Substance evidence
Cannabidiol (CBD)
The anxiety evidence, the dose gap, and the liver signal.
What Its Legal Status Actually Means
Read this before anything below it. Legal to sell is not the same fact as reviewed, verified, or approved.
There is no FDA-approved psychiatric indication for CBD — not for anxiety, depression, PTSD, insomnia, or any other mental health condition. One cannabidiol product is FDA-approved: Epidiolex (plant-derived purified CBD oral solution), indicated for seizures associated with Lennox-Gastaut syndrome, Dravet syndrome, or tuberous sclerosis complex in patients 1 year and older. Everything else sold as CBD is an unapproved consumer product: the 2018 Farm Bill removed hemp (cannabis containing no more than 0.3% delta-9 THC by dry weight, 7 U.S.C. 1639o) from the Controlled Substances Act, but the FDA has stated it is unlawful to add CBD to food or to market it as a dietary supplement, denied three industry citizen petitions asking for a rulemaking to permit supplement marketing (January 2023), and concluded that existing food and supplement frameworks are not appropriate for CBD and that a new pathway from Congress is needed. In practice that means retail CBD is sold with no premarket review of potency, purity, or claims.
Quoted whole from the 2026-08-18 study harvest. Not summarized, not shortened.
What the Research Shows
5 records from the 2026-08-18 harvest, each checked against PubMed and Europe PMC on that date. How a study was designed, how many people were in it, and who paid for it are printed above what it found, because those three facts decide how much the finding is worth.
Study 1 · 2026
- Meta-analysis of randomized trials
- 2,477 participants
- Top-tier journal
- Independently funded
- Preregistered
- Declared conflicts of interest
gold standard98/100Top of the evidence hierarchy, independently funded.
The most recent systematic review and meta-analysis of randomised trials of cannabinoids as a primary treatment for mental disorders — 54 trials, 2,477 participants, searched to May 2025 — found no significant effect on anxiety outcomes and no randomised evidence at all for depression. Cannabinoids raised the odds of any adverse event (OR 1.75, 95% CI 1.25-2.46; number needed to harm 7) without raising serious adverse events, and the authors concluded that routine use of cannabinoids for mental disorders is currently rarely justified.
How this record was checked
PubMed record 41856154 fetched via NCBI efetch 2026-08-18; title, authors, journal, year, DOI, 54 trials and n=2477 confirmed from the abstract, along with the funding line "The National Health and Medical Research Council" read directly off the record's FUNDING field (hence INDEPENDENT) and PROSPERO registration CRD42023392718. Record carries an "Erratum in" notice (Lancet Psychiatry 2026;13(8):e11) but no retraction flag.
Study 2 · 2011
- Randomized trial
- 24 participants
- Reputable journal
- Funding not established
early signal50/100Suggestive but preliminary. Not settled.
Read this first: Conducted by the University of São Paulo CBD group (Zuardi, Crippa, and colleagues). The 2011 record carries no conflict-of-interest statement, but the same group's later publications disclose that several of these authors are co-inventors on a CBD patent that the university has licensed to pharmaceutical companies, so independence from a commercial interest in CBD cannot be assumed.
Twenty-four never-treated patients with social anxiety disorder received a single 600 mg oral dose of CBD or placebo 90 minutes before a simulated public speaking test; the CBD group reported significantly less anxiety, cognitive impairment, and discomfort during the speech than placebo, ending up close to the untreated healthy comparison group. This is a one-dose laboratory challenge in 24 people, not a treatment trial — it says nothing about whether taking CBD over weeks improves social anxiety in daily life.
How this record was checked
PubMed record 21307846 fetched via NCBI efetch 2026-08-18; title, authors, journal, year, DOI, and n=24 randomised patients (12 CBD, 12 placebo, plus 12 untreated healthy controls) confirmed from the abstract. PMC record 3079847 fetched the same day via NCBI efetch (db=pmc) contains no funding or conflict statement, so funding is recorded UNKNOWN.
Study 3 · 2019
- Randomized trial
- 57 participants
- Reputable journal
- Mixed funding
- Declared conflicts of interest
moderate58/100Reasonable evidence with real limitations.
