What if Depression is a Metabolic Problem?

World Travel
May 2
4 min read

Nineteen adults walked into a twelve-week trial in the UK with mental illness their psychiatrists had largely run out of ideas for. Most had cycled through years of medication, therapy, ketamine, transcranial magnetic stimulation. Seven had been admitted to psychiatric units, several more than once. Then they went on a ketogenic diet.

Twelve weeks later, mean depression scores had fallen 62 per cent. Anxiety scores had fallen 46 per cent. Eight participants were in remission for depression, nine for anxiety. Nobody dropped out.

The study, published in March in Frontiers in Nutrition, is small, retrospective and conducted by a researcher with a direct financial stake in the program being evaluated. It proves nothing on its own. But it lands inside a quiet shift in psychiatric research that is starting to ask an awkward question. What if some forms of depression, anxiety and even psychosis are not chemical imbalances in the conventional sense, but metabolic problems in the brain?

The mechanism, as the paper sets it out, draws on six interlocking pathways now circulating in the literature: impaired brain glucose metabolism, insulin resistance, neurotransmitter imbalance, oxidative stress, mitochondrial dysfunction and neuroinflammation. Stripped of the jargon, the premise is that the brain of a person with severe psychiatric illness may be struggling to make energy efficiently from glucose, and that switching the body's fuel source to ketones, the molecules produced when carbohydrates are restricted, can restore something that medication cannot reach.

This is not a fringe position. Chris Palmer at Harvard's McLean Hospital has been publishing case reports on ketogenic interventions in bipolar disorder and schizophrenia for several years. Shebani Sethi at Stanford reported in 2025 that 16 college students with major depressive disorder recorded a 69 per cent reduction in symptoms over 10 to 12 weeks on a therapeutic ketogenic diet, delivered alongside standard care. Trials are underway in the United States and Europe. Mainstream psychiatry has not adopted the model, but it has stopped dismissing it.

What distinguishes the new UK paper is its delivery. The programme, run by Integrative Ketogenic Research and Therapies, ran entirely online, in a group format, over six months. Participants prepared their own meals from a structured food list, tracked macronutrients in apps such as MyFitnessPal or Cronometer, and pricked their fingers each morning to measure blood ketones. Eight hours of education was followed by weekly group support calls and access to a moderated private community. Participants registered ketone levels above the threshold for nutritional ketosis 85 per cent of the time.

That matters because face-to-face metabolic psychiatry programs, however promising, will never reach the people most in need of them. In the UK, around one million people were waiting for mental health treatment in 2024, with some waiting up to four years for secondary care. Australia tells a similar story, particularly outside metropolitan areas. A programme that can be delivered through a screen, with a glucose meter and a food list, is a different proposition from one requiring access to a specialist clinic.

The participants themselves were a heterogeneous group: generalised anxiety disorder, bipolar I and II, recurrent depression, ADHD, autism, paranoid schizophrenia, premenstrual dysphoric disorder. Most had cycled through years of conventional and alternative treatment. The paper's author, Erin Bellamy of the University of East London, frames this not as a representative sample but as evidence that the program is reaching a chronic subgroup the existing system has not helped.

Caution is warranted. Bellamy owns the company being evaluated, a conflict declared in the publication but worth restating clearly: this is a researcher assessing her own paying clients, in a wellness-adjacent field where commercial ketogenic coaching is expanding faster than the evidence base supporting it. There was no control group. The sample is tiny. Several participants experienced worsening symptoms during the trial, two of them in connection with significant life events, others for no reason the paper can identify. Reported transition side effects included irritability, fatigue, headaches and disrupted sleep, although these resolved within a fortnight.

What the study offers, on its own terms, is a feasibility signal. The diet is hard, but people stuck with it. The structure, education first, then community support, appears to have held a fragile cohort together for three months. The biomarkers line up with the symptom changes. None of that proves causation, and Bellamy is careful not to claim it does. She is calling for a larger pilot across the full six-month program, with broader outcome measures and longer follow-up, before any conversation about integration into mainstream care.

The deeper question the paper raises is not whether keto works, but what it would mean if it did. A psychiatric model built around metabolism rather than chemistry would reorganise a great deal: how the illness is diagnosed, who treats it, what insurance covers, and whether food, in the end, sits closer to the centre of the conversation than the pharmaceutical industry has so far been willing to consider.