Observer variation in visual assessment of forelimb horseshoe characteristics on thoroughbred racehorses

Objective -

To assess the accuracy and reliability of a visual method of evaluating horseshoe characteristics. Animals - 1,199 Thoroughbred racehorses.

Characteristics of 1 forelimb horseshoe were visually assessed on horses immediately prior to racing by 5 field observers at 5 major racetracks in California. Characteristics evaluated included horseshoe type; toe grab height; and the presence of a rim, pad, and heel traction devices. Sensitivity and specificity for observer assessment of horseshoe characteristics were calculated by comparing observer assessments to a postmortem laboratory standard for horses that died within 48 hours of a race. Intraobserver agreement was assessed in a subset of horses by comparing horseshoe observations made before and after the horse's race. Interobserver agreement was evaluated by comparing horseshoe assessment among observers who examined the same subset of horses prior to racing on select days. Results - The sensitivity and specificity of this visual method of evaluating horseshoe characteristics were good and ranged from 0.75 to 1 and 0.67 to 1, respectively. Agreement beyond chance (weighted kappa values) between observers and the laboratory standard for toe grab height was fair (0.60 to 0.62). Intraobserver and interobserver agreements (kappa values) were high (0.86 to 0.99 and 0.71 to 1, respectively).

Visual observation of horseshoes can be a feasible and reproducible method for assessing horseshoe characteristics prospectively in a large cohort of horses under racing conditions.