Published on 23 September 2026

Analysis · Depression · Psychopharmacology

Racemic ketamine or intranasal esketamine: similar score drops, different remission rates

Q: Can we conclude that esketamine is superior to racemic ketamine?

No. The study is observational, product allocation is not random, assessment is open-label, and power on the secondary outcomes is insufficient. The remission gap is statistically significant and consistent across sensitivity analyses, but its confidence interval is wide and the authors themselves call it preliminary, for lack of a mechanistic rationale and given the possibility of residual confounding. It generates a serious hypothesis, not a demonstration of superiority.

Q: Is non-inferiority on the MADRS decrease firmly established, then?

It is met according to the criterion the authors had set: less than a three-point difference between groups, with power of at least 0.97 to detect a small effect. This is a solid result within the limits of the design, but it still comes from a non-randomized, open-label, single-center comparison. It does not amount to a demonstration of equivalence between the two molecules.

Q: What exactly does the target trial emulation framework mean?

It is a method that consists of first defining the randomized trial one would have wanted to run, then reconstructing that structure from observational data, while explicitly checking four assumptions: exchangeability, consistency, positivity, and no interference. It disciplines the analysis and makes its assumptions explicit. It does not replace randomization and does not remove any unmeasured confounder; the authors say so themselves.

Q: Do the cost differences transpose elsewhere?

The amounts cited in the publication come from a 2019 US source, not from Swiss data or from any other national data. They depend on insurance coverage, health system contracts, and wholesale prices. Reimbursement conditions in the reader's own jurisdiction have not been verified here and should be established from the relevant official sources at the time of reading.

Q: Why does blood pressure tolerability differ between the two products?

The study reports a larger mean systolic and diastolic elevation under racemic ketamine, without establishing a mechanism. The authors note that the racemic formulation requires roughly double the dose by mass, and that titration was flexible up to 168 mg per session. They judge the difference unlikely to be clinically relevant. This point cannot be settled from this single series.

◆ Collection
Journal of Affective Disorders · 2026; 400: 121208 · Sarlon et al.
DOI 10.1016/j.jad.2026.121208
PMID 41577046
Scientific 68
Editorial 78

Editorial collections The Monday Morning Patient A Study Under the Microscope

The essentials

Seventy-six outpatients with treatment-resistant depression, followed at a university psychiatric center in Basel between February 2021 and February 2025, received either off-label intranasal racemic ketamine or intranasal esketamine under its approved protocol. On the mean drop in MADRS score between day one and day twenty-eight, the two products came within less than a point of each other, well under the predefined three-point non-inferiority margin. On remission, the gap ran the other way: 38.7% of patients on esketamine reached remission against 15.6% on racemic ketamine, with a wide confidence interval. The work is retrospective, non-randomized, conducted at a single center, and power for the secondary endpoints falls short of 80%. The authors themselves describe their results as hypothesis-generating and their analysis as intermediate.

Context

Since intranasal esketamine was approved for treatment-resistant depression, in March 2019 in the United States and December 2019 in Europe according to the authors, cost and access have become a recurring question in services treating these patients. Racemic ketamine has been used since the 1960s, first as an anesthetic; at subanesthetic doses given intravenously it is now an established intervention in depression, supported by several randomized trials and meta-analyses. The intranasal racemic route, by contrast, remains far less studied. To the authors’ knowledge, no study had directly compared the two intranasal formulations in this population.

The stakes are not purely economic. If both products lower scores by the same margin but one carries more patients across the remission threshold, then the average conceals two different distributions of response, and the choice is not reducible to a budget arbitration.

