Ketogenic nutrition and the mind
Since 1921 the ketogenic diet has been an epilepsy therapy. For a few years now it has been studied in bipolar disorder, psychosis and depression, and the results look very different depending on the diagnosis.
More from the nutrition cluster
Foundations from this category
In psychiatry the ketogenic diet is neither a fad diet nor a magic bullet. It belongs to the most serious attempts to also understand mental illness as a disturbance of brain metabolism. And that attempt has come much further in bipolar disorder than in depression. Anyone who skips this difference is either selling a diet or missing one of the most interesting developments of the past twenty years.
You are scrolling through your phone. Someone writes that four months of eating ketogenic made their bipolar disorder calmer than anything before it. Underneath, hundreds of comments. One half is celebrating, the other half is angry.
And somewhere in between sits your question, the one that drowns in the noise: is there something to this.
I find both camps too quick. Anyone who talks about a magic bullet here has probably not read the studies. Anyone who talks about a fad diet usually has not read them either. Because this way of eating does not come from the diet world at all. It comes from neurology, it is more than a hundred years old, and for a few years now it has been studied seriously in psychiatry.
So this text sorts the studies by diagnosis. That is exactly the point almost everyone skips. In bipolar disorder the data look different than in depression, and in anxiety different again.
What to expect here
- Why an epilepsy therapy from 1921 ended up in psychiatry
- The metabolic hypothesis: the brain as an energy problem
- What ketones do, and which part of that is only shown in animals
- Bipolar disorder: why the trail is clearest here
- Depression: why the first randomised test is sobering
- Psychosis: striking case series, no controlled data
- Risks, contraindications and interactions
- What you can do without strict ketosis
- Twelve questions I am asked about this most often
A therapy from the year 1921 arrives in psychiatry
For many people, ketogenic sounds like bacon, butter and a trend that will disappear again soon. That is understandable. The loud part of this way of eating is often not serious.
The medical part is older than your grandparents. In 1921 a way of eating was formulated at the Mayo Clinic that rebuilds the metabolism of fasting without anyone fasting. Very few carbohydrates, a lot of fat. The liver produces ketone bodies, and the brain can use them as fuel. Children with severe epilepsy had fewer seizures on it.
You often read that this goes back to Hippocrates. What ancient texts document is fasting as a seizure treatment, not a ketogenic diet. The modern form is a hundred and five years old.
A London research group randomised 145 children with daily seizures in whom at least two antiepileptic drugs had failed. One group started ketogenic straight away, the other only three months later.
After three months the mean seizure frequency in the diet group was 62.0 percent of the starting value, in the control group 136.9 percent. 38 percent of the diet group reached more than 50 percent fewer seizures, in the control group 6 percent.
For you this means two things. A way of eating can measurably change something in a neurological illness. And the most common side effects were constipation, vomiting, low energy and hunger.
Neal EG et al. Lancet Neurol. 2008;7(6):500-506. DOI: 10.1016/S1474-4422(08)70092-9 [RCT, n=145]The sober benchmark comes from Cochrane. The 2020 review covers 13 randomised trials with 932 participants. In children the chance of seizure freedom was clearly higher, in adults nobody reached it, and all 13 trials counted as highly prone to bias because of missing blinding. So even in the best studied area the evidence is not as firm as many people think.
Martin-McGill KJ et al. Cochrane Database Syst Rev. 2020. DOI: 10.1002/14651858.CD001903.pub5 [Meta-analysis, k=13, n=932]
The image of the ketogenic diet as “just a diet” is misleading for this topic. In neurology it is a metabolic therapy with a side effect profile, follow up appointments and stopping criteria.
That is exactly the perspective I would like to see in psychiatry: an intervention with goals, measurements and a plan for the case that it does not fit.
And now you understand why anyone came up with the idea of testing this old therapy in psychiatry at all.
The hypothesis behind it: the brain as an energy problem
Do you know that feeling when your head simply runs empty on the third bad day in a row. Not sad, not anxious. Empty. As if the power supply had been turned down.
In classical psychiatry we ask first about messenger substances and about the life story. Both matter, and for many people both carry a lot. The metabolic perspective adds one more question: how well is this brain supplied with energy. Where the line between exhaustion and depression runs is something I describe in the article on burnout and depression.
This question has a surprisingly solid epidemiological basis.
A Dutch research group followed 601 adults for nine years who had never had depression or an anxiety disorder at the start. Three markers of insulin resistance were measured.
14 percent developed major depression. A higher triglyceride to HDL ratio came with a hazard ratio of 1.89. Those who newly developed prediabetes in the first two years had a hazard ratio of 2.66 afterwards.
For you this means: metabolism came before mood in time. That proves no cause, but it does challenge the obvious counter explanation, because nobody was depressed at the start.
