Does Spravato work? What the trials found

Nearly all the efficacy data comes from three studies. Knowing what each one actually asked matters more than any single percentage.

TRANSFORM-2 — the four-week efficacy trial

Adults aged 18 to under 65 with treatment-resistant depression. Everyone was started on a new oral antidepressant; participants were randomised to add either esketamine nasal spray or a placebo spray. Four weeks.

Primary endpoint, from the FDA label: change in MADRS depression score at week 4. The esketamine group improved by 4.0 points more than placebo (95% confidence interval: 0.6 to 7.3). MADRS runs 0 to 60. A four-point difference is real and it is statistically significant; it is also, on its own, a moderate effect.

Response and remission at day 28, from the FDA's own medical review: response 69% on esketamine versus 52% on placebo; remission 53% versus 31% (analysis set n=101 and n=100).

Note what "placebo" means here. It does not mean untreated. Both arms began a new oral antidepressant, which is why over half the placebo group responded. Any page quoting "69% response" without that comparator is telling you something misleading.

One more honesty note: ClinicalTrials.gov reports different figures for the same trial — 63.4% versus 49.5% for response, 48.2% versus 30.3% for remission — on a larger analysis set. Both are correct for their respective analysis sets. We quote the FDA review's figures and name the set rather than picking whichever pair looks better.

SUSTAIN-1 — the relapse-prevention trial

A different question: among people who had already responded, does continuing esketamine delay relapse? Published in JAMA Psychiatry in 2019.

Among stable remitters, 26.7% relapsed on continued esketamine versus 45.3% on the oral antidepressant alone (hazard ratio 0.49; 95% CI 0.29–0.84). Among stable responders, 25.8% versus 57.6% (hazard ratio 0.30; 95% CI 0.16–0.55).

This is the most encouraging dataset for the drug, and it deserves a caveat: it only describes people who had already responded. It says nothing about whether you will be one of them.

The monotherapy trial

The 2025 study supporting the monotherapy approval, published in JAMA Psychiatry. Esketamine alone versus placebo, no oral antidepressant in either arm.

Primary endpoint, from the FDA label: MADRS change at day 28. 56 mg improved by 5.1 points more than placebo (95% CI 2.3 to 7.9); 84 mg by 6.8 points more (95% CI 4.1 to 9.5). Participants: 86, 95, and 197 respectively.

We are not publishing response or remission percentages for this trial. The sources we could reach disagree, because they use different thresholds for remission — and quoting a remission rate without knowing which definition produced it would be exactly the kind of borrowed-number error this site exists to avoid. The MADRS changes above are from the label and are solid.

These two words get used interchangeably in marketing copy and mean quite different things clinically.

Response conventionally means at least a 50% reduction in depression score from where you started. If you began severely depressed, a 50% improvement can still leave you meaningfully depressed. Response is real improvement; it is not recovery.

Remission means your score has dropped below a threshold taken to indicate minimal symptoms. It is a higher bar, and it is closer to what most people mean when they ask whether something "worked".

Notice that remission rates in these trials are consistently lower than response rates — 53% versus 69% in TRANSFORM-2. When a page quotes a single impressive percentage without saying which of these it is, it is usually the response figure.

There is also a definitional wrinkle worth knowing: different studies set the remission threshold at different MADRS scores. That is precisely why we are not quoting remission percentages from the monotherapy trial — a rate is meaningless without the definition attached.

Every limitation below is a reason to hold the numbers above loosely.

Trial populations are selected. Participants met specific entry criteria, and people with significant medical comorbidities, substance use disorders, or certain psychiatric conditions were typically excluded. If that describes you, the trial population was not people like you.

Follow-up is bounded. TRANSFORM-2 ran four weeks. Depression is measured in years. A four-week result is not a statement about where you will be in two years.

The comparator was not "nothing". Both arms of TRANSFORM-2 started a new oral antidepressant. The trials measure what esketamine adds, not what it does from a standing start.

Group averages hide individual variation. A 4-point average difference is compatible with some participants improving dramatically and others not at all. The average describes the group; nobody experiences the average.

These trials do not compare Spravato to the alternatives. Nothing here tells you whether Spravato would work better for you than ECT, TMS, another medication, a different combination, or psychotherapy. Those comparisons largely have not been done head to head, and we will not manufacture them.

Nobody can tell you your odds. Not this site, and not your prescriber — though your prescriber can at least weigh your specific history, which we cannot.

If you are reading a page about success rates after failing several medications already, you have probably had this experience: something was presented as promising, you allowed yourself to hope, and it did not work. It would be dishonest to end this page without acknowledging that Spravato is sometimes that too.

Some things that are true and are not consolation prizes.

Not responding to Spravato is not a verdict on you. Treatment response in depression is poorly predicted by anything currently measurable. A medication not working is information about the medication in your particular biology, not evidence about your character or your prospects.

It is not the last option. ECT has the strongest efficacy evidence of any treatment for severe treatment-resistant depression and remains substantially under-used, partly because of a public reputation that is decades out of date. TMS, medication strategies not yet tried, augmentation approaches, and structured psychotherapy alongside medication are all real avenues. There are also treatments in trials that were not available a few years ago.

The four-week assessment point is a decision point, not a deadline. If response is partial, dose and schedule adjustments are a normal part of the process rather than a sign of failure.

Say it plainly to your prescriber. "This isn't working and I'm running out of hope" is clinically important information, not a complaint. It is also worth saying before things get worse rather than after.

And if you are in a bad place right now: call or text 988. It is not only for people who are actively suicidal — it is for anyone in distress, including someone worn down by a treatment that did not work.