BACKGROUND: Chest X-ray interpretation is an important skill in the diagnosis of congestive heart failure (CHF) by emergency physicians. OBJECTIVES: This study evaluated the ability of emergency physicians to recognize CHF on chest X-ray and the effect of level of training and confidence upon accuracy of interpretation. METHODS: This was a prospective, blinded study in which 24 patients with an elevated brain natriuretic peptide, low ejection fraction, and diagnosis of CHF were retrospectively identified. In addition, 31 patients without CHF were identified and used as controls. These 55 chest X-rays were presented to emergency attending and housestaff and a radiologist. We calculated the accuracy of the raters' diagnoses, and measured their confidence in that diagnosis and their level of training. RESULTS: Physicians correctly identified the CHF chest X-rays 79% of the time (sensitivity 59%, specificity 96%; positive likelihood ratio 14.6, negative likelihood ratio 0.43). Accuracy ranged from a low of 78% among first-year residents to a high of 85% among attending, and from 73% (confidence rating of 3/5) to 91% (confidence rating of 5/5). Increasing confidence was significantly correlated with accuracy across the spectrum (p = 0.001). An accuracy of 95% among radiologists suggests that a negative X-ray does not rule out CHF. CONCLUSIONS: High specificity (96%) and low sensitivity (59%) suggest that emergency physicians are excellent at identifying CHF on X-ray when present, but under-call it frequently. Sensitivity may be much higher in real life given clinical correlation. Both increased level of training and higher confidence significantly improved accuracy. Copyright Â
BACKGROUND: Chest X-ray interpretation is an important skill in the diagnosis of congestive heart failure (CHF) by emergency physicians. OBJECTIVES: This study evaluated the ability of emergency physicians to recognize CHF on chest X-ray and the effect of level of training and confidence upon accuracy of interpretation. METHODS: This was a prospective, blinded study in which 24 patients with an elevated brain natriuretic peptide, low ejection fraction, and diagnosis of CHF were retrospectively identified. In addition, 31 patients without CHF were identified and used as controls. These 55 chest X-rays were presented to emergency attending and housestaff and a radiologist. We calculated the accuracy of the raters' diagnoses, and measured their confidence in that diagnosis and their level of training. RESULTS: Physicians correctly identified the CHF chest X-rays 79% of the time (sensitivity 59%, specificity 96%; positive likelihood ratio 14.6, negative likelihood ratio 0.43). Accuracy ranged from a low of 78% among first-year residents to a high of 85% among attending, and from 73% (confidence rating of 3/5) to 91% (confidence rating of 5/5). Increasing confidence was significantly correlated with accuracy across the spectrum (p = 0.001). An accuracy of 95% among radiologists suggests that a negative X-ray does not rule out CHF. CONCLUSIONS: High specificity (96%) and low sensitivity (59%) suggest that emergency physicians are excellent at identifying CHF on X-ray when present, but under-call it frequently. Sensitivity may be much higher in real life given clinical correlation. Both increased level of training and higher confidence significantly improved accuracy. Copyright Â
Authors: Matthew D Li; Nishanth Thumbavanam Arun; Mishka Gidwani; Ken Chang; Francis Deng; Brent P Little; Dexter P Mendoza; Min Lang; Susanna I Lee; Aileen O'Shea; Anushri Parakh; Praveer Singh; Jayashree Kalpathy-Cramer Journal: Radiol Artif Intell Date: 2020-07-22