FDA Proposal

Why I wrote a REMS

When the Food and Drug Administration approves a drug whose risks a label alone cannot manage, it attaches a Risk Evaluation and Mitigation Strategy, a REMS. For psilocybin and MDMA, that is where the decisions that matter will be made: who may sit with a patient, with what training, in what setting, under what watch, and what happens after the patient goes home.

On 14 September 2026 the agency held its first public hearing on psychedelic drugs as a class and asked four questions about exactly those conditions. Nobody I could find had drafted the instrument itself. I have spent years in the part of this work the hearing kept pointing at, the months after the session, so I wrote the draft. I should also say what I want from this. Not to be a therapist; the limits of therapy are real, and I have lived on both sides of them. A seat inside the system with a name the system recognizes, and chaplaincy is the door I am looking at, which is why the tenth element asks for one and says so. The draft is below, in full, versioned and dated, and it will keep changing as people correct me.

How a proposal like this reaches the agency. By law the drug's sponsor writes the REMS and the agency negotiates it. A practitioner has four doors: a comment to an open docket, which is this one; the public hearing portion of the advisory committee meeting that will review the first application; a citizen petition, which obliges the agency to answer in writing within 180 days and has changed a REMS before; and the sponsors themselves, who read a docket before a reviewer does. This proposal goes through the first door on 5 October and is written to fit the other three.

Two companions

My comment to the docket. Six asks in under 5,000 characters, filed under my name with this proposal attached. Public and permanent. The docket is open to anyone until 11:59 p.m. Eastern on 5 October 2026.
[The Day the Sacred Was Ruled Out of Scope](LINK TO POST 37). My account of the hearing for anyone who was not in the room: who spoke, who could not risk speaking, and the argument the proposal turns into clauses.

A third piece, addressed to the body that would someday certify practitioners, will be added here when it is ready.

A REMS for the Room

A practitioner's proposal to the Food and Drug Administration, and to the sponsors who will draft the first one

Version 1.2 · September 2026 · Submitted to docket FDA-2026-N-7542 as an attachment to the author's public comment · A working document, offered for correction

What this is, and what it is not

This is a proposal for one instrument, addressed to the office that holds it.

The Food and Drug Administration cannot license a guide, write a scope of practice, or decide what anyone is paid. Those belong to the states, to the American Medical Association, to Medicare and the insurers. Saying so first matters, because an ask filed at the wrong office is a no on a technicality rather than a no on the merits.

What the agency can do is attach conditions to a drug's approval through a Risk Evaluation and Mitigation Strategy, and the strongest of those conditions reach the setting, the personnel, the monitoring, and the record. The setting is therefore within the agency's reach, and the question is what shape the agency gives it. I. Glenn Cohen of Harvard Law stated the consequence in April 2026: what kind of REMS the agency attaches to an approval "will heavily determine the delivery model, the cost involved, and how many Americans can ultimately get access." The design of the REMS is, in effect, access policy.

So far as I can find, no peer-reviewed article, no advocacy organization, and no published filing has drafted one for this class. The nearest scholarship, Fahrenkopf in the Food and Drug Law Journal, analyzes the esketamine program as the federal template and names the unsolved gap without filling it. That absence is the reason this document exists. It is a first draft, offered for correction.

This document does not invent a safety standard. It does four things instead. It sets out what the agency has already required of drugs carrying similar risks, so nothing here asks for a power the agency lacks. It surveys what the field has already built, with hour counts and thresholds, so the agency is not asked to draft language two states already enforce and a third is adopting. It describes one practice in the open, so the ask is not abstract. Then it proposes ten elements, each with its statutory hook, its precedent, and the outside standard the agency could adopt rather than compose. Where an element is novel it says so, and its weaknesses are named inside it.

Where the evidence is thin, this document says so before a reviewer has to find it.

How this document relates to the comment it accompanies

This proposal is attached to my public comment to docket FDA-2026-N-7542, which makes six asks in under 5,000 characters, and to my published account of the 14 September 2026 hearing, which carries the argument for a general reader. The comment states the asks. This document supplies the elements behind them, with statutory hooks and precedent. The list below maps one to the other, so a reviewer can move between them.

Ask 1, spiritual care as a competency, not a religious claim. Element 10. Hook: (f)(3)(B).
Ask 2, a named role for the months after. Elements 7 and 8. Hooks: (f)(3)(D) and (E).
Ask 3, experience as a safety competency. Elements 2, 3, 4, and 5. Hooks: (f)(3)(A) and (B).
Ask 4, certify the room by what happens inside it. Elements 1 and 6. Hooks: (f)(3)(B), (C), and (E).
Ask 5, let the support stand in daylight. Part VIII, and outside the REMS. Not a REMS matter.
Ask 6, a participant record, and one table. Element 9, and Part XI. Hooks: (f)(3)(F) and (f)(5).

The comment's second ask describes the cadence I run in my own practice, contact within seventy-two hours, again at two weeks, and monthly through six months. Element 8 asks the agency to require less than that: two documented safety contacts, at 24 to 72 hours and at 6 to 8 weeks, using validated instruments. The practice cadence is what one practitioner does. The element is the minimum a regulator can defend on current evidence.

Who is writing, and from where

I am a psychedelic guide and integration coach, a Marine Corps combat veteran, a former diplomat, an ordained minister, and a founding member of the Veteran Mental Health Leadership Coalition.

For much of my adult life I was a patient in the traditional mental health system, treated for post-traumatic stress after the Marines and for depression classified as treatment-resistant.

Since the Marines I have spent fifteen years in healthcare innovation, including as chief operating officer at a company behind an FDA-cleared biometric device for nervous system regulation, work with a state health department on opening public data, and years teaching pharmaceutical executives to design for health rather than for the management of pain. None of it reached what was wrong with me.

I now sit with adults past twenty-one who chose this path for themselves, close to five hundred ceremonies and thousands of hours of integration after them, offered within the harm reduction tradition. I do not supply any controlled substance. I am also the integration partner at a ketamine clinic in Arizona, where the medicine is prescribed and the setting licensed, and I support clinical trials in this field.

I write, then, from three positions: former patient, current practitioner, and operator. I have made mistakes in this work and have been corrected by peers, which is an argument for the accountability structures proposed here rather than against the person proposing them.

Part I · The instrument, and its limits

A REMS is authorized by 21 U.S.C. 355-1. Where a label alone will not manage a serious risk, the agency may require elements to assure safe use. Under 355-1(f)(3) those elements are a closed list of six:

(A) prescribers have particular training or experience, or are specially certified
(B) pharmacies, practitioners, or health care settings are specially certified
(C) the drug is dispensed to patients only in certain health care settings
(D) the drug is dispensed with evidence or other documentation of safe-use conditions
(E) each patient is subject to certain monitoring
(F) each patient is enrolled in a registry

Taken together, the six elements describe the setting: who may be present, with what training, in what kind of facility, under what monitoring, with what documented. Three features of the statute shape any ask.

The sponsor proposes, and the agency negotiates. Under 355-1(a) the applicant submits a proposed REMS with the application, and under (g) the applicant may propose modifications at any time. There is no notice-and-comment rulemaking for an individual REMS. The agency's own 2019 guidance on when a REMS is necessary says it may consider information from external experts, advisory committees, the literature, and professional societies in reaching that judgment, so a proposal like this one is addressed to the agency and to the sponsors at once, and reaches the sponsors first.

