No clue about clubbing – interobserver agreement on the presence of clubbing in interstitial lung diseases
Epidemiology, diagnosis and monitoringRationale
Clubbing is considered a clinical feature of idiopathic pulmonary fibrosis (IPF) and other fibrotic interstitial lung diseases (ILDs), and believed to be associated with poor prognosis. There is no 'gold standard' to assess clubbing. In daily practice physicians determine clubbing at sight, however, we do not know how reliable this is. We aimed to investigate the agreement of clubbing assessment among pulmonary healthcare providers.
Consecutive outpatients with fibrotic ILDs of a tertiary referral center were asked to participate. Photographs of the hands and index fingers of these patients were made and displayed in a digital survey. This survey was send to international pulmonary physicians, residents and ILD specialist nurses, who scored presence of clubbing and their confidence of diagnosis. K alpha was used to asses interrater reliability and independent sample t-tests for differences between groups.
We included photographs of 87 patients with fibrotic ILDs, 42% with IPF, 12% underlying collagen vascular disease, 12% unclassifiable fibrosis, 8% chronic hypersensitivity pneumonitis, and 27% other fibrotic ILDs. Mean age of patients was 67 years (range33-84), 62% male, mean FVC 81% (range 25-145) and TLCOc 55% (range 20-104). The presence of clubbing was assessed by 45 raters; 40% male, 49% pulmonary physician, 38% pulmonary resident and 13% ILD specialist nurse, 69% work in a university hospital and 49% are specialized in ILD.
Clubbing was scored as present in 36% of cases, with a range of 7-79%. Only in 11% of cases all raters unanimously agreed on the presence or absence of clubbing. Raters were 89% sure of their judgment in those cases; this is significantly higher than the cases in which they did not unanimously agree (79%, p <0,001) (figure 1). Male raters were significantly more confident of their overall judgements (84%) than female raters (77%, p=0,031). The interrater reliability for the assessment of clubbing was low (α = 0,27 95%CI = 0,06-0,48). No differences are found between ILD specialized physicians(α = 0,23) versus non-specialized physicians (α = 0,33), pulmonary physicians (α = 0,32), residents (α = 0,36) and nurses (α = 0,33), university hospital (α = 0,28) versus non-university hospital (α = 0,25), and <5 years (α = 0,32) versus >5 years of experience (α = 0,25).
Interobserver agreement on the presence of clubbing in patients with ILDs is poor. This implies that the message on the diagnostic and prognostic value of clubbing in ILD should be nuanced.
Figure 1 a. 100% agreement on presence of clubbing, average confidence 90% b. 53% scored clubbings as present and 47% as absent, average confidence 75%