Acute mood-elevating properties of microdosed LSD in healthy volunteers: a home-administered randomised controlled trial

Microdosing — taking sub-hallucinogenic doses of psychedelics such as LSD — has become a common community practice, with retrospective surveys reporting a variety of mood and cognitive benefits. Randomised controlled trials to date, however, have detected only modest acute effects and no persistent changes from repeated dosing; the investigators note that many prior trials used a small number of doses administered in laboratory settings, which may lack ecological validity relative to community practice. Biological plausibility for beneficial effects has been suggested by objective measures (for example, changes in functional connectivity and circulating BDNF), and theory about experience-dependent plasticity implies that drug-induced neurobiological changes might need real-world stimulation to translate into measurable behavioural effects. To address the ecological gap, Murphy and colleagues conducted a randomised, double-blind, placebo-controlled trial that used repeated self-administered microdoses of LSD over six weeks. The study aimed to test acute (dose-day) and durational (pre‑to‑post intervention) effects on mood, expectancy, trait measures and cognition in healthy male volunteers, while also assessing safety, adverse events (AEs), blinding, and the influence of expectancy in a home-administration setting more representative of community microdosing.

This was a double-blind randomised controlled trial conducted between April 2021 and April 2022. Eighty healthy male volunteers were randomised 1:1 to receive either 10 μg LSD base (n = 40) or an inactive placebo (water; n = 40). Doses were administered sublingually by 1 ml oral syringe every three days for six weeks (14 doses total). The first dose was taken under clinic supervision; the remaining 13 doses were self-administered at home. Participants were naïve to microdosing (prior psychedelic use permitted if >1 year prior). Dose compliance was monitored via participant-recorded videos submitted on dose days. Inclusion/exclusion criteria and detailed procedures were prospectively published; some protocol amendments (including a dose‑titration procedure) were introduced and are reported in supplements. Daily measures were collected by nightly SMS-linked REDCap questionnaires from baseline through the final visit. Daily assessments included open-field AEs (coded with MedDRA), a categorical blinding guess and confidence VAS, and 16 visual analogue scales (VAS) rating momentary changes across domains (centred on 'Usual' = 0, range -50 to +50). Pre- and post-intervention measures were administered at Baseline (~1 week before first dose) and at a Final visit (~2 days after last dose) and included expectancy/experience ratings, personality (BFI-2), absorption (MODTAS), cognitive flexibility (DFlex), mindfulness (FFMQ), emotion scales (NIH Toolbox and DASS), and cognitive tasks from the NIH Toolbox. Statistical analysis used both frequentist linear mixed-effects models (lmerTest) and Bayesian MCMC models (brms/Stan) to assess Group x Day interactions for daily measures and Group x Visit interactions for pre/post measures. Multiple comparisons were handled with Bonferroni-adjusted alpha thresholds for each domain in frequentist tests and with shrinkage priors (horseshoe) in the Bayesian models; exploratory follow-up Bayesian analyses used normal priors. The primary analytic set was intention-to-treat (ITT) including all randomised participants, with a per-protocol (PP) set excluding participants with early discontinuation, delayed final assessments (>2 days after last dose), missing data, or not having received exactly 14 doses. Blinding was summarised using the Bang Blinding Index (BBI). Safety monitoring included blood pressure and heart rate measures.

