Article Text

This article has a correction. Please see:

Towards evidence based medicine for paediatricians
    1. Evidence-based On Call, Cairns Library, John Radcliffe Hospital, Oxford OX3 9DU, UK
    1. Bob Phillipsbob.phillips{at}

    Statistics from

    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.

    In order to give the best care to patients and families, paediatricians need to integrate the highest quality scientific evidence with clinical expertise and the opinions of the family.1 Archimedes seeks to assist practising clinicians by providing “evidence based” answers to common questions which are not at the forefront of research but are at the core of practice.

    A word of warning. The topic summaries are not systematic reviews, though they are as exhaustive as a practising clinician can produce. They make no attempt to statistically aggregate the data, nor search the grey, unpublished literature. WhatArchimedes offers is practical, best evidence based answers to practical, clinical questions.

    The format of Archimedes may be familiar. A description of the clinical setting is followed by a structured clinical question. (These aid in focusing the mind, assisting searching,2 and gaining answers.3) A brief report of the search used follows—this has been performed in a hierarchical way, to search for the best quality evidence to answer the question.4 A table provides a summary of the evidence and key points of the critical appraisal. For further information on critical appraisal, and the measures of effect (such as number needed to treat, NNT), books by Sackett et al and Moyer et al may help.5 ,6 To pull the information together, a commentary is provided. But to make it all much more accessible, the clinical bottom line is highlighted.

    Readers wishing to submit their own questions—with best evidence answers—are encouraged to read the Instructions for Authors at

    Critical appraisal note: evidence of equivalence versus no evidence of difference

    When a randomised study compares two therapies and finds no difference in an important outcome between the two, does this provide evidence of equivalence or merely an absence of evidence of effectiveness? The practical answer requires integration of the study with clinical expertise.  If a dichotomous outcome is present (dead versus not, hospitalised versus sent home) then a variety of measures can be generated. Of these, those which give information of the risk reduction (relative, absolute, or its inverse; the number needed to treat) are useful. With continuous outcome measures, estimates of difference can also be produced, but may be more difficult to interpret. A confidence interval can be produced for each of these measures, within whose limits the true effect is likely to fall. This degree of uncertainty is the first element to be considered in the question. As a rule of thumb, if a study cannot exclude a 20% difference between the treatments, it is probably not evidence of equivalence.  The second element requires knowledge of the disease and outcome. If the outcome is fatal, disfiguring or had serious morbidity, perhaps a greater degree of certainty is required. There is also the rest of the clinical literature on the subject—does this result agree with the general tenor of evidence or does it stick out?  Combining these aspects—precision, importance, and congruence—allows an answer to the question of equivalence versus lack of difference.


    Are routine chest x rays helpful in the management of febrile neutropenia?


    A friendly, coryzal 5 year old girl with acute lymphocytic leukemia attends with another episode of febrile neutropenia. According to departmental protocol, her admission includes a chestx ray. You wonder as to the value of this routine irradiation.

    Structured clinical question

    In a 5 year old girl with febrile neutropenia [patient] does routine chest radiography [intervention] assist in management decisions or diagnose occult pneumonia [outcome]?


    Secondary sources—nil. SumSearch—“neutropenia” AND “radiography” AND filter “diagnosis”. Search results—67 individual articles found, three relevant.


    See table 1-1.

    Table 1-1


    There is no good quality study addressing the use of chest radiographs in uncomplicated febrile neutropenia. Two of these studies are consistent with clinical feeling—lack of abnormal signs or symptoms in children with febrile neutropenia rules out pneumonia. The meth-odological weaknesses would tend to favour this—with one study having clinical features as part of the reference standard, and the second tending to fail to perform chest radiography on children without symptoms. The third study only gives data on respiratory signs (ignoring symptoms) and has a subsequently reduced sensitivity and improved specificity.

    Clinical bottom line

    • Pneumonia was uncommon in children with febrile neutropenia (∼3%)

    • An absence of respiratory signs and symptoms made pneumonia very unlikely

    • Routine chest x rays seem unnecessary.


