Learning confounds algometric assessment of mechanical thresholds in normal dogs

Abstract

Objective: To perform algometric readings in normal dogs in a design that would assess possible confounding factors.

Study design: Prospective study.

Animals: Skeletally mature spayed female, intact male and castrated male retriever or retriever mix dogs without orthopedic or neurologic disease (n = 19).

Methods: Twelve common surgical sites were selected for algometric pressure testing. Threshold response was defined as a conscious recognition of the stimulus, and recorded in Newtons. Sites were tested in the same order, and the testing sequence repeated 3 times on each side of the dog. Dogs were tested in the morning and evening of the same day and was repeated 10-14 days later, allowing 4 separate data collections for each dog.

Results: Data were analyzed using ANOVA or ANCOVA. When all the data were included in the analysis, dog (P < .0001), order (P < .0001), site (P < .0001), site order (P = .0217), time (P < .0001), day (P < .0001) and repetition (P < .0001) all significantly affected the algometer readings. When only the first reading for each site was included in the analysis, dog (P < .0001), site (P < .0001) and sex (P < .0001) all significantly affected algometer readings.

Conclusion: These results suggest that learning occurred over repeated collection time points, with dogs anticipating the stimulus and reacting at lower thresholds.