How to read a skin study without a science degree

A label says “clinically proven” and, for once, cites an actual paper. You look it up and hit a wall of jargon: “randomized, vehicle-controlled, double-blind, n=42, p<0.05.” You don’t need a science degree to get through it. You need to know where to look and what to skip. Here’s a map.

The layout: every study has the same five rooms

A 1997 BMJ series on how to read a paper lays out the standard structure, which most medical journals still follow. It’s called IMRaD: Introduction (why the authors did the study), Methods (how they did it and how they analyzed it), Results (what they found) and Discussion (what they think it means). In front of all that sits the abstract, a one-paragraph summary.

The BMJ author’s key piece of advice is counterintuitive: if you’re deciding whether a paper is worth your time, do it “on the design of the methods section and not on the interest of the hypothesis, the nature or potential impact of the results, or the speculation in the discussion.”

In other words, skip to the middle. The abstract is the trailer; the methods are the film.

The abstract: useful, but written to sell

Read the abstract to learn what the study was about, then treat it with mild suspicion. Abstracts are short, written by the authors, and tend to lead with the best result. The word “significant” in an abstract means “unlikely to be chance”, not “large” or “noticeable”. A “significant” 3% change in a lab reading may be real and still invisible on your face.

Two things the abstract should tell you clearly: what kind of study this was (a trial? a survey? a review of other studies?) and roughly how many people were in it. If it doesn’t say, that’s your first flag.

The methods: the five questions that matter

This is where you spend your time. The BMJ series suggests starting with three orientation questions: why was the study done, what type of study is it, and was that design appropriate for the question? For a skin care product, you can boil that down to five practical checks:

1. Who was studied? Age, skin type, condition. A cream tested on 30 healthy 25-year-olds tells you little about an eczema-prone toddler or your mother’s winter hands. Look for “inclusion criteria”.
2. How many? Find the “n”. A 2024 paper in The Oncologist shows what goes wrong with small numbers: sampling ten people at a time from two known populations, the authors got “different/opposite findings” from one draw to the next, purely by chance. Bigger groups quiet that noise.
3. Compared with what? Look for the words “control”, “vehicle” (the same cream without the active ingredient), “placebo” or “comparator”. If everyone got the product and nothing was compared, the study can’t separate the product from time, weather and attention.
4. Who knew what? “Randomized” means a coin flip chose the groups. “Double-blind” means neither participants nor assessors knew who got what. “Open-label” means everyone knew. The less blinding, the more room for wishful grading.
5. What was measured, and when? Find the “primary outcome”. Was it an instrument reading, a trained grader’s score, or a questionnaire? Was it measured at two weeks or twelve? Was it decided before the study or picked afterwards?

If you answer those five, you understand the study better than most people who quote it.

The results: numbers with their clothes on

A few habits help here.

Look for the actual numbers, not just “improved”. How much did the product group change, and how much did the control group change? The difference between those two is the effect. A 40% improvement sounds great until you see the control group improved 35%.

Look for a range around the number. Good papers give a “confidence interval”, a range within which the true effect probably lies. The 2014 review of evidence-based dermatology urges readers to stop dividing results into “significant” and “not” and instead use these ranges “to estimate a range of likely effects.” A wide range means the study isn’t sure, whatever the headline says.

Look for what happened to everyone. If 60 people started and 40 finished, ask why 20 left. Irritation? Boredom? The paper should say.

The discussion and limitations: where honesty lives

The discussion is the authors’ interpretation, and it’s the most optimistic room in the house. Read it, but read the limitations paragraph twice. This is where authors admit the sample was small, the study was short, there was no control group, or the funder designed the trial. It’s often the most useful paragraph in the paper, and it’s usually near the end, just before the conclusion.

Also check two things that sit outside the main text: who funded the study, and whether the authors declare a conflict of interest. Neither disqualifies the work. Both tell you how hard to look.

Red flags, in order of how often you’ll see them

• No control group. Before-and-after in a single group is the most common design in skin care and the least informative.
• Tiny sample. Anything under about 20 is a pilot, whatever it’s called.
• Surrogate measures. The study measured a lab value or a cell-culture result rather than something you’d notice. The 2014 dermatology review warns of trials built around “a clinically dubious outcome measure” instead of “what is really important to patients.”
• “Up to.” A 2012 study commissioned by the US Federal Trade Commission found that many consumers read “up to X%” as a promise of X%. In a paper, “up to” usually means the single best participant.
• Short duration. One application, one day, one week. Skin conditions cycle over months.
• A mismatch between paper and label. The study tested an ingredient at one concentration in a lab; the label claims results for a finished product on your face.
• “Data on file.” Not a study you can read. Treat it as a claim without a source.

Finding studies yourself

You can do this for free. PubMed is a search engine run by the US National Library of Medicine that indexes millions of biomedical papers. Type the ingredient and the condition (for example, an emollient ingredient plus “atopic dermatitis”), then use the filters on the left: article type lets you narrow to “randomized controlled trial” or “systematic review”, and you can restrict by publication date or age group.

Many results show only an abstract. For the full paper, look for a “free full text” link, or search PubMed Central (PMC), the library’s free full-text archive, which lets you filter by date and search inside articles. When the full text isn’t free, the abstract plus the methods summary usually answers the five questions above.

One caution from the library’s own guidance: filters based on article type rely on indexing that can lag, so very recent papers may drop out when you apply them. If something looks missing, try again without the filter.

If you’re reading studies because a skin problem isn’t settling, eczema (atopic dermatitis, the itchy inflamed rash), acne, a rash that keeps coming back, that’s a sign to get it looked at rather than to keep reading. If it isn’t better after two weeks, see your doctor or pharmacist.

What to remember: go to the methods first, and answer five questions: who, how many, compared with what, who was blinded, what was measured and when. Read the numbers with their comparison group attached, read the limitations paragraph twice, and treat “up to”, tiny samples and lab-dish outcomes as flags. PubMed and PMC put the papers within reach of anyone with a browser.