The study at a glance

Population
Seventy-six outpatients aged 18 and older with treatment-resistant depression, followed at the University Psychiatric Clinics of Basel between February 2021 and February 2025. Thirty-one on esketamine, forty-five on racemic ketamine. Excluded were patients referred only for maintenance treatment (5), those who received fewer than seven induction administrations (4), and those who discontinued induction for an adverse effect (3).
Intervention
Intranasal racemic ketamine, 28 to 168 mg per session, twice weekly for four weeks, that is, eight administrations during the induction phase. Flexible titration according to age, tolerability, clinical response, and comorbidities.
Comparator
Intranasal esketamine, 28 to 84 mg per session under the approved titration protocol, the same induction schedule, and the same monitoring framework.
Primary outcome
Change in total MADRS score between day one and day twenty-eight, with a predefined non-inferiority margin of less than three points and a stated power of at least 80% on that threshold. Secondary outcomes: remission (MADRS 12 or lower), response (at least a 50% MADRS decrease), sensitivity analyses at thresholds of 10 and 7, safety and tolerability.
Design
Single-center, retrospective, open-label observational study on routine clinical data, analyzed under the target trial emulation framework. Repeated-measures analysis of variance, odds ratios with confidence intervals, Bayesian multilevel modeling, and chi-square tests with Yates correction. Intention-to-treat analyses, run under R 4.4.0. Oxford level of evidence 3b.

Quality control

Point checked	Verdict
Pre-registered protocol	No registration mentioned
FindingThe publication mentions no registration number. It does state that the three-point non-inferiority margin was predefined, as was the non-inferiority criterion on remission, set at retaining at least half of esketamine’s effect. A transparency statement signed by the guarantor author accompanies the article.
Randomization	Absent
FindingProduct allocation depended on clinical and administrative considerations: ineligibility for esketamine for lack of sufficient prior treatment trials, denial of insurance coverage, the patient’s wish to start without delay, or a prior favorable experience with racemic ketamine. This is the structuring limitation of this work, and the authors place it first among their limitations.
Blinding	Absent
FindingOpen-label study, presented as such. The authors write that patient expectations and clinician beliefs about an approved versus an off-label product may have influenced symptom reporting and clinical assessment.
Baseline comparability	Satisfactory on baseline data
FindingAge, sex, body mass index, baseline MADRS, diagnosis, and severity are balanced in Table 1, all p values above 0.10, which led the authors not to adjust for covariates. Two caveats: the proportion of women differs by nearly twofold without reaching significance (25.8% versus 46.7%, p = 0.110), and the authors acknowledge that concomitant psychotropic treatments differed between groups and may have contributed to the results.
Statistical power	Uneven across endpoints
FindingPower of at least 0.97 to detect a small effect on the primary outcome, close to 1.0 for a moderate effect. But only 74% to detect an odds ratio of 0.5 on the dichotomous secondary outcomes, short of the 80% target the authors set for themselves. The study’s most striking result is also its least well-supported one, and the authors themselves note that under low power a significant result can be a false positive.
Protocol comparability	Asymmetric
FindingFlexible titration on one side, a standardized algorithm on the other. The authors note that this flexibility produced variable durations of stable dosing at the time of the day-28 assessment, and that it may have contributed to the lower remission observed in the ketamine group.
Funding and conflicts of interest	No specific funding
FindingThe publication states that the research received no specific grant from any funding body, commercial or non-profit, and that the authors have nothing to declare. The protocol was approved by the ethics committee of Northwestern and Central Switzerland. Data are stated to be available from the corresponding author on reasonable request.

Results

38.7%
Remission rate under intranasal esketamine at day twenty-eight, against 15.6% under intranasal racemic ketamine. Odds ratio 3.43, 95% confidence interval 1.16 to 10.12; number needed to treat 4; Bayesian posterior probability of superiority 98.6%.

