161 Million Views, 20 Health Claims, and What the Survivors Had in Common

How much of the health advice that reaches you is actually true?

Nobody knows, and the reason is dull: answering it takes two jobs that almost nobody does together. Somebody has to measure what people actually watch. Somebody else has to sit down and read the studies behind it. Those are different skills, different budgets, and different weeks.

So we did both. Over the past month we checked 20 health claims the slow way — reading the trials, rating how well each one held up — and we also measured how far each claim had travelled on YouTube. Put the two columns beside each other and you get something we have not seen anywhere else: 161,268,256 views, sorted by whether the claim underneath survived being read.

Here is the whole thing. Nothing is for sale on this page.

Two thirds of the attention sat on claims that did not hold up

Of the 161,268,256 views we measured, 110,785,923 — 68.7% — sat on claims we rated Preliminary or Unsupported. That is: claims where the research is not yet good enough to tell you anything, or where we went looking and it simply was not there.

The single biggest topic in the set is high cortisol symptoms, at 19,864,237 views. It is rated Unsupported.

And then there is creatine gummies: 6,571,036 views for a product with zero published analyses of what is actually in it. Not a bad study. Not a mixed result. Nobody has ever tested one and written it down.

Before you take that number anywhere: we picked these topics

This is the part most pages would put at the bottom in grey text, so we are putting it here instead.

This is not a random sample of health content. We chose these 20 topics, and we choose things that look shaky — that is the job. A claim-checking desk that picked topics at random would spend most of its week confirming that vegetables are good for you.

So 68.7% does not mean 69% of health advice is wrong. It means that among the claims we thought were worth checking, that is where the attention landed. Anyone who quotes this page as the first thing has misread it, and we would rather say so now than be cited badly later.

What it is good for is the next question.

So is there a shortcut? Something you can check in ten seconds?

That is the question worth asking, because almost nobody is going to read forty papers before deciding whether to buy a jar of something. If reading the research is the only defence, then the research may as well not exist.

We had a hunch, so we tested it against our own table before writing it down here.

Three rows argue with us, and they are the interesting ones

A pattern you only notice when it agrees with you is not a pattern. So here are the rows that break it.

• Beetroot juice — sells something (juice and nitrate shots), and rated Supported anyway.
• Psyllium husk — sells something (psyllium powder), and rated Established anyway.
• Oil pulling — costs nothing (swishing oil already in your kitchen), and still only reached Preliminary.

Look at the two that survived. Psyllium husk is a fiber that has been sold in a plain tub since before most supplement brands existed. Beetroot juice is a vegetable. Both are commodities — generic, unbranded, available at any grocery store, with nobody in particular owning the upside if you believe in them.

Now look at the ones that failed: sea moss, NAD+, methylene blue, colostrum, creatine gummies. Every one of those is a branded category with a markup and a story.

To be explicit, because this is the kind of rule that gets flattened in the retelling: this is a heuristic about where to aim your scepticism, not a verdict about any product. It tells you which claims deserve a closer look. It cannot tell you what is true. Only reading the studies does that, which is why the rest of this site exists.

What we are watching next — and how you will know if we were wrong

This page is not finished. It is a running tally, and every verdict we publish adds a row to it. So here is what we are watching, stated in advance so it counts.

The thing that would prove us wrong: the “nothing to sell” column filling up with failures. It has 5 entries today and one of them — oil pulling — already came back Preliminary. If that column keeps growing and keeps failing, then what we found is not a pattern about commerce at all. It is a pattern about which claims are easy to test, and we will say so on this page rather than quietly dropping the section.

What to watch for in the wild, meanwhile. Three shapes show up again and again in the rows that failed, and once you have seen them you cannot unsee them:

• A mechanism doing the work of a result. The claim explains how something would work, in convincing detail, and never gets around to whether it did. Detail about mechanism is not evidence of effect.
• A real study about the wrong people. The trials exist and are perfectly good — they were just run in patients recovering from cancer surgery, or in mice, and the claim quietly transferred to you.
• A product nobody has ever tested as a product. The ingredient has research behind it. The thing in the jar has none, and the two get talked about as if they were the same object.

Every one of those has a verdict on this site showing it in action, linked in the table above. We will keep adding rows, and when the numbers move we will update this page rather than writing a new one, so the tally stays honest.

If you want to see what earns each of the four ratings, we wrote that down separately: how we read a study. And nobody pays us to reach a conclusion — there is nothing to buy on this page, and there never will be on a page that carries a verdict.