Which aspects of the everyday behavior of older dogs correlate with performance on a visuo-spatial memory test and the canine cognitive dysfunction rating scale (CCDR)?

The behavioral aspects of aging in the domestic dog have primarily been investigated through owner-reported scales that measure specific behavioral signs of aging, and laboratory-based memory and cognitive tests. We need to know more about how aging affects everyday behavior and functionality in owned dogs. This study tested a methodology for identifying patterns in everyday physical and behavioral function associated with behavioral tests and scales. Fifty-seven family dogs aged 8+ years were included. Owners completed the Canine-Cognitive-Dysfunction-Rating-Scale (CCDR), and each dog underwent a visuo-spatial memory test (VSMT) and a physical examination. Owners also completed a 112-item checklist of everyday behaviors and activities that a normally functioning dog should be able to perform. Feature selection was performed using a series of orthogonal projections to latent structures (OPLS) models that sequentially excluded checklist items with low variable importance to projection until R²Y and Q² were optimized. All OPLS models were strong and significant (R²Y up to 0.579, <i>p</i> < 0.0001), and resulted in three short (8-11 item) subscales from the checklist whose scores correlated with their relevant test counterpart: CCDR score (r = -0.63, <i>p</i> < 0.0001), VSMT score (r = 0.658, <i>p</i> < 0.0001), and physical health score (r = -0.649, <i>p</i> < 0.0001). The content of the subscales provided valuable insights into the everyday behavioral correlates of the tests. For example, a pattern of items describing mood, motivation, mobility, vision, memory and trainability was associated with better VSMT performance, but better performance in the CCDR scale was associated with items relating to mobility, exercise tolerance, vision, and hearing. This indicates that in older dogs a substantial proportion of the variability of the results of tests like the VSMT and CCDR, can be accounted for by physical and sensory health issues, and that patterns of everyday behavior are correlated with these tests. We also created an OPLS model for age, as a reference point for comparison. Our results indicate that using multivariate statistics to perform feature selection can identify systematic relationships between everyday behavior and tests and clinical scales, to provide valuable insights into the real-world effects of brain aging.