Three mailed questionnaire surveys were conducted in 1997-1998. The first was sent to practising veterinarians in Finland. The second survey was sent to breeders and the third to owners of the same sixteen breeds (Table 1). All surveys were basically identical in order to enable comparisons between the groups, but some minor adjustments were made, mostly in phrasing and terminology, so that the questionnaire was understandable to each group. The veterinary survey was sent to all 680 veterinarians categorized in the membership files of the Finnish Veterinary Association as practicing. The breeds included in the study were chosen according to their participation and success versus failure in health programmes (Table 2). Each breed's success or failure was evaluated in epidemiological studies (IV, V). For the breeder survey, ten breeders with the highest number of puppy registrations for the previous five years and ten others were included in each breed. For the owner survey twenty dog owners who had purchased a dog in 1997 and twenty who had purchased one in 1993 were randomly chosen in each breed.
Table 1. Number of sent questionnaries and recieved answers in surveys to Finnish veterinarians (Study I), dog breeders (Study II) and dog owners (Study III).
Table 2. Breeds included in the survey of Finnish dog breeders' and dog owners' attitudes about control programmes for canine genetic diseases and each breed's inclusion criteria.
Ophthalmological examination records of 18 146 dogs that were born in 1988-1996 and officially screened before 1 January 1997 were included in the study. Only those breeds with more than 700 registered puppies during the previous five years and official screening programmes started before that were chosen. For Dobermanns and collies another dataset was also included, consisting of all the dogs born in years 1988-1997 that were officially examined before 12 December 1997. This data gave scorings for PHTVL in Dobermanns and CEA in collies.
This study included hip dysplasia records for 64 349 dogs in 22 breeds. Nineteen of these breeds have had official control programmes for at least five years and the number of registered puppies in each breed has exceeded 700 during the last five years. Rottweilers were also included in the study as a model for a breed with active voluntary screening. On the other hand Saint Bernard Dogs were included as a model of a breed that is commonly thought to be characterized by severe hip dysplasia and whose breeders commonly oppose breeding restrictions.
Costs of ophthalmological examination were calculated using estimated lowest, average and highest costs per dog. Costs of finding affected dogs were calculated separately.
For hip dysplasia the average screening cost per dog was estimated to be FIM450 ($80), including veterinary fees and the Kennel Club's screening fees. In the calculation models an average puppy price of FIM3600 ($635) and an average veterinary fee of FIM500 ($90) were used. To calculate the benefits, the change in the number of dysplastic dogs and those having severe dysplasia (grades D or E) was estimated on the basis of discovered change in dysplasia prevalence between the dogs that were screened before 1988 and those that were born in 1988-1995. In those breeds with increased dysplasia prevalence the number of saved animals was thus negative, leading to negative values in calculations of benefits in relation to costs. All calculations were made within a breed. Models 1 and 2 are based on the assumption that all dysplastic versus severe dysplastic dogs are replaced immediately (Swenson et al., 1997A). In the other models it is assumed that a certain number of dysplastic dogs are replaced immediately and others need additional veterinary care and medication because of hip dysplasia, resulting in extra costs (Table 3).
In those breeds with significant changes in dysplasia prevalence, a calculation was made of the number of animals that should be saved from hip dysplasia in order to reach a benefit/ cost ratio of 1 with Model 1. In order to be able to better compare the breeds the difference between the actual and expected number of saved animals as a ratio of the number of examined animals in every breed was also calculated.
Hip dysplasia records of 10 706 German Shepherds from the Finnish Kennel Club's data bases were used in order to estimate the heritability of hip dysplasia and environmental effects and to predict breeding values and to estimate genetic change achieved with the breeding programme.
Statistical analyses were performed using Statistix TM Version 4.1. Analytical Software, Tallahassee, Fl., USA. In Studies VI - VII the programme package PEST (Groeneveld, 1990), including a variance component estimation programme REML VCE4 (Groeneveld, 1997), was used.
In Studies I-III The Wilcoxon signed rank -test was used to analyze the differences of the importance of the various sources of information. The same test was used for comparison of the frequency of different reasons to euthanize young dogs (Studies I-II). The Friedman two-way analysis of variance by ranks followed by multiple comparison tests was used for ranking the importance of different health actions and diseases. In addition, for the comparison of different key groups' opinions of the level of knowledge between each group, the Friedman analysis was followed by multiple comparison tests. Differences between each subgroup were compared with the Mann-Whitney-u or the Kruskal-Wallis tests (Studies II-III).
In Study IV the 2-test was used to calculate the significance of the number of CEA- or PHTVL-affected animals in the beginning and at the end of the study period and the differences between the various groups of pooled data for Dobermanns and collies.
In Study V 2-test was used to analyze the within-breed heterogenity of the study groups for hip dysplasia. Fisher's exact test was used for comparisons between breeds with or without breeding restrictions.
In Studies VI and VII fixed effects were first tested with a logistic regression model. Variance and covariance components for hip dysplasia were estimated by applying the Restricted Maximum Likelihood (REML) method (Patterson and Thompson, 1971). The BLUP (best linear unbiased prediction) procedure (Henderson, 1984) with the PEST programme (Groeneveld, 1990) was used to estimate the fixed effects and breeding values. The heterogenity between the groups with different parental HD scores in different age groups was tested with the 2-test.