Metanoia's Public Comment to the FDA: Psychedelic Data Standards That Count Healthy Adults

Docket No. FDA-2026-N-7542: Considerations for Potential Future Therapeutic Use of Psychedelic Drugs; Public Hearing; Request for Comments

Submitted to: Regulations.gov

Topic addressed: Topic 4, Best Practices for Data Collection and Standardization

Submitted by: Metanoia, a California nonprofit public charity (EIN 39-2328100), 2108 N St #5956, Sacramento, CA 95816-5712. Contact: main@metanoia-ca.org · +1 (209) 432-7113.

Scope of this comment

This comment addresses Topic 4 only. It does not address:

• the safety or effectiveness of any drug product or the merits of any pending or anticipated application
• the scheduling status of any substance under the Controlled Substances Act
• the legalization or decriminalization of psychedelic substances or the merits of any State or local program authorizing their use
• religious, ceremonial, or personal (non-medical) use of psychedelic substances
• or any individual dispute, enforcement matter, or complaint regarding a specific practitioner or entity

Non-medical use appears in this comment only as a category of data source, and only with respect to how records of such use should be structured. Where this comment describes an existing State program, it does so solely to characterize what that program's records contain, and it takes no position on the program's merits. We rely here on the Agency's own statement that, while it does not seek comment on the merits of State or local programs authorizing psychedelic use, "FDA welcomes input on data collection from such programs as described above," and on the Agency's stated interest in coordination across data sources, including State programs, "to characterize both medical and non-medical use."

One further point of position. The Agency's final guidance for industry, Psychedelic Drugs: Considerations for Clinical Investigations (July 2026), maps the clinical trial setting under an investigational new drug application. This comment concerns records that arise outside that setting. The guidance itself states that the Agency may weigh "the public health effects of the drug," including risks related to "nonmedical use" for "patients and the broader population," and it asks sponsors to review adverse events reported in the context of "recreational or unapproved use." The records that would feed both of those exercises are the records this comment is about.

Disclosure

Metanoia has no financial relationship with, and holds no financial interest in, any entity developing, manufacturing, or expecting to provide products or services related to psychedelic drugs or substances. Metanoia is a single entity with no parent, subsidiary, or affiliate. It does not provide, sell, or source any controlled substance, and does not provide therapy or medical services.

Metanoia's published long-term organizational plan contemplates the future formation of a separate for-profit research affiliate, limited in scope to contracted institutional research. No such entity has been formed, no jurisdiction has been selected, and no agreement with any such entity exists. We state this here because it is described on our public website, and we would rather disclose a contemplated structure than have the Agency learn of it from a source other than us. Should such an entity be formed, Metanoia will disclose it in any subsequent submission to this or any other docket.

We place this disclosure at the front rather than the back because the recommendations below concern how evidence about psychedelic drugs should be recorded, and a reader is entitled to know what interest the party proposing them holds before reading them.

Summary of recommendations

1. Specify a minimum common data element (CDE) set before repositories are built. Harmonization retrofitted onto repositories designed independently costs far more than specifying elements in advance, and some elements, notably anything recorded before exposure, cannot be recovered afterwards at all.

2. Add a prospectively recorded purpose of use element as the structural analogue of indication. Without it, linked real-world records cannot be separated reliably into medical and non-medical strata, which is the characterization the Agency has said it seeks.

3. Adopt a standard case definition and fixed observation windows for persisting adverse effects, together with a re-exposure censoring rule.

4. Require a data provenance block on every record: prospective versus retrospective capture, reporter type, and the retention denominator. With those three fields, records of differing evidentiary quality can be pooled without being conflated.

Each recommendation is a specification. None depends on any view about whether, where, or by whom psychedelic drugs should be used.

Why the sequence matters now

A note on the public hearing of September 14, 2026. Metanoia was among the speakers randomly selected from the requests received. The Agency stated at that hearing that selection was not based on the anticipated substance or viewpoint of a requestor's comments, and that selection to speak does not constitute endorsement by FDA or participating federal partners of any speaker, organization, viewpoint, product, position, or regulatory outcome. We record the proceedings here for one reason only: the sequence this comment asks for was asked for independently, the same afternoon, by presenters with whom we have no connection. Four of them, speaking respectively for a State task force, an independent regulatory consultancy, a real-world evidence company, and a territorial research institute, each made a version of the same request. One asked the Agency to publish a minimum element set first, observing that a registry produces volume rather than evidence, and that the notice contemplates coordinated registry networks without common data elements. One asked that the objective be not simply more data but compatible data, designed from the start to be interoperable, so that a State program becomes a reliable feed for a coordinated registry rather than a silo. One asked the government to convene the field now to agree on a minimal data collection floor, and named drug provenance among the points it should cover. One observed that integrating such fields now takes effort, but that re-engineering software networks and health registries after launch would be complex, expensive, and would delay safety monitoring. We do not restate their positions, and we do not speak for them. We record the convergence because it bears on sequence, which is the one thing this comment asks the Agency to settle now. (note 2)

