Trial of Ketamine Masked by Surgical Anesthesia in Depressed Patients
This triple-masked, randomised, placebo-controlled trial (n=40) of adults with major depressive disorder (MDD) found no short-term effect on depression severity (measured by MADRS) after a single dose of intravenous ketamine (35mg/70kg) compared to placebo (saline) during anaesthesia for routine surgery.
Authors
- Cianfichi, L. J.
- Flohr, J. R.
- Hack, L. M.
Published
Abstract
Background Ketamine may have antidepressant properties, but its acute psychoactive effects complicate successful masking in placebo-controlled trials.Methods In a triple-masked, randomized, placebo-controlled trial, 40 adult patients with major depressive disorder were randomized to a single infusion of ketamine (0.5 mg/kg) or placebo (saline) during anesthesia for routine surgery. The primary outcome was depression severity measured by the Montgomery-Åsberg Depression Rating Scale (MADRS) at 1, 2, and 3 days post-infusion. The secondary outcome was the proportion of participants with clinical response (≥50% reduction in MADRS scores) at 1, 2, and 3 days post-infusion. After all follow-up visits, participants were asked to guess which intervention they received.Results Mean MADRS scores did not differ between groups at screening or pre-infusion baseline. The mixed-effects model showed no evidence of effect of group assignment on post-infusion MADRS scores at 1 to 3 days post-infusion (-5.82, 95% CI -13.3 to 1.64, p=0.13). Clinical response rates were similar between groups (60% versus 50% on day 1) and comparable to previous studies of ketamine in depressed populations. Secondary and exploratory outcomes found no evidence of benefit for ketamine. 36.8% of participants guessed their treatment assignment correctly; both groups allocated their guesses in similar proportions.Conclusion A single dose of intravenous ketamine compared to placebo has no short-term effect on the severity of depression symptoms in adults with major depressive disorder. This trial successfully masked treatment allocation in moderate-to-severely depressed patients using surgical anesthesia. While it is impractical to use surgical anesthesia for most placebo-controlled trials, future studies of novel antidepressants with acute psychoactive effects should make efforts to fully mask treatment assignment in order to minimize subject-expectancy bias. (ClinicalTrials.gov number, NCT03861988)
Research Summary of 'Trial of Ketamine Masked by Surgical Anesthesia in Depressed Patients'
Introduction
Ketamine is a dissociative anaesthetic that has shown rapid-acting antidepressant effects in patients with major depressive disorder (MDD), including treatment-resistant cases. Prior randomised trials using a single intravenous infusion of 0.5 mg/kg report rapid therapeutic onset and measurable response and remission rates within 24 hours. However, the drug’s obvious acute psychoactive effects make participant masking difficult in placebo-controlled trials, creating a risk of subject-expectancy bias that could inflate apparent treatment effects. This concern applies broadly to other fast-acting psychoactive therapeutics as well. Lii and colleagues designed a randomised, triple-masked, placebo-controlled trial that used routine surgical anaesthesia to conceal administration of ketamine or saline, with the primary aim of determining whether a single 0.5 mg/kg infusion of ketamine given during maintenance anaesthesia produces greater short-term antidepressant effects than placebo. The hypothesis was that ketamine would reduce depressive symptoms more than inert saline within the first 3 days post-infusion in adults with moderate-to-severe MDD. The study therefore addresses both efficacy and the feasibility of rigorous masking in ketamine trials.
