Psychedelic Harm Reduction: What the Evidence Actually Shows

The short answer

The Chacruna Institute posted an announcement on September 16, 2026 for a free virtual forum held September 23: "Building Tripsafe Communities: Why We Need Psychedelic Harm Reduction," with Tomislav Majić (Charité Berlin), Maha N. Mian (Suffolk University/UCSF), and Helena Aicher (Zurich/Basel). You can read it here: https://chacruna.net/building-tripsafe-communities-why-we-need-psychedelic-harm-reduction/

The page contains no statistics and no study citations. It's framing — "hidden risks... that often go unaddressed in the enthusiasm surrounding their therapeutic promise." We did not attend, there is no public recording we could find, and we are treating the event strictly as a hook.

The substance is elsewhere. Majić edited *Psychedelic Harm Reduction*, volume 77 of Current Topics in Behavioral Neurosciences (Springer, 2026) — 26 chapters including "Classifying Psychedelic-Related Complications," "Psychedelic-Related Psychosis," "HPPD," and "Psychedelic Drug Checking" (https://link.springer.com/book/10.1007/978-3-032-21107-1). He is also a co-author on the phase 2b depression trial that missed its primary endpoint this year.

So: what does the published record actually say about harm, and does harm reduction fix it?

Oregon's first real cohort: 346 people, four serious reactions, one reported

Evidence tier: prospective observational cohort, single-arm, no control group. This is the strongest real-world safety data that exists, and it still cannot establish cause.

The OPEN network enrolled 346 adults across 24 of Oregon's 26 licensed service centers with 83 facilitators, November 2024 to March 2026, with 93.1% follow-up at one week and 90.2% at three months. DOI 10.1001/jamanetworkopen.2026.30608; full text at https://pmc.ncbi.nlm.nih.gov/articles/PMC13491119/

The safety numbers:

• Four participants (1.2%) had adverse behavioral reactions requiring medical attention — one hospitalization, three ED visits, doses 25–50 mg.
• All four were psychedelic-naive, and all four were seeking mental health treatment. That is the single most actionable finding in the paper.
• New thoughts of dying or suicide: 3 (1.0%) at one week, 6 (1.9%) at three months.
• Participants who called the experience harmful: 1.6% at one week, rising to 2.3% at three months.

And the line that should reframe every "Oregon has had almost no adverse events" post you have read: facilitators reported only one of the four serious reactions through Oregon's mandated channel. The authors say plainly that state adverse-event data "may underestimate true adverse reaction rates."

That matters because the aggregate state numbers are what get quoted. Yu, Tafur, Moreno and Dahmer analyzed Oregon Health Authority dashboard archives for calendar 2025 in *Frontiers in Psychiatry* (May 13, 2026): 5,935 clients, 5,375 sessions, behavioral adverse events at 2.42 per 1,000 sessions and medical at 2.79 per 1,000, with 7 hospital transports across the year (https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2026.1777387/full). Note the definition: Oregon counts an "adverse reaction" only if it requires emergency services or provider contact *during a session*. Anything that surfaces two weeks later is invisible to that system.

Cheung, Propes and Yaden at Johns Hopkins made the same point in *International Journal of Drug Policy* (Sept 2026, DOI 10.1016/j.drugpo.2026.105384): reactions were "relatively uncommon: however, Oregon's reporting framework sets a high threshold for reportable events," and broad eligibility combined with non-clinical supervision "may raise several safety concerns, particularly when clients with clinical symptoms seek services."

One more caveat the authors flag themselves: the cohort was about 5% of Oregon clients in that window, 86.7% White, 76.6% college-educated, and 29.0% had household income over $200,000. Baseline symptoms were relatively low and psychiatric diagnoses were unknown. This is not a representative sample of everyone who will ever take psilocybin.

The 10% number — and why it is not "10% got HPPD"

This is the finding that got the least coverage and deserves the most care.

