Resident in Pediatrics Boston Children's Hospital Boston Children's Hospital Boston, Massachusetts, United States
Background: Chest radiograph (CXR) is common for pediatric pneumonia evaluation, though it is not clear how specific radiographic patterns impact clinical decision-making.
Objective: We assessed the impact of viral radiographic features on pneumonia diagnosis and antibiotic prescribing.
Design/Methods: Children 3 months to 18 years who had a CXR for pneumonia evaluation in a pediatric emergency department (ED) between May 2015 and March 2020 were enrolled. Prior to CXR, physicians indicated their clinical suspicion for pneumonia and antibiotic plan. Viral features (peribronchial cuffing, perihilar markings, interstitial infiltrate) and pneumonia features (consolidation, opacity, infiltrate) were identified on CXR. The relationship between radiographic features and pneumonia diagnosis was assessed by clinical suspicion for pneumonia prior to CXR. Antibiotics were classified as either narrow-spectrum, broad-spectrum, or macrolide. Chi-square and McNemar’s analyses were used to assess differences in pneumonia diagnosis by CXR features and in pre-CXR antibiotic plan versus treatment.
Results: Of the 1,121 patients, 400 (35.7%) had normal CXRs, 370 (33.0%) had viral features alone, 177 (15.8%) had both viral features and radiographic pneumonia, and 174 (15.5%) had radiographic pneumonia alone (Table 1). There was no significant difference in pneumonia diagnosis in patients with viral features alone versus normal CXRs (8.6% vs. 8.0%, p=0.75, Figure 1). Pneumonia diagnosis was less common in children with both viral features and radiographic pneumonia versus radiographic pneumonia alone (54.3% vs. 86.2%, p<0.001). There were no differences in the pre-CXR antibiotic plan and treatment for patients with normal CXRs (10.3% vs. 8.0%, p=0.24) or viral features alone (12.1% vs. 8.6%, p=0.10, Figure 2). Children with radiographic pneumonia were treated more often as compared to the pre-CXR plan (86.2% vs. 23.4% p<0.001). Most patients (58.6%) received narrow-spectrum antibiotics. Macrolide monotherapy was common for normal CXRs (43.9%) and CXRs with viral features alone (23.1%) compared to radiographic pneumonia (6.6%) (Figure 3). Conclusion(s): Antibiotic prescribing rates are similar among children with normal CXRs and CXRs with viral features alone, but lower for children with both viral features and radiographic pneumonia compared to children with radiographic pneumonia alone. Given that most CXRs demonstrate viral features or are normal, CXRs may be useful in avoiding antibiotic overuse.
Table 1: Patient Baseline Characteristics
Figure 1: Impact of Chest Radiograph Classification on Pneumonia Diagnosis for All Patients and Stratified by Level of Clinical Suspicion for Pneumonia.
Figure 2: Impact of Chest Radiograph Classification on Pneumonia Diagnosis. Pre-radiographic antibiotic plan represents the proportion of patients who would have received antibiotics and a pneumonia diagnosis based on the history and clinical exam alone. Post-radiographic antibiotics given represents the proportion of patients who received antibiotics and a pneumonia diagnosis after incorporating the chest radiograph into medical decision making.
Figure 3: Pneumonia Diagnosis Stratified by Chest Radiograph Classification and Antibiotic Class Prescribed. Narrow-spectrum presents an aminopenicillin. Broad-spectrum represents an aminopenicillin with a beta-lactamase inhibitor, cephalosporin, fluoroquinolone, vancomycin, or piperacillin and tazobactam.
Authors/Institutions: Alexandra T. Geanacopoulos, Boston Children's Hospital, Brookline, Massachusetts, United States; Susan C. Lipsett, Boston Children's Hospital, Cambridge, Massachusetts, United States; Alexander W. Hirsch, Boston Children's Hospital, Boston, Massachusetts, United States; Mark Neuman, Boston Children's Hospital, Boston, Massachusetts, United States