Methods
This was an investigator-initiated, university-sponsored, triple-masked, randomised, placebo-controlled trial conducted at Stanford University Medical Center. Adults scheduled for elective non-cardiac, non-intracranial surgery who screened ≥10 on the PHQ-8 were further assessed; inclusion required a confirmed diagnosis of MDD, a current episode of ≥4 weeks, English literacy, BMI 17–40 kg/m2, and a combined MADRS plus HADS score of ≥31. Exclusion criteria included pregnancy, moderate-to-severe substance use disorder, psychotic-spectrum disorders, significant neurological disease, recent clinically significant thyroid dysfunction, high-dose chronic opioid use (≥90 MME/day), and high suicide risk. Concurrent psychotherapy and antidepressant medications were permitted if stable for ≥4 weeks. Forty participants were randomised in blocks to receive either a single intravenous ketamine infusion (0.5 mg/kg diluted to 40 ml, infused over 40 minutes) or 40 ml normal saline infused over 40 minutes. Randomisation and drug compounding were managed by the investigational drug service. Masking extended to participants, investigators, and direct care providers including anaesthesiologists; processed EEG monitoring (SedLine Patient State Index, PSI) guided anaesthetic depth and the study infusion was started after induction and surgical incision during maintenance anaesthesia (target PSI 25–50). Anaesthesiologists received the study drug in an unlabeled syringe and were instructed not to alter anaesthesia in response to EEG changes during the infusion except to avoid patient awareness. Use of nitrous oxide was minimised. The primary outcome was the Montgomery-Åsberg Depression Rating Scale (MADRS) measured on postoperative days 1, 2 and 3. The pre-specified secondary outcome was clinical response (≥50% reduction in MADRS from screening baseline); remission (MADRS ≤12) and other measures (HADS, Brief Pain Inventory–SF for pain, inpatient and outpatient opioid use in MME) were exploratory. Outcomes were assessed in hospital and by video/telephone after discharge; participants were also asked to guess treatment assignment at day 14. A priori sample size calculation targeted 38 participants for 80% power (based on previous trial effect sizes) and two additional participants were added for potential attrition, yielding N=40. The analysis used an intention-to-treat framework. The pre-registered primary analysis was a mixed-effects model for repeated measures with fixed effects for group, time (days), and their interaction, plus random intercepts and slopes and an unstructured covariance matrix. An alternative non-prespecified mixed model analysing change from the pre-infusion baseline (day 0) was also used. Sensitivity analyses adjusted for baseline MADRS differences and excluded participants exposed to nitrous oxide. Cohen’s kappa was calculated post hoc to quantify agreement on treatment guesses.
Results
Participants were recruited between February 2020 and August 2022 and randomised 1:1 to ketamine or saline (N=40). The mean age was 51 years; 70% were female, 65% white, 87.5% non-Hispanic, and 62.5% employed. At screening both groups had moderate depression (screening MADRS: ketamine mean 27.7, placebo 30.6) and comparable treatment-resistance scores (Maudsley Staging Method). Pre-infusion MADRS on the day of surgery were similar (ketamine 25.1 [SD 8.3], placebo 29.9 [SD 7.0]). Surgical and anaesthetic characteristics were broadly similar between arms; all but one participant underwent general anaesthesia. Three participants (two ketamine, one placebo) received nitrous oxide ≥50% for ≥1 hour. For the primary outcome, the pre-registered mixed-effects model found no evidence of an effect of group assignment on MADRS scores across post-infusion days 1–3 (95% CI -13.3 to 1.64, p=0.13). The rate of change in MADRS also did not differ between groups. An alternative model analysing change from the pre-infusion baseline similarly showed no between-group difference (95% CI -6.25 to 7.89, p=0.82). Missing MADRS data did not exceed 5% at any visit and were not imputed. Both groups showed substantial mean decreases from day 0 to day 1: ketamine mean change −12.4 (SD 9.2), placebo −14.7 (SD 9.0), corresponding to average reductions of 46% and 48% respectively. Clinical response (≥50% reduction from screening) on post-infusion day 1 occurred in 60% of the ketamine group and 50% of the placebo group; response rates remained similar on days 2 and 3. Remission (MADRS ≤12) occurred in 50% vs 35% of ketamine and placebo participants on day 1 and was 40% in both groups by day 3. HADS trajectories paralleled MADRS findings. Cumulative opioid consumption and most pain outcomes did not differ between groups; average pain at day 14 was numerically similar (ketamine 4.8 [SD 1.5], placebo 3.7 [SD 2.0]). Hospital length of stay differed between groups (means 1.9 [SD 1.7] vs 4.0 [SD 3.3] days, p=0.02), with the authors reporting longer stays in the placebo group. Safety and protocol deviations: one participant retrospectively failed to meet the pre-infusion symptom threshold but was included in ITT analyses. One death occurred in the ketamine group five days post-infusion; the investigators attributed this event to underlying cardiovascular disease rather than trial procedures. Sensitivity analyses adjusting for pre-infusion MADRS and excluding participants exposed to nitrous oxide did not materially change results. At final follow-up, 36.8% of participants guessed their treatment correctly; Cohen’s kappa was 0.33, indicating fair agreement and suggesting effective masking.
