She checked every box. Two sessions of preparation, psychoeducation, rapport-building. A therapist in the room. A comfortable, non-clinical space. Two sessions of integration afterward. She was a participant in COMP001, COMPASS Pathways' phase 2b trial of synthetic psilocybin for treatment-resistant depression—the closest thing psychedelic medicine has to a gold standard for safety.
She came out of it with worsening suicidal ideation and a prolonged period of eating so little that she needed active support to stabilize.
She was not careless. She was not at a warehouse party. She was one of twelve participants who experienced serious adverse events in that single supervised trial, and she experienced two of them herself.
She was one of twelve participants who experienced serious adverse events in that single supervised trial, and she experienced two of them herself.
This is the detail that punctures the reigning story about who gets hurt.
The reassuring orthodoxy goes like this: psychedelics are dangerous mainly in the chaotic recreational world—high doses, adulterated product, polysubstance mixing, no supervision, and users with hidden psychiatric vulnerabilities. Clean up the setting, screen the person, put a trained sitter in the room, and the harm ledger empties out.
There is real truth in this. The case literature does concentrate serious acute events in unsupervised, high-dose, or polysubstance contexts. Persistent problems—prolonged psychosis, triggered mania, maladaptive behavioral shifts—are genuinely uncommon in controlled trials, and have historically been attributed to predisposing conditions.
But the orthodoxy quietly does two jobs at once. It catalogs the risks, and it locates them all somewhere else—in the festival crowd, in the unscreened, in the already-ill. The COMP001 participant shows why that second move fails.
Start with the predisposition list, because it is longer and more specific than the breakthrough narrative tends to admit.
A personal or family history of psychosis or bipolar disorder is the classic red flag, and for good reason. A systematic review of case reports found 17 documented cases of schizophrenia-spectrum disorder and 17 of affective disorder—depression, mania, or both—following psychedelic use, some emerging after a single dose. Recovery was the rule for the affective cases. It was not the rule for the schizophrenia-spectrum ones.
Trauma is its own axis of risk. One naturalistic study found that people with a history of adverse childhood experiences were more likely to report psychological difficulties afterward, with the risk climbing as the ACE burden climbed. Psychedelic states can surface recovered memories—accurate or not—experienced with overwhelming felt reality, a problem the field openly admits it does not know how to handle.
Then there is the body. Classic psychedelics are broadly well tolerated, but psilocybin and its metabolite weakly inhibit the hERG potassium channel that governs cardiac repolarization—meaning people genetically predisposed to prolonged QT, or taking interacting drugs, sit closer to a cardiac edge than the average trial volunteer.
Cardiovascular disease widens the danger further. Arrhythmias and vascular events have occurred in people with underlying heart conditions on MDMA; ketamine's acute hemodynamic effects make it contraindicated in aneurysmal or hemorrhagic disease and risky in uncontrolled hypertension.
And there is a hard toxidrome ceiling that dose and heat can breach regardless of intent.
Consider Mr B, 23, no significant medical history, who collapsed at an outdoor festival after hours of dancing in hot weather, having taken what he believed was LSD. He seized in the ambulance. He arrived with a core temperature of 39.7°C, rigid, hallucinating, in rhabdomyolysis, his sodium dangerously low.
That cascade—hyperthermia driving muscle breakdown, sympathetic overdrive, hyponatremia compounded by hypotonic fluids—is not a psychological event. It is a physiological ceiling that dose, exertion, and heat can push a person straight through. Set and setting have nothing to offer once the temperature climbs.
Youth belongs on the list too. In one deliberately designed negative-outcomes study, 53% of the completers were under 25, and young age recurred as a causal thread in the severe cases.
So the honest contraindication catalog is substantial: psychosis and bipolar history, family loading, trauma and ACEs, QT genetics, cardiovascular disease, youth. Screen aggressively and you shrink the risk. This is the orthodoxy at its strongest.
Now the turn. Because the most disturbing cases in the literature did not happen to the unscreened festival crowd or the psychiatrically loaded. They happened to people with clean histories, inside supervised or expert contexts.
Return to that same negative-outcomes study of 32 completers. It was built specifically to capture the responses the positive-outcome trials miss—self-reported negative reactions lasting more than 72 hours. Among those completers, 37.5% received a new psychiatric diagnosis that emerged after their psychedelic use. Anxiety arose or worsened in 87%.
The researchers named a common mechanism: a pro-plasticity drug, frequently "over-dosed," colliding with adverse conditions or vulnerability. Not a bad room. The drug itself, doing what it does to a malleable brain—sometimes in the wrong person, sometimes simply in too large a quantity.
That phrase—pro-plasticity drug—is the crux. These compounds are being celebrated precisely because they open the brain to change. A tool that pries the mind open does not become safe merely because someone kind is sitting beside you.
Which brings us to the case that should end the fiction that supervision neutralizes the risk.
Dr. A, a 71-year-old clinical psychologist with no prior personal or family psychiatric history, enrolled in a psychedelic therapy facilitator-training program that involved taking repeated high doses of psilocybin mushrooms herself.
Over the following four months she developed an unremitting constellation of symptoms—severe sleep impairment, anhedonia, suicidal ideation. Her chief complaint on admission to an inpatient psychiatric unit: "I'm stuck in a loop."
She had no history to predict this. She was, by profession, a mental health expert. She was inside a formal program run by people who present themselves as guides.
And here is the part that indicts the culture around the drug: when her symptoms worsened, her psychedelic therapy trainers advised her against seeking psychiatric support. The delay was not incidental. It was steered.
Her symptoms ultimately resolved only after a course of electroconvulsive therapy—the most intensive intervention conventional psychiatry has for severe, treatment-resistant depression.
Hold these three cases together. A gold-standard trial participant with worsening suicidality and disordered eating. A 23-year-old in a full toxidrome. A 71-year-old expert hospitalized, then given ECT, after being told to stay away from psychiatry. None of them is the cautionary figure the orthodoxy points to.
The pattern is not "unsafe settings produce harm." It is that harm reaches into the safest-looking settings, sometimes carried by the very drug the setting was built to deliver.
There is also the matter of timing. Case reports describe psychotic and affective states emerging after a single use, and flashback-type phenomena persisting for months—sometimes years. The consent conversation that promises a bounded, supervised session cannot honestly promise a bounded window of risk.
This is not an argument to shut the field down. The efficacy signals are real, and the FDA breakthrough designation for treatment-resistant depression did not come from nowhere.
It is an argument that enthusiasm has outrun the safety accounting. The contraindication list needs to be treated as long and taken seriously—not as a formality that clears most applicants, but as a genuine filter for psychosis and bipolar loading, trauma burden, cardiac and QT risk, and youth.
Consent has to be honest about delayed reemergence: that a difficult state can arrive or return long after the sitter has gone home, and that "helpful" and "harmful" can coexist, as the COMP001 participant's own account—she found the experience beneficial despite it all—makes uncomfortably clear.
And the field has to abandon the most convenient fiction of all: that a comfortable room and a trained presence neutralize a drug engineered to make the brain more plastic. Supervision is a safeguard. It is not an antidote.
The other ledger exists. It is not confined to the people we would prefer to blame.