In 2022, Winchester published the results of a study led by Professor Andrew Knight, which claimed the headline: "Vegan diets may be the healthiest to feed pet dogs, say researchers"1,2.
For the study, 2,639 dog owners were invited to report their own opinion of seven indicators of their dog's health.
The researchers also looked at the prevalence of 22 specific health disorders, reported as being based on 'predictions by vet assessments', finding 49% prevalence in traditional meat diets, 43% for raw meat diets and 36% for vegan diets.
However, it was not clear in the report what "predictions by vet assessments" actually means. Whose predictions? The vets'? Or the owners'?
Closer examination of the original study reveals it was "owners' predictions of vet assessments".
Owner-reported opinion, otherwise known as anecdote, is of course notoriously unreliable, and if there is one thing more unreliable than owner-reported opinion, it would surely be owner-reported opinion of their vet's opinion.
Furthermore, the original study was of an observational design, and therefore could only suggest correlation, not causation.
Lastly, the original analysis only looked at the relationship between diet and owner observations/opinion, without considering the impact of other factors assessed by the study, such as the age of the dog and whether the owner was a vegan.
For the new study3, therefore, researchers at the University of Liverpool led by Professor Alex German re-analysed the data used for the original study, in particular correcting the data for confounding variables.
When analysed in this way, significant illness in dogs correlated most strongly with the age of the dog, whether it was on medication, and the frequency of visit to the vet.
Compared to these, the association between owner opinions of health and feeding vegan dog food was minimal.
Co-author of the second study, Richard Barrett-Jolley, Professor of Neuropharmacology, said: “We know how seriously owners take their pet’s health.
"By revisiting and further interrogating these data, we have been able to draw more nuanced insights. “Crucially, we cannot draw a firm conclusion as to what diet type is actually best for dogs; this was never possible given the nature of the original dataset and study design.
"However, we can conclude that variables other than dog diet are more strongly associated with owner opinions about the health of their dog.”
References
PS: Whilst you're here, take a moment to see our latest job opportunities for vets.