Watson KT et al. Am J Psychiatry. 2021;178(10):914-920. DOI: 10.1176/appi.ajp.2021.20101479 [Cohort, n=601]The Canadian group around Cynthia Calkin studied 121 adults with bipolar disorder. Those who were also insulin resistant or had type 2 diabetes showed threefold higher odds of a chronic course and eightfold higher odds of not responding to lithium prophylaxis. The difference in the lithium response is the most striking number in the whole field: 3.2 percent versus 36.8 percent.
In psychosis a similar pattern shows up, and it does so before medication is involved. A meta-analysis across 12 studies and 1,137 people with a first, untreated psychosis found raised insulin resistance and impaired glucose tolerance. A second meta-analysis across 28 studies looked at the spread instead of the averages: not everyone is affected, but there is a subgroup with clearly stronger changes.
There is an uncomfortable point on top of that. Antipsychotics can dampen symptoms and burden metabolism at the same time. A meta-analysis across 27 randomised trials showed a more than doubled risk of clinically meaningful weight gain for most of these drugs. That is not an argument against these medicines, it is an argument for keeping metabolism in view. You will find the background in insulin resistance and leptin and insulin.
When a review article in the most important psychiatric journal describes bipolar disorder as a possible illness of brain energy metabolism, that is no longer an outsider position.
Berk M, Walder K, Kim JH. World Psychiatry. 2025;24(1):47-49. DOI: 10.1002/wps.21266The guiding question of metabolic psychiatry is not “which messenger substance is missing”. It is “how well does this brain get to its energy”.
From a KPNI perspective this is no competitor to the classical view, it is a second layer. Nerves, immune system and metabolism are not three subjects there, they are one system with one shared bottleneck: the mitochondria.
And now you understand why a fuel of all things came into view.
What ketones do in the brain, and which part is still hypothesis
Imagine two roads leading to your brain. One carries glucose, the other carries ketones. Normally the sugar road is the main artery and the second one lies quiet.
It gets interesting when the first one narrows.
Lever one, the backup fuel. A Canadian review summarised PET studies: in mild to moderate Alzheimer's disease the glucose uptake of the brain is reduced while ketone uptake stays unchanged. The sugar road narrows, the ketone road stays open. Whether the same holds in bipolar disorder or depression is plausible and not proven.
Lever two, gas and brake. This is where quotes online go wrong most often. The popular sentence is: ketones raise GABA and lower glutamate. The original paper supports only the first half.
A Swedish research group measured 17 amino acids in the cerebrospinal fluid of 26 children with refractory epilepsy, before the ketogenic diet and four months later.
GABA, taurine, serine and glycine rose clearly. Glutamate and aspartate stayed unchanged. Children with more than 50 percent seizure reduction had higher GABA values than those who did not respond.
For you this means: the calming side rose measurably, the excitatory side stayed the same in the spinal fluid. Anyone claiming otherwise is quoting this study loosely.
Dahlin M et al. Epilepsy Res. 2005;64(3):115-125. DOI: 10.1016/j.eplepsyres.2005.03.008 [Cohort, n=26]Twenty years later a second measurement level was added. A group in Edinburgh studied adults with bipolar disorder using magnetic resonance spectroscopy, so directly in the tissue. There, glutamate plus glutamine fell by 11.6 percent in the anterior and by 13.6 percent in the posterior cingulate. Two measurement levels, two statements. They belong next to each other, not on top of each other.
Lever three, silent inflammation. Beta hydroxybutyrate, the most important ketone body, blocked the NLRP3 inflammasome in cell culture and in mouse models, a central switch of silent inflammation. Important: these are cell and animal data. The sentence “ketogenic eating dampens neuroinflammation in the human brain” is not covered by them.
Lever four, more power plants. In young rats a three week ketogenic diet led to more mitochondria in the hippocampus. Brain slices from these animals kept synaptic transmission going even at low glucose. This picture is exactly what stands behind metabolic psychiatry. It is documented in the rat brain, not in the human one.
Four levers sound like four proofs. But they are four very different levels of proof. Two of them were measured in humans, two come from cell culture and animal models.
If somebody presents all four to you in the same tone of voice, you now know where you can ask.
And now you understand why the mechanisms sound more convincing than the clinical results.
Bipolar disorder: this is where the trail is clearest
Many people with bipolar disorder know this double burden. The illness itself, and on top of it the metabolic changes that come along over the years: weight, blood pressure, blood sugar, triglycerides.
That is exactly why the field grew here first. These studies measure psychiatric and metabolic endpoints at the same time.
At the Stanford department for metabolic psychiatry a research group accompanied 23 people with schizophrenia or bipolar disorder and metabolic abnormalities over four months. All of them stayed on their medication.
