Ranking bovine pain-related behaviors using a logistic regression algorithm

• Published in: Applied animal behaviour science Vol. 271; p. 106163
• Main Authors: Trindade, Pedro Henrique Esteves, da Silva, Gustavo Venâncio, de Oliveira, Flávia Augusta, Luna, Stelio Pacca Loureira
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.02.2024
• Subjects: algorithms; analgesia; animal behavior; Animal welfare; appetite; Behavior; castration; Cattle; data collection; farms; pain; Pain assessment; Quality of life; regression analysis; sheep; Animal welfare; Pain assessment; Behavior; Cattle; Quality of life
• ISSN: 0168-1591
• DOI: 10.1016/j.applanim.2024.106163

Pain identification in farm animals is a welfare challenge worldwide. In the last decade, our research team validated the Unesp-Botucatu Cattle Acute Pain Scale (UCAPS) based on specific pain or discomfort behaviors and alterations to maintenance behaviors when the cattle experiences pain. Recently, for other specific-species pain scales, such as sheep, pigs, and horses, we have used algorithms to weigh the pain-related behaviors, ranking their heterogeneous importance. Because the ranking of cattle pain-related behaviors' importance is still unknown, we aimed to investigate whether UCAPS pain-related behaviors have different statistical weights and rank their importance. We used a dataset comprising behavioral scores from 40 bulls in the perioperative period subjected to surgical castration in our previous study. Bulls were filmed between 16 and 24 hours before castration, between 1 and 4 hours after castration, 1 hour after analgesia, and 24 hours after the end of the surgery. Three experienced observers assessed all videos blind to the timepoint. After watching each video, the observers decided whether they felt analgesia should be indicated for the individual (expert assessment) and score UCAPS behaviors. Each pain-related behavior was weighted using a multilevel binomial logistic regression. Among the 17 UCAPS behaviors, four were not significant (‘Activity 1: restless, moves more than normal or lies down and stands up with frequency’, ‘Appetite 1: hyporexia’, ‘Wagging tail’, and ‘Licking wound’), suggesting minor importance. The algorithm highlighted ‘Kicking’, ‘Hind limbs extended’, ‘Extends neck’, ‘Head down’, and ‘Interaction 2’ (apathetic: may be isolated or may not accompany the other animals; does not react to tactile, visual and/or audible environmental stimuli) respectively, as the five most important pain-related behaviors within this sample. Statistical weighting might contribute to the acute pain scoring approach and selecting behaviors in the psychometric validation process for pain scales. In conclusion, the statistical weighting based on multilevel binomial logistic regression raised evidence for a ranking of the importance of UCAPS pain-related behaviors in determining the need for analgesia in bulls after surgical castration. These findings contribute to a better understanding of bulls' behavioral responses to acute pain, potentially improving pain identification and monitoring, and enhancing welfare. •The algorithm evidenced different importance levels of behaviors used to monitor pain.•Behaviors related to pain, discomfort or maintenance were the five most important ones.•Statistical weighting could be used for validation of the bovine pain scale.•Kicking, extending hind limbs or neck, and head down were the most important behaviors.