Does Your Pediatric Patient Have Pneumonia?
Abstract & Commentary
Source: Rothrock SG, et al. Do published guidelines predict pneumonia in children presenting to an urban ED? Pediatr Emerg Care 2001;17:240-243.
It is difficult to predict the presence of pneumonia in sick children with respiratory symptoms in the emergency department (ED). While numerous criteria for signs and symptoms abound in the literature, controversy remains about when to obtain a chest x-ray (CXR) for definitive diagnosis. To assess the validity of recently published guidelines, Rothrock and colleagues, at an urban ED in Orlando, evaluated evidence-based criteria generated by a Canadian task force. These experts suggested in 1997 that the absence of a four-fold cluster of tachypnea, respiratory distress, crackles, and diminished breath sounds accurately excluded pneumonia.
Rothrock studied 329 children younger than age 5. All had CXRs for possible pneumonia after clinical examination by pediatric ED physicians, who completed a computerized questionnaire for each case. Patients were excluded in instances of trauma, foreign body ingestion, or submersion. Board certified radiologists interpreted all radiographs, defining pneumonia as "pneumonia or an infiltrate."
In these cases, 130 of 329 patients (40%) had fever, 104 (32%) had cough, and 73 (23%) had respiratory distress (grunting, gasping, or retractions). Tachypnea was categorized according to World Health Organization criteria: more than 60 breaths per minute (bpm) for neonates younger than 1 month old, more than 50 bpm for infants 2-12 months old, and more than 40 bpm for children 1-5 years old. Preliminary diagnoses, prior to viewing radiographs, included upper respiratory infection (48%), pneumonia (38%), asthma (6%), sepsis/meningitis (2%), and aspiration (1%). Upon radiographic evaluation, 67 of 329 children (20%) were diagnosed with pneumonia. The authors applied the Canadian guidelines to the Orlando ED children, calculating a sensitivity of 45% (95% CI, 33-58%), specificity of 66% (95% CI, 60-72%), positive predictive value of 25% (95% CI, 18-34%), and negative predictive value of 82% (95% CI, 77-87%).1 The single clinical finding most predictive of pneumonia was respiratory distress, with sensitivity and specificity of 25% and 18%, respectively.
The authors conclude that the published criteria from the 1997 Canadian task force are insufficiently sensitive for detection of pneumonia in infants and young children.
Commentary by Michael Felz, MD
The sensitivity of less than 50% and positive predictive value of only 25% do not support the widespread application of these criteria, based on this analysis of more than 300 children in Orlando. The authors are quick to point out prior studies of predictive symptoms and signs for pneumonia in which data are equally conflicting. They further cite the general lack of consensus on when to perform a CXR in ill children, despite studies touting certain symptoms singly or in combination. As an example, the best predictor of pneumonia in the current study was respiratory distress, although the sensitivity of 25% is hardly persuasive.
I reviewed the paper from the Canadian task force.1 These experts in pediatric infectious disease and microbiology held six consensus meetings and examined published evidence from developed countries only. They analyzed five studies of a total of 1100 children, 201 of whom had radiographic pneumonia. Sensitivity and specificity were calculated for four clinical findings: respiratory distress, tachypnea, crackles, and decreased breath sounds. Absence of all four excluded pneumonia with high specificity, but the authors stated "no finding in itself can be used to diagnose or rule out pneumonia." These guidelines, then, suggest when not to order a CXR, not when to do so.
So what are busy practitioners to conclude? It seems to me that a child presenting with fever and respiratory symptoms needs a CXR if tachypnea (age appropriate), retractions, crackles, or reduced breath sounds are documented. Conversely, the absence of this four-fold cluster of clinical manifestations almost would appear to eliminate the need for a chest film, based on the 82% negative predictive value derived from the Orlando study. My colleagues in our pediatric ED further reminded me that an oxygen saturation of less than 96% is an indication for a CXR in an ill child with respiratory symptoms.
Immunization against Hemophilus influenzae and Streptococcus pneumoniae has shifted the etiologic agents for pediatric pneumonia to favor viral pathogens in the majority of cases. I suspect that revised guidelines are forthcoming that will incorporate criteria more specific for infection by respiratory syncytial virus (RSV), parainfluenza, influenza, rhinovirus, and adenovirus. In the meantime, clinicians would be wise to heed age-specific definitions of tachypnea, and be vigilant for signs of respiratory distress. I will welcome more definitive decision rules for pediatric pneumonia diagnosis and indications for CXRs once upcoming prospective studies are completed.
1. Jadavji T, et al. A practical guide for the diagnosis and treatment of pediatric pneumonia. Canad Med Assoc J 1997;156:S703-S711.
Dr. Felz, Associate Professor, Department of Family Medicine, Medical College of Georgia, Augusta, is on the Editorial Board of Emergency Medicine Alert.