The Exclusion Criterion: Psilocybin Meets the Patients the Trials Left Out

Before I was a CE provider, I answered phones for the 988 Suicide and Crisis Lifeline. I talked with people experiencing suicidal ideation every day, and I learned that "suicidal ideation" is not one thing: it arrives in every kind, every variety, every severity, from a passive wish to not wake up that has lasted twenty years to an active plan taking shape in the next hour. Those calls left me with two convictions I hold with equal force. I do not want to recommend a treatment that could increase suicidal thinking. And I desperately want treatments that can reduce it.

A new study puts both convictions under pressure at once, which is why it's worth a careful read rather than a headline. The relationship between psilocybin and suicidal ideation has just acquired its first dedicated trial data, in a population psychedelic research has spent two decades deliberately excluding.

Researchers at Sheppard Pratt enrolled 20 adults with major depressive disorder and chronic suicidal ideation, defined as daily suicidal thoughts lasting an hour or more for at least three months, plus at least two failed antidepressant trials. Each received a single 25 mg dose of COMP360 synthetic psilocybin inside a heavyweight protocol: medication washout, three preparation sessions totaling six or more hours, an eight-hour supervised dosing session, and three integration sessions afterward.

The results, on their face, are striking. On the Modified Scale for Suicidal Ideation (MSSI), the primary outcome, scores dropped a mean of 13.95 points at week 3 (Cohen's *d* = 1.73). Three-quarters of participants met response criteria at week 3; 45% reached full remission of suicidal ideation, and 35% were still in remission at week 12. Depression scores (MADRS) fell 17 to 20 points across timepoints. There were no serious adverse events.

Now the other side of the ledger, which the press release does not lead with. This was an open-label, single-arm study: every participant knew they received psilocybin, and there was no comparison group, so expectancy effects are uncontrolled and the authors say so plainly. Twenty people at one specialized academic center. Sixty percent restarted psychiatric medications after week 3, which muddies any claim about durability. And two participants, 10% of the sample, showed increased suicidal ideation relative to baseline; in one of them, the increase persisted. In a study this size, that's two people. In a treatment delivered at scale, that's a percentage you'd want to understand very well first.

Early evidence, then, that a single dose of psilocybin with intensive psychological support may reduce chronic suicidal ideation, for carefully selected patients, with real uncertainty about durability and a small signal in the wrong direction for a minority. That's the accurate sentence. It's less quotable than the headline version, and it's the one I'd want a colleague to carry into a consultation.

The most important thing about this study may be who was in it rather than what it found. For two decades, active suicidality has been a near-universal exclusion criterion in psychedelic trials. The logic was defensible: an unproven compound, a destabilizing acute experience, a population where the worst outcome is irreversible, and institutional review boards doing what they exist to do.

But notice what that caution built. The patients most likely to ask about psilocybin, people with treatment-resistant depression and chronic suicidal thinking, are precisely the people about whom the trial literature could say almost nothing. Clinicians have been fielding questions from a population the evidence base was designed not to include. Exclusion criteria protect trial participants; they also quietly decide who gets data and who gets extrapolation. The people who answer crisis lines, and the people who call them, have been living on the extrapolation side.

Worth being precise about how far this study moves that line: it enrolled chronic ideation with C-SSRS severity 3–4, and still excluded anyone at level 5 (active plan with intent) in the prior three months, along with psychosis history, bipolar I, and recent substance use disorders. The most acute population remains unstudied. This is the first careful step, not the arrival.

Honesty about this study requires holding it next to a less comfortable dataset. In psilocybin trials for treatment-resistant depression, serious adverse events have included suicidal ideation and intentional self-injury, including in Compass's own Phase 2 program, and the professional literature has begun discussion of adverse events differentiated by substance and context rather than treating psychedelics as a single safety category.

Both facts are true at once: a dedicated trial in chronically suicidal patients recorded no serious adverse events and large reductions in ideation, and the broader TRD trial record includes suicidality among its serious adverse events. These aren't contradictory findings to be resolved in favor of whichever one matches your priors. They're the boundaries of the territory risk assessment actually lives in: a treatment that may reduce suicidal ideation in many patients and may be followed by worsening in some, studied so far in samples too small to tell us who is who.

For clinicians working in or near KAP and psychedelic care, a few concrete implications:

• Baseline severity grading matters more, not less. The study's own architecture turned on the C-SSRS: levels 3–4 were eligible, level 5 was not. If your screening can't make that distinction reliably, you can't map your client onto this evidence at all.
• The monitoring window is post-acute. The two worsening cases showed up in scale scores over weeks, not as dosing-day emergencies. Structured reassessment at the one-week and three-week marks, not just a same-day check, is what this data pattern argues for.
• Washout and medication restarts are clinical events. Participants tapered medications before dosing and most restarted afterward; in chronically suicidal patients, both transitions are risk windows in their own right.
• The honest patient conversation includes both halves. "Early evidence suggests it may help, a small number got worse, and nobody can yet predict which group you'd be in" is harder to say than either the hype or the dismissal. It's also the only version that's true.

There's a values question under all of this that I don't want to leave implicit. Who gets studied is a justice question. Chronic suicidality is concentrated among people the system has already failed at least twice, by definition, in this sample. Researching them carefully, rather than excluding them indefinitely, is the field beginning to take its own access rhetoric seriously. The same logic should reach the populations still outside the criteria.

For the fuller screening picture, the Comprehensive KAP Assessment Bundle (2 CEs) pairs suicide assessment with medical and psychological screening.

Twenty adults with major depression and chronic suicidal ideation each received one 25 mg dose of psilocybin inside an intensive support protocol. Suicidal ideation scores dropped a mean of about 14 points at week 3, three-quarters responded, and there were no serious adverse events. But it was open-label, single-arm, and small, so expectancy effects are uncontrolled.

The evidence is genuinely mixed. This dedicated trial recorded no serious adverse events and large reductions in ideation, yet two of twenty participants (10%) showed increased suicidal ideation, persisting in one. Broader depression trials have also listed suicidality among serious adverse events. It may help many and worsen some, in samples too small to predict who.

The trial enrolled chronic ideation at C-SSRS severity levels 3–4 and still excluded level 5 (active plan with intent) in the prior three months, plus psychosis history, bipolar I, and recent substance use disorders. The most acute population remains unstudied. This is a first careful step into a group psychedelic research has long excluded, not an arrival.

Baseline severity grading matters more, not less: the study turned on the C-SSRS, with levels 3–4 eligible and 5 excluded. The two worsening cases appeared in scores over weeks, so reassess at one and three weeks, not just dosing day. Medication washout and restarts are their own risk windows, and patient conversations should name both halves of the evidence.