At-Home Telehealth-Supported Subcutaneous Ketamine Therapy in Adults With Moderate to Severe Depression, Anxiety, or PTSD: A Real-World Observational Study of Safety, Feasibility, and Clinical Outcomes in a Large, Heterogeneous Cohort in the United States

Parks and colleagues begin by noting that major depressive disorder, generalised anxiety disorder and PTSD remain major causes of disability despite longstanding research into medication and psychotherapy. They argue that a large gap persists between evidence-based treatment and real-world access, and that even when care is available, current first-line options often act slowly, have limited remission rates, or are associated with attrition and side effects. Existing ketamine approaches, particularly intravenous and intranasal administration, can produce rapid effects but are constrained by cost, logistics and access barriers. Against this background, the paper sets out to evaluate a telehealth-mediated, at-home subcutaneous ketamine protocol. The authors state three main aims: to assess safety, to examine symptom change across depression, anxiety and trauma-related distress, and to evaluate feasibility in a real-world setting. They position subcutaneous delivery as a potentially higher-bioavailability alternative to sublingual at-home ketamine, with the possibility of combining clinical precision with remote accessibility. The study is presented as a large retrospective observational analysis of routine clinical data from a US telehealth ketamine service, and the authors frame it as an early large-scale test of this model across multiple psychiatric conditions.

The study used retrospective chart review data from Mindbloom, a US direct-to-consumer telehealth ketamine provider operating across 38 states. Patients were screened by psychiatric clinicians using intake information, self-report measures, a synchronous video consultation and prescription monitoring programme checks. The sample consisted of adults with moderate to severe symptoms, defined as PHQ-9 scores of at least 10 for depression, GAD-7 scores of at least 10 for anxiety, or PCL-5 scores of at least 33 for PTSD. The extracted text lists exclusions including ketamine use disorder, active psychotic or manic symptoms, primary psychotic disorder, recent suicidality, uncontrolled hypertension, serious cardiopulmonary disease, pregnancy and other severe systemic disease. Some higher-risk cases, such as active substance use disorders or severe trauma histories, required case-by-case review and external support confirmation. The protocol followed an at-home subcutaneous ketamine regimen delivered with remote clinician oversight and guide support. Patients received education, self-administration training and a mandatory return demonstration. A peer treatment monitor had to be present for sessions, and patients were supplied with a blood pressure monitor, injection kit and ondansetron for nausea prevention. Treatment began with 0.5 mg/kg and could be titrated by clinicians in 10-20 mg increments, with the first three doses sent after eligibility confirmation and the next three after a follow-up video consultation. The analysis focused on the first six sessions, and the maximum dose was capped at 120 mg or 1.2 mg/kg, whichever was lower. Outcomes were assessed at baseline and after sessions 2, 4 and 6 using the PHQ-9, GAD-7 and PCL-5. Side effects were queried after sessions 2, 4 and 6 using a single self-report item, with additional adverse events recorded in the electronic health record. The primary analyses used random-intercept mixed-effects models with cubic splines for time to handle non-linear symptom trajectories and variable timing between assessments. Models adjusted for age, sex, prior psychiatric diagnosis and final dose. The authors also calculated descriptive statistics, percent change, paired t-tests and sensitivity analyses using multiple imputation and last observation carried forward. No power calculation was performed because this was a retrospective analysis of existing data.

