How we check a claim before we publish it
We started from the opposite end of the usual process. Instead of writing an article and then looking for support, we took every factual statement our own program already makes, put them in a list, and went looking for a health authority document that says each one.
That list has 1,228 statements in it. Most of them didn't survive.
Who we check against
Each statement is compared against published documents from these bodies, plus systematic reviews found through searches we wrote down before we ran them:
- ACOG American College of Obstetricians and Gynecologists
- ASRM American Society for Reproductive Medicine
- NICE UK National Institute for Health and Care Excellence
- Cochrane Cochrane systematic reviews
- CDC US Centers for Disease Control and Prevention
- NIH ODS NIH Office of Dietary Supplements
- WHO World Health Organization
- NHS UK National Health Service
A statement only counts as checked when the source document has actually been retrieved and a passage found in it that says the statement. That passage gets copied out word for word and shown next to the statement, so you can see whether we've read it fairly rather than taking our word for it.
The three labels we publish
These aren't the same strength, so we don't show them as if they were. Every statement in the library carries one of them.
Multiple health authorities agree
More than one independent health authority states it.
This is the only bar we treat as strong enough to use in advertising.
Stated in a health authority's guidance
One health authority states it in published guidance.
Solid, but it rests on a single body's reading.
Supported by a systematic review
A systematic review of the research supports it.
Research consensus rather than an authority's own guidance.
122 of the 1,228 statements reached one of those three bars. 94 of them are on the evidence library pages so far.
What we hold back
The rest of the list is the more useful half, honestly. Here's where it went:
Emerging evidence
Early or mixed research. We don't put it on these pages, because a statement labeled honestly as thin still reads as advice once it's on a page.
Not enough support found
We looked and couldn't find an authority document that says it. It goes to a list to be rewritten or dropped, not to a page.
Contradicted
An authority document says something different. These are the ones we most want to find, and they never get published.
Those statements haven't been quietly deleted. They're a working queue, and where one of them appears in our own program material, the material is what gets changed.
What review has actually happened
This is the part most sites are vague about, so here it is plainly.
The checking described above is automated. Software retrieved each source document, found the passage, and judged whether the passage actually supports the statement. That is why the label on every claim says machine-verified, and it's also why we publish the passage itself: the check you can run on us is the one that counts.
Clinical review is a separate step and it hasn't happened. We've written down who is allowed to do it, what they have to be qualified in, and that a separate person has to check the evidence record rather than the same person signing off twice. Nobody holds either role yet.
So every claim card says machine-verified, clinical review pending, and it will keep saying that until it isn't true. When a reviewer is appointed, the statements change one at a time as they're read, not all at once because a role got filled.
Where we use a number
A lot of authority documents carry intake figures, and those figures are the most useful and most dangerous thing on the page.
Our rule is that a number stays attributed. It appears inside the quoted passage, with the body that published it named and linked. We don't restate it in our own voice as an amount for you to take, because the moment we do, we're prescribing, and we're not qualified to.
Educational information, not medical advice. Talk to your own clinician before changing anything you take.
Found something wrong? The corrections page explains how to tell us and what happens next.