Executive Order 14401 of April 18, 2026, which this notice cites as part of its background, directs the Secretary of Health and Human Services, through the Advanced Research Projects Agency for Health, to allocate at least $50 million from existing funds to support and partner with State governments that have enacted or are developing programs to advance psychedelic drugs for serious mental illnesses, "including through Federal funding, technical assistance, and data sharing," and directs HHS and FDA to collaborate with the Department of Veterans Affairs and the private sector "to increase clinical trial participation, data sharing, and real-world evidence generation regarding psychedelic drugs" (91 FR 21709, April 22, 2026). Since that Executive Order, FDA and the Department of Veterans Affairs have executed a memorandum of understanding, effective July 13, 2026, that frames the exchange of clinical trial data, safety information, and real-world data between them, names treatments "addressed in the two EOs (e.g., psychedelic drugs)" among the subjects such an exchange may support, and specifies no data element, no format, and no vocabulary (MOU 225-26-007). HHS has since announced, through ARPA-H, the EVIDENT initiative, funded at up to $139.4 million, which "will allocate at least $50 million to match state government investments in psychedelic research for populations with serious mental illness" and lists a shared data foundation among its four technical areas (ARPA-H, April 21, 2026). Federal money is therefore about to reach State programs, with data sharing named among its purposes and a shared data foundation already on the implementing agency's agenda. Common data elements specified before that build-out cost very little. Harmonization retrofitted afterwards is expensive, and the elements that must be recorded before an exposure occurs cannot be retrofitted at any price. This comment is addressed to that window.

The final guidance adds a second reason. It asks sponsors of drugs with a history of human use to conduct "a systematic review of the literature to identify AEs associated with recreational or unapproved use," to treat such events as "predefined adverse events of special interest," and to base an analysis of "the scope of abuse and associated harms" on "a systematic review of the published literature and other available epidemiological data sources." That review can only be as good as the records it draws on. We have not identified a shared case definition, a shared set of observation windows, or a shared denominator convention in use across those records, and estimates from different sources therefore cannot be compared. Recommendations 3 and 4 describe the minimum that would make the review the guidance calls for feasible.

The problem these recommendations address

The Agency proposes to create data repositories that draw on electronic health records, claims, registries, pharmacy data, adverse event reports, patient and clinician reported outcomes, surveillance systems, federally funded surveys, and State programs. The value of such a repository depends on whether records arriving from these sources can be pooled. At present they cannot, for three reasons documented in the peer reviewed literature.

We have not identified a core outcome set in use. A PRISMA-guided review of patient reported outcome measures used in clinical research on classical psychedelics (three databases, two independent reviewers, studies conducted in clinical settings from 1990 onward, with electronic searches run through May 22, 2023) examined 93 papers reporting 93 unique trials and identified 17 different rating scales in use. The authors call explicitly for the development of a core outcome set to enable comparison across compounds, participants, and settings, and they recommend that instruments be designed to assess the setting of an experience, the same data element the Agency lists under Topic 4 (Hovmand, Poulsen and Arnfred, 2024). Seventeen instruments that cannot be interchanged, across 93 trials, is the documented baseline as of that search date. Instruments introduced since can only have raised the count. A repository built on that base inherits the variance.

The available instruments measure the acute experience rather than the durable outcome. The five instruments used most often in the trials that review covered were the Five-Dimensional Altered States of Consciousness Questionnaire (54 of 93 trials), visual analog and Likert scales developed for individual trials (27), the Hallucinogen Rating Scale (23), the 30-item Mystical Experience Questionnaire (15), and the States of Consciousness Questionnaire (14). Each of them quantifies an altered state during or shortly after exposure. None measures what the Agency's longitudinal follow-up bullet contemplates. Instruments aimed at the weeks after exposure have begun to appear since that search. The Afterglow Inventory, validated in 2025, is one example, and it measures positive subacute effects rather than persisting adverse ones (Majić et al., 2025). That is welcome, and it leaves the problem where it was: records from different sources are comparable only when the sources share a set of elements, and we have not identified any such set in adopted use. The presence of trial-specific scales in second place in the review matters on its own: a measure built for one trial cannot, by construction, be pooled with any other.

Recorded purpose and measured outcome come from different frames. In what its authors describe as the largest prospective survey of naturalistic psilocybin use to date, motivations were captured before exposure, in a survey completed immediately prior to use (n = 1,802), and respondents could select more than one. They reported self-exploration (n = 1,461, 81.1%), mental health (n = 1,284, 71.3%), therapy (n = 863, 47.9%), creativity (n = 787, 43.7%), recreation (n = 680, 37.7%), productivity (n = 403, 22.4%), and physical health (n = 258, 14.3%) (Nayak et al., 2023). (note 1) Outcomes were then reported on measures of anxiety, depression, alcohol misuse, cognitive flexibility, emotion regulation, spiritual wellbeing, extraversion, neuroticism, and burnout.

