Study analysis · JAMA Psychiatry · 2026
This diet slashed depression in 6 weeks—then vanished. Here’s why ketones had nothing to do with it.
A super strict low-carb diet made people with severe depression feel a little better for 6 weeks, but then it wore off—and it wasn’t because of ketones.
Overview
What the study found
The study in plain English — the bottom line, every takeaway we extracted, and what to do with them.
In simple terms
This study is like a fair test where two groups of people with serious depression tried different diets. One group ate very low-carb food, and the other ate more veggies and healthy fats. The low-carb group felt a little better after 6 weeks, but not by much—and after 12 weeks, both groups felt the same. So we can't say the low-carb diet definitely caused the improvement.
What’s the bottom line?
Scientists tested if a strict low-carb diet (ketogenic diet) helps people with depression that didn't improve with medicine, compared to a healthy plant-based diet.
How strong is this study?
This study was pretty well done because it randomly assigned people to groups, made sure both groups got the same amount of help, and followed up with almost everyone. But it gave the low-carb group free meals and more money, which might have made people feel better for reasons other than the diet. So while it's one of the best types of studies, we still need to be careful about trusting the results too much.
0 / 100
- COI disclosureconflicts of interest not disclosed
- Data availabilitydata not shared
- Code availabilitycode not shared
76 / 100
- Randomization+20/20
- Blinding+9/15
- Control group+15/15
- Sample size (n=88)+7.1/20
- Follow-up+10/10
100 / 100
100 / 100
- P-values+15/15
- Effect size+20/20
- Confidence intervals+15/15
- Pre-registration+15/15
Each component is scored out of 100 and then capped by the study design — a case series cannot reach the ceiling a randomised trial can, however well it is reported.
Where it sits
RCT reviewsReviews of RCTs (Meta-analyses)
Max 100Randomized TrialsRandomized Trials
Max 90Reviews of Cohort StudiesReviews of Cohort Studies
Max 85Cohort StudiesCohort Studies
Max 72Reviews of Case-Control StudiesReviews of Case-Control Studies
Max 63Case-Control StudiesCase-Control Studies
Max 58Cross-Sectional & Case SeriesCross-Sectional & Case Series
Max 50Expert OpinionExpert Opinion
Max 572 / 100
Probability of being correct
Participants are randomly assigned to treatment or control groups, minimizing bias. The gold standard for testing whether an intervention causes an effect.
This design can establish causation. Although this is a randomized controlled trial, the effect size is small, the primary outcome is borderline significant (p=0.05), and the effect disappears at 12 weeks and in sensitivity analyses using alternative models. The control group also received substantial support, making it difficult to isolate the specific effect of ketosis. Therefore, while causation is plausible, it is not definitively established.
No Conflicts
No conflicts of interest identified
No conflicts of interest or funding sources were disclosed in the provided text, and there is no evidence of industry involvement or author affiliations that suggest bias.
Independent Analysis Safeguards
- The data analyst (M.G.) did not deliver the intervention.
- Statistical analysis plan was finalized prior to database locking.
- Analyses were conducted using Stata and R with intention-to-treat and prespecified methods.
- Sensitivity analyses were performed to assess robustness of results.
Although the study appears methodologically rigorous with blinding, pre-specified analysis, and independent data analysis, the absence of any declared funding or conflict of interest section raises a minor transparency concern. However, without evidence of industry ties or funder influence, the overall risk of bias is assessed as low.
Key takeaways
- 01
After 6 weeks, people on the low-carb diet felt slightly better (PHQ-9 score dropped 2.2 points more than the other group).
- 02
But after 12 weeks, both groups felt the same.
- 03
The better mood only happened in people with very severe depression (score ≥20), and it had nothing to do with ketones in their urine.
- 04
A 2-point drop on the PHQ-9 scale is small — it’s not enough to feel dramatically better for most people, and it didn’t last.
Surprising findings
- The control group improved just as much as the keto group over time—both dropped over 8 points on the PHQ-9 scale.People assume the keto group’s improvement was due to diet—but the control group (just eating more veggies and swapping fats) improved nearly as much, suggesting the real driver was attention, hope, or structure.
- Only 20% of keto participants kept the diet after support ended—even though they had free meals and coaching.Keto advocates claim it’s sustainable—but here, even with perfect support, 80% quit within 6 weeks. This isn’t a lifestyle; it’s a short-term experiment.
Practical takeaways
If you have severe treatment-resistant depression (PHQ-9 ≥20), consider a short-term, supervised keto trial—only if you can get free meals and weekly coaching.
