Established care
Dietary therapies
Ketogenic diet, modified Atkins, MCT, low glycaemic index treatment, gluten-free casein-free, GAPS, elimination diets
One of these is an established antiseizure treatment with randomised evidence and a place in guidelines. The others are widely used in autism with much weaker support. The word “diet” covers both, which is why they are set out separately here.
Either a medicines regulator has licensed it for this use, or clinical practice guidelines recommend it on the strength of controlled trials. This is the standard everything else on the scale is measured against — and for most children, the treatments at this level are the ones that will actually change their day.
Where it stands, condition by condition
The same therapy can be well supported for one problem and completely untested for another. This is the single most common place families are misled.
| Condition | Evidence | What that means here |
|---|---|---|
| GLUT1 transporter deficiency, pyruvate dehydrogenase deficiency | Established care | Here the ketogenic diet is not an adjunct but the treatment itself: it supplies the brain with the fuel it cannot otherwise obtain. This is also why the diagnosis matters — the condition is treatable and will not be found without testing for it. |
| Drug-resistant epilepsy in children | Established care | Randomised evidence supports the ketogenic diet as an option after antiseizure medicines have failed, with roughly a third to a half of children achieving a meaningful reduction in seizures. It is recommended in international guidance and delivered by a specialist team. |
| Specific epilepsy syndromes — infantile spasms, Dravet, myoclonic-atonic epilepsy, Lennox-Gastaut, FIRES | Established care | These are the syndromes in which dietary therapy is most often considered early rather than late, and in several of them response rates are higher than in drug-resistant epilepsy generally. |
| Gluten-free, casein-free diet for autistic features | Not supported by evidence | Systematic reviews have not found adequate evidence of benefit, and the better-designed double-blind studies with controlled food challenges were negative. The separate and important exception is coeliac disease, which must be diagnosed and treated if it is present. |
| GAPS diet | Not supported by evidence | No clinical trial evidence of any kind. It is among the most restrictive regimens offered to children, and in a child who already eats a narrow range the nutritional risk is substantial. |
| Elimination and few-foods diets in ADHD | In clinical trials | Some children do respond, and controlled studies of restricted few-foods diets report an effect in a subgroup. The burden on the family is high and dietetic supervision is needed, so this is a considered trial rather than a default. |
| Removing food additives or sugar to change behaviour | Early research only | Effects reported in controlled studies are small and inconsistent. Sugar in particular has been tested repeatedly in blinded conditions without supporting the belief that it causes hyperactivity. |
What it is
The ketogenic diet is a medically supervised, high-fat, low-carbohydrate regimen that shifts the body's fuel from glucose to ketone bodies. It exists in several forms of decreasing strictness — classical ketogenic, medium-chain triglyceride, modified Atkins and low glycaemic index treatment — which allows the burden to be matched to the family.
It is a prescribed treatment, not a way of eating. It is initiated and monitored by a team including a specialist dietitian, with baseline investigations, defined follow-up and planned discontinuation.
Exclusion diets — gluten-free casein-free, GAPS, various elimination regimens — remove categories of food on the reasoning that they contribute to behaviour or development. They are usually begun at home without dietetic input, and that is where most of the risk sits.
The distinction matters because the two are not comparable. One is an antiseizure treatment with randomised evidence behind it; the others are dietary changes with much weaker support and real nutritional consequences in a child who already eats selectively.
How it is meant to work
Ketone bodies provide an alternative fuel for the brain and alter neuronal metabolism, neurotransmitter balance and excitability. Several mechanisms have been proposed and none fully explains the effect — a situation similar to cannabidiol, where the clinical result arrived before the explanation.
In GLUT1 deficiency the reasoning is direct and complete: glucose cannot cross into the brain in sufficient quantity, and ketones bypass the defective transporter.
For exclusion diets in autism the proposed mechanism is that incompletely digested peptides from gluten and casein cross the gut and affect the brain. This hypothesis has been investigated and has not been supported.
What has actually been tested
- The ketogenic diet is supported by randomised controlled trials in children with drug-resistant epilepsy, with meaningful seizure reduction in a substantial minority, and it appears in international consensus guidance on when and how to use it.
- Response is often apparent within weeks, which allows a defined trial period with a clear decision point rather than open-ended continuation.