Two things follow, and both are structural rather than critical of that study. First, when the recorded purpose of an exposure and the instrument used to measure its outcome come from different frames, the resulting record cannot be interpreted without knowing both, and most data sources currently capture neither in structured form. Second, and more consequential for a repository: in this cohort the same individuals endorsed self-exploration (81.1%) and mental health (71.3%), and nearly half also endorsed therapy (47.9%). Purpose in this population is neither binary nor single-valued. Any scheme that infers medical or non-medical status from a missing indication field will misclassify a large share of exactly this population. Recommendation 2 addresses that problem.

Two further features of the existing naturalistic evidence base bear directly on data quality.

Attrition is severe, and the choice of denominator is unstandardized. The Nayak cohort passed through six sequential surveys with the following sample sizes:

• 8,006 at consent
• 2,833 at the assessment two weeks before use
• 1,802 the day before
• 1,551 at 1 to 3 days after
• 1,182 at 2 to 4 weeks after
• and 657 at 2 to 3 months after

At 2 to 4 weeks, 1,055 of 1,182 respondents (89.3%) reported no persisting negative effects, leaving 127 who reported at least one. At 2 to 3 months, 609 of 657 (92.7%) reported none, leaving 48. Those counts are among the few prospective estimates available, and they are stable facts. The rate derived from them is not. The same 127 participants are 10.7% of the 1,182 who completed that window, 4.5% of the 2,833 person pre-exposure baseline, or 1.6% of the 8,006 who consented. The same 48 participants are 7.3%, 1.7%, or 0.6% on those three denominators. All three denominators are defensible, all three are present in the same study, and we have not identified a convention that governs the choice. A single count of persisting adverse effects can therefore enter a repository as a rate that differs by nearly a factor of seven. The study's authors identify response attrition as a limitation bearing on generalizability and bias.

Retrospective before-and-after designs remain common. A cross-sectional online survey completed by 2,510 adults, each reporting at least one lifetime psychedelic experience, asked participants to complete measures of depression, anxiety, and emotional well-being retrospectively for the periods before and after psychedelic exposure. Thirteen percent of that sample (n = 330) endorsed at least one harm from their use (Raison et al., 2022). Records produced by this design are informative, but they are not equivalent to prospectively captured records, and a repository that does not distinguish the two will silently mix them.

The two State programs whose rules we have read define adverse events on incompatible clocks. Oregon fixes a window around the session. Colorado starts its clock when the licensee is notified. Neither defines a persisting effect, and a repository that ingests both without recording which convention produced each count is pooling two different quantities.

The Agency lists State programs for psychedelic use among the sources a repository would coordinate across. Oregon's program operates today under published administrative rules that function, for this purpose, as a data dictionary, and those rules are worth reading before their output is ingested. We describe Oregon's rules in detail because that program has been evaluated prospectively and its reporting rules are the ones we have read line by line. We have since read the corresponding Colorado rules, and we refer to them below where they bear on a recommendation. We have not reviewed the rules of any other State program, and nothing here should be read as a statement about them.

Under the Oregon administrative rules currently in effect:

• an "adverse reaction" means an adverse behavioral or medical reaction "that occurred during an administration session and does not require transport to a hospital"
• an "adverse behavioral reaction" and an "adverse medical reaction" each mean a client reaction "that required contacting emergency services or receiving care from a medical care provider that occurred during an administration session"
• and a "severe adverse reaction" is one requiring transport to a hospital (OAR 333-333-1010(2), (3), (4), (85))

A further category, the "post-session reaction," extends the window to "a medical or behavioral reaction occurring within 72 hours of the client's release from an administration session that was likely related to psilocybin consumption and resulted in contacting emergency services or the client receiving care from a medical care provider" (OAR 333-333-1010(59)). Service centers report counts in each category to the Authority in aggregate, quarterly (OAR 333-333-4910, both rules last amended effective January 1, 2026).

Three properties follow. We state them descriptively. The definitions are administrable and auditable, they do not depend on any instrument, and the program does look beyond the session itself, which are real virtues.

The observation window closes 72 hours after release. An effect that first presents at two weeks or at three months, the intervals at which the prospective cohort described above measured persisting negative effects, falls outside every category in the rule. Records produced by this program can therefore contribute to the acute half of the Agency's bullet on "the acute and long-term effects of psychedelic drugs," and to a narrow margin beyond it, but not, by construction, to the long-term half.

