“Psychedelic-assisted therapy” is still used as if it named one treatment. The four most visible psilocybin-class programs no longer fit under it. Three of them use synthetic psilocybin and one uses a deuterated psilocin analogue, and each has been built into a different product: a different patient, a different number of doses, a different staffing model and a different amount of structured therapy around the drug.
Those differences are written into trial registrations and annual reports, and they decide what a clinic has to provide and what a payer is asked to cover. They are also easy to overstate. On the measure that drives clinic economics most directly, the length of a supervised dosing day, the programs that have put a number on record are closer together than their marketing suggests.
Four patients, four regimens
Compass Pathways is developing COMP360, synthetic psilocybin, for treatment-resistant depression, defined in its Phase 3 registrations as failure of two to four medications in the current episode. Its 2025 annual report describes both pivotal trials as monotherapy. One tested a single 25 mg dose against placebo. The other tested two doses three weeks apart at 25 mg, 10 mg or 1 mg. Both met their primary endpoints, and the company expects to complete its rolling New Drug Application this quarter.
Usona Institute is testing synthetic psilocybin in major depressive disorder without a treatment-resistance requirement. Its Phase 3 trial, uAspire, assigned participants to one oral dose of 25 mg, 5 mg or inactive placebo, and eligible participants may be offered open-label 25 mg during a one-year follow-up. It enrolled 238 adults. The registry records primary completion in January 2026, and as of October 12 neither the registry nor Usona’s own site shows topline results.
Helus Pharma’s HLP003 is not psilocybin. It is a deuterated analogue of psilocin, the active metabolite, and it is being tested as an adjunct. The registrations for its two pivotal trials, APPROACH and EMBRACE, require participants to have had an inadequate response to a stable antidepressant and state that they “will continue on their current antidepressants.” The regimen is two doses, 16 mg in APPROACH and 8 mg or 16 mg in EMBRACE, about three weeks apart. The registry records APPROACH’s primary completion on September 4, topline data are expected this quarter, and the company is targeting a possible application in 2028.
Incannex Healthcare’s PSX-001, synthetic psilocybin, is aimed at generalized anxiety disorder and is the earliest of the four. Its completed Phase 2 trial gave two doses alongside psychotherapy. The company’s annual report, filed September 25, says it is adjusting the development plan, including the design of the Phase 2 study.
So before any question of delivery arises, the four are headed toward different labels: a later-line monotherapy, a broader depression indication, an add-on to existing medication, and an anxiety treatment. A payer writing coverage criteria for one would not be describing the patients eligible for another.
The support model is where they separate most
The sharper divergence is in what surrounds the dose, and here the registry language is more revealing than any presentation.
Usona’s is the most staff-intensive protocol on record. Its registration describes a “Set and Setting” protocol with preparatory meetings with two facilitators, “a 7-10 hour dosing session” supervised by the same two facilitators, and “4 hours of post-dose integration sessions.” Group psychosocial support continues through the follow-up year.
Compass’s annual report describes three phases, preparation, administration and follow-up, with a dosing session that “lasts approximately six to eight hours” and “at least one healthcare professional” present throughout. It adds that in its trials those professionals “refrain from intervening with the patient’s experience, unless required for safety reasons.” The company has spent 2026 drawing that line more firmly. Its second-quarter release refers to a “standardized monitoring-and-support approach” and cites an analysis of its PTSD trial in which almost 80 percent of the dosing session was spent in silence. The trial registrations still describe the drug as “administered with psychological support,” and list the intervention as given “under supportive conditions.”
Incannex sits at the other end. Its annual report says PSX-001 “is designed to be used in combination with psychological therapy from trained therapists,” and describes the Phase 2 regimen as two dosing sessions plus “up to 11 non-drug, specialist psychotherapy sessions” over ten weeks. That is a psychotherapy program with a drug inside it, not a drug with monitoring around it.
