Assessment of physicians' proficiency in reading chest radiographs for pneumoconiosis, based on a 60-film examination set with two factors constituting eight indices

Abstract

Two hundred and thirty-three individuals read chest x-ray images (CXR) in the Asian Intensive Reader of Pneumoconiosis (AIR Pneumo) workshop. Their proficiency in reading CXR for pneumoconiosis was calculated using eight indices (X1-X8), as follows: sensitivity (X1) and specificity (X2) for pneumoconiosis; sensitivity (X3) and specificity (X4) for large opacities; sensitivity (X5) and specificity (X6) for pleural plaques; profusion increment consistency (X7); and consistency for shape differentiation (X8). For these eight indices, one-way analysis of variance (ANOVA) and Scheffe's multiple comparison were conducted on six groups, based on the participants' specialty: radiology, respiratory medicine, industrial medicine, public health, general internal medicine, and miscellaneous physicians. Our analysis revealed that radiologists had a significant difference in the mean scores of X3, X5, and X8, compared with those of all groups, excluding radiologists. In the factor analysis, X1, X3, X5, X7, and X8 constituted Factor 1, and X2, X4, and X6 constituted Factor 2. With regard to the factor scores of the six participant groups, the mean scores of Factor 1 of the radiologists were significantly higher than those of all groups, excluding radiologists. The two factors and the eight indices may be used to appropriately assess specialists' proficiency in reading CXR.

Keywords: Chest x-ray image; Factor analysis; International Labor Organization Classification; Pneumoconiosis; Sensitivity; Specificity.

Fig. 1.

Factor loading with two factors by using orthogonal varimax rotation.