Interobserver Reliability of the Animal Welfare Indicators Welfare Assessment Protocol for Horses

On-Farm Welfare Assessment of Horses: The Risks of Putting the Cart before the Horse

“…In conclusion, the results of both studies based on the application of the AWIN protocol are quite congruent: They indicate a low prevalence of behavioral problems and good health conditions, especially the second study [48]. The first study confirms the prevalence of obesity of horses in developed countries.…”

“…Another study, with the primary aim to test inter-observer reliability (see below) took place in Germany on 10 farms and included a total of 435 horses of various ages, sexes and breeds, although they were mostly warmbloods [48]. Breeding, leisure and sport horses were involved, and housing varied from single stall to groups housed outdoors.…”

“…While hair coat assessment has been validated for goats, a "rough coat" being associated with thinness and lower quality in hair mineral assays [55], it has, to our knowledge, never been validated (i.e., clearly correlated with physiological/health or behavioral measures) for horses, even though Pritchard et al [24] and Czycholl et al [48] found some inter-observer reliability. This measure presents a specific problem for horses as horses that are allowed time for free exercise outdoors may not have "shiny glossy hair" when back in their stall, as horses readily roll in mud when possible.…”

“…Although restricted to single-housed horses, extension for assessment of group-housed horses is underway. Czycholl et al's [48] study, with two observers assessing the same horses at the same time, showed that inter-observer reliability was overall good, especially when using a kappa agreement coefficient corrected for both bias and distribution of data, the Prevalence Adjusted Bias Adjusted Kappa, PABAK. Kappa calculations yielded scores between 0.35 and 1 (ex: stall size), the highest scores being logically for resource-based data (e.g., stall size and exercise).…”

“…The same is true for QBA. Moreover, although studies report correlations between the results of QBA and some behavioral tests, the question of the validity of QBA remains under question [48,53], as for all measures entirely based on subjective assessments. As mentioned above, being playful does not necessarily mean being happy for an adult horse [13], and human representations of the significance of behaviors may be influenced by a variety of factors, such as culture, access to a reference population and personal experience (e.g., [6]).…”