Helus is the least defined, and the gap between its documents is worth noticing. Its October corporate presentation says the duration of HLP003’s acute effects, about four to six hours, “fits well into existing interventional day clinic setting,” and it does not describe a support model at all. Its trial registrations do: every arm receives “manualized psychological support performed by facilitators.” The company’s own filings name that model, EMBARK, and describe a therapist training program behind it. Whether any of it is intended to carry into a label is not stated anywhere.
Three components that are often blurred need to stay separate here. Supervision and clinical monitoring are present in all four programs. Preparation and integration sessions are present in all four trial protocols, at very different intensities. Structured psychotherapy is claimed by only one. A program that drops the word therapy has not dropped the room, the staff or the observation period.
The dosing day is long in every version
That last point is where the fragmentation argument has to be qualified.
Compass’s trials ran six-to-eight-hour sessions, and its management has told investors it expects about six hours of required monitoring, against two for esketamine nasal spray. Usona’s protocol specifies seven to ten hours. Helus claims four to six hours of acute effects but has published no discharge criteria or monitoring requirement for HLP003. Incannex’s annual report gives no session length for PSX-001.
None of the three on record is a short visit. Helus makes the comparison itself: its presentation says its other clinical candidate, an intramuscular deuterated DMT for anxiety, showed acute effects of about 90 minutes and discharge readiness within about three hours in Phase 1, and describes that profile, not HLP003’s, as fitting the “existing SPRAVATO treatment paradigm.” By the company’s own framing, the oral program is a day-clinic product.
The economic variable, then, is less the hours per session than what fills them and how often they recur. Three mechanisms follow from the protocols.
Staffing is the first. The Category III billing codes created for psychedelic medication monitoring, effective January 2024, are hourly, with one code for the first clinician and separate add-on codes for a second clinician and for clinical staff. A two-facilitator protocol and a one-professional protocol are different claims under that structure. Medicare lists the codes without a national payment rate, so the structure says who can bill and not what they would be paid.
Frequency is the second. Compass tested one dose and two doses, and its single-dose trial included a blinded retreatment stage. Usona offers a further dose to participants who meet set criteria. Helus’s extension study permits up to three further doses. Helus charts the result as two clinic visits a year against 34 to 56 for esketamine, with a footnote that the HLP003 profile “is illustrative and is subject to further validation in Phase 3 studies.” Fewer visits reduces patient burden and the referral radius a clinic needs. It also means each patient generates far less recurring activity for a site than an esketamine patient does. Whether that trade is attractive depends on how the long session is paid, which no payer has yet said.
Background medication is the third. An adjunctive product leaves the patient on an existing prescription and inside an existing prescriber relationship. A monotherapy, tested in trials that required participants to discontinue prohibited medications, raises questions a label will have to answer about tapering and about who manages the patient between doses. Usona’s registration does not state its antidepressant policy, so its position on that axis is not established.
What the protocols cannot tell you
Trial protocols are designed to produce clean evidence, not to describe a commercial service, and several of these differences may not survive approval.
A label and a risk evaluation and mitigation strategy can require less than a protocol did, or more. Compass’s expectation of six hours and no psychotherapy requirement is the company’s expectation; the FDA has not finalized either. Preparation and integration sessions that every trial participant received may or may not become conditions of use, and that single decision would move the cost of treatment more than any difference between the molecules.
Nor do the protocols rank the models. No trial has compared a therapy-intensive regimen with a monitoring-only one, so there is no evidence here that more support improves outcomes or that less support preserves them. Helus’s own slide carries the necessary caution: no head-to-head comparisons have been made, and results from different trials may not be comparable.
Usona’s economics are also unlike the others’. A nonprofit sponsor with the most staff-intensive protocol and no disclosed commercial plan may not be optimizing for clinic throughput at all.
What can be said is narrower and more useful than “the category is fragmenting.” The four programs have already diverged on who the patient is, how many doses are given, how many people are in the room and how much therapy is attached. Where a session length is on record, they have not diverged on the need for a supervised room for most of a day. Who supplies that room, and on what terms, is a separate question. The first approved label will settle which of the trial-era requirements become permanent, and every program behind it will be measured against that document, not against its own protocol.