Dog Disease Networks Reveal Age-Linked Comorbidity Patterns Relevant to Human Health
Scientists mapped 160 canine diseases across 26,614 dogs, uncovering how conditions cluster together—and how these patterns intensify with age.
Summary
Researchers from the Dog Aging Project built the first large-scale comorbidity networks in companion dogs, mapping statistical associations among 160 health conditions in over 26,000 dogs. Using a Poisson binomial test adjusted for age, sex, sterilization status, breed background, and weight, the team identified well-known disease pairings—such as diabetes with cataracts and hypertension with chronic kidney disease—alongside less-studied links like proteinuria and anemia. A directed network incorporating reported diagnosis timing revealed likely disease sequences, including diabetes preceding cataracts and dry-eye disease leading to corneal ulcers. Age-stratified analysis showed disease networks became denser and more centralized in senior dogs, mirroring comorbidity patterns seen in aging humans. These findings advance veterinary informatics and suggest companion dogs are a valuable real-world model for studying human aging and multimorbidity.
Detailed Summary
Comorbidity—the co-occurrence of two or more health conditions—tends to increase with age and is a central concern in geroscience. Yet most comorbidity network research has focused on humans, with limited, small-scale work in animal models. Companion dogs are a compelling exception: they share living environments and many diseases with humans, receive sophisticated veterinary care, and age in ways that closely mirror human aging. Understanding canine comorbidity networks could therefore yield insights applicable to both species.
This study leveraged the Dog Aging Project (DAP), a large longitudinal U.S. cohort study, using cross-sectional health data from its first annual survey. The final analytical sample included 26,614 dogs reporting at least one of 160 health conditions (those occurring in at least 60 dogs). The cohort was approximately balanced by sex and reproductive status, with ~46% spayed females and ~46% neutered males. A Poisson binomial test was used to assess whether condition co-occurrences exceeded chance, adjusting for age, sex, sterilization status, breed background (purebred vs. mixed), and body weight—covariates known to influence disease risk in dogs.
The undirected comorbidity network confirmed several well-established disease associations: diabetes co-occurring with cataracts and blindness, and hypertension co-occurring with chronic kidney disease (CKD). Importantly, the network also surfaced less-studied associations, such as proteinuria with anemia, suggesting potential new directions for veterinary research. A directed comorbidity network—constructed using owner-reported dates of condition onset—added temporal resolution, supporting known clinical sequences: diabetes preceding cataracts, elbow/hip dysplasia before osteoarthritis, and keratoconjunctivitis sicca (dry eye) preceding corneal ulcers.
Age-stratified analysis divided dogs into Young Adult, Mature Adult, and Senior groups. Global network centrality measures—reflecting how interconnected and hub-dominated the disease network is—increased monotonically with age and were highest in the Senior group. Critically, the hypertension–CKD association only emerged in the Senior subgroup, underscoring how some comorbidities are age-gated and may be missed without life-stage-specific analysis. This mirrors findings in human aging research, where multimorbidity accelerates in later life.
The study demonstrates that large-scale owner-reported veterinary data, when rigorously analyzed with appropriate covariate adjustment, can generate clinically meaningful and statistically robust comorbidity maps. These networks provide a foundation for improved canine healthcare management, evidence-based veterinary practice, and translational aging research. Limitations include the cross-sectional nature of the baseline data, reliance on owner-reported diagnoses (which may introduce recall or ascertainment bias), and the absence of clinical confirmation for many conditions. Future longitudinal DAP waves will enable more rigorous causal inference about disease sequencing.
Key Findings
- Diabetes–cataract and hypertension–CKD comorbidities confirmed in 26,614 dogs using network analysis.
- Proteinuria–anemia association identified as a novel, less-studied canine comorbidity link.
- Directed network shows diabetes precedes cataracts and dry eye precedes corneal ulcer temporally.
- Disease networks grow denser and more centralized as dogs age; hypertension–CKD only appears in seniors.
- Covariate-adjusted Poisson binomial test provides robust framework for large-scale comorbidity mapping.
Methodology
Cross-sectional owner-reported health data from 26,614 DAP dogs with at least one of 160 conditions (minimum prevalence n=60) were analyzed. A Poisson binomial test assessed pairwise co-occurrence significance, adjusting for age, sex, sterilization status, breed background, and weight. A directed network was constructed using owner-reported diagnosis onset dates to infer temporal disease sequences.
Study Limitations
The baseline DAP data are cross-sectional, limiting causal inference despite the directed network analysis. Health conditions are owner-reported without clinical verification, introducing potential recall and ascertainment bias. The cohort is skewed toward sterilized dogs in the U.S., which may limit generalizability to intact or internationally diverse canine populations.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: