Comparison of visual lameness scores to gait asymmetry in racing Thoroughbreds during trot in-hand

Regular monitoring of movement asymmetry with inertial measurement units (IMUs) to aid in the diagnosis of the underlying cause of a lameness is feasible. Normal ranges for specific groups of horses may be required, with consideration of expert veterinary opinions for both asymmetry screening and lameness assessment. The aim of this study was to determine movement asymmetry values compared with expert lameness scores to enable screening for lameness in Thoroughbreds in race training. IMU gait assessment during in-hand trot-up was performed in 25 racehorses undergoing routine gait analysis or lameness examination at the Singapore Turf Club. Video recordings were graded numerically (0–5) for lameness by six experienced racehorse veterinarians. Inter-observer agreement and consistency were determined. Median lameness scores were used to calculate sensitivity and specificity for head, withers and pelvic movement asymmetry. Guideline values for aligning movement asymmetry values with expert opinions about forelimb and hindlimb lameness were determined from receiver operating characteristics (ROC). Inter-observer agreement was poor to fair, inter-observer consistency was good (intraclass correlation coefficient: 0.667 for forelimbs and 0.617 for hindlimbs). ROCs indicated higher discriminative power for hindlimb lameness using pelvic asymmetry (90% sensitivity, 93% specificity) compared with forelimb lameness using head asymmetry (69% sensitivity, 89% specificity) or withers asymmetry (44% sensitivity, 89% specificity). When compared to expert lameness scores from videos of a limited number of Thoroughbred racehorses, preliminary guideline values for movement asymmetry screening for forelimb lameness (>|14.5 mm|) and hindlimb lameness (>|7.5 mm|) are higher than previously reported clinical thresholds of >|7 mm| for head movement and >|4 mm| for pelvic movement asymmetry.