Burden is a statutory constraint rather than a courtesy. Under 355-1(f)(2)(C) elements shall not be unduly burdensome on patient access, "considering in particular" patients with serious conditions and "patients who have difficulty accessing health care (such as patients in rural or medically underserved areas)." Under (f)(2)(D) elements shall, to the extent practicable, conform with those used for drugs with similar risks. Under (f)(3)(A), certification methods must be made widely available, including in frontier areas. By the published accounts of the September hearing, a retired Navy SEAL told the agency from the podium that a fifteen-minute physician-availability clause alone could close qualified clinics across the rural West. That objection is the (f)(2)(C) test, stated from the field. (One note on language: "minimally burdensome" appears in guidance and commentary, not in the statute. The tests are "not unduly burdensome on patient access" and "minimize the burden on the health care delivery system.")

The agency is already obliged to ask people like us. Under 355-1(f)(5) the Secretary shall, through the Drug Safety and Risk Management Advisory Committee, or another of its advisory committees, seek input from patients, physicians, pharmacists, and other health care providers on how elements may be standardized so as not to burden access or the delivery system, and shall periodically evaluate whether existing elements meet those tests. This document is submitted under that subsection.

Part II · What the agency has already required, and what it has already said

Four programs that establish the shape

None is a psychedelic. Each answers a question a psychedelic approval will raise.

Esketamine, 2019 to present. The closest analogue in the pharmacopoeia: a dissociative given in a clinic, with a period afterward in which the patient cannot be sent home. The original approval imposed five of the six elements, all but prescriber certification: certified settings, certified pharmacies dispensing only to those settings, nothing going home, self-administration under direct observation, and at least two hours of monitoring followed by a clinical-stability assessment before leaving. A 2020 modification split the requirements by inpatient and outpatient setting, so the agency has already accepted that one regime need not fit every room.

Brexanolone. A sixty-hour infusion for postpartum depression. Its label required continuous pulse oximetry, sedation assessment every two hours, a provider on site throughout, facility and patient enrollment, and that a patient be accompanied during interactions with her own children. That last condition matters more than its subject. It is proof the agency will write a rule about who else is in the room when the risk calls for it. The approval was withdrawn in April 2025 at the sponsor's request. The precedent stands.

Clozapine, released June 2025. After a docket and a joint advisory committee, the agency removed the REMS, finding its elements "no longer necessary" and noting that "the REMS can sometimes impede access." The most encouraging precedent in the set and the most demanding. A REMS is a living instrument, so a first version need not be perfect. It also means a first version will be lived inside by real patients for years.

iPLEDGE. The cautionary tale, and the agency knows it. A 2011 study of 129,544 pregnancy-capable patients found the program had not significantly reduced exposure. A 2019 study found non-white patients had higher rates of interruption from the system's logistics. At the March 2023 advisory committee, seventeen of twenty-two voting members voted against the nineteen-day lockout, and the program's own pregnancy registry came into question. A registry design fixed at approval is lived with for decades.

What the precedent establishes

The agency has already required, for one drug or another:

that a setting be certified
that a named provider be present throughout an administration
that a patient be monitored for a stated minimum period
that a clinical-stability assessment precede discharge
that equipment be present on site
that a patient be enrolled in a registry
that another human being be present during a vulnerable interval

Each of those applies directly to a supervised psychedelic session. The instrument fits. What it has not previously been asked to cover is the period after discharge.

What the agency has already said about this class

Two documents matter more than any argument a practitioner can make, because the agency wrote them.

The final guidance. Psychedelic Drugs: Considerations for Clinical Investigations, finalized 14 July 2026, already specifies two monitors for the duration of a session: a lead with graduate-level training and clinical experience in psychotherapy, licensed to practice independently, and an assistant with a nursing or bachelor's degree and a year of clinical experience. Where the lead is not a physician, one on call within fifteen minutes. The agency removed the draft's list of qualifying professions in response to requests for broader eligibility. It also asks that sponsors follow subjects beyond week twelve, with blinded long-term follow-up typically to twelve months.

Two consequences follow. The agency has already committed to a two-person model with tiered credential floors, so an element carrying that forward is continuity rather than novelty. And the agency has already asked for twelve-month follow-up before approval. A REMS whose watch ends at discharge, or at Oregon's seventy-two hours, would be less rigorous than the agency's own pre-approval guidance. The agency has specified everything except the period after discharge.

The Complete Response Letter. Issued to Lykos Therapeutics on 8 August 2024 and released publicly in September 2025, it cited failure to systematically document abuse-related adverse events, inconsistency between the pivotal protocol and the training manual in defining adverse events, inadequate therapist oversight with patients acutely impaired, inconsistent therapy protocols across sites, several unreported adverse events found on inspection, and an inability to separate the drug effect from the psychotherapy. It recommended an independent audit of all study records. And it asked for standardized discharge-readiness criteria covering both psychological and physiological assessment.

That letter carries everything below. The agency does not have to be persuaded that the human wrapper around the molecule is a regulatory object with a safety signal it cannot currently see. It said so, in writing, about the drug nearest approval. Discharge-readiness criteria are the agency already reaching past the acute session. This document cites the agency's own findings throughout.

Part III · What the field has already built

This is the part of the record that has not been assembled in one place, and it is the reason this document is as long as it is.

The common assumption, inside the field and outside it, is that psychedelic practice is unregulated because no standards exist. That is not true. Standards exist in quantity. Two states enforce them and a third is adopting them. Several are quantified, threshold-based, and published free. What does not exist is any mechanism to make them stick to a practitioner who stops caring what the field thinks.

So the agency is not being asked to draft a standard of care. It is being asked to select from a shelf that is already full, and to supply the one thing those standards lack, which is enforcement.

1. Training: where three states landed

Oregon, OAR 333-333-3050 and 3070: 128 didactic hours for applications from 2025, up from 120, across nine subject areas; a 40-hour practicum with 30 in the room; 10 hours of consultation inside the practicum.
Colorado, 4 CCR 755-1-3 and 755-1-4: 150 didactic hours across fourteen subject areas with per-area hours; a 40-hour practicum with 30 in the room; a separate and continuing 40 hours of consultation.
New Mexico, proposed 7.35.3 NMAC: 65 didactic hours plus 10 with simulated patients; a practicum of 100 hours for facilitators and 120 for practitioners, of which 80 are on administration days with fourteen or more patients; consultation not specified.
The independent programs: 140 to 250 didactic hours; commonly a 40-hour practicum; consultation varies.

Oregon and Colorado legislated independently and produced the identical forty-hour practicum with thirty in the room, and every major program adopted it. The independent programs cluster in the same didactic band: the California Institute of Integral Studies at 140 to 150, the Integrative Psychiatry Institute at 140 or more, Naropa at 150 and 200, Berkeley at 160, InnerTrek at 160 or more. No independent program sits below 100. The clustering reflects two state codes setting the market standard. New Mexico, writing last, went the other way: lighter on the classroom and much heavier on the room. Whatever the agency takes from the three, the direction of the newest rule is toward supervised hours with real patients.

Four further things are settled.

Consultation as an ongoing structure separate from training. Colorado's extra forty hours, Compass's remote group mentoring, Lykos routing adherence findings to supervisors. Nobody in this field believes training ends at certification.
The two-person session with differentiated roles. The Lykos dyad, the agency's lead and assistant, Colorado's Facilitator and Clinical Facilitator, New Mexico's practitioner and facilitator. One architecture, two people, unequal credentials, at least one licensed.
Personal experience with the medicine: permitted, capped, screened, never required. Colorado caps trainee dosing at ten of thirty practicum hours with screening, Oregon at four hours, and the 2024 JAMA Network Open consensus holds it should be neither required nor prohibited, since no evidence shows an outcome difference.
Ethics, boundaries, and touch already carry dedicated hours. Colorado allocates 35 of its 150 to them. The field has already decided boundary violation is the characteristic harm of this modality and priced it into the curriculum.

