Do collagen drinks and supplements work for skin?
The evidence is contested, and two large reviews disagree. A 2025 meta-analysis in the American Journal of Medicine pooled 23 randomised trials with 1,474 participants. Across all of them, collagen supplements improved skin hydration, elasticity and wrinkles. When the authors split the trials by who paid, trials not funded by pharmaceutical companies showed no effect, while trials funded by them did. The highest-quality trials also showed no significant effect. The authors concluded there is currently no clinical evidence to support collagen supplements to prevent or treat skin ageing.
A 2026 umbrella review, which is a review of reviews, came to a warmer view. It gathered 16 systematic reviews covering 113 trials and 7,983 people, and reported that skin outcomes were consistently favourable. A third analysis, in 2023, pooled 26 trials of hydrolysed collagen and found improved hydration and elasticity against placebo, while noting several biases in the trials and calling for larger ones. Its authors declare no conflict of interest.
What is a meta-analysis, and what is an umbrella review?
A meta-analysis pools the numbers from several trials of the same question, so that many small trials can act like one larger one. It is only as good as the trials that go in, and a pool of weak trials stays weak. An umbrella review goes one level up: it collects the systematic reviews and meta-analyses themselves. Both are useful, and neither can repair a flaw inside the trials underneath. Reading the fine print on who funded the trials, and how well they were run, is part of reading either one.
Why do the reviews disagree?
Because what goes into a pooled result decides what comes out. In the 2025 analysis, the positive overall result did not hold up once the paper looked at funding and quality separately. The benefit disappeared in the trials not funded by pharmaceutical companies and in the high-quality trials. It remained in the pharmaceutical-funded trials, and in the low-quality trials for elasticity. Funding does not prove a trial is wrong, and it is a reason to look harder at how the trial was run.
An umbrella review adds up existing reviews, and those reviews draw on overlapping trials. So it repeats the strengths of those trials and also their weaknesses. Its abstract does not report a funding statement, or a split by funding source or by trial quality, which is the split that changed the picture in the other analysis. That is my reading of the abstract, and the full paper may say more.
Small numbers add to the difficulty. The 2025 analysis averaged roughly 64 people per trial (1,474 people across 23 trials), and the 2023 analysis roughly 66 (1,721 across 26). Trials of that size can be swung by a few people. The trials in these reviews also use different products, doses and lengths of use, and measure different things. The 2023 analysis found that the hydration effect varied with the source of collagen and how long people took it.
What did the trials actually measure?
Skin hydration, elasticity and wrinkles, as the reviews describe them. A difference on a measuring instrument may not be a difference you can see in the mirror, and the abstracts do not say how large the changes were in everyday terms. The abstracts do not say whether participants themselves noticed a difference. Ask what was measured before you decide what a study proves.
Does drinking collagen feed the collagen in your face?
The pieces reach the blood, and the trials do not show what happens after that. A 2025 randomised trial in healthy volunteers gave a single dose of collagen hydrolysate and followed blood samples for six hours. The amino acids hydroxyproline, glycine and proline, and short chains of them called peptides, entered the bloodstream. That shows absorption. It does not show that those pieces go to the skin, or that the body uses them to build collagen there. The authors work for a collagen ingredient company, which is worth knowing when reading the result.
Collagen in food and drinks is broken down in digestion, like other proteins. The body then decides where the pieces go. The skin trials above test whether skin measurements change after weeks of taking collagen, which is the more relevant question.
How do I check a collagen product in Malaysia?
Look for a registration or notification number on the label and search for it. The National Pharmaceutical Regulatory Agency (NPRA) runs a public search, called QUEST3+, for registered and notified products. A listing tells you the regulator knows the product, and it does not tell you the drink improves skin. That question is what trials are for, and the label's claims are not trial results.
Also read what the product says it does. A pack that promises to erase lines or "rebuild" collagen is making a claim the trials above do not support. Whitening creams are a separate hazard, and the mercury check is explained in whitening creams and mercury.
What should I do if I still want to try one?
Treat it as a small experiment. Choose a product that is listed with NPRA, tell your doctor or pharmacist if you take regular medicines or are pregnant or breastfeeding, and decide in advance what result would make you stop. Keep the ordinary things going too, such as the sun protection discussed in sunscreen in a tropical climate.
Injected and energy-based collagen treatments are a different category with different evidence and different risks, and they are set out in three roads to collagen. What skinboosters can and cannot do is in what skinboosters actually fix. None of that shows a drink is useless or that a treatment is better. Each is judged on its own evidence.
Where the evidence splits by who paid, the sensible response is patience. Two big reviews disagree, and only better trials will settle it.
