How to read psychedelic research without losing the caveats.

Replace “does it work?” with a question naming a population, intervention, comparator and outcome. Record the species and setting too. A study of mice, healthy volunteers or people with a diagnosed condition has a different scope. Relevance to another population is an additional inference, not a detail to omit.

Cochrane treats mismatches between the research question and available evidence as a question of indirectness. That distinction helps explain why a biologically interesting result may not answer a clinical question. Cochrane guidance on evidence certainty.

Write down the actual assay or outcome. In the PsychDoc source record, dendritic spine imaging, functional brain connectivity and symptom scales describe different observations. A headline about “rewiring” does not make them interchangeable. Ask what the measurement can establish and which extra inference is needed to connect it to the claim.

Keep follow-up beside the result: during an experiment, days later and months later are separate questions. A finding at one visit does not establish persistence beyond it. Check which outcome and analysis were specified in advance. CONSORT 2025 sets reporting expectations for randomized trials, including outcomes, participant flow and analysis.

Improvement from baseline and improvement relative to a control are different comparisons. Look for the comparator, randomization and who knew the assigned intervention. Record whether participants, staff or outcome assessors could infer assignment, and whether that knowledge could affect this particular outcome.

Next, compare enrollment with the number analyzed. Why are outcomes missing, and do the reasons differ between groups? A small dropout percentage alone does not settle the issue. Cochrane assesses bias for a specific result, including randomization, deviations from interventions, missing data, measurement and selective reporting. Cochrane risk-of-bias guidance.

Count underlying studies and participants, not just papers. A second analysis of one trial is not a second independent trial. Check cohort reuse, research groups, funding and declared conflicts; a shared author is a reason to inspect the relationship, not proof that results are invalid.

For a systematic review, inspect the search date, eligibility rules and included-study list. PRISMA 2020 provides reporting guidance that helps readers see how a review was assembled. A reporting checklist does not itself validate the conclusions.

Read the effect estimate and uncertainty together. Ask whether the interval permits materially different conclusions, whether outcomes agree across studies, and whether harms received comparable attention. Cochrane's certainty framework considers bias, inconsistency, indirectness, imprecision and publication bias. PsychDoc's displayed tiers should not be read as formal GRADE assessments.

Open the psilocybin and brain rewiring dossier and compare its records for Shao 2021 and Siegel 2024. The former concerns dendritic spines in mice; the latter concerns functional connectivity in healthy human volunteers. They do not measure the same thing in the same population.

Before calling them agreement or contradiction, compare their outcomes and follow-up windows. Use the evidence map to find each source and its linked claims, then read the original papers. The map is a navigation aid; the dossier's proposed explanation remains a hypothesis with explicit limitations and pending independent expert review.

Fill one record per claim. Leave missing information marked as unknown.

1. Question: What exact claim am I evaluating?
2. Source: What is the paper, version, publication status and persistent link?
3. Population: Which species, cohort and setting were studied?
4. Measurement: What outcome or assay was used, and at what time point?
5. Comparison: What control, allocation and masking procedures were reported?
6. Participants: How many enrolled, were analyzed and were lost to follow-up?
7. Result: What is the effect estimate, uncertainty and reported harm?
8. Independence: Are participants, datasets or research programs shared?
9. Interpretation: What is observed, what is inferred, and what remains unknown?
10. Review: What source or qualified assessment could change this reading?

Cohort

The group of participants studied. Several papers can describe one cohort.

Assay

The test or measurement method used to observe a phenomenon.

Comparator

The condition against which the intervention is assessed.

Attrition

Loss of participants or measurements during follow-up; its reasons matter.

This is a reading aid, not a validated appraisal instrument or medical advice. About PsychDoc · Editorial policy · Discuss a research briefing