Paul Austin posted the number that should have stopped every clinical trial strategist mid-scroll: more than 20,000 people have taken legal psilocybin in Oregon since 2023, and until last week, no one in the regulatory ecosystem could tell you what happened to them. That silence was not a gap in the science. It was a structural failure in how we recognize evidence when it arrives in the wrong container.
Now OHSU’s Todd Korthuis and his team have cracked that container open. The study, published in JAMA Network Open and covering 2,363 participants across Oregon and Colorado’s regulated psilocybin services, represents the largest real-world longitudinal dataset on psilocybin ever assembled. Per Austin’s post, that sample is six times larger than every clinical trial of a classic psychedelic since 1951, combined. Depression scores dropped 49% at two weeks. Anxiety dropped 51%. Wellbeing rose 22%. In the OHSU-followed cohort of 346 clients, moderate-to-severe depression prevalence fell from 42% to 17% at three months. PTSD from 48% to 17%. Life satisfaction climbed from 63% to 82%.
The clinical trial industry needs to sit with that for a moment before reaching for the methodological red pen.
What the RCT Cannot See
The instinct to immediately discount naturalistic evidence is understandable. The OHSU study has no control group. The cohort opted into research at roughly a 5% participation rate among all Oregon program users, per Austin’s post. The demographic skew is severe: 87% white, 29% earning over $200,000 annually, with session costs ranging from $900 to $5,000. OHSU President Shereef Elnahal noted directly that this “does limit the generalizability of the findings.” These are real limitations and they matter.
But the counterintuitive reading of this dataset is more important than its caveats. The Oregon program’s eligibility criteria consist of three exclusions: lithium, active psychosis, and active suicidal ideation. No diagnosis required. No prescription. Facilitators need state training, not clinical licensure. Participants came primarily out of curiosity (74%) or self-improvement goals (71%), not because they were treatment-seeking patients pre-screened and optimized for RCT inclusion. This is, in other words, approximately what population-level psychedelic access looks like, which is precisely what no Phase 2 or Phase 3 trial can simulate.
The signal that emerged from that openness should rewrite several assumptions held by trial designers and pharma strategists alike. Lucas Pawlik’s analysis of the Thompson et al. dataset (DOI: 10.64898/2026.08.11.26360133) identifies two findings that directly challenge prevailing clinical assumptions. First, participants taking more than 30mg did not show greater improvements in depression or anxiety than those taking 30mg or less. Second, more intense mystical experiences were only weakly associated with better mental health outcomes. If dose-response linearity and mystical-experience intensity have been baked into trial designs as predictors of efficacy, the Oregon data suggests those bets need revisiting before they reach a pivotal study.
The safety profile compounds the case for taking this evidence seriously. Across 346 closely followed participants, there were four serious behavioral reactions (1.2%), one hospital transport, and zero serious medical reactions, per Austin’s post. The four adverse cases share a profile that any protocol designer should study: all were psychedelic-naive, all arrived with pre-existing mental health conditions in treatment-seeking mode, all took 25 to 50mg doses, and two were taking antidepressants. Chris Stauffer’s read of the data confirms that harm did not distribute randomly across the cohort. Risk sorted by readiness, not by molecule. That is a finding with direct implications for screening protocol design in every ongoing IND application in this space.
Pharma Is Betting Billions on a Bridge That Doesn’t Exist Yet
While OHSU was publishing naturalistic outcome data, Big Pharma was writing nine-figure checks in the other direction. Greg Mattingly’s breakdown of current M&A activity documents four transactions that together represent several billion dollars in committed or contingent capital: Eli Lilly’s deal with AtaiBeckley worth up to $3.8 billion (including BPL-003, a 5-MeO-DMT compound for treatment-resistant depression), AbbVie’s move on Gilgamesh’s bretisilocin for up to approximately $1.2 billion, Otsuka’s $1.225 billion deal with Transcend centered on methylone for PTSD, and Johnson & Johnson reportedly leading an $85 million Series C for Delix Therapeutics, a non-hallucinogenic neuroplastogen platform on which J&J holds a right of first negotiation following Phase 2 results.
That last deal is the tell. J&J is not buying the psychedelic experience. It is buying the downstream neuroplasticity mechanism, structurally insulated from the regulatory and scheduling complexity that the classic psychedelics carry. Mattingly frames this pattern correctly: the pharma wager is on rapid-acting neuroplasticity platforms, with the hallucinogenic phenotype treated as a feature to be engineered around rather than preserved. That is a commercially rational strategy, but it creates a specific problem for the evidence ecosystem.
