Diagnostic sensitivity and interobserver agreement of radiography and ultrasonography for detecting trochlear ridge osteochondrosis lesions in the equine stifle

Abstract

Osteochondrosis lesions commonly occur on the femoral trochlear ridges in horses and radiography and ultrasonography are routinely used to diagnose these lesions. However, poor correlation has been found between radiographic and arthroscopic findings of affected trochlear ridges. Interobserver agreement for ultrasonographic diagnoses and correlation between ultrasonographic and arthroscopic findings have not been previously described. Objectives of this study were to describe diagnostic sensitivity and interobserver agreement of radiography and ultrasonography for detecting and grading osteochondrosis lesions of the equine trochlear ridges, using arthroscopy as the reference standard. Twenty-two horses were sampled. Two observers independently recorded radiographic and ultrasonographic findings without knowledge of arthroscopic findings. Imaging findings were compared between observers and with arthroscopic findings. Agreement between observers was moderate to excellent (κ 0.48-0.86) for detecting lesions using radiography and good to excellent (κ 0.74-0.87) for grading lesions using radiography. Agreement between observers was good to excellent (κ 0.78-0.94) for detecting lesions using ultrasonography and very good to excellent (κ 0.86-0.93) for grading lesions using ultrasonography. Diagnostic sensitivity was 84-88% for radiography and 100% for ultrasonography. Diagnostic specificity was 89-100% for radiography and 60-82% for ultrasonography. Agreement between radiography and arthroscopy was good (κ 0.64-0.78). Agreement between ultrasonography and arthroscopy was very good to excellent (κ 0.81-0.87). Findings from this study support ultrasound as a preferred method for predicting presence and severity of osteochondrosis lesions involving the femoral trochlear ridges in horses.