Inter- and intraobserver agreement in interpretation of CT features of medial coronoid process disease

OBJECTIVES To evaluate inter- and intraobserver reliability of the assessment of computed tomography features commonly used in the identification and classification of medial coronoid process disease and to assess inter- and intraobserver variability in the identification of the percentage ulna sclerosis from single transverse computed tomography images. METHODS Eight observers, on two occasions, reviewed 84 standardised single transverse computed tomography images acquired at the level of the apex of the medial coronoid process. Observers assessed: medial coronoid process disease, coronoid process fragmentation, osteophytes, sclerosis grade and sclerosis delineation with normal bone defined using a sclerometer. Cohen's kappa and intraclass correlation coefficient were calculated. RESULTS Inter-observer agreement was fair to moderate for identification of computed tomography changes consistent with medial coronoid process disease, moderate to almost perfect for fragmentation, and moderate for osteophyte detection. Agreement was poor for sclerosis grading. Percentage sclerosis measured with a sclerometer had moderate to almost perfect inter- and intraobserver agreement. CLINICAL SIGNIFICANCE These findings suggest that subjective assessment of computed tomography images is less repeatable than more quantitative methods.