What the Evidence Shows | Metabolic Psychiatry Research

The honest summary, before the detail. Metabolic psychiatry — the idea that some mental illness involves how the brain produces and uses energy, and that a ketogenic diet might therefore help — is a real and active research area. Most of its more striking findings are in bipolar disorder and schizophrenia rather than in depression. For depression specifically there is now one randomised trial, in people whose antidepressants had not worked: the ketogenic group improved by about two more points on a depression questionnaire at six weeks, a difference that only just reached significance, had faded by twelve weeks, and which the authors themselves called of uncertain clinical relevance. Beside it sits a three-person case series. A review of eating only animal foods found nine human studies in all and concluded that long-term adherence cannot currently be recommended. This is a field worth watching. It is not a treatment, and nobody should read it as one.

This page is not a reason to change your treatment. Nothing below shows that food can replace talking therapy or medication, and nothing below is a reason to stop or alter a prescription. If you take antidepressants, keep taking them unless your prescriber advises otherwise. NHS guidance on treating depression.

The distinction almost everything gets wrong

The stronger results in metabolic psychiatry come from studies of bipolar disorder and schizophrenia. They are not findings about depression, and this site never presents them as if they were.

It matters because of how this research travels. A pilot study in bipolar disorder and schizophrenia produces a striking number — a 32% fall in symptom scores — and that number then appears in articles about depression, in social media posts about depression, and on websites selling things to people with depression. The study did not look at depression. Bipolar disorder and schizophrenia are different conditions, with different biology and different treatments.

For depression specifically, the entire evidence base is one randomised trial, a three-person case series, and a trial that has not reported. That is it.

Why the observational research is weaker than it looks

Population studies that find meat-eaters reporting better mood than people who avoid meat cannot tell you which way the arrow points. Depression takes away appetite, energy, concentration and interest — all the things that cooking and shopping depend on. People who are depressed may well eat differently because they are depressed.

Those studies also compare meat-eaters with vegetarians and vegans. That is a different question from whether eating only animal foods does anything, which is what this site is actually about. Nobody has run that study.

Every study this site relies on.

Mixed or borderline

DIME

Gao et al., JAMA Psychiatry, February 2026

Who was studied Depression — treatment-resistant

What it tested: The first proper randomised trial of this idea in depression. Eighty-eight UK adults aged 18 to 65 with a PHQ-9 score of 15 or more who had already tried at least two antidepressants without enough benefit. Six weeks on a ketogenic diet or on a carefully matched comparison diet, with everyone followed to twelve weeks.
What it found: At six weeks the ketogenic group improved more: a fall of 10.5 points on the PHQ-9 depression questionnaire against 8.3 in the comparison group. The adjusted difference between the groups was 2.18 points, with a confidence interval running from 4.33 down to 0.03 and a p-value of 0.05 — in other words, only just distinguishable from no difference at all. No serious adverse events occurred.
What it cannot tell you: By twelve weeks the gap between the groups had shrunk to 1.85 points and was no longer statistically significant. The authors describe the clinical relevance as uncertain, note the effect was modest and did not show up in most of the secondary measures, and report that fewer than one in ten participants managed to keep the diet going once the study support stopped. This is the strongest evidence that exists in depression, and it is a borderline six-week result that faded.

A ketogenic diet for treatment-resistant depression: a randomised clinical trial · registry NCT06091163

Reported a benefit

Calabrese case series

Calabrese, Frase & Ghaloo, Frontiers in Nutrition, May 2024

Who was studied Depression and anxiety — three patients

What it tested: Three adults aged 32 to 36 with major depression and generalised anxiety, treated in a specialist private metabolic-psychiatry practice with a personalised animal-based ketogenic diet for twelve to sixteen weeks.
What it found: All three reached remission on the questionnaires used — two within seven weeks and one at twelve. Anxiety improved faster than depression. The patients also lost roughly eleven to fifteen per cent of their body weight.
What it cannot tell you: Three people. No control group, no randomisation, and the records were looked at afterwards rather than collected as a trial. Everyone was a paying patient of a practice that specialises in this treatment, which is about as far from a representative sample as it is possible to get. The authors themselves present it as a reason to run a trial, not as evidence that the diet works. It cannot tell you that this diet does anything, and it is quoted on this site only because leaving out the most-cited paper in the field would be its own kind of dishonesty.

Complete remission of depression and anxiety using a ketogenic diet: case series · doi:10.3389/fnut.2024.1396685

Reported a benefit

Stanford pilot

Sethi et al., Psychiatry Research, 2024

Who was studied Bipolar disorder and schizophrenia — not depression

What it tested: Twenty-three adults with bipolar disorder or schizophrenia who also had metabolic problems, on a ketogenic diet for four months while staying on their existing psychiatric medication. Roughly 10% of calories from carbohydrate, 30% protein, 60% fat, with no calorie counting.
What it found: Participants with schizophrenia showed a 32% reduction on the Brief Psychiatric Rating Scale, and overall clinical severity scores improved by about 31%. The metabolic results were striking: by the end of the study none of the participants still met the criteria for metabolic syndrome.
What it cannot tell you: Everyone received the diet — there was no control group and no randomisation, so there is nothing to compare the improvement against, including the simple effect of four months of close clinical attention. Twenty-three people. And, most importantly for this site: these were people with bipolar disorder and schizophrenia. This trial says nothing directly about depression, and anybody presenting it as if it did is misleading you.

