Purpose: The purpose of this study was to determine how often a second-opinion interpretation of interstitial lung disease (ILD) by an academic cardiothoracic radiologist is discordant with the initial interpretation by a nonacademic radiologists and how often the clinical diagnosis determined by multidisciplinary consensus agrees with the initial and second-opinion interpretations.
Methods: This retrospective study included 364 consecutive second-opinion CT examination reports of imaging from nonacademic radiology practices from July 2014 to May 2016. The second-opinion interpretations, provided by seven fellowship-trained cardiothoracic radiologists, were compared with the initial interpretations and the clinical diagnoses determined by multidisciplinary consensus.
Results: Two hundred ninety-six consecutive reports met the inclusion criteria, and two hundred had findings of ILD. The initial interpretations lacked specific diagnoses in 41% of reports, but the second-opinion reports lacked specific diagnoses in only 7%. When a diagnosis was provided, the second-opinion diagnosis disagreed with the initial interpretation in 25% of cases. The clinical-consensus diagnosis was concordant with that of the academic radiologists 85% of the time but concordant with the initial interpretation only 44% of the time. The academic radiologists' diagnostic sensitivity was higher than that of the initial radiologists for the four most common diagnoses: usual interstitial pneumonitis (0.91 versus 0.4), sarcoidosis (0.94 versus 0.60), hypersensitivity pneumonitis (0.79 versus 0.17), and nonspecific interstitial pneumonitis (0.72 versus 0.14).
Conclusions: Academic cardiothoracic radiologists were more likely to provide specific diagnoses for ILD, and these diagnoses were more likely to be concordant with the multidisciplinary consensus.
Keywords: Interstitial lung disease; interpretation discrepancy; radiology consultation; second opinion.
Copyright © 2019 American College of Radiology. Published by Elsevier Inc. All rights reserved.