Ten percent is an order of magnitude above the "rare" framing that circulates in psychedelic media. But read what was measured: self-reported symptoms at one and three months, not DSM-5 HPPD diagnoses. One participant out of 346 reported significant distress from them.

That gap is exactly the distinction Žuljević (Split) and Majić (Charité) draw in their chapter on flashbacks, HPPD and reactivations in *Psychedelic Harm Reduction* (https://link.springer.com/chapter/10.1007/7854_2025_610). Their clinical read, based on case series and clinical experience: "many patients presenting with post-psychedelic complications assume they have HPPD; however, only a minority actually meet diagnostic criteria." Most cases resolve spontaneously within a year. Chronic cases are described as very rare.

The part worth knowing before you need it: there is no evidence-based treatment for HPPD. What clinicians use — clonidine, benzodiazepines, antipsychotics — comes from case reports, not trials.

So both of the popular framings are wrong. "HPPD is vanishingly rare, don't worry about it" understates how often people notice lingering visual changes. "One in ten people get HPPD" overstates a clinical disorder from a symptom checkbox. The accurate version is duller: transient visual after-effects are reasonably common, distressing persistent ones are not, and nobody knows how to treat the persistent ones well.

The efficacy picture is messier than the press releases

Tier: human RCT. Mertens, Koslowski, Betzler, Evens, Majić and Gründer published EPISODE in *JAMA Psychiatry* this year (DOI 10.1001/jamapsychiatry.2026.0132): 144 adults with treatment-resistant depression, German outpatient sites, triple-blind, active placebo of 100 mg nicotinamide. The six-week HAM-D response rate was 17.0% for 25 mg psilocybin versus 10.6% for placebo — odds ratio 1.73, P = .19. The primary endpoint failed. Secondary measures gave what the authors call "exploratory evidence of a clinically meaningful effect." Safety: more dosing-day suicidal ideation on 25 mg (4% vs 1–2%) and one case of HPPD. Majić, a speaker at the Chacruna forum, is a co-author. We did not read the full text or its funding statement, so we are not characterizing the trial's independence either way.

Tier: human RCT, feasibility. Rucker et al. in *Nature Medicine* (Aug 6, 2026) ran psilocybin-assisted therapy for TRD inside the NHS: 60 randomized, 59 completed, MADRS adjusted mean difference −10.41 at week 3 and −12.92 at week 6, Cohen's d −1.70 (https://www.nature.com/articles/s41591-026-04541-0). That is a big effect. It also came with blinding failure: 100% of psilocybin participants and 70% of placebo participants correctly guessed their allocation. The trial excluded bipolar disorder, psychotic disorder, substance dependence, personality disorder, dementia and any suicide attempt in the past year — and the authors note that excluding high-risk people attenuates the safety signals you can observe. Four serious adverse events total, three in the open-label extension, and the only one judged possibly related was in the placebo arm.

Tier: systematic review. Marinis, Clarke, Guerin, Guastella and Bedi assessed side-effect reporting across 24 psilocybin trials and 917 participants in *BJPsych Open* (2025, DOI 10.1192/bjo.2025.10847). Against the CONSORT Harms standard, only six trials (25%) rated high quality; median adherence was 50%. All nine RCTs were rated high risk of bias for side-effects as an outcome, largely because of functional unblinding. They flag interpersonal harms and existential crises as barely measured at all.

Tier: company press release, not peer-reviewed. Compass Pathways reported COMP005 (258 participants, 32 US sites, single 25 mg dose, MADRS difference −3.6, p<0.001) in June 2025, and COMP006 (581 participants, 25 mg vs 10 mg vs 1 mg, 25 mg vs 1 mg difference −3.8, p<0.001) in February 2026. In COMP006, 73% of treatment-emergent adverse events occurred on dosing days and 83% resolved within 24 hours; six participants (2%) had treatment-emergent serious adverse events; suicidal ideation was under 1% across both trials, and the one serious adverse event of suicidal behavior occurred in the 1 mg arm. Compass says it expects to complete an NDA in Q4 (https://ir.compasspathways.com/News--Events-/news/news-details/2026/Compass-Pathways-Successfully-Achieves-Primary-Endpoint-in-Second-Phase-3-Trial-Evaluating-COMP360-Psilocybin-for-Treatment-Resistant-Depression/default.aspx).

