Research
Ketamine Therapy Statistics
Ketamine therapy statistics describe different treatments, patients and outcomes; they do not add up to one reliable success rate. This reference separates intravenous ketamine, at-home sublingual ketamine and esketamine nasal spray, and keeps each result attached to its sample size and follow-up period.
Sources checked November 17, 2026. This is a selected evidence reference, not a systematic review, a treatment recommendation or an estimate of your chance of improving.
Download the ketamine study data (CSV). Each row includes its original source, endpoint, denominator information and an interpretation limit.
Ketamine study results at a glance
These results belong to the study named in each row. Differences between rows are not head-to-head comparisons.
| Study and treatment | Population and sample | Result and time point | Main limit |
|---|---|---|---|
| Zarate et al., 2006: intravenous ketamine | Treatment-resistant major depression; 18 enrolled, 17 received ketamine | The day after ketamine, 71% met response criteria and 29% met remission criteria; 35% maintained response for at least one week | Small crossover experiment; a one-day result is not a long-term outcome |
| Murrough et al., 2013: intravenous ketamine vs midazolam | Treatment-resistant major depression; 73 randomized | At 24 hours, response was reported in 64% with ketamine vs 28% with midazolam | Single infusion; active comparator; short primary follow-up |
| ELEKT-D, 2023: intravenous ketamine vs ECT | Treatment-resistant major depression without psychosis; 403 randomized, 365 started assigned treatment | Over the three-week treatment phase, response was 55.4% with ketamine vs 41.2% with ECT | Open-label noninferiority trial in people referred to ECT clinics |
| Hull et al., 2022: at-home sublingual ketamine with support | 1,247 treatment completers with sufficient data | Over four weeks, 62.8% reported at least a 50% improvement on the PHQ-9 depression questionnaire | Uncontrolled completer analysis; self-reported symptoms |
| Mathai et al., 2024: at-home telehealth-supported ketamine | Real-world depression data; 11,441 patients analyzed | Adverse events were detected in 3.0%–4.8% of participants in the reported analyses | No control group; flexible treatment; event detection is not complete risk measurement |
| ESCAPE-TRD, 2023: esketamine vs quetiapine | Treatment-resistant depression; 676 randomized; both groups continued an SSRI or SNRI | Week-eight remission: 91/336 (27.1%) with esketamine vs 60/340 (17.6%) with quetiapine | Different medicine and endpoint from intravenous or oral ketamine studies |
| Janik et al., 2025: esketamine monotherapy | Adults with treatment-resistant depression; 378 in the reported efficacy population | At day 28, the adjusted MADRS score difference vs placebo was −5.1 points for 56 mg and −6.8 points for 84 mg | A symptom-scale difference is not a response percentage |
What does “ketamine therapy success rate” mean?
A success rate is meaningful only after “success” is defined. A study may count a large symptom reduction, a low final symptom score, continued improvement months later, or another outcome.
| Term | What it describes | What it does not establish |
|---|---|---|
| Response | Meeting the study's improvement threshold | Full symptom relief or permanent recovery |
| Remission | Reaching the study's low-symptom threshold | A cure or freedom from future episodes |
| Mean score change | The group's average change on a symptom scale | The share of people who improved |
| Sustained response | Continuing to meet an improvement rule across specified visits | Benefit beyond the observation period |
| Treatment completion | Finishing the course or assessments used in an analysis | That everyone who started benefited |
| Adverse-event rate | Events identified under that study's monitoring rules | A complete measure of all short- and long-term harms |
The same number can answer very different questions. “Half of the group had a response at one day” is different from “half remained well at six months.” Always carry the time point with the statistic.
Why intravenous and at-home results should stay separate
Route and setting are part of the evidence. A result for an infusion in a clinic does not establish the result for an oral lozenge taken at home. A telehealth program's results also cannot be assumed to apply to every other program.
