Ketamine and Spravato, explained

A sixty-year-old anesthetic became the first new kind of antidepressant in a generation. Here is the whole story for patients and clinicians: where it came from, why the nasal spray is only half the molecule, what it does in a synapse, what the trials actually showed, how dose and route compare, how long it lasts, what else it might treat, and what we still don't know.

By Reid Robison, MD. 4,886 words, with a molecule you can turn over, a synapse you can scroll, and a calculator for the question I get most.

I have been working with ketamine since 2011, when a grant from the Intermountain Foundation let me run my first study of it and I led Janssen's intravenous ketamine trial in Salt Lake City. Since then I've been part of thousands of ketamine sessions and Spravato doses, piloted group ketamine-assisted psychotherapy at an eating disorder treatment center, and started the clinic that gave Utah its first Spravato. I say this not as a credential but as a disclosure: I am not neutral about this molecule. I have watched it do things nothing else in psychiatry does, and I have watched it do nothing at all, and I've spent a lot of the last fifteen years trying to tell in advance which it's going to be.

The question I get most, from patients and from colleagues, is some version of is Spravato just ketamine? The honest answer takes a drawing, a little chemistry and about twenty trials, so here it all is.

History: from the battlefield to the dolphin tank to the clinic

Ketamine was made in 1962 by Calvin Stevens at Parke-Davis, as a gentler cousin of phencyclidine, and first given to a human in 1964 by Edward Domino and Guenter Corssen in a Michigan prison.^1,2 Domino's wife, Toni, supplied the word that stuck when the company balked at "dreaming": dissociative. It was approved in 1970, and shipped to Vietnam in quantity, because it does something no other anesthetic does: it keeps the airway reflexes and blood pressure up while the patient goes somewhere else.³

• 1962Synthesized at Parke-Davis as CI-581.
• 1964First human doses, Jackson State Prison, Michigan. Volunteers describe floating, and being "in outer space."¹
• 1970FDA approval as an anesthetic; widely used in Vietnam as a battlefield anesthetic.
• 1970sJohn Lilly, inventor of the isolation tank, takes ketamine in it and comes back with reports of dolphin group minds. The psychonaut era begins; the term itself is Ernst Jünger's, from 1970.⁴
• 1978Marcia Moore's Journeys into the Bright World, the first book on ketamine as a psychedelic. She died the following winter after an overdose.⁵
• 1985–Evgeny Krupitsky in St Petersburg runs ketamine psychedelic therapy for alcoholism, a decade of work the West mostly ignored.⁶
• 1994John Krystal at Yale gives subanesthetic ketamine to healthy volunteers as a model of schizophrenia. The dose he settled on, 0.5 mg/kg over 40 minutes, is the one every depression trial would use.⁷
• 1990sClub use rises; the DEA makes it Schedule III in 1999.
• 2000Berman and colleagues at Yale give that same infusion to seven depressed patients. Depression lifts within hours. Nobody had seen that before.⁸
• 2006Zarate at the NIMH replicates it in treatment-resistant patients: 71% respond at 24 hours.⁹
• 2011Janssen begins developing the S-enantiomer as a nasal spray. I ran their intravenous study in Salt Lake City that year.
• 2019Spravato approved by the FDA for treatment-resistant depression (March), the first mechanistically new antidepressant since the 1980s; approval for depression with suicidal thinking follows in 2020, and as a monotherapy in 2025.¹⁰
• 2023Two NEJM trials in one year: esketamine beats quetiapine; IV ketamine is non-inferior to ECT in outpatients.^11,12

Three things about that timeline are worth holding onto. The dose came from a schizophrenia study, not a depression study. The psychedelic use came before the psychiatric use, by thirty years. And the drug has been generic the whole time, which is why a pharmaceutical company had to find a half of it to patent.

Chirality: one molecule, two hands

Ketamine has one carbon atom with four different things attached to it. Any such carbon can be built two ways, which are mirror images of each other and can't be rotated into one another, the way your left hand can't be rotated into your right. Chemists call the two forms enantiomers and label them R and S. Ordinary ketamine, the stuff in the anesthesia cart and the IV bag, is a racemic mixture: exactly half R, half S. Esketamine, the es- being the letter S, is the S half purified out. Spravato is esketamine. That's the whole difference in chemistry, and it is not nothing.

