Microdosing with psychedelics to self-medicate for ADHD symptoms in adults: A prospective naturalistic study

Haijen and colleagues frame adult attention deficit hyperactivity disorder (ADHD) as a prevalent condition with persistent functional impairments and high comorbidity, and note that many adults either discontinue first-line pharmacological treatments or do not respond adequately to them. The introduction summarises how microdosing (MD) with classical psychedelics (small, repeated doses of LSD, psilocybin, etc.) is reported anecdotally to improve concentration, productivity and mood, and that limited experimental work in neurotypical volunteers has shown small-dose effects on attention and time perception. The authors argue that prospective, behaviourally assessed data are needed in adults with ADHD because prior work has relied largely on retrospective self-reports. The study therefore set out to prospectively assess changes in self-reported ADHD symptoms, well-being, and temporal processing in adults who intended to begin MD on their own initiative. Using a naturalistic design, the investigators hypothesised that ADHD symptoms would decrease and well-being would increase after four weeks of MD compared to baseline, that improvements in ADHD symptoms would correlate with changes in well-being, and that time perception performance would improve. Secondary aims examined whether concurrent use of conventional ADHD medication or presence of psychiatric comorbidity modified outcomes and explored characteristics of participants who did not improve after four weeks.

This was an online, prospective naturalistic survey of adults who intended to start microdosing to self-manage ADHD symptoms. Recruitment occurred via an online advertisement placed on a microdosing information website. Interested individuals provided written informed consent one to three days before initiating MD and completed a baseline survey; follow-up surveys were administered at two and four weeks after enrolment. Participants were asked to keep a diary recording the substances and doses they used during the study period. Surveys could be paused and completed later. Data collection occurred during the COVID-19 pandemic period noted in the paper. Inclusion for analysis required scoring above a clinical cut-off on at least one subscale of the Conners' Adult ADHD Rating Scale—Short Version (CAARS-S:SV), with a t-score ≥ 65 used to indicate clinically elevated symptoms. The sample therefore included adults with a formal ADHD diagnosis and individuals reporting severe ADHD complaints. Baseline measures collected demographic information, psychiatric/neurological/physiological diagnoses (allowing multiple selections), lifetime psychedelic experience, and current or prior use of conventional ADHD medications. Primary and secondary outcomes were assessed at baseline, 2 weeks and 4 weeks. ADHD symptoms were measured with the CAARS-S:SV (subscales: inattention; hyperactivity/impulsivity; ADHD index; plus a DSM-IV total symptom score). Well-being was measured with the WHO-5 Well-Being Index. Temporal processing was assessed with an auditory time reproduction task (TRT) presenting six intervals (1000 ms, 1500 ms, 3200 ms, 3700 ms, 5500 ms, 6000 ms); performance was expressed as a relative difference score (estimated minus actual interval, divided by actual interval). Instructions discouraged explicit counting and participants performed two practice and twelve test trials (each interval twice). Statistical analysis used linear mixed models (LMMs) with Time (0W, 2W, 4W) as a within-subject factor; Interval was an additional within-subject factor for the TRT analysis. Akaike's information criterion determined the best-fitting covariance structure. Missing data were handled via restricted maximum-likelihood estimation. Significant main effects were followed by Bonferroni-corrected pairwise comparisons. An alpha of 0.05 was used. The investigators also calculated frequencies of non-responders (no improvement or worsening in CAARS DSM-IV total t-score) and Pearson correlations between change scores of ADHD symptoms and WHO-5 well-being.

