Real-World Effectiveness of At-Home Sublingual Ketamine Telehealth for Anxiety and Depression: A Retrospective Observational Study

Retrospective real-world analysis (15 US states; January–November 2021; institutional review board approved as exempt) of de-identified routine clinical and quality-management data from a telemedicine ketamine-assisted therapy (KAT) service. Patients completed online eligibility screening followed by a video-based intake with a prescribing psychiatric clinician confirming diagnosis and at-home safety. Inclusion: age ≥18, reliable internet access, clinician-assigned diagnosis of depression (PHQ-9 ≥10) and/or anxiety (GAD-7 ≥10). Exclusion: ketamine allergy, current substance dependence, history of opioid use disorder, active/past psychosis or mania, uncontrolled cardiac disease, pregnancy/nursing, recent suicidal intent or attempt. N=1,247 patients with sufficient outcome data identified from a chart review of approximately 2,848–4,334 records (discrepancy noted in the paper text). Treatment: 300–450 mg rapidly dissolving sublingual ketamine tablets (initial dose ~5 mg/kg, adjusted for dissociative response), mailed to patients; administered at home with a physically present peer monitor and real-time guide support via text during each 8-hour session. Guides provided 30-minute pre- and post-session video calls plus daily text check-ins. Clinicians conducted follow-up video consultations 1–2 days after the first session. Outcomes: PHQ-9 (depression) and GAD-7 (anxiety) assessed at baseline, after Session 2 (week 2), and after Session 4 (week 4). Safety outcomes: C-SSRS suicide screening (baseline), AUDIT, DAST-10, plus real-time vital sign monitoring. Dissociation measured with a 3-item adaptation of the Clinician Administered Dissociative States Scale. Primary analyses included response rate (≥50% score reduction), remission (follow-up score <5 with baseline ≥10), and clinically significant change (≥5-point reduction below clinical threshold). Growth Mixture Modelling (Mplus 8, parallel PHQ-9 and GAD-7 processes) identified latent trajectory classes.