That inference is now measurable rather than only logical. A prospective evaluation of this same program, published in August 2026, enrolled 346 clients of Oregon's regulated psilocybin services between November 2024 and March 2026 at 24 of the 26 service centers then holding an active license, with retention of 93.1% at one week, 92.8% at one month, and 90.2% at three months. Serious behavioral adverse reactions were recorded in connection with four of the 346 sessions (1.2%), and no serious medical adverse reactions were recorded. Of those four, one was reported by the facilitator during the administration session and resulted in transport to a hospital. The other three were identified only through participant follow-up after the session: at one week in two cases and at three months in the third, in each case through a participant's report of a health care or emergency department visit attributed to the experience. For two of those three, the facilitator recorded no adverse reaction during the session at all. The study's authors state that State mandates collect safety events occurring within three days of services, and conclude that a reporting mandate focused on the day of services may underestimate true adverse reaction rates (Korthuis et al., 2026).

We cite that finding for its structure and not for its magnitude, and we are explicit about what it does not establish. Four events in a single cohort is a small number. The study is a single-arm observational cohort and supports no causal inference. Its authors report that it reached approximately 5% of the people who received these services in the period, and identify selection bias and limited generalizability as limitations. The same study reports improvements on its mental health, well-being, and life satisfaction measures, and this comment takes no position on those findings. What the finding does establish is narrower and sufficient: in the State program whose prospective evaluation we have identified, the events visible inside the reporting window and the events visible only outside it stood at one to three, and the three were visible only because someone asked the participant weeks or months later. That is a measurement property of the window. It is not a judgement about the program, and it is not an argument for or against it.

The threshold is a care-contact threshold rather than a symptom threshold. What is counted, in every category including the post-session one, is an emergency services contact, a contact with a medical care provider, or a hospital transport. This yields a hard, linkable numerator that requires no rating scale, which is why we propose a healthcare contact element in Recommendation 1. It is a utilization count, and it cannot be interchanged with an incidence count derived from symptom reporting in the published literature.

The reported quantity is an aggregate count, and the denominator is separable but not attached. Clients may ask that their records be withheld from the data submitted to the Authority, and the number of clients who so request is itself a reported figure, which is more than most sources provide. Even so, a count transmitted without the denominator it belongs to reproduces, at the level of an entire State, the ambiguity documented above within a single study.

Colorado, the second State whose rules we have read, counts on a different clock. A facilitator there must report a life-threatening or otherwise serious adverse health event to the Director within 24 hours (4 CCR 755-1, Rule 6.23, current version effective December 5, 2025), and a licensed natural medicine business must report a possible adverse health event to the Division within two business days (1 CCR 213-1, Rule 3015, current version effective August 12, 2026). Neither period is measured from the session. The business rule runs from the moment the licensee is notified of the event, and the facilitator rule states a twenty-four-hour deadline without fixing any starting point at all. Neither is therefore an observation window, and the rules we read provide no scheduled post-session contact and no periodic report of events arising after a session. Colorado's event definition also reaches a symptom or an illness rather than only a contact with the health care system, so its counts and Oregon's are not interchangeable. We describe what the rules require and take no position on either program.

A presenter at the September 14 hearing, a licensed facilitator working in the Oregon program, raised a related point from the opposite direction: that the Federal government is asking for collaboration, data, and real-world evidence, while this proceeding expressly excludes remarks on religious, ceremonial, and personal non-medical use, and that Oregon, one of the main sources of that evidence, built a non-medical, supported-use model in which there is no diagnosis, no prescription, and no referral. (note 2) This comment takes no position on that tension, and does not ask the Agency to resolve it. We note only that it presents first as a records problem: the records that model produces have no indication field, and Recommendation 2 is the specification that allows such records to be ingested and stratified without anything being decided about the model that produced them.

We raise this for one reason. A repository that ingests counts from these State sources alongside incidence estimates from the naturalistic literature will be combining quantities that are defined differently, bounded differently in time, and denominated differently. The provenance block in Recommendation 4 and the case definition in Recommendation 3 exist so that such records can be pooled without being conflated. Metanoia submitted substantially the same recommendations to the Oregon Health Authority on September 17, 2026, in its rulemaking on OAR chapter 333, division 333.

None of this is an argument about the substances. It is an argument that the Agency's Topic 4 objectives (interoperability, data quality, longitudinal follow-up, and adverse event characterization) cannot be met on the current measurement base without specification.

One property of that specification is worth stating plainly, because it was audible on both sides of the September 14 hearing. A fixed case definition, fixed observation windows, and an explicit denominator are not instruments of either optimism or caution. They are what prevents an underestimate of harm and an overestimate of benefit alike. Presenters who disagreed sharply with one another about psychedelic drugs asked, in different words, for the same specification, and that is a property of specifications rather than a coincidence of that afternoon.

Recommendation 1: a minimum common data element set

We recommend that any Coordinated Registry Network or comparable repository require the following elements. Fields are grouped to match the Agency's own list (drug, dosage, indication, setting, patient characteristics).

Recommendation 2: purpose of use as the structural analogue of indication

The Agency's list of common data elements includes indication. Records originating from non-medical data sources have no indication field, because no indication was assigned. If those records enter a repository with that field empty, they become indistinguishable from records where the field was simply not collected, and the medical and non-medical stratification the Agency describes cannot be performed on the pooled data.