Of 356 individuals who consented and began the survey, 247 completed the baseline survey; after excluding two responses judged non-serious and 12 participants without elevated CAARS t-scores, 233 respondents were analysed. Most participants had prior psychedelic experience (80%). ADHD was reported by 159 participants (68% of the analysed sample), and 54% of those with ADHD reported at least one comorbid diagnosis; depression, anxiety, PTSD and dyslexia were common comorbidities. More than half of the diagnosed participants had previously stopped conventional ADHD medication; current conventional medication use during the study varied. About half of respondents completed the diary item reporting the substance used: among those, 78% reported psilocybin/psilocin, 12% novel lysergamides, 9.5% LSD, and one respondent reported ayahuasca. Self-reported mean doses were reported for these substances but diary completion was incomplete. On the primary outcome, CAARS-S:SV DSM-IV total symptoms t-scores decreased over time (main effect of Time, p = .000, ηp2 reported). Pairwise comparisons showed significant reductions from baseline to 2 weeks (Δ2W–0W = −10.17, p = .000) and to 4 weeks (Δ4W–0W = −15.43, p = .000), and a further reduction between 2 and 4 weeks (Δ4W–2W = −5.26, p = .007). A Time × Medication use interaction was observed (F(3,241.4) = 2.86, p = .038, ηp2 = .034): individuals using conventional ADHD medication alongside MD had higher CAARS scores at 2 weeks (β = 8.7, p = .006), indicating a smaller early decrease, but no difference was found at baseline or 4 weeks. The Time × Comorbidity interaction for the DSM-IV total score was not significant. Analyses of CAARS subscales paralleled the total score findings. Time produced main effects on ADHD index, inattention, and hyperactivity/impulsivity t-scores. Significant Time × Medication use interactions were present for the ADHD index (F(3,247.6) = 3.65, p = .013, ηp2 = .042) and Inattention (F(3,250.6) = 3.45, p = .017, ηp2 = .040), with medication users showing smaller decreases at 2 weeks (ADHD index β = 8.04, p = .001; Inattention β = 10.92, p = .002). A Time × Comorbidity interaction emerged for the ADHD index (F(3,232.7) = 2.73, p = .045, ηp2 = .034) that reflected higher baseline ADHD index scores in those with comorbid diagnoses (β = 3.30, p = .011); no comorbidity effects were found at follow-ups. No Time interactions were observed for the Hyperactivity/Impulsivity subscale. At the four‑week time point, 9 of 47 respondents (19.1%) were classified as non-responders (no improvement or worsening in CAARS DSM-IV total t-score); three of these nine were using conventional medication alongside MD. Some non-responders had transient improvement at 2 weeks and worsened by 4 weeks. Well-being (WHO-5) improved significantly over time (F(2,119.4) = 31.50, p = .000, ηp2 = .345). Well-being scores increased from baseline to 2 weeks (Δ2W–0W = 16.47, p = .000) and to 4 weeks (Δ4W–0W = 17.80, p = .000), with no significant change between 2 and 4 weeks. A Time × Comorbidity interaction indicated lower baseline well-being for participants with comorbid diagnoses (β = −6.61, p = .009), but comorbidity did not alter follow‑up scores. No Time × Medication interaction on WHO-5 was detected. Changes in ADHD symptom scores correlated moderately and negatively with changes in well-being (2W: r = −0.367, p = .003; 4W: r = −0.471, p = .001), indicating that larger reductions in ADHD symptoms were associated with greater increases in well-being. For time perception, there was no main effect of Time on TRT relative difference scores (F(2,59.2) = 0.63, p = .534). Interval length produced the expected effects on reproduction accuracy (longer intervals more likely underestimated). A three-way Time × Interval × Medication use interaction was significant (F(18,97.78) = 2.52, p = .002, ηp2 = .317); follow-up estimates showed that, for the shortest interval (1000 ms), individuals using conventional ADHD medication alongside MD had higher relative difference scores (over-reproduction) at 2 weeks (β = .186, p = .038) and 4 weeks (β = .353, p = .004) compared with those not using conventional medication. No medication effects were found for other intervals. The authors note small sample sizes for medication subgroups at follow-ups (n = 16 at 2W; n = 9 at 4W) and substantial variability in TRT performance, which limit confidence in these specific findings. Attrition was substantial: only 47 respondents completed the four‑week assessment (about 20% of the analysed baseline sample), and diary completion about substances was incomplete, leaving many substance- and dose-related data missing.

Haijen and colleagues interpret the findings as preliminary evidence that self-initiated microdosing is associated with reductions in self-reported ADHD symptoms and increases in well-being over a four‑week period in adults with diagnosed or clinically elevated ADHD symptoms. They emphasise that symptom reductions were observable at two weeks and continued to improve by four weeks, and that symptom change was moderately correlated with enhanced well-being. The authors report that concurrent use of conventional ADHD medication appeared to delay early symptom improvement (at 2 weeks) but did not alter outcomes at 4 weeks; comorbidity did not change MD effects on main outcomes at 4 weeks. Regarding temporal processing, the investigators did not find overall improvement after MD; instead, an over-reproduction of the shortest interval was observed in participants using conventional medication alongside MD, a finding they treat cautiously because of small subgroup sizes and high variability. The discussion situates these results alongside prior retrospective reports and limited experimental findings in healthy volunteers, noting concordance with anecdotal and survey-based claims that MD can aid concentration and mood. The authors consider their prospective, naturalistic design a strength because it reduces some uncertainty inherent in retrospective self-reports and allowed repeated behavioural and self-report assessments using validated instruments. Inclusion of a cognitive task alongside clinical questionnaires is highlighted as an additional asset. Key limitations acknowledged by the investigators include the lack of a placebo-control group and the consequent inability to rule out expectancy/placebo effects, recruitment via a MD website that likely biased the sample towards individuals favourably disposed to MD, self-selection and high dropout (only about 20% of baseline participants completed the four‑week measures), incomplete diary data about the substances and doses used, and limited information about formulation, storage and route of administration. The CAARS-S:SV is based on DSM‑IV criteria and uses older normative data, which the authors note as a limitation. Concerning the time reproduction task, they point out that two trials per interval may have been insufficient to detect subtle MD effects. Finally, small numbers of participants using conventional medication at follow-up limit interpretation of interactions involving medication. For future work, the authors recommend placebo-controlled and blinded studies (including approaches that allow participant blinding in naturalistic settings), standardised dosing and administration protocols, more extensive cognitive testing (working memory, attention, executive function), and exploration of MD effects in other disorders that include attentional symptoms. Despite the caveats, they conclude that this study provides initial prospective evidence suggesting MD may have therapeutic value for adults with ADHD or severe ADHD complaints, and that controlled trials are needed to confirm the magnitude and specificity of the effects.