Editorial process
What has to be true before a page publishes
The checks that run on every build
These are not a policy document, they are code. If one fails, nothing ships.
- Claims. No dosing, no protocol instructions, no disease-treatment claim including implied ones, and no presenting a non-diagnostic device as a diagnostic test.
- Corpus provenance. No statistic can render without its denominator, its scope and the exact query behind it. A bare count will not compile.
- Evidence tiers. Every cited source carries a tier, and preprints, abstracts and registries cannot be cited without a stated limitation.
- Commercial placement. No paid link may appear on the about, methodology, corpus or disclosures pages. Any link that would is a build failure.
- Dark by default. No commercial link renders unless its programme is both approved and explicitly switched on.
- Orphans and broken links. Every page must be reachable and every internal link must resolve.
How we choose topics
From what patients actually discuss, measured across 7.2 million comments, rather than from keyword volume. We publish the measurements, including for the topics we decided against, so our choices are auditable.
Updating and corrections
Prices are re-verified on a stated cadence and every one renders the date it was last checked. When we get something wrong we log it publicly rather than editing quietly. The correction log.
Use of AI
We use AI tooling for research, drafting and for building the checks described above. It is not permitted to be the source of a fact. Every figure on this site traces to a data file, a named study, or a published query, and the personal sections of this site are written by a person about their own experience or they are not written at all.