There Are Two Kinds of “No Evidence”, and Only One of Them Means the Claim Is Wrong
You have read the sentence a hundred times. There is no evidence for that.
Did it mean nobody checked, or that somebody checked and found nothing?
Those are opposite situations, and the same five words cover both. One says a claim was put to the test and failed. The other says the test was never run — so the claim might be perfectly true, and nobody on earth can currently tell you.
We have been keeping track. Below are the ones we hit while checking claims one at a time, sorted into which kind they are.
Say this part out loud first
An untested claim is not a disproven claim. Nothing in the first list below has been shown to be false. It has been shown to be unchecked, which is a different and much weaker thing.
We are labouring this because the drift is so easy. A site that collects gaps starts treating every gap as a debunk, and within a year it is the site that says nothing works. An absence is a description of the literature, not a verdict on the claim.
What an absence does tell you is that anybody stating the claim confidently — in either direction — is going beyond what is known. That includes us.
Kind one: nobody ran the study (19 of them)
Each line names the specific study that does not exist. That precision is the point — “no evidence” is exactly the vague phrase we are trying to replace.
| Creatine gummies No published analysis of a commercial creatine gummy, across four literature searches. Nobody has bought a jar and measured what is in it. | unsupported |
| Lymphatic drainage massage Every trial studied people with lymphedema after cancer treatment. For weight loss, contouring, detox or bloating in healthy adults: no trials at all. | unsupported |
| Oil pulling The trials measure plaque, gum inflammation, bacteria and breath. For tooth whitening, and for drawing out toxins — the two claims doing most of the selling — nothing. | preliminary |
| Testosterone boosters Researchers took the 16 best-selling products, pulled out all 37 ingredients, and found 105 trials of the ingredients. Trials of the actual products: none. | unsupported |
| Castor oil packs The mechanism needs castor oil to cross skin. We could not find a single study measuring whether it does. | unsupported |
| Beef tallow on skin The scoping review’s own conclusion: what is needed is research on tallow as a cosmetic product in humans. The thing being sold is the thing untested. | unsupported |
| Testosterone screening Stated plainly in the clinical guideline: no clinical trials have assessed whether screening strategies actually help. | unsupported |
| A home test for loose skin No validated way to tell loose skin from fat at home. Not a pinch, not a threshold. Nobody has published one. | unsupported |
| Seed oils, as you actually eat them The trials tested swapping saturated fat for seed oil. Seed oil added on top of a modern diet — the real exposure — has never been trialled. | unsupported |
| Walking after meals, for weight Good evidence for blood sugar. We went looking for trials of post-meal walking and body weight and found none. | supported |
| Food combining, for digestion Promoted continuously since 1929. For the claim that separating foods eases bloating or digestion: no trials. Not negative ones. None. | unsupported |
| Berberine as “nature’s Ozempic” The whole pitch rests on raising GLP-1. That has been measured in rats. In a person, nobody has measured it. | unsupported |
| Cold plunges, at plunge length The calorie figure comes from studies scaling hours of cold to a full day. What three minutes delivers, nobody has measured. | unsupported |
| Sea moss and heavy metals Sold as removing heavy metals; the national reference work lists it as containing arsenic, lead, cadmium and mercury. Whether these cancel out in a body is unmeasured. | unsupported |
| Creatine and water retention The best available review says so in its own limitations: there is an absence of trials structured to assess water retention. Nobody has run the one that would settle it. | established |
| Intermittent fasting, for hard outcomes A Cochrane review found no trial has reported a hard cardiovascular outcome. No heart attacks, no strokes, no deaths measured. | preliminary |
| Any cooking oil, against death No trial has ever tested a specific cooking oil against mortality and produced a clean answer — in either direction, for any oil. | unsupported |
| Today’s apple cider vinegar gummies In 2005 researchers tested eight ACV tablets and could not confirm they contained apple cider vinegar. Nobody has run the equivalent test on the gummies since. | preliminary |
| Most of the Blue Zones Ten regions carry the claim. Three are well characterised. One has never been studied at all. | preliminary |
Kind two: somebody ran it, and it came back empty (3 of them)
Here is the contrast. These claims got the study the first list is waiting for. It arrived, it was properly run, and it found nothing.
