Optimal clinical management of children receiving dietary therapies for epilepsy: Updated recommendations of the International Ketogenic Diet Study Group

This is not a single experiment but an **expert consensus guideline**—a synthesis of all published research (randomized trials, cohort studies, case series) combined with a survey of 31 international experts from 17 centers. The authors updated the 2009 version of these guidelines to reflect a decade of new evidence.

The guideline covers **four major dietary therapies**:

**Classic ketogenic diet (KD):** High fat (typically 4:1 ratio of fat to protein+carbohydrate by weight), very low carbohydrate. Requires strict medical and dietitian supervision.

**Modified Atkins diet (MAD):** Less restrictive than classic KD. Limits carbohydrates to ~10–20 grams/day but does not restrict protein, fat, or calories. No fasting or hospital initiation required.

**Medium-chain triglyceride diet (MCT):** Uses MCT oil (from coconut/palm oil) to provide ~60% of calories, allowing more carbohydrate and protein than classic KD.

**Low glycemic index treatment (LGIT):** Limits carbohydrates to those with glycemic index <50, typically 40–60 grams/day. Least restrictive.

The guideline addresses **patient selection, contraindications, initiation methods (fasting vs. non-fasting, inpatient vs. outpatient), supplementation, side effect management, monitoring frequency, and discontinuation protocols.**

Outcome measures across the reviewed studies:

**Primary:** ≥50% reduction in seizure frequency (responder rate)

**Secondary:** Seizure freedom, quality of life, cognitive/behavioral improvements, side effect profiles, tolerability (how long patients stay on the diet)

The guideline synthesizes data from **multiple studies** spanning:

**Age range:** Infants as young as 6 weeks through adolescents (the guideline is pediatric-focused, though they note increasing adult data)

**Population:** Children with drug-resistant epilepsy (failed ≥2 antiseizure medications), including specific epilepsy syndromes (infantile spasms, Dravet syndrome, Lennox-Gastaut syndrome, tuberous sclerosis complex, myoclonic-atonic seizures, Glut1 deficiency syndrome, pyruvate dehydrogenase deficiency)

**Sample sizes across key trials:** The 4 randomized controlled trials cited had sample sizes ranging from 38 to 145 children. The consensus survey included responses from 31 experts representing centers that collectively treat thousands of children annually.

**Setting:** International—17 countries represented across 31 authors (55% from outside the United States)

The guideline does not report a single measurement protocol but synthesizes findings from studies that used:

**Seizure diaries:** Parents/caregivers recorded seizure type, frequency, and duration daily

**Responder rate:** Proportion of patients achieving ≥50% reduction in seizure frequency from baseline (typically a 4-week baseline period before diet initiation)

**Seizure freedom:** Complete cessation of seizures (typically confirmed over ≥3 months)

**Ketone levels:** Measured via blood beta-hydroxybutyrate (target 2–5 mmol/L) or urine acetoacetate (moderate to large)

**Side effect monitoring:** Standardized questionnaires and clinical assessments at each follow-up visit (typically every 1–3 months)

**Laboratory monitoring:** Blood tests for electrolytes, lipids, liver function, kidney function, carnitine, selenium, zinc, vitamin D, and complete blood count—typically at baseline, then every 3–6 months

**Growth parameters:** Weight, height, head circumference plotted on growth charts

**Survey of expert practice:** 20-question survey emailed to all 31 authors, with responses compared to the 2009 survey to identify trends

**Study design:** This is an **expert consensus guideline**—not a single experiment. The authors conducted a systematic review of published literature (primarily Class IV evidence, with some Class II and III randomized trials) and combined it with a survey of expert clinical practice.

**How the consensus was built:**

1. **Author selection:** 31 experts (25 neurologists, 6 dietitians) from 17 countries, each with at least one first- or senior-authored peer-reviewed publication on KDTs. No more than 3 from any single center.

2. **Literature review:** Authors were assigned sections based on expertise and instructed to cite peer-reviewed publications. Dr. James Wheless assigned levels of evidence using the American Academy of Neurology classification.

3. **Survey:** A 20-question survey was emailed to all 31 participants to capture group opinion on controversial topics. Results were incorporated as percentage responses.

4. **Endorsement:** The final guideline was reviewed and endorsed by the Child Neurology Society, The Charlie Foundation, and Matthew's Friends.

