What does "clinically proven" actually mean?
Imagine you are standing in a pharmacy, supermarket, health food shop or, perhaps more dangerously, doomscrolling social media late at night. There are two products competing for your attention. One claims to support healthy skin, improve sleep or reduce the appearance of wrinkles. The other carries two reassuring words: clinically proven.
Well, that settles it, doesn’t it? Break out the credit card and add to cart. Scientists must have been involved. There were probably laboratories, white coats, clipboards and serious looking people peering at significant graphs. Somewhere, one imagines, a researcher has solemnly stamped PROVEN across a large dossier before returning it to the scientific vault.
Except, of course, that science does not work quite like that.
“Clinically proven” is an extraordinarily persuasive phrase precisely because it sounds far more specific than it often is. It invites us to make the mental leap from “someone has conducted some research” to “science has established that this product works”. Unfortunately, those are not necessarily the same thing.
So, when you see the words “clinically proven”, the most useful response is not immediate belief or cynical dismissal. It’s a much simpler question: “Proven to do what, exactly?”
“Clinical” sounds impressive, but what does it actually tell us?
In scientific research, clinical studies generally involve human participants rather than isolated cells or animal models. In marketing, however, terms such as “clinically tested” and “clinically proven” can be considerably less informative than they sound. That matters: if someone wants to claim that a product improves human health, evidence that it produced an interesting reaction in a collection of cells is not the same as evidence that it produces a meaningful benefit in actual people.
However, the mere existence of a clinical study tells us remarkably little about its quality.
A study could involve 3,000 participants or 12. It could run for five years or five days. It could compare the treatment with a placebo, another treatment or nothing at all. Participants could be randomly allocated to groups, or they could choose their own. Researchers could know who received which treatment, or they could be blinded to that information. The outcome could be an important improvement in health, or a tiny change in a laboratory measurement whose significance to everyday life is either uncertain or unproven.
Yet all of these could, in ordinary conversation, be described as clinical research.
That is why “clinically tested” and “clinically proven” should not be treated as synonyms for “definitively shown to work”.
Regulation varies between countries, but the underlying scientific question remains the same: does the evidence actually justify the claim being made? In the United States, for example, the Federal Trade Commission (FTC) requires health-related advertising claims to be supported by “competent and reliable scientific evidence”.[1] Its guidance says that health benefit claims will generally require randomised, controlled human clinical testing, although the appropriate evidence depends on the claim and there are circumstances in which such trials are not feasible. Crucially, the strength of evidence claimed in the advertising must itself be accurate.
The label on the product is therefore the beginning of our investigation, not the end of it.
Not all clinical trials are created equal
Imagine that I announce I have developed a cream that improves concentration. Yes, a cream. Please try to keep an open mind, if not an open wallet. It contains Quantum Intellectual Peptides™.
I give the cream to ten people and ask them whether they feel more alert afterwards. Seven say yes.
Can I say it has been clinically tested? I have certainly tested something on humans.
Can I reasonably conclude that the cream caused an improvement in concentration? Not remotely.
Perhaps the participants expected it to work. Perhaps they were my parents, siblings and close friends and wanted to please me. Perhaps they happened to sleep particularly well the night before. Perhaps “more alert” is not a reliable measure of concentration. Perhaps I quietly forgot to mention the three people who said it did nothing.
This is why study design matters.
Randomisation
In a randomised trial, participants are assigned to different groups by chance rather than according to their preferences or the researchers’ choices.
Suppose people who are already particularly health-conscious disproportionately choose to take a new supplement. Six months later, they may be healthier than people who did not take it, but perhaps they also exercise more, smoke less, eat differently and have better access to healthcare.
Randomisation helps distribute such differences between groups so that, on average, the groups are more comparable before the intervention begins. It is one reason randomised trials can support causal conclusions with much greater confidence than many non-randomised designs.[2]
Control groups
We also want to know what happened to people who didn’t receive the intervention.
Without a control group, an improvement after treatment can easily be mistaken for an improvement because of treatment.
