medwireNews: Researchers have developed a clinical prediction model to assess the likelihood of different serious bacterial infections (SBIs) in children presenting at the emergency department with fever.
Rianne Oostenbrink (Erasmus MC-Sophia Children's Hospital, Rotterdam, the Netherlands) and colleagues found that factors such as an oxygen saturation of less than 94% and an increased breathing rate were important predictors for pneumonia. Raised C-reactive protein levels were predictive both for pneumonia and other SBIs like septicemia/meningitis and urinary tract infections. The presence of chest wall retractions and low oxygen saturation, on the other hand, was useful to rule out SBIs other than pneumonia.
Writing in the BMJ, the team points out that "to differentiate children who have a benign self limiting viral infection from the small proportion with serious bacterial infections, many prediction models have been proposed. Most of these prediction models have not, however, been validated."
Included in the study were 1750 children aged between 1 month and 15 years presenting to a Dutch hospital with fever between 2003 and 2005 and a second population of 967 children presenting to a second Dutch hospital in 2007. Also included were 487 children presenting to a UK hospital in 2005 and 2006 who were used as the validation cohort.
The researchers used the C statistic to estimate the ability of the model to discriminate between patients with different diagnoses. For predicting pneumonia in the external validation cohort, the C statistic was 0.81; however, the discriminative ability of the model to predict other SBIs was substantially lower, at 0.69.
When levels of C-reactive protein were excluded from the model, it performed significantly less well than when they were included.
Oostenbrink and colleagues also found that a low-risk threshold of less than 2.5% was useful to rule out, and a high-risk threshold of 10.0% or more was useful to rule in, the presence of pneumonia and other SBIs.
They conclude: "If further validated, this model may well support doctors in their diagnostic work-up and therapeutic decision making."
medwireNews (www.medwirenews.com) is an independent clinical news service provided by Springer Healthcare Limited. © Springer Healthcare Ltd; 2013
Free full text