By the end, none of them still met the criteria for a metabolic syndrome. With good adherence to the diet, visceral fat tissue fell by 36 percent and HOMA-IR by 27 percent. The overall clinical impression improved by 31 percent, sleep quality by 19 percent.
For you this means: impressive numbers, single arm design. Without a control group it stays open what would have happened without the diet. The metabolic part is more robust than the psychiatric one.
Sethi S et al. Psychiatry Res. 2024;335:115866. DOI: 10.1016/j.psychres.2024.115866 [Case Series, n=23]In a French clinic the charts of 31 adults with hard to treat mental illness were reviewed retrospectively. All of them additionally received ketogenic food, for between 6 and 248 days.
Three people did not last 14 days. In the remaining 28 the Hamilton depression score fell from 25.4 to 7.7 and the MADRS from 29.6 to 10.1. Weight, blood pressure, blood sugar and triglycerides improved as well.
For you this means: the numbers are large, the design is weak. In a clinic, psychotherapy, daily structure and distance from everyday life all act at the same time. This observation deserves a controlled trial, it does not replace one.
Danan A et al. Front Psychiatry. 2022;13:951376. DOI: 10.3389/fpsyt.2022.951376 [Case Series, n=31]The most interesting single study comes from Edinburgh and is regularly misunderstood online. 27 people with bipolar disorder, mood stable at the start, kept a modified ketogenic diet for six to eight weeks. On the clinical scales nothing changed. That sounded like a failure, but it was a question of the starting point: whoever starts in the balanced range has barely any room upwards.
It got interesting across the course of the day. In the 14 people with reliable daily self ratings, the ketone level was positively linked with mood and energy and negatively with impulsivity and anxiety. On top came the change in the glutamate signal in the cingulate already mentioned.
There is also indirect evidence without ketosis. In a quadruple blinded randomised trial with 45 people with treatment resistant bipolar depression, insulin resistance was targeted with metformin. In the eleven people where that worked, the depression scores improved clearly from week 6 on, with large effect sizes up to week 26.
Three independent observation series show a consistent pattern, and the metabolic basis is well documented. But there is no randomised, controlled trial of ketogenic food in bipolar disorder. Everything comes from single arm pilot studies and case series.
“There are studies on it” and “it is proven” are two different sentences. In bipolar disorder the first one holds, not the second.
No reason to dismiss the topic. A reason to read every percentage with the addition: observed in a single arm pilot study.
And now you know why I listen carefully in this condition and still put things cautiously.
Depression: why the data are thinner than the headlines
If you live with depression, you are probably hoping for good news here. I do not want to disappoint you, and I do not want to tell you anything untrue either.
In depression the evidence base is better than in bipolar disorder. Better means: tested more strictly. And that is exactly why the results are more modest.
An Oxford research group randomised 88 adults with treatment resistant depression to two six week food programmes: a ketogenic diet from prepared meals, or an active comparison diet with more vegetables and better fats. Both groups received the same amount of counselling and attention.
Depressive symptoms fell clearly in both groups. The difference was 2.18 PHQ-9 points after six weeks and 1.85 points after twelve weeks, the latter no longer statistically meaningful. No secondary endpoint showed a difference. There were no serious adverse events.
For you this means: on the ketogenic diet people felt better. On the comparison diet almost equally better. The endpoints had been published two years earlier, so this trial did not look for its question afterwards.
Gao M et al. JAMA Psychiatry. 2026;83(4):331-340. DOI: 10.1001/jamapsychiatry.2025.4431 [RCT, n=88]A Canadian group evaluated 50 studies with 41,718 people and separated randomised designs from weaker ones.
Across ten randomised trials the effect on depressive symptoms was minus 0.48 standardised mean difference, so moderate. Across nine randomised trials on anxiety there was no association at all. In the uncontrolled studies both values came out much better.
For you this means: part of the benefit comes from the design, not from the food. The effect was largest where ketosis had been confirmed in the blood.
Janssen-Aguilar R et al. JAMA Psychiatry. 2026;83(1):13-22. DOI: 10.1001/jamapsychiatry.2025.3261 [Meta-analysis, k=50, n=41,718]A systematic review from 2023 traces the state before that. Of 1,377 screened papers, twelve studies with 389 people remained, nine of them case reports, not a single one of high quality. And one detail that almost never appears in the German language internet: in some people the symptoms came back after the diet ended.
This is where the only well built randomised test of ketogenic food in depression exists so far. Against an active comparison diet it found only a small advantage that was no longer significant after twelve weeks. The meta-analysis across ten randomised trials shows a moderate effect. Together that means: a signal, not a breakthrough.
The most honest question in this field is not “does it do anything”. It is: is it the ketosis itself, or is it everything that comes with it.
