Reading Clinical Claims: What Sits Behind 8 Weeks and 87 Percent
Reading clinical claims starts with the control group, not the percentage: topical trials can only be vehicle-controlled, and the vehicle is itself a cream.
Iris Vance writes as a researcher, not a dermatologist or licensed esthetician. Nothing here diagnoses or treats. How we work.
Reading Clinical Claims: What Sits Behind 8 Weeks and 87 Percent
Reading clinical claims well means starting somewhere other than the number on the box. Three things decide what a skincare trial result is worth: what the control group was given, who did the measuring, and how long it ran. The percentage is downstream of all three.
The most useful of the three is also the least known, and it is structural rather than a matter of any one company cutting corners.
Why topical trials cannot be placebo-controlled
A drug trial can give the control group an inert pill. A cream trial cannot give the control group nothing, because something has to go on the skin. A systematic review of over-the-counter vitamin A products in the Journal of Clinical and Aesthetic Dermatology states the constraint directly: It must be remembered that trials of topical treatments can only be vehicle-controlled; they cannot be placebo-controlled.
The consequence follows immediately. The vehicle is the formula minus the active, which means it is a cream, and creams do things. As the review puts it, The vehicles used are almost certainly going to have some occlusive or emollient properties, with temporary and potentially longer-term effects on the skin.
So the control group in a skincare trial is not an untreated group. It is a group using an unlabeled moisturizer. That is why the review notes that The effect of vehicle or placebo in clinical trials is well-recognized and can be relatively high with topical treatments, similar to what is observed in analgesic or antidepressant trials.
What to do with that when reading a claim. A product that beats its vehicle has cleared a real bar. A product reporting only that users improved, with no vehicle arm at all, has demonstrated that putting a cream on skin changes skin. Those are very different claims, and advertising rarely distinguishes them.
Who did the measuring
The second variable is the instrument, and it splits into two kinds.
Clinical evaluation means a person looking at skin and scoring it. It is standard, and it is vulnerable in a specific way the review names: a change in facial expression or lighting can alter how facial lines look, independent of whether anything about the skin changed.
The alternative is instrumental. In the review's description, Profilometry, on the other hand, offers an objective and quantitative method to evaluate the effects of topical treatments on the signs of (photo-)aging, by measuring the surface topography of the skin rather than judging it by eye.
The gap between the two shows up in the results. Of the nine trials in that review, Five trials performed profilometry and only one of those reported a statistically significant improvement against vehicle.
And a third kind, which is not measurement at all. Consumer self-assessment, the source of most "87% of users agreed" figures, is a satisfaction survey run on people who know they are using the product. It belongs in the marketing column, not the evidence column, and this site grades it as Limited wherever it appears, per how we rate skincare evidence.
How long is long enough
Duration is the easiest thing to check and the most often quietly wrong.
The benchmark worth carrying comes from the prescription end of the same category: trial evidence puts visible improvement in photoaged skin at three to six months of tretinoin, with much of it arriving after six, which the retinoid strength ladder sets out with the source attached.
Against that, an eight-week cosmetic trial reporting a clear anti-aging effect is claiming to have moved faster than the strongest molecule in the class. It might be right. It is at least a reason to look at what was measured and against what. How long skincare takes to work collects the documented timelines by category, and most of them are longer than the trials that get advertised.
The checklist a good trial publishes
The review lists what proper reporting of a randomized controlled trial includes, and it doubles as a checklist for reading one: primary endpoint designation, power calculation, sample size determination, intention-to-treat analysis, well-described randomization and double-blinding processes, and reporting of subject demographics and baseline characteristics.
Two of those are worth translating.
Power calculation and sample size. A properly designed trial works out in advance how many people it needs for a difference to be meaningful. A study that recruits whoever was available and finds a result has no defense against that result being noise.
Intention-to-treat analysis. Everyone randomized is counted, including the people who dropped out. Analyzing only those who finished quietly removes the people for whom the product was irritating or useless, which is the group whose absence flatters the result most.
None of this is exotic. It is the standard the same review applies when it concludes that the positive trials it found had major methodological flaws affecting design and analysis.
Why real life underperforms the trial
Even a well-run trial reports a number that will not transfer intact. The review names the reason: Clinical trials often have results that are more impressive than what is observed in real life because of subjects' adherence to treatment, which is essential in trials and, therefore, emphasized and monitored.
Trial participants are reminded, supervised and counted. Nobody does that at home in week nine. The honest expectation for a product with a genuine trial behind it is somewhat less than the trial reported, applied for somewhat longer than you intended.
Translating the claim on the box
| What the claim says | What it may actually mean | What to ask |
|---|---|---|
| "Clinically proven" | Some study exists, of unstated design | Proven against what: vehicle, or nothing? |
| "87% of users agreed" | Consumer self-assessment survey | Was anything measured, or only reported? |
| "Visible results in 2 weeks" | Short trial, often clinical scoring by eye | Faster than tretinoin manages, on what measurement? |
| "Dermatologist tested" | Someone with the title was involved | Tested for what, with what result? Covered in skincare label claims |
| "Reduces wrinkles by 30%" | Instrumental measure, or a scoring scale | Against vehicle, and over what period? |
| "Improvement vs. baseline" | Compared with the same people before | Compared with a control group, or only with themselves? |
The last row is the one worth memorizing. Against baseline means against nothing.
FAQ
What does vehicle-controlled mean in skincare? The control group receives the same formula with the active ingredient left out. It is the only kind of control a topical trial can run, since something has to be applied, and the vehicle itself usually has occlusive or emollient properties.
Are skincare clinical trials reliable? It depends entirely on design. The systematic review discussed here found nine randomized, double-blind, vehicle-controlled trials of over-the-counter retinol products and judged the positive ones to have major methodological flaws. A trial existing is not the same as a trial being informative.
What is a good sample size for a skincare study? There is no single number. What matters is whether the trial ran a power calculation to determine the size it needed before recruiting, which is one of the reporting standards the review lists.
Why do skincare studies use self-assessment? Because it is cheap, fast and produces high percentages. Participants know what they are using, so the result reflects satisfaction rather than measurement, which is why claims sourced this way are graded as weak evidence here.
How long should a skincare trial last? Long enough for the endpoint being claimed. For photoaging, the prescription benchmark in the literature is three to six months, so a cosmetic trial claiming comparable results in eight weeks is asserting something faster than the best-documented option in the category.
Almost every skincare claim can be sorted with three questions, none of which require a background in statistics: what did the control group put on their face, who or what did the measuring, and for how long. A claim that survives those is worth taking seriously, and one that avoids stating them has usually answered them already.
Sources
- Evidence for the Efficacy of Over-the-counter Vitamin A Cosmetic Products in the Improvement of Facial Skin Aging: A Systematic Review, The Journal of Clinical and Aesthetic Dermatology