Red Alert 3: The null hypothesis in epilepsy

A suggestion? The null hypothesis for people working in the field of epilepsy should be that "there is no epilepsy". The doctor is therefore required to disprove the null hypothesis. Yes, even for people with a diagnosis of epilepsy. This applies to every consultation. Any epileptologist worth their salt does this all the time, I'm sure.

If there is any doubt*, it is far better not to treat someone for epilepsy and to wait until further data (history, EEG recordings, videos or video EEG) removes any doubt. Sometimes this requires that everyone should be patient and await further seizures to provide an unambiguous diagnosis. Any harm that accrues from waiting is massively exceeded by the harms from treating, for argument's sake, one in 20 patients incorrectly with antiseizure medications. Think carefully about that; one in 20 patients means that if you are correct in 95% of your diagnoses of epilepsy, you are exposing one in 20 of your patients to unnecessary, prolonged treatment with antiseizure medication. This is inevitably harmful; consider the impact of the diagnosis of epilepsy on the person, the stigma of the condition, the "stickiness" of the diagnosis, the potential cognitive, teratogenic and other side-effects of antiseizure medications and the cumulative direct costs of antiseizure medications, consultations, blood tests, CT and MRI scans, emergency unit visits and the indirect costs. Hence, even if one was correctly "95% certain" of the diagnosis, which in medicine generally is high certainty, in the case of people with seizures there are substantial untoward consequences. Hopefully you will recognise the value of being frugal with the diagnosis of epilepsy and the value of not being afraid to wait for further information, some of which may only accrue with the passage of time. In the face of diagnostic uncertainty, get more history and/or before war EEG recordings, where you must be frugal with the interpretation of apiculate waves as representing inter-ictal epileptiform discharges. If there continues to be doubt, video-EEG is the best way to establish a confident diagnosis.

It is generally a very bad idea to use a trial of therapy as a means of setting uncertainty about the diagnosis. Consider the following:

1. The seizures stop; is this a direct antiseizure effect, a placebo response (which includes the natural history the), coincidental or does the patient have epilepsy? Many patients with functional seizures initially respond to antiseizure medications. There are several other conditions which respond to antiseizure medications, including migraine and bipolar mood disorders. Hence, the strategy of the trial of therapy is generally diagnostically useless. 

2. Seizures diminish in frequency; is this a treatment effect, the placebo response or just the natural variability of the frequency of the epileptic or non-epileptic seizures?

3. The seizures continue as before; are these epileptic seizures that are drug resistant or does this indicate an alternative diagnosis, such as functional seizures?

As you may see from the above, a trial of therapy to establish the diagnosis is a bad strategy. I have never seen a study assessing its usefulness, but I would aver that this strategy has an unacceptably poor sensitivity and specificity.

*As you probably recognise, there is no such thing as diagnostic certainty. One really needs to ask yourself, "what, in percentage terms, is the probability that I am incorrect"? In doing so, we are our own greatest enemies as we tend to be overconfident. So, subtract at least a few percent. Then work on a strategy of arriving at a satisfactorily confident diagnosis. Remember that your cellphone is a very good diagnostic aid; phone as many people who have witnessed the patient's seizures as possible. Don't forget, waiting and asking family members, friends and colleagues to video as many seizures as possible is a good diagnostic strategy.