Exploratory factor analysis of signalment and conformational measurements in Thoroughbred horses with and without recurrent laryngeal neuropathy.

Background: Conflicting results have been reported for risk factors for recurrent laryngeal neuropathy (RLN) based on resting endoscopic evaluation and comparison of single conformation traits, with many traits correlated to one another. Objectives: To simplify identification of signalment and conformation traits (i.e. variables) associated with RLN cases and controls diagnosed with exercising overground endoscopy (OGE) using exploratory factor analysis (EFA). Study design: Prospective cohort. Methods: Pearson's rank correlation was used to establish significance and association between variables collected from n = 188 Thoroughbreds from one stable by observers blinded to OGE results. Exploratory factor analysis was conducted on nine variables for cases and controls; common elements between variables developed a factor, with variables grouped into three factors for cases and controls respectively. Correlation (loading) between each variable and factor was calculated to rank relationships between variables and cases/controls, with factors retrospectively named based on their underlying correlations with variables. Results: Numerous inter-correlations were present between variables. Most strongly correlated in cases were wither height with body weight (r = 0.70) and ventral neck length (r = 0.68) and in controls body weight with rostral neck circumference (r = 0.58). Wither height (r = 0.61) significantly loaded the top-ranked factor for cases ('height(RLN)'), explaining 25% of conformational variance. Ventral neck length (r = 0.69) and age (r = 0.57) significantly loaded the second-ranked factor for cases ('neck length(RLN)'), explaining 16% of conformational variance. Rostral neck circumference (r = 0.86) and body weight (r = 0.6) significantly loaded the top-ranked factor for controls ('body size(CON)'), explaining 19% of the variance. Wither height (r = 0.84) significantly loaded the second-ranked factor for controls ('height(CON)'), explaining 13% of the variance. Main limitations: Horses had not reached skeletal maturity. Conclusions: Exploratory factor analysis allowed weightings to be determined for each variable. Wither height was the predominant conformational feature associated with RLN. Exploratory factor analysis confirms aggregated conformational differences exist between RLN cases and controls, suitable for future evaluations.