A total of 3,041 patients were included in the overall analysis, drawn from 3,870 initially approved for subcutaneous treatment between January 2024 and October 2025. The mean age was 44.7 years, and 52.4% were female. Most participants reported prior psychiatric treatment. Only 19 patients (0.6%) switched from subcutaneous to sublingual administration, suggesting good acceptability of the protocol. Symptom scores improved across all three measured domains over the six-session course. The authors report adjusted marginal means showing a decline in PHQ-9 from 14.64 at baseline to 6.30 after six sessions, GAD-7 from 13.06 to 6.09, and PCL-5 from 46.7 to 27.5. They describe these changes as statistically significant and large, with effect sizes (dz) ranging from 1.35 to 1.58. Minimal clinically important difference was reached by 81.8% of patients with depression, 80% with anxiety and 84.6% with PTSD. The extracted text indicates that symptom reductions were already evident early in treatment and continued across assessment waves. Side effects were uncommon. After sessions 2, 4 and 6, 3.1%, 3.2% and 2.8% of respondents, respectively, reported side effects. The most common effects were lower abdominal pain at earlier waves and memory loss at the final wave. The authors state that dose was not associated with side-effect risk. By the end of the study, 0.2% reported injection-related discomfort and 0.5% switched away from subcutaneous treatment. Over the 22-month reporting period, three serious adverse events were reported, corresponding to an SAE rate of 0.08% (3/3943). One involved a psychotic episode after the first treatment and was considered likely related to ketamine. Two patients died by suicide during treatment; the paper says attribution to treatment could not be determined. In the mixed-effects models, assessment wave was significantly associated with improvement in PHQ-9, GAD-7 and PCL-5 scores. The authors report that history of a psychiatric disorder was associated with higher symptom scores on all three measures, while higher final dose was weakly associated with higher PHQ-9 and GAD-7 scores. Older age predicted lower GAD-7 scores, and female sex was associated with higher PHQ-9 scores. Sensitivity analyses with multilevel multiple imputation and last observation carried forward gave broadly similar results to the main analysis.

Parks and colleagues interpret the findings as support for their three main hypotheses: that at-home subcutaneous ketamine can be implemented safely, can produce rapid and clinically meaningful symptom reductions, and is feasible in routine practice. They emphasise the low adverse event rate, the very low rate of patients switching administration routes, and the substantial improvements in depression, anxiety and PTSD symptoms across the six-session course. In their view, the results suggest that a supervised remote model can achieve outcomes comparable to clinic-based ketamine approaches while preserving an acceptable safety profile. The authors place these findings in the context of earlier ketamine literature and argue that subcutaneous delivery may help overcome some of the limitations of both intravenous and sublingual administration. They note that sublingual treatment can be limited by variable absorption and patient discomfort, whereas subcutaneous injection offers higher bioavailability and more precise dosing. They also compare the low rate of reported side effects with the higher burden seen with conventional pharmacological treatments, while acknowledging that a direct side-by-side comparison is not possible from this study. Several limitations are explicitly acknowledged. The study is observational and retrospective, so it cannot establish superiority over other treatments or isolate the effects of ketamine from the effects of guide support, clinician contact or the broader integration process. The authors state that ethnicity data were not collected routinely, limiting assessment of group differences and generalisability. They also note that attrition and missing data may introduce bias, although sensitivity analyses suggested the findings were robust. The variable timing of assessment waves meant that the model captured symptom change across sessions while accounting for heterogeneous spacing, but not the precise effect of treatment timing itself. Safety interpretation is also cautious because serious adverse events were rare but included a psychotic episode and two suicides early in treatment, with treatment attribution uncertain. The authors suggest that future studies should disentangle the relative contributions of support and ketamine itself, examine the role of integration, and test the model under stricter time-anchored protocols. They conclude that the real-world design provides useful baseline evidence of safety, feasibility and clinical utility for at-home subcutaneous ketamine in a large, diverse US cohort.

The authors conclude that telehealth-supported subcutaneous ketamine can be delivered at scale at home with acceptable safety and without serious clinical complications in most cases. They argue that remote safeguards such as clinician check-ins, video-verified injection competence and home blood pressure monitoring may allow ketamine care to be decentralised without losing safety. They further state that subanaesthetic home dosing, when combined with clear remote instruction and oversight, is suggestive of clinical benefit and may offer a lower-cost alternative to clinic-based ketamine pathways, including those governed by the more restrictive REMS requirements for esketamine.