    Clare Collins (Research Fellow, Oxford Vaccine Group, Oxford) [clare.collins{at}] Matthew Fenton (Registrar in Paediatric Oncology, John Radcliffe Hospital, Oxford) Bob Phillips (Junior Fellow, Centre for Evidence-based Medicine)


    1. 1-1.
    2. 1-2.
    3. 1-3.

    Does dexamethasone improve blood pressure in hypotensive ill neonates?


    A 25 week gestation baby, birth weight 695 g is ventilated for respiratory distress syndrome. Invasive blood pressure monitoring at 2 hours of age showed a mean of 23–25 mm Hg. The blood pressure did not improve over the next 24 hours, in spite of three intravenous boluses of 0.9% saline and concurrent infusions of dopamine and dobutamine at 15 μg/kg/min.

    A colleague suggests that dexamethasone might help to improve the baby's blood pressure.

    Structured clinical question

    In hypotensive preterm infants [patient] does treatment with dexamethasone [intervention] increase blood pressure [outcome]?


    Secondary sources—nil. Search strategy—“hypotension” AND “(dexamethasone OR steroid)” AND “newborn” AND “(clinical trial)”. Search results—three articles found, two relevant.


    See table 2-2.

    Table 2-2


    The study by Gassmaier and Weston was well constructed in terms of randomisation, blinding, and intention to treat analysis. Dexamethasone administration, after treatment with volume boluses, dopamine, and adrenaline infusion, improved BP such that adrenaline was discontinued in 63% babies (compared with 11% of placebo group).

    The paper by Bourchier and Weston supports the idea that bolus steroids are a useful adjunct to conventional treatments for hypotension in sick, ventilated preterm infants. Dopamine and hydrocortisone both appeared to be effective (p = 0.108) in the treatment of hypotension refractory to treatment with fluid bolus. However, if five babies received hydrocortisone, one additional baby remained hypotensive, compared with similar babies who received dopamine. Confidence intervals for this NNT are wide (3, 45), suggesting a larger study would show statistical difference at the 5% level.

    Dexamethasone appears to be a useful adjunct to the commonly used pathway for treating hypotension in neonates (fluid bolus ± dopamine ± dobutamine).

    Although no adverse events related to steroid use are reported in either paper, no long term follow up is reported and caution is warranted as there is emerging evidence of increased risk of cerebral palsy, following postnatal dexamethasone use in babies at risk for chronic lung disease (Shinwell et al,Arch Dis Child Fetal Neonatal Ed2000;83:F177–F181), without improvement in mortality (Halliday and Ehrenkranz. Early postnatal (<96 hours) corticosteroids for preventing chronic lung disease in preterm infants.Cochrane Database of Systematic Reviews, Issue 1, 2001).

    Clinical bottom line

    • Dexamethasone improved blood pressure in ill, ventilated neonates.


    Richard Nicholl (Consultant Neonatologist) [DrNicholl{at}] on behalf of The Northwick Park Neonatal Journal Club, Northwest London NHS Hospitals Trust


    1. 2-1.
    2. 2-2.

    Does nebulised adrenaline reduce admission rate in bronchiolitis?


    A 4 month old infant attends the emergency department in the late morning with bronchiolitis. It is the first episode of wheeze. Clinically, there is moderate indrawing and recession, tachypnoea (RR = 50), reasonable air movement on auscultation, and the oxygen saturation is 94% in air. You want to admit the infant, but the mother is breast feeding and keen to get home by 3 pm, when her other children get home from school. You have heard that in North America, nebulised adrenaline has been used in some cases and admission has been avoided.

    Structured clinical question

    In an infant with bronchiolitis [patient] does nebulised adrenaline (compared to other treatments) [intervention] reduce the need for admission [outcome]?


