Antidepressant effects of a psychedelic experience in a large prospective naturalistic sample

Depression treatment advances have been limited since the 1980s, and interest in psychedelic-assisted interventions has re-emerged since the mid-2000s. Earlier controlled trials and population studies have indicated possible antidepressant and anxiolytic effects after one or a few psychedelic doses, and prior work has emphasised the importance of contextual or "set and setting" variables and the quality of the acute psychedelic experience (for example mystical-type experiences, challenging experiences and emotional breakthrough) in mediating longer-term outcomes. This study aimed to extend that literature into a large prospective naturalistic sample by examining changes in depressive and anxiety symptoms before versus after a self-initiated psychedelic experience. Specifically, the investigators tested whether drug dose, motive for use (with a focus on a medicinal motive), prior psychedelic experience and acute experience quality (emotional breakthrough, mystical-type and challenging experiences) predicted symptom change. Depressive symptoms measured by the Quick Inventory of Depressive Symptomatology (QIDS-SR-16) served as the primary outcome, assessed at baseline and 2 and 4 weeks post-experience via an online prospective survey platform.

This was a prospective naturalistic observational study using an online recruitment and survey platform (Psychedelic Survey). Eligible participants were adults (≥18 years) with good English comprehension who planned to take a psychedelic drug in the near future; planned drugs included classic serotonergic psychedelics and others (psilocybin, LSD/1P-LSD, ayahuasca, DMT/5‑MeO‑DMT, mescaline, salvia, iboga/ibogaine). Participants provided informed consent online, gave an email and expected date of experience, and received emailed survey links timed to four assessment points: about 1 week before the experience (baseline), 1 day after, 2 weeks after and 4 weeks after. The present analyses used data from baseline, 1 day, 2 weeks and 4 weeks. To align the sample with clinical relevance, individuals with minimal depressive symptoms at baseline (QIDS < 6) were excluded, producing an analysis sample of N = 302 at baseline. Key measures included the QIDS-SR-16 (primary outcome) and a short-form Trait Anxiety measure (STAI-T) as the main secondary outcome. Acute experience measures collected one day after the session comprised the Mystical Experience Questionnaire (MEQ), Challenging Experience Questionnaire (CEQ) and Emotional Breakthrough Inventory (EBI). Drug dose was categorised in LSD-equivalents (low to extremely high). Baseline covariates included age, gender, education, nationality, psychiatric history, prior psychedelic use (frequency categories) and motives for use (rated on a 4-point scale), with particular focus on a medicinal motive. Setting variables (retreat, therapeutic framing, recreational framing, number present, presence of a welfare-responsible person) were also recorded. Statistical analyses were performed in Stata 15. Descriptive analyses and Spearman correlations explored bivariate associations. Linear mixed models (LMM) were used to assess change over time and to handle missing data, with time treated as a repeated effect, an unstructured covariance, and a random intercept to model within-subject variance. Models controlled for age, gender and education, and some models included baseline QIDS as a covariate. Predictors evaluated in LMMs included drug dose, medicinal motive and prior psychedelic experience and their interactions with time. A separate LMM examined acute-experience predictors (MEQ, CEQ, EBI) and their interactions with time; because MEQ and EBI were collinear (r = 0.52), secondary models excluded one or the other to assess robustness. Estimated mean changes (M-change) and standardised effect sizes (Cohen's d) were reported; standardised regression coefficients were computed to show predictor effects.