Outcome	Result
MADRS decrease at four weeks	Less than a one-point difference between arms
ReadingMean baseline MADRS 27.5 (standard deviation 8.1) overall, 17.7 (8.9) after treatment, a mean decrease of 9.7 points (8.3). By group: esketamine, a 9.3-point decrease (standard deviation 9.1); racemic ketamine, a 10.0-point decrease (standard deviation 7.7). The gap stays well under the three-point non-inferiority margin. Primary endpoint met.
Effect of time	F(1,74) = 104.54; p < 0.001
ReadingGeneralized eta squared 0.25. Both groups improve markedly. In the absence of a control arm, this improvement cannot be attributed to the product alone.
Treatment effect and interaction	F(1,74) = 0.70; p = 0.41
ReadingGeneralized eta squared 0.007, and no interaction between time and treatment either (F(1,74) = 0.15; p = 0.70; eta squared 0.0005). No detected difference between the two products on the mean trajectory. An absence of detected difference is not proof of equivalence, and that reservation cuts both ways; it is the high power on this endpoint that makes the non-inferiority argument credible here, not the non-significant p value alone.
Remission, MADRS 12 or lower	38.7% versus 15.6%; OR 3.43
ReadingTwelve of thirty-one patients against seven of forty-five. Odds ratio of 3.43 in favor of esketamine, 95% confidence interval 1.16 to 10.12; the publication also reports the same relationship in reciprocal form, 0.29 (0.10 to 0.86), whose upper bound under 0.5 causes the predefined non-inferiority criterion on this point to fail. Chi-square 4.09; p = 0.043; Cohen’s omega 0.23. Bayesian posterior probability of superiority 98.6%. Power 74%. The confidence interval spans a modest odds ratio to a very high one: the direction of the gap is consistent, its magnitude is not.
Sensitivity analyses on remission	Same direction at thresholds of 10 and 7
ReadingFor MADRS 10 or lower: 35.5% versus 11.1% (p = 0.0229; OR 0.23; 95% CI 0.07 to 0.74). For MADRS 7 or lower: 19.4% versus 2.2% (p = 0.0328; OR 0.09; 95% CI 0.01 to 0.83), that is, six patients against a single one. The numbers involved are very small.
Response, at least a 50% MADRS decrease	41.9% versus 31.1%; not significant
ReadingOdds ratio 0.63 (95% CI 0.24 to 1.62); number needed to treat 10; chi-square 0.53; p = 0.468; omega 0.08. The Bayesian analysis here gives a 49.2% probability of esketamine’s superiority, the equivalent of a coin flip. The observed advantage therefore concerns crossing the remission threshold, not the response threshold.
Blood pressure	Larger rises under racemic ketamine
ReadingSystolic: mean rise of 7.44 mmHg (standard deviation 8.63) under racemic ketamine against 3.16 mmHg (5.62) under esketamine (t = 2.62; p = 0.011; d = 0.57). Diastolic: 6.91 mmHg (4.73) against 4.33 mmHg (3.93) (t = 2.59; p = 0.012; d = 0.58). The authors judge these differences statistically significant and of moderate magnitude, but unlikely to be clinically relevant.
Other adverse events and adherence	Comparable profiles
ReadingOne discontinuation in the esketamine group (3.2%, fatigue), two in the ketamine group (4.4%: severe dissociation; hypotension and bradycardia probably related to a comorbidity). The most frequent event was transient blood pressure elevation (8 patients), then nausea (5), managed with amlodipine or ondansetron; two patients received lorazepam for acute anxiety. No between-group difference on dissociative symptoms or on anxiety (all p values above 0.1).
Cost of the induction phase	Figures drawn from an earlier US source
ReadingThe discussion reports no cost measured at the Basel center itself. It cites amounts published in 2019 in the United States: $5,664 to $8,142 for the first month of esketamine induction, against a drug cost of $42 per session for compounded intranasal racemic ketamine, excluding monitoring and administration costs. Across the eight induction sessions, the corresponding gap is on the order of seventeen to twenty-four fold, not a hundred or a hundred fifty fold. These figures do not transpose to the Swiss health system, nor to any other health system, without independent verification.

Critical appraisal

Domain	Risk of bias
Selection of participants	High
FindingAccess to esketamine depended in part on financial coverage and treatment history. The two groups may therefore differ on unmeasured characteristics that are themselves linked to prognosis. The authors write that these factors are not fully captured by the variables recorded at inclusion.
Classification of interventions	Moderate
FindingFlexible titration against standardized titration: it is not quite the same intervention being compared on each side. The authors acknowledge that this difference could have advantaged or disadvantaged either product.
Confounding	Moderate
FindingThe target trial emulation framework structures the analysis and requires naming the assumptions of exchangeability, consistency, positivity, and no interference. It does not create the comparability that randomization would have provided, and the authors state that residual confounding cannot be excluded. No covariate adjustment was performed.
Outcome measurement	Moderate
FindingMADRS rated open-label. One point deserves flagging: in this health system, only the esketamine group had to document a predefined clinical improvement for reimbursement to continue. The authors themselves cite this arrangement as a possible source of bias, on the side of the product that achieves the better remission rate.
Selective outcome reporting	Moderate
FindingNo registered protocol is mentioned. Conversely, the thresholds and sensitivity analyses are described as predefined, the result unfavorable to the less costly product appears in the abstract, and the discrepancy between mean non-inferiority and failure on remission is discussed by the authors, which limits the concern of an opportunistic emphasis.