- For gluten-free casein-free diets in autism, systematic reviews conclude that the evidence is insufficient. Studies using blinded food challenges, where neither family nor assessor knew whether the child had received the excluded food, did not find the effects reported in unblinded settings.
- That gap between unblinded and blinded results is informative rather than dismissive: it indicates how much of the observed change comes from expectation and from the other things that alter when a family restructures a child's diet.
- The GAPS diet has no trial evidence. It is not that the trials were negative; they have not been done.
What we still do not know
- Which child will respond to the ketogenic diet — no reliable predictor exists outside the specific metabolic conditions where it is the definitive treatment.
- Optimal duration, and how best to withdraw it in a child who has responded.
- Whether the less strict variants are equivalent to the classical diet across all syndromes.
- Whether a defined subgroup of autistic children genuinely responds to exclusion diets, which would require blinded challenge studies in that subgroup to answer.
Risks and unwanted effects
- The ketogenic diet has a defined and manageable adverse-effect profile that requires monitoring: constipation, growth faltering, raised lipids, kidney stones, acidosis, vitamin and mineral deficiency, and carnitine depletion. These are reasons for supervision, not reasons to avoid the treatment.
- Restrictive diets in children who already eat selectively are the main concern on this page. Many autistic children have a narrow accepted food range. Removing further categories can produce inadequate intake, micronutrient deficiency and, in documented cases, serious nutritional illness.
- Calcium and vitamin D fall predictably when dairy is removed without replacement, and bone health in children with motor disability is already a concern.
- The social effect is underestimated: a child on a restrictive diet cannot eat what other children eat at school or at a birthday, and that matters.
- A diet begun without a dietitian and without a review date tends to continue indefinitely whether or not it helped.
Questions to ask before you agree
A centre that is doing good work will welcome these questions and answer them in writing.
- If the ketogenic diet is proposed: which variant, which team, and which baseline investigations and monitoring schedule?
- How long is the trial period, what are we measuring, and at what date do we decide?
- If an exclusion diet is proposed: what exactly are we expecting to change, and how will we tell whether it did?
- Who is checking my child's growth, intake and micronutrients while the diet is in place?
- Has coeliac disease been excluded properly, rather than assumed?
More in this section
Early detection and early intervention
The single highest-value thing in this whole field, and the one least often discussed in clinics selling treatment. Cerebral palsy can be identified in the first months of life, and that is when therapy does the most.
Established careGoal-directed, task-specific training
Not a brand of therapy but the principle underneath the ones that work: the child practises the actual task they are trying to learn, with enough repetition to matter.
Established careEarly behavioural and developmental intervention
Early structured support for an autistic child is recommended care. Which kind, how many hours, and towards what goal are all still genuinely argued over — and the arguing is the useful part.
Established careConstraint-induced therapy and bimanual training
For a child who uses one hand much more than the other, these are among the best-evidenced interventions in paediatric neurology — and both work, for different things.
Established carePhysiotherapy and strength training
Well supported when it is built around function and genuinely loaded — much weaker when it is passive stretching and generic exercise.
Established careOccupational therapy
The discipline that works on what the child actually does all day — dressing, eating, writing, playing, getting through a school morning.
Established careSpeech, language and communication
For a child who cannot speak clearly, giving them a way to communicate is among the most consequential interventions available — and the fear that it will stop them talking is unfounded.
Established careSpasticity and tone management
Medical and surgical treatment of tone, which works best when it is tied to a functional goal rather than to a number on a tone scale.
In clinical trialsTechnology-assisted rehabilitation
Useful mainly as a way of delivering more practice, not as a treatment in its own right — and that distinction decides whether it is worth the money.
In clinical trialsNamed physiotherapy approaches
Families are usually asked to choose between brands. The evidence says the brand matters far less than whether the session is active, goal-directed and repeated often enough.
In clinical trialsSensory therapies
Two things are sold under one word. Manualised Ayres Sensory Integration has been tested in trials; the sensory add-ons sold around it mostly have not.
Established careVision and cerebral visual impairment
The most commonly missed problem in children with brain injury. A child who cannot interpret what they see is often described as inattentive, uncooperative or more delayed than they are.