Discussion
Lii and colleagues interpret their findings as showing no short-term antidepressant benefit of a single 0.5 mg/kg intravenous ketamine infusion administered during surgical anaesthesia compared with saline, based on MADRS scores on post-infusion days 1–3. Randomisation appeared effective given similar baseline MADRS scores, and both clinician-rated (MADRS) and patient-rated (HADS) measures showed comparable decreases in both groups. The magnitude and variance of the observed decreases were similar to prior awake-ketamine studies, but those prior trials often lacked effective masking. The investigators highlight successful participant masking as a major strength: fewer than half of participants guessed their allocation correctly and guess distribution was similar across arms. They suggest that inadequate masking in prior ketamine trials may have allowed subject-expectancy bias to inflate apparent treatment effects, and that rigorous masking—here achieved via surgical anaesthesia—substantially reduced the between-group effect size. Nonetheless, the authors acknowledge alternative explanations. Anaesthetic agents used during surgery might have independent antidepressant effects (for example, nitrous oxide, propofol, or inhaled agents under certain conditions), opioid exposure could modulate ketamine’s efficacy, and baseline heterogeneity such as longer median current episode duration in the ketamine arm could affect responsiveness. Key limitations the investigators acknowledge include not assessing whether anaesthesiologists remained blind to allocation and not measuring participant treatment expectancies prior to randomisation, so they cannot conclusively attribute the smaller-than-anticipated effect to reduced expectancy alone. Surgical heterogeneity was not controlled at recruitment, and the trial cannot fully distinguish between a true null effect of ketamine and an occlusion of ketamine’s effect via placebo-like mechanisms maintained in the absence of unmasking. The single observed death in the ketamine arm is reported and discussed as likely related to pre-existing cardiovascular comorbidity rather than the intervention. The authors recommend that future trials of rapid-acting psychoactive antidepressants prioritise rigorous masking to reduce subject-expectancy bias, while noting the impracticality of using surgical anaesthesia broadly.
Conclusion
Using surgical anaesthesia to mask treatment allocation, this randomised, triple-masked trial found no difference in depression severity between a single intraoperative 0.5 mg/kg ketamine infusion and saline on post-infusion days 1–3, and no evidence of benefit on secondary or exploratory outcomes. The investigators conclude that inadequate masking in prior ketamine trials may have contributed to larger reported effect sizes and that future studies of acute psychoactive antidepressants should strengthen masking procedures to limit subject-expectancy bias. They note, however, that surgical anaesthesia is not a practical masking strategy for routine trials.
View full paper sections
CONCLUSION
A single dose of intravenous ketamine compared to placebo has no short-term effect on the severity of depression symptoms in adults with major depressive disorder. This trial successfully masked treatment allocation in moderate-to-severely depressed patients using surgical anesthesia. While it is impractical to use surgical anesthesia for most placebo-controlled trials, future studies of novel antidepressants with acute psychoactive effects should make efforts to fully mask treatment assignment in order to minimize subject-expectancy bias. (ClinicalTrials.gov number, NCT03861988)
INTRODUCTION
Ketamine, a dissociative anesthetic with multiple molecular targets, is associated with rapidacting antidepressant effects in patients with major depressive disorder (MDD) , including those with treatment-resistant depression (TRD). Across studies, an intravenous infusion of 0.5 mg/kg ketamine produces a clinical response in 41% and remission in 19% of patients with TRD at 24 hours. Therapeutic effects appear within 2 hours of a single ketamine infusion. In most randomized controlled trials (RCTs) of ketamine for depression, participant masking has been nearly impossible given the drug's obvious acute psychological effects. Inadequate masking presents a major confound to interpreting studies of ketamine, as well as other rapid-acting psychoactive therapeutics such as psilocybin and methylenedioxymethamphetamine (MDMA), to the extent that most investigations do not report on the success of participant masking. Incomplete masking may lead to subject-expectancy bias, which occurs when a research subject has an expectation for a given result that influences the reported outcome. Subject-expectancy bias may contribute to overestimation of treatment effect sizes in antidepressant trials involving ketamine. We utilized ketamine's established safety in surgical settings by conducting a randomized placebo-controlled trial in which the administration of ketamine was masked by other surgical anesthetics. The primary aim of this study was to determine whether ketamine, given at a dose of 0.5 mg/kg over 40 minutes during surgical anesthesia, produces a greater antidepressant effect than placebo. We recruited patients with depression severity comparable to previous studies and analyzed similar follow-up time points. We hypothesized that ketamine is superior to an inert placebo (0.9% sodium chloride, i.e., normal saline) in reducing depression symptoms within the first 3 days post-infusion in a population of adults with moderate-to-severe levels of MDD.
