Editorial Policy
Last updated 2026-08-07
Last updated: 14 August 2026
This policy governs how BioRx selects, sources, grades, reviews and corrects the material it publishes. It applies to every contributor, whether staff, freelance or volunteer.
Sourcing standards
We work from primary literature first. Every substantive claim on BioRx traces to a citeable source, and we prefer the original study to any summary of it.
Sources we use, in order of preference:
- Systematic reviews and meta-analyses in peer-reviewed journals.
- Randomised controlled trials and other registered clinical trials, including their registry records (ClinicalTrials.gov, ISRCTN, EU CTR).
- Regulatory assessments and monographs: MHRA, EMA, FDA, WHO.
- Peer-reviewed observational and preclinical studies.
- Reputable secondary references (pharmacopoeias, standard textbooks) for background and mechanism.
Sources we do not treat as evidence: manufacturer marketing material, vendor product pages, press releases unaccompanied by a published study, podcasts, forums, social media, anecdote, and articles whose only support is another article.
Preprints may be cited where they are the only available evidence, but are always identified as not peer-reviewed and are never used to upgrade an evidence tier.
Retractions. We check the retraction status of cited work at review. Retracted sources are removed and the affected claims are re-examined; where a retraction changes a conclusion we publish a correction.
Citations include enough detail for a reader to find the original: authors, year, journal, and a DOI or permanent identifier wherever one exists. Where a source sits behind a paywall we say so.
The four-tier evidence scale
Every claim on BioRx is graded. The tier reflects the strength of the evidence, not the attractiveness of the finding.
| Tier | Label | What it means |
|---|---|---|
| Tier 1 | Established in humans | Supported by multiple well-conducted randomised controlled trials in humans, a systematic review or meta-analysis of such trials, or a regulatory approval based on that evidence. Effects, and their limits, are reasonably well characterised. |
| Tier 2 | Emerging human evidence | Some human data exists (early-phase trials, small or single RCTs, or consistent observational studies), but the evidence base is limited in size, duration, replication or quality. Direction of effect is plausible; magnitude and safety are not settled. |
| Tier 3 | Preclinical only | Evidence comes from animal models. Results may be striking, but species differences, dose scaling and model limitations mean nothing here can be assumed to transfer to humans. |
| Tier 4 | Mechanistic, theoretical or anecdotal | In vitro work, computational modelling, extrapolation from related compounds, or reported experience. This tier tells you what someone has proposed, not what has been shown. |
Rules of application:
- A page is graded per claim, not per compound. A single compound may carry Tier 1 evidence for one endpoint and Tier 4 for another, and we display them separately.
- Tiers never round up. Where evidence is genuinely borderline, we assign the lower tier and explain why in the text.
- Tier 3 and Tier 4 material is published with an explicit statement that it has not been demonstrated in humans.
- Absence of evidence is stated as absence of evidence, never as evidence of absence, and never as an implied benefit.
Review cadence
| Content type | Reviewed at least every |
|---|---|
| Compound entries with active clinical research | 6 months |
| Compound entries with a stable evidence base | 12 months |
| Explainers, background and methodology pages | 12 months |
| Safety-relevant content, following any new regulatory action | Promptly, and in any event within 30 days of the change coming to our attention |
Reviews are triggered on schedule and also on event: a new trial publication, a regulatory decision, a safety signal, a retraction, or a credible reader challenge. Each page displays its last reviewed date, and a review that results in no change is still recorded as a review.
Conflicts of interest
BioRx has no commercial relationship with any manufacturer, supplier, clinic, laboratory or distributor of the compounds it covers. We take no advertising, sponsorship, affiliate revenue or paid placement of any kind.
Every contributor completes a declaration of interests before writing and updates it whenever their position changes. Declarable interests include employment or consultancy, research funding, equity holdings, patents or applications, paid speaking or advisory roles, and close personal relationships with an interested party.
Where a relevant interest exists:
- it is disclosed on the page in question, in plain language;
- the contributor is normally recused from grading evidence on that compound; and
- the entry receives review by a contributor with no declared interest.
Undeclared interests discovered after publication are treated as a correction event: the disclosure is added, the affected content is re-reviewed, and the change is logged.
Corrections policy
We will get things wrong. When we do, we would rather fix it visibly than quietly.
How to report an error. Use the contact form with the page URL, the passage concerned, and (where possible) the source that shows the error. Anyone may report an error; you do not need to explain who you are or why you are asking.
What happens next. We acknowledge corrections within 2 working days and aim to resolve them within 10 working days. Where a claim is safety-relevant and we cannot immediately verify it, we may amend or temporarily remove the passage while we check.
How we classify and act:
- Typographical or formatting : corrected silently.
- Factual correction not affecting meaning : corrected, with a dated note in the page's change log.
- Material correction affecting a claim, tier or conclusion : corrected, with a visible correction notice at the top of the page stating what was wrong, what it now says, and the date.
- Retraction : where a claim cannot be supported at all, the content is withdrawn and replaced with a dated explanation. We do not delete the page silently.
We do not remove accurate, properly sourced content because a party is unhappy with it. We will always consider a substantiated argument that it is wrong.