To establish evidence for the validity and reliability of three commonly used pain
scales in dogs when assessed by video by specialists in anesthesia.
Mixed-method test–retest observational study.
A group of six American College of Veterinary Anesthesia and Analgesia board-certified
specialists and 31 postoperative dogs.
The evaluators scored 31 dogs using a visual analogue scale (VAS), numeric rating
scale (NRS), and Glasgow pain scale (GPS). The evaluators individually scored the
dogs using all three scales together and subsequently, at 3 month intervals, using
each of the scales apart. Then, all evaluators in one room reviewed 23 of the videos.
A camera was positioned for video and audio recording of discussion about the videos.
Intra- and interobserver reliability was determined using a two-way random model intra-class
correlation coefficient (ICC).
Linear regression indicated a strong correlation among all scales when assigned together
(VAS versus NRS, p < 0.0001, R2 = 0.93; VAS versus GPS, p < 0.0001, R2 = 0.59; and NRS versus GPS, p < 0.0001, R2 = 0.61) and apart (VAS versus NRS, p < 0.0001, R2 = 0.68; VAS versus GPS, p < 0.0001, R2 = 0.40; and NRS versus GPS, p < 0.0001, R2 = 0.47). Posture, appearance, vocalization, stiffness, interaction between the animal
and a person and response to palpation were identified as important variables for
assessing pain. Intra-observer reliability produced average ICC values of 0.90 for
VAS, 0.89 for NRS and 0.85 for GPS. Interobserver reliability produced average ICC
values when scores were assigned together (VAS: 0.93, NRS: 0.93 and GPS: 0.93) and
when done separately (VAS: 0.91, NRS: 0.93 and GPS: 0.95).
Conclusions and clinical relevance
The preferred use of the VAS and NRS over the use of the GPS should be cautiously
considered for research applications when experts are observers. Revisions of the
GPS to clarify descriptors and remove or modify items that may not be associated with
pain in dogs should be considered.