Evaluation of radiographic heart size measurement repeatability and reproducibility among veterinary students with varying training levels

Vertebral heart scale (VHS) and vertebral left atrial size (VLAS) are commonly used to assess the heart size. It remains unknown if observers with different levels of training within the same protocol would improve the repeatability and reproducibility. This study aims to describe the measurement errors and to assess if different levels of experience can influence the outcome. Radiographs of healthy dogs and dogs with myxomatous mitral valve disease were included. Nine veterinary students were divided into three groups and followed a learning phase by measuring a set of 10, 30 and 50 images respectively, followed by an assessment of measurement error. Next, each group measured 20 radiographs twice, and the measurements of an experienced observer were used to calculate the reproducibility and repeatability. For VHS and VLAS measurements, 67.8% and 54.1% respectively, had at least one error. Interobserver agreement for VHS was comprised between 0.948 and 0.993 while for VLAS between 0.819 and 0.978. Intraobserver agreement for VHS was comprised between 0.959 and 0.995, while for VLAS between 0.71 and 0.994. This study shows that the measurement errors are common, however all students obtained a substantial to almost perfect agreement, suggesting that measuring experience does not influence the result.