Interobserver variation in the diagnosis of bronchiectasis on high-resolution computed tomography

The purpose of our study was to determine interobserver variation in the analysis of high-resolution computed tomography (HRCT) in the lungs of patients with clinically suspected bronchiectasis. HRCT scans of 88 patients were analysed independently by three radiologists with variable experience in thoracic radiology using a subjective scoring system to record bronchi as normal, mildly abnormal or severely abnormal. The presence, severity and distribution of bronchial dilatation and bronchial wall thickening were recorded. Kappa values were calculated for assessment of interobserver agreement. Agreement between the three readers was good for the detection of bronchiectasis (kappa 0.78) and assessment of its severity (0.68), detection of bronchial wall thickening (0.64) and moderately good for the assessment of its severity (0.58) on a per-patient basis. When individual lobes were analysed, agreement was moderately good for the detection of abnormal bronchi (0.59). Agreement on the extent of abnormal bronchi using five categories was only fair (0.39), but was good when differences of one category were ignored (0.63). Interobserver variation with HRCT in suspected bronchiectasis appears satisfactory for comparative studies.