Practiceby Lab 10 Get started ›
Crossfire

Is a ketogenic diet worth trying, or do its risks outweigh its benefits?

What the experts say

Georgia A. EdeMD · psychiatrist, nutritional and metabolic psychiatry

For a serious mood or psychotic illness that intensive standard treatment has not controlled, medically supervised carbohydrate restriction is worth considering as an addition to treatment. It is not established, and she says so in the same papers.

What it rests onA retrospective analysis of 31 hospitalised adults with major depressive disorder, bipolar disorder or schizoaffective disorder, which she wrote: all 28 who stayed on the diet beyond two weeks improved, 43 percent reached clinical remission, 64 percent were discharged on less medication. Its own limitations section states that the extent to which the diet contributed cannot be ascertained, that nobody was blinded, and that several dietary variables changed at once. She is separately a co-author of a 2026 review titled 'The ketogenic diet is not for everyone', and first author of a 2026 consensus listing absolute contraindications. Her declared interests on the 2022 paper are stock options in DietDoctor.com; by 2026 she declares a USD 750 CME course, book royalties, eight advisory boards and conference honoraria.

Source: frontiersin.org ↗
Neal BarnardMD, FACC · founding president, Physicians Committee for Responsible Medicine

For most people, the risks of a very-low-carbohydrate diet outweigh the benefits, and it should not be a long-term strategy.

What it rests onA narrative review in Frontiers in Nutrition on which he is senior author of seven (Crosby L, Davis B, Joshi S, Jardine M, Paul J, Neola M, Barnard ND; Front Nutr 8:702802, 28 Jul 2021). Its conclusion states that LDL cholesterol can rise, sometimes dramatically, that pregnant women on such diets are more likely to have a child with a neural tube defect even when supplementing folic acid, that the foods increasing on these diets are linked to higher risk of chronic kidney disease, cardiovascular disease, cancer, diabetes and Alzheimer's disease, and that for most individuals the risks of such diets outweigh the benefits. He serves without compensation as president of the Physicians Committee for Responsible Medicine and receives book royalties and lecture honoraria.

Christopher M. PalmerMD · psychiatrist, McLean Hospital and Harvard Medical School

For a serious mental illness that has not responded to standard treatment, a medically supervised ketogenic diet is worth asking a clinician about. It is a medical intervention, not a diet, and he says himself it is not established.

What it rests onTwo uncontrolled case reports of remission in chronic schizophrenia, set against a three-year study of over 6,000 medicated adults with schizophrenia in which 4 percent reached full relief. His own peer-reviewed writing says randomised controlled trials are still needed to demonstrate efficacy and safety, and his own 2026 case series documents hypomania or mania in nine people starting the diet. He declares royalties on Brain Energy and ownership of MH2, a practice that provides the therapy.

Source: psychologytoday.com ↗

Overview

They are not talking about the same person or the same indication, and the most quotable sentence from each of them is the one most likely to hide that.

Ede and Palmer are both addressing supervised therapeutic use in people whose serious mood or psychotic illness has failed intensive standard treatment: Ede's strongest evidence was delivered as three hospital-prepared meals a day with a psychiatrist reviewing patients six days a week, and Palmer's comes with a care team, baseline labs and a named contraindication list. Barnard is addressing general adults adopting a ketogenic diet long term for weight and chronic disease, plus two specific risk groups, pregnancy and chronic kidney disease. The review he rests on does not discuss depression, schizophrenia or bipolar disorder anywhere.

Flattening this into 'keto: good or bad' misrepresents all three. One more caution: the most inflammatory quotations in the accompanying press release are from Lee Crosby, the review's first author, not from Barnard.

Where they agree

Substantial, and all three state it in print. Barnard's conclusion opens by conceding the therapeutic case, that ketogenic diets reduce seizure frequency in some people with drug-resistant epilepsy, and Ede's consensus records the same at 97.9 percent agreement.

All three exclude pregnancy and breastfeeding, and all three exclude advanced kidney disease: Barnard from the review, Ede from both her 2014 article and her 2026 consensus, where breastfeeding is an absolute contraindication, and Palmer via the Delphi panel's absolute contraindication list. All three accept that LDL cholesterol can rise substantially, the risk Barnard's review names, which Ede's own co-authored 2026 review reflects by listing familial hypercholesterolaemia among relative contraindications.

None of them claims a ketogenic diet is a general-population health strategy, Ede explicitly tells general readers to make simpler changes first, and both she and Palmer say in their own peer-reviewed writing that the psychiatric case is not yet established.

What would settle it

For Ede: the randomised controlled trial of 100 adults with schizophrenia and bipolar disorder she reports as under way in Australia, which she names in her own 2025 editorial as the first of its kind. For Palmer: the randomised controlled trials he says are needed and reports roughly twenty of as under way.

For Barnard: long-term outcome data separating supervised therapeutic use with lipid monitoring from self-directed low-carbohydrate eating, a distinction the review does not draw, and a direct test of whether the chronic-disease risks it identifies apply at the same magnitude under clinical supervision.

Practice does not pick a winner here

You do, and your call is dated, kept, and yours to change. Follow the people you already trust and what they publish turns into a short daily practice.

Get started

Free to sign up and use. We do not ask for a card.