Related Experiment Videos
External Validation of Risk Scores for Pediatric Pneumonia
Joan L Robinson1, James D Kellner2, Jennifer Crotts2
1From the Faculty of Medicine & Dentistry, University of Alberta, Edmonton, Alberta, Canada.
Background:
Neither the Pneumonia Risk Score (PRS) nor the VALS-DANCE diagnostic tool has been externally validated using a reference standard specific for bacterial pneumonia. We used a previously published cohort of children with suspected bacterial pneumonia to externally validate these scores.
Methods:
Children attending Canadian pediatric emergency departments with suspected pneumonia had extensive standardized data collected that were provided to an expert panel to reach a consensus diagnosis. This diagnosis was used to externally validate both clinical scores. Proportions, point estimates and 95% confidence intervals (CIs) were calculated, as appropriate, for the classifications and cutoffs recommended for PRS and VALS-Dance.
Results:
In a cohort of 236 children, the PRS classified 69 as low risk, 155 as moderate risk and 12 as high risk for "radiographic" pneumonia. Of these, 11/69 (15.9%), 44/155 (28.4%) and 5/12 (41.7%), respectively, had bacterial pneumonia, yielding a receiver operating characteristic curve of 0.64 (95% CI 0.56-0.71). The PRS low-risk grouping yielded negative likelihood ratio (-LR) 0.56 (95% CI 0.24-0.88) and the high-risk grouping a positive LR (+LR) 2.10 (95% CI -0.23 to 4.42) for bacterial pneumonia. In a cohort of 94 children, the VALS-DANCE Step1 cutoff yielded +LR 2.97 (95% CI 1.52-4.08) and -LR 0.19 (95% CI 0.05-0.33) for distinguishing bacterial from viral pneumonia. In a cohort of 53 children, the VALS-DANCE Step2 cutoff yielded +LR 13.1 (95% CI 0.83-208.8) and -LR 0.65 (95% CI 0.51-0.84) for distinguishing typical from atypical bacterial pneumonia.
Conclusions:
The PRS may have limited utility in the emergency department in guiding clinicians as to whether children with suspected pneumonia need antibiotics. For children hospitalized with community-acquired pneumonia, the VALS-DANCE Step1 diagnostic tool may have some utility for ruling out, but little for confirming bacterial pneumonia. For those with bacterial pneumonia, VALS-DANCE Step2 has uncertain utility for confirming typical and little utility for diagnosing atypical pneumonia.
