Rater Agreement on Gait Assessment during Neurologic Examination of Horses

Background Reproducible and accurate recognition of presence and severity of ataxia in horses with neurologic disease is important when establishing a diagnosis, assessing response to treatment, and making recommendations that might influence rider safety or a decision for euthanasia. Objectives To determine the reproducibility and validity of the gait assessment component in the neurologic examination of horses. Animals Twenty-five horses referred to the Royal Veterinary College Equine Referral Hospital for neurological assessment (n=15), purchased (without a history of gait abnormalities) for an unrelated study (n=5), or donated because of perceived ataxia (n=5). Methods Utilizing a prospective study design; a group of board-certified medicine (n=2) and surgery (n=2) clinicians and residents (n=2) assessed components of the equine neurologic examination (live and video recorded) and assigned individual and overall neurologic gait deficit grades (0-4). Inter-rater agreement and assessment-reassessment reliability were quantified using intraclass correlation coefficients (ICC). Results The ICCs of the selected components of the neurologic examination ranged from 0 to 0.69. Backing up and recognition of mistakes over obstacle were the only components with an ICC>0.6. Assessment-reassessment agreement was poor to fair. The agreement on gait grading was good overall (ICC=0.74), but poor for grades <= 1 (ICC=0.08) and fair for ataxia grades >= 2 (ICC=0.43). Clinicians with prior knowledge of a possible gait abnormality were more likely to assign a grade higher than the median grade. Conclusion and Clinical Importance Clinicians should be aware of poor agreement even between skilled observers of equine gait abnormalities, especially when the clinical signs are subtle.