Symptoms fluctuate. People recover naturally. Behaviour changes when people know they are being studied. Expectations matter. If everyone receiving my miraculous concentration cream reports feeling more alert after a week, that sounds interesting. If an equivalent group receiving an inactive cream reports exactly the same improvement, then the miracle has become considerably less miraculous.
The FTC specifically identifies comparison with a control group as an important principle in assessing health research.[3]
Blinding
Then there is blinding, which is far less gruesome than it sounds. Blinding means keeping relevant people unaware of which intervention a participant received. Depending on the study, that might include participants, healthcare providers, researchers assessing outcomes or data analysts. Where blinding is possible and appropriate, it can reduce the possibility that expectations influence behaviour, treatment or measurement.[4]
This becomes especially important when outcomes are subjective. If I know I have received an expensive new anti-ageing cream rather than the placebo, I may scrutinise my face in the mornings with considerable optimism (and without my glasses). If the researcher examining my photograph also knows which cream I used, even an entirely conscientious researcher may be influenced by that knowledge.
Human beings are not machines: this remains true even when we are wearing lab coats.
Size matters, but not in the way you think
Now imagine a properly controlled trial finds that a product works. Excellent! How many people were studied?
A trial involving 18 people can provide useful preliminary information, but small studies generally produce less precise estimates, are more vulnerable to chance variation and may fail to represent the wider population who will eventually buy the product. Adequately powered larger studies of comparable quality can therefore give us greater confidence, although size cannot rescue poor study design.
We should also ask who those participants were.
A treatment tested exclusively on healthy men aged 20 to 30 has not automatically been demonstrated to work equally well in women aged 70. Evidence obtained from people with a particular condition cannot simply be transferred to healthy people without justification. The FTC itself gives the example of research conducted in a particular patient population being insufficient to support a broader claim aimed at healthy adults.[5]
So when an advert says that a product is “clinically proven to improve skin hydration” we might reasonably ask: “In whom?” Twenty university students? Five hundred adults? People with unusually dry skin? For how long? Under what conditions?
Science becomes much more informative when we stop admiring the adjective clinical and start examining the nouns and numbers that follow it.
Statistically significant does not necessarily mean significant to you
Here we encounter one of the most useful distinctions in medical research: the difference between statistical significance and a clinically meaningful effect.
A study may detect a difference that is unlikely to be explained by chance, yet that difference may be so small that it has little practical importance.
Suppose a sleep product is associated with an average increase in sleep duration of three minutes per night. With a sufficiently large study, that difference might conceivably be statistically detectable, but in real terms an extra three minutes of sleep per night is unlikely to make you less reliant on your morning coffee.
Marketing department: CLINICALLY PROVEN TO INCREASE SLEEP DURATION
Customer: “Wonderful. By how much?”
Marketing department: “We would prefer to return to the first sentence.”
The FTC explicitly warns that statistically significant results must also translate into benefits that are clinically meaningful to consumers.[6]
Whenever you encounter an impressive percentage, therefore, ask for the actual effect.
A 50 percent improvement sounds enormous. An outcome changing from 2 people in 10,000 to 3 people in 10,000 is also a 50 percent relative increase.
Percentages are not dishonest, but they do need context.
One study is not the final word
Suppose the trial was large, randomised, controlled and blinded. The result was both statistically significant and clinically meaningful.
Now we have something genuinely interesting, but we still do not have the scientific equivalent of the Ten Commandments engraved on stone.
Studies sometimes produce unusual results by chance. Methodological problems may emerge later. Other researchers may fail to reproduce the findings. A result may apply only under particular conditions.
This is why replication and the wider body of evidence matter.
If ten good studies have examined a treatment and nine find no meaningful benefit while one finds a spectacular effect, advertising based entirely upon the spectacular study would give a distorted picture of the evidence.
That is cherry picking, and it is particularly powerful because the evidence being quoted may be perfectly genuine. Suppose researchers measure 20 outcomes and only one produces an impressive result. Reporting that one outcome as though it had always been the main question can make chance findings look much more convincing. Trial registration and pre-specification help reveal whether researchers changed outcomes or analyses after seeing the data.