Why does it matter which hand? Because receptors are also handed. The NMDA receptor, ketamine's main target, binds the S form about three to four times more tightly than the R form, so the S enantiomer is roughly twice as potent as the racemic mixture and you need about half the milligrams.² That is the pharmacological case for Spravato: more of the active hand, less of the other, lower dose. There are two complications. First, the R form is not inert: in mice it is the better antidepressant, longer-lasting and with less dissociation, which is a puzzle if NMDA blockade is the whole story.¹³ A small open-label pilot of IV arketamine in humans looked promising; controlled trials have been less convincing, and the puzzle is open.¹⁴ Second, the only randomized head-to-head in depressed patients, from Brazil, gave IV esketamine at half the dose of IV racemic ketamine and found them non-inferior, which is exactly what the chemistry predicts and tells you nothing about which hand does the work.¹⁵

So: is Spravato just ketamine? It is half of ketamine, the half that fits the receptor best, delivered through the nose at a dose adjusted for both facts. Whether that half alone is better, worse or the same as both halves together is a question no large trial has answered directly, and the indirect comparisons lean, if anything, toward the IV racemic drug.¹⁶

Mechanism: what happens in a synapse

The textbook account, which is probably mostly right, goes like this. Ketamine blocks the NMDA receptor, a glutamate-gated channel. At the low doses used for depression it blocks them preferentially on fast-firing inhibitory interneurons, the cells whose job is to keep the cortex's pyramidal neurons in check.¹⁷ With the brake off, pyramidal cells fire more and glutamate floods the synapse. The flood lands on AMPA receptors, the other glutamate receptor, and the resulting activity makes the postsynaptic cell release BDNF, which binds TrkB, which switches on mTORC1, which drives local protein synthesis, which, within hours, grows new dendritic spines.¹⁸ The depressed prefrontal cortex has fewer synapses than it should; ketamine grows some back, fast.¹⁹ Scroll the synapse.

The trials: what was studied and what the numbers mean

There are two bodies of evidence and they don't mix well. For racemic ketamine, mostly intravenous, there are dozens of investigator-led trials, nearly all small, nearly all a single infusion against saline or midazolam, with the outcome measured at 24 hours to a week. For esketamine, there is one company's Phase 3 program: large, long, against a placebo spray, with an oral antidepressant started in both arms at the same time, and the outcome measured at four weeks. The ketamine trials ask "does it work?"; the esketamine trials ask "does it add to a new antidepressant, and keep working?" Here are the ones that matter, as response and remission rates rather than effect sizes, because that's what you can picture.

What the numbers mean, in order. First, a single infusion of ketamine reliably produces a response in roughly half to two-thirds of people who have failed several antidepressants, within a day, against a placebo rate near zero to a quarter. That has been replicated enough times that it is no longer in doubt.³² Second, Spravato's margin over placebo is real and modest: about four points on the MADRS, a sixty-point scale, in the one short-term Phase 3 trial that succeeded, with the two others (a fixed-dose trial, and a trial in people over 65) missing their primary endpoints.^26,27,28 Both arms in those trials also started a new antidepressant, which is part of why the placebo response was so high. Third, the relapse-prevention trial is Spravato's best evidence: in people who had already remitted, continuing the spray cut relapse from 45% to 27%.²⁹ Fourth, the two suicidality trials showed faster improvement in depression but no significant effect on the suicidality scale itself; the 2020 label expansion rests on the depression result.^33,34 Fifth, 2023 gave us the comparisons the field wanted: esketamine beat quetiapine head-to-head, and IV ketamine was non-inferior to ECT in outpatients, though a Swedish trial in sicker inpatients found ECT better.^11,12,30

The question nobody has run the trial to answer: IV ketamine versus Spravato, same patients, same schedule. Meta-analyses that compare across trials favour the IV drug, with the caveat that the IV trials were shorter, smaller, and measured earlier, when the effect is biggest.¹⁶ My clinical read, for what it's worth, is that IV ketamine is somewhat stronger per session, Spravato is more consistent and easier to sustain, and the difference matters less than the question of what happens after week four.

Dose, route and bioequivalence

The standard infusion is 0.5 mg/kg of racemic ketamine over 40 minutes: for a 70 kg adult, 35 mg, of which 17.5 mg is the S enantiomer. Spravato comes as 56 or 84 mg of esketamine, of which the label says about 48% is absorbed through the nose.¹⁰ Oral ketamine has a bioavailability around 17–29%, intramuscular about 93%, sublingual troches around 30%, intranasal racemic around 45%.^35,36 Oral dosing has one small positive placebo-controlled trial behind it, three times a week for three weeks.³⁷ If you want a crude way to compare any dose by any route with the reference infusion, multiply the dose by the fraction absorbed and by the fraction that is S-ketamine. That's what this does.

