When a dog develops diabetes, what happens next? How likely is it that cataracts follow? If a dog has hip dysplasia, what are the chances osteoarthritis develops down the line? These questions matter to every veterinarian building a monitoring plan and every owner trying to stay ahead of their dog's health.

A study published in PLOS Computational Biology in August 2025 provides the first data-driven answers at population scale. Drawing on owner-reported health data from 26,614 dogs enrolled in the Dog Aging Project, researchers at the Fred Hutchinson Cancer Center built comorbidity networks that map which conditions tend to appear together — and in some cases, which ones come first.

What the study did

The researchers, led by Antoinette Fang and Jing Ma, extracted owner-reported health condition data from the Dog Aging Project, a long-term longitudinal cohort study of companion dogs across the United States. They included conditions reported in at least 60 dogs, yielding 160 distinct health conditions. Only dogs with at least one reported condition were included (n = 26,614).

The team constructed two types of networks:

1. Undirected comorbidity networks using a Poisson binomial test, adjusting for age, sex, sterilization status, breed background (purebred vs. mixed-breed), and weight. These networks show which conditions co-occur more often than expected by chance.

2. Directed comorbidity networks that accounted for the reported timing of condition onset. These reveal potential temporal sequences — Condition A before Condition B.

Networks were further stratified by age group (young, adult, senior), allowing the researchers to see whether comorbidity patterns shift as dogs age.

Key findings from the comorbidity network

The study identified 269 statistically significant condition pairs after adjusting for multiple testing and confounders. Several findings confirmed well-established veterinary knowledge, while others surfaced less-studied relationships.

Diabetes and cataracts: the strongest cascade

The directed network showed that diabetes mellitus tends to be reported before cataracts — consistent with the known pathophysiology of diabetic cataracts in dogs, where sustained hyperglycemia causes osmotic lens swelling and opacification. The undirected network also confirmed a strong association between diabetes and blindness, reflecting the downstream visual consequence when cataracts go untreated.

This finding is not new clinically, but seeing it quantified at this scale reinforces the importance of early diabetic monitoring. For veterinarians managing a newly diagnosed diabetic dog, the network data underscore why ophthalmologic monitoring should begin at diagnosis, not after the owner reports vision changes.

Hypertension and chronic kidney disease: an age-dependent signal

Hypertension and chronic kidney disease (CKD) were significantly associated — but primarily in the senior age stratum. This mirrors what is seen in human medicine, where the hypertension-CKD axis becomes more prominent with advancing age. In younger dogs, the pair did not reach significance, suggesting that routine blood pressure monitoring becomes particularly important as dogs enter their senior years, especially those with early renal markers like elevated SDMA or proteinuria.

Proteinuria and anemia: a novel finding

One of the less-studied associations flagged by the network was the co-occurrence of proteinuria and anemia. While this relationship has been described in human nephrology (anemia of chronic kidney disease, driven by decreased erythropoietin production), it has received limited attention in the veterinary literature. The finding suggests that clinicians finding proteinuria on urinalysis should consider concurrent assessment of packed cell volume or a complete blood count, particularly in dogs with concurrent CKD.

Ear infections as a network hub

Otitis externa (ear infections) emerged as a hub node connecting to multiple allergic skin conditions: seasonal allergies, food-related skin allergies, and atopic dermatitis. The network also linked ear infections to keratoconjunctivitis sicca (KCS, or dry eye), likely reflecting the shared immune-mediated or allergic underpinnings across mucosal surfaces.

For the practicing veterinarian, this hub structure reinforces the clinical pattern where recurrent ear infections in a dog should prompt investigation of underlying allergic disease rather than being treated as isolated events.

Orthopedic cascades: dysplasia before osteoarthritis

The directed network revealed temporal sequences consistent with clinical progression: elbow dysplasia and hip dysplasia were reported before osteoarthritis. Elbow dysplasia and hip dysplasia also co-occurred, reflecting the shared developmental orthopedic disease pathway.

Patellar luxation and tracheal collapse also co-occurred — a pairing that may reflect the brachycephalic and toy-breed predisposition to both conditions rather than a direct pathophysiological link.