No psilocybin product is FDA-approved for anything. An NDA has not been submitted.

Does harm reduction work? The honest answer is "unproven"

Tier: survey, self-report, post-hoc. Pleet, White, Zamaria and Yehuda published a study of 884 callers to the Fireside Project peer-support telephone line in *Psychedelic Medicine* (1(2):69–73, 2023). Callers reported that the line de-escalated them from psychological distress (65.9%), that they may have been harmed without it (29.3%), that they may otherwise have called 911 (12.5%), and that they may have gone to an ED (10.8%).

Those are the numbers people cite. Here is the part that usually gets dropped: the second author, Joshua White, is Fireside Project's founder. The service operator co-authored the evaluation of its own service, and the outcome measure is callers speculating about a counterfactual that never happened.

That is not an accusation of bad faith. It is a description of the evidence tier. Nobody has run a controlled trial of Zendo-style in-person peer support. We looked. The defensible framing is that harm reduction is plausible, cheap, low-risk and widely endorsed — and unproven by controlled evidence. Anyone calling it "evidence-based" is upgrading a caller survey.

Where the evidence for harm is much harder is the unregulated product side. FDA's investigation into Diamond Shruumz-brand chocolate bars, cones and gummies closed with 180 illnesses, 73 hospitalizations, 3 potentially associated deaths across 34 states (final counts as of Nov 4, 2024). FDA lab analysis found muscimol, 4-AcO-DMT, psilocin, kavalactones, ibotenic acid — and pregabalin, a prescription drug, in a product sold for microdosing (https://www.fda.gov/food/outbreaks-foodborne-illness/investigation-illnesses-diamond-shruumz-brand-chocolate-bars-cones-gummies-june-2024). CDC's *MMWR* (74(1), Jan 9, 2025) described four Arizona patients, two of them adolescents, hospitalized with seizures, loss of consciousness, respiratory depression and cardiac abnormalities; two required intubation (https://www.cdc.gov/mmwr/volumes/74/wr/mm7401a3.htm). All recovered within 24 hours. CDC's recommendation was to stop consuming these products and to use caution with anything claiming mushroom-based psychoactives.

That is what a *documented* harm signal looks like — and it is the one place where a harm-reduction intervention with a real mechanism, drug checking, has an obvious target. Majić's volume includes a chapter on it. Background on that particular brand: /brands/diamond-shruumz

Screening, interactions, and what is actually legal

Interactions, tier: systematic review of case reports and small human studies. Halman, Kong, Sarris and Perkins reviewed drug–drug interactions with classic psychedelics in the *Journal of Psychopharmacology* (38(1):3–18, online Nov 20, 2023) — 52 studies from 36 reports (https://pmc.ncbi.nlm.nih.gov/articles/PMC10851641/).

• Lithium: three case reports, ages 21–29, describing earlier onset and increased hallucinatory and psychological LSD effects. You will see a much scarier lithium claim circulating — dozens of case reports, half involving seizures. That figure is not from this review, and we could not read the paper it traces to. We are not printing numbers we have not verified. What we can say: Oregon's regulator treats lithium as a hard stop. See /guides/psilocybin-and-lithium
• SSRIs: fluoxetine markedly diminished LSD effects in 18 patients with delayed onset in 8; sertraline decreased effects in 11; escitalopram with psilocybin in 23 participants reduced anxiety and ego dissolution without blocking positive mood. See /guides/psilocybin-and-ssris
• MAOIs: phenelzine nearly abolished the subjective LSD response in two cases, nialamide blocked reactions entirely in 14, isocarboxazid attenuated them in four. See /guides/psilocybin-and-maois

