Reliability of radiological assessment of ulnar trochlear notch sclerosis in dysplastic canine elbows

Abstract

The objectives of this study were to quantify the sensitivity and specificity of visual assessment of radiographs of the canine elbow in detecting ulnar trochlear notch sclerosis, to establish interobserver and intra-observer variation for the presence and grade of sclerosis and to quantify the effect of radiographic exposure on observer grading. Mediolateral elbow radiographs were obtained from Labrador retrievers (n=34) aged between six and 18 months. Radiographs from dogs with an arthroscopic diagnosis of fragmented medial coronoid process (n=17) and those from a control population (n=17) were subjected to observer grading for the presence or absence of and the grade of ulnar trochlear notch sclerosis. Interobserver and intra-observer variation and observer sensitivity and specificity were calculated. Digital data from the ulnar trochlear notch were correlated with mean observer grade to quantify the effect of radiographic exposure on observer grade. Interobserver agreement was "fair" (kappa=0.251 to 0.369) and intra-observer agreement was "moderate" to "substantial" (kappa=0.462 to 0.667). The sensitivity of observer assessment was 72 per cent with a specificity of 22 per cent. Mean observer grade was not significantly correlated with the degree of radiographic exposure (P=0.70). Ulnar trochlear notch sclerosis is a phenomenon associated with fragmented medial coronoid process. However, interobserver agreement in grading this feature is only fair, being identified by observers with moderate sensitivity but with relatively poor specificity. This low specificity may predispose to overdiagnosis in clinical cases. Intra-observer agreement is moderate to substantial, suggesting that individuals can reliably quantify this radiological feature on multiple occasions. The ability of observers to assess the degree of sclerotic change is not significantly affected by radiographic exposure.