Ketogenic diet intervention on metabolic and psychiatric health in bipolar and schizophrenia: a pilot trial · registry NCT03935854

Flags risks

Scoping review

Lietz, Dapprich & Fischer, Nutrients, 2026

Who was studied All-animal-food diets — general health, not mental health

What it tested: A systematic sweep of everything published on eating only animal foods. The authors found nine human studies in total: five case studies, two surveys of social-media users, one exploratory study and one modelling exercise.
What it found: Individual papers reported weight loss, greater fullness and some improvement in inflammatory or metabolic markers. The authors judged the overall evidence very limited, citing small samples, short durations and the absence of control groups.
What it cannot tell you: The risks found are the reason this review belongs on the site. The authors flagged likely shortfalls in vitamin C, vitamin D, calcium, magnesium, iodine and fibre, and raised LDL and total cholesterol. Their conclusion was that long-term adherence to this way of eating cannot be recommended at present. Nine studies, almost all of them case reports and surveys, is not a body of evidence.

A scoping review of the current evidence, potential benefits and risks of all-animal-food diets · doi:nutrients-18-00348

Mixed or borderline

Meat and mental health

Dobersek et al., Critical Reviews in Food Science and Nutrition, 2021

Who was studied Meat eaters vs vegetarians and vegans — not all-animal diets

What it tested: Twenty studies pooled together, covering 171,802 people: 157,778 who ate meat and 13,259 who did not.
What it found: People who ate meat reported slightly lower depression and anxiety than people who avoided it. The effect sizes were small — 0.22 for depression and 0.17 for anxiety on a scale where 0.2 is conventionally called small.
What it cannot tell you: Three things, all of which matter. It is observational, and the authors say plainly that the evidence precludes drawing causal conclusions — people who are depressed may change how they eat, rather than the other way round. It compares meat-eaters with vegetarians and vegans, so it tells you nothing at all about eating only animal foods, which is a different question. And four of the authors declare in the paper that they have previously received funding from the Beef Checkoff, through the National Cattlemen’s Beef Association. That does not make the finding wrong, but you are entitled to know it.

Meat and mental health: a meta-analysis of meat consumption, depression and anxiety · doi:10.1080/10408398.2021.1974336

No results yet

KETO-MOOD

University Psychiatric Clinics Basel, registered 2023

Who was studied Depression — trial underway

What it tested: An eight-week assessor-blinded randomised comparison of a very-low-carbohydrate, high-fat diet against an active comparison diet, added to standard psychiatric care. The registry lists 120 planned participants.
What it found: Nothing yet. No results have been published. The registry gives an estimated completion date of July 2027.
What it cannot tell you: It is listed here so that nobody mistakes an ongoing trial for a finished one. Registered trials sometimes report late, report differently from their protocol, or do not report at all. Until it publishes, it adds nothing to what is known.

Ketogenic diet for depression — randomised controlled trial · registry NCT06105762

The risks, which belong on the same page

A site that lists only the encouraging findings is marketing, not education. The scoping review above flagged likely shortfalls in vitamin C, vitamin D, calcium, magnesium, iodine and fibre, and raised LDL and total cholesterol. Its authors' conclusion was that long-term adherence to this way of eating cannot be recommended at present.

That is the considered view of the people who read everything published on it. It sits here, on the same page as the encouraging case series, because both are true at once.

What a fair reading adds up to

This is not a treatment for depression. No trial has tested it as a replacement for therapy or medication, and none of the results above would support doing so.
The one randomised trial in depression was borderline and did not last. About two points on a questionnaire at six weeks, gone by twelve, in people whose antidepressants had already failed. The authors called the clinical relevance uncertain.
The stronger findings are in other conditions. Bipolar disorder and schizophrenia, in a single-arm pilot of twenty-three people with no control group.
The most-shared paper describes three patients. At a private clinic that specialises in this approach.
The risks are real and documented. Nutrient shortfalls and raised cholesterol, from the only systematic review of this way of eating.
Hardly anybody keeps it up. Fewer than one in ten trial participants were still following the diet once the study support ended.

Where to go from here

If your mood has been low for more than a couple of weeks, or it is interfering with ordinary life, the step that changes things is an appointment. In England you can refer yourself directly to NHS talking therapies without going through a GP.

If you want to discuss this research with a clinician, take the actual papers rather than the headlines — every one of them is linked above. For individual dietary advice, particularly about a pattern this restrictive, ask your GP to refer you to a registered dietitian.

How this page is maintained. Every study above was checked against the published paper or the trial registry, and the link is given so you can read it yourself. Findings are quoted with their limitations rather than without, and the funding declarations are reported where they exist. If a study is later retracted, corrected or overturned, this page gets changed rather than quietly left alone. No clinical review of this page has taken place. Last checked 25 October 2026.

Read the practical guides