We recommend a two-part element in place of a single indication field.

(a) Clinical indication present: whether a clinical indication diagnosed by a licensed clinician was present at the time of exposure, recorded as yes, no, or unknown. This single field carries the medical versus non-medical distinction explicitly rather than by inference from a missing value.

(b) Purpose of use: coded to a closed vocabulary, recorded as one designated primary purpose plus any number of secondary purposes, with a free-text field retained for terms not yet in the vocabulary, and with the timing of capture recorded (before exposure or after exposure).

Multi-valued capture is required for the element to work at all. In the prospective cohort described above, the seven motivation categories drew 5,736 endorsements from 1,802 respondents, an average of about 3.2 motivations per person, and self-exploration, mental health, therapy, and creativity were each endorsed by more than 40% of that group. A single-valued purpose field applied to that population would force a coder to discard most of what was recorded, and the discarded values are the ones that carry the medical and non-medical signal.

A second and independent dataset, collected inside a licensed and supervised State program rather than in a naturalistic survey, reaches the same structure. The Oregon cohort described above recorded, at baseline and with multiple responses permitted, the reasons its 346 participants sought services: curiosity (257, 74.3%), to improve or change themselves (247, 71.4%), to treat mental health symptoms (223, 64.5%), to process trauma (178, 51.4%), connection with others, the Earth, or nature (166, 48.0%), spiritual reasons (101, 29.2%), professional burnout (69, 19.9%), to reduce substance use (47, 13.6%), physical health symptoms (45, 13.0%), coping with death (34, 9.8%), and fun, enjoyment, or celebration (31, 9.0%) (Korthuis et al., 2026). Those categories drew 1,398 endorsements from 346 participants, an average of about four per person, and the two most frequently endorsed were not clinical. The point is not the ranking, which is specific to that program and that period. The point is that the multi-valued structure is not an artifact of naturalistic self-selection: it appears just as clearly in a supervised, licensed, State-regulated setting, and a single-valued field would discard most of what those participants recorded there too.

A second State has already adopted an element of this shape. Colorado's facilitator rules require the collection of "Reasons the participant sought natural medicine services, including whether the participant had any diagnosed physical or behavioral health condition for which the participant sought natural medicine services," and a separate field recording "Whether the goal of facilitation was for clinical or experiential purposes (or both)" (4 CCR 755-1, Rule 6.22(C)(1)(e) and (i), current version effective December 5, 2025). Colorado's healing center rules require deidentified information on "the intention or desired outcome for the Administration Session, including whether the Administration Session is intended to treat physical or behavioral conditions, or is for wellness services" (1 CCR 213-1, Rule 8035(C)(5), current version effective August 12, 2026). At the statutory level, Colorado authorizes release from its public health database for "reporting on services, participant motivations, demographics, and outcomes" (Colo. Rev. Stat. 25-1.5-120.5(3)(c)(III)), and requires facilitators, on and after July 1, 2026, to provide the State health department with "information provided by a participant prior to and following the participant's administration session" (Colo. Rev. Stat. 12-170-108(5)(b)(V)). Those two statutory provisions are conditioned on funding, and we have not verified whether the funding has been received. We describe what the rules require and take no position on the program. What Colorado's rules do not supply is a closed vocabulary for the reasons, a field recording when the reason was captured, or any alignment with Oregon's differently worded element (OAR 333-333-4910(2)(b)). That is the gap Recommendation 2 is addressed to, and two States having reached the same element independently is the strongest argument we can offer that it belongs in a minimum set.

The Oregon study also establishes feasibility, as does the naturalistic cohort. Purpose was captured prospectively, in a web survey completed immediately before exposure, from 1,802 people, without clinical infrastructure. Prospective purpose capture at scale outside a clinical setting has already been done.

Timing is the substantive part of this recommendation. Purpose recorded after an exposure is subject to the same recall and consistency effects that make retrospective outcome measurement weaker than prospective measurement, and it can be revised in light of the outcome. Purpose recorded before exposure is a fixed reference point against which a later outcome can be assessed. Where a data source can capture anything before exposure, purpose should be among the earliest elements captured, because unlike almost every other element in the set it cannot be reconstructed later.

The Agency has already accepted this logic in the trial setting. The final guidance states that "an expectancy evaluation questionnaire (i.e., assessing subjects' expectations about potential drug effects) pre-randomization and at the end of treatment to improve data interpretability despite functional unblinding should be considered." Expectancy and purpose are different variables, and we do not suggest the guidance recommends capturing purpose of use. The principle it adopts is the one that matters here: a subjective context variable that is not captured before exposure leaves the resulting data harder to interpret. Recommendation 2 carries that principle from the trial record to the non-clinical record.

Metanoia is developing a closed vocabulary for this element and an accompanying outcome instrument referenced to it, for non-clinical exposure records. Both are intended for open publication as described below.