Less sugar, fewer heavily processed foods, more structure, weekly conversations, weight loss. The active comparison arm shows how much of that you can have without ketosis. How much lifestyle as a whole can move is in the article on lifestyle as therapy.
And now you understand why in depression I talk about other food routes first and about ketosis only after that.
Psychosis and schizophrenia: striking cases, no proof
There are numbers in this field at which I looked twice on first reading.
In the French chart review the PANSS score fell from 91.4 to 49.3 points in the ten people with schizoaffective disorder. In the Stanford pilot study the BPRS scores fell by 32 percent in the participants with schizophrenia. Shifts like that are rarely seen in psychiatry.
And still this is no proof. Both observations come from unblinded investigations without a control group. On top of that come the historical triggers of the field: two case reports from 2017 and two from 2019 in which the authors describe psychotic symptoms receding on ketogenic food. Four people in total, the beginning of a research story and not its result.
What makes me thoughtful here is the prehistory from section two. The disturbed glucose metabolism is already there in a first psychosis, before the first medication is given, and the variability meta-analysis suggests a particularly affected subgroup. If there is a benefit here, it would probably be found there and not in everyone.
The reported changes are the largest in the field, the designs the weakest. A randomised trial does not exist. What is well documented is the metabolic abnormality before treatment starts, not the dietary therapy.
Large numbers from small, uncontrolled investigations are no reason for enthusiasm and no reason for mockery. They are a reason for a proper trial.
That is exactly what is happening right now. Until results are in, this area stays a hypothesis with striking individual observations.
And now you know why the most impressive numbers of all rest on the weakest ground.
The edges: alcohol withdrawal, anxiety and eating disorders
Three more areas belong here. They show how differently this field looks at its edges.
At the US American NIAAA, people in inpatient alcohol withdrawal were randomised over three weeks to a ketogenic diet (19 people) or a standard diet (14 people). The background: in alcohol dependence the brain metabolises more acetate instead of glucose.
The ketogenic group needed fewer benzodiazepines in the first week of withdrawal and reported less craving. In a separate rat experiment a preceding ketogenic diet lowered alcohol intake.
For you this means: a very direct test of the metabolic idea. If one fuel drops out, a backup fuel may ease the withdrawal symptoms. The sample is small, and the animal part stays an animal part.
Wiers CE et al. Sci Adv. 2021;7(15):eabf6780. DOI: 10.1126/sciadv.abf6780 [RCT, n=33] plus [In vivo, rat]In anxiety it looks different, and that is the clearest negative statement in this text. Across nine randomised trials the large meta-analysis found no association between ketogenic food and anxiety symptoms. In the uncontrolled studies, by contrast, it looked like a large effect.
And then there is one area where I become especially attentive. There is an open pilot study with five adults who had been weight rehabilitated after anorexia nervosa but still suffered from the thoughts and behaviours of the eating disorder. They began a ketogenic diet with the explicit goal of keeping their weight, followed by six ketamine infusions. Several eating disorder scales improved, one person relapsed four months later without ketogenic food.
Five people, open design, close professional supervision and an explicit goal of weight stability. This study is no argument that a ketogenic diet would be suitable in an eating disorder. It shows rather under which safety conditions something like this is researched at all.
With a current or past eating disorder, a strictly rule based way of eating can activate old patterns again, even when it is medically justified. Restriction stays restriction. If this concerns you, please read the article on eating disorders between body and mind first and talk to the professional treating you before you change anything.
Nine randomised trials together show no association with anxiety symptoms. The positive reports come from uncontrolled investigations. A textbook example of how much a study design can change a result.
A review from 2024 screened 30 studies across all neuropsychiatric diagnoses, from autism through ADHD to addiction. The result is a map, not a recommendation.
The sentence “ketogenic nutrition in mental illness” is therefore too coarse. It only becomes useful with a diagnosis behind it.
And now you know why a diagnosis stands in front of so many statements in this text.
Risks, limits and what medical supervision actually means
Suppose you still think: I would like to discuss this with the person treating me. Exactly the right reflex. Because this part is missing from most articles online.
Psychiatric medication is never changed on your own
In all the studies in this text the participants stayed on their psychiatric medication, the food came on top of it. None of these papers examined whether medication can be replaced by ketogenic food.
Stopping or reducing antidepressants, mood stabilisers and antipsychotics abruptly can trigger withdrawal symptoms and relapses. In bipolar disorder and psychosis that can become dangerous.
With lithium there is one more reason: in the first ketogenic weeks the fluid and sodium balance often shifts noticeably, and the blood level can change through that. This is no detail, it is a reason for monitoring.
A ketogenic change of diet with a psychiatric illness therefore belongs in medical supervision. Not because you were not allowed to decide for yourself, but because two systems are being moved at once: your metabolism and your medication.
