I haven't written (much) about epilepsy until now because it's so, well, uncontroversial. But it's come to my attention that a lot of people who need to know about it, don't. So I'll start with a bit of history.
Around 400 BC, the physician Hippocrates (of Hippocratic Oath fame) wrote a book called On the Sacred Disease, about epilepsy. In it, he identifies the disease as a brain disorder rather than a demonic infestation, and recommends fasting as a treatment. Some centuries later, as recounted in the Bible, a man brought an epileptic child to Jesus for an exorcism, His disciples having failed. The child is miraculously cured on the spot, but Jesus takes a moment to point out that "this kind goeth not out but by prayer and fasting."
And so it goes until the 1920s, when the effectiveness of this treatment, combined with its long-term impracticality, led to a great deal of study of the biology of starvation, in hopes of discovering a treatment that could yield similar benefits in the long run. Enter the ketogenic diet, for a while. After 1939 and the development of phenytoin, the diet fell into disfavor until roughly the 1990s, which saw a resurgence of research on the matter that has continued unabated to this day.
Instead of linking to individual studies to back up my claim that dietary treatment is a mainstream approach to epilepsy, and especially pediatric epilepsy, I'll link to a whole pile of them and let you poke around. They all say more or less the same thing: the ketogenic diet is a very effective, but imperfect treatment for intractable epilepsy. Why the emphasis on "intractable"? I've yet to find a study that attempts the dietary therapy except as a last resort, after numerous drug therapies have failed. This is a problem.
First, I can't help but be cynical about the pharmaceutical profit motive behind the universality of this approach. But more importantly, and someone will correct me if I'm making an unjustified logical leap here, but the success rates must be artificially low, because it is only ever tried against the most difficult cases - those for which no other treatment works. And still it succeeds.
That's not the only reason to suspect that the (already impressive) success rates ought to be higher than those reported. For example, if the duration of the disease is a factor in its tenacity, then allowing it to persist long enough to vet half a dozen drug therapies puts the diet (and indeed, the patient) at an unfair disadvantage when it's kept on the bench until all else has failed. The possibility that duration plays a role is basically speculative, but weakly supported by the observation that the diet is somewhat less effective in treating adult epilepsy. Additionally, we can't afford to forget that these studies are generally published by epilepsy specialists, whose benefactors obviously include the manufacturers of antiepileptic drugs. So there exists a clear funding bias in favor of downplaying the efficacy of nonpharmaceutical treatments.
It may even be true - this is pure speculation but can't currently be ruled out - that drug therapies act on the disease or on the brain in such a way as to make it less susceptible to the effects of the diet. But enough of the weak arguments: I shall now break out the big guns and explain why the (amazing) success rates are as low as they are.
First, most of the so-called ketogenic diets administered range from 5:1 to 3:1 or even 2:1, meaning that 17% to as high as 33% of calories are from carbohydrates, assuming no patients are sneaking the occasional cupcake. Even on the best of these diets it will be a challenge to achieve and maintain a state of ketosis, and the rest don't deserve to be called ketogenic. If there are any carbohydrates at all in bacon or ribeyes, that's how many I consume on a typical day, and my only motivation is to stay generally healthy. I would think that an epilepsy patient, his parents and his doctor have a much stronger incentive to adhere as rigidly as possible to the treatment presented as the last best hope, but these studies suggest otherwise.
A better, but still imperfect, approach has been to use the Atkins diet, which at least in its initial stages is a ketogenic diet, and one with a more favorable ratio than any of the above. Observe the table on the second page of this paper on treating epilepsy with the Atkins diet, and notice the relationship between success and level of ketosis. Also noteworthy here are the relationships between success and age, and success and the number of antiepileptic drugs tried before attempting dietary treatment.
Yet another factor confounding the success rates is that many studies fail to even use what I would call food to administer the diet, then express disappointment and confusion at the dropout rate. Along with the other causes I've listed to be skeptical of the failure rate, this drives home a suspicion, at least to me, that a properly administered dietary therapy stands an excellent chance of reducing seizures in childhood epilepsy by 100%.
What would this therapy look like? It would be extremely low in carbohydrates, closer to 20:1 than 2:1. It would be administered at home by well-informed parents with some oversight from a doctor. It would consist of actual food - meat, eggs, dairy products etc. - and not powdered protein shakes. These well-informed parents might even lend support, if the child is old enough to justify it, by going on the diet themselves and removing any nonconforming foods from the house. They'd be surprised by their own results, unless they read this blog regularly.