Recommendation 3: case definition and observation windows for persisting adverse effects

The Agency asks for approaches to defining, identifying, and reporting adverse events in the context of acute and long-term effects. We have not identified a standard case definition for a persisting adverse effect in use in the field, nor standard observation windows, nor a convention for handling re-exposure during follow-up. As a result, the small number of prospective estimates that exist, such as the counts described above, cannot be pooled with each other or compared across sources.

The same gap was described independently at the September 14 hearing. A clinical researcher recounted having documented, under the current framework, an adverse event whose content was an increase in a participant's well-being, and asked whether that is the kind of event the Agency intends to have captured. The presenter pointed to oncology's Common Terminology Criteria for Adverse Events as a precedent that, in the presenter's description, fixes not only the terms but the criteria for what counts as an adverse event and at what severity, and said that psychedelic medicine needs a glossary with standardized criteria as well as standardized terms. (note 2) We record that account rather than adopting any characterization of the oncology criteria as our own. The distinction drawn, between terms and criteria, is the distinction this recommendation turns on: a shared vocabulary without shared thresholds produces records that look comparable and are not. Recommendation 3 is a proposal for the criteria half.

The absence of a case definition is not only a cross-study problem. Within the naturalistic prospective cohort cited above, persisting negative effects at 2 to 4 weeks can be read as 10.7% (the composite: any of the four listed effects) or as 4.7% (n = 55 of 1,182, the largest single category, mood fluctuations). The four categories drew 144 endorsements from 127 participants, which confirms that they overlap and cannot be summed. Both readings are accurate. Nothing in the record marks which convention a downstream extractor should apply, and the two differ by more than a factor of two before any question of denominator arises. A repository that ingests one number per study without capturing the composite-versus-category convention is averaging incompatible quantities.

Re-exposure during follow-up is common enough to distort any of these estimates. In the same cohort, 26% of the participants remaining at the 2 to 4 week follow-up, and 50% of those remaining at 2 to 3 months, had taken at least one further dose of psilocybin by the time they answered. An effect recorded at three months may therefore follow a second or third exposure rather than the one being studied, and nothing in an unflagged record says which.

We recommend the following.

Case definition. We propose defining a persisting adverse effect as a new or worsened symptom present at or beyond 30 days after exposure and not present during the 30 days preceding exposure. The thresholds are proposals. What matters is that a definition be fixed in advance and shared across sources. Expected acute effects occurring within the pharmacological duration of the exposure are recorded separately and are not counted as persisting effects.

Fixed observation windows at 7, 30, 90, and 365 days after exposure, so that estimates from different sources refer to the same intervals. The 365-day window aligns with the final guidance, which describes "blinded long-term follow-up (typically 12 months)" as the most informative design for detecting durable effects and the need for retreatment. A non-clinical record that stops at 90 days cannot be compared with a trial record that runs to a year.

Severity and functional impact, graded on a stated scale and recorded separately from symptom presence.

Re-exposure censoring. Any re-exposure during a follow-up window is recorded with its date, and records with re-exposure before the window closes are flagged and analyzed separately. Without this rule, an effect attributed to one exposure may in fact follow a later one.

Denominator convention. Every reported rate states its denominator explicitly, either completers at that window or the enrolled cohort, and repositories should require both to be reportable.

Recommendation 4: structured setting and data provenance

Setting. The Agency lists setting among the common data elements it identifies under Topic 4, and the review cited above independently recommends that instruments be designed to assess it. That review defines setting as the culture, place, and situation in which consumption occurs, including the other individuals present (therapist, guides, peers), the decor, the activities performed, and the beliefs shared among the group. The structured fields we propose in Recommendation 1 decompose that definition into recordable values, with the physical environment category standing in for place and decor. They are not a new construct.

We have not identified a validated instrument for setting that has come into standard use. The review notes one developed specifically for ayahuasca sessions and calls for a broader equivalent, and an international Delphi consensus published in 2025, the ReSPCT guidelines, has since specified 30 setting variables that psychedelic clinical trials should report (Pronovost-Morgan et al., 2025). ReSPCT is a reporting standard for trials. The fields we propose are a recording schema for non-clinical records, and we intend to map them to the ReSPCT items rather than to compete with them. Until a validated instrument exists, we recommend that setting be captured as those structured fields rather than as free text. Structured capture of supervision, licensure status of any supervising person, group size, preparatory contact, and pre-exposure screening yields a usable variable immediately and does not require waiting for instrument development.