The field's own consensus curriculum. Passie and twenty-one co-authors, including Mithoefer, Grob, Greer, Nielson, Gorman, Phelps, Belser, Guss, Vermetten, and Gasser, published a model curriculum in the Journal of Psychopharmacology: two to three years, completed psychotherapy training as prerequisite, and supervised practice in three graded stages, observation as assistant, then a minimum of five consecutive sessions as active therapist with a supervisor present and debriefing after each, then independent practice under supervision. Two figures in it are the only ones of their kind. Supervisors must have 400 hours conducting this therapy with actual patients. And the Czechoslovakian curriculum of the 1960s and 70s required five controlled self-experiences and thirty supervised sessions with five different patients. Their verdict on research-only training is the crux: programs built around a study manual "will not prepare therapists to deliver SAP outside of the research environment."

2. Competency assessment: the instrument nobody uses

Here is the widest gap in the field, and it does not track hour counts at all.

At one end, a well-known year-long program certifies on attendance at eighty percent of study groups. At the other sits an instrument almost nobody outside the trials has read. The MAPS adherence ratings manual, version 6.3, rates recorded sessions against itemized behavioral criteria, 24 items for preparatory sessions with separate sets for experimental and integrative, scored yes or no. Raters are licensed or in-training clinicians with a year of psychotherapy internship and a year of trauma work, qualified through about 65 hours over five months. Threshold: 75 percent agreement with a gold standard, inter-rater reliability at ICC 0.75 or above, remediation capped at two rounds. Deviations escalate to supervisors and medical monitors.

This is the most developed competency instrument published in the field: quantified, threshold-based, with a remediation ladder and a rater pathway. It is free, and it was built for the drug class in question. Two supporting instruments use the same architecture: Cybin's EMBARK-CT certifies on live roleplay with remediation below 80 percent, and the Compass fidelity scale by Tai and colleagues requires four supervised psilocybin sessions before leading independently.

Against that, the states. Oregon's exam is open-book, on the regulations, unlimited retakes. A jurisprudence exam, and a good one, but it tests knowledge of the rules rather than conduct in the room. Colorado requires no examination. The Board of Psychedelic Medicine and Therapies has a blueprint and, as of this writing, no examination administered and no candidate certified. The field possesses a free, drug-specific competency instrument with a numeric threshold, and the only credentials anyone can earn measure attendance or rule knowledge.

3. Who trains the trainers

Three mechanisms exist and have never been combined. Oregon, from September 2025, requires at least one instructor at an approved program to hold a facilitator license. Colorado requires two faculty with at least one licensed, practicum supervisors licensed, a program director with documented expertise in adult education, biennial review with withdrawal available, and a Distinguished Educator license, the only dedicated train-the-trainer credential for this work in American law. The consensus curriculum supplies the numeric floor of 400 patient hours. Nobody requires all three, and the national bodies, the university certificates, and the sponsor programs specify no trainer qualification at all.

4. Ethics: the content is settled, the consequence is not

Across the MAPS code, the practice guidelines, the JAMA consensus panel, and the rules of Oregon and Colorado, the following are not in dispute. Sexual contact with a client is categorically prohibited. Only the post-termination bar varies, from one year to two to permanent. Supportive touch requires affirmative advance consent in ordinary consciousness, is confined to a narrow anatomical range, and is refusable at any moment. Altered states create suggestibility that leaves the client unable to protect their own interests, so the practitioner bears the entire burden of the boundary. Two practitioners, or recording, reduce abuse risk. Dual relationships are prohibited.

The specifics are drafted and in force. Oregon OAR 333-333-5120 limits supportive touch to hugs or hands, feet, and shoulders, with prior written consent. Colorado 4 CCR 755-1-6 requires a written physical touch contract executed in a preparation session. The JAMA consensus, twenty-seven panelists, holds that a patient "should always be able to refuse supportive touch during a session, regardless of capacity at the time of refusal." Oregon built its own code of conduct from the counseling association, psychological association, and MAPS codes. A state regulator lifted enforceable rules from a voluntary field code, which is the clearest proof the field's ethics are fit to codify.

And now the other half. Roughly a dozen concordant ethics documents, and exactly two enforcement bodies, both state regulators created by ballot measure. The most-cited pledge publishes no signatory roster and no removal procedure. Its successor describes accountability as something that "emerges organically from peer relationships," which is a description of having no mechanism. The practice guidelines filed into the agency's own docket say in their own text that "there is currently no requirement that practitioners complete such programs, nor is there an accreditation mechanism to set standards." Those guidelines expired by their own terms in August 2025, were never revised, and the association that wrote them appears, from its final tax filing, to have ceased operations in 2024.

The case that proves it. A participant in a MAPS-sponsored Phase 2 trial was treated on Cortes Island by a married co-therapist pair. Session video shows pinning, cuddling, blindfolding. The sponsor did not review the videos for six years. One therapist later had sexual contact with her while she remained enrolled. He held no license, so the provincial college had no jurisdiction and could only ask him to stop using the title. The licensed co-therapist's complaint sat three years. The civil suit settled confidentially, sealing the record. The sponsor's sanction was a private blacklist. That was inside a federally regulated trial, with video, the most surveilled setting this field has.

The enforcement record in the legal states is administrative rather than ethical. Oregon's own 2025 summary records 51 complaints, 60 investigations, and 22 final orders, 18 of them civil penalties and four license revocations. The summary does not say whether any revocation concerned a boundary violation, and I could not find a report of one. Colorado publishes a discipline roster and no summary. The deficiency is in enforcement, not in content.

5. Consent: the problem this class creates

The literature has converged on dynamic consent, ongoing communication that lets a patient change their decision during care rather than at a single signing, because these compounds increase suggestibility and place the practitioner in a position of undue influence over a person who cannot advocate for themselves. A second problem has no clean solution: a patient cannot fully know what they are consenting to, because the experience may alter the values by which they would judge it. The practice guidelines already require consent in writing, in ordinary consciousness, revisited at multiple points. Colorado's touch contract is the nearest thing in force to a template. What does not exist is a field-endorsed informed consent template, as distinct from the state forms. That is a blank a practitioner's proposal can offer to help fill.

6. Integration: named by everyone, defined once, owned by nobody

This is the heart of the matter, and where this document must be most careful, because the strongest single piece of evidence points the wrong way. The concession comes first.

The evidence does not support a claim that integration improves clinical outcomes. A 2025 meta-analysis in General Hospital Psychiatry found no association between therapy hours and outcomes across sixteen studies, including integration hours analyzed separately (b = −0.11, p = .170), across a range of 4.5 to 18 hours, with the floor never tested. A 2026 scoping review of 44 articles found no quantitative evidence that integration improves outcomes and named it a critical gap. The practice guidelines say so themselves: the guideline "intentionally does not define the frequency of contact ... given the inadequacy of existing data to define a best practice." This document therefore does not propose any cadence as clinically optimal.

What the trials did. The pivotal MDMA protocol, NCT03485287: a 90-minute integrative session the morning after, four phone check-ins over seven days, further sessions at two and four weeks, and the Columbia Suicide Severity Rating Scale at scheduled visits. Across trials the median is two sessions per dose, the first at 24 to 48 hours.