The FTC guidance consequently stresses the importance of considering the totality of relevant scientific evidence, rather than selecting supportive findings while ignoring better quality contradictory research. It also warns against selectively reporting positive outcomes when a study has measured several things.[7]
This is also why systematic reviews can be so valuable. Instead of asking only, “What did this study find?”, researchers examine the broader collection of relevant studies, assess their quality and consider what the evidence shows overall.
One swallow doth not a summer make, and one clinical trial doth not a scientific consensus make.
What about peer review?
Peer review is another phrase that sometimes acquires magical properties in public discussion.
If a study has been peer reviewed, other specialists have examined it before publication. That is useful because it provides a layer of scrutiny and can identify errors, weaknesses or unsupported conclusions.
But peer review is quality control, not canonisation.
A peer-reviewed study can still be small, poorly designed, contradicted by subsequent research or simply wrong. Conversely, the fact that research has not yet undergone peer review does not automatically make it false.
The better question is not: “Has this been peer reviewed?” but: “What does peer review add to the other evidence I have about the quality of this research?”
That may be less satisfying than a simple tick box, but critical thinking has an unfortunate habit of making things more complicated just when we were hoping to finish early.
Who paid for the research?
Funding deserves attention too, although this requires more nuance than simply declaring: INDUSTRY FUNDED! UNTRUSTWORTHY! CASE CLOSED!
Companies have legitimate reasons to fund research into their own products. Indeed, someone has to pay for clinical trials, and conducting serious research is expensive.
The potential problem is not that commercial funding automatically makes a study false. It is that financial interests can create incentives that influence which questions are investigated, which outcomes receive attention, how results are interpreted and which findings reach publication.
So ask:
Who funded the study?
Were researchers independent?
Was the study registered in advance?
Were the planned outcomes specified before the results were known?
Have independent groups replicated the findings?
Most importantly, does the result fit the wider evidence?
The source of funding is a reason to look more carefully, not an excuse to stop looking.
Was the whole product tested, or merely one ingredient?
This is particularly important with supplements, cosmetics and wellness products.
Imagine a face cream containing twelve ingredients. One of those ingredients has been studied and shown, under particular conditions, to affect skin hydration. The manufacturer then advertises the cream as: FORMULATED WITH A CLINICALLY PROVEN INGREDIENT
Read that sentence carefully. It does not necessarily tell us that the cream itself has been clinically shown to produce the advertised effect. The dose, formulation, delivery method and population all matter. An ingredient that behaves one way in isolation does not automatically make every product containing it effective.
The FTC specifically warns that evidence concerning an ingredient may not substantiate claims about a finished product when other ingredients or differences in formulation mean the evidence cannot simply be transferred.[8]
The distinction is minute in linguistic terms but potentially enormous in scientific terms.
So what should you ask?
The next time “clinically proven” appears on a bottle, advert or website, resist both the temptation to be impressed and the temptation to roll your eyes.
Instead, ask:
Proven to do what?
Was the actual product tested, or only an ingredient?
How many people took part?
Who were they?
Was there an appropriate comparison or control group?
If the study was testing cause and effect, were participants randomly assigned where this was feasible?
Was blinding used where appropriate?
How large was the actual effect?
Was the effect clinically meaningful, rather than merely statistically significant?
Has the finding been replicated?
What does the wider body of evidence show?
Who funded the research, and can I find the study myself?
You do not need a doctorate in epidemiology to ask these questions. You do not even need to answer all of them every time. Their purpose is to interrupt the mental leap from “sciencey words are present” to “therefore the claim is true”.
“Clinically proven” may describe a claim supported by excellent research. It may describe a modest finding that marketing has inflated beyond recognition. It may refer to one ingredient rather than the product in your hand. It may rest on a small preliminary study, while better research points elsewhere.
The phrase in isolation cannot tell you which.
Good evidence does not become stronger because the packaging is reassuring, the percentage is enormous or the person in the advertisement has a stethoscope draped strategically around their neck.
When you next encounter those comforting words, there is no need to recoil from them. Just add three more: “Clinically proven how?”
Comments