg EW,
      4. et al
      . Repeatability of the Manchester Triage System for children. Emerg Med J 2010;27:512–6.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Harrell FE,
      2. et al
      . Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression and Survival Analysis. New York, NY: Springer 2001.
    23. ↵
      1. Bonadio WA
      . Evaluation and management of serious bacterial infections in the febrile young infant. Pediatr Infect Dis J 1990;9:905–12.
      OpenUrlPubMedWeb of Science
    24. ↵
      1. Craig JC,
      2. Williams GJ,
      3. Jones M,
      4. et al
      . The accuracy of clinical symptoms and signs for the diagnosis of serious bacterial infection in young febrile children: prospective cohort study of 15 781 febrile illnesses. BMJ 2010;340:c1594.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Stanley R,
      2. Pagon Z,
      3. Bachur R
      . Hyperpyrexia among infants younger than 3 months. Pediatr Emerg Care 2005;21:291–4.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Jaffe DM,
      2. Fleisher GR
      . Temperature and total white blood cell count as indicators of bacteremia. Pediatrics 1991;87:670–4.
      OpenUrlAbstract/FREE Full Text
    27. ↵
      VassarS*t*ats. Website for Statistical Computation. http://faculty.vassar.edu/lowry/VassarStats.html (accessed 4 April 2011).
    28. ↵
      1. Harrell FE Jr,.,
      2. Lee KL,
      3. Mark DB
      . Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361–87.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. FitzGerald G,
      2. Jelinek GA,
      3. Scott D,
      4. et al
      . Emergency department triage revisited. Emerg Med J 2010;27:86–92.
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Steiner MJ,
      2. DeWalt DA,
      3. Byerley JS
      . Is this child dehydrated? JAMA 2004;291:2746–54.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      NICE clinical guideline 47. Feverish Illness in Children – Assesment and initial management in children younger than 5 years. London, UK: NICE, 2007.
    32. ↵
      1. Verboon-Maciolek MA,
      2. Thijsen SF,
      3. Hemels MA,
      4. et al
      . Inflammatory mediators for the diagnosis and treatment of sepsis in early infancy. Pediatr Res 2006;59:457–61.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Van Steensel-Moll HA,
      2. Van der Voort E,
      3. Bos AP,
      4. et al
      . Respiratory syncytial virus infections in children admitted to the intensive care unit. Pediatrie 1989;44:583–8.
      OpenUrlPubMed
    34. ↵
      1. Black CP
      . Systematic review of the biology and medical management of respiratory syncytial virus infection. Respir Care 2003;48:209–31; discussion 231–3.
      OpenUrlPubMed
    35. ↵
      1. Almond SC,
      2. Summerton N
      . Diagnosis in General Practice. Test of time. BMJ 2009;338:b1878.
      OpenUrlFREE Full Text
    36. ↵
      1. Thompson MJ,
      2. Ninis N,
      3. Perera R,
      4. et al
      . Clinical recognition of meningococcal disease in children and adolescents. Lancet 2006;367:397–403.
      OpenUrlCrossRefPubMedWeb of Science
    37. ↵
      1. Thompson M,
      2. Coad N,
      3. Harnden A,
      4. et al
      . How well do vital signs identify children with serious infections in paediatric emergency care? Arch Dis Child 2009;94:888–93.
      OpenUrlAbstract/FREE Full Text
    38. ↵
      1. Rinderknecht AS,
      2. Ho M,
      3. Matykiewicz P,
      4. et al
      . Referral to the emergency department by a primary care provider predicts severity of illness. Pediatrics 2010;126:917–24.
      OpenUrlAbstract/FREE Full Text

    Footnotes

    • Funding RN is supported by a grant from ZonMW, the Dutch Organization for Health Research and Development, and Erasmus MC Doelmatigheid. RO is supported by an unrestricted grant from Europe Container Terminals B.V.

    • Competing interests None.

    • Ethics approval This study was conducted with the approval of the Ethics Committee, Erasmus MC, Rotterdam, The Netherlands.

    • Provenance and peer review Not commissioned; externally peer reviewed.