|Diagnostic value of clinical features at presentation to identify serious infection in children in developed countries: a systematic review
|Van den Bruel A, Haj-Hassan T, Thompson M, Buntinx F, Mant D
This generally well-conducted review concluded that cyanosis, rapid breathing, poor peripheral perfusion and petechial rash should be used routinely to identify serious infection. Although the conclusions reflect the evidence presented, each clinical feature was assessed in only a few, poor quality, studies. Therefore, the conclusions should be viewed with some caution.
To evaluate the diagnostic value of clinical features of serious infection in children presenting to ambulatory care settings in developed countries.
MEDLINE, EMBASE, CINAHL, DARE and Medion were searched without language restrictions to June 2009. Search terms were reported. Reference lists of retrieved articles and NICE guidelines published prior to 2008 were also searched, and experts in the field contacted.
Diagnostic accuracy and prognostic studies evaluating the accuracy of clinical signs of serious infection in children aged between one month and 18 years presenting to ambulatory care settings in Europe, the USA, Australia, New Zealand and Japan, were eligible for inclusion. Studies had to recruit an appropriate patient spectrum, use a valid reference standard, and report sufficient data to construct a 2x2 table of test performance. Studies were excluded if they were conducted in children with pre-existing immune suppression, or if they recruited 50% or more patients outside the target age group (unless results were reported separately for the children of interest).
The age of the included children ranged from newborn to 16 years; most studies included children under the age of six. The prevalence of serious infection ranged from 0.78 to 55.3%. Most studies were conducted in emergency departments.
Two independent reviewers selected studies; disagreements were resolved by a third reviewer.
Assessment of study quality
Study quality was assessed by one reviewer using the 14-criteria QUADAS (Quality Assessment of Diagnostic Accuracy Studies) tool; results were checked by a second reviewer; disagreements were resolved by discussion. Studies were rated A (all criteria met), B (studies not rated A, C or D), C (potential for progression bias, no independent reference standard, interpreters of reference standard not blinded to results of index test), or D (potential for verification bias, interpreters of index test not blinded to results of reference standard).
Data to construct a 2x2 table of test performance were extracted by one reviewer and checked by a second, from which positive and negative likelihood ratios (LR+/-) with 95% confidence intervals (CI) were calculated. Probability of illness, with 95% confidence interval, after a positive and negative test was determined. Disagreements were resolved by discussion. Where a cell contained zero, 0.5 was added to all cells. Where a study used a range of cut-off points, the cut-off with the highest positive likelihood ratio and lowest negative likelihood ratio was used.
Methods of synthesis
The studies were combined in a narrative synthesis. Differences between studies were discussed in the text and study details tabulated. Clinical features were considered of prognostic value for ruling infection when the positive likelihood ratio was more than 5.0 in at least one study, or ruling out infection when the negative likelihood ratio was less than 0.2 in at least one study. Prevalence of infection of up to 5% was considered low, 5 to 20% intermediate, and over 20% high. Some summary estimates of sensitivity and specificity calculated using a bivariate model were presented, but only from sensitivity analyses where studies thought to be causing heterogeneity were excluded.
Results of the review
Thirty studies were included (n=14,453 children; range 72 to 3,981); 27 studies were prospective, of which 24 were cross-sectional, and 20 recruited consecutive children. Of the 30 studies, one was rated A, two rated B, 18 rated C, and nine rated D.
The clinical features with the greatest prognostic value for detecting serious infection in low prevalence settings were parental concern that the illness was different from previous illnesses (LR+ 14.4, 95% CI 9.3 to 22.1; one study) and the clinicians instinct that there was something wrong (LR+ 23.5, 95% CI 16.8 to 32.7; one study).
The most valuable circulatory and respiratory prognostic factors for detecting serious infection were cyanosis (LR+ range 2.66 to 52.2; three studies), rapid breathing (LR+ range 1.26 to 9.78; three studies), shortness of breath (LR+ range 1.11 to 9.30; three studies), and markers of poor peripheral circulation (LR+ range 2.39 to 38.8; seven studies). Petechial rash also had high prognostic value (LR+ range 6.18 to 83.7; six studies)
The clinical features with the greatest prognostic value for excluding serious infection was a five-stage decision tree (LR- 0.04, 95% CI 0 to 0.26; one study), and the absence of shortness of breath combined with the absence of parent or clinician concern excluding pneumonia (LR- 0.07, 95% CI 0.01 to 0.46; one study).
Cyanosis, rapid breathing, poor peripheral perfusion and petechial rash should be used routinely to identify serious infection, but serious illness will still be missed without effective precautionary measures.
The authors addressed a clear research question supported by appropriate inclusion criteria. Several relevant sources were searched, but the search for unpublished studies was limited. Each stage of the review process was conducted in duplicate, reducing the risk of error and bias. Study quality was assessed using appropriate criteria.
The decision to make the main synthesis a narrative was appropriate, given the clinical heterogeneity between studies. Of the clinical features deemed to have prognostic value that were assessed in more than one study, only petechial rash had a likelihood ratio greater than that stated as being required for significant prognostic value, in all studies. Most clinical features were evaluated in one or few studies.
This was a generally well-conducted review, and the conclusions reflect the evidence presented, but the assessment for each clinical feature is based on only a few, poor quality, studies. Therefore the conclusions should be viewed with some caution.
Implications of the review for practice and research
Practice: The authors stated that: the clinical features identified by WHO for use in developing countries can be used in the initial assessment of children in developed countries; there should be more emphasis placed on parental concern during the diagnostic process; infection can only be ruled out if several clinical features are considered together; and there is a need for safeguards to ensure that children with serious illness who do not have red flag symptoms at presentation are not missed.
Research: The authors stated that the level of risk at which clinical action should be taken needs to be identified.
Health Technology Assessment Project 07/37/05; National Institute for Health Research National School for Primary Care Research.
Van den Bruel A, Haj-Hassan T, Thompson M, Buntinx F, Mant D. Diagnostic value of clinical features at presentation to identify serious infection in children in developed countries: a systematic review. Lancet 2010; 375(9717): 834-845
Subject indexing assigned by NLM
Acute Disease; Ambulatory Care; Bacteremia /diagnosis; Child; Dehydration /diagnosis /etiology; Developed Countries; Gastroenteritis /diagnosis; Humans; Infection /diagnosis; Meningitis /diagnosis; Meningococcal Infections /diagnosis; Pneumonia /diagnosis; Primary Health Care
Date abstract record published
This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.