Read this first: The disclosure states that STI-Pharm supplied the CBD at no cost and that four authors are co-inventors of a fluorinated-CBD patent; the University of São Paulo has licensed that patent and holds an agreement with Prati-Donaduzzi to develop a synthetic CBD product for indications including anxiety disorders. Public grant money plus donated study drug and patent interests is why funding is recorded MIXED.
Fifty-seven healthy men were randomised to 150 mg, 300 mg, or 600 mg of oral CBD or placebo before a simulated public speaking test. Only the 300 mg dose significantly reduced anxiety; 150 mg and 600 mg were no different from placebo. More CBD is not better, and the dose that worked here is far above what retail CBD products deliver.
How this record was checked
PubMed record 30328956 fetched via NCBI efetch 2026-08-18; title, authors, journal, year, DOI, and the n=57 allocation (150 mg n=15, 300 mg n=15, 600 mg n=12, placebo n=15) confirmed from the abstract, and the conflict-of-interest statement read off the same record. Europe PMC full text (PMC6781714) fetched 2026-08-18: acknowledgements state the study was supported by CNPq and FAPESP grants (CNPq/MS/SCTIE/DECIT 26/2014; 466805/2014-4), read alongside the donated-drug and patent disclosures.
Study 4 · 2015
- Narrative review
- Reputable journal
- Funding not established
contested33/100Weak or heavily disputed. Treat as a question, not an answer.
Read this first: The low score reflects study design, not disputed findings: this is a narrative review with no systematic search, no pooled estimate, and no readable funding statement, so it carries little independent evidentiary weight even though its cautious conclusion has held up.
The most-cited review of CBD for anxiety concluded that animal evidence is strong but that human evidence is limited to single acute doses, with very few studies in people who actually have an anxiety disorder and essentially none on chronic dosing. Ten years on, that gap has still not been filled by adequately powered trials.
How this record was checked
PubMed record 26341731 fetched via NCBI efetch 2026-08-18; title, authors, journal, year, and DOI confirmed. PMC record 4604171 fetched the same day via NCBI efetch (db=pmc) returns front matter and abstract only, with no funding or conflict statement, so funding is UNKNOWN and independence is left unchecked.
Study 5 · 2017
- Cross-sectional study
- 84 participants
- Top-tier journal
- Mixed funding
- Declared conflicts of interest
early signal43/100Suggestive but preliminary. Not settled.
Read this first: Funded by the Institute for Research on Cannabinoids, a nonprofit on whose board three of the authors sit (unpaid), and two authors disclose personal fees from numerous cannabis and cannabinoid companies. The result nonetheless cuts against those commercial interests, which weakens the usual direction-of-bias concern.
Laboratory analysis of 84 CBD products bought online found that 26% contained less CBD than the label claimed and 43% contained more — roughly seven in ten were mislabelled — and delta-9 THC was detected in 21% of samples. Whatever the evidence says about a given CBD dose, a retail product cannot be assumed to contain the dose on the bottle.
How this record was checked
PubMed record 29114823 fetched via NCBI efetch 2026-08-18; title, authors, journal, year, and DOI confirmed. PMC record 5818782 fetched the same day carries the full funding statement (Institute for Research on Cannabinoids, a 501(c)(3) supported by individual donations) and the conflict disclosures; the research letter's body text is not in PMC, so the figures (84 products, 26% under-labelled, 43% over-labelled, THC in 21%) were confirmed from the peer-reviewed commentary indexed at PMC6024459 (Pediatr Neurol Briefs 2018;32:3), which cites this paper directly. Europe PMC also lists NIDA grant R01 DA040460 against the record, hence MIXED rather than INDUSTRY or INDEPENDENT.
The Risks
Nobody checks this purchase the way a prescription is checked. No prescriber reviews it against the rest of what you take, no pharmacist screens the interaction, and no regulator confirms the dose in the capsule before it is sold. That is why this section sits here at full length rather than as a footnote.