What the states require. Oregon OAR 333-333-5260: a duty to offer, with no timing, no minimum, no content, and no documentation of outcome, plus a nondirective mandate that arguably forecloses directed safety inquiry. Colorado is materially stronger: rule 6.19(B) specifies content including evaluation of reactions with referral, and 6.19(A) prohibits a separate fee for the first integration session. New Mexico's Medical Psilocybin Act, which requires a working program by 31 December 2027, goes furthest in framing, making preparation, administration, and integration definitional elements of the service itself, but the proposed rule sets no content, timing, or count for patient care. The only fully specified three-phase protocol in the state is the University of New Mexico group trial under an IND, with six integration sessions and peer-led follow-up at four, six, and ten weeks. Its lead author sits on the state advisory board, so it is the likely template. It is still a trial design, not a rule.

Ascertainment is what the evidence does support, and a regulator should find that more persuasive. Korthuis and colleagues at Oregon Health & Science University followed 346 adults through 24 of Oregon's 26 licensed centers and published in JAMA Network Open this August. Retention 93.1 percent at one week, 92.8 at one month, 90.2 at three months. Four serious behavioral reactions, 1.2 percent, all psychedelic-naive, all with baseline psychiatric conditions, all at 25 to 50 mg, and three of the four surfaced at follow-up rather than in the session. Facilitators, under Oregon's mandatory reporting regime, reported one of the four. Six of seven participants with facilitator-reported reactions themselves reported nothing at any timepoint, so neither party to the session is a sufficient reporter. Persistently worsening depression ran 3.0, 4.4, and 4.5 percent at one week, one month, and three months. New thoughts of dying or suicide ran 1.0, 0.6, and 1.9 percent. Depression is still rising at the end of observation, and suicidal thoughts are highest at the last point measured. Persistent perceptual disturbance was reported by 10.0 percent at one month and 8.7 at three, figures that passive reporting does not capture. The authors' own conclusion is that state mandates limited to three days "underestimate actual safety events over time."

The instrument, not the population, produces the number. Oregon's reportable reaction is one within 72 hours of release that led to emergency services or medical care, on a form that states on its face, "This form is not submitted to OPS unless requested." The state's 2025 data shows 2.42 adverse behavioral reactions and 2.79 medical per 1,000 sessions across 5,935 clients. The research cohort in the same state found 1.2 percent serious. The comparison is imperfect, per participant against per session, serious against all, and an enrollment window against a calendar year. Allowing for all of that, the order of magnitude is the point. Same state, same period, different instrument.

The long tail nobody measures. Evans and colleagues surveyed 608 people reporting difficulties beyond 24 hours, in PLOS ONE. About one third persisted over a year, about one sixth beyond three. The duration distribution is bimodal, peaking under one week and over one year with a dip around one month. That shape is an empirical argument about placement even where the evidence cannot support a number: a single check-in at one month sits at the exact trough of detection. The sample is self-selected, so it gives no incidence rate. That is itself a finding. No population-based incidence estimate for prolonged post-psychedelic difficulty exists, in any setting.

The natural experiment already running. Ketamine has been federally legal since 1970 and is the one medicine of this class available nationwide. The field's 2026 Delphi consensus recommends "some form of psychological aftercare" and sets its monitoring standard at 30 to 60 minutes. A recommendation for the months, a standard for the half hour. A 2025 nationwide survey found over 40 percent of clinics dispensing for at-home use. No published figure exists for the proportion offering integration. My own count, from clinic partners, is fewer than one in six, and I label it as mine.

Who pays. The CPT Category III codes 0820T, 0821T, and 0822T all describe monitoring during administration. There is no psychedelic-specific integration code, and as Category III codes they carry no relative value units.

7. Measurement: three consensus documents, zero adopted

The most solvable problem in the survey, because the work is published and nobody has picked it up.

Service and outcome. The OPEN consensus measures, a three-round e-Delphi with 36 experts spanning trials, ceremonial and Indigenous settings, and state services: 39 core measures, including total dose administered, documentation of touch and sexual boundaries, respect for boundaries as a client-reported outcome, and a back-up facilitator and medical referral pathway. It recommends a brief safety checklist to the facilitator within one day and to the client at one week.

Setting. The ReSPCT guidelines in Nature Medicine: 30 variables across physical environment, session procedure, therapeutic framework, and subjective experience. Sixty-eight percent of the panel held that setting spans recruitment through follow-up.

Harm. Palitsky and colleagues at Emory audited 25 instruments and found behavioral, relational, and spiritual domains inconsistently covered or absent. Their framework adds sociocultural, psychospiritual, interpersonal, behavioral, and perceptual domains, and psychotherapy-related harm including exploitation, boundary violation, and over-dependence as a formal category. New-or-worsened from baseline, attribution assessed separately from detection, symptom level rather than diagnosis level, 54 defined terms, assessment to six months, and informant reporting from family. It also warns against the "no bad trips" framing because it hands the patient the job of interpreting their own distress. That is the data dictionary the agency's September hearing asked for.

Why it matters now. Across 24 psilocybin trials, only six achieved high-quality side-effect reporting and only two had long-term follow-up. Two Oregon studies asking about persistent perceptual disturbance with different instruments got 10.0 and 2.3 percent. That gap is instrumentation, not biology. And the states are diverging: New Mexico's proposed rule requires adverse event reports within two calendar days, including whether the safety plan functioned, while collecting no demographic or veteran-status data at all, and Oregon collects quarterly demographics on a public dashboard. Complementary, and a federal element could require both.

Cost decides who can comply. A complete battery is available for zero dollars: PHQ-9, whose footer reads "No permission required to reproduce, translate, display or distribute," GAD-7, the PCL-5 from the Department of Veterans Affairs, the WHO-5, the MEQ-30, the Challenging Experience Questionnaire, the Emotional Breakthrough Inventory, the Watts Connectedness Scale, and PROMIS short forms. The paid alternatives cost about $3.40 per administration or are licensed by participant count. A per-patient license fee is a per-patient access barrier. If the agency names instruments, it should name free ones.

8. What the field does not have

Stated plainly, because a reviewer will find it anyway. No accredited practitioner credential: the board founded in 2022 has a blueprint and, three years past its announced launch, no examination and no certificant. No accreditation of training: the guidelines expired and no accrediting body can be shown to operate, and the agency should verify before relying on either claim. No evidence that integration works. No removal mechanism that reaches a person: two state regulators, a sponsor's private blacklist, a college that can ask a man to stop using a title. No consensus instrument in use: three exist, none is required anywhere.

Part IV · What one practice does

A practitioner asking a regulator to require something should be willing to show what it looks like when a person does it. The following is the arc I run.

I offer it as one worked example rather than a model. It is not validated, has not been compared against anything, and its outcomes have been examined by nobody but me and my clients. What it demonstrates is feasibility. Every burden proposed below is a burden I already carry, which is the only standing I have to propose it.

Screening, and the no. Every inquiry is screened before anything is scheduled: personal and family history of psychosis and bipolar disorder, current medications including any taper, cardiovascular history, and whether anyone in the person's life will be there afterward. I turn people away, and the reasons are written down. That accumulated judgment, across the field, is the discernment map of the whole practice, and it is recorded almost nowhere.
Preparation. Multiple sessions before anything is taken, covering intention, history, what can happen, what will not be promised, and the specific matter of touch. Consent in ordinary consciousness, in writing, revisited rather than signed once, with a pre-agreed way to stop.
The arc is measured. Eight assessments: baseline, readiness, the acute experience within hours, insight and breakthrough close to the session, then repeated symptom, well-being, and connectedness measures across the months. Intake shapes preparation, session data shapes integration, and what I learn about my own blind spots is how a practitioner improves in a craft with no supervisor in the room.
The session. I remain present for the full duration.
After. Contact within seventy-two hours. Again at two weeks. Then monthly through six months, reachable between. I have run that cadence for years with real people and know what it costs, which is the only reason I can ask for a version of it.
The record. Everything lives in a HIPAA-compliant environment on software built for the purpose. Each session produces a plan and what I call an Inner Snapshot, a rendering of a client's inner world they can read months later. The record belongs to the person who lived the hours, and I guard it accordingly, which is why Element 9 is written the way it is.
Accountability. Peer consultation, and the willingness to be told something about myself and stay workable.

