Keto diet may ease depression

Major depressive disorder, bipolar disorder and schizophrenia have previously been linked to mitochondrial dysfunction, insulin resistance, reduced glucose consumption by the brain and systemic inflammation.

Ketogenic diets are high-fat, moderate-protein, and very low-carbohydrate diets. They entered clinical practice about a hundred years ago as a non-fraught treatment option for severe, drug-resistant epilepsy.

With long-term carbohydrate restriction, the main source of energy for the brain shifts from glucose to ketone bodies - beta-hydroxybutyrate, acetoacetate and acetone. This switch is known to affect mitochondrial function, oxidative stress levels and inflammatory signalling pathways.

Previously, possible effects of ketodiet on gamma-aminobutyric acid (GABA) and glutamate signalling, gut microbiota and neuronal network stability have been reported, as well as overlaps with the pharmacodynamics of normotimics (mood stabilising drugs).

Initial clinical data and case descriptions indicated improvements in mood, anxiety, cognitive function, body weight and quality of life on ketogenic regimes, but also raised safety concerns in certain groups - for example, people with mitochondrial DNA deletion or taking multiple medications.

In the study, "Ketogenic Diets and Depression and Anxiety," published in the journal JAMA Psychiatry, scientists conducted a systematic review and meta-analysis to assess the association between ketogenic diets and adult mental health, with a focus on depressive and anxiety symptoms.

A total of 50 studies involving 41,718 people aged 18 to 70 years from 15 countries were included in the analysis, of which 23 papers were conducted in the US.

The sample included both psychiatric and non-psychiatric groups: people with major depressive disorder, bipolar disorder, schizophrenia, generalised anxiety disorder, post-traumatic stress disorder, and patients with medical conditions such as obesity. Mental health outcomes were assessed only using validated psychiatric scales including, for example, the Patient Health Questionnaire-9 (PHQ-9) and Generalised Anxiety Disorder-7 (GAD-7).

Study types included 14 randomised clinical trials, 17 quasi-experimental papers, five analytical cross-sectional studies, six clinical case series and eight individual clinical observations. The publications span from 1965 to 2025, with a marked increase in papers starting in 2019 and peaking in 2024.

Depressive symptoms were assessed in 41 studies and were the most frequent outcome, anxiety symptoms in 29 papers. Twenty-two studies included participants with formal psychiatric diagnoses, while 33 papers focused on non-psychiatric populations, where mental health indicators were more likely to act as secondary endpoints.

Results for depression

The randomised clinical trials on depressive symptoms included 10 studies involving 631 participants. Pooled analysis showed a standardised mean difference (SMD) of -0.48, interpreted as a small-medium association in favour of the ketogenic diet compared with control diets.

In trials that used biochemical monitoring of ketosis (measurement of ketone body levels), the effect was more pronounced, with an SMD of -0.88. In trials without ketone monitoring, the association was small and statistically insignificant. Subgroup comparison showed a difference of -0.84, suggesting a potentially modifying role of ketosis control.

When studies were divided by the presence of a high-carbohydrate diet in the control group, papers without this comparison showed a large and significant association with improvement in depressive symptoms (SMD -1.49), whereas studies with high-carbohydrate comparator diets showed no significant effect. The difference in standardised mean differences between these subgroups was -1.37.

Obesity status also mattered. In the group without obesity, pooled trial data showed a large and statistically significant association (SMD -0.88), whereas in obese participants the association was small and statistically insignificant (-0.11).

The intensity of the intervention was also found to be important. Very low-carbohydrate diets (10% of energy from carbohydrates) formed a separate subgroup, whereas low-carbohydrate diets provided 11-20% of energy from carbohydrates.

Very low-carbohydrate regimes showed a significant large association with improvement in depressive symptoms (SMD -0.79), whereas less stringent low-carbohydrate diets showed a nonsignificant effect (-0.05). The difference between these subgroups was -0.75. Dividing by intervention duration (4-6 weeks, 8-9 weeks, 10 weeks or more) showed no significant differences between subgroups.

Quasi-experimental studies confirmed the "signal" on depression. Nine such studies showed a pooled standardised mean difference of -0.66, which the authors interpret as a medium strength association. In studies with ketone monitoring, the SMD was -0.62 and without monitoring -0.88, but no statistically significant differences between subgroups were found.

Mixed results for anxiety and other symptoms

The anxiety-focused randomised clinical trials included nine studies and 672 participants. Pooled analysis yielded an SMD of -0.03, interpreted as no significant association between ketogenic diets and change in anxiety symptoms compared with comparator diets.

Subgroup analyses by ketone control traits, use of low-fat diets as a control, obesity status, level of carbohydrate restriction, and duration of intervention showed no consistent modifying factors; no subgroup with high-carbohydrate comparators showed a significant improvement in anxiety.

The quasi-experimental data looked somewhat different. In six of these studies that assessed anxiety symptoms, the pooled standardised mean difference was -0.58, indicating an improvement in anxiety of moderate strength within the groups themselves on the ketogenic intervention. However, subgroup analysis showed no significant differences according to ketone control, presence of neurological disease, other comorbid conditions, or duration of intervention.

The authors note that depressive and anxiety symptoms tended to improve across populations - in major depressive disorder, generalised anxiety disorder, PTSD, schizophrenia, bipolar disorder and in non-psychiatric medical groups.

Reports from case series and single clinical observations have described reductions in psychotic symptoms and mood stabilisation in people with schizophrenia or bipolar spectrum disorders following ketogenic or closely related regimens. One pilot study in PTSD showed significant reductions in symptoms, but the authors emphasise the need for further trials in trauma-associated conditions.

Cautious conclusions for practice

The researchers conclude that ketogenic diets may have therapeutic potential for depressive symptoms, with more pronounced results observed where nutritional ketosis is biochemically confirmed. At the same time, data on anxiety in randomised trials remain preliminary and equivocal.

The larger improvements recorded in quasi-experimental studies, according to the authors, may be due to both stricter adherence to the ketodiet protocol and the greater vulnerability of such designs to systematic errors.

The generalisability of the findings is limited by variations in diet composition, types of regimens compared, level of adherence support, symptom rating scales used and study quality, as well as the short follow-up period in many of the papers. The authors emphasise that the pooled estimates do not reflect a single universal effect and do not prove that the ketogenic diet is the cause of improvement; changes in symptoms are likely to vary between patients.