What the paragraph below covers:
- Suicidal thoughts or behavior
- Liver injury
- Drug interactions
- What is actually in the product
- Stomach and gut
- Sedation
CBD is not inert. In the controlled epilepsy trials that supported Epidiolex, 12-13% of CBD-treated patients had ALT elevations above three times the upper limit of normal versus 1% on placebo, with some cases requiring hospitalisation; risk rises with dose and with concomitant valproate. CBD inhibits and induces multiple CYP and UGT enzymes, so it can raise or lower blood levels of other drugs — the clobazam interaction is well documented, and the same mechanism applies to many psychiatric medications. Somnolence, sedation, diarrhoea, and reduced appetite are common, and the approved labelling carries a warning for suicidal behaviour and ideation. Separately, retail CBD products are frequently mislabelled: in a JAMA analysis of 84 products bought online, only about three in ten contained the labelled amount of CBD and 21% contained detectable THC. The FDA has publicly flagged potential liver harm, drug interactions, and possible harm to the male reproductive system with long-term CBD use.
What Is Not Known
The boundaries of the section above, in the harvest’s own words. What it excluded, why, and where it looked and found nothing. An absence of evidence is not evidence of absence, and it is not evidence of benefit either.
What Was Left Out, and Why
Quoted from the harvest, where these notes sat above the study list rather than after it. Where a note says “below” it means the studies in the section above.
- Epidiolex epilepsy trials (Devinsky 2017 NEJM and successors) excluded: they are the FDA-approved seizure indication, not a psychiatric one, and including them would inflate the apparent strength of the mental-health evidence base.
- Open-label and retrospective CBD "anxiety clinic" case series excluded in favour of the randomised and meta-analytic evidence below.
- Hauser 2023 Cochrane review of cannabis-based medicines for cancer pain was verified (PMID 37283486) but dropped as off-topic for mental health.
- The dose gap is the story: every controlled trial showing an anxiolytic signal used a single oral dose of 300 mg or 600 mg of pharmaceutical-grade CBD. No controlled trial has tested the far smaller amounts in retail CBD oils, gummies, or vapes, and no controlled trial has tested repeated daily dosing for an anxiety or mood disorder.
The Retraction Check
PubMed efetch records for all five PMIDs (21307846, 30328956, 26341731, 29114823, 41856154) inspected 2026-08-18 for "Retraction in" flags; none present. Two carry "Erratum in" notices (31672337-family errata do not apply here; 41856154 has Lancet Psychiatry 2026;13(8):e11), noted in the relevant records.
Questions for a Prescriber or Pharmacist
This page does not tell anyone to take Cannabidiol (CBD) or to avoid it. It is not able to: it does not know what else you take, what you have tried, or what you are treating. These are the questions each study above raises, written to be asked out loud.
- The largest trial review to date found no benefit of cannabinoids for anxiety and no randomised evidence for depression — is there any reason my case would be different from the people in those trials?
- The trial people cite for CBD and social anxiety used one 600 mg pharmaceutical dose in 24 people before a lab speech task — does that tell us anything useful about taking CBD daily for my anxiety?
- This trial found only the 300 mg dose helped — 150 mg and 600 mg did nothing — so how would I know whether any over-the-counter CBD product is anywhere near a dose that has ever been tested?
- The review everyone quotes says the human CBD evidence is acute-dosing only — has anything since then actually tested taking it every day for an anxiety disorder?
- If most online CBD products don't contain what the label says and one in five had detectable THC, how would I verify what is actually in a product before taking it with my other medications?
And the one to ask at the counter, whatever the answers above turn out to be: given everything I am already taking, what would you want to know about Cannabidiol (CBD) before I put it in the same body?
Where This Page Comes From
Transcribed from the study-harvest-2026-08-18 record for cannabidiol, built into this page by scripts/supplement-ingest.mjs on 2026-09-07. Nothing here is fetched at page load, and nothing here was written by a model without a citation behind it.
| Source file | study-harvest/2026-08-18/cannabidiol.yaml |
|---|---|
| Source repo | REPTechnologies/avalo-backend |
| SHA-256 | b41668e4271c48c95fc5cf5e98458679e5042cf513dfd88adba7d4d716179f97 |
| Also searched as | CBD, hemp extract, CBD oil, Epidiolex, full-spectrum hemp |
The Other Substances
- 5-HTP
- Ashwagandha
- Cannabis and THC
- Creatine
- Kratom
- L-Theanine
- Magnesium
- Melatonin
- Methylene Blue
- Microdosing
- N-Acetylcysteine (NAC)
- Omega-3
- Phenibut
- SAMe
- St John’s Wort
- Valerian
Do any supplements actually work? asks the question across all of them. The full library has the medication records, the funding investigations, and the glossary.