What I am building, and why I name it. Most clinics that tried to staff integration found a salaried therapist cost more than uneven demand could carry, and the weeks after went back to the patient alone. No agency chose that. The architecture produced it. I am building an outsourced integration and harm reduction layer clinics can contract instead of staff, on the cadence above, delivered by coaches, peers with lived experience, and licensed therapists under common training, with peer consultation and a misconduct process built in.

I name my interest plainly. The role I am asking the agency to require is one practitioners are prepared to fill, in the open, under whatever rule the agency writes, and I stand to benefit if it is adopted. I will share the cadence, the consent language, and the instrument set with any registry or guidance effort at no cost. And every element below is written to be satisfiable by a solo practitioner with free instruments and a telephone, which is the test I would want applied to anyone else's self-interested proposal.

Part V · The proposal: ten elements

Each element names its statutory hook, its precedent, and the existing standard the agency could incorporate rather than draft.

Element 1 · Certify the setting, not a single profession

Hook: 355-1(f)(3)(B) and (C). Precedent: esketamine's certified settings.

A certified setting attests to its protocols, personnel, physical space, emergency plan, and record-keeping. That is how esketamine already works, and it is the mechanism least likely to foreclose care where clinicians are few. Certification should attach to what a setting can do rather than to the type of building, since Colorado already permits administration in a private residence, in hospice, and at home for a participant who is dying, and a standard written for hospitals will not reach those settings.

The standard already exists, and the agency already holds it. The Professional Practice Guidelines for Psychedelic-Assisted Therapy, twelve guidelines from licensure through screening, safe settings, monitoring during and after, and integration, were filed into the agency's own guidance docket as comment 0146. Two cautions. The guidelines expired in August 2025 and their authoring body appears to have dissolved, so incorporate the text rather than the body. And they specify no hours, no supervision, and no competency method, which is why Elements 3 and 4 exist.

Element 2 · Two monitors, tiered by capability, with eligible licenses named

Hook: 355-1(f)(3)(A) and (B). Precedent: the agency's own final guidance.

The guidance already requires two monitors with differentiated credential floors and a fifteen-minute physician standard. The ask is continuity plus breadth. Carry the structure from the trial into the approval, and name in the REMS which licenses may be certified at each tier, because an unstated eligibility list becomes an exclusion in practice.

Both tiers are drafted. Colorado's facilitator definitions distinguish a Facilitator from a Clinical Facilitator, who must hold licensure as a psychologist, social worker, counselor, physician, nurse practitioner, or physician assistant. New Mexico's proposed rule separates a certifying clinician, a licensed practitioner, and a facilitator who holds no professional license at all, qualifying through a department-approved practicum rather than a degree, as supervised support. A medical model with a lay tier in the room, which is the staffing reality of a six-hour session.

The pool the guidance drew is too narrow, and the people who told the agency so are worth naming. In a comment to the guidance docket dated 25 August 2023, Lieutenant General Martin Steele, Brett Waters, and David Esselman, with signatories from eighteen organizations including the Navy SEAL Foundation and fifteen individual experts, asked that palliative-care physicians, clinical pharmacists, and advanced practice nurses be eligible lead monitors, and that the assistant tier admit peer support specialists, veteran peer specialists in particular, without a bachelor's degree. The argument is (f)(2)(C) itself. An element that admits only psychiatrists will not survive its own rural-access test, and, as one nurse filed to this docket, twenty-six years of emergency and trauma nursing is capability whatever the billing code says.

Element 3 · Adopt the training structure the field converged on, by reference to the codes that enforce it

Hook: 355-1(f)(3)(A) and (B). Precedent: two state codes and every major program.

Require, as a condition of practitioner certification, didactic instruction at or above the state floors, 128 hours in Oregon and 150 in Colorado, across enumerated subject areas; 40 hours of supervised practicum with at least 30 at actual administration sessions; and ongoing consultation as a continuing obligation. Point at Oregon's 3050 and 3070 and Colorado's 755-1-3 and 755-1-4. The agency is not setting a number. It is ratifying one the field set twice, independently. If the agency wants a content list, Colorado's fourteen itemized subject areas with per-area hours supply one at no drafting cost, including 25 hours of ethics and 10 of boundaries and touch.

Two provisions complete it. Trainer qualification: at least one instructor and every practicum supervisor holding a current license to practice this work, on Oregon's and Colorado's model. Trainee experiential exposure: permit, cap, screen, and credit it, on Colorado's ten of thirty hours or Oregon's four. Settled practice the agency can adopt without entering a live ethical argument.

Element 4 · Assessed competency against a published instrument with a numeric threshold, and name the instrument

Hook: 355-1(f)(3)(A). Precedent: the sponsor's own trials, in the drug class under review.

Do not require completion of training. Require demonstrated adherence. Point at the MAPS adherence manual:itemized behavioral criteria per session type, independent review of recorded sessions, 75 percent agreement with a gold standard, ICC at or above 0.75, remediation capped at two rounds. Certify the raters as well as the practitioners, since the manual also specifies who may rate and how they qualify. That is what converts a REMS from an attendance record into a competence check, with a tool that is published, free, and already used at scale in the trials the agency is reviewing.

Named weakness. Fidelity to a manual is not competence. A practitioner can rate well and still be the wrong person in a room, and no instrument in this field has been validated against client harm, because the outcome data does not exist. This manual was also in use in the trials whose oversight the agency found inadequate, so it is demonstrably necessary and demonstrably not sufficient. The Complete Response Letter's findings are about what was done with the ratings, which is why this element requires the rater pathway and the escalation rather than the manual alone. It buys a floor, not an assurance, and the agency should say so when it writes it.

Element 5 · Use the post-approval window to require the supervised sitting that prohibition forbids

Hook: 355-1(f)(3)(A) and (B). Precedent: none federally. Novel.

Supervised hours with a Schedule I substance cannot lawfully be earned in most of the country, so the one requirement that most obviously separates competence from attendance is the one almost no program can impose. The workarounds are unequal to it: two legal states, trips abroad no regulator can verify, trial hours the consensus curriculum says do not prepare a therapist for practice, breathwork held not equivalent, or nothing at all, which is the common case.

The catch-22 binds trainees before lawful access exists. It does not bind practitioners after approval. Once a drug is distributed under a REMS, supervised sessions with real patients are lawful everywhere it is dispensed. A REMS makes every certified site a lawful practicum site. So require that before a practitioner serves as lead monitor independently, they complete a graded sequence inside certified sites: observe as assistant, then a minimum of five consecutive sessions as active therapist with a supervisor present and debriefing after each, then independent practice under continuing supervision. Point at the consensus curriculum for the shape, the Compass model's four supervised sessions, and New Mexico's proposed practicum of eighty hours on administration days with fourteen or more patients as proof the states are already writing it.

Write it as a post-certification condition, never a grandfathered prerequisite, and the catch-22 never binds.

Named weakness. This adds cost and time when the field will be shortest of practitioners, and the first supervisors will be people whose own hours came from trials or from the legal states. The honest version is a phase-in with a stated end date. And New Mexico's proposed rule shows how easy the drafting error is: it authorizes practitioners, facilitators, and healing-center staff to possess and administer the drug and is silent on students, while requiring students to conduct supervised sessions. If that silence survives adoption, nobody can complete a practicum. Any federal element should say in terms that a trainee under supervision at a certified site is a lawful participant.

