Ketamine for Social Anxiety Disorder: A Randomized, Placebo-Controlled Crossover Trial

Social anxiety disorder (SAD) is common and causes substantial functional impairment, yet roughly one-third to one-half of patients do not achieve meaningful symptom relief with current evidence-based treatments such as selective serotonin reuptake inhibitors (SSRIs), venlafaxine, or cognitive behavioural therapy. Converging neuroimaging, spectroscopy, and preclinical data implicate glutamatergic abnormalities in SAD, including elevated glutamate or glutamine in regions such as the anterior cingulate and thalamus. Ketamine, an N-methyl-D-aspartate (NMDA) receptor antagonist that modulates glutamate signalling, has produced rapid antidepressant and anxiolytic effects in prior controlled studies of depression and preliminary work in obsessive-compulsive disorder and post-traumatic stress disorder. Taylor and colleagues therefore designed a proof-of-concept, double-blind, randomised, placebo-controlled crossover trial to test whether a single intravenous infusion of ketamine (0.5 mg/kg over 40 minutes) reduces social anxiety symptoms in adults meeting DSM-5 criteria for SAD. The trial aimed to characterise the time course of effects over two weeks post-infusion and to compare clinician-rated and self-reported anxiety measures, recognising uncertainty about durability and the optimal outcome measures for ketamine's anxiolytic effects.

Design and participants: The study was a double-blind, randomised, placebo-controlled crossover trial conducted at Yale University from 2014–2016. Eighteen adults aged 18–65 meeting DSM-5 criteria for SAD by structured interview were enrolled. Inclusion required a baseline Liebowitz Social Anxiety Scale (LSAS) score indicative of at least moderate social anxiety. Comorbid anxiety and mood disorders were permitted; other lifetime Axis I diagnoses were excluded. Participants on psychiatric medication were eligible if doses had been stable for at least one month and for SSRIs/SNRIs/clomipramine at least two months. Substance use disorders (other than tobacco) and positive drug screens were exclusionary. Recruitment sources included clinicaltrials.gov and community outreach. Intervention and procedure: Each participant received two infusions in random order: ketamine 0.5 mg/kg IV over 40 minutes and matched placebo (normal saline), separated by a 28-day washout. Randomisation was performed by the investigational drug service and concealed from study investigators. Blinded raters conducted clinician-rated assessments but were not present during infusions; separate raters at the infusion assessed dissociative/psychotomimetic effects. Participants were asked not to discuss infusion experiences with blinded raters to reduce inadvertent unblinding. Outcomes and timing: Assessments occurred pre-infusion, 3 hours post-infusion, and on days 1, 2, 3, 5, 7, 10 and 14 post-infusion. The primary clinician-rated outcome was the LSAS (0–144). The primary self-report outcome was a visual analogue scale for anxiety (VAS-Anxiety, 0–100). Secondary measures included the State-Trait Anxiety Inventory—State subscale (STAI-S) and the Hamilton Depression Rating Scale (HDRS-17) for depressive symptoms; dissociative effects were measured with the Clinician Administered Dissociative States Scale (CADSS). The LSAS was modified to ask about symptoms since the previous assessment. Statistical analysis: Analyses used SAS 9.4. Normality of continuous outcomes was confirmed by Shapiro–Wilk tests. The investigators tested for carryover effects by comparing baseline ratings between sequence groups (ketamine-first versus placebo-first) using paired t-tests; evidence of carryover for a given outcome led to analysing only data from the first phase for that outcome. For outcomes without carryover, mixed-effects linear models included treatment, infusion order, time, and relevant interactions; non-significant order terms were removed. Restricted maximum likelihood estimation with an autoregressive covariance structure was used. Cohen's d with confidence intervals was reported for treatment effects at each time point. Treatment response definitions followed prior literature: >35% reduction in LSAS and >50% reduction in VAS from baseline at any point ≥24 hours post-infusion. The Wilcoxon signed-rank test compared responder proportions. Alpha was set at 0.05. The extracted text indicates that depression analyses included only participants above an HDRS-17 threshold suggestive of mild to moderate depression, but the exact cutoff symbol was not clearly extracted.