In 2026 an international expert panel put together 33 recommendations for the first time in a Delphi process: who counts as suitable, what should be measured, how to proceed. Important for placing this: a Delphi consensus is an ordered expert opinion, not evidence. And several of the eight people involved are at the same time authors of the primary studies quoted here. That belongs said openly.
When caution is appropriate
Eating disorder, current or in the history
In my view the most important filter. If at all, then only with close professional supervision and a clear eye on weight.
SGLT2 inhibitors and glucose lowering medication
Together with very few carbohydrates, SGLT2 inhibitors can raise the risk of ketoacidosis at normal blood sugar. On insulin or other antidiabetic drugs the requirement can drop quickly. Both need medical steering.
Liver, kidney and inherited metabolic disorders
A very high fat diet can change the load on these organs. In disorders of fatty acid breakdown it is dangerous. That belongs clarified before the start.
Pregnancy and breastfeeding
For a psychiatric question, reliable data are missing entirely. No area for experiments.
Unstable phases
An acute manic, severely depressive or psychotic phase is not the moment for a large change of diet. Stability comes first.
What supervision means in practice
- Clarify beforehand: diagnosis, medication list, kidney and liver values, blood lipids, blood sugar, HbA1c.
- During the switch: check electrolytes and fluid balance, and with lithium the blood level.
- Measure instead of guessing: without proof in the blood nobody knows whether ketosis is present at all.
- Watch the mind: document mood, sleep and drive, ideally together with a second person.
- Plan the exit: relapses after stopping the diet have been described. Whoever starts should know beforehand how things continue.
- Take side effects seriously: constipation, vomiting, low energy and hunger are well documented.
Medical supervision here is no formality and no legal sentence at the end of an article. It is the difference between an intervention you can observe and stop again, and a self experiment with an illness in which a relapse has real consequences.
And now you understand why I wrote this section longer than any study section before it.
What you can do without strict ketosis
Maybe you are thinking right now: a lot of effort for an uncertain result. That thought is fair.
The good news: there is a much lower threshold door in the same direction, and in depression it is better documented than the ketogenic one.
An Australian research group randomised 67 adults with moderate to severe depression, most of whom were already receiving psychotherapy, medication or both. One group received seven individual counselling sessions on a modified Mediterranean diet, the other a social contact programme with an identical schedule.
After twelve weeks 32.3 percent of the food group reached remission compared with 8.0 percent in the control group. The effect size was a Cohen d of 1.16, the number needed to treat 4.1.
For you this means: in depression there is a randomised tested dietary intervention that forces nobody to cut out fruit or to measure ketones. Its effect size lies clearly above the one the ketogenic trial reached against its active comparison.
Jacka FN et al. BMC Med. 2017;15(1):23. DOI: 10.1186/s12916-017-0791-y [RCT, n=67]Between ordinary eating and strict ketosis there is also a large field: the stability of your blood sugar. Smaller swings can mean fewer afternoon crashes, calmer sleep and less craving. That is no ketosis, but it goes in the same metabolic direction.
It becomes visible when you measure it. A glucose sensor over two weeks can show you which meals move your blood sugar. More on that in 14 days with a glucose sensor and blood sugar spikes. Without technology, unprocessed food and quality over quantity are the better entry point.
The choice is not ketogenic or nothing. Between the two lies almost everything that makes a metabolism calmer.
Whoever starts there loses nothing. And can still decide later whether a stricter form comes into question at all.
The metabolic hypothesis deserves to be taken seriously and is mechanistically well founded. The therapeutic evidence is promising and untested in bipolar disorder, tested and modest in depression, and negative in anxiety. In none of these areas does a way of eating replace psychotherapy or medication.
If you want to move forward with your own situation and not only read: below this article you will find the option to book an appointment.
And now you have both together: why this field is exciting, and where its limits run.
Frequent questions about ketogenic nutrition and mental health
Can a ketogenic diet ease depression?
The honest answer is: maybe a little, and less than many headlines suggest. The first properly randomised test in treatment resistant depression included 88 adults. After six weeks the ketogenic group was 2.18 PHQ-9 points better than an active comparison diet, after twelve weeks the gap was 1.85 points and no longer statistically meaningful. A meta-analysis across 50 studies and 41,718 people found a moderate effect on depressive symptoms across ten randomised papers. That is a signal, but not a breakthrough.
How long does it take before something changes in mood on a ketogenic diet?
In the available studies the interventions ran between six weeks and four months. The randomised depression trial measured its primary endpoint after six weeks. The Stanford pilot study ran four months, the inpatient case series between 6 and 248 days. In the Edinburgh pilot study self reported mood even swung from day to day with the ketone level. A reliable time frame for an individual person cannot be taken from these data. Whoever uses the first two weeks as a yardstick is usually still measuring the switch itself.