Element 6 · Continuous presence, with a monitoring minimum indexed to dose

Hook: 355-1(f)(3)(E). Precedent: esketamine's two-hour floor and stability assessment.

Two hours is the right idea at the wrong scale for a compound whose effects run six. Point at OAR 333-333-5250, which indexes minimum session duration to dose, from one hour below 2.5 mg to six hours at 35 to 50 mg, with start and end times recorded, and at 5200, which defines continuous monitoring as visual and audio contact throughout. A drafted element, tested across thousands of sessions. Adapt the table, keep the stability assessment esketamine already requires, and permit group administration at a participant-to-monitor ratio the sponsor proposes and the agency reviews, since veterans heal in units and couples arrive in twos. New Mexico's proposed ratios, one practitioner per eight and one facilitator per two, govern supervised training rather than routine care and should be cited for what they are.

Named weakness. One New Mexico patient advocate has filed to this docket arguing the person in the room is itself the adverse event, and asking for an option to be alone with the medicine (comment 0102). That position deserves consideration before continuous presence is written as an absolute.

Element 7 · A documented discharge and transport plan, made before the session begins

Hook: 355-1(f)(3)(D). Precedent: esketamine's readiness-to-leave assessment, and the Complete Response Letter's own request for standardized discharge-readiness criteria.

Colorado's standards of practice require a documented discharge plan before an administration session begins and a transport plan before the participant leaves. Three of Oregon's four serious reactions surfaced at follow-up rather than in session. The danger is not concentrated in the hour the current instrument watches, and the planning has to happen before the patient is impaired.

Element 8 · A documented safety contact at two defined intervals, using a validated instrument

Hook: 355-1(f)(3)(D) and (E). Precedent: the pivotal protocol the agency already reviewed. The element as such is novel.

The proposal. Require the certified setting to complete and document a structured safety contact:

at 24 to 72 hours after discharge, and again at 6 to 8 weeks after administration
each administering, at minimum, a validated suicidality instrument and a validated symptom measure for the indication
in person, by telehealth, or by telephone
with non-completion after documented attempts recorded as a reportable data element
with findings above threshold triggering a defined referral pathway

Scope. This is not a therapy requirement. The word integration arguably should not appear in the element, because it is a contested clinical construct and naming it invites the objection that the agency is regulating psychotherapy. This is an ascertainment requirement: the minimum instrument capable of detecting the adverse events this class produces.

The argument, in the order a regulator will accept it. The approval's safety database was generated under a protocol with structured post-dose contact and a suicidality instrument at scheduled visits, so marketing without any such contact means delivering the drug under conditions materially different from those in which its safety was characterized. A REMS exists to preserve the conditions of safe use. Without scheduled contact, delayed-onset events have no reporting pathway: a patient whose depression worsens at ten weeks does not attribute it to a dose in month one, and has no one to tell. The one natural experiment shows passive reporting missing three of four serious events. And in the Oregon research cohort, 90.6 percent attended at least one integration session. Those were volunteers who consented to follow-up, so expect the selection objection, and still: a mandate formalizes what nine in ten of them already did. The marginal burden is documentation and a free instrument.

On the intervals. The early anchor matches what the trials did. The late anchor targets the window where the Oregon figures for worsening depression and suicidal thoughts are at their highest, and sits after the one-month trough in the bimodal distribution of extended difficulties, so a single late contact is not placed at the point of minimum detection.

Who would staff it. Fireside Project's peer-support line reports 40,000 conversations and 700 trained volunteers on its own site, and its peer-reviewed evaluation with UCSF and Mount Sinai co-authors found 65.9 percent of 848 surveyed callers reported de-escalation from distress. Heroic Hearts runs three coaching sessions before and three after for veterans. New Mexico's proposed unlicensed facilitator tier shows a state regulator admitting the same workforce to the room. The workforce for a telephone contact with a free instrument exists, and it is not psychiatrists.

Named weaknesses, in full, because this is the element most likely to be challenged. The intervals are chosen by convergence and inference. No study has compared schedules or shown that contact at these times prevents harm. Directed safety screening is not nondirective, so the element collides with Oregon's mandate and may require the assessment to be separable from any integration encounter, which raises contacts and cost. Requiring an attempt on a patient who has disengaged raises autonomy problems this document does not resolve. And the strongest objection deserves full strength: this would be the first REMS element in history mandating a psychosocial encounter, and there is no controlled evidence that the encounter mitigates any harm. Detection is not mitigation. The correct instrument for an unanswered question is a study, not a condition on every patient's access. A mandated contact raises cost in a field with no integration code, the patients priced out are disproportionately the sickest, and the counterfactual to a somewhat-monitored patient is often an unmonitored one underground.

Where the objection lands. It is strongest against mandating integration and weakest against mandating a documented safety assessment at defined intervals using a validated instrument, which is not psychotherapy, is auditable, can be delivered by telephone, and mirrors surveillance the agency imposes routinely. This element therefore proposes the ascertainment requirement only.

And the settlement it represents. Laurel Kilgour of the Psychedelic Bar Association warned from the podium that an overly zealous risk framework pushes patients toward unregulated alternatives. Kevin Sabet of the Foundation for Drug Policy Solutions filed that "the threshold question is not how to implement psychedelic therapies safely; it is whether the current evidence justifies implementing them at all" (comment 0017), and asks for twelve-month follow-up before approval. The access advocate and the skeptic are pointing at the same hole from opposite sides: harm outside the monitored window, to people outside the monitored system. A two-contact ascertainment requirement, inexpensive enough not to displace patients into unregulated settings and long enough to observe the delayed signal, is a requirement both positions can accept.

One provision to make it real. Require that the mandated contacts be included in the cost of the certified episode and not separately billed, on Colorado's no-separate-fee model. Whether the agency may set payment terms through a REMS is doubtful, and this document flags the doubt. If the authority is not there, the ask belongs at the American Medical Association and the Centers for Medicare and Medicaid Services, where the missing integration code also lives.

Element 9 · A registry that is distributed rather than central, on the architecture the agency already operates

Hook: 355-1(f)(3)(F) and (E). Precedent: the agency's own Sentinel Initiative.

The nearest template on the agency's shelf is the esketamine registry, whose enrollment form collects first name, middle initial, last name, date of birth, email, phone, street address, city, state, ZIP, and sex, held by the manufacturer in a database of every outpatient who received the drug. For a Schedule III drug that is uncomfortable. For a Schedule I substance, where enrollment alone is adverse evidence, it is untenable.

Why enrollment is the exposure. In Wilson v. Lynch, 835 F.3d 1083 (9th Cir. 2016), a woman was refused a firearm because she held a state medical marijuana registry card. She said she had never used the substance. The Ninth Circuit held the card alone gave reasonable cause. The entry was the evidence. Beyond that: clearance guidelines treat Schedule I use as disqualifying regardless of state law, the privacy rule permits disclosure for national security and on an administrative subpoena with no judge, and immigration inadmissibility can attach to an admission without a conviction. Every veteran with a clearance, every parent in a custody proceeding, every green-card holder will run that arithmetic before walking in. A registry they cannot trust produces two harms: the patients it exposes, and the far larger number it drives into the space where nothing is measured. A privacy-protective design is therefore the precondition for safety surveillance, not a competitor to it.

The design, assembled from mechanisms the government already operates.