Level of evidence

Scientific68

Editorial78

Oxford level of evidence 3b. Confidence is reasonable on one descriptive point: in this center, the two products produced comparable score decreases over four weeks, with sufficient power to make that comparison informative. Confidence is low on the remission gap, which rests on a small sample, nineteen events in total, 74% power, a very wide confidence interval, non-random allocation, and open-label assessment, in a setting where reimbursement of only one of the two products was conditioned on documented improvement. This gap reads as a hypothesis to test, not as an established hierarchy. The authors themselves describe their own analysis as intermediate and call for prospective studies.

The colleague test

What a hospital psychiatrist experienced in treatment-resistant depression might say about this work, between two consultations.

“On paper, the racemic version does just as well for a fraction of the cost. Except two and a half times fewer patients reach remission, and remission is what I care about, not the mean slope. Except also that on response, there’s nothing left. Retrospective, open-label, seventy-six patients, allocation partly driven by who is paying: I note the hypothesis, I change nothing.”

Translated for practice: this work feeds a discussion about access in settings where esketamine is unavailable or not reimbursed. It does not settle a treatment choice, and it does not license presenting the two formulations as interchangeable.

What you can do with this

Non-inferiority on a mean score decrease and failed non-inferiority on the remission rate are not contradictory: they describe two different distributions of response behind the same average. The clinical threshold and the mean slope do not tell the same story, and this is the explanation the authors themselves retain.
Esketamine’s advantage appears only on remission. On response, defined as at least a 50% score decrease, the two products do not differ. This contrast is worth keeping in mind before the result gets reduced to a single figure.
To a patient who has read that ketamine costs a hundred times less than esketamine, it is fair to answer that the price gap is real and substantial, but that the amounts cited in this publication are 2019 US estimates covering only the drug’s cost, excluding monitoring and administration, and that they correspond to a ratio on the order of seventeen to twenty-four for the induction phase. A direct clinical comparison now exists; it remains preliminary.
This study was conducted in Switzerland, where intranasal esketamine was the only authorized intranasal formulation and intranasal racemic ketamine was administered off-label, after information and written consent. Authorization, marketing, and collective reimbursement are three distinct regulatory planes, and a coverage decision never states that a treatment is ineffective. The authorization, prescribing, and reimbursement conditions in the reader’s own jurisdiction have not been verified here and should be checked against current official regulatory sources before this comparison informs a prescribing decision.
Blood pressure monitoring is part of the administration protocol in both groups, with measurement every twenty minutes and observation for at least ninety minutes after administration. Mean systolic elevation was larger with racemic ketamine in this series, though the authors do not consider it clinically significant.
Worth watching: the prospective trials the authors call for, and the continued data collection at their center, the only means likely to settle the remission question.

Frequently asked questions

Can we conclude that esketamine is superior to racemic ketamine?

No. The study is observational, product allocation is not random, assessment is open-label, and power on the secondary outcomes is insufficient. The remission gap is statistically significant and consistent across sensitivity analyses, but its confidence interval is wide and the authors themselves call it preliminary, for lack of a mechanistic rationale and given the possibility of residual confounding. It generates a serious hypothesis, not a demonstration of superiority.

Is non-inferiority on the MADRS decrease firmly established, then?

It is met according to the criterion the authors had set: less than a three-point difference between groups, with power of at least 0.97 to detect a small effect. This is a solid result within the limits of the design, but it still comes from a non-randomized, open-label, single-center comparison. It does not amount to a demonstration of equivalence between the two molecules.

