Martin-McGill · The Cochrane database of systematic reviews 2020 · systematic review and meta-analysis · n=13 studies (932 participants)

Ketogenic diets for drug-resistant epilepsy.

Cited 399 times in the scientific literature.

Level 1 - systematic review of randomized trials

Systematic review and meta-analysis of randomized controlled trials

PubMed 32588435 · doi:10.1002/14651858.CD001903.pub5 · record verified 2026-08-29

What was done

This Cochrane systematic review and meta-analysis evaluated randomized and quasi-randomized controlled trials of ketogenic diets (including classic ketogenic diets, medium-chain triglyceride diets, and modified Atkins diets) versus usual care or alternative ketogenic protocols for drug-resistant epilepsy across all ages. Authors extracted data independently, assessed risk of bias, and conducted intention-to-treat meta-analyses reporting risk ratios (RRs) with 95% confidence intervals (CIs) for seizure freedom, seizure reduction (50% or greater), and adverse effects.

What was found

Thirteen studies with 932 participants (711 children, 221 adults) were included. In children compared to usual care (4 studies, n = 385), ketogenic diets increased seizure freedom (RR 3.16, 95% CI 1.20 to 8.35; P = 0.02; very low certainty) and seizure reduction of 50% or more (RR 5.80, 95% CI 3.48 to 9.65; P < 0.001; low certainty). In adults compared to usual care (2 studies, n = 141), zero participants achieved seizure freedom, and seizure reduction did not reach statistical significance (RR 5.03, 95% CI 0.26 to 97.68; P = 0.29; very low certainty). In child diet comparisons, up to 55% achieved seizure freedom on a classic 4:1 ketogenic diet and up to 25% on a modified Atkins diet. Gastrointestinal symptoms (vomiting, constipation, diarrhea) were the most common adverse events.

Why it matters

Ketogenic dietary therapies provide an effective non-pharmacological treatment option that substantially reduces seizures in children with drug-resistant epilepsy. However, evidence supporting their efficacy in adults remains inconclusive.

Limits

All 13 included trials had high risk of performance and detection bias due to lack of blinding. Studies had small sample sizes causing serious imprecision, leading to low to very low GRADE certainty across all outcomes. Adult data were limited to only 221 participants across studies.

Cited by