Distributed, not central. Copy Sentinel. A common data model plus a query-distribution protocol, not a database of patients. Each site holds its own records, the operator distributes analysis programs, and sites return aggregate results only. The agency's own principle is that it "does not receive or possess data with personally identifiable information," across 541.5 million patient identifiers as of October 2025. There is no argument it does not scale. Say in the REMS text that the esketamine model is foreclosed. ETASU (F) says registry, not centrally identified registry, and for a Schedule I therapeutic the manufacturer has no operational need for a home address.
Site-held pseudonyms. A site-generated identifier, the crosswalk kept by the site and nobody else.
A Certificate of Confidentiality as a condition of operation. The agency already issues them for FDA-regulated products under 42 U.S.C. 241(d). Protection runs in perpetuity, the information becomes immune from legal process, and it extends to vendor copies. Statutory novelty and cost: none.
Part 2 equivalence for every record regardless of indication. Import the operative protections of 42 CFR Part 2: the §2.12(d)(1) bar on using records to initiate or substantiate criminal charges, binding anyone who obtains them; §2.64 notice and hearing before a noncriminal order; §2.65's heightened standard for orders investigating a patient; §2.67's bar on undercover agents. Otherwise confidentiality turns on diagnosis: a patient treated for alcohol use disorder gets the strongest protection in federal law and a patient treated for PTSD gets the weakest, for no principled reason.
Minimization, no secondary use, real opt-out. No name, address, email, phone, or full date of birth. Age band and state, with small-cell suppression. No field enters the model without a written statement of the safety question it answers. State affirmatively that registry data may not be used for employment, licensure, clearance, immigration, custody, firearms eligibility, or underwriting, and may not be matched against any law-enforcement or benefits database. And an opt-out that does not gate treatment, because a REMS that conditions the medicine on surveillance consent will be evaded. The reporting duty sits on the site, which is licensed, inspectable, and has no clearance to lose.
The fields, because three consensus documents already specify them. Adopt the 39 OPEN measures, the 30 ReSPCT variables, and the 54 Palitsky terms, with attribution assessed separately from detection. Name free instruments. Take the one-week and one-month points from the two Element 8 contacts, then follow at three, six, and twelve months by patient-reported instrument sent remotely, with an extended-difficulty screen at 24 and 36 months. Those later points are questionnaires, not visits, and the burden sits on the registry operator rather than the setting. Twelve months is the agency's own pre-approval standard, and Oregon held 90 percent retention at three months. And record what cannot be added later: every baseline, every acute-session measure, the screening denominator including everyone turned away and why, per-patient dose and product, the concomitant-medication taper timeline, and prior psychedelic experience.
Sunset. Reassess every field and every element on a schedule, with the presumption at each review being removal unless retention is justified. The iPLEDGE experience shows the alternative.

Named weaknesses. ETASU (F) says each patient is enrolled, and an opt-out sits in tension with that text. The resolution proposed here is that enrollment of the treatment episode at the site is pseudonymous and mandatory, which keeps the denominator, while the patient's opt-out governs the longitudinal follow-up and any linkage. And the third-party use bars reach further than a REMS can. The sponsor and its sites can be bound. A court cannot. The Certificate of Confidentiality does that work, and Part VII says so.

Element 10 · Spiritual care available on request, drawn from a credential that already exists

Hook: 355-1(f)(3)(B), what a certified setting must be able to provide. Precedent: the federal government's, not the agency's.

The hearing notice ruled religious, ceremonial, and personal use out of scope. The exclusion is sound as a statement of jurisdiction, and this proposal accepts it. It does not describe what occurs in a supervised session, and this element addresses that gap without asking the agency to rule on any belief.

The ask is narrow on purpose. A certified setting must be able to make a board-certified chaplain or equivalent available at the patient's request, and must document that the offer was made. Available, not assigned. A patient who wants nothing of the kind receives nothing of the kind.

That standard is more rigorous than anything this field has built. Board Certified Chaplain status requires a qualifying master's degree, four units of clinical pastoral education at 400 hours each, faith-group endorsement, 2,000 hours of experience, and 31 competencies demonstrated before a live peer committee, re-examined every five years. New Mexico's proposed curriculum already requires a module on the spiritual, existential, religious, and theological aspects of psilocybin therapy, so a state regulator has named the subject even where no state has named the profession. The federal government already relies on it: Medicare's hospice benefit has required spiritual counseling on the team since it began in 1982, now at 42 CFR 418.64, and the Veterans Health Administration pairs chaplains with clinicians in moral injury groups. Neither asked a regulator to rule on a church.

The reason to ask. In my own practice, a substantial share of clients who stated no spiritual interest at intake described the experience afterward in spiritual or existential terms. The trials measure the same phenomenon with validated instruments. The agency is not being asked to endorse a belief. It is being asked to recognize a common clinical event, whose least prepared subjects are the patients who enrolled for symptom relief. A patient with no framework for that experience in the weeks after is a safety concern, not a religious one. My comment to the docket states my own position on this in the first person; this document keeps to the regulatory ask.

My interest, disclosed. I am ordained, have founded and served in psychedelic churches, and hold no chaplaincy credential. I am considering the training. If the agency writes this element, I may one day be among those a certified setting admits under it, and the reader should weigh the ask with that in mind, as with the disclosure in Part IV.

Named weakness. The chaplaincy profession has not been consulted on this, by me or by anyone in this field so far as I can find. This element needs a board-certified chaplain to tell me what I have got wrong before it is worth filing as written.

Part VI · The burden test, run on ourselves

Under 355-1(f)(2)(C) and (D), every element answers to access and to the delivery system. Running the test on your own proposal before a reviewer does is the minimum courtesy.

Element 1, certified setting. Access burden low; certification attaches to the site rather than to a scarce specialty. Delivery-system burden: conforms to the esketamine model already running.
Element 2, two tiered monitors. Access burden moderate, and lower than the alternative; naming licenses broadly is what protects rural access. Delivery system: mirrors the agency's guidance and two states' tiers.
Element 3, state-floor training with a 40/30 practicum. Access burden real, and front-loaded on the practitioner; the risk is workforce supply in the first two years. Delivery system: already law in two states and the market standard.
Element 4, assessed competency. Access burden low for patients, real for practitioners and for the rater pool, which does not exist at scale. Delivery system: uses a free instrument already in the pivotal trials.
Element 5, supervised sitting. The heaviest element, and the one most likely to bind supply in year one; phase it, with an end date. Delivery system: novel, and possible only because approval makes the site lawful.
Element 6, dose-indexed monitoring. Access burden real; a six-hour session needs six hours of staffing, which is the honest cost of the drug rather than a regulatory addition. Delivery system: adapted from a table in force in Oregon.
Element 7, discharge and transport plan. Access burden low; paperwork already routine in procedural sedation. Delivery system: Colorado requires it today.
Element 8, two safety contacts. Access burden low for the patient, real for the setting; two contacts, not a course of care. Delivery system: novel as a REMS element, and the no-separate-fee provision may exceed the agency's authority.
Element 9, distributed registry. Negative access burden; it removes an access barrier the central alternative creates. Delivery system: copies a system the agency runs across 541 million patient identifiers.
Element 10, chaplain on request. No access burden; declining costs nothing. Delivery system: uses an existing credential and an existing federal precedent.

Where this proposal is most vulnerable. Elements 3, 4, and 5 together are a real workforce constraint in the first years, and the patients who will feel it are in the places (f)(2)(C) names. The honest response is to phase the standard rather than weaken it, widen the eligible licenses as far as safety allows, and say plainly that a two-year squeeze is a cost traded against a class of harm the field has shown it cannot detect or discipline on its own. Reasonable people will weigh that trade differently.

Part VII · What is not being asked for, and two objections answered

Precision about the ask is the argument.

