Written by Kimiko Dunbar
Spoon Feed
In this AI-generated case scenario community acquired pneumonia (CAP) study, CXR ordering preferences weakly correlated with the Pneumonia Risk Score decision tool, and physicians primarily relied on CXR results to guide antibiotic use.
Snap, crackle… CXR
Diagnosis of community acquired pneumonia (CAP) remains variable. Despite IDSA guidelines to avoid CXR if not requiring hospitalization, many children are still receiving imaging. This study aimed to better understand physician practices in diagnosis of pneumonia and empiric antibiotic use to inform the application of CAP clinical decision making tools. Authors examined clinician preferences for CXR and antibiotic use in CAP using AI-generated case scenarios. In a cross-sectional survey of 172 physicians evaluating 839 cases, crackles (OR 4.17), prior pneumonia (OR 2.38), chest pain (OR 1.90) and fever (OR 1.82) most influenced CXR decisions, while past hospitalization (OR 4.24), focal decreased breath sounds (OR 3.86) and crackles (OR 3.45) predicted empiric antibiotic use. Ultimately, CXR findings dictated final antibiotic decisions. The study compared factors associated with physician suspicion for pneumonia with two prediction models, the Pneumonia Risk Score (PRS) and the Catalyzing Ambulatory Research in Pneumonia Etiology and Diagnostic Innovations in Emergency Medicine (CARPE DIEM) model. Suspicion weakly correlated with PRS (spearman coefficient 0.25, CI: 0.18-0.32), but not with CARPE DIEM (spearman coefficient 0.02, CI: -0.05-0.09). Limitations include selection bias and reliance on synthetic data. Findings support integrating clinical decision support tools.
How will this change my practice?
This study demonstrates one of the many areas in which physicians practice outside of evidence. Admittedly, I too would prefer to have a CXR in a patient diagnosed with CAP and can understand the overutilization of resources. I’m looking forward to the development of clinical decision making tools to align my practice with evidence.
Editor’s note: CARPE DIEM found poor to moderate agreement on clinical exam findings – such as crackles, rhonchi, or even decreased breath sounds – in children with suspicion of pneumonia. The IDSA recommendation to treat empirically based on history and exam, without CXR, is questionable, since exam findings are unreliable. ~Clay Smith
Sources
Preferences for Management of Pediatric Pneumonia: A Clinician Survey of Artificially Generated Patient Cases. Pediatr Emerg Care. 2025 Jan 1;41(1):41-49. doi: 10.1097/PEC.0000000000003231. Epub 2024 Jul 1. PMID: 38950412
