Created by Dr. Kelly M. Cassidy, 2007
Date last edited: 26 December 2007
Longevities based on the Vet School Data (Patronek 1997) were always less than longevities based on questionnaires
Vet School Data longevities were, on average, 60% (± 11%) of questionnaire longevities (Range 38% to 80%). (Based on a comparison of longevities of 35 breeds for which Patronek et al., 1997 had breed data and for which there was a combined sample size of more than 20 from questionnaire data. Longevities from questionnaires were based on weighted averages of all questionnaires from all countries. ) For example, the median lifespan of Boxers in the vet school study was 6.0 yrs (N=455), while the weighted average of three questionnaire studies for Boxers was 10.0 (Total N from 3 questionnaires, 295)
The Vet School Data is a biased sample. Dogs referred to vet schools often have uncommon conditions or conditions for which treatment is too specialized for most veterinary practices. Vet schools are also less likely to see dogs euthanized for "old age" and no vet practice sees dogs that die of "old age" at home (unless the dog is necropsied).
Questionnaires may also be biased. Owners may have faulty memories. Breed fanciers wanting to present "their" breed in the best light may mentally make excuses for dogs that die young and omit those dogs from surveys. If a person new to the breed got a dog that died young, that person may have gotten discouraged and moved on to another breed and been less likely to contribute to a breed survey than a person with better luck with the breed. Nonetheless, the questionnaires probably give a more accurate picture of a breed's lifespan than vet school studies.
If the ratio of longevities from Vet School data to questionnaire data had a low standard deviation, then a breed's longevity obtained from questionnaire data could be predicted with reasonable accuracy from the more convenient and much larger Vet School database. Unfortunately, there is such a large range of differences between Vet School and questionnaire longevities, that the vet school database is not very useful for ballpark estimates of longevities likely to be obtained by a questionnaire.
There may be some insights into breed health to be gleaned by knowing the relative lifespan of a breed based on vet school data vs questionnaire data, but I cannot see what it might be.
Data Set of vet school data vs questionnaire data
Longevity in different countries, from questionnaires
(Update note. Added Bichon Frise August 18, 2007, to bring breed total
Note: The addition of breeds beyond 20 or so caused essentially no change in results, so no new breeds will be added after Flat-coated Retriever to this analysis.)
There are two UK questionnaires, the Kennel Club (UK) survey and Michell's (1999) British Owner survey. For the USA and Canada, there is no multi-breed questionnaire survey, but there are dozen of individual breed surveys, usually conducted by breed clubs. Most of the breed club surveys included dogs in countries other than the USA or Canada, but USA and Canadian dogs comprised the vast majority of dogs in almost all of the surveys. Most of the American breed club surveys included 1%-5% Canadian dogs, with a smattering of other countries (usually less than 5%).
I compared longevities of breeds for which the sample size for both the UK and USA/Canada surveys were at least 20 dogs. For the UK, the weighted mean (weighted by sample size) of the (UK) KC and British Owner survey data was used for each breed. In almost all cases, the (UK) KC sample size was larger than the British Owner survey sample size. See the Breed Data page for a break-down of the two data sources.
There were 23 breeds that met the criteria of a total sample size of >20 for both regions. On average, longevities for breeds in the UK were about a year (0.8 ± 1.7 yrs) longer than longevities for the same breeds obtained from USA/Canada breed surveys.
Three breeds had a substantially higher longevity in the USA/Canada data, ironically the English Cocker Spaniel (12.1 yrs in the USA/Canada compared to 11.3 in the UK), the Airedale Terrier (11.8 yrs in the USA/Canada compared to 10.8 yrs in the UK), and the Australian Shepherd (12.4 yrs in the USA/Canada compared to 9.0 yrs in the UK). The Australian Shepherd had the lowest UK sample size (22), barely meeting my inclusion criteria of at least 22, so may not be a good estimate.
Three breeds (Field Spaniel, Italian Greyhound, and Manchester Terrier) had longevities of more than 3 years longer in the UK survey than in the USA/Canada surveys.
It is not clear whether the differences in survey longevities for the two areas represent real differences in breed lifespans or represent differences in survey methodologies or characteristics of the owners who responded to the surveys. If the survey differences are real, the reasons would be of great interest to dog owners.
Data set of longevities for breeds in UK vs USA surveys