The at-home studies above included clinical and psychosocial support; they did not study unsupervised recreational use. The 2022 report uses treatment completers with adequate data, while the 2024 analysis describes a larger real-world dataset without a control group. These are observations of care, not randomized proof that the medication alone caused the change. 2022 study; 2024 study.
Keeping those designs visible prevents two mistakes: treating all improvement as a drug effect, and treating a selected group of respondents as everyone who began treatment.
What the larger controlled trials add
Controlled trials help answer a more specific question: what happened under the tested protocol compared with the assigned alternative?
ELEKT-D tested noninferiority to ECT in nonpsychotic treatment-resistant depression. Its reported response gap was 14.2 percentage points, with a 95% confidence interval from 3.9 to 24.2 points. That finding should not be broadened to every person for whom ECT might be considered. Original trial.
ESCAPE-TRD tested esketamine nasal spray against extended-release quetiapine, with an oral antidepressant in both groups. Its week-eight outcome was remission, not the response endpoint used by ELEKT-D. Original trial.
The 2025 monotherapy trial reported 86 participants in the 56 mg group, 95 in the 84 mg group and 197 in the placebo group in its efficacy population. The respective placebo-adjusted day-28 differences had 95% confidence intervals of −7.91 to −2.33 and −9.48 to −4.07 MADRS points. Original trial.
These are useful results. They still answer different clinical questions, so this page does not combine them into a treatment ranking or pooled success percentage.
What can the safety statistics tell us?
Safety figures show what was recorded within a study's population, monitoring and observation period. They should be read alongside treatment benefits, not used as a claim that treatment is risk-free.
An at-home cohort cannot directly be compared with a clinic trial if the programs asked about events in different ways or followed people for different lengths of time. A low reported percentage also cannot rule out uncommon serious events.
The current SPRAVATO prescribing information carries boxed warnings for sedation, dissociation, respiratory depression, abuse and misuse, and suicidal thoughts and behaviors. It requires monitoring for at least two hours after administration and restricted REMS distribution. Those requirements describe SPRAVATO; they are not a safety certification for compounded ketamine.
The FDA also explains that compounded drugs are not FDA-approved: it does not verify their safety, effectiveness or quality before marketing. Approval, prescribing practice and evidence of benefit are separate questions.
How to check a ketamine statistic before using it
Before quoting a figure, find six details in the source:
- Treatment: the exact drug, route, setting and accompanying care.
- Population: diagnosis, selection rules and who was excluded.
- Denominator: randomized participants, treated participants, completers or people with a completed follow-up.
- Outcome: response, remission, score change, adverse event or another measure.
- Time: the assessment point and whether later follow-up exists.
- Design: randomized comparison, open-label observation, survey or administrative report.
For example, use “55.4% response during a three-week intravenous ketamine trial in nonpsychotic treatment-resistant depression,” with the study citation. Avoid turning it into “ketamine works for 55.4% of people.” The second wording drops the population, treatment and time period that make the number interpretable.
Data downloads and definitions
The study CSV records one selected endpoint per study, or per explicitly separated trial comparison. Blank numeric fields mean a value was not extracted; they do not mean zero. Percentages are stored as percentage values, so 55.4 means 55.4%, not a decimal probability.
The data dictionary defines the fields and explains which comparisons must stay separate. Sources remain the authority for the full protocol and results.
For psilocybin, MDMA and Oregon service-use data, see the separate psychedelic therapy statistics reference. Those datasets answer different questions and are not pooled with this one.
How this reference was prepared
We selected original studies that illustrate important differences in treatment route, research design, outcome and follow-up. We checked the linked reports or their original published abstracts, the current SPRAVATO label and FDA compounding information on November 17, 2026.
This is a curated reference. It does not claim to include every study, calculate an overall treatment effect, compare providers, estimate market size or measure all long-term risks. We did not collect patient records or conduct a clinical trial. Study publication years are shown beside the results; the check date does not make an older result new research.