    Secondary sources—Cochrane Library (2001): “bronchiolitis”, two systematic reviews (one irrelevant—anticholinergics and wheeze);Clinical Evidence (Issue 4): “child health—bronchiolitis”, two systematic reviews (one irrelevant—adrenaline not included); DARE: “bronchiolitis”, five systematic reviews (three irrelevant; two relevant SRs were by same authors—one referenced in Cochrane and one referenced in journal).

    PubMed clinical queries: “bronchiolitis” AND “epinephrine” [therapy, sensitive]—eight references (three irrelevant to question).

    MedLine [1966 to Dec 2000] (Ovid): “bronchiolitis” OR “bronchitis” AND [“epinephrine (exp)” or “catecholamines”]; LIMIT to “clinical trial”—13 references (eight irrelevant to question).

    Five papers addressed the question of nebulised adrenaline and bronchiolitis (one of them specifically answering the question).


    See table 3-3.

    Table 3-3


    There is only one study (Menon et al) that specifically answers the question; this study shows a reduction in hospital admission, and the study group is similar to the patient in the clinical scenario.

    A systematic review that includes adrenaline as one of a number of bronchodilators fails to show significant differences in outcomes compared to placebo. However, adrenaline has an α adrenergic action which is thought to be important in bronchiolitis (as well as the β adrenergic bronchodilatory effects it has). The positive effect of adrenaline may therefore have been diluted in the systematic review by the inclusion of agents that have little or no effect.

    The Menon et al study compared adrenaline with salbutamol, which is not routinely used in the UK in this condition. For this reason, data on studies comparing adrenaline to placebo in bronchiolitis are also presented. Studies comparing the two show a benefit of adrenaline over placebo as well as benefit over pure β adrenergic agonists.

    It is thought that the α adrenergic properties of adrenaline are important in bronchiolitis, as the vasoconstriction of the pulmonary vessels reduces mucosal oedema and exudate, thereby reducing airway obstruction.

    The regime used was 3 ml of 1/1000 adrenaline nebulised at arrival and 30 minutes later. The infants were then observed for at least two hours.

    Currently, a multicentre trial in the UK comparing nebulised adrenaline with placebo is under discussion.

    Clinical bottom line

    • Nebulised adrenaline probably reduced hospital admission in bronchiolitis

    • Nebulised adrenaline appeared superior to salbutamol and placebo in relieving symptoms in bronchiolitis.


    Maud Meates (Consultant Paediatrician, North Middlesex Hospital, London) [mmeates{at}]


    1. 3-1.
    2. 3-2.
    3. 3-3.
    4. 3-4.
    5. 3-5.

    Supplementary materials

    • Archimedes Web Only Pages

      Archimedes has gathered the summaries of the evidence to answer this month's questions. Unlike the paper edition, these electronic pages give extra information. Each table links to critical appraisals of the studies, and also to web pages which may explain some unfamiliar terms.

      Click on a question to see the summary table, and click on each citation to see a more detailed critique.

      Are routine chest x rays helpful in the management of febrile neutropenia?

      Does dexamethasone improve blood pressure in hypotensive ill neonates?

      Does nebulised adrenaline reduce admission rate in bronchiolitis?

    • Archimedes - Towards Evidence-Based Paediatrics

      The principles of evidence-based healthcare have been widely accepted across a variety of health care settings, and a textbook of Evidence-based Pediatrics and Child Health is available to educate and inform paediatricians about its' practice. Their definition is "the integration of clinical information obtained from a patient with the best evidence available from clinical research and experience, and the application of this knowledge to the prevention, diagnosis or management of disease in that child" [1]. In order that this may be performed efficiently, summaries of the best evidence are helpful to guide practice.

      The format we are using is based on the Best BETS ( template. To register a question, and to submit completed Archimedes topics, please check the BestBETS site or email bob.phillips{at} The creation of such a topic summary follows this process:

        • Selection of a clinical scenario

        Scenarios which trigger searches for evidence-based medicine are best if they address common questions, which are important for the patients, doctor, health system or all three.

