Prevalence and study design
The present study estimated a prevalence of epilepsy of 9.5%. As very few studies have investigated the prevalence of epilepsy in canine populations, and because prevalence may vary greatly from breed to breed due to variable genetic background, data for comparisons are not available.
A prevalence of epilepsy about 1.0% has been reported from university-based populations (not breed-specific) [21, 22]. Compared to this, the prevalence estimated at 9.5% in the present study appears to be high. A study by Famula and Oberbauer (2000), showed that 21% of 938 Belgian Tervueren registered in the American Kennel Club had epilepsy . However, in 26 of these dogs only one seizure was observed, and thus this subpopulation did not fulfil the criteria of epilepsy used in the present study as defined by the International League Against Epilepsy (ILAE): "epilepsy is a condition characterized by recurrent epileptic seizures, numbering two or more" .
One might argue that owners that have a dog with epilepsy are more inclined to participate in a questionnaire survey than owners without a dog with epilepsy. If we assume that none of the dogs among the non-responders were positive, then a prevalence of 3.9% (49 dogs out of 1,248) would have been the obtained. This would imply that we have overestimated the prevalence. However, this hypothesis is counteracted by the fact that during the study, we contacted ten dog owners, who did not respond to the mailed questionnaire due to the inconvenience of filling it out and bringing it to the mailbox, even though they had a positive attitude towards the investigation. In these dogs, the investigators could confirm a diagnosis of epilepsy in all ten cases. Had they fulfilled the inclusion criteria of participating in the prevalence study by returning the mailed questionnaire, the prevalence estimate reported would have increased to 11.1%. The ten dogs provide evidence that there may be more hidden epileptic individuals in the population, whose owner did not return the mailed questionnaire due to either inconvenience or unwillingness to communicate the presence of epilepsy in their dog.
Moreover, we did not contact the owners who stated that their dogs did not have epilepsy.
Also adding to our belief that the calculated prevalence is an underestimate is the finding that in the present study, the mean age of onset of epilepsy was found to be 3.3 years. A number of dogs younger than three years may very well have been too young for having expressed the disease at the time of the investigation.
Both living and dead dogs with epilepsy were included when calculating prevalence. Excluding dead dogs would pose a risk of excluding a group of dogs with a higher prevalence of epilepsy (that might have been euthanized motivated by the severity of the epileptic condition) than the rest of the population, thus creating selection bias. Regarding recall bias, it was judged that owners would recall the rather serious signs of epilepsy even if they occurred up to 10 years back.
The mailed questionnaire was designed to detect as many possibly epilepsy-positive individuals as possible with the attendant risk of producing preliminary false-positives. A broader definition is used to identify possible cases in the study population (similar to a test with a high sensitivity but a low specificity).
Double-checking the positive responders (self-reported epilepsy) from the mailed questionnaire investigation by conducting a telephone follow-up interview emphasized the importance of validating the responses given by laymen (corresponding to a test with a high specificity ruling out the false-positive cases). Had this not been done, the study would have reported 64 epilepsy positive dogs instead of 49, the false positives leading to overestimated prevalence.
The return rate of the questionnaire was 42% which is acceptable in terms of the reliability of the calculated prevalence risk, but contains the limitation that we have described: who is willing to respond and who definitely does not want to respond.
In clinical studies the case definition is of uttermost importance for the reliability of the results presented. In the present study, the study design was chosen with the specific purpose to search for epilepsy positive individuals in the total population over ten years, as we wanted to determine the prevalence of epilepsy and investigate the nature of the disease in the breed in a defined time period. By nature, such a study design (including both living and dead dogs) does not allow the investigators the possibility of performing a standardized clinical evaluation of the dogs, and therefore the diagnosis of epilepsy must be based on seizure phenomenology and history as reported by the owners.
It is well known, that in epilepsy the diagnosis is to a great extend based upon the seizure history reported by the owner, and in both human and canine epileptic patients, the seizure history often holds the key to the diagnosis [20, 23]. As described by the International League Against Epilepsy (ILAE), the clinical manifestation of epilepsy consists of a sudden and transitory abnormal phenomenon which may include alterations of consciousness, motor, sensory, autonomic or psychic events, perceived by the patient or an observer . Seizure phenomenology and history is increasingly being used for diagnosing epilepsy in humans . The validity of the diagnosis of epilepsy in the dogs presented in this study is supported by the fact that in 71% of the dogs, the general practitioner had independently of our investigation established a clinical diagnosis of epilepsy, and in the remaining dogs we did not find any indications of disease other than epilepsy when performing a thorough and extensive interview.
Seizure characteristics and risk factors
The present study found that focal seizures, with or without secondary generalization were the most common type of seizures. Our findings are in accordance with previous studies of seizure distribution in dogs with epilepsy [4, 11, 22, 24]. It should be stressed, that an unusually high number of individuals (25%) in this study experienced focal seizures without secondary generalization (focal seizure phenomenology not followed by convulsions), and therefore special attention should be paid to signs of focal seizure phenomenology in this breed in order to avoid that a diagnosis of epilepsy in some individuals is missed [25, 26].
The Belgian shepherds are very active dogs that are easily "wand up" and in general react strongly to stressful situations involving anxiety or a high degree of excitement. In eleven dogs (22%), stressful situations would often act as an external trigger of seizures. Stress is a known trigger of seizures in humans with epilepsy and has also been reported in Labrador retrievers .
Unfortunately, information about the risk factors among dogs without epilepsy was not available, since the owners of these dogs were not contacted. This makes it impossible to identify risk factors for epilepsy.
A slight predilection for epilepsy has been reported for male dogs [6, 9]. However, no indication of this was found in our study where around 2/3 of the cases were females. Other studies have found no difference between the sexes [8, 11, 12, 28].
We found a highly significant risk of being euthanized because of epilepsy in intact animals as opposed to neutered dogs (P = 0.007). This result may indicate that a sparing effect of neutralization on seizure frequency and seizure severity exists. This could be explained by the fact that sex hormones influence seizure threshold. The convulsant/anticonvulsant effects of estrogens and progesterone, respectively, are demonstrated in women with catamenial epilepsy (seizure clustering around the time of menses) [29, 30]. There is an increase in seizure frequency, immediately preceding the menstrual period, which correlates to a decrease in progesterone level, and an additional increase in seizure frequency immediately preceding the time of ovulation, correlating with a high estrogen level with no simultaneous increase in progesterone. At the end of ovulation, simultaneously with an increase in the progesterone level, the seizure frequency decreases . Also, neutering may help to decrease a high level of hyperactivity/aggression/stress and thereby act as a protector against potential seizure provocation.
Finally, 5.2% of the overall study population was euthanized because of aggression showing that epilepsy is not the only grave problem experienced by the Groenendael and Tervueren variants.