Lameness and osteoarthritis formed another significant pair, as did hip dysplasia and lameness, capturing the progressive functional decline that accompanies joint disease.

Parasite co-infections

Hookworms and roundworms were significantly associated, as were roundworms and tapeworms. Coccidia and Giardia co-occurred. Anaplasmosis and Lyme disease — both tick-borne — showed significant co-occurrence, highlighting shared vector exposure risk.

These parasite pairs likely reflect environmental co-exposure and shared susceptibility in dogs with inadequate parasite prevention, rather than biological interaction between the organisms.

Toxin co-exposures

Chocolate ingestion and grape/raisin ingestion were significantly associated, likely reflecting dogs with indiscriminate dietary habits rather than any pharmacological interaction. Chocolate ingestion also co-occurred with human medication ingestion, consistent with households where dogs have access to both food items and medications.

Age-stratified networks: disease clusters grow denser with age

A key finding was that comorbidity networks become denser in older dogs. Young dogs' networks were relatively sparse, with few significant pairs. Adult dogs showed more connections, and senior dogs showed the densest networks with the most interconnections.

This mirrors what is seen in human geroscience: multimorbidity is fundamentally an aging phenomenon. As dogs accumulate conditions over time, previously independent disease processes begin to interact, creating feedback loops that accelerate decline.

The practical implication is that monitoring intensity should increase with age — not just for individual conditions, but for the clusters of conditions that tend to co-occur. A senior dog diagnosed with one condition should be proactively screened for its common network neighbors.

What this means for veterinary practice

The comorbidity network approach offers several takeaways for clinicians:

Screen for network neighbors, not just the presenting complaint. When a dog is diagnosed with diabetes, the network says: start monitoring for cataracts. When a dog develops proteinuria, check for anemia. When a dog has recurrent ear infections, investigate allergic disease.

Adjust monitoring by age. Senior dogs with CKD should have blood pressure assessed regularly. Young dogs with hip dysplasia should begin joint health monitoring before clinical osteoarthritis appears.

Use parasite co-infection data to guide prevention counseling. The strong co-occurrence of hookworms/roundworms and anaplasmosis/Lyme disease reinforces the importance of broad-spectrum parasite prevention rather than targeting individual organisms.

Recognize hub conditions. Ear infections, dental disease, and skin allergies function as hub nodes connected to multiple other conditions. Addressing these early and thoroughly may have downstream benefits beyond the immediate complaint.

Limitations to keep in mind

The study relies on owner-reported health conditions, which introduces reporting bias. Owners may underreport conditions they do not recognize or conditions their veterinarian has not discussed with them. The data also represent owner reports from a specific time window rather than longitudinal clinical diagnoses confirmed by veterinary examination.

Additionally, the comorbidity network identifies associations, not causation. A significant co-occurrence does not mean one condition causes the other — it may reflect shared risk factors, shared breed predisposition, or the general accumulation of disease with age.

The dataset reflects the Dog Aging Project cohort, which may not be fully representative of the entire U.S. dog population. Dogs whose owners enroll them in a longitudinal health study may differ systematically from the general pet population.

Why this research matters beyond dogs

The companion dog shares environments, exposures, and many age-related disease pathways with humans. Mapping canine comorbidity networks provides a translational window into human multimorbidity — a growing challenge as human populations age. The diabetes-cataract cascade, the hypertension-CKD axis, and the age-dependent densification of disease networks are all patterns visible in both species, reinforcing the dog's role as a model organism for aging research.

The researchers note that because pet dogs share homes, environments, and many age-related diseases with humans, mapping how canine illnesses cluster and cascade offers a window into the same multimorbidity processes that affect people — and may point to earlier detection and prevention strategies for both species.

Sources

• Fang A, Kumar L, Creevy KE, Promislow DEL, Ma J, Dog Aging Project Consortium. Constructing the first comorbidity networks in companion dogs in the Dog Aging Project. PLOS Computational Biology. 2025;21(8):e1012728. https://doi.org/10.1371/journal.pcbi.1012728
• Fang A et al. Source code and data: https://github.com/aeyfang/pbc-dap-network
• PubMed entry: https://pubmed.ncbi.nlm.nih.gov/40811246
• Dog Aging Project: https://dogagingproject.org