State of the evidence in mental health, 2026
Every entry on Shrinkopedia carries an evidence rating: Strong, Mixed or Limited. Counted across the whole encyclopedia, 29 of 66 conditions (44%) and 19 of 26 treatments (73%) rate Strong. 7 conditions rate Limited, and no treatment does. The full dataset is free to download and reuse with credit.
What the numbers say
Most of what people look up has a real research base behind it. 89% of conditions rate Strong or Mixed, which means there's at least a body of guidelines, trials or careful clinical description to work from.
The thin spots are concentrated. 5 clusters hold all 7 of the Limited ratings, and personality disorders account for 3 of them. Within that one cluster the gap is wide. Borderline personality disorder rates Strong, with several therapies tested in randomized trials, while schizoid and histrionic personality disorder rate Limited.
No treatment on the site rates Limited. The 7 rated Mixed are listed below. Mixed doesn't mean a treatment doesn't work. It means the studies don't yet agree on how well, or for whom.
Conditions by clinical cluster
| Cluster | Strong | Mixed | Limited | Total |
|---|---|---|---|---|
| Anxiety and related disorders | 6 | 1 | 0 | 7 |
| Obsessive-compulsive and related | 2 | 4 | 0 | 6 |
| Depressive and mood disorders | 5 | 1 | 1 | 7 |
| Trauma and stress-related | 1 | 2 | 0 | 3 |
| Psychotic and related | 1 | 2 | 1 | 4 |
| Personality disorders | 1 | 5 | 3 | 9 |
| Dissociative disorders | 0 | 3 | 0 | 3 |
| Sleep-wake disorders | 3 | 2 | 0 | 5 |
| Neurodevelopmental | 2 | 3 | 0 | 5 |
| Disruptive, impulse-control, and conduct | 1 | 0 | 1 | 2 |
| Neurocognitive | 0 | 1 | 0 | 1 |
| Substance use and addictive | 3 | 2 | 0 | 5 |
| Eating and feeding | 4 | 0 | 1 | 5 |
| Somatic symptom and related | 0 | 1 | 0 | 1 |
| Perinatal mental health | 1 | 3 | 0 | 4 |
| Grief, burnout, and other | 0 | 2 | 0 | 2 |
Some conditions sit in two clusters (postpartum depression appears under mood and perinatal, for example), so cluster totals add up to more than 66.
Conditions rated Limited
Limited means there's little solid evidence either way. It says nothing about whether a condition is real or whether it can be treated. People with these diagnoses are treated every day; the treatments just haven't been tested in many good trials.
- ARFID (avoidant/restrictive food intake disorder)
- Delusional disorder
- DMDD (disruptive mood dysregulation disorder)
- Histrionic personality disorder
- Intermittent explosive disorder
- Paranoid personality disorder
- Schizoid personality disorder
Treatments not rated Strong
- Acceptance and commitment therapy (ACT) (Mixed)
- Family therapy (Mixed)
- Group therapy (Mixed)
- Ketamine therapy (Mixed)
- Lifestyle interventions (Mixed)
- Mindfulness-based therapy (Mixed)
- Psychodynamic therapy (Mixed)
Every condition and treatment
Conditions (66)
Treatments (26)
| Treatment | Rating | Last medically reviewed |
|---|---|---|
| Acceptance and commitment therapy (ACT) | Mixed | May 24, 2026 |
| Antipsychotics | Strong | June 8, 2026 |
| Behavioral activation | Strong | May 24, 2026 |
| Benzodiazepines | Strong | June 8, 2026 |
| Cognitive behavioral therapy (CBT) | Strong | May 24, 2026 |
| Dialectical behavior therapy | Strong | July 13, 2026 |
| ECT | Strong | May 24, 2026 |
| EMDR | Strong | May 24, 2026 |
| Exposure therapy | Strong | May 24, 2026 |
| Family therapy | Mixed | May 24, 2026 |
| Group therapy | Mixed | May 24, 2026 |
| Interpersonal psychotherapy | Strong | July 11, 2026 |
| Interpersonal therapy (IPT) | Strong | May 24, 2026 |
| Ketamine and esketamine | Strong | July 11, 2026 |
| Ketamine therapy | Mixed | June 8, 2026 |
| Lamotrigine | Strong | July 11, 2026 |
| Lifestyle interventions | Mixed | May 24, 2026 |
| Lithium | Strong | July 11, 2026 |
| Metformin | Strong | June 8, 2026 |
| Mindfulness-based therapy | Mixed | May 24, 2026 |
| Mood stabilizers | Strong | June 8, 2026 |
| Psychodynamic therapy | Mixed | May 24, 2026 |
| Psychotherapy | Strong | May 24, 2026 |
| SNRIs | Strong | June 8, 2026 |
| SSRIs | Strong | June 8, 2026 |
| TMS | Strong | May 24, 2026 |
How the ratings are made
Each entry is rated once, as a whole, when it's written, and the rating is checked again at every medical review. We work down a fixed hierarchy of sources: systematic reviews and clinical practice guidelines first, then randomized controlled trials, then observational and cohort studies, and expert consensus last. Strong means consistent, good-quality evidence. Mixed means the evidence points somewhere but isn't consistent, or a topic is well described clinically without a large research base behind it. Limited means there's little solid evidence either way. The full policy is on how we review evidence.
Comparison pages (such as bipolar I vs bipolar II) are left out of the counts because they rate two conditions at once. Symptoms and terms are in the dataset but not in the headline figures. For them the rating describes how well studied the concept is, which isn't the same question.
What this doesn't prove
A rating describes the research behind an entry. It doesn't predict how any one person will respond to a treatment, and a Strong rating doesn't mean a condition is easy to treat.
These are editorial ratings made by one team against a written hierarchy. They aren't a formal GRADE assessment, and a different team could rate a handful of entries differently. We publish every rating so anyone can check them.
The encyclopedia covers what readers look up, not a random sample of psychiatry. The percentages describe these 66 conditions and 26 treatments, not the field as a whole.
Download the data
The full dataset has one row per entry: type, title, URL, rating, the kind of evidence it rests on, clinical cluster and every review date.
- evidence-ratings.csv (157 rows)
- evidence-ratings.json (same data, with totals)
Free to reuse under Creative Commons Attribution 4.0. Credit "Shrinkopedia, medically reviewed by Shariq Refai, MD, MBA" and link to this page. See reuse and embedding for the rest of what you can use freely. Journalists can find background and contact details on the press page.
We update this report once a year. Ratings change between editions when an entry's evidence moves, and those changes are logged on corrections and updates.
Educational reference, not medical advice. If you're in crisis in the US, call or text 988, or call 911 in an emergency.
How to cite this page
- Short
State of the evidence in mental health, 2026. Shrinkopedia, medically reviewed by Shariq Refai, MD, MBA. https://shrinkopedia.com/research/state-of-the-evidence/- APA
Shrinkopedia. (2026, September 24). State of the evidence in mental health, 2026. Medically reviewed by Shariq Refai, MD, MBA. https://shrinkopedia.com/research/state-of-the-evidence/- MLA
"State of the evidence in mental health, 2026." Shrinkopedia, 24 Sept. 2026, https://shrinkopedia.com/research/state-of-the-evidence/.
Click a citation to select it.
Your next step in The Shrink Network
- For plain-language summaries of individual studies, including what each one doesn't prove, see AnxietyResearch.org.
- For the editorial standards every Shrink Network site follows, see ShrinkNetwork.com.