The Oregon real-world data tells us something about psilocybin in its full experiential form, delivered by non-clinical facilitators to a general adult population, with the mystical experience intact and unmanaged. The compounds pharma is developing are optimized to strip exactly that experience out. The datasets will not translate cleanly between those two realities, and the regulatory frameworks that govern one are not designed for the other.
Anya Ragnhildstveit and her co-authors at the Cambridge Psychedelic Research Group address this translation problem directly in their Nature Medicine piece on moving psilocybin from research to practice. The gap between what controlled trials demonstrate and what regulated service delivery produces is not simply a methodological artifact. It reflects a genuine structural discontinuity: clinical trials optimize for efficacy signals in diagnosed populations under maximum therapist support, while Oregon’s model optimizes for access in a general population under facilitator support. The outcomes may look similar on a depression scale, but the mechanisms producing them and the populations experiencing them are different enough that regulators treating the evidence as interchangeable would be making a category error.
The Monitoring Gap That No One Is Funding
Carly Newhouse’s accounting of the VA’s psychedelic research portfolio adds a useful institutional counterpoint: the VA is now on its 20th clinical trial exploring psychedelic-assisted therapies, with more than $23 million distributed through the VA’s Office of Research and Development, budget stability unaffected by HHS cuts, and a tailwind from the April 2026 executive order directing acceleration of FDA evaluation for psychedelic compounds with Breakthrough Therapy designations. That is a federal research infrastructure operating at genuine scale, insulated from the funding volatility that has hampered SAMHSA and other HHS entities.
But the possible suicidality signal flagged in the Thompson et al. dataset, described by Pawlik as requiring better monitoring, sits in a category that neither the VA’s clinical trial machinery nor pharma’s IND infrastructure is currently designed to capture at the population level. The Oregon and Colorado programs collectively served tens of thousands of clients. The adverse event reporting mechanisms in those programs were not built to the pharmacovigilance standards that an IND protocol would require. James L.’s Maloca Healthcare platform, designed as a first-of-its-kind pharmacovigilance program for psychedelic experiences across retreat settings in Brazil, Costa Rica, Catalonia, and the United States, is one emerging attempt to close that gap. It will not close it at the scale that the regulatory question now requires.
Ken Weingardt frames the central question cleanly: what happens when psilocybin moves out of the research laboratory and into the real world? The Oregon data offers an intriguing early answer. But “intriguing early answer” is not an evidentiary standard that the FDA has any existing framework to incorporate, and the agency has shown no public appetite to build one. The pharma deals are closing. The trials are enrolling. The VA is on its 20th study. And the largest real-world dataset in the history of psychedelic research is sitting outside every regulatory pathway that would make it count.
The population that can afford $900 to $5,000 per session is already generating the data. The regulatory system that would let anyone else benefit from it has not been designed yet.
References
- Paul F. Austin, LinkedIn post on OHSU JAMA Network Open psilocybin study, 2026
- Shereef Elnahal, LinkedIn post on OHSU psilocybin real-world study publication, 2026
- Lucas Pawlik, LinkedIn post on Thompson et al. Oregon/Colorado dataset (DOI: 10.64898/2026.08.11.26360133), 2026
- Greg Mattingly, LinkedIn post on Big Pharma psychedelic M&A activity, 2026
- Anya Ragnhildstveit, LinkedIn post on Nature Medicine piece, Cambridge Psychedelic Research Group, 2026
- Carly Newhouse, LinkedIn post on VA psychedelic clinical trial portfolio, 2026
- James L., LinkedIn post on Maloca Healthcare pharmacovigilance platform, 2026
- Ken Weingardt, LinkedIn post on OHSU Oregon psilocybin services outcomes, 2026
- Chris Stauffer, LinkedIn post on Oregon psilocybin services safety and outcomes data, 2026
Moe Alsumidaie, MBA, MSF, is founder and Chief Editor of Vanguard Publications, which publishes Clinical Trial Vanguard, Pharma Vanguard and BullScope, and Head of Research at CliniBiz. He has two decades in clinical trial operations and data science, with earlier roles at Genentech, Abbott Vascular and Stanford University Medical Center, and is a guest lecturer in clinical trial sciences at Rutgers University.