**What this design can and cannot prove:**

**Can prove:** Expert agreement on best practices based on available evidence. Trends in clinical management over the past decade. Identification of areas where evidence is strong vs. weak.

**Cannot prove:** Causal efficacy of any specific diet (that requires randomized trials). The guideline is only as good as the underlying studies—most are Class IV (uncontrolled case series). The survey reflects opinion, not experimental data.

**Key methodological weakness:** The guideline explicitly states that "unless mentioned, all evidence is Class IV." This means most recommendations are based on uncontrolled observational studies, not randomized trials. The survey component is opinion-based, though from highly experienced clinicians.

**Efficacy (across all KDTs):**

**~40–50% of children** achieve ≥50% seizure reduction (responder rate) across all epilepsy types

**~10–15% become seizure-free** (complete cessation)

These rates are consistent across the 4 randomized trials cited (Class II and III evidence)

**Specific conditions with higher efficacy (≥60–70% responder rate):**

Glucose transporter 1 deficiency syndrome (Glut1DS): KDT is the **treatment of choice**—not just for seizures but for the underlying metabolic defect

Pyruvate dehydrogenase deficiency (PDHD): KDT bypasses the metabolic block

Epilepsy with myoclonic-atonic seizures (Doose syndrome): ~60–70% responder rate

Infantile spasms: Especially when caused by tuberous sclerosis complex

Tuberous sclerosis complex: ~60% responder rate

Dravet syndrome: ~60–70% responder rate

Children with gastrostomy tubes: Easier to administer the diet, higher compliance

**Conditions with moderate efficacy (40–50% responder rate, similar to general KDT population):**

Lennox-Gastaut syndrome: 51% responder rate in one study

Other unspecified drug-resistant epilepsies

**When to start KDT:**

Consensus: After a mean of **2.6 failed antiseizure medications** (SD 0.9)

88% of experts believe KDT should be considered **very early** for the 7 high-efficacy conditions listed above

Previously, KDT was reserved as a "last resort" after 4+ failed medications

**Initiation methods (trends from 2009 to 2018):**

**Fasting before initiation:** Previously standard (2–3 day fast to induce ketosis). Now only **18% of centers** routinely fast (down from 40% in 2009)

**Inpatient initiation:** Previously universal. Now **52% of centers** start the diet as an outpatient (up from 0% in 2009)

**Classic KD vs. MAD:** Classic KD remains most common for infants/young children. MAD increasingly used for adolescents and adults

**Side effects (most common):**

Gastrointestinal: Constipation (most common), nausea, vomiting, diarrhea

Growth: Weight loss or poor weight gain (requires careful calorie adjustment)

Kidney stones: ~3–7% of children (can be reduced with oral potassium citrate)

Dyslipidemia: Elevated cholesterol and triglycerides (usually transient, resolves within 1–2 years)

Acidosis: Transient metabolic acidosis during initiation

Selenium deficiency: Can cause cardiomyopathy if prolonged

Bone health: Reduced bone mineral density with long-term use

**Discontinuation:**

Average duration of KDT: **2 years** (range 1–5 years)

**~20% of children** maintain seizure reduction after diet discontinuation

Weaning should be gradual (over 2–3 months) to avoid seizure rebound

If no benefit after **3 months**, the diet is typically discontinued

**Plain English translation:**

If 100 children with drug-resistant epilepsy start a ketogenic diet, **40–50 will have their seizures cut in half or more.** Of those, **10–15 will become completely seizure-free.**

For specific conditions like Glut1 deficiency or Dravet syndrome, **60–70 out of 100 children** will have their seizures cut in half or more.

This is roughly **2–3 times better** than what you'd expect from adding another antiseizure medication after 2 have already failed (where responder rates are typically 15–25%).

The effect is **not immediate**—it typically takes **2–4 weeks** to reach full ketosis and see maximum benefit, though some children respond within days.

The effect is **reversible**—if the diet is stopped, seizures typically return to baseline within days to weeks (except in ~20% who maintain improvement).

**What the authors acknowledge:**

Most evidence is Class IV (uncontrolled case series)

No head-to-head trials comparing the 4 dietary therapies

Limited data on long-term outcomes (>5 years)

Survey reflects opinion, not experimental data

Pediatric-focused; adult data not systematically reviewed

No data on cognitive/behavioral outcomes as primary endpoints in most studies

**What a critical reader would note:**

**No blinding:** KDTs cannot be blinded (the diet is obvious to families and clinicians). This introduces expectation bias—parents who invest significant effort in the diet may over-report improvement.