TRIAL OVERSIGHT
This was an investigator-initiated, university-sponsored trial. The trial protocol was approved by the institutional review board at Stanford University and all participants gave written informed consent. Randomization and drug compounding were handled by Stanford Health Care Investigational Drug Service. The first and last authors designed the trial. The first author analyzed the data and wrote the first draft of the manuscript. The second through sixth performed the trial and collected the data. The penultimate author provided content expertise and advised trial design. The overall trial was overseen by the last author. The authors vouch for the accuracy and completeness of the data and for the fidelity of the trial to the protocol. There was no industry involvement in the collection or analysis of the data, and no agreements were in place between the authors and any commercial entity.
PARTICIPANTS
Adults undergoing elective non-cardiac, non-intracranial surgery were recruited from preoperative clinics at Stanford University Medical Center. The 8-item Patient Health Questionnaire (PHQ-8) was distributed to patients through a perioperative mental health screening service. To be eligible for the study, patients must score ≥ 10 on the PHQ-8, corresponding with at least moderate depression. Research staff screened electronic health records (EHR) of patients scheduled for surgery who scored ≥ 10 points on the PHQ-8; those without exclusion criteria documented in the EHR were introduced to the study and consented for an additional screening visit. Surgical clinics also referred patients with symptomatic depression who expressed interest in the trial. These patients were contacted by research staff for a telephone pre-screen, and qualifying patients were consented for an additional screening visit. At this visit, research staff collected information on demographics, medical, and psychiatric history, including level of antidepressant treatment resistance assessed by the Maudsley Staging Method (MSM). Inclusion and exclusion criteria were assessed via a hybrid approach of corroborating EHR data with patient self-report. Inclusion criteria included English literacy, body mass index of 17-40 kg/m 2 , a diagnosis of MDD (single or recurrent) and a major depressive episode of ≥ 4 weeks duration prior to screening. The diagnosis of MDD was confirmed by the Mini International Neuropsychiatric Interview Module A. Participants must also have had a combined score of ≥ 31 from the Montgomery-Åsberg Depression Rating Scale(MADRS) and the Hospital Anxiety and Depression Scale 16 (HADS). These scales were administered in-person, or by secure video conference or telephone, which have been validated for the PHQ, HADS, and MADRS. Exclusion criteria included pregnancy, breastfeeding, moderate or severe substance use disorder, history of schizophrenia or schizoaffective disorder, dementia or other amnestic cognitive disorder, history of surgery involving the brain or meninges, encephalitis, meningitis, or degenerative central nervous system disorder, clinically significant thyroid dysfunction within the past 6 months, and chronic use of ≥ 90 morphine milligram equivalents (MME) per day. Patients at high risk of suicidal behavior on the Columbia-Suicide Severity Rating Scale 21 were also excluded. Concurrent psychotherapy and antidepressant therapy were allowed if therapy was stable for ≥ 4 weeks prior to screening.
TRIAL DESIGN AND PROCEDURES
This was a triple-masked, randomized, placebo-controlled trial. Twenty participants were allocated to a single dose of intravenous ketamine (0.5 mg/kg diluted into 40 ml of normal saline, infused over 40 minutes using a programmable pump). Another twenty participants were allocated to 40 ml of normal saline infused similarly over 40 minutes. Pharmacy staff randomized participants using computerized block randomization with 5 blocks of 8. The participant, investigators, and direct care providers (e.g., anesthesiologists) were masked. Unmasking occurred after all 40 participants progressed through all follow-up timepoints. Processed electroencephalography (EEG) from a SedLine® device (Masimo Corporation, Irvine, California, USA) was used to confirm anesthetic depth, measured by the device's Patient State Index (PSI). To ensure participant masking, the infusion was initiated after anesthetic induction and surgical incision, during maintenance anesthesia (PSI of 25 to 50, consistent with the manufacturer recommendations for general anesthesia). The study drug was provided to the anesthesiologist in an unlabeled syringe. Anesthesiologists, masked to patient group allocation, administered routine anesthesia tailored to the surgical procedure and patient comorbidities; they were asked to avoid altering the anesthetic in response to any perceived changes to the processed EEG during the study drug infusion (excepting excessively high PSI values indicating the patient was at risk of intraoperative awareness). Anesthesiologists were asked to minimize use of nitrous oxide (N2O), which has reported antidepressant effects. Agents used for anesthetic maintenance included intravenous propofol and inhaled sevoflurane or isoflurane. A standard multimodal analgesic regimen was used, consisting of intravenous opioid and acetaminophen, with or without intravenous lidocaine.
