527 - Radiographic Uncertainty and Antibiotic Prescribing in Pediatric Pneumonia

Publication Number: 2512.527

Alexandra T. Geanacopoulos, Boston Children's Hospital, Jamaica Plain, MA, United States; Michael Monuteaux, Boston Children's Hospital, Boston, MA, United States; Kyle Nelson, Harvard Medical School, Boston Children's Hospital, Boston, MA, United States; Susan Lipsett, Boston Children's Hospital, Boston, MA, United States; Alexander Hirsch, Boston Children's Hospital, Boston, MA, United States; Janine Amirault, Boston Children's Hospital, Boston, MA, United States; Cynthia A. Gravel, Boston Children's Hospital, Boston, MA, United States; Mark Neuman, Harvard Medical School, boston, MA, United States

Background: Over one million U.S. children are diagnosed with community-acquired pneumonia (CAP) in ambulatory settings annually, yet the diagnostic utility of chest radiograph (CXR) is unclear. Many clinicians perform CXR, while others prescribe empiric antibiotics, reflecting challenges in the history and exam, mixed CXR evidence, and CXR availability. Equivocal CXR may drive antibiotic overuse, but data is limited.

Objective: We evaluated how radiographic uncertainty (i.e. equivocal CXR) influences antibiotic prescribing in children undergoing CAP evaluation.

Design/Methods: We prospectively evaluated children 3 months-18 years old with CXR for possible CAP in a large pediatric emergency department (ED) between March 2015 and April 2025, excluding those with complex chronic conditions, current antibiotics, and bacterial coinfection. Three pediatric ED physicians blinded to clinical data categorized CXR reports as no pneumonia, equivocal, and pneumonia. Treating attending physicians indicated whether they would have prescribed antibiotics without the CXR (i.e. pre-CXR plan). We dichotomized children by pre-CXR antibiotic plan and performed adjusted logistic regression models to evaluate the association between CXR interpretation and antibiotic prescription. We performed sensitivity analyses to explore how different equivocal results (i.e. favor bronchiolitis/atelectasis, equal probability bronchiolitis/atelectasis and pneumonia, favor pneumonia) influence decision-making.

Results: Of 2,118 children evaluated, most were < 5 years (64%) with fever (75%), cough (92%), and dyspnea (60%). CXRs showed no pneumonia in 73%, pneumonia in 13%, and equivocal findings in 14%. Attendings planned to treat 297 (14%) children with antibiotics pre-CXR and 587 (28%) received treatment (Figure 1). Among children with no pre-CXR antibiotic plan, an equivocal CXR (adjusted OR (aOR): 38.5, 95% CI 26.2-57.6) and pneumonia on CXR (aOR 374.0, 95% CI 191.0-826.0) were associated with prescribing versus no pneumonia on CXR (Table 1). Among children with a pre-CXR plan for antibiotics, an equivocal CXR (adjusted OR (aOR): 37.6, 95% CI 12.4-166.0) and pneumonia on CXR (aOR 43.9, 95% CI 16.1-158.0) were also associated with increased prescribing. In sensitivity analyses, all equivocal categories were associated with increased prescribing (Table 2).

Conclusion(s): Radiographic uncertainty, present in 1 in 7 children with possible CAP, is associated with increased antibiotic prescribing. CXRs may inadvertently drive antibiotic use, highlighting the urgent need for decision support to guide targeted CXR use and management of diagnostic uncertainty.

Figure 1. Pre-CXR antibiotic plan vs. post-CXR antibiotic decision, stratified by CXR interpretation.
Proportion of patients with suspected CAP whom the attending clinician would have treated with antibiotics without CXR (light blue) vs. proportion of patients who received antibiotics after CXR (dark blue), stratified by final CXR interpretation. CXR categories include: no pneumonia (normal CXR, bronchiolitis/atelectasis), equivocal (favor bronchiolitis/atelectasis, equal probability of bronchiolitis/atelectasis or pneumonia, favor pneumonia), and pneumonia. Error bars represent the Wilson 95% confidence intervals for each point estimate.

Table 1: Logistic regression models evaluating the association between CXR interpretation and antibiotic prescribing in children with and without pre-CXR antibiotic plan


Table 2: Sensitivity analysis evaluating the association between 5-level definition of CXR interpretation and antibiotic prescribing in children with and without pre-CXR antibiotic plan