What each number is, and what it is not
Veterinary research reports several different quantities and they are not interchangeable. We store the metric type with every figure and show it on the page, rather than converting everything into one number that looks tidy.
- One-year period prevalence The share of dogs seen for a condition during a year of primary veterinary care. From VetCompass clinical records.
- Cross-sectional owner-reported prevalence The share of living surveyed dogs reported to have a condition. Undercounts what owners do not notice.
- Screening registry proportion The share of dogs submitted to OFA that evaluated abnormal. Submission is voluntary and skewed toward dogs expected to pass, and a radiographic finding is not the same as clinical disease.
- Median survival Median age at death in a measured population.
Why we no longer publish a single lifetime health cost
An earlier version of this site converted all of the above into a modelled lifetime risk and multiplied it by a treatment price to produce one expected health figure per breed. An external review in August 2026 found that this was not defensible, and we agreed. Converting a screening proportion or an owner survey into lifetime risk claims something the studies did not measure, and multiplying a radiographic finding by a surgical price assumes a link that has not been demonstrated.
We now publish two things separately. The cost of owning the dog, where every input is sourced and the components genuinely add together. And the documented health risks, each shown in the units the study used, with the treatment price stated where a matching cost source exists. A condition with no matching cost source is left unpriced rather than given a plausible-looking number.
What we exclude from totals
Organ-system groupings such as "skin disease" are shown for context but never added alongside the specific conditions they contain, because that would count the same disease twice. All-dog population rates are shown as context and never presented as breed figures. Age-of-onset ranges are display aids, are not sourced, and feed no calculation.
Geography
The strongest health datasets are British, because no US equivalent to VetCompass exists. Cost data is American. Every page says which is which. A French Bulldog's airway is shaped the same in Ohio as in Oxfordshire, but where breeding lines genuinely differ we say so.
Corrections
Every figure is re-verified against a committed copy of its source by a script that fails the build if anything disagrees. If you have a source that contradicts something here, send it.