Health Technology Assessment (Mar 2012)

Systematic review and validation of prediction rules for identifying children with serious infections in emergency departments and urgent-access primary care

  • M Thompson,
  • A Van den Bruel,
  • J Verbakel,
  • M Lakhanpaul,
  • T Haj-Hassan,
  • R Stevens,
  • H Moll,
  • F Buntinx,
  • M Berger,
  • B Aertgeerts,
  • R Oostenbrink,
  • D Mant

DOI
https://doi.org/10.3310/hta16150
Journal volume & issue
Vol. 16, no. 15

Abstract

Read online

Background: Although the vast majority of children with acute infections are managed at home, this is one of the most common problems encountered in children attending emergency departments (EDs) and primary care. Distinguishing children with serious infection from those with minor or self-limiting infection is difficult. This can result in misdiagnosis of children with serious infections, which results in a poorer health outcome, or a tendency to refer or admit children as a precaution; thus, inappropriately utilising secondary-care resources. Objectives: We systematically identified clinical features and laboratory tests which identify serious infection in children attending the ED and primary care. We also identified clinical prediction rules and validated those using existing data sets. Data sources: We searched MEDLINE, Medion, EMBASE, Cumulative Index to Nursing and Allied Health Literature and Database of Abstracts of Reviews of Effects in October 2008, with an update in June 2009, using search terms that included terms related to five components: serious infections, children, clinical history and examination, laboratory tests and ambulatory care settings. We also searched references of included studies, clinical content experts, and relevant National Institute for Health and Clinical Excellence guidelines to identify relevant studies. There were no language restrictions. Studies were eligible for inclusion if they were based in ambulatory settings in economically developed countries. Review methods: Literature searching, selection and data extraction were carried out by two reviewers. We assessed quality using the quality assessment of diagnostic accuracy studies (QUADAS) instrument, and used spectrum bias and validity of the reference standard as exclusion criteria. We calculated the positive likelihood ratio (LR+) and negative likelihood ratio (LR–) of each feature along with the pre- and post-test probabilities of the outcome. Meta-analysis was performed using the bivariate method when appropriate. We externally validated clinical prediction rules identified from the systematic review using existing data from children attending ED or primary care. Results: We identified 1939 articles, of which 35 were selected for inclusion in the review. There was only a single study from primary care; all others were performed in the ED. The quality of the included studies was modest. We also identified seven data sets (11,045 children) to use for external validation. The most useful clinical features for ruling in serious infection was parental or clinician overall concern that the illness was different from previous illnesses or that something was wrong. In low- or intermediate-prevalence settings, the presence of fever had some diagnostic value. Additional red flag features included cyanosis, poor peripheral circulation, rapid breathing, crackles on auscultation, diminished breath sounds, meningeal irritation, petechial rash, decreased consciousness and seizures. Procalcitonin (LR+ 1.75–2.96, LR– 0.08–0.35) and C-reactive protein (LR+ 2.53–3.79, LR– 0.25–0.61) were superior to white cell counts. The best performing clinical prediction rule was a five-stage decision tree rule, consisting of the physician’s gut feeling, dyspnoea, temperature ≥ 40 °C, diarrhoea and age. It was able to decrease the likelihood of serious infections substantially, but on validation it provided good ruling out value only in low-to-intermediate-prevalence settings (LR– 0.11–0.28). We also identified and validated the Yale Observation Scale and prediction rules for pneumonia, meningitis and gastroenteritis. Limitations: Only a single study was identified from primary-care settings, therefore results may lack generalisability. Conclusions: Several clinical features are useful to increase or decrease the probability that a child has a serious infection. None is sufficient on its own to substantially raise or lower the risk of serious infection. Some are highly specific (‘red flags’), so when present should prompt a more thorough or repeated assessment. C-reactive protein and procalcitonin demonstrate similar diagnostic characteristics and are both superior to white cell counts. However, even in children with a serious infection, red flags will occur infrequently, and their absence does not lower the risk. The diagnostic gap is currently filled by using clinical ‘gut feeling’ and diagnostic safety-netting, which are still not well defined. Although two prediction rules for serious infection and one for meningitis provided some diagnostic value, we do not recommend widespread implementation at this time. Future research is needed to identify predictors of serious infection in children in primary-care settings, to validate prediction rules more widely, and determine the added fvalue of blood tests in primar-care settings. Funding: The National Institute for Health Research Health Technology Assessment programme.

Keywords