Supplement

Collagen peptides

Protein · Hydrolyzed collagen type I & III

Collagen peptides are among the most heavily advertised skin products of all. Meta-analyses find better skin moisture and elasticity across all studies; the only analysis that separates studies by funding source, however, found these effects only in manufacturer-funded studies.

What collagen peptides are

Collagen is the body’s most abundant structural protein. As a food supplement, it is sold hydrolyzed, i.e. broken down into short fragments that can be absorbed.

The mechanism remains disputed, because collagen is digested like any other protein. That ingested fragments are reassembled into collagen specifically in the skin is an assumption, not an established process.

Why funding decides everything here

There are numerous randomized studies on the effect on skin, measured with instrument readings for moisture and elasticity, i.e. surrogate markers. Taken together, they look positive. The decisive test: sort them by origin and quality.

A 2025 analysis is so far the only one that has done exactly that. The result reverses the overall picture: the positive finding disappears as soon as the manufacturer-funded studies are taken out. The problem is not a single study but the structure of the research field.

What is well supported

Skin moisture and elasticity: clear effects in the largest meta-analysis. Pu 2023 pooled 26 randomized studies with 1,721 participants and found Z=4.94 (p less than 0.00001) for hydration and Z=4.49 (p less than 0.00001) for elasticity. This is the broadest available analysis of the effect on skin.
Even the most critical analysis finds effects in the overall picture. Myung and Park 2025 analyzed 23 randomized studies with 1,474 participants. Across all studies, there were significant improvements in hydration, elasticity and wrinkles. Only sorting by funding source and methodological quality changes the picture – the initial finding itself is undisputed and the same in both analyses.
Knee osteoarthritis: significant effect on pain, moderate certainty of evidence. Lin 2023 found a standardized mean difference of −0.58 (95% CI −0.98 to −0.18; p=0.004) in four randomized studies with 507 participants. The certainty of evidence according to GRADE was rated moderate – the highest GRADE rating given in the evidence reviewed on this topic.

What the studies show

Separated by funding: no effect

A meta-analysis of 23 randomized studies with 1,474 participants (Myung and Park 2025) is the only analysis that stratifies by funding source. Across all 23 studies, there were significant improvements in skin moisture, elasticity and wrinkles. Studies without pharmaceutical or manufacturer funding showed no effect on any of these three measures; only industry-funded studies found any. Sorting by methodological quality produced the same pattern: no significant effect in any category in the high-quality studies, an elasticity gain exclusively in the low-quality ones. The conclusion in the original: There is currently no clinical evidence to support the use of collagen supplements to prevent or treat skin aging.

The counter-calculation without stratification

A meta-analysis of 26 randomized studies with 1,721 participants (Pu 2023) reports clear effects on skin moisture (Z=4.94; p less than 0.00001) and elasticity (Z=4.49; p less than 0.00001). The authors themselves name several sources of bias; they did not separate the studies by funding. The contrast between the two analyses is the core of this page: largely the same studies, and the difference in results arises solely from sorting by origin and quality.

Wrinkles: statistically borderline, clinically marginal

A meta-analysis of 19 randomized studies with 1,341 participants (Nukaly 2026) on oral and topical peptides in general, not only collagen, found an overall mean difference of 0.27 for wrinkles at p=0.04; this was driven by oral polypeptides. The value is barely statistically notable and clinically marginal, and no statement specifically about collagen can be derived from it.

Joints: a pain signal on a weak basis

For knee osteoarthritis, there are four randomized studies with 507 participants (Lin 2023). For pain, the standardized mean difference was −0.58 (95 percent confidence interval −0.98 to −0.18; p=0.004) with I-squared at 68 percent, meaning the individual studies differed strongly from one another. The certainty of evidence according to GRADE was considered moderate – with a high risk of bias in all four studies. For adverse events, the odds ratio was 1.66 (0.99 to 2.78; p=0.05), with very low certainty of evidence.

Where the data stop

No skin effect without manufacturer funding. In the independently funded studies, skin moisture, elasticity and wrinkles remained unchanged. The frequently cited overall finding is produced by the industry-funded studies.
No effect in high-quality studies. The only positive finding after sorting by quality, an elasticity gain, comes exclusively from low-quality studies. Two sorting criteria, the same result.
No evidence for cartilage structure or preserved function. The joint data concern reported pain. Whether cartilage would be preserved or a joint replacement avoided has not been studied.
On hair, nails and gut, there is nothing in the evidence reviewed. These promises are simply not covered by the studies analyzed.

Status, approval and legal

Collagen peptides are sold as food supplements; the evidence reviewed contains no assessment by a public authority, and approval as a medicine was not the subject of the studies analyzed.

The decisive statement here comes not from an authority but from the body of studies itself – as stated word for word in the conclusion of the 2025 meta-analysis.

Safety

Tolerability is considered good; no serious problems were reported in the studies analyzed. If you are allergic to fish or beef, pay attention to the source of the raw material.

One caveat remains: the odds ratio for adverse events from the joint meta-analysis, at 0.99 to 2.78, reaches just below 1, and the certainty of evidence was considered very low. The finding is no more suitable as an all-clear than as a warning.

BK-Score Thin human evidence

Human evidence		5
Mechanism		5
Safety data		9
Hype gap		4
Track record of use		8

There are RCTs on skin moisture and joint complaints, but they are small and mostly manufacturer-funded. The mechanism remains disputed, because collagen is digested like any protein. Very heavily advertised.

The score rates the state of knowledge, not the substance. “Safety data 9” means well studied – not harmless. “Track record of use 9” means used long and widely – that is not proof of