Has the ketogenic diet been studied in bipolar disorder?
Studied yes, proven no. There is a single arm pilot study from Stanford with 23 people included, a retrospective review of 31 inpatient charts and a pilot study from Edinburgh with 27 participants including brain imaging. None of these papers had a control group for the ketogenic intervention. That means: the observations are consistent and interesting, but nobody can say how much of it belongs to the food and how much to the intensive care, the structure and the weight loss.
What is metabolic psychiatry anyway?
Metabolic psychiatry is a way of looking, not a set therapy. It does not ask first which messenger substance is missing, but how well the energy supply of a brain works. The starting point are observations like these: in a nine year Dutch cohort of 601 people, insulin resistance predicted newly occurring depression, and even in a first untreated psychosis the glucose metabolism is measurably altered. In this field the ketogenic diet is only one of several tools under study. It replaces neither psychotherapy nor medication.
Can I stop my antidepressants if I eat ketogenic?
No, and explicitly not on your own. In all the studies quoted here the participants stayed on their psychiatric medication, the food came on top. Stopping antidepressants, mood stabilisers or antipsychotics suddenly can trigger withdrawal symptoms and relapses. With lithium there is the additional point that fluid and salt balance can shift in the first ketogenic weeks, which influences the blood level. Every change of medication belongs in the hands of the doctor treating you.
What side effects does a ketogenic diet have on the mind?
Psychological side effects have hardly been recorded systematically so far, and that is exactly a gap. From epilepsy research, low energy, hunger, constipation and vomiting are well documented. The switching phase can come with tiredness, irritability and changes in sleep, which is hard to tell apart from a worsening when you live with a mental illness. A systematic review from 2023 pointed out explicitly that side effects and dropout rates often go unreported in this field. That is why observation from outside belongs to it.
Who is a ketogenic diet not suitable for?
Caution is warranted with a current or past eating disorder, in pregnancy and breastfeeding, with advanced liver or kidney disease, with certain inherited metabolic disorders and in people with strongly swinging blood sugar on medication. With an eating disorder in the history, a strictly rule based way of eating can reactivate old patterns, even when it is medically justified. This point is almost never mentioned online and in my view it is the most important filter of all.
Which medications go badly with a ketogenic diet?
Particular care applies to SGLT2 inhibitors, because together with very few carbohydrates they can raise the risk of ketoacidosis at normal blood sugar. With insulin and other glucose lowering agents the requirement can drop quickly, which favours hypoglycaemia. With lithium the level can change through fluid and sodium shifts. Antipsychotics with a strong metabolic profile call for regular checks anyway. None of these combinations is automatically forbidden, but none of them belongs in a self experiment.
What happens if I stop the ketogenic diet again?
That is one of the most honest open questions. A systematic review from 2023 with twelve studies and 389 people reports that in some people the symptoms came back after the diet was stopped. In the pilot study on anorexia as well, one of five people relapsed four months after the end of treatment and without ketogenic food. That does not mean every step back is preprogrammed. It means the question of what comes afterwards belongs discussed before the beginning.
Is intermittent fasting the same as a ketogenic diet?
No. With intermittent fasting you shift the time window in which you eat. With a ketogenic diet you permanently change the composition. Both can lead to ketone bodies in the blood, but to a very different extent and with very different reliability. The psychiatric studies have consistently looked at a low carbohydrate way of eating, not at an eating window. Whoever transfers the study results to intermittent fasting transfers them to something that was not studied that way.
Do I have to measure ketones, and with what?
In the studies measurements were taken, and for good reason. The 2026 meta-analysis found stronger associations with depressive symptoms precisely in those studies that had confirmed ketosis in the blood. Without measurement nobody knows whether ketosis is present at all. Blood tests for beta hydroxybutyrate from the fingertip are the usual route, breath and urine tests count as less precise. The international expert consensus from 2026 names measurement and monitoring as a standard part of such a therapy.
What does a medically supervised ketogenic switch cost roughly?
Reliable cost figures for the German speaking region do not exist, and I do not want to invent any. What can be named are the components: medical consultations, laboratory checks before and during the switch, test strips for ketone measurement, in many cases nutritional therapy support and often a higher shopping price for the food itself. For a psychiatric question this is not covered by statutory health insurance in Germany. Whoever considers it should work out the costs beforehand, so that the switch does not fail on money after eight weeks.
Where this topic leads next
The metabolic perspective does not end with ketosis. It shows up everywhere energy, inflammation and the nervous system belong together.
Keto and the mind
Evidence sorted by diagnosis
this articleLifestyle as therapy
What sleep, movement and food can move in depression
Silent inflammation
The invisible brake between metabolism and nervous system
Burnout and mitochondria
When energy production itself becomes the bottleneck
And if the insulin thread from the second section keeps occupying you, you will find it in detail in the article on insulin resistance.