The benefit fades after 6 weeks, doesn’t improve anxiety or cognition, and 80% of people quit after support ends. It’s not a long-term solution.
medium confidenceWhy this study matters
The 6-Week Miracle That Faded
In an 88-person trial, people on a ketogenic diet (<30g carbs/day) saw a 2.18-point greater drop in depression scores (PHQ-9) than those on a plant-rich diet at 6 weeks (P=0.05). But by 12 weeks, the difference disappeared—both groups felt the same.
People think keto fixes mental health long-term—but this study proves the effect is fleeting, even with perfect adherence and free meals. It’s not a cure, just a temporary bump.
Only Severe Cases Benefited
The diet’s benefit was almost entirely in people with the worst depression (PHQ-9 ≥20), where scores dropped 4.73 points more than the control group. Those with moderate depression (PHQ-9 15–19) saw zero improvement.
If you’re mildly depressed, keto won’t help—but if you’re in crisis, it might give you a short lifeline. This isn’t a universal fix; it’s a targeted tool for the most desperate cases.
Ketones Didn’t Cause the Mood Boost
Despite measuring urine ketones twice weekly, researchers found zero correlation between ketone levels and mood improvement. People with high ketones didn’t feel better than those with low or none.
The whole keto-for-brain theory hinges on ketones—but this study blows that up. The benefit came from something else: maybe the structure, the support, or the placebo effect.
Want the whole report?
Detailed mode opens the full scientific breakdown — every score component, the methodology, conflicts of interest, the evidence analysis behind each claim, and the raw study data.
Overview
What the study found
The study in plain English — the bottom line, every takeaway we extracted, and what to do with them.
Not medical advice. For informational purposes only. Always consult a healthcare professional. Terms
Scientists tested if a strict low-carb diet (ketogenic diet) helps people with depression that didn't improve with medicine, compared to a healthy plant-based diet.
Research results
After 6 weeks, people on the low-carb diet felt slightly better (PHQ-9 score dropped 2.2 points more than the other group). But after 12 weeks, both groups felt the same. The better mood only happened in people with very severe depression (score ≥20), and it had nothing to do with ketones in their urine.
What this means - more context
A 2-point drop on the PHQ-9 scale is small — it’s not enough to feel dramatically better for most people, and it didn’t last.
To assess whether a ketogenic diet (KD) improves depression in adults with treatment-resistant depression (TRD) compared to a matched control diet.
In an 88-participant RCT, a KD with <30g carbs/day and intensive support reduced PHQ-9 scores by 2.18 points more than a phytochemical-rich control diet at 6 weeks (P=0.05), but this effect vanished at 12 weeks. No benefits were seen for anxiety, cognition, or quality of life. Ketone levels did not correlate with mood changes. The effect was stronger in severe depression (PHQ-9≥20), with a 4.73-point difference at 6 weeks.
Methods Used
Single-blind RCT with 88 adults with TRD (PHQ-9≥15) randomized 1:1 to a ketogenic diet (<30g carbs/day, prepared meals, weekly dietetic support) or a phytochemical-rich control diet (extra fruits/vegetables, unsaturated fats, matched support). Primary outcome: change in PHQ-9 at 6 weeks; secondary outcomes at 12 weeks. Intention-to-treat analysis with linear mixed-effects models.
Main Finding
A ketogenic diet produced a modest, statistically significant reduction in depression severity (mean PHQ-9 difference: −2.18 points; 95% CI, −4.33 to −0.03; P=0.05) at 6 weeks compared to a control diet, but this effect was not sustained at 12 weeks (−1.85 points; P=0.10) and was not correlated with ketone levels. The benefit was significantly stronger in severe depression (PHQ-9≥20): −4.73 points at 6 weeks.
Confidence Level
Moderate. This is a well-conducted RCT with high follow-up (93%), intention-to-treat analysis, and matched control. However, the primary effect was borderline significant (P=0.05), sensitivity analyses weakened it, and the clinical relevance is uncertain due to small effect size and lack of secondary outcome benefits.
Study Flags
Red Flags
- •Primary outcome was borderline significant (P=0.05)
- •Effect not sustained at 12 weeks
- •No correlation between ketone levels and mood improvement
No biological mechanisms were identified in this study. This may be an epidemiological, observational, or survey-based study that reports associations rather than proposing causal biological pathways.
Surprising Findings
The control group improved just as much as the keto group over time—both dropped over 8 points on the PHQ-9 scale.