Recruitment screened 136 men and randomised 80 participants (40 LSD, 40 placebo). Demographics were majority New Zealand European; only male participants were recruited. Treatment and assessment adherence was high: one missed visit out of 240, no missed home doses out of 1,022, four unverified doses due to video failure, and three incomplete daily questionnaires out of 3,939. Blood pressure and heart rate analyses raised no safety concerns. Adverse events were common but mostly mild. The proportion of participants reporting at least one AE was 85.0% in the LSD group versus 80.0% in placebo (OR = 1.4, 95% CI [0.4, 4.8], Fisher's exact p = 0.77). No severe or serious AEs were reported. The only AE with a statistically significant between-group difference was 'feeling jittery' (LSD 32.5% vs placebo 7.5%; OR = 5.62, 95% CI [1.6, 27.7], p = 0.01). Following early reports of overstimulation, a titration procedure was added: seven participants (six LSD, one placebo) had doses titrated and three of those (all LSD) subsequently discontinued; a fourth LSD participant discontinued without titration. In the discontinuation cases, feelings of overstimulation and anxiety were reported and resolved within two weeks with supportive care. Blinding analyses indicated effective blinding in the placebo group (BBI = -0.05, 95% CI [-0.11, 0.014]) but partial unblinding in the LSD group (BBI = 0.419, 95% CI [0.35, 0.48]). Conditional probability of a correct categorical guess was 0.61 in the LSD group and 0.36 in the placebo group; confidence ratings were positive toward LSD in the LSD group and negative in placebo. Daily VAS analyses of 16 scales using mixed models identified seven scales with significant Group x Day interaction effects after correction: 'angry', 'connected', 'creative', 'energy', 'happy', 'irritable', and 'well'. Bayesian modelling largely corroborated these findings: 95% credible intervals excluding zero for the same measures except 'angry', and very high posterior probabilities (>99.9%) that effects existed for 'energy', 'happy', 'irritable', and 'well', with >99% probability for 'connected' and 'creative'. Inspection of means showed that increases in 'energy', 'connected', 'creative', 'happy' and 'well' were driven by dosing days; decreases in 'angry' and 'irritable' on dose days suggested transient reductions with return to baseline on subsequent non-dose days. Including pre-intervention expectancy as a covariate did not eliminate these interaction effects, and expectancy did not predict subsequent daily scores. An analysis restricted to doses where participants reported 'Don't know' for allocation still showed a highly probable effect for 'energy' (>99.9%) and a probable effect for 'well' (>99%), suggesting some dose-day effects remained even when participants were uncertain of allocation. Effect sizes were generally smallest for the first dose given in the laboratory compared with subsequent home doses, notably for 'creative' and 'energy'. Pre/post expectancy/experience ratings showed significant Group x Visit interactions in the ITT analysis for retrospective beliefs about 'energy', 'happy' and 'connected' (Bonferroni-corrected), with only 'energy' (and 'well' in the PP analysis) remaining significant in PP. Bayesian analysis supported a highly probable true change in retrospective 'energy' beliefs and probable changes for 'happy' and 'connected'. In these cases participants in the LSD arm retrospectively rated greater belief that microdosing had caused positive changes than did placebo participants, exceeding their pre-intervention expectancy. For pre/post trait, emotion and cognition measures, no Group x Visit effects survived Bonferroni correction in either ITT or PP analyses. Bayesian analyses with conservative shrinkage priors were null; a highly exploratory Bayesian analysis without strong shrinkage suggested possible effects (for example, DASS Anxiety, NIH Meaning and Purpose, NIH Pattern Comparison Speed) but the authors considered these insufficiently robust given multiple comparisons. The study did not support the pre-registered primary hypotheses: MODTAS (p = 0.92) and BFI-2 Openness (p = 0.55) showed no credible change in the LSD group.

Murphy and colleagues interpret their findings as indicating that microdosing 10 μg LSD in healthy adult men is relatively safe overall but can produce transient mood-elevating and prosocial effects on dosing days, specifically increases in reported creativity, connectedness, energy, happiness and overall wellness alongside reductions in irritability and anger on dose days. These dosing-day-limited changes did not translate into persisting improvements on psychometric trait, mood or cognitive measures over the six-week intervention period in this psychologically healthy sample. The investigators highlight that a subset of participants experienced feelings of overstimulation that in some instances led to anxiety and study discontinuation; dose titration mitigated some of these cases. They therefore recommend considering lower initial doses, flexible scheduling and responsive titration in future trials, especially while predictive biomarkers of adverse response remain unknown. The authors also emphasise the ecological validity finding that first doses given in a laboratory setting produced smaller acute effects than subsequent home doses, suggesting that laboratory-only dosing paradigms may substantially underestimate effects experienced in real-world contexts. Blinding was incomplete in the LSD group, yet the authors report that acute effects on energy and wellness persisted when analyses were restricted to dose days where participants reported uncertainty about allocation and when controlling for pre-intervention expectancy. From this they conclude that expectancy and unblinding alone do not fully explain the observed acute dose-day effects, while acknowledging the need for active placebos and stratified analytic approaches in future patient trials to better isolate pharmacological effects. Null durational results are addressed cautiously: the lack of longitudinal changes may reflect ceiling or floor effects in a healthy male sample, the limited generalisability given the male-only and predominantly European ancestry sample, and disruptions caused by the Covid-19 pandemic which delayed some final assessments and support interpretation of the per‑protocol analyses. Overall, the authors suggest that acute mood-elevating and pro-social dose-day effects could plausibly be beneficial in clinical populations characterised by anhedonia, but state that this remains untested and that claims of enduring therapeutic benefit are not supported by the present data.

This double-blind RCT of repeated home-administered 10 μg LSD microdoses in healthy adult males found credible acute mood and prosocial effects on dosing days but no evidence of sustained changes in personality, mood or cognitive function after six weeks. Adverse effects were generally mild, yet a minority experienced overstimulation and anxiety; dose titration alleviated some cases. The investigators conclude that microdosing appears relatively safe in this population but caution against untested claims of durable therapeutic benefit, and they recommend future trials use active placebos, consider dose titration, and evaluate effects in clinical populations and more diverse samples.