OUTCOME MEASURES
The primary outcome was the MADRS score measured 1, 2, and 3 days post-infusion, as previous studies have found the greatest antidepressant effect occurs within the first 72 hours of a single ketamine infusion. The same sample was assessed at 1, 2 and 3 days post-infusion, as participant dropout only occurred after outcomes were assessed on day 3. Additional assessments were made 5, 7, and 14 days post-infusion and used for exploratory analyses. The MADRS is a clinical rating scale used widely in antidepressant trials; it consists of 10 items which measure the severity of depression in individuals, with a total score ranging from 0 to 60, and higher scores indicating more severe depression. Trained clinical research personnel administered the MADRS. The secondary outcome was clinical response, defined as ≥50% reduction in MADRS scores from screening baseline. Remission, defined as MADRS score ≤12 in our study, was treated as an exploratory outcome. Other exploratory outcomes included the HADS score, postoperative pain intensity and cumulative opioid use. The HADS is a self-reported questionnaire used to assess the severity of anxiety and depression in hospital patients; it consists of 14 items (7 items measuring anxiety, 7 items measuring depression) with a total score ranging from 0 to 42, with higher scores indicating more severe symptoms. Postoperative pain was assessed by the Brief Pain Inventory Short Form modified for postoperative use (BPI-SF). The BPI-SF measures severity of pain and its impact on daily functioning; it consists of 9 items, each using a numeric rating scale from 0-10. Inpatient postoperative opioid use, calculated as total daily MME, was abstracted from the EHR for each day of hospitalization. Outpatient postoperative opioid use, via pill counts, was obtained during remote assessments. At 14 days post-infusion, participants were asked to guess their treatment arm. Outcomes were assessed in person during postoperative hospitalization, and by video or telephone after discharge.
STATISTICAL METHODS
Intention-to-treat (ITT) analysis was performed for the primary outcome. A mixed-effects model for repeated measures was the analysis strategy pre-registered before data unmasking to evaluate the antidepressant superiority of ketamine to placebo on postoperative days 1, 2, and 3. The following fixed effects were included in the model: group, time in days, and the interaction between group and time. We included random effects for intercepts and slopes to account for variation in MADRS scores and differential treatment effects. An alternative, non-prespecified mixed-effects model using change from pre-infusion baseline scores on the day of surgery was also used to analyze the primary outcome. An unstructured covariance matrix was used in all mixed models described in this study. We also calculated Cohen's kappa statistic, in a post-hoc analysis, to assess the level of agreement between groups regarding their guesses on treatment allocation. All analyses were performed using RStudio software (version 2022.07.1 for MacOS). The lme4 package was used for mixed-effects modeling. Our sample size estimation was derived from an a priori power analysis for the primary outcome. In a randomized controlled trial of ketamine versus active placebo performed by Phillips et al., participants had a mean decrease of 10.9 points (standard deviation [SD] 8.9) in MADRS total score relative to pre-infusion scores compared with a mean decrease of 2.8 points (SD 3.6) with midazolam. For reference, the minimum clinically important difference on the MADRS is estimated to range from 3 to 9 points. Using these results, we computed an estimated total sample size of 38 participants at a two-sided alpha level of 0.05 and 80% power to detect this difference if using parametric testing. An additional 2 participants were added to account for potential attrition, for a total of 40 participants. Interim analyses were not performed.