Feasibility is again established rather than assumed. The prospective cohort described above captured setting in structured form from 1,551 respondents 1 to 3 days after exposure:

• whether the person was alone (n = 667, 43.0%), with friends also using (n = 399, 25.7%), with a sober friend acting as a sitter (n = 255, 16.4%), with a shaman or guide (n = 23, 1.5%), with a guided group (n = 22, 1.4%), or with a therapist (n = 17, 1.1%)
• and whether the location was home (n = 1,081, 69.7%), outdoors in nature (n = 245, 15.8%), a religious or spiritual setting (n = 39, 2.5%), a concert or festival (n = 18, 1.2%), a party (n = 11, 0.7%), or another public place (n = 11, 0.7%)

In this cohort the categories implying a licensed clinician account for roughly 1% of exposures. That is why supervision presence, supervisor role, and supervisor licensure must be three separate fields. Collapsed into one, the distinction that matters most for risk characterization disappears.

Provenance. Three fields on every record:

• whether capture was prospective or retrospective
• whether the reporter was the subject, a supervising or attending person, or a clinician
• and the retention denominator for the window in question

The permitted values for the second field belong to the setting in which the record was created, since a State program names a facilitator where a survey names an observer, and what matters is that the field exist and be coded to a stated list. These are inexpensive to collect. Without them, records of materially different evidentiary weight are pooled indistinguishably.

The reporter type field is not a formality, and the Oregon cohort shows why. In that study, facilitator reports and participant reports of adverse event type and severity often diverged. Of seven participants for whom a facilitator recorded an adverse reaction of special interest during the administration session, six reported no challenges or adverse effects at any follow-up point. Of the four participants with serious adverse reactions, two had no facilitator-reported reaction during the session at all (Korthuis et al., 2026). The same exposure produces different records depending on who is reporting, and neither record is simply wrong. A repository that does not record which of them is speaking cannot detect that divergence, cannot reconcile it, and will present the two as though they were the same kind of observation.

We add one field to the provenance block for a reason that was raised at the hearing and that we had not specified: the jurisdiction in which a record was captured, and the primary language in which it was captured. A repository built on a sample frame that omits a jurisdiction cannot report that omission unless the field exists, and a record captured through translation is not evidentially identical to one captured in the respondent's first language. (note 3)

What Metanoia offers

Metanoia is a California 501(c)(3) public charity conducting research and public education on natural psychedelics in healthy adults. We are preparing (a) a draft minimum common data element set for non-clinical exposure records and (b) a longitudinal outcome instrument referenced to a prospectively recorded purpose of use.

Metanoia intends to publish both under the Creative Commons Attribution 4.0 International license, without fee or additional licensing restriction, and to make them available to the Agency, to State programs, to academic investigators, and to any registry operator wishing to use or adapt them. Neither will incorporate items drawn from any proprietary instrument. We would welcome the opportunity to contribute this material to any common data element development process the Agency or its federal partners convene, including work with State programs supported under Executive Order 14401.

Our disclosure appears at the front of this comment rather than here.

Notes

1. Metanoia presented oral remarks at the September 14, 2026 hearing and cited this figure. The figure as spoken was imprecise. The correct value is the one given here: 47.9%, n = 863 of 1,802 (Nayak et al., 2023). We correct it on our own initiative.

2. Characterizations of remarks made at the public hearing of September 14, 2026 are taken from the hearing webcast as transcribed by automated captioning. They have not been checked against the official transcript, which the Agency has indicated will be posted to the docket. For that reason they are given in paraphrase rather than as quotations, and no presenter is named here. Metanoia will identify the presenters and correct any characterization in a supplemental filing once the official transcript is available, if that would assist the Agency.

3. These elements were proposed by other presenters at the September 14, 2026 hearing, and we adopt them with attribution:

• the concurrent non-drug intervention element and its timing, by a presenter speaking on cochlear-implant rehabilitation
• the reproductive stage and hormonal status element, by a presenter speaking on women's health
• the jurisdiction and primary language elements, by a presenter speaking for a research institute in Puerto Rico
• and the concomitant psychoactive exposure element, by a presenter speaking for a State health department

Names are omitted for the reason given in note 2.

Sources

Literature

1. Hovmand OR, Poulsen ED, Arnfred S. Assessment of the acute subjective psychedelic experience: A review of patient-reported outcome measures in clinical research on classical psychedelics. J Psychopharmacol. 2024;38(1):19–32. Epub 2023 Nov 16. PMID 37969069. PMCID PMC10851631. doi:10.1177/02698811231200019

2. Nayak SM, Jackson H, Sepeda ND, et al. Naturalistic psilocybin use is associated with persisting improvements in mental health and wellbeing: results from a prospective, longitudinal survey. Front Psychiatry. 2023;14:1199642. PMID 37795509. PMCID PMC10545967. doi:10.3389/fpsyt.2023.1199642

3. Raison CL, Jain R, Penn AD, Cole SP, Jain S. Effects of Naturalistic Psychedelic Use on Depression, Anxiety, and Well-Being: Associations With Patterns of Use, Reported Harms, and Transformative Mental States. Front Psychiatry. 2022;13:831092. PMID 35370864. PMCID PMC8965278. doi:10.3389/fpsyt.2022.831092

