Validation of a method to predict the etiology of pneumonia in children

Introduction. Identifying etiology in children with pneumonia requires time and technical resources, not always available. Therefore, the initial management of pneumonia is often based on clinical, laboratory, and/or radiographic data. Clinical prediction rules based on a combination of factors could increase diagnostic accuracy. Objective. To validate the diagnostic accuracy of a clinical prediction rule (Bacterial Pneumonia Score, BPS) to distinguish bacterial from viral pneumonia in children. Population and methods. This observational, diagnostic test evaluation study was performed among January 2004 and December 2006. Children aged 1 month to 15 years old, hospitalized for pneumonia in whom a bacterial or viral etiology was identified were included. Children with chronic pulmonary disease, congenital heart disease, admission to the intensive care unit, underlying immunologic disease, mixed viral and bacterial infection, or inability to identify viral or bacterial pathogens were excluded. Admission data were recorded (age, temperature, WBC count and chest radiograph evaluation). BPS was then calculated (range -3 to 15 points), taking into account that a BPS >= 4 suggests bacterial pneumonia. Results. We included 82 patients aged 1 to 96 months with pneumonia (79% viral and 21% bacterial). A BPS >= 4 predicted bacterial pneumonia with sensibility: 94%, specificity: 34%, positive predictive value: 27%, and negative predictive value: 95%. Conclusion. The BPS was accurate on identifying most children with bacterial pneumonia, who required antibiotic therapy.