Broader interaction map: /guides/psilocybin-drug-interactions

Law. Oregon's Measure 109 passed in November 2020 and is codified at ORS 475A; OHA licenses manufacture, transport, delivery, sale and service provision, applications opened January 2, 2023, and centers opened in summer 2023 (https://www.oregon.gov/oha/ph/preventionwellness/pages/oregon-psilocybin-services.aspx). Eligibility is adults 21+ with no medical diagnosis, prescription or referral required — a preparation session with a licensed facilitator is mandatory before administration. The program excludes people with active suicidal ideation or psychosis, or who have taken lithium in the past 30 days. Details: /legal-status/oregon

Colorado's 2022 ballot measure decriminalized personal cultivation, possession, consumption and sharing, and created state-licensed healing centers for adults 21+ under the Department of Revenue's Natural Medicine Division. Colorado Public Radio reported on February 4, 2026 that 34 healing centers were licensed with more than a dozen applications pending, with pricing around $3,500 for preparation-administration-integration at two centers and pre-screening covering medications, health concerns and family history. Colorado has published no outcome study comparable to Oregon's. /legal-status/colorado and /vs/oregon-vs-colorado-psilocybin-therapy

New Mexico legislators approved psilocybin access within the health care system in 2025 under the Medical Psilocybin Act, SB 219, with a statutory deadline of December 31, 2027 (https://www.nmlegis.gov/Sessions/25%20Regular/bills/senate/SB0219JUS.pdf). /legal-status/new-mexico

As of June 2026, 28 other states were considering policies to expand access. And on July 14, 2026 FDA finalized "Psychedelic Drugs: Considerations for Clinical Investigations," finalizing its June 2023 draft and covering CMC, nonclinical work, clinical pharmacology, abuse potential, trial design and safety monitoring (https://www.fda.gov/regulatory-information/search-fda-guidance-documents/psychedelic-drugs-considerations-clinical-investigations).

PsyBear takeaways

1. Every serious reaction in Oregon's cohort happened to a psychedelic-naive person. Four out of 346, all first-timers, all seeking mental health treatment. If you are new to this, you are the population the safety data is describing. Start at /guides/safe-trip and /guides/trip-preparation.

2. The official numbers undercount and the authors say so. Facilitators reported one of four serious reactions through Oregon's mandated channel. Oregon only counts events that happen during a session. Treat any "adverse event rate" quoted from a state dashboard — or from a commercial provider directory — as a floor, not a measurement.

3. Ignore the unsourced safety stats floating around. "Zero hospitalizations," "0.15% adverse event rate" — we could not source those, and they contradict the peer-reviewed record, which documents one hospitalization among 346 cohort participants and seven hospital transports in Oregon's 2025 data.

4. Lingering visual symptoms are more common than you have been told; HPPD is not. 10.0% reported symptoms at one month; one participant of 346 reported significant distress. And there is no evidence-based treatment if it persists.

5. The efficacy case is real but unsettled. A 144-person phase 2b missed its primary endpoint. An NHS RCT with a large effect had total blinding failure. All nine RCTs in a 2025 systematic review were rated high risk of bias for side-effects as an outcome. Two phase 3 wins exist as company press releases. That is a field mid-argument, not a closed case.

6. Harm reduction is worth doing and is not yet proven. A caller survey co-authored by the service's founder is the best peer-reviewed evidence we found. Do it anyway — a sitter, a screen, a plan, an integration conversation cost almost nothing. Just don't call it evidence-based. /guides/trip-sitter and /guides/psychedelic-integration-workbook

7. Screening is the intervention with the clearest rationale. Lithium in the past 30 days, active psychosis, active suicidality: Oregon's own exclusions. Serotonergic and MAOI interactions are documented. If you are on medication, that conversation belongs with a clinician before anything else. /guides/psilocybin-drug-interactions

