Impact of chest radiography for children with lower respiratory tract infection: a propensity score approach.

Emmanuelle Ecochard-Dugelay,Vincent Gajdos,Jocelyne De Laveaucoupet,Francis Perreaux,Muriel Beliah,Philippe Labrune,Hubert Ducou-Lepointe,Ralph Epaud,Caroline Boisson,Jean Bouyer

doi:10.1371/journal.pone.0096189

Abstract

BackgroundManagement of acute respiratory tract infection varies substantially despite this being a condition frequently encountered in pediatric emergency departments. Previous studies have suggested that the use of antibiotics was higher when chest radiography was performed. However none of these analyses had considered the inherent indication bias of observational studies.ObjectiveThe aim of this work was to assess the relationship between performing chest radiography and prescribing antibiotics using a propensity score analysis to address the indication bias due to non-random radiography assignment.MethodsWe conducted a prospective study of 697 children younger than 2 years of age who presented during the winter months of 2006–2007 for suspicion of respiratory tract infection at the Pediatric Emergency Department of an urban general hospital in France (Paris suburb). We first determined the individual propensity score (probability of having a chest radiography according to baseline characteristics). Then we assessed the relation between radiography and antibiotic prescription using two methods: adjustment and matching on the propensity score.ResultsWe found that performing a chest radiography lead to more frequent antibiotic prescription that may be expressed as OR = 2.3, CI [1.3–4.1], or as an increased use of antibiotics of 18.6% [0.08–0.29] in the group undergoing chest radiography.ConclusionChest radiography has a significant impact on the management of infants admitted for suspicion of respiratory tract infection in a pediatric emergency department and may lead to unnecessary administration of antibiotics.

Highlights

Lower respiratory tract infection (LRTI) is one of the most common conditions encountered in pediatric emergency departments (PED) during the winter months, and its management consumes substantial health-care resources.[1]
We found that performing a chest radiography lead to more frequent antibiotic prescription that may be expressed as OR = 2.3, confidence interval (CI) [1.3–4.1], or as an increased use of antibiotics of 18.6% [0.08–0.29] in the group undergoing chest radiography
Chest radiography has a significant impact on the management of infants admitted for suspicion of respiratory tract infection in a pediatric emergency department and may lead to unnecessary administration of antibiotics

Summary

Introduction

Lower respiratory tract infection (LRTI) is one of the most common conditions encountered in pediatric emergency departments (PED) during the winter months, and its management consumes substantial health-care resources.[1]. Radiographic findings are often nonspecific, observational studies have reported a higher prevalence of prescription of antibiotics among children in whom CR has been performed.[6,7] the effects of CR may have been overestimated in these reports owing to a likely indication bias, as these children had not been randomly selected for radiography To avoid such a bias, Rosenbaum and Rubin suggested using the propensity score (PS), defined as the probability of receiving a particular treatment or procedure conditional on the observed covariates: two subjects from the two groups compared (CR or no CR) with the same propensity score can be considered as if they had been randomly assigned to the two groups.[8]. None of these analyses had considered the inherent indication bias of observational studies

Methods

Results

Discussion

Conclusion