About this archive

Where the research comes from

Everything comes from PubMed, the medical literature database of the United States National Library of Medicine. For each topic a fixed search runs against that database, anchored on the subject headings the Library itself assigns to an article.

That anchoring matters more than it sounds. Searching on loose words pulls in the wrong field entirely: “collagen” appears just as often in research on knee osteoarthritis, and that has nothing to do with your skin. The subject heading is a librarian's judgement that an article is about this — and that is a far steadier foundation than a word count.

The archive currently holds 4,693 publications, from 1966 to 2026. Of those, 1,200 were selected as carrying the evidence, spread over 5 topics.

How a study gets its grade

Not by a language model, but by a rule you can read back. PubMed records for every article what type of study it was and whether it was done in people, animals or cells. We translate that into a number from 1 to 5 — the evidence ladder — and the reason is stored alongside it, so a wrong grade can be traced to the rule that produced it rather than to “the computer said so”.

Where the rule cannot establish something, a study gets no number at all rather than a guessed one.

What you will not find here

• No product advice and no recommendations. We describe what exists; we prescribe nothing.
• No advertising, no affiliate links, no sponsored content, nothing for sale beyond the subscription itself.
• No medical advice. Every study links back to the original publication, so you — or your doctor — can check where a statement came from.

If a manufacturer helped fund a study and that was declared, it says so on the page. If nothing is stated, that means nobody wrote it down — not that the research was independent.

Still being built

The plain-language write-ups are still being written. Until they are there, each study page shows the abstract exactly as the researchers published it.