Observer Variation in Bronchographic Diagnosis

Observer variation* in the interpretation of chest radiographs has been estimated by a number of workers (Birkelo, Chamberlain, Phelps, Schools, Zacks and Yerushalmy, 1947; Cochrane and Garland, 1952; Fletcher and Oldham, 1949; Garland, 1949; Springett, 1956; Stradling and Johnston, 1953; Yerushalmy, Harkness, Cope and Kennedy, 1950; Yerushalmy, Garland, Harkness, Hinshaw, Miller, Shipman and Zwerling, 1951; and Zwerling, Miller, Harkess and Yerushalmy, 1951) and the mechanisms involved have been studied by Newell, Chamberlain and Rigler (1954), and by Clayson, Frew, McIntosh, McWhirter, McKinley and Stein (1955). In a recent investigation into the tomographic diagnosis of active tuberculous cavitation (Gandevia and Stradling, 1957), it seemed that the chief source of observer variation lay in the interpretation, in terms of pathology, of shadows which both observers had seen rather than in a failure to see a “ring shadow” or failure to agree upon the presence of an air space of some kind. In order to test the truth of this hypothesis we decided to study observer variation in bronchographic diagnosis. In bronchiectasis the shadows cast by the contrast medium are large, localised, sharply demarcated and in excellent contrast to the background. In addition proximal and unaffected bronchi serve for comparison with doubtfully abnormal areas. On the other hand, the bronchographic diagnosis of chronic bronchitis is based upon the recognition and careful interpretation of multiple tiny lesions, or upon minor changes in the bronchial walls, for which there may be no appropriate “control” areas for comparison.