| Creatine and hair loss Forty-five men randomised, thirty-eight completed, blinded, twelve weeks. They measured DHT, the DHT-to-testosterone ratio, hair density, follicular units and hair thickness. The result: nothing. | unsupported |
| The post-workout testosterone spike About as direct a test as you can run: identical arm curls, once with a deliberate hormone surge and once without. Strength rose equally in both. In the paper’s words, no effect of hormone elevation. | unsupported |
| Food combining, for weight Fifty-four people, calories and macronutrients matched. Separated group lost 13.7 lb (6.2 kg), mixed group lost 16.5 lb (7.5 kg). No significant difference, and the numbers ran the wrong way. | unsupported |
One claim manages to be both at once
Food combining — the idea that you should not eat protein and starch in the same meal — has been promoted continuously since 1929, and it makes two promises.
The weight promise was tested. Fifty-four people, calories and macronutrients matched, which is exactly the right way to isolate the separating. The separated group lost 13.7 lb (6.2 kg); the mixed group lost 16.5 lb (7.5 kg). No significant difference, and the numbers ran the wrong way.
The digestion promise has never been tested at all. Not negatively. Not once. Ninety-seven years of promotion and zero trials.
Same diet, same page, two completely different epistemic situations. If you only ever heard “there is no evidence for food combining”, you would not know that half of it was measured and half of it was ignored.
So which kind should worry you more?
The untested kind, and it is not close.
A claim that was tested and failed is a solved problem. It is unpleasant for whoever was selling it, but the information exists and you can act on it.
An untested claim is worse for you specifically, because it cannot be resolved by reading harder. There is no paper to find. Everyone confidently discussing it — the people selling it and the people mocking it — is working from the same nothing.
And notice which claims end up in the first list. Not obscure ones. Creatine gummies, testosterone boosters, lymphatic drainage, apple cider vinegar gummies — large, busy markets where the specific question a buyer would ask has never been put to a test.
Our read on why: the study that settles a claim is only worth funding to somebody who does not already have your money. A market can run for decades on a plausible mechanism, and nobody inside it has a reason to spend on the experiment that could end it. That is a guess about incentives, not a finding — but it fits the shape of the list better than coincidence does.
How to tell which one you are reading
This is the part worth taking away, because it works on claims we have never touched.
- Ask what the study would have looked like. If someone tested this properly, who was in it, what got measured, how long? If a page cannot tell you that, it is probably reporting an absence without knowing it.
- Watch for ingredient evidence standing in for product evidence. “Studies show ashwagandha…” is a claim about a compound. Whether the capsule in your hand contains a useful amount of it is a separate question, and usually an unasked one.
- Watch for the population swap. Real trials, run in patients after cancer surgery or in rats, quietly re-pointed at a healthy adult. The evidence is genuine; it is just about somebody else.
- “Not proven” and “disproven” are not synonyms — and anyone using them interchangeably is telling you something about their argument rather than the research.
What happens to this page next
It grows, and we would rather it grew in one direction than the other.
Every absence here is recorded as an absence in our searching, not a proof that nothing exists. If a study turns up — if somebody finally buys the jar of gummies and runs the assay — that entry moves to the second list or off the page, and we will say which. A register like this is only worth anything if entries can leave it.
The one we most want to see filled: creatine gummies. Seventy-four thousand searches a month, millions of views arguing about what is inside them, and not one published analysis. It is the cheapest unanswered question on this entire page, and it has stayed unanswered because answering it serves nobody who currently profits from the argument.
The four ratings above are the ones on every verdict here, and we wrote down how we read a study so you can see what earns each. Two companion pages measure different things from the same corpus: what health advice actually reaches people, and who paid for the study.
Nobody pays us to reach a conclusion, and there is nothing for sale on this page.