Practice Overview Of Canine Health (POOCH)
18 years ago
7243 views
Posted
3rd March, 2006 00h00
An epidemiological investigation of ill health in dogs and any temporal association with vaccination
D.S. Edwards, W.E. Henley, E.R. Ely, J.L.N. Wood
Following a magazine survey, one anti-vaccination campaign group claimed that ten percent of dogs suffered illness within three months of vaccination, implying that canine vaccination was responsible for this ill health due to impairment of the immune system.
An epidemiological investigation, entitled practice overview of canine health (POOCH), was conducted to determine whether a temporal association between canine vaccination and ill health in dogs exists. The study population consisted of dogs that had used the services of a veterinary practice within the previous twelve months. Practices were randomly selected from a national list and individual dogs were randomly selected from the databases of participating practices.
Postal questionnaires were sent by the veterinary practices, with a practice covering letter, a letter explaining the study and a reply paid envelope. One questionnaire, with the name of a specific dog, was sent per household. The questions regarded any signs of ill health in the dog within the two-weeks preceding completion of the questionnaire; its vaccination history and questions about the dog, its household and any flea or worming treatments administered. Owners who had not responded to the questionnaire after three weeks were sent a reminder card. The last vaccination dates for participating dogs were also obtained from the practices. The study was designed to have 80% power to detect a two-fold increase in signs of ill-health for dogs vaccinated within the last three months, compared to dogs vaccinated prior to this.
The data from completed questionnaires were checked for errors before being recorded in a customised MS Access database, following which it was examined using standard approaches for categorical data analysis. Multivariable logistic regression modelling was used to adjust estimates of the effect of recent vaccination on risk of ill-health for potential confounders. Variables were retained in models if they significantly reduced model deviance (likelihood ratio c 2 statistic P <0.05). A form of non-linear mixed effects modelling, called generalized logistic regression, was used to determine whether clustering of illness in dogs occurred at the practice or breed level. Equivalence between the risks of ill-health amongst dogs vaccinated recently and those not vaccinated recently was tested by use of the Hauck-Anderson procedure.
Twenty-eight practices participated in the study and 3,966 useable questionnaires were received and entered onto a computerised database. Over 19% of dogs had signs of ill health within two weeks of questionnaire completion with the frequency of specific signs of illness reported ranging from 0.5 to 54.5 per 1,000 dogs. The dogs ' ages ranged from two weeks to 23 years (median six years four months). Twenty three percent of dogs were recently vaccinated (<3mo). The percentage of ill health amongst dogs recently vaccinated (<3mo) was 16.4% compared with 18.8% of dogs that had not received any vaccination for over three months.
Vaccination within the previous three months was not significantly associated with ill-health at the univariable level. No statistical evidence of clustering at the veterinary practice or breed level was detected in the generalised logistic regression modelling. Logistic regression modelling indicated that signs of ill health within two weeks of questionnaire completion significantly increased with age.
Following adjustment for age in multivariable analyses, recent vaccination was not found to be significantly associated with ill health (P = 0.2) Multivariable analyses adjusted for age also showed that the number of prior vaccinations was not associated with signs of ill health. T he lack of significant association between recent vaccination or number of vaccinations and ill-health was not confounded by other variables. Further multivariable analyses found no significant overall temporal associations between signs of ill-health and recent vaccination within preceding time intervals of 1 through to12 months. Analyses using the Hauck-Anderson procedure indicated that recently vaccinated dogs had similar levels of ill health to dogs not recently vaccinated. Within the acceptance limits of +/- 5%, recently vaccinated (<3mo) and non-recently vaccinated groups were equivalent.
The POOCH study was designed to provide the necessary quantitative data with sufficient power to test the hypothesis that vaccination is temporally associated with canine ill-health. Questionnaire responders were blinded to the study hypothesis in order to ensure the validity of the study. Health in vaccinated and un-vaccinated animals was not compared because of the difficulty in identifying a comparable population of un-vaccinated dogs: Veterinary practices only see small numbers of unvaccinated dogs and those that are seen are likely to be exposed to different environmental risk factors than vaccinated dogs, as the decision to vaccinate is likely, for example, to be influenced by the socio-economic status of the owner. Using other vaccinated dogs as controls and looking for a temporal association between vaccination and ill-health should have reduced the effect of unmeasured confounders.
More from