4. Majić T, Schmidt TT, Gröticke A, et al. The Afterglow Inventory (AGI): Validation of a new instrument for measuring subacute effects of classic serotonergic psychedelics. J Psychopharmacol. 2025;39(5):474–488. PMID 40165350. PMCID PMC12099019. doi:10.1177/02698811251326937

5. Pronovost-Morgan C, Greenway KT, Roseman L, et al. An international Delphi consensus for reporting of setting in psychedelic clinical trials. Nat Med. 2025;31(7):2186–2195. PMID 40461819. PMCID PMC12283393. doi:10.1038/s41591-025-03685-9

6. Korthuis PT, Cook RR, Gregoire D, et al. Safety and Mental Health Outcomes of Oregon State-Regulated Psilocybin Services. JAMA Netw Open. 2026;9(8):e2630608. PMID 42616497. PMCID PMC13491119. doi:10.1001/jamanetworkopen.2026.30608

Regulatory sources

7. Food and Drug Administration. Considerations for Potential Future Therapeutic Use of Psychedelic Drugs; Public Hearing; Request for Comments. 91 FR 43095–43098 (July 14, 2026). FR Doc. 2026-14155. Docket No. FDA-2026-N-7542.

8. Executive Order 14401 of April 18, 2026, Accelerating Medical Treatments for Serious Mental Illness. 91 FR 21709–21711 (April 22, 2026).

9. Food and Drug Administration, Center for Drug Evaluation and Research. Psychedelic Drugs: Considerations for Clinical Investigations. Guidance for Industry. July 2026. Docket No. FDA-2023-D-1987. Notice of availability: FR Doc. 2026-14158 (July 14, 2026).

10. Oregon Administrative Rules, OAR 333-333-1010 (Definitions) and OAR 333-333-4910 (Required Reporting of Data), Oregon Health Authority, Public Health Division. Both last amended by PH 21-2025, effective January 1, 2026.

11. Advanced Research Projects Agency for Health. ARPA-H announces first research teams for $139 million initiative to transform behavioral health and bring new, more precise and effective treatments to Americans. April 21, 2026. arpa-h.gov.

12. Memorandum of Understanding between the United States Food and Drug Administration and the United States Department of Veterans Affairs. MOU 225-26-007, signed and effective July 13, 2026. fda.gov.

13. Colorado Department of Regulatory Agencies, Office of Natural Medicine Licensure. Natural Medicine Licensure Rules and Regulations, 4 CCR 755-1, Rules 6.22 and 6.23. Current version effective December 5, 2025.

14. Colorado Department of Revenue, Natural Medicine Division. Colorado Regulated Natural Medicine Rules, 1 CCR 213-1, Rules 3015 and 8035. Current version effective August 12, 2026.

15. Colo. Rev. Stat. 25-1.5-120.5 and 12-170-108, as added and amended by Senate Bill 25-297 (2025), Concerning the Implementation of Colorado's Natural Medicine Initiative, signed June 3, 2025.

Verification note. Article metadata for literature sources 1 to 3 was retrieved from PubMed and checked on August 13, August 16, and September 12, 2026. Metadata and abstracts for sources 4 and 5 were retrieved from PubMed on September 12, 2026. Participant counts, motivation and setting distributions, persisting-effect counts, and re-exposure proportions attributed to source 2, the review methodology, instrument frequencies, and definition of setting attributed to source 1, and the sample and harm figures attributed to source 3 were checked against the full text of those articles in PubMed Central on September 12, 2026. Source 6 was read in full in PubMed Central on September 15, 2026. The cohort size, retention figures, serious adverse reaction counts and their timing, the reasons participants gave for seeking services, the divergence between facilitator and participant reports, and the authors' statements about the scope of State reporting mandates were each checked against the full text on that date. Sources 7 and 8 were read in full in the Federal Register on August 17, 2026, and again on September 12, 2026. Source 9 was read in full on August 30 and September 12, 2026. The definitions quoted from source 10 were read in the Oregon Secretary of State's Administrative Rules database on August 17 and September 12, 2026. Source 11 was read on arpa-h.gov on September 12, 2026. Source 12 was read in full on fda.gov on September 22, 2026. Sources 13 and 14 were read in full on September 22, 2026 in the Code of Colorado Regulations published by the Colorado Secretary of State, in the versions that database marks as current, and Rule 6.22 was compared with the preceding version and found identical. Source 15 was read in the enrolled final act published by the Colorado General Assembly on the same date. The two Colorado statutory provisions cited in Recommendation 2 are conditioned by that act on the receipt of funding, and we have not verified whether the funding has been received or whether the database it establishes is operating. Where this comment states a percentage that does not itself appear in the cited article, that percentage is arithmetic performed on counts reported in the article, and the underlying counts are given alongside it so the calculation can be checked. Characterizations of the September 14, 2026 public hearing rest on the hearing webcast as transcribed by automated captioning, and carry the status set out in note 2.