Sublingual Ketamine for Depression and Anxiety: A Retrospective Study of Real-World Clinical Outcomes

Depressive and anxiety disorders cause substantial disability and reduced quality of life, and current pharmacotherapies have important limitations: selective serotonin reuptake inhibitors (SSRIs) are slow to act and ineffective in a sizeable minority, while benzodiazepines produce more rapid symptomatic relief but carry sedation, cognitive effects and a high risk of long-term dependence. Ketamine has emerged as a rapidly acting antidepressant with growing evidence of benefit in treatment-resistant depression and preliminary signals for efficacy in generalized anxiety disorder (GAD) and social anxiety disorder (SAD). Most clinical research has used intravenous administration, leaving questions about the feasibility, safety and effectiveness of less invasive, at‑home delivery methods such as sublingual lozenges. Swanson and colleagues designed the present study to examine real-world clinical outcomes following repeated, at-home subanesthetic sublingual ketamine provided through a private telehealth service. The primary aim was to evaluate changes in depression, generalized anxiety and social anxiety symptoms over three follow-up time points. Secondary aims included describing safety (side effects and adverse events), the tendency toward longer‑term use, patient-reported changes in functioning and a comparison of outcomes between treatment-resistant and non-resistant depression cases. The investigators frame this work as a retrospective assessment of routine clinical care to inform feasibility and signal detection for future controlled trials.

This study is a retrospective secondary analysis of de-identified patient records from a private telehealth practice (Wondermed) offering at‑home sublingual ketamine for adults with depression, GAD and/or SAD. Records were eligible if the patient was prescribed ketamine between May 2022 and April 2023 and completed at least one post-treatment assessment within 60 days of baseline. The University of California, Los Angeles IRB determined the analysis did not meet the definition of human subjects research and granted exemption. Patients underwent a clinical evaluation by trained physician assistants and nurse practitioners who applied eligibility screening and contraindication checks; prescription was limited to those judged low-risk by the providers. The typical prescription was a 1‑month supply of sublingual lozenges, most commonly 150–200 mg once weekly; clinicians could individualise dosing within a 50–400 mg range and a maximum frequency of twice weekly. Renewal required completion of a post-treatment questionnaire and a remote follow-up consultation. The intervention package also included an integration programme of daily exercises (journaling, meditation, breathwork, cognitive exercises) and hour-long soundscapes to use during sessions. Outcome data included baseline demographics and clinical characteristics and up to three follow-ups at approximately 30, 60 and 90 days. Primary outcomes were changes in PHQ-9 (depression), GAD-7 (generalised anxiety) and SAD-D-10 (social anxiety) scores. Secondary outcomes were safety (side effects/adverse events as reported after the first month and at follow-ups), prevalence of long-term use (ongoing treatment past 6 months measured by prescriptions dispensed), and patient-reported well-being changes. For subgroup analyses, the sample was divided by baseline severity thresholds: PHQ-9 ≥ 10 for depression, GAD-7 ≥ 10 for GAD, and SAD-D-10 ≥ 15 for SAD. Within the depression subgroup, treatment‑resistant depression (TRD) was defined as prior use of two or more antidepressants without adequate response; a non-resistant cohort included those with one prior inadequate trial, discontinuation for side effects, or no prior antidepressant use. The investigators note that detailed dose/duration data for prior antidepressant trials were not available. Because the outcome data did not meet assumptions for parametric tests, nonparametric methods were used. The Skillings–Mack test assessed changes from baseline to 1, 2 and 3 months while accounting for missing repeated measures, with Dunn's post hoc corrections; Friedman's ANOVA was applied for the subset completing all three follow-ups. Effect sizes (Cohen's d with 95% confidence intervals), and categorical endpoints—response (≥50% symptom reduction), remission (follow-up score < 5), deterioration (increase ≥5 points), and meaningful clinical improvement (MCI, ≥20% reduction)—were reported. Descriptive statistics summarised side effects, prescription counts and patient-reported well-being. Analyses were performed in R.

