Do collagen supplements improve skin and joints? Decoding the claim

Collagen powders and capsules are marketed on two separate promises at once, smoother skin and healthier joints, and the trial evidence behind those two promises has moved in opposite directions in just the last two years. That makes collagen an unusually clean case of a claim that has to be split in two rather than answered with a single verdict.

The skin claim, and the 2025 reanalysis that undercut it

For most of the last decade, the case for collagen and skin rested on trials like Proksch and colleagues (Skin Pharmacology and Physiology, 2014), a double-blind, placebo-controlled study of 69 women aged 35 to 55 that found measurable improvements in skin elasticity after 8 weeks of daily collagen peptides, using a product supplied by the manufacturer that funded the study. Trials of this shape, small, short, positive, and funded by a collagen manufacturer, became the backbone of the marketing claim, and pooled together in earlier meta-analyses they looked like a real, if modest, effect.

A 2025 systematic review and meta-analysis (Myung and Park, American Journal of Medicine) went further than simply pooling the trials: it split them by funding source. Looking at all 23 randomized controlled trials together (1,474 participants), collagen supplements did appear to significantly improve skin hydration, elasticity, and wrinkles. But when the analysis separated trials funded by a pharmaceutical or supplement company from trials that were not, the industry-funded trials showed the benefit and the independently funded trials did not. The same pattern held when the trials were split by methodological quality: the higher-quality studies showed no significant effect on any of the three outcomes, while the significant improvement in the pooled result was concentrated in the lower-quality trials (which showed a benefit for skin elasticity). The authors’ own conclusion was blunt: there is currently no clinical evidence to support the use of collagen supplements to prevent or treat skin aging, once funding and quality are accounted for. This is one of the clearer documented cases of a supplement’s evidence base evaporating under a bias-corrected reanalysis, and it should carry real weight against the skin half of the claim.

The joint claim, and evidence that has gotten stronger, not weaker

The joint half of the claim has moved in the opposite direction. An earlier meta-analysis (García-Coronado and colleagues, International Orthopaedics, 2019) pooled 5 randomized placebo-controlled trials and found a significant reduction in osteoarthritis symptoms, measured by the WOMAC index, but with high heterogeneity between the small trials it included. That heterogeneity was a real limitation: a critical commentary the following year (von Hippel, International Orthopaedics, 2021) argued directly that the meta-analytic evidence available at the time did not support strong conclusions either way, a fair caution against reading a handful of small, variable trials as a settled answer.

A larger, more recent analysis has since addressed much of that gap. Liang and colleagues (Osteoarthritis and Cartilage, 2024) pooled 35 randomized controlled trials, 25 of them contributing to the primary analysis (2,856 patients), and used trial sequential analysis, a method built specifically to check whether the accumulated evidence is large enough to support a firm conclusion rather than reflecting noise or early stopping. It found a small-to-moderate reduction in pain (standardized mean difference -0.35) and improvement in function (standardized mean difference -0.31), graded moderate to high certainty, with the trial sequential analysis indicating the evidence had reached a large enough total sample to support a definitive conclusion, and no increase in withdrawals or adverse events compared with control. This is a materially more robust body of evidence than the skin literature, though a standardized mean difference in that range is still a modest effect on a subjective pain and function scale, and given how clearly funding shaped the skin trials, it is a fair open question, not addressed in the published abstract, how much of the joint literature carries the same industry-funding pattern.

Where the two claims genuinely diverge

Putting the two bodies of evidence side by side, they no longer support the same verdict. The skin claim rests on a literature that a rigorous, bias-stratified 2025 reanalysis found does not hold up once industry funding and study quality are accounted for. The joint claim rests on a larger, more recent, and more methodologically careful synthesis that found a real, statistically robust, if modest, effect, with an explicit statistical check confirming the evidence base is large enough to trust. Marketing that treats “collagen improves skin and joints” as a single, unified claim is not representing either finding accurately.

Where a documented condition changes the calculation

None of the trials discussed here enrolled people with a diagnosed collagen-synthesis disorder or a specific dermatologic or rheumatologic condition being separately treated; they tested generally healthy adults with skin-aging concerns or radiographically confirmed osteoarthritis, taking a supplement on top of, not instead of, standard care. Whether collagen supplementation makes sense alongside an existing joint condition, other supplements, or medications is a question that depends on the specific diagnosis and treatment plan, and belongs with the clinician managing that condition rather than with a general population-level trial average.

The honest bottom line

Collagen for skin appearance is a claim whose supporting evidence has been shown, by the field’s own most recent and most careful analysis, to depend heavily on who funded the trial, with independently funded and higher-quality studies finding no effect. Collagen for joint symptoms in osteoarthritis rests on a larger, newer, and considerably more rigorous body of evidence that does show a real, if modest, benefit. Those two findings pull in different directions, which is why this claim earns a mixed verdict rather than a single answer, and why the skin and joint promises printed on the same bottle deserve different levels of trust.