Run the default and you'll see the puzzle: 84 mg of esketamine through the nose puts roughly twice as much S-ketamine into the blood as the standard infusion, and yet nobody thinks Spravato is twice as strong. The reason is the rest of the pharmacology: the spray is absorbed over half an hour or more and peaks lower; the infusion delivers the R half too, with its own metabolites; and the relationship between exposure and antidepressant effect is not linear anywhere in this literature.

Two findings from the trials bear on dose. In the esketamine program, 84 mg was not clearly better than 56 mg, and the fixed-dose trial missed significance at both.^27,26 In the IV literature, there is a dose-response: a 2020 dose-ranging trial of 0.1 to 1.0 mg/kg found the standard and higher doses clearly active and 0.1 mg/kg not clearly better than placebo, and the Australian subcutaneous trial got its result only after raising doses toward 0.9 mg/kg.^38,31 The practical upshot is that the standard infusion sits near the middle of the useful range, and that very low doses, including some of what is sold as "microdosed" ketamine, probably don't do much.

How long it lasts

The honest answer is: not long enough. After a single infusion, the effect peaks at 24 hours and, in most responders, is gone within a week; the median time to relapse in the single-infusion trials is a few days to two weeks.^9,24 A course of six infusions over two weeks produces a longer response, typically measured in weeks rather than days, and Phillips' trial found that weekly maintenance infusions held the response for most of those who had one, while those who stopped relapsed at a median of around two to three weeks.^39,40 Spravato's schedule is built on the same shape: twice weekly for four weeks, weekly for four, then weekly or every other week for as long as it's working.¹⁰ The mechanism section explains why: spines grow in a day and prune over weeks.

Safety, briefly

Acutely: dissociation (which is the point for some and the problem for others), a rise in blood pressure and heart rate for an hour or so, nausea, dizziness, and in a few people anxiety during the session. These are why Spravato requires two hours of observation after each dose under its REMS program, and why I'd want the same for any infusion.^10,41 Longer term, the concerns come from heavy recreational use rather than clinical dosing: bladder damage (ketamine cystitis), memory impairment and dependence are well documented in people using grams a day, and are the reason clinical protocols cap frequency and ask about use between sessions.^42,43 Whether twice-weekly clinical dosing for years carries any of that risk is, honestly, not known; the longest Spravato follow-ups run a few years and are reassuring but not definitive. At-home sublingual ketamine by telehealth, which grew enormously after 2020, has one large open-label effectiveness study behind it and no controlled trial; the consensus statements of the field have been consistent that the drug's safety record was built in supervised settings.^44,45,46

Beyond depression: the evidence, ranked

Everything below is off-label, including IV ketamine for depression itself; only Spravato has an FDA indication. The grades are mine: A means replicated randomized trials, B means at least one adequate randomized trial, C means small or conflicting randomized data, D means case series, and E means hope. "RCTs" counts randomized controlled trials I'd consider informative.

• Aon-label
  Treatment-resistant unipolar depression Spravato FDA 2019

  The body of evidence above. Replicated for IV racemic ketamine across many trials; Phase 3 program for esketamine.³²

• A−5+ RCTs
  Suicidal ideation, acute rapid

  An individual-patient meta-analysis of ten trials found a single infusion reduced suicidal thoughts within a day in about half of patients, independent of the effect on depression.⁴⁷ Esketamine's own suicidality trials improved depression but did not separate from placebo on the suicidality measure.^33,34 Nothing here tells you about suicide itself.

• B+2 RCTs
  Bipolar depression

  Two NIMH crossover trials in patients on lithium or valproate: response at day 1 of 71% and 79% against saline, with no manic switches.^48,49 Small, consistent, and in a condition with few good options.

• B2 RCTs
  Alcohol use disorder

  A single infusion paired with motivational therapy reduced heavy drinking over three weeks; a British trial of three infusions with relapse-prevention therapy cut heavy-drinking days at six months.^50,51 Krupitsky's Russian work from the 1980s said the same thing in a different language.⁶

• B−3 RCTs, mixed
  PTSD

  A single infusion beat midazolam at 24 hours; six infusions over two weeks beat midazolam at two weeks in civilians.^52,53 A larger multi-site trial in veterans and active-duty personnel found no dose-related benefit.⁵⁴ The civilian–military split may be about chronicity, moral injury, or chance.