People assume the keto group’s improvement was due to diet—but the control group (just eating more veggies and swapping fats) improved nearly as much, suggesting the real driver was attention, hope, or structure.
Practical Takeaways
If you have severe treatment-resistant depression (PHQ-9 ≥20), consider a short-term, supervised keto trial—only if you can get free meals and weekly coaching.
RCT reviewsReviews of RCTs (Meta-analyses)
Max 100Randomized TrialsRandomized Trials
Max 90Reviews of Cohort StudiesReviews of Cohort Studies
Max 85Cohort StudiesCohort Studies
Max 72Reviews of Case-Control StudiesReviews of Case-Control Studies
Max 63Case-Control StudiesCase-Control Studies
Max 58Cross-Sectional & Case SeriesCross-Sectional & Case Series
Max 50Expert OpinionExpert Opinion
Max 572 / 100
Probability of being correct
Participants are randomly assigned to treatment or control groups, minimizing bias. The gold standard for testing whether an intervention causes an effect.
Human RCT
Subject
High probability
on the GRADE evidence scale
This study is like a fair test where two groups of people with serious depression tried different diets. One group ate very low-carb food, and the other ate more veggies and healthy fats. The low-carb group felt a little better after 6 weeks, but not by much—and after 12 weeks, both groups felt the same. So we can't say the low-carb diet definitely caused the improvement.
No conflicts of interest were detected in this study. No score impact.
Strengths
- Randomized controlled trial with concealed allocation
- High follow-up rate (93% at 12 weeks)
- Active control group with matched support and blinding
Weaknesses
- Primary outcome p-value is borderline (p=0.05)
- Effect size (Cohen d = -0.68) is modest and below prespecified clinical threshold
- No significant effect at 12 weeks or in per-protocol analysis
Methodology
Evidence Keywords
Statistical Reporting
Not medical advice. For informational purposes only. Always consult a healthcare professional. Terms
Scientists tested if a strict low-carb diet (ketogenic diet) helps people with depression that didn't improve with medicine, compared to a healthy plant-based diet.
Research results
After 6 weeks, people on the low-carb diet felt slightly better (PHQ-9 score dropped 2.2 points more than the other group). But after 12 weeks, both groups felt the same. The better mood only happened in people with very severe depression (score ≥20), and it had nothing to do with ketones in their urine.
What this means - more context
A 2-point drop on the PHQ-9 scale is small — it’s not enough to feel dramatically better for most people, and it didn’t last.
To assess whether a ketogenic diet (KD) improves depression in adults with treatment-resistant depression (TRD) compared to a matched control diet.
In an 88-participant RCT, a KD with <30g carbs/day and intensive support reduced PHQ-9 scores by 2.18 points more than a phytochemical-rich control diet at 6 weeks (P=0.05), but this effect vanished at 12 weeks. No benefits were seen for anxiety, cognition, or quality of life. Ketone levels did not correlate with mood changes. The effect was stronger in severe depression (PHQ-9≥20), with a 4.73-point difference at 6 weeks.
Methods Used
Single-blind RCT with 88 adults with TRD (PHQ-9≥15) randomized 1:1 to a ketogenic diet (<30g carbs/day, prepared meals, weekly dietetic support) or a phytochemical-rich control diet (extra fruits/vegetables, unsaturated fats, matched support). Primary outcome: change in PHQ-9 at 6 weeks; secondary outcomes at 12 weeks. Intention-to-treat analysis with linear mixed-effects models.
Main Finding
A ketogenic diet produced a modest, statistically significant reduction in depression severity (mean PHQ-9 difference: −2.18 points; 95% CI, −4.33 to −0.03; P=0.05) at 6 weeks compared to a control diet, but this effect was not sustained at 12 weeks (−1.85 points; P=0.10) and was not correlated with ketone levels. The benefit was significantly stronger in severe depression (PHQ-9≥20): −4.73 points at 6 weeks.
Confidence Level
Moderate. This is a well-conducted RCT with high follow-up (93%), intention-to-treat analysis, and matched control. However, the primary effect was borderline significant (P=0.05), sensitivity analyses weakened it, and the clinical relevance is uncertain due to small effect size and lack of secondary outcome benefits.
Study Flags
Red Flags
- •Primary outcome was borderline significant (P=0.05)
- •Effect not sustained at 12 weeks
- •No correlation between ketone levels and mood improvement
No biological mechanisms were identified in this study. This may be an epidemiological, observational, or survey-based study that reports associations rather than proposing causal biological pathways.
