Observer variability and sensitivity of radiographic diagnosis of canine medial coronoid disease

Medial coronoid disease (MCD) is a very common form of elbow joint disease and it's radiographic diagnosis can be challenging since it is frequently based on the detection of rather subtle primary or secondary changes than on a large primary lesion. We hypothesized that accuracy of radiographic diagnosis of MCD is highly dependent on training and experience level.

Radiographs of 102 canine elbows were evaluated for MCD by four observers with different levels of training and experience. All elbows underwent CT scans and arthroscopy. Sensitivity and specificity of radiographic and CT interpretation was determined using arthroscopy as a gold standard. Interobserver and intraobserver agreement (reliability and repeatability) were assessed by using Cohen's Kappa (κ) statistic.

The sensitivity (92.4-96.7%) of the two experienced observers was almost comparable to that of CT (100%) and significantly higher than that of the two less experienced observers (77.2-80.4%). Reliability of the radiographic diagnosis of MCD was better between observers with higher experience level (κ= 0.74) than between observers of lower or different experience levels (κ=0.07-0.42). Repeatability was better in experienced (κ= 0.73-0.88) than in less experienced observers (κ= 0.31-0.42).

Our results confirm that training and experience play important roles in reaching high sensitivity, reliability and repeatability for the radiographic diagnosis of MCD.

Although radiography is inferior to CT in imaging of the medial coronoid process itself, sensitivity of radiographic diagnosis MCD can be significantly improved with observer experience almost reaching that of CT. Therefore, it is advised that radiographic screening for MCD should be performed by specialists experienced in the radiographic evaluation of elbow joint disease.