• C+1 RCT
  Cocaine use disorder

  One infusion plus mindfulness training: 48% abstinent at the end of five weeks against 11% with midazolam.⁵⁵ One trial, one group, never replicated.

• C2 small RCTs
  Anxiety disorders, including social anxiety

  A dose-ranging crossover in refractory anxiety and a crossover in social anxiety both positive, both tiny.^56,57

• C1 small RCT
  OCD

  A fifteen-person crossover: half responded at one week against none on saline, in patients without comorbid depression.⁵⁸ Promising, unreplicated, and a mechanism that makes sense (OCD is a glutamate disorder on several lines of evidence).

• C−1 RCT
  Chronic pain, complex regional pain syndrome

  A ten-day outpatient infusion protocol reduced pain against placebo for weeks.⁵⁹ Anesthesiologists have used it this way for decades with more enthusiasm than trials.

• Dcase series
  Eating disorders

  Mills' 1998 series of fifteen patients with compulsive eating-disorder behavior, nine of whom improved after intermittent infusions, is still the largest controlled-looking report.⁶⁰ Our group ketamine-assisted psychotherapy series at a residential eating disorder program found improvements in depression and anxiety, with no controls.⁶¹ A 2021 narrative review found a handful of cases and one open pilot, and nothing randomized.⁶² I work in this area, and this grade is accurate.

Missing from the ladder on purpose: "general wellness," "burnout," and the various things telehealth ketamine is marketed for that don't have a trial attached. The absence of evidence there is real.

Ketamine-assisted psychotherapy, and the three ways to use the drug

There are three honest ways to work with ketamine, and they are often confused, so here they are side by side.

The case for pairing ketamine with therapy is coherent: the drug opens a window of plasticity, lowers defenses, and sometimes produces an experience people find meaningful, and the window closes in days. The evidence for it is thinner than the case. The largest dataset is a 2019 report of 235 patients across three practices doing ketamine-assisted psychotherapy, with good outcomes and no comparison group.⁶³ A small Yale pilot found that cognitive behavioral therapy after a course of infusions slowed relapse.⁶⁴ The most rigorous trial to date is a strange and encouraging one: Price and colleagues gave a single infusion and then four days of a brief, automated, computer-delivered self-association exercise, and found it extended the antidepressant effect out to a month, against the same infusion with a sham exercise.⁶⁵ In alcohol use disorder, the British trial found the ketamine effect but little added benefit from the therapy arm, which is the only controlled test of the specific question in that population.⁵¹ A 2022 review of the field concluded what I'd conclude: the models are plausible, the mechanisms line up, and the controlled trials that would settle it have not been run.^66,67

My own practice, for the record: I do all three, I think the psycholytic model is underrated and the psychedelic model oversold, and I think the single most useful thing a therapist does around ketamine is not the session itself but the week after it.

The gaps

• IV ketamine versus Spravato, head to head. The comparison patients ask about and insurers decide on, and it has never been run at scale with the same schedule in the same people.
• Ketamine with therapy versus ketamine without. The field's favorite claim and its biggest missing trial. The Price study shows how cheaply the question can be asked; nobody has asked it with a human therapist and a real sample size.
• Which hand. Whether R-ketamine, S-ketamine or the mixture is the better antidepressant is a question the animal data and the human data answer differently.
• Who responds. Fifteen years in, there is no validated predictor of response. Not a biomarker, not a clinical profile, not a dissociation score. The best predictor is still the first infusion.
• Long-term safety of clinical dosing. We know what grams a day does. We don't know what 56 mg twice a month for ten years does, because nobody has done it for ten years yet.
• Maintenance. How long to continue, how to stop, and whether anything (an oral antidepressant, lithium, therapy, exercise) lets you stop sooner. The relapse-prevention trial says stopping is bad; nothing says how to make stopping safe.
• Dose and route, properly. No adequately powered trial has compared oral, sublingual, intranasal and IV ketamine at matched exposure. The calculator above is the state of the art, which is a problem.
• Everything outside depression. One good trial each, at best, for most of the ladder.
• The at-home question. Telehealth ketamine reached a lot of people fast. Its safety case is an open-label study and an absence of catastrophes, which is not the same thing as evidence.

None of that is a reason not to use it. It is a reason to be precise about what we're doing when we do.

This post is educational and isn't medical advice, and the calculator is for understanding, not prescribing. Ketamine is a controlled substance with real risks, and Spravato is dispensed only under supervision for good reasons. If you're considering either, talk to a clinician who does this work and will tell you the parts that didn't fit on this page. If you're in crisis, in the US you can call or text 988 any time.