Scientific sources
- Neal EG, Chaffe H, Schwartz RH et al. The ketogenic diet for the treatment of childhood epilepsy: a randomised controlled trial. Lancet Neurol. 2008;7(6):500-506. DOI: 10.1016/S1474-4422(08)70092-9 · PMID: 18456557 [RCT, n=145]
- Martin-McGill KJ, Bresnahan R, Levy RG, Cooper PN. Ketogenic diets for drug-resistant epilepsy. Cochrane Database Syst Rev. 2020;6(6):CD001903. DOI: 10.1002/14651858.CD001903.pub5 · PMID: 32588435 [Meta-analysis, k=13, n=932]
- Watson KT, Simard JF, Henderson VW et al. Incident Major Depressive Disorder Predicted by Three Measures of Insulin Resistance: A Dutch Cohort Study. Am J Psychiatry. 2021;178(10):914-920. DOI: 10.1176/appi.ajp.2021.20101479 · PMID: 34551583 [Cohort, n=601]
- Calkin CV, Ruzickova M, Uher R et al. Insulin resistance and outcome in bipolar disorder. Br J Psychiatry. 2015;206(1):52-57. DOI: 10.1192/bjp.bp.114.152850 · PMID: 25323142 [Cohort, n=121]
- Calkin CV, Chengappa KNR, Cairns K et al. Treating Insulin Resistance With Metformin as a Strategy to Improve Clinical Outcomes in Treatment-Resistant Bipolar Depression (the TRIO-BD Study). J Clin Psychiatry. 2022;83(2):21m14022. DOI: 10.4088/JCP.21m14022 · PMID: 35120288 [RCT, n=45]
- Perry BI, McIntosh G, Weich S et al. The association between first-episode psychosis and abnormal glycaemic control: systematic review and meta-analysis. Lancet Psychiatry. 2016;3(11):1049-1058. DOI: 10.1016/S2215-0366(16)30262-0 · PMID: 27720402 [Meta-analysis, k=12, n=1,137]
- Pillinger T, McCutcheon RA, Howes OD. Variability of glucose, insulin, and lipid disturbances in first-episode psychosis: a meta-analysis. Psychol Med. 2023;53(7):3150-3156. DOI: 10.1017/S0033291721005213 · PMID: 37449481 [Meta-analysis, k=28]
- Barton BB, Segger F, Fischer K et al. Update on weight-gain caused by antipsychotics: a systematic review and meta-analysis. Expert Opin Drug Saf. 2020;19(3):295-314. DOI: 10.1080/14740338.2020.1713091 · PMID: 31952459 [Meta-analysis, k=27]
- Cunnane SC, Courchesne-Loyer A, St-Pierre V et al. Can ketones compensate for deteriorating brain glucose uptake during aging? Implications for the risk and treatment of Alzheimer's disease. Ann N Y Acad Sci. 2016;1367(1):12-20. DOI: 10.1111/nyas.12999 · PMID: 26766547 [Mechanism Review]
- Dahlin M, Elfving Å, Ungerstedt U, Åmark P. The ketogenic diet influences the levels of excitatory and inhibitory amino acids in the CSF in children with refractory epilepsy. Epilepsy Res. 2005;64(3):115-125. DOI: 10.1016/j.eplepsyres.2005.03.008 · PMID: 15961283 [Cohort, n=26]
- Youm YH, Nguyen KY, Grant RW et al. The ketone metabolite beta-hydroxybutyrate blocks NLRP3 inflammasome-mediated inflammatory disease. Nat Med. 2015;21(3):263-269. DOI: 10.1038/nm.3804 · PMID: 25686106 [In vitro] and [In vivo, mouse]
- Bough KJ, Wetherington J, Hassel B et al. Mitochondrial biogenesis in the anticonvulsant mechanism of the ketogenic diet. Ann Neurol. 2006;60(2):223-235. DOI: 10.1002/ana.20899 · PMID: 16807920 [In vivo, rat]
- Campbell IH, Campbell H. The metabolic overdrive hypothesis: hyperglycolysis and glutaminolysis in bipolar mania. Mol Psychiatry. 2024;29(5):1521-1527. DOI: 10.1038/s41380-024-02431-w · PMID: 38273108 [Mechanism Review]
- Berk M, Walder K, Kim JH. Past, present and future of research on brain energy metabolism in bipolar disorder. World Psychiatry. 2025;24(1):47-49. DOI: 10.1002/wps.21266 · PMID: 39810675 [Review]
- Sethi S, Wakeham D, Ketter T et al. Ketogenic Diet Intervention on Metabolic and Psychiatric Health in Bipolar and Schizophrenia: A Pilot Trial. Psychiatry Res. 2024;335:115866. DOI: 10.1016/j.psychres.2024.115866 · PMID: 38547601 [Case Series, n=23]