PARTICIPANTS
Participant recruitment occurred between February 2020 and August 2022. For participant flow through the clinical trial, see the CONSORT diagram (Figure). The screening visit occurred between 27 days and 16 hours prior to surgery (mean (SD) 5.1 (4.6) days). The mean age of trial participants was 51 years; they were mostly female (70%), white (65%), non-Hispanic (87.5%), employed (62.5%), and never smoked (65%). At screening, both groups had moderate levels of depression (ketamine: mean MADRS score = 27.7, placebo: 30.6) and moderate levels of treatment resistance(ketamine: mean MSM score = 8.3, placebo: 7.5). The presence of symptomatic depression was also supported by the HADS (ketamine: mean score = 24.6, placebo = 24.7). Although current MDD episode durations were longer in ketamine group, the difference did not reach significance (ketamine: median = 38 months, placebo: 17). Both groups also scored similarly on the BPI-SF at screening, except for two questions in which participants in the ketamine group reported having more pain interference with sleep (mean 7.7 versus 5.5, p=0.02) and enjoyment of life (7.6 versus 5.7, p=0.04). Other characteristics were similar between groups (Table). Tablesummarizes participants' surgical and anesthetic characteristics. Participants in both study arms had similar levels of disease burden, as measured by the American Society of Anesthesiologists (ASA) Physical Status Classification, as well as the Charlson Comorbidity Index. Patients presented to a range of surgical departments which were distributed similarly between the two groups. With regards to anesthetic type, all except one underwent general anesthesia. One participant in the ketamine group had monitored anesthesia care with a neuraxial block; however, the depth of anesthesia was within study parameters. Two participants in the ketamine arm and one participant in the placebo arm received N2O at a concentration of ≥ 50% for ≥1 hour. Use of preoperative and intraoperative opioids and length of surgery were similar between groups.
OUTCOMES
For the primary outcome on post-infusion days 1, 2, and 3, the mixed-effects model (Tableshowed no evidence of effect of group assignment on MADRS scores (95% CI -13.3 to 1.64, p=0.13). The MADRS rate of change also did not differ between groups (95% CI -1.54 to 4.93, p=0. 30). An alternative model using change from pre-infusion baseline MADRS scores on the day of surgery ("day 0") also showed no between-groups difference in change scores (95% CI -6.25 to 7.89, p=0.82). The rate of change for the MADRS change scores also did not differ between groups (95% CI -1.96 to 4.62, p=0.43). Missing MADRS scores among enrolled participants did not exceed 5% at any visit; therefore, missing data were not imputed for the primary outcome. Pre-infusion baseline MADRS scores on day 0 did not differ between trial groups (ketamine: 25.1 [SD 8.3]; placebo: 29.9 [SD 7.0]). From day 0 to day 1, the average change in MADRS scores was -12.4 points (SD 9.2) in the ketamine group and -14.7 points (SD 9.0) in the placebo group, corresponding to a mean decrease of 46% and 48% on the MADRS, respectively. In both groups, MADRS scores increased slightly on day 2 relative to the nadir at day 1, but the decrease from baseline persisted through all follow-up time points up to day 14 (Figure). The HADS scores also followed a trajectory similar to the MADRS scores (Extended Data Table). At the end of the follow-up period, participants were asked to guess which intervention they had received; 36.8% of all participants guessed correctly, and the distribution of guesses between groups was comparable, with Cohen's kappa = 0.33, indicating fair agreement between groups (Figure). MADRS change scores relative to day 0 are visualized in Figure. Our secondary outcome was clinical response, defined as ≥50% reduction in MADRS scores from screening baseline. On post-infusion day 1, 60% and 50% of participants in the ketamine and placebo group, respectively, met criteria for clinical response. Rates of clinical response in both trial groups remained similar to each other on post-infusion days 2 and 3 (Figure). We also analyzed remission, which we defined in our study as MADRS score of ≤12. Remission occurred in 50% and 35% of participants in the ketamine and placebo group, respectively, on post-infusion day 1; this difference closed on post-infusion day 3, when 40% of both groups remained in remission (Table). Cumulative opioid consumption did not differ between groups (Figure). Of note, no participants received either preoperative or postoperative methadone or buprenorphine, nor were any maintained on opioid antagonist therapy. Average pain intensity on the BPI-SF at 14 days post-infusion was not different between groups (ketamine: 4.8 [SD 1.5]; placebo: 3.7 [SD 2.0]). Pain interference on the BPI-SF at day 14 post-infusion also did not differ between groups (ketamine: 6.2 [SD 2.2]; placebo: 5.7 [SD 3.5] ). Hospital length of stay was longer in the placebo group (mean 1.9 [SD 1.7] days versus 4.0 [SD 3.3] days, p=0.02).