**Selection bias:** Children who start KDT are highly motivated families. Those who drop out early (due to difficulty or lack of effect) are often not included in final analyses.

**No placebo control:** Most studies are uncontrolled. The 4 randomized trials used "waitlist" or "continued medications" controls, not a true placebo diet.

**Industry funding:** The Charlie Foundation (a parent advocacy group) funded the consensus process. While not pharmaceutical industry, this could introduce bias toward positive recommendations.

**Sample heterogeneity:** Studies mix different epilepsy types, ages, and diet types, making it hard to know which diet works best for which child.

**Short follow-up:** Most studies report outcomes at 3–6 months. Long-term adherence and side effects are poorly documented.

**No standardized outcome definition:** Some studies use ≥50% reduction, others use ≥90% reduction or seizure freedom. This makes cross-study comparison difficult.

For someone running their own n=1 experiment (note: KDTs are medically supervised—do not attempt without a doctor and dietitian):

### What to test

**If you have drug-resistant epilepsy:** Consider the **modified Atkins diet (MAD)** as the most feasible self-experiment. It requires no fasting, no hospital stay, and no strict calorie counting.

**Dose:** Limit carbohydrates to **10–20 grams/day** (roughly 1 slice of bread worth). Eat unlimited fat (butter, oil, cream, avocado, fatty meats) and moderate protein. No sugar, grains, fruits (except small berries), or starchy vegetables.

**Alternative:** Low glycemic index treatment (LGIT)—limit carbs to 40–60 grams/day, choosing only low-GI foods (glycemic index <50). Less restrictive but possibly less effective.

### Minimum meaningful duration

**4 weeks minimum** to assess response. Full ketosis takes 2–4 weeks to achieve.

**3 months** to determine if the diet is truly effective. If no improvement by 3 months, it's unlikely to work.

**2 years** is the typical treatment duration if effective, then gradual weaning.

### What to measure

**Primary:** Seizure frequency (daily diary—type, duration, severity). Compare 4-week baseline to each 4-week treatment block.

**Ketone levels:** Blood beta-hydroxybutyrate (target 2–5 mmol/L) or urine ketone strips (moderate to large). Measure daily during initiation, then weekly.

**Side effects:** Constipation (track bowel movements), weight (weekly), energy levels, mood, sleep quality

**Lab work (medically supervised):** Baseline and every 3 months: electrolytes, lipids, liver/kidney function, selenium, zinc, vitamin D, carnitine, complete blood count

### Key confounds to control for

**Medication changes:** Do not change antiseizure medications during the diet trial. If you must, wait 4 weeks after the change before assessing diet effect.

**Illness:** Fever, infections, and vomiting can disrupt ketosis and trigger breakthrough seizures. Note these in your diary.

**Carbohydrate "cheats":** Even small amounts (5–10g) can break ketosis for 24–48 hours. Be strict.

**Calorie intake:** Undereating can cause weight loss and metabolic stress. Overeating (especially carbs) prevents ketosis. Work with a dietitian to calculate appropriate calories.

**Hydration:** Dehydration worsens ketosis side effects (constipation, kidney stones). Drink 2–3 liters of water daily.

**Timing:** Ketone levels fluctuate throughout the day. Measure at the same time each day (morning is standard).

### What a positive result would look like

**≥50% reduction** in seizure frequency compared to your 4-week baseline, sustained for at least 4 weeks

**Seizure freedom:** Complete cessation of seizures for ≥3 months (this is the gold standard)

**Ketone levels consistently ≥2 mmol/L** (blood) or moderate-large (urine)

**No severe side effects** that outweigh the benefit (e.g., if you lose >5% body weight, develop kidney stones, or have severe constipation, the diet may need adjustment or discontinuation)

**Improvement within 2–4 weeks** of achieving stable ketosis (some respond within days)

**Critical warning:** Ketogenic diets are **medically supervised treatments** for epilepsy. Do not start one without a neurologist and dietitian. The diet can cause serious side effects including kidney stones, growth failure, selenium deficiency (which can cause heart damage), and severe dyslipidemia. Children on KDTs require regular blood monitoring. This guideline is for understanding the evidence, not for self-prescribing.