Do published guidelines predict pneumonia in children presenting to an urban ED?

Objectives: In 1997, a Canadian task force published evidence-based guideli nes for diagnosing pediatric pneumonia, concluding that the absence of each of four signs (ie, respiratory distress, tachypnea, crackles, and decrease d breath sounds) accurately excludes pneumonia. The study was performed to evaluate the accuracy of these guidelines in predicting pneumonia in young children. Methods: This was an observational study conducted over a 4-month period at an urban emergency department with 80,000 annual visits, approximately 20% of which were children less than or equal to5 years old. Consecutive child ren less than or equal to5 years old who underwent chest radiography were e nrolled. Prior to ordering radiographs, treating physicians were required t o enter specific patient signs and symptoms into a computerized database. W orld Health Organization criteria were used to define tachypnea. Sensitivit y, specificity, and predictive values of the task force guidelines in predi cting pneumonia were calculated. Results: Three hundred twenty-nine children, including 67 (20%) with pneumo nia, were enrolled. Guidelines were 45% sensitive (95% confidence interval (CI) = 33-58) and 66% specific (95% Cl = 60-72) for diagnosing pneumonia. P ositive and negative predictive values were 25% (95% Cl = 18-34) and 82% (9 5% CI = 77-87), respectively. Conclusion: Previously published evidence-based guidelines for excluding pe diatric pneumonia were found unreliable in this study.