Agreement between multiple objective and subjective equine lameness evaluators

Abstract

Lameness evaluations, which tend to be subjective in manner, are a critical component of equine musculoskeletal health assessments. While typically sufficient for in-field diagnostics, scientific studies often require more precise and repeatable techniques. In recent years, several methods to assess equine gait asymmetries have become available. This study sought to evaluate the agreement between 2 veterinary evaluators, an AI-based smart phone application (Sleip), and 2 inertial sensor systems (Equinosis Q and Equisym). Eighteen research horses with naturally occurring lameness were jogged in hand on a straight 30-m asphalt path for 3 round trips. Data output from each observer or system was converted to a 0-to-3 scale based on recorded degree of asymmetry for each limb. Agreement in asymmetry scores were quantified via weighted Cohen's κ, and P-values were less than 0.05 unless noted. Objective evaluator agreement (κ = 0.70) was higher than subjective evaluator agreement (κ = 0.41) across all limbs. Objective systems had a higher agreement with hind limb asymmetries (κ = 0.61) when compared with the subjective evaluators (κ = 0.15, P > 0.05). A similar pattern for forelimb asymmetries was present between objective (κ = 0.81) and subjective evaluators (κ = 0.67), although the difference in agreement was less pronounced. Pairwise weighted Cohen's κ scores are presented in Table 1. Individual agreement scores between the objective systems were generally higher and less variable than the veterinary evaluators. Data from this project suggest that objective systems are less apt to variability when detecting subtleties in hind limb asymmetries compared with straight-line veterinary examination. Further, the objective systems had consistently higher agreement. Limitations of this study included only jogging horses in a straight line as lunging and flexion tests would likely have affected the agreement between subjective and objective evaluators. This is some of the first work comparing the Equisym to other objective systems while also showing these systems have better asymmetry detection capabilities and agreement compared with veterinary examination when horses trot in a straight line.