        • Definition of the clinical question

        Structured clinical questions improve one's efficiency when searching for answers, and narrow down the area of enquiry. What might start as "How do you treat croup?" becomes focussed into "In a child with croup, does nebulised budesonide work better or faster than oral dexamethasone?" The structured clinical question has three (or four) parts: the patients, the intervention/test, possibly a comparison, and the outcome(s) of interest.

        • Search for answers

        Searches are best if they look for the best quality evidence first, then work down the hierarchy of evidence ( In order to do this, searches are first performed on secondary sources (Cochrane Library, Best Evidence and Clinical Evidence), then repeated for either more recent information or any information on a Medline system. PubMed, the NLM's freely available Medline service, is the commonly used search engine. If a systematic review is found (and of good quality), then a search forwards in time, to see if anything has been published after the review, is necessary.

        The technique is to use a broad search on the secondary sources - for instance the patients' condition. If this produces an unreasonably large number hits, narrow with the intervention. If this is still an unreasonably large number, try using the comparator or an outcome measure until a reasonable number are listed, then scan the abstracts. The search words can be free text (if only a few hits are found) or MeSH terms (indexed words) if too many articles are discovered.

        The search on Medline follows a similar strategy, but with the use of Clinical Question Filters.

        Another person should repeat this search, and a consensus about the results reached.

        • Appraise the evidence

        The appraisal of evidence is the important step in the process. All studies require appraisal - from Cochrane reviews to case series from Orthopaedics Weekly. There are many texts assisting the critical evaluation of studies, the classic text is by Sackett et al [2], or the first chapters in the Evidence-based Pediatrics and Child Health textbook [1].

        • Create a critically appraised topic (CAT)

        The distillation of the critical appraisal into a CAT is important to allow others to evaluate the conclusions reached. A CAT is a detailed summary of the critical appraisal of the individual paper. In order to do this, CATMaker, a programme from the Oxford Centre for Evidence-based Medicine has been developed (, although any format is acceptable.

        • Summarise as a best evidence topic (BET)

          The best evidence topic is the final accumulation of the critical appraisal. The strict format allows the casual reader to extract important information quickly and easily. Some notes on the template are below:


        Introductory title

        A question to tease the reader into your article.


        Short description of the clinical episode.

        Structured clinical question

        Identifying the 3 or 4 parts of the question


        Details of the search strategy used, and the number of studies found. This is abbreviated greatly in the paper version.

        Study results

        As a table; this identifies the studies, the population group and size, the study design (and level of evidence), the key outcomes and their results. Any vitally important points of critical appraisal are also noted here.


        A short narrative, summary of the evidence and its application.

        Clinical bottom lines

        A summary of the important messages from the BET - 3 at the most.

        In general, the past tense is preferred for all answers based on studies of a lower quality than a good systematic review, and the use of terms such as 'may' or 'might' for information of very low quality.


        Name and designations, only please.


        The BETs within Archimedes are the best evidence summaries a real clinician can produce. They represent the best of the available evidence applied in real paediatric settings. They can be created within journal clubs, training schemes or by interested clinicians at any level of training. In time, this archive will demonstrate the strengths and weaknesses of the evidence upon which paediatric care is based.

        A template is available here

        To register a question, and to submit completed Archimedes topics, please check the BestBETS site or email bob.phillips{at}



        BestBETs are an original format developed by Kevin Mackway-Jones and Simon Carley at the Manchester Royal Infirmary.


        (1) Evidence-based Pediatrics and Child Health. Moyer V, Elliot EJ, Davis RL et al. BMJ Books: London 2000

        (2) Evidence-based Medicine: How to Practise and Teach EBM. Sackett DL, Straus SE, Richardson WS, Rosenberg W, Haynes RB Second Edition. Churchill Livingstone: Edinburgh, 2000.

      Linked Articles

      • Archives this month
        BMJ Publishing Group Ltd and Royal College of Paediatrics and Child Health
        BMJ Publishing Group Ltd and Royal College of Paediatrics and Child Health