Sample sizes were N = 302 at baseline, N = 182 at 1 day, N = 155 at 2 weeks and N = 109 at 4 weeks. Mean age was 27.4 years (±10.1), 70.9% were male, 74.2% had undergraduate education or higher, and 46% reported a prior psychiatric diagnosis. Baseline mean QIDS was 9.9 (±3.7) and baseline STAI-T was 47.4 (±12.5). Most participants had prior psychedelic experience (about 80% had used more than once). LSD and psilocybin were the most commonly used drugs and a majority took moderate-to-high doses. More than half reported using psychedelics in a therapeutic setting; regarding companionship and support, 39% had others present with guidance, about one-third had others without guidance and about one-third were alone. Predictors of attrition included younger age, lower education, lower conscientiousness and higher extraversion. Primary clinical outcomes showed significant reductions in depressive and anxiety symptoms at 2 weeks, with no further significant change from 2 to 4 weeks. QIDS decreased from baseline to 2 weeks with an estimated M-change = -4.37 (SE = 0.32), p < 0.001, Cohen's d = 1.18; baseline to 4 weeks M-change = -4.17 (SE = 0.37), p < 0.001, Cohen's d = 1.13. STAI-T decreased from baseline to 2 weeks (M-change = -6.17, SE = 0.90, p < 0.001, Cohen's d = 0.49) and from baseline to 4 weeks (M-change = -5.33, SE = 1.05, p < 0.001, Cohen's d = 0.43). When stratified by baseline depression severity, all groups showed significant QIDS reductions from baseline to 2 weeks and little change from 2 to 4 weeks. Participants with moderate depressive symptoms decreased to a mean of M = 6.97 (SE = 0.62) at 2 weeks (p < 0.001; Cohen's d = 1.52). Those with severe baseline symptoms decreased to M = 6.34 (SE = 1.00) at 2 weeks (p < 0.001; Cohen's d = 3.28); a non-significant increase in the severe group from 2 to 4 weeks suggested a tendency toward symptom recurrence (M-change = 2.19, SE = 1.32, p = 0.096). In a subsample with moderate-to-severe baseline symptoms, very large effect sizes were reported (Cohen's d = 2.4 at 2 weeks and d = 2.1 at 4 weeks). Predictor analyses for change in QIDS from baseline to 2 weeks identified several significant associations after controlling for age, gender, education and baseline QIDS. Having a stronger medicinal motive was associated with larger reductions in depressive symptoms (interaction B = -0.991, standardised β = 0.316, p = 0.002). Higher drug dose predicted larger decreases (B = -0.824, β = 0.196, p = 0.029). Conversely, greater lifetime prior psychedelic use predicted smaller decreases in depressive symptoms (B = 0.508, β = 0.213, p = 0.033). In models testing acute experience measures, the Emotional Breakthrough Inventory score predicted greater reductions in QIDS (interaction B = -0.024, β = -0.201, p = 0.039). The MEQ was not a significant predictor in models that included the EBI, and when the EBI was excluded the MEQ remained non-significant (B = -0.026, β = -0.155, p = 0.095); the MEQ and EBI displayed collinearity (r = 0.52).

Nygart and colleagues report that, in this large prospective naturalistic sample, self-initiated psychedelic experiences were followed by large reductions in depressive symptoms and moderate reductions in trait anxiety at 2 and 4 weeks. The magnitude of depressive symptom change (Cohen's d ≈ 1.1 overall; larger in those with more severe baseline symptoms) was noted to be comparable to effect sizes reported in controlled trials, prompting the authors to suggest that some beneficial outcomes seen in clinical settings may translate into real-world use under certain circumstances. The investigators highlight several predictors consistent with prior work on context and acute experience. Higher psychedelic dose and having a medicinal motive were associated with greater improvement, while greater prior lifetime psychedelic exposure predicted smaller improvements. Emotional breakthrough during the acute session was the only acute-experience measure that significantly predicted 2-week antidepressant response when included alongside the MEQ and CEQ; the MEQ did not independently predict response in models that included the EBI, an outcome the authors attribute in part to collinearity between mystical- and emotional-breakthrough measures. The discussion situates these findings within existing literature linking intensity and specific qualities of the acute psychedelic experience to therapeutic outcome, and emphasises psychological insight and emotional processing as plausible mediators. The authors acknowledge several limitations inherent to the naturalistic uncontrolled design. The sample was self-selected, skewed young, male and well educated, and many participants were experienced psychedelic users; these factors limit generalisability and may bias results. Attrition was substantial and may have introduced bias despite analyses showing that acute-experience quality and depressive symptom severity did not predict dropout; demographic and personality variables did predict attrition. The absence of a control group, potential expectancy effects linked to medicinal motives, and reliance on self-report instruments were also noted. The researchers therefore advise caution in interpreting causality and call for further controlled and longitudinal observational studies to replicate and extend these findings. Finally, the authors reflect on implications: while the results add naturalistic evidence supporting therapeutic potential and underscore the importance of dose, motive and acute experiential factors, they recommend careful public messaging about context-dependency and safety as policy and decriminalisation debates progress.

The study concludes that naturalistic psychedelic experiences were associated with robust reductions in depressive symptoms and moderate reductions in trait anxiety at 2 and 4 weeks among participants with baseline depressive symptoms. Larger improvements were predicted by higher drug dose, a stronger medicinal motive, fewer prior psychedelic uses and experiencing an emotional breakthrough during the session. Given the uncontrolled and self-selected nature of the sample and substantial attrition, the authors emphasise that these data are vulnerable to bias and that controlled and further observational research is needed to confirm and clarify these associations.