What exactly does the target trial emulation framework mean?

It is a method that consists of first defining the randomized trial one would have wanted to run, then reconstructing that structure from observational data, while explicitly checking four assumptions: exchangeability, consistency, positivity, and no interference. It disciplines the analysis and makes its assumptions explicit. It does not replace randomization and does not remove any unmeasured confounder; the authors say so themselves.

Do the cost differences transpose elsewhere?

The amounts cited in the publication come from a 2019 US source, not from Swiss data or from any other national data. They depend on insurance coverage, health system contracts, and wholesale prices. Reimbursement conditions in the reader’s own jurisdiction have not been verified here and should be established from the relevant official sources at the time of reading.

Why does blood pressure tolerability differ between the two products?

The study reports a larger mean systolic and diastolic elevation under racemic ketamine, without establishing a mechanism. The authors note that the racemic formulation requires roughly double the dose by mass, and that titration was flexible up to 168 mg per session. They judge the difference unlikely to be clinically relevant. This point cannot be settled from this single series.

Annotated bibliography

Source study. Sarlon J, Thomi D, Brühl AB, Liwinski T, Lang UE. Real-world comparison of intranasal racemic ketamine and esketamine in treatment-resistant depression: A retrospective observational study. Journal of Affective Disorders. 2026; 400: 121208. DOI 10.1016/j.jad.2026.121208. Open access publication under a Creative Commons license. Funding: the research received no specific grant from any funding body, commercial or non-profit. Conflicts of interest: the authors declare having nothing to declare. PMID 41577046. No supplementary material accompanied the document consulted: only the published abstract was accessible for this English version, and no data appearing solely in the full text or its supplementary material is reproduced here beyond what the French version, verified against the full text, already establishes.

Context references, cited by the study. The following works are cited by the source study, and their bibliographic details are taken from its reference list; they were not consulted in full text for this verification. Sarlon J, Liwinski T, Bruehl AB, Thomi D, Lang UE. Intranasal racemic ketamine in treatment-resistant depression: efficacy and tolerability in outpatient treatment. Journal of Affective Disorders Reports. 2025; 21: 100954. DOI 10.1016/j.jadr.2025.100954. This is the uncontrolled case series from the same center, of which the present work is the comparative continuation.

Bahji A, Vazquez GH, Zarate CA. Comparative efficacy of racemic ketamine and esketamine for depression: a systematic review and meta-analysis. Journal of Affective Disorders. 2021; 278: 542 to 555. DOI 10.1016/j.jad.2020.09.071. Meta-analysis cited by the authors for a superior efficacy of intravenous racemic ketamine over intranasal esketamine, which runs counter to their own remission result and grounds their caution.

An D, Wei C, Wang J, Wu A. Intranasal ketamine for depression in adults: a systematic review and meta-analysis of randomized, double-blind, placebo-controlled trials. Frontiers in Psychology. 2021; 12. DOI 10.3389/fpsyg.2021.648691. External comparison reference: a weighted mean difference of 4.09 MADRS points (95% CI 2.18 to 6.00) against placebo across five trials.

Hernán MA, Robins JM. Using big data to emulate a target trial when a randomized trial is not available. American Journal of Epidemiology. 2016; 183: 758 to 764. DOI 10.1093/aje/kwv254. The methodological framework claimed by the study.

Bahr R, Lopez A, Rey JA. Intranasal esketamine (Spravato) for use in treatment-resistant depression in conjunction with an oral antidepressant. P&T. 2019; 44: 340 to 375. The sole source of the dollar amounts cited in the study’s discussion, including the $5,664 to $8,142 for the first month of induction and the $42 per session for compounded racemic ketamine.

Point still to verify. The regulatory status of intranasal esketamine and of intranasal racemic ketamine in treatment-resistant depression was not verified from the documents consulted here, which describe only the Swiss situation. These elements should be established against the official regulatory sources of the reader’s own jurisdiction before this comparison is transposed elsewhere.

Editorial collections

The Monday Morning Patient A Study Under the Microscope