Surprising Findings
The control group improved just as much as the keto group over time—both dropped over 8 points on the PHQ-9 scale.
People assume the keto group’s improvement was due to diet—but the control group (just eating more veggies and swapping fats) improved nearly as much, suggesting the real driver was attention, hope, or structure.
Practical Takeaways
If you have severe treatment-resistant depression (PHQ-9 ≥20), consider a short-term, supervised keto trial—only if you can get free meals and weekly coaching.
RCT reviewsReviews of RCTs (Meta-analyses)
Max 100Randomized TrialsRandomized Trials
Max 90Reviews of Cohort StudiesReviews of Cohort Studies
Max 85Cohort StudiesCohort Studies
Max 72Reviews of Case-Control StudiesReviews of Case-Control Studies
Max 63Case-Control StudiesCase-Control Studies
Max 58Cross-Sectional & Case SeriesCross-Sectional & Case Series
Max 50Expert OpinionExpert Opinion
Max 572 / 100
Probability of being correct
Participants are randomly assigned to treatment or control groups, minimizing bias. The gold standard for testing whether an intervention causes an effect.
Human RCT
Subject
High probability
on the GRADE evidence scale
This study is like a fair test where two groups of people with serious depression tried different diets. One group ate very low-carb food, and the other ate more veggies and healthy fats. The low-carb group felt a little better after 6 weeks, but not by much—and after 12 weeks, both groups felt the same. So we can't say the low-carb diet definitely caused the improvement.
No conflicts of interest were detected in this study. No score impact.
Strengths
- Randomized controlled trial with concealed allocation
- High follow-up rate (93% at 12 weeks)
- Active control group with matched support and blinding
Weaknesses
- Primary outcome p-value is borderline (p=0.05)
- Effect size (Cohen d = -0.68) is modest and below prespecified clinical threshold
- No significant effect at 12 weeks or in per-protocol analysis
Methodology
Evidence Keywords
Statistical Reporting
Scoring
How strong is this study?
This study was pretty well done because it randomly assigned people to groups, made sure both groups got the same amount of help, and followed up with almost everyone. But it gave the low-carb group free meals and more money, which might have made people feel better for reasons other than the diet. So while it's one of the best types of studies, we still need to be careful about trusting the results too much.
0 / 100
- COI disclosureconflicts of interest not disclosed
- Data availabilitydata not shared
- Code availabilitycode not shared
76 / 100
- Randomization+20/20
- Blinding+9/15
- Control group+15/15
- Sample size (n=88)+7.1/20
- Follow-up+10/10
100 / 100
100 / 100
- P-values+15/15
- Effect size+20/20
- Confidence intervals+15/15
- Pre-registration+15/15
Each component is scored out of 100 and then capped by the study design — a case series cannot reach the ceiling a randomised trial can, however well it is reported.
Where it sits
RCT reviewsReviews of RCTs (Meta-analyses)
Max 100Randomized TrialsRandomized Trials
Max 90Reviews of Cohort StudiesReviews of Cohort Studies
Max 85Cohort StudiesCohort Studies
Max 72Reviews of Case-Control StudiesReviews of Case-Control Studies
Max 63Case-Control StudiesCase-Control Studies
Max 58Cross-Sectional & Case SeriesCross-Sectional & Case Series
Max 50Expert OpinionExpert Opinion
Max 572 / 100
Probability of being correct
Participants are randomly assigned to treatment or control groups, minimizing bias. The gold standard for testing whether an intervention causes an effect.
This design can establish causation. Although this is a randomized controlled trial, the effect size is small, the primary outcome is borderline significant (p=0.05), and the effect disappears at 12 weeks and in sensitivity analyses using alternative models. The control group also received substantial support, making it difficult to isolate the specific effect of ketosis. Therefore, while causation is plausible, it is not definitively established.
No Conflicts
No conflicts of interest identified
No conflicts of interest or funding sources were disclosed in the provided text, and there is no evidence of industry involvement or author affiliations that suggest bias.
Independent Analysis Safeguards
- The data analyst (M.G.) did not deliver the intervention.
- Statistical analysis plan was finalized prior to database locking.
- Analyses were conducted using Stata and R with intention-to-treat and prespecified methods.
- Sensitivity analyses were performed to assess robustness of results.
Although the study appears methodologically rigorous with blinding, pre-specified analysis, and independent data analysis, the absence of any declared funding or conflict of interest section raises a minor transparency concern. However, without evidence of industry ties or funder influence, the overall risk of bias is assessed as low.