> **SUPERSEDED — historical record only.** This document has been replaced by `PSILOCYBIN_INDEPENDENT_REPLICATION_PROTOCOL_V0.3.md`. It is retained because earlier releases link to it. Do not use it for review, extraction, or registration.

# Protocol v0.1 — Controlled 25 mg Psilocybin Safety in Major Depressive Disorder

**Linked estimand:** EST-AE-PSI-001  
**Date:** 23 July 2026  
**Status:** Draft frozen before full screening and extraction. Not registered in PROSPERO or OSF. No pooled estimate has been calculated.

## Review question

Among adults aged 21–65 with major depressive disorder who meet contemporary controlled-trial eligibility criteria, what are the 0–24-hour and 0–7-day risks of prespecified serious medical and severe psychiatric/behavioral outcomes after one 25 mg oral dose of synthetic psilocybin with preparation, continuous monitoring and post-session support, compared with placebo or a minimal-dose/inactive control delivered with comparable support?

## Eligibility

### Include

- Randomized controlled trials in adults with MDD or a clearly separable MDD subgroup.
- A single 25 mg oral synthetic-psilocybin arm.
- Placebo, inactive, active-placebo or minimal-dose control.
- Structured preparation and monitored dosing session.
- Safety observation covering at least 24 hours; 7-day data extracted when available.
- Full publication, registry results, regulatory report or sponsor report with arm-level safety data.

### Exclude from the primary direct stratum

- Unsupervised or natural-mushroom exposure.
- Doses other than 25 mg when the 25 mg arm cannot be separated.
- Open-label studies without a control arm.
- Populations without MDD, unless used only for a prespecified indirect safety appendix.
- Repeated-dose regimens when first-dose outcomes cannot be separated.

Treatment-resistant depression is retained in an **indirect population stratum**, not silently pooled with the primary MDD population.

## Outcomes

Report each outcome separately; do not combine them into a single harm score.

1. Serious adverse event under the study/regulatory definition.
2. Death.
3. Hospital or emergency transfer.
4. ICU-level care or organ support.
5. Severe anxiety, panic, psychotic or behavioral event requiring rescue medication, extended observation, restraint, transfer or hospitalization.
6. Self-harm, suicide attempt, or sustained clinically important worsening.
7. Rescue medication and extended observation as process outcomes, even when not classified as serious adverse events.

## Search sources

- MEDLINE/PubMed
- Embase
- PsycINFO
- Cochrane CENTRAL
- ClinicalTrials.gov
- WHO ICTRP
- Reference lists and forward citation searching
- Regulatory/sponsor documents when publicly available

No language restriction at search stage. Searches begin in 2018 and are updated through the final search date.

## Draft PubMed strategy

```text
((psilocybin[MeSH Terms] OR psilocybin[Title/Abstract] OR COMP360[Title/Abstract])
AND
("depressive disorder, major"[MeSH Terms] OR major depressive disorder[Title/Abstract]
 OR MDD[Title/Abstract] OR treatment-resistant depression[Title/Abstract])
AND
(randomized controlled trial[Publication Type] OR random*[Title/Abstract]
 OR placebo[Title/Abstract] OR controlled[Title/Abstract]))
AND (2018:3000[pdat])
```

The final strategies must be peer-reviewed by an information specialist and reported under PRISMA-S.

## Screening

Two reviewers independently screen titles/abstracts and full texts using `psilocybin-screening-form.csv`. Disagreements are resolved by discussion or a third reviewer. Exclusion reasons are retained for the PRISMA flow diagram.

## Data extraction

Two reviewers independently extract:

- eligibility and baseline risk;
- exact formulation and dose;
- preparation, monitoring and post-session support;
- comparator and blinding procedures;
- randomized/dosed/analyzed denominators;
- outcome definition, ascertainment and window;
- event counts by arm;
- rescue medication, extended observation and transfer;
- missing follow-up;
- funding, sponsor role and author conflicts;
- registry/publication discrepancies.

## Risk of bias

Use RoB 2 at the result level. Functional unblinding is not assumed to equal high risk automatically; reviewers document how it could affect deviations, co-interventions, reporting and outcome measurement for each result.

## Synthesis

- Primary reporting: arm-level risks per 1,000 administrations and risk differences.
- Use exact binomial intervals for study-level risks.
- Do not apply a continuity correction that turns zero events into evidence of zero risk.
- Pool only clinically and methodologically compatible studies.
- When rare events and sparse studies make a pooled model unstable, publish study-level estimates and a narrative synthesis.
- Separate direct MDD/placebo evidence from indirect TRD or minimal-dose-control evidence.
- Report support-setting heterogeneity rather than treating “psychological support” as a single uniform intervention.

## Missing evidence

Search registries for completed-but-unreported studies and compare prespecified outcomes with publications. Contact authors/sponsors for missing arm-level 7-day safety counts when necessary. Missing reports remain visible and may block certainty claims.

## Certainty

Certainty is rated by outcome and evidence stratum, considering risk of bias, inconsistency, indirectness, imprecision and missing evidence. The review will not assign a single certainty grade to “psilocybin.”

## Prespecified stopping rule

The review may conclude **insufficient evidence for a precise serious-event risk** even when no serious events were observed. That is a valid result, not a failed review.

## Reporting standards

The final report will follow PRISMA 2020 and PRISMA-S. Protocol amendments receive a dated changelog entry before affected results are inspected.