Supplement Brief publishes material corrections when factual errors affect the interpretation of evidence, study attribution, doses, outcomes, or safety information. This page records those corrections and significant issues identified before publication.
Urolithin A vs NMN
What was wrong
- Third-party AI brief described a full sleep RCT protocol (actigraphy, PSQI, metabotype stratification) with no ClinicalTrials.gov registration.
- Liu et al. (JAMA Network Open 2022) was labeled cognitive evidence — the trial measured muscle endurance only.
- ENDURO was described as including football players; the published trial enrolled male distance runners only (NCT04783207).
- NCT07060898 (~650 participants) was described as a polysomnography sleep trial; sleep is a secondary self-report endpoint on a brain-health app study.
What we did
- Guide written from primary papers and trial registries only.
- ATLAS (Singh et al., Cell Reports Medicine 2022): pre-specified primary endpoint peak power output (no win vs placebo) separated from significant hamstring isokinetic torque at 500 mg and 1000 mg; handgrip did not separate from placebo between groups.
- Liu 2022 framed as muscle endurance. ENDURO limited to runners. NCT07060898 described with correct endpoint hierarchy.
Metformin and B12 deficiency
What was wrong
- Draft materials implied metformin causes neuropathy at population level; Yang 2019 meta-analysis reports deficiency risk, not higher neuropathy prevalence at meta level.
- Methylcobalamin marketed as superior to cyanocobalamin without comparative RCT support (PMID 25820384).
What we did
- Published guide limits neuropathy framing to clinical risk when deficiency is missed.
- Forms section debunks methyl-default marketing with Obeid et al. cited explicitly.
L-citrulline vs L-arginine for ED
What was wrong
- Cormio 2011 frequently misreported as 3 g/day instead of 1.5 g/day.
- PMID 33587304 meta-analysis mis-attributed to Chen J instead of Xu Z (Andrologia 2021; e14007, not e13995).
- Schwedhelm pharmacokinetics paper mis-cited as PMID 17953784 instead of 17662090.
- 2024 network meta-analysis mis-attributed to Sansone et al. instead of Barbonetti et al.
What we did
- Body citations and source list aligned to PubMed records.
- Verification notes added to the article research section.
- Last updated set to 2026-07-24 on the live guide.
Nitric oxide ED supplements
What was wrong
- Hub draft inherited a Chen 2021 label on PMID 33587304 — first author is Xu Z et al., Andrologia 2021.
What we did
- Hub cites Xu Z et al. with correct PMID and article ID.
- Cross-link to the citrulline pillar for the full citation audit trail.
Pure typos, punctuation, CSS, and internal link tweaks are omitted unless they changed a factual claim. Entries marked Caught pre-publish document errors caught before first publication — often from AI briefs, influencer slides, or mislabeled trial summaries already circulating in the category.
How we correct guides
- Factual errors, wrong doses, mis-attributed studies, and outcomes described backwards (secondary sold as primary) are fixed in the article body and sources list.
- Substantive fixes bump the guide’s Last updated date shown on the article page. Typo and formatting-only edits do not.
How verified claims are counted
A verified claim is a specific numeric or citation-backed statement in a guide — a trial result, dose, safety interaction, primary endpoint, sample size, funding source, or effect size — checked against its primary source during pre-publish review.
Totals are intended to be reproducible: a reader following this definition and our published guides should arrive at the same order of magnitude, not an exact match sentence-by-sentence.
Counts as one verified claim
- A body sentence with a source reference where the cited field was opened and matched (PubMed abstract, full text, or ClinicalTrials.gov registry).
- Primary-endpoint checks: pre-specified outcome, direction of effect, and whether the result reached statistical significance vs placebo.
- Citation integrity checks: first author, journal, year, article ID, and PMID/DOI match the source list entry.
- Numeric checks: dose, duration, n=, and effect size match the primary paper or registry record.
Does not count
- General framing, hedging, or buy/skip recommendations without a specific verifiable fact.
- A source listed only in the bibliography with no corresponding factual statement in the body.
- The same fact verified twice because it appears in two sentences — counted once.
- Typo, formatting, or CSS fixes that do not change a factual claim.
A public verification summary (total claims checked vs corrections logged) will appear on this page once the log reaches 10 entries and totals have been audited against this definition. Until then, only material corrections and pre-publication rejections are listed at the top of this page. See also methodology.
Report an error
We welcome corrections.
Email [email protected] with the guide URL, the sentence that looks wrong, and a PMID, DOI, or ClinicalTrials.gov ID if you have one. We review every message; we do not provide personal medical advice.