Not a class-wide templated REMS. Lawrence Goldkind of the Uniformed Services University argued from the podium against a single REMS for the class, because the compounds differ. He is right about the compounds, and the reply is this: the duration of vulnerability varies by compound. The existence of a vulnerable period does not. So the standard should be a common obligation with compound-specific windows. The elements above are a menu with stated hooks, not a form.
Not a loosening. Edward Jacobs of Johns Hopkins told the agency that real-world risk is more likely to be underestimated than overestimated, and that it is better to start with stronger protections and adjust as evidence accumulates than to tighten after harm. He deserves a yes or no. Yes. Every element here that binds supply carries a phase-in or a sunset, and the registry carries both, which is the clozapine lesson: the agency has shown it will loosen when the evidence arrives.
Not a psychotherapy mandate in the label. These elements condition the setting and the safe-use record, not the content of anyone's clinical hour.
Not a single-profession gate. See (f)(2)(C), and Element 2.
Not a registry that functions as a list. See Element 9, which is written as the argument against one.
Not a ruling on ceremony, church, or sacrament. Those sit outside the agency's jurisdiction, and this proposal accepts the line the notice drew.
Not a chaplain assigned to a patient who did not ask.
Not anything requiring new statutory authority. Every element is drawn from the six already in 355-1(f)(3), with two flagged exceptions. The no-separate-fee provision in Element 8 may be a payment term the agency cannot set. And the use bars in Element 9 bind the sponsor, its vendors, and its sites, which a REMS reaches, and cannot bind a court, an employer, or an immigration officer, which it does not. Those belong in the Certificate of Confidentiality and, in the end, in statute. The document asks for them anyway, so the agency sees the whole shape of what a trustworthy registry needs.

Part VIII · Four asks beyond the REMS

A REMS is one instrument. Four things this proposal wants sit outside it and belong on the record anyway, so the agency is not asked to do through a REMS what it should do through something else.

The label. A REMS binds certified settings. The label reaches every clinician who ever prescribes and outlives any REMS, as clozapine demonstrated. State the administration setting, the observation period, and the discharge assessment in the label itself, under 21 CFR 201.57, and require a Medication Guide under 21 CFR Part 208. Patients should have a plain-language account of what the days after administration may involve.

Postmarketing requirements. Under section 505(o)(3), 21 U.S.C. 355(o)(3), the agency may require studies to assess signals of serious risk. The signal is in the published record: three of four serious reactions in Oregon surfaced after the session. Make durability and post-session events a requirement rather than a commitment. A requirement is published and reported to Congress. A commitment carries neither obligation. This is also the right home for the question Element 8 cannot answer: whether post-session contact reduces harm has never been tested, and the correct instrument for an untested question is a study.

State data. The agency publishes postmarketing studies satisfied with real-world evidence and has issued guidance on registries. Oregon runs the only mandatory psilocybin safety reporting stream in the country, quarterly since January 2025, and New Mexico proposes two-day adverse event reporting. The ask is not that the agency commit to using state data, but that it state what would make such data fit for purpose. Right now the states are treated as a parallel universe, and one of them is already collecting the safety data the federal system will wish it had.

What gets measured. A depression trial measures depression. What clients describe afterward is functioning, connection, the willingness to be in a room with their family. In the 2021 psilocybin and escitalopram comparison, psilocybin outperformed on well-being and on work and social functioning, all of it secondary and therefore underpowered. If the instruments cannot see it, the trials cannot show it, and the payers will never cover it. The agency runs a qualification process for clinical outcome assessments under section 507, 21 U.S.C. 357, and a requestor need not be a sponsor. So the ask comes with an offer: confirm that a practitioner and patient coalition is an eligible requestor, and say what a letter of intent would need to contain.

Part IX · The body that would measure it

A REMS can require a competency it cannot itself measure. The agency can say a practitioner must demonstrate adherence to a published standard. It cannot build the instrument, train the raters, convene the committee, or revoke anything. That belongs to a certifying body, and the one that intends to do it, the Board of Psychedelic Medicine and Therapies, has been working on its first examination since 2022 with no candidate yet certified. If the agency writes Element 4 by naming a competency standard, the body that certifies against that standard acquires the only real leverage it will ever have. That is a further reason to name the standard, and it is also the limit of what this proposal can do: the examination itself is that body's work, not the agency's and not mine.

Part X · What this proposal does not have

Stated once, in one place, so a reviewer does not have to assemble it.

No controlled evidence that the central novel element works. Element 8 mandates a contact whose harm-mitigation efficacy has never been tested. The argument is ascertainment, and it is only as good as the reader finds it.
No cost estimate. Elements 3, 4, 5, and 8 impose real costs and this document quantifies none of them. That is a deficiency in a filing that invokes (f)(2)(C).
No consultation with the chaplaincy profession, whose credential Element 10 leans on.
No accredited practitioner credential to point at, which is why Element 1 certifies settings against published guidelines rather than individuals against a credential.
No peer review. Nothing here has been through the review a sponsor's regulatory team and the agency's reviewers will give it.
Several citations need verification before filing, named rather than buried. New Mexico's 7.35.3 is a proposed rule with its hearing on 2 October 2026, and its adopted producer rule is not yet on the official state code server, so certified copies should be requested. Oregon's 128-hour figure is the 2025 revision, and the public mirror of the rule still shows 120, so cite the state's current text. The status of any training accreditation body could not be confirmed either way. Oral remarks from the September hearing rest on trade accounts until the transcript is posted, and the accounts disagree on affiliations. And the ketamine integration estimate is mine.

Part XI · Who this is addressed to

The agency. The elements, the hooks, and the burden test are written to be read by a reviewer. The statutory door is 355-1(f)(5). When the Drug Safety and Risk Management Advisory Committee takes up a psychedelic REMS, seat a patient who has been through a supervised session and a licensed state-program facilitator who has sat with strangers through hundreds of hours, as voting members. The charter provides for it.

This proposal reaches the agency as a comment to docket FDA-2026-N-7542, which the hearing notice says will be weighed equally with oral testimony. It is written so that it can also be presented in the open public hearing of the advisory committee meeting on the first application, under 21 CFR 14.29, and, if the first approved REMS omits the period after discharge, filed as a citizen petition under 21 CFR 10.30, which obliges the agency to respond in writing within 180 days. Petitions have changed a REMS before. FDA-2012-P-0818 produced class-wide opioid labeling changes the agency carried into the REMS, and FDA-2019-P-1534 shaped the mifepristone program. The agency has also reopened a REMS on public input without a petition: the iPLEDGE modifications after the 2023 advisory committee, and the clozapine release after patients told the 2024 committee the program impeded access.

The sponsors. A REMS is proposed by the applicant. Whoever files first writes the first draft of the room, and a practitioner's comment reaches them faster than it reaches a reviewer.

The states and the payers, reading over the agency's shoulder. The handoff is the hazard. The agency writes the conditions first, and the states, the insurers, and the Drug Enforcement Administration build on whatever shape is left them. A narrow certification becomes a narrow profession, then a narrow billing code, then a patient in a rural county with no eligible provider. Which is why a comment addressed to the agency is worth writing even about the things the agency cannot give.

Reference shelf

Statute and regulation

21 U.S.C. 355-1 · REMS authority, the six elements, the burden tests, the input obligation
21 CFR 201.57 and 21 CFR Part 208 · labeling and the Medication Guide
42 CFR 418.64 · hospice spiritual counseling since 1982
42 CFR Part 2 · confidentiality of substance use disorder records
45 CFR 164.512 · law enforcement and national security disclosures
42 U.S.C. 241(d) · Certificates of Confidentiality
Wilson v. Lynch, 835 F.3d 1083 (9th Cir. 2016)