Seizure‐precipitating factors in dogs with idiopathic epilepsy

Dear Editor,

1.

With great interest, I have read the article of Dr. Forsgård et al., concerning the clinical important topic of seizure‐precipitating factors in dogs with idiopathic epilepsy. In our referral clinic, I am confronted with dogs with idiopathic epilepsy and often I have discussions with the owners whether there are factors that can precipitate seizures. Although I am pleased to see a scientific approach to sort out this clinical uncertainty, I do have some concerns about the methodology the authors used. Hence, I also have concerns about the conclusions.

The authors have chosen a cross‐sectional design to answer the clinical question whether seizure‐precipitating factors do exist in dogs with idiopathic epilepsy. Studying causal associations between risk factors and seizures, this design is not appropriate to make inferences about causality. Temporality is an important aspect to make causality more likely. Bradford Hill described temporality (the cause always precedes the result)1 as 1 of the criteria for causality. A cross‐sectional design does not explicitly meet this criterion.

The study design the authors chose is prone to possible recall bias:2 the methods the author used to observe the exposure to risk factors is an interview according to a predefined questionnaire. Owners are more likely to remember stress‐related situations that happened shortly before seizures occurred. It is not unlikely that situations that were as stressful, but not followed by seizures, are forgotten and ignored. It is difficult to score the exposure to subjective factors, once the outcome (seizures) is already known, especially in a self‐assessment scoring system. The varying time span between precipitating factor and seizure introduces another uncertainty in making conclusions on causation. The Bradford Hill criterion of biologic plausibility1 is not met here. There is not a clear medical explanation for stress to cause seizures 1‐2 days later. Especially with varying intervals between supposed precipitating factors and seizures, the role of chance cannot be underestimated. What is the probability for a dog with idiopathic epilepsy of having a seizure 1‐2 days after a hot day, if hot weather was not a precipitating factor, so by chance alone? If an owner recognizes 9 different seizure‐precipitating factors, and if in 78% of the dogs only sporadic seizures had a precipitating factor, how confident can we be about the factor really being precipitating? With the questionnaire method used, the seizures are declared precipitated by the highly biased owner, seeking for explanations and patterns. This is demonstrated by the big difference between the open‐ended question and the reported factors from the checklist.

Another reason for survey studies not being suitable for causal research is the absence of a control group, which is unexposed to the determinants under study.2 The researchers tried to answer the question how often the precipitating factors do occur in dogs with idiopathic epilepsy, without being followed by seizures. They formed a small control group consisting of dogs with epilepsy, without precipitants (or owners who did not acknowledge the existence of factors that precipitate seizures?). Whether this is a valid control group is questionable, because of possible information bias introduced by the self‐assessment questionnaire/interview approach.

Based on clinical experience, I tend to believe that precipitating factors do exist, because I recognize the described factors in the stories that owners of dogs with idiopathic epilepsy tell me. But as this personal experience is extremely prone to bias, I need unbiased causal research to rely on. Only if there was evidence for causality, the conclusions that seizure‐precipitating factors are common and avoiding these factors could achieve better treatment outcomes would be proper. After reading the study of Dr. Forsgård et al., I am still left with the clinical uncertainty about seizure‐precipitating factors.