Intra- and interobserver reliability estimates for identification and grading of upper respiratory tract abnormalities recorded in horses at rest and during overground endoscopy

Abstract

Background: Previous studies support good intra- and interobserver agreements for endoscopic evaluation of various upper respiratory tract (URT) diseases in horses. However, these studies mainly assessed resting endoscopic examination videos and/or focussed on a single URT abnormality. Objectives: To estimate intra- and interobserver agreement for identification and grading of all URT abnormalities from resting and overground endoscopy (OGE) videos of Thoroughbreds. Study design: Blinded, fully crossed design. Methods: Resting and OGE URT videos for n = 43 Thoroughbreds were retrospectively chosen based on identification of common URT disorders. The videos were randomly evaluated in duplicate by 4 raters blinded to all information including prior URT disorder(s) diagnosis. Abnormalities were graded using well-described ordinal scales. Intra- and interobserver agreements were estimated using Cohen's weighted κ and Krippendorff's α, respectively. Results: Intraobserver agreement was perfect/nearly perfect for arytenoid symmetry at exercise, epiglottic entrapment and epiglottic retroversion, substantial for arytenoid asymmetry at rest, palatal dysfunction (PD), medial deviation of the aryepiglottic folds (MDAF), pharyngeal mucus and epiglottic grade at exercise and moderate for vocal fold collapse (VFC), ventromedial luxation of the apex of the corniculate process of the arytenoid (VLAC), nasopharyngeal collapse (NPC) and epiglottic grade at rest. Interobserver agreement was substantial for arytenoid symmetry at exercise and PD and moderate for arytenoid asymmetry at rest, MDAF, VLAC and epiglottic entrapment. It was only fair for VFC, epiglottic grade at exercise, epiglottic retroversion, pharyngeal mucus and NPC and poor for epiglottic grade at rest. Main limitations: Sample size was insufficient to allow assessment of the effect of one abnormality on the grading of another abnormality. Conclusions: Observers were consistent in grading URT disorders. However, significant disparity in grading existed between observers for some conditions affecting reliability.