A predictive model has excellent discrimination for radiographic community-acquired pneumonia (CAP) in children, according to a study published in Pediatrics. Sriram Ramgopal, MD, and colleagues constructed a predictive model for radiographic CAP based on clinical features using data from patients aged 3 months to 18 years with signs of lower respiratory infection. Radiographic CAP was identified in 22.2% of 1,142 patients. Increasing age, prolonged fever duration, tachypnea, and focal decreased breath sounds were positively associated with CAP in an adjusted analysis. There was a negative association between CAP and rhinorrhea and wheezing. A bootstrapped reduced model with three variables—age, fever duration, and decreased breath sounds—had an area under the receiver operating characteristic of 0.80. Among 229 children with a predicted risk of less than 4%, 5.7% had radiographic CAP (sensitivity, 94.9%); among 229 children with a predicted risk of greater than 39%, 61.1% had CAP (specificity, 90.0%). “When predicted risk is low or high, this model may allow for decreased chest radiograph use,” Dr. Ramgopal and colleagues wrote.
