Diagnostic accuracy of a clinical diagnosis of idiopathic pulmonary fibrosis: an international case–cohort study

We conducted an international study of idiopathic pulmonary fibrosis (IPF) diagnosis among a large group of physicians and compared their diagnostic performance to a panel of IPF experts.A total of 1141 respiratory physicians and 34 IPF experts participated. Participants evaluated 60 cases of interstitial lung disease (ILD) without interdisciplinary consultation. Diagnostic agreement was measured using the weighted kappa coefficient (κ_w). Prognostic discrimination between IPF and other ILDs was used to validate diagnostic accuracy for first-choice diagnoses of IPF and were compared using the C-index.A total of 404 physicians completed the study. Agreement for IPF diagnosis was higher among expert physicians (κ_w=0.65, IQR 0.53-0.72, p<0.0001) than academic physicians (κ_w=0.56, IQR 0.45-0.65, p<0.0001) or physicians with access to multidisciplinary team (MDT) meetings (κ_w=0.54, IQR 0.45-0.64, p<0.0001). The prognostic accuracy of academic physicians with >20 years of experience (C-index=0.72, IQR 0.0-0.73, p=0.229) and non-university hospital physicians with more than 20 years of experience, attending weekly MDT meetings (C-index=0.72, IQR 0.70-0.72, p=0.052), did not differ significantly (p=0.229 and p=0.052 respectively) from the expert panel (C-index=0.74 IQR 0.72-0.75).Experienced respiratory physicians at university-based institutions diagnose IPF with similar prognostic accuracy to IPF experts. Regular MDT meeting attendance improves the prognostic accuracy of experienced non-university practitioners to levels achieved by IPF experts.