- Danan A, Westman EC, Saslow LR, Ede G. The Ketogenic Diet for Refractory Mental Illness: A Retrospective Analysis of 31 Inpatients. Front Psychiatry. 2022;13:951376. DOI: 10.3389/fpsyt.2022.951376 · PMID: 35873236 [Case Series, n=31]
- Campbell IH, Needham N, Grossi H et al. A pilot study of a ketogenic diet in bipolar disorder: clinical, metabolic and magnetic resonance spectroscopy findings. BJPsych Open. 2025;11(2):e34. DOI: 10.1192/bjo.2024.841 · PMID: 39995103 [Case Series, n=27]
- Gao M, Kirk M, Knight H et al. A Ketogenic Diet for Treatment-Resistant Depression: A Randomized Clinical Trial. JAMA Psychiatry. 2026;83(4):331-340. DOI: 10.1001/jamapsychiatry.2025.4431 · PMID: 41637092 [RCT, n=88]
- Gao M, Kirk M, Lash E et al. Evaluating the efficacy and mechanisms of a ketogenic diet as adjunctive treatment for people with treatment-resistant depression: A protocol for a randomised controlled trial. J Psychiatr Res. 2024;174:230-236. DOI: 10.1016/j.jpsychires.2024.04.023 · PMID: 38653031 [Review, study protocol]
- Janssen-Aguilar R, Vije T, Peera M et al. Ketogenic Diets and Depression and Anxiety: A Systematic Review and Meta-Analysis. JAMA Psychiatry. 2026;83(1):13-22. DOI: 10.1001/jamapsychiatry.2025.3261 · PMID: 41191382 [Meta-analysis, k=50, n=41,718]
- Dietch DM, Kerr-Gaffney J, Hockey M et al. Efficacy of low carbohydrate and ketogenic diets in treating mood and anxiety disorders: systematic review and implications for clinical practice. BJPsych Open. 2023;9(3):e70. DOI: 10.1192/bjo.2023.36 · PMID: 37066662 [Systematic Review]
- Yousufzai W, Singh M, Ahmadi L et al. Measuring the effects of ketogenic diet on neuropsychiatric disorder: A scoping review. Prog Neuropsychopharmacol Biol Psychiatry. 2025;136:111205. DOI: 10.1016/j.pnpbp.2024.111205 · PMID: 39592022 [Systematic Review]
- Wiers CE, Vendruscolo LF, van der Veen JW et al. Ketogenic diet reduces alcohol withdrawal symptoms in humans and alcohol intake in rodents. Sci Adv. 2021;7(15):eabf6780. DOI: 10.1126/sciadv.abf6780 · PMID: 33837086 [RCT, n=33] plus [In vivo, rat]
- Calabrese L, Scolnick B, Zupec-Kania B et al. Ketogenic diet and ketamine infusion treatment to target chronic persistent eating disorder psychopathology in anorexia nervosa: a pilot study. Eat Weight Disord. 2022;27(8):3751-3757. DOI: 10.1007/s40519-022-01455-x · PMID: 35997954 [Case Series, n=5]
- Ede G, Bernstein M, Calabrese L et al. Awareness and best practices in using ketogenic therapy to treat serious mental illness: a modified Delphi consensus. Front Nutr. 2026;13:1749406. DOI: 10.3389/fnut.2026.1749406 · PMID: 41816235 [Review, Delphi consensus]
- Jacka FN, O'Neil A, Opie R et al. A randomised controlled trial of dietary improvement for adults with major depression (the SMILES trial). BMC Med. 2017;15(1):23. DOI: 10.1186/s12916-017-0791-y · PMID: 28137247 [RCT, n=67]
- Palmer CM, Gilbert-Jaramillo J, Westman EC. The ketogenic diet and remission of psychotic symptoms in schizophrenia: Two case studies. Schizophr Res. 2019;208:439-440. DOI: 10.1016/j.schres.2019.03.019 · PMID: 30962118 [Case Series, n=2]
- Palmer CM. Ketogenic diet in the treatment of schizoaffective disorder: Two case studies. Schizophr Res. 2017;189:208-209. DOI: 10.1016/j.schres.2017.01.053 · PMID: 28162810 [Case Series, n=2]
- Calkin C, Kamintsky L, Friedman A. Reversal of insulin resistance is associated with repair of blood-brain barrier dysfunction and remission in a patient with treatment-resistant bipolar depression. Bipolar Disord. 2022;24(5):553-555. DOI: 10.1111/bdi.13199 · PMID: 35247029 [Case Series, n=1]