Key Takeaways

The best real-world safety data we have — Korthuis et al., JAMA Network Open, Aug 19, 2026, 346 adults in Oregon's licensed program — found that 1.2% had adverse behavioral reactions needing medical attention, and all four of those people were psychedelic-naive. Facilitators reported only one of the four through Oregon's mandated channel, so the state's own dashboard undercounts. Self-reported HPPD-type symptoms showed up in 10.0% at one month, which is far above the "rare" framing, though only one participant of 346 reported significant distress. Meanwhile the largest academic RCT of the year (EPISODE, JAMA Psychiatry) missed its primary endpoint, and the peer-reviewed evidence that peer-support harm reduction works is one post-hoc caller survey co-authored by the service's own founder. Harm reduction is cheap, plausible, and untested by controlled trials. Screening is the part that actually has a signal behind it.

FAQ

How risky is psilocybin in Oregon's regulated program?

In the strongest real-world data available — Korthuis et al., JAMA Network Open, Aug 19, 2026, a single-arm cohort of 346 adults across 24 licensed Oregon service centers — 4 participants (1.2%) had adverse behavioral reactions requiring medical attention, resulting in 1 hospitalization and 3 ED visits. All four were psychedelic-naive and all four were seeking mental health treatment. The design was uncontrolled, which the authors say precludes causal inference.

Do 10% of psilocybin users get HPPD?

No. The Oregon cohort found HPPD-type symptoms self-reported by 10.0% of participants at one month and 8.7% at three months, but those were self-reported symptoms rather than DSM-5 diagnoses, and only 1 participant of 346 reported significant distress. Žuljević and Majić note in Current Topics in Behavioral Neurosciences vol. 77 that only a minority of people who assume they have HPPD meet diagnostic criteria, that most cases resolve spontaneously within a year, and that no evidence-based treatment exists.

Does Oregon's state adverse-event data tell the full story?

No, and the researchers say so directly. Facilitators reported only 1 of the 4 serious reactions in the JAMA Network Open cohort through Oregon's mandated channel, so state data "may underestimate true adverse reaction rates." Oregon also defines a reportable adverse reaction narrowly — one requiring emergency services or provider contact during a session — so anything surfacing later is not captured.

Is there proof that psychedelic harm-reduction services reduce harm?

Not from controlled trials. The one peer-reviewed evaluation we found is a survey of 884 callers to the Fireside Project helpline (Psychedelic Medicine, 2023), in which 65.9% said the line de-escalated them from distress and 29.3% said they may have been harmed without it. The service's founder is a co-author, and the measure is callers speculating about a counterfactual. Harm reduction is plausible and cheap; it is not trial-proven.

Did the big psilocybin depression trials all succeed?

No. The EPISODE phase 2b trial (JAMA Psychiatry, 2026), 144 adults with treatment-resistant depression, missed its primary endpoint: 17.0% six-week HAM-D response on 25 mg versus 10.6% on active placebo, OR 1.73, P = .19. An NHS feasibility RCT (Nature Medicine, Aug 2026) showed a large effect but had complete blinding failure, with 100% of psilocybin participants correctly guessing their allocation. Compass Pathways' two phase 3 results are company press releases, not peer-reviewed publications.

Which medications are a problem with psilocybin?

Halman et al. (Journal of Psychopharmacology, 2023) reviewed 52 studies: three lithium case reports described earlier onset and intensified psychological effects; SSRIs including fluoxetine, sertraline and escitalopram diminished effects; MAOIs such as phenelzine, nialamide and isocarboxazid attenuated or entirely blocked responses. Oregon's program excludes anyone who has taken lithium in the past 30 days, or who has active suicidal ideation or psychosis.

Related resources

• Safe Trip Guide
• Psilocybin Drug Interactions
• Psilocybin and Lithium
• Oregon Legal Status
• Oregon vs Colorado Psilocybin Therapy
• Magic Mushroom Risks