Sample and retention: Of 18 randomised participants, 17 (94%) completed both infusion phases; one participant missed the second (placebo) infusion for non-study reasons. All ketamine infusions were completed without reported serious adverse outcomes in the extracted text. Carryover effects: A significant carryover effect was detected for the LSAS: participants who received ketamine in the first period had lower LSAS baselines before the second period than those who received placebo first (mean difference 15.35, t = 2.84, df = 15, p = 0.012). No significant carryover was detected for VAS-Anxiety, HDRS-17, or STAI-S. Clinician-rated social anxiety (LSAS): Restricting to phase 1 for the LSAS due to carryover, mixed-model analyses showed a significant Time effect (F 8,124 = 3.0, p = 0.004) and a significant Time × Treatment interaction (F 9,115 = 2.6, p = 0.01), favouring ketamine over saline. Significant between-group mean differences (ketamine minus placebo) occurred at day 2 (diff = 22.67 ± 7.28, effect size (ES) = 1.10, F = 4.9, p = 0.04), day 5 (diff = 22.33 ± 7.28, ES = 1.09, F = 4.7, p = 0.04), and day 10 (diff = 28.72 ± 7.36, ES = 1.37, F = 7.6, p = 0.01). Six of 18 subjects (33.33%) met the predefined LSAS responder criterion (>35% reduction) after ketamine versus 0 of 17 after placebo (Wilcoxon z = 2.24, p = 0.025). Subscale analyses indicated significant Time and Time × Treatment effects for LSAS Fear and Avoidance subscales, and for Social Anxiety and Social Avoidance subscales; Performance Anxiety and Performance Avoidance subscales did not show significant Time × Treatment interactions. Maximum CADSS score and age at SAD onset did not significantly moderate LSAS effects. Self-reported anxiety (VAS, STAI-S): On the VAS, time alone was significant (F 10,141 = 5.8, p < 0.001) but the Time × Treatment interaction was not (F 10,141 = 0.4, p = 0.95) and the treatment main effect approached but did not reach conventional significance (F 1,27.4 = 3.4, p = 0.07). No individual time point showed a significant ketamine advantage on VAS. Nevertheless, responder analysis found 16 of 18 (88.89%) achieved >50% VAS reduction after ketamine versus 9 of 17 (52.94%) after placebo (Wilcoxon z = 2.12, p = 0.034). STAI-S scores showed significant treatment effects: Time (F 7,167 = 5.4, p < 0.001) and Treatment (F 1,47 = 8.2, p = 0.006); between-group differences favouring ketamine were observed at days 1 (diff = 7.76 ± 2.30, ES = 0.83, F = 5.7, p = 0.02), 7 (diff = 8.53 ± 2.28, ES = 0.92, F = 7.0, p = 0.01) and 10 (diff = 8.39 ± 2.64, ES = 0.78, F = 5.0, p = 0.03). Sensitivity analyses restricted to phase 1 produced similar STAI-S findings. Depressive symptoms: Among 12 participants with comorbid depressive symptoms at baseline (HDRS-17 mean = 22, SD = 5), ketamine did not produce significant reductions in HDRS-17 scores in either the full crossover analysis or phase 1–only analyses (time and interaction terms not significant). The authors note few participants met criteria for major depressive disorder and the study was underpowered for depression outcomes. Safety and tolerability: Ketamine was generally well tolerated. Common adverse effects reported in the extracted text included paresthesias and numbness (n = 9), feeling disconnected from reality (n = 7), visual changes (n = 6), and confusion or difficulty thinking (n = 6); the text breaks off during a list that included dizziness, so the full adverse-event list is not completely available in the extraction.

Taylor and colleagues interpret their findings as initial, proof-of-concept evidence that a single ketamine infusion reduces clinician-rated social anxiety symptoms in adults with SAD over the short term, while effects on self-reported momentary anxiety (VAS) were not observed in mixed-model analyses. The study also found higher proportions of treatment responders after ketamine on both clinician-rated (LSAS) and self-reported (VAS) definitions, and significant improvements on the STAI-S at several post-infusion time points. The authors situate these results within prior literature showing ketamine's rapid anxiolytic effects in depression and preliminary benefits in obsessive-compulsive disorder and post-traumatic stress disorder. They note that, despite statistically significant reductions, the overall sample on average continued to exhibit moderate social anxiety after ketamine and that some individual responders still retained moderate symptoms shortly after infusion. Ketamine did not significantly reduce depressive symptoms in this sample, a null result the investigators attribute to the small number of participants with current major depressive disorder, generally milder depressive severity in the cohort, and limited statistical power. Key limitations acknowledged by the study team include inadequate blinding using saline—17 of 18 participants correctly identified when they had received ketamine—which may bias self-report and even blinded clinician ratings. The LSAS showed carryover effects through 28 days, prompting reliance on first-period data for that outcome; the authors suggest ketamine's anxiolytic effects may have enabled behavioural activation (increased social exposure), which could have sustained benefit but was not systematically measured. Other limitations include the small sample size and the crossover design given apparently durable anxiolytic effects. For future research the authors recommend larger, parallel-group trials using an active comparator such as midazolam to improve blinding, prioritising blinded clinician ratings over single-item VAS measures, incorporating physiological anxiety measures (for example skin conductance), exploring optimal dosing and repeated administration schedules, and examining potential genetic moderators (for example BDNF rs6265). They caution that these results are preliminary and should not be taken as definitive evidence for long-term use of ketamine in anxiety disorders, but suggest that ketamine-like glutamatergic compounds merit further investigation for SAD and other anxiety disorders.