From 1,028 screened records, 431 patients met inclusion criteria after excluding those without post-treatment assessments, late follow-ups, or invalid data. The analysed sample had mean age 43.6 (SD 10.9) years and was 49.2% female. The mean interval between assessments was about 34 days. Based on baseline severity thresholds, 249 patients met criteria for the depression subgroup, 271 for generalized anxiety and 194 for social anxiety, although the extracted text contains inconsistent denominators for the social anxiety subgroup in some tables (see note below). Safety: Eighty-one of 431 patients (18.8%) reported minor side effects, most commonly dizziness, headache and nausea; all side effects resolved within 24 hours without medical assistance. One patient reported transient increased anxiety/depression that resolved within 24 hours. There were no reports of misuse, addiction, diversion, or serious adverse events (including suicidal behaviour or self-harm) in the analysed records. Regarding treatment duration, 285 of 431 patients (66.1%) received three or fewer prescriptions before stopping, and 397 of 431 (92.1%) stopped after six or fewer prescriptions, indicating limited continuation beyond six months for most patients. Primary clinical outcomes: Across all three subgroups, statistically significant improvements from baseline were observed at each follow-up (p < 0.001). Reported meaningful clinical improvement (MCI, ≥20% reduction) rates in the depression subgroup were 188 of 249 (75.5%) at 1 month, 85 of 106 (80.2%) at 2 months and 47 of 53 (88.7%) at 3 months. In the generalized anxiety subgroup MCI rates were 225 of 271 (83.0%) at 1 month, 108 of 119 (90.8%) at 2 months and 39 of 49 (80.3%) at 3 months. For social anxiety the paper reports 196 of 237 (82.7%) at 1 month, 89 of 102 (87.3%) at 2 months and 47 of 56 (83.9%) at 3 months; however, earlier in the results the social anxiety subgroup size is given as 194, so the denominators for social anxiety outcomes are inconsistent in the extracted text. Treatment response rates were highest at the 3‑month follow-up across subgroups, remission rates rose over time, and deterioration was uncommon (less than 3% overall). The authors note that no deterioration was directly attributed to treatment and that some patients who initially worsened later improved. Functioning and well-being: Self-reported positive changes were common: 391 of 431 (90.7%) reported overall improvements at 1 month, with 188 of 194 (96.9%) at month 2 and 100 of 101 (99.0%) at month 3. Commonly endorsed changes included improved ability to calm down under stress and increased hopefulness about the future. Treatment‑resistant status comparison: Within the depression subgroup, 38 patients met the study's TRD definition and 101 were classified as non-resistant; 110 patients lacked the necessary prior-treatment information and were excluded from this comparison. The extracted text states there was no significant difference in baseline depression severity between TRD and non-resistant cohorts but does not provide the baseline means in the available extraction. Outcomes appeared comparable between cohorts. Long-term treatment beyond six months was infrequent in both groups (3 of 101 non-resistant patients and 5 of 38 TRD patients continued past six months).

Swanson and colleagues interpret their findings as indicating that repeated, at‑home sublingual ketamine was associated with statistically and clinically meaningful reductions in symptoms of depression, generalized anxiety and social anxiety across three follow-up time points in this real-world telehealth cohort. Dosing sessions were generally well tolerated, with only transient, minor side effects and no serious adverse events reported in the analysed records. The investigators suggest that careful patient selection and remote monitoring practices used by the provider may have contributed to the favourable safety profile. The authors emphasise the relevance of the observed benefits for GAD and SAD given limited prior data, and note that care delivered by trained physician assistants and nurse practitioners appeared feasible and safe—an important point in light of workforce shortages. They further propose that ketamine's rapid onset could justify earlier consideration in selected patients rather than reserving it solely for treatment‑resistant cases, particularly when rapid reduction of suicidal ideation is clinically desirable. The study team also highlights that many patients reported improved functioning and well‑being, and they posit that the adjunctive integration resources (meditation, journaling, breathwork and soundscapes) may have supported these outcomes; they recommend further work to define optimal ways to combine ketamine with supportive therapies. The authors acknowledge key limitations: the retrospective observational design, absence of a comparator or control group, and limited generalisability given recruitment from a single private telehealth provider. They note an important selection effect introduced by the provider’s follow-up procedures—post‑treatment assessments were performed as part of prescription renewal, which excluded patients who completed only one month of treatment from outcome analyses and may bias results. Other limitations include incomplete data on prior antidepressant trial dose/duration and heterogeneity in dosing regimens. Finally, the investigators call for prospective randomised trials and longer-term studies to better establish optimal dosing, duration, candidate selection and the abuse/misuse risk of repeated at‑home sublingual ketamine.

In this retrospective, real-world cohort, repeated low-dose sublingual ketamine delivered for at‑home use via a telehealth service was associated with significant reductions in depression, generalized anxiety and social anxiety symptoms, high rates of patient‑reported functional improvement, and a low incidence of transient side effects. No serious adverse events were recorded in the analysed sample. The authors conclude that further controlled and long-term research is needed to define optimal dosing strategies, duration of treatment, patient selection criteria and the longer-term safety and misuse risks of this mode of ketamine delivery.