Detecting Pain in Horses via Facial Expressions

Many people enjoy horseback riding and equestrian sports, but sometimes these activities cause pain, not pleasure, for the animal. This is particularly true when a horse has musculoskeletal pain, which can be difficult to detect until the horse shows significant lameness and poor performance.

Early detection of musculoskeletal pain is critical for horses’ welfare and the successful treatment of their lesions so that they can perform at their best. Sue Dyson, MA, VetMB, PhD, DEO, FRCVS, head of clinical orthopaedics at the Centre for Equine Studies of the Animal Health Trust in the United Kingdom, and her colleagues recognized the need to develop an ethogram to describe facial expressions in horses being ridden in order to identify pain. Results of this initial study were published in November 2016.¹

RELATED:

• Welfare Quality and Stress in Working, Breeding Horses
• WVC 2017: Pain Control for the Aging Horse

Dr. Dyson and her team then conducted a second study, published in March 2017, to evaluate whether ridden horses’ expressions could reveal musculoskeletal pain.² Their aim was to ascertain whether the facial expression ethogram they developed could be adapted to a pain scoring system for ridden horses to differentiate lame and nonlame animals.

Developing the Ethogram

According to Dr. Dyson, owners may misconstrue many of a horse’s facial expressions and actions as being behavioral rather than pain related. “For many years I’ve observed that horses that are uncomfortable show differences in their facial expressions, including when they are being ridden,” she said. She also has noticed that owners are often slow to realize when their horses are in pain. “They tend to attribute behavior such as bucking or kicking out after a jump to naughtiness or stubbornness, when it’s quite possi- bly a manifestation of pain.”

Using previous publications and still photographs of 150 lame and nonlame ridden horses, Dr. Dyson and her team created an ethogram and an associated training manual.¹ To determine whether a broad range of people could apply the ethogram consistently to recognize facial expressions in horses, they selected 13 people from diverse backgrounds, including veterinarians, veterinary technicians, horse owners, equine studies graduates, equine technicians, an equine veterinary nurse, and a British Horse Society instructor, to be assessors. After being trained through the manual and lecture, the assessors used the ethogram to evaluate still lateral photographs of 27 horses’ heads during training. The following features were evaluated:

• Eyes: open, semi-closed, or closed; normal expression or intense stare; eye shape; and whether the sclera is exposed
• Ear position: both ears erect and parallel, both erect and to the side, 1 ear forward and 1 back, or both ears back
• Mouth: closed or open, tongue position, and muzzle shape
• Head position: relative to the vertical

The horses’ features were graded using “yes,” “no,” or (if impossible to determine) “cannot see.” Based on some anomalies in the assessments, the researchers adjusted the ethogram.

After additional training, the assessors blindly evaluated 30 images of the heads of horses that were either lame or not lame. The assessors’ mean inter-rater agreement was 87%, and there was no difference in scoring related to professional background. Although the mean percentage of overall agreement was 80%, there was a large standard deviation due to inconsistency in assessment of the horses’ eyes and muzzle.

The researchers concluded that the facial expression ethogram specific for ridden horses (FEReq) could be applied accurately by people with different backgrounds in the horse industry, including owners, to describe a horse’s facial expressions while being ridden.

Testing the Ethogram

The researchers had 2 objectives for the second study: first, to assess differences in FEReq scores misconstrued between still photographs of the by owners as being heads of lame and nonlame horses behavioral rather than while they were being ridden; second, to use observations of facial pain related. expressions using the ethogram to develop a pain score for each anatomic area. The study’s analyst, an equine veterinarian with additional training in equine behavior, used the FEReq and training manual to blindly evaluate 519 photographs taken while the horses were ridden in a trot and canter.

The photographs of 76 lame horses and 25 nonlame horses included “before” and “after” images of 7 horses that had undergone diagnostic analgesia to eliminate their lameness. A pain score was applied to each facial marker of the ethogram, including head position and appearance of the horse’s ears, eyes, nostrils, mouth, lips, and tongue. A total pain score was determined for each still photograph, and an overall total pain score was then calculated for each horse.

Pain scores were significantly higher in the lame horses than in the nonlame group. In addition, scores for total pain, head position, and ear appearance were significantly decreased in the lame horses after analgesia was provided. The best indicators of pain for a ridden horse were twisting the head, being above or severely above the bit, ear position (both ears back, 1 ear forward and 1 backward, 1 ear to the side, or 1 backward), open mouth with exposed teeth, the bit pulled to 1 side, and certain eye features (partially or fully closed, tension caudal to the eye, intense stare, or exposed sclera). Nonlame horses were more likely to be on the bit, with ears held forward, eyes open, and lips slightly separated.

This study demonstrated that a trained equine veterinarian could determine the status of a horse, either lame or not lame, by analyzing photos using the FEReq combined with the pain score. Significant differences between lame and nonlame horses included eye position and facial expression, the presence of orbital tension in the muscles dorsal and caudal to the eye, and whether the horse’s eyes were open or closed. However, the features of the ethogram should be considered in conjunction with the pain score, not just a single observation of 1 aspect of the horse’s head while the horse is being ridden.

The authors also stated that they “need to determine whether analysis of still images, video recordings, or live horses using these limited parameters could accurately distinguish between sound and lame horses.” In addition, the authors noted that the “results apply to schooling-type work in the Olympic equestrian disciplines and are not necessarily transferable to other equestrian sports in which different head and neck positions are preferred. However, it seems unlikely that different head and neck positions should induce alterations in ear posture or the appearance of the eyes unless there was associated discomfort.”

Looking Ahead

The next phase of the study was to develop a whole-horse ridden ethogram based on assessment of video recordings of lame and nonlame horses. This incorporates the facial expressions as well as other aspects of the horse’s behavior.³ Significant differences were observed between nonlame and lame horses, with unwillingness to go, crookedness, hurrying, changing gait spontaneously, poor-quality canter, resisting, and stumbling and toe dragging occurring more frequently in lame horses in addition to changes in facial expression and head and neck posture. The next phase will investigate changes in whole-horse behavior before and after musculoskeletal pain has been abolished by diagnostic analge-sia. The horse is clearly trying to communicate, according to Dr. Dyson, and we as veterinarians need to be aware of this.

Dr. Bentz is a 1997 graduate of the University of Pennsylvania School of Veterinary Medicine. She is co-founder of Academic Veterinary Solutions, serves as equine chair for the Pennsylvania Veterinary Medical Association's Scienti c Program Committee, teaches at University of Pennsylvania and Manor College, and is a consultant for Veterinary Information Network.

References:

• Mullard J, Berger JM, Ellis AD, Dyson S. Dvelopment of an ethogram to describe facial expresions in ridden horses (FEReq). J Vet Behav Clin Appl Res. 2017(Mar-Apr);18:7-12. doi: 10.1016/j.jveb.2016.11.005
• Dyson S, Berger J, Ellis AD, Mullard J. Can the presence of musculoskeletal pain be determined from the facial expressions of ridden horses (FEReq)? J Vet Behav Clin Appl Res. 2017;19(May-June):78-89. doi: 10.1016/j.jveb.2017.03.005
• Dyson S, Berger J, Ellis A, Mullard, J. Development of an ethogram for a pain scoring system in ridden horses and its application to determine the presence of musculoskeletal pain. J Vet Behav Clin Appl Res. 2018;23(Jan-Feb):47-57. doi:10.1016/j.jveb.2017.10.008