Radiographic findings and etiologic diagnosis in ambulatory childhood pneumonias

Abstract

The chest roentgenograms of 128 consecutive ambulatory children with radiologic pneumonia were read independently and without clinical information by a faculty general pediatrician (Ped), a pediatric radiologist (R-P) and a general radiologist (R-G). The films were classified as normal, indicative of a viral or bacterial process, or indeterminate. Readings were compared with results of viral titers and bacterial cultures. Agreement between any two observers in classifying films, measured by unweighted Kappa, while statistically significant (p less than 0.001) for any pair, was low. There was no significant difference between the agreement scores of Ped/R-P, Ped/R-G, and R-P/R-G. Twenty-one patients had fourfold viral titer increases (N = 16) or positive bacterial cultures of blood or pulmonary aspirate (N = 5). The sensitivity of viral readings for titers increases varied from 19% to 68% depending on observer type; the sensitivity of bacterial readings for positive bacterial cultures varied from 60% to 80%. The three observers agreed on a correct reading in only three children with viral and three with bacterial pneumonia. Because of poor observer agreement and appreciable false-negative errors when viral and bacterial readings were compared to titer increases and positive bacterial cultures, respectively, we conclude that radiographic findings are poor indicators of etiology diagnosis in ambulatory childhood pneumonias and, of themselves, are an insufficient data base for making therapeutic decisions.