PROTOCOL VIOLATIONS AND SAFETY
One participant in the ketamine group initially met inclusion criteria at screening but was retrospectively found not to have maintained her symptom severity on the morning of surgery (additive MADRS and HADS score of 30, below the minimum of 31). This participant was randomized and included in the ITT analysis. No protocol deviations related to the administration of the study drug occurred in this trial. Adverse events were evaluated at every visit. Notably, one death in the ketamine group occurred 5 days post-infusion, which triggered the unmasking of treatment assignment. The patient had been discharged from the hospital on postoperative day 3, with no surgical or anesthetic complications documented prior to discharge. Subsequently, the patient experienced a witnessed cardiac arrest at home; advanced cardiac life support was initiated by paramedics and continued in the emergency room until the patient expired.
SENSITIVITY ANALYSES
We tested whether our results were sensitive to: 1) a possible difference in pre-infusion baseline MADRS scores, and 2) the exclusion of participants who received nitrous oxide, an anesthetic which may also have antidepressant qualities. We adjusted for a possible difference in preinfusion baseline MADRS scores by including it as a fixed covariate and specifying random effects only for slopes in our mixed-effects model; this showed no between-group difference in MADRS scores (95% CI -8.41 to 4.49). When we applied our original mixed-effects model for the primary outcome after excluding 3 participants who received nitrous oxide at a concentration of 50% for at least 1 hour, there was also no between-group difference in MADRS scores (95% CI -13.3 to 2.74, p=0.13).
DISCUSSION
This randomized, triple-masked trial compared the short-term antidepressant efficacy of ketamine with placebo in adults with moderate-to-severe depression. There was no effect of treatment on our primary outcome, MADRS scores on days 1, 2, and 3 post-infusion. Baseline MADRS scores obtained at screening and on the day of surgery did not meaningfully differ between groups, supporting the effectiveness of randomization. In both trial groups, the observed decrease in MADRS score at day 1 was similar to, or exceeded, the decreases observed in previous ketamine trials in awake patients. The variance in MADRS change scores observed in our study is also comparable to previous studies in awake patients, supporting our a priori power calculation to detect a between-group difference. The HADS, an alternative patient-rated depression scale, yielded a similar score trajectory as the MADRS and strengthens our conclusion that ketamine and placebo did not differentially impact mood in this trial. Participant retention was excellent, with no loss to follow-up occurring within the primary outcome window. Notably, one participant death occurred in the ketamine group. However, this severe adverse event was attributed to underlying cardiovascular comorbidities rather than a direct result of trial procedures, consistent with previous analyses of cardiovascular safety outcomes after intravenous ketamine infusion. Remarkably, rates of clinical response and remission were similarly high in both groups and on par with respective rates reported in a recent meta-analysis. The surprisingly robust clinical response and remission rate observed in both arms of this trial raises the question of whether anesthetics besides ketamine may have antidepressant effects. N2O has been shown to improve depression symptoms in patients with TRD. However, only 3/40 participants in our study were exposed to N2O at a concentration of 50% for ≥ 1 hour, making it unlikely to impact depression scores at the group level, as confirmed in our sensitivity analyses. Propofol infusions and inhaled isoflurane have also shown antidepressant properties when given at doses that suppress EEG activity ("burst suppression") for 15 minutes, over multiple administrations, which differs substantially in depth and timing from standard surgical anesthesia used in our study (the recommended PSI range of 25-50 avoids burst suppression). Opioids are also routinely used during surgical anesthesia, and recent evidence shows that blocking opioid receptors attenuates the antidepressant effect of ketamine. No patients were on opioid antagonist or partial agonist therapy, and the average daily MME use in both groups prior to surgery was relatively low. While it is possible that opioid exposure interferes with the antidepressant effect of ketamine, we would expect smaller than usual effects across groups; we saw effect sizes equivalent to prior reports of non-surgical patients receiving ketamine. Baseline heterogeneity in psychiatric characteristics could potentially explain the smaller-thananticipated difference in post-treatment depression scores. Although clinical and sociodemographic characteristics were largely similar between trial groups, there was a notable difference in current MDD episode length-with the ketamine group having a longer median episode duration (38 months) compared to the placebo group (17 months). A longer current MDD episode may predict more treatment resistance to traditional antidepressants. However, studies comparing characteristics of responders and non-responders to ketamine therapy have been mixed, with some studies showing that current MDD episode length impacts treatment responsewhile other studies do not. We also cannot rule out the effect of surgical heterogeneity between groups, which we did not control for in our recruitment design; however, between-group differences in case counts did not exceed 3 for any surgical department, and intraoperative factors, including length of surgery and types of anesthetic used, did not meaningfully differ between groups. Other studies have also evaluated the effect of ketamine on mood ratings in surgical patients; however, numerous methodological limitations prevent direct comparison to studies of intravenous ketamine for depression in the psychiatric literature. Frequently, these trials were not conducted in a population likely to meet criteria for moderate-to-severe MDD. In the Prevention of Delirium and Complications Associated with Surgical Treatments (PODCAST) study-the largest study to date comparing ketamine with saline in surgical patients-depression scores were analyzed as a secondary outcome from a study designed to evaluate the efficacy of ketamine for postoperative delirium in patients >60 years old. Notably, only 9.6% of participants met the PHQ-8 cutoff for moderate depression preoperatively, and no diagnostic data or clinician-rated scales were reported. Kudoh et al. and Liu et al. enrolled surgical patients with mild to moderate depression severity. A 2021 RCT testing ketamine during surgical anesthesia required moderate-to-severe depression (MADRS ≥ 22) for eligibility; however, these participants underwent intracranial tumor resection, a population we excluded due to the possibility of mood and personality changes associated with cortical lesions and resections of such lesions. A key strength of our trial was the evaluation of participant masking. At their last follow-up visit, patients in both groups allocated their guesses in similar proportions and fewer than half guessed correctly. The intervention was effectively masked-an uncommon finding among antidepressant trials with ketamine. Assessment of masking is also rare among RCTs involving ketamine and electroconvulsive therapy (ECT), an important antidepressant treatment delivered to briefly anesthetized patients. Most RCTs evaluating the effect of adjunctive ketamine on ECT outcomes have found no benefit of ketamine given at doses of 0.5 mg/kg or greater. Of the RCTs reviewed by McGirr et al., only one reported on masking effectiveness. Outcome expectancy related to the stated intent of the trial may drive apparent treatment effects. Previous studies of ketamine in surgical patients generally find that when patients are recruited to test ketamine's antidepressant effect as a primary outcome, depression scores decrease postoperatively. Conversely, patients in the PODCAST study, who were recruited to participate in a trial focused on reducing postoperative delirium, depression symptoms (a secondary outcome) worsened slightly in the postoperative period. One limitation of our study is that we did not assess the blind of the anesthesiologists who administered the study drug. While it is possible that close inspection of the intraoperative processed EEG could reveal changes consistent with a 0.5 mg/kg subanesthetic ketamine infusion, we specifically instructed the anesthesiologists to avoid altering the patient's anesthetic in response to the processed EEG during drug infusion (barring large excursions in PSI that correlate with patient awareness). Nonetheless, we cannot exclude the possibility that anesthesiologists who guessed the patient's treatment allocation may have altered their anesthetic in a way that influenced postoperative mood. Our results suggest that when differential subject-expectancy bias is minimized with successful masking, the treatment effect size of ketamine is reduced considerably. However, a major limitation of our study is that we did not measure treatment expectancies prior to randomization. Therefore, we cannot definitively conclude that subject-expectancy bias mediates the causal relationship between effective masking and smaller treatment effect sizes. Regardless of the intervention being tested, subject expectations of a positive outcome-also known as hope-may drive large decreases in depression symptoms seen in antidepressant trials. Our trial design cannot distinguish between a null-effect of ketamine for depression and an occlusion of ketamine's antidepressant effect through a placebo-like mechanism maintained in the absence of unmasking.
CONCLUSION
This trial utilized surgical anesthesia to successfully mask the allocation of a single antidepressant dose of ketamine or placebo in a sample of depressed patients and found that depression scores at 1, 2, and 3 days post-infusion did not differ between trial groups. Secondary and exploratory outcomes also found no evidence of benefit for ketamine over placebo. These findings differ from those of prior antidepressant trials with ketamine conducted without adequate masking, where the large effect sizes reported may reflect expectancy bias. Our results suggest that ketamine may actually be ineffective for the short-term treatment of MDD. While it is impractical to use surgical anesthesia for most placebo-controlled trials, future studies of novel antidepressants with acute psychoactive effects should make stronger efforts to mask treatment assignment to minimize the effects of subject-expectancy bias. use proposal for review and approval. Please direct all inquiries and requests to bheifets@stanford.edu.