Syrian Hamster Grimace Scale picture generation and data collection.

This procedure was adapted from Rat Grimace Scale (RGS): The Manual.⁴⁵ Images of hamster facial expressions were analyzed from 9 animals that underwent laparotomy, with each serving as its own control. By using the PlayMemories Home software (Sony) Save Frame feature, the images were generated from the 5-min V-BC videos taken at 1300 the day before surgery (day 2) and at 1300 on the day of surgery (day 3; approximately 6 h after the procedure). An observer (AE) who was blind to which day the videos were taken, watched blinded videos and captured still images at 1-min intervals when at least one side of the face was visible and resolution was optimal. Ninety images were generated, comprising 5 ‘baseline’ images from day 2 and 5 ‘postsurgery’ images from day 3 for each animal. The images were cropped to exclude the body, were brightened, and were presented to scorers in a digital slide presentation in a blind and random fashion by the observer.

The facial features, termed ‘action units’ (AU), evaluated were: orbital tightening, whisker change, ear changes, and nose–cheek flattening. The scale was derived after initial observation of generated images and used elements as described in grimace scales developed for rats and mice.^22,45 Orbital tightening, whisker change, and ear changes were adapted from the Mouse Grimace Scale, and nose–cheek flattening was adapted from the Rat Grimace Scale. Normal hamsters had wide-open eyes, gently sloping whiskers, a distinct crease between their cheek and nose bulge, and slightly forward-facing ears. For orbital tightening, the palpebral fissure was narrowed, and hamsters appeared to be squinting. For whisker change, the whiskers stood out straight from the cheeks and appeared to clump together. For ear changes, the ears were folded and appeared narrower from the side and generally pointed caudally; as seen from the front, the ears were less visible or had tips that point laterally. For nose–cheek flattening, the separation (crease) between the nose and cheek area was less distinct, and the whole side of the face appeared to be flattened.

Four scorers were trained in SHGS and then evaluated hamster images. The scorers were 3 veterinary technicians (BG, AS, KF) with experience assessing animals for pain and 1 board-certified laboratory animal veterinarian (JH), all of whom are female. They were trained on the AU and score severity in a 30-min session using representative photos and received 3 photos for practice scoring. During the scoring sessions, scorers were given representative photos and written descriptions of the severity of each AU. Scorers rated the presence and intensity of each AU on a scale of 0 to 2 for practice images and the 90 experimental images. A score of 0 meant that the AU was not present, a score of 1 denoted that the AU was moderately visible, and a score of 2 was that the AU was pronounced. The total SHGS score for an image was the average of the intensity ratings for each AU. For each image, scorers assessed the animal as painful or not painful (pain–no pain). The reliabilities of the SHGS and the individual AU were determined via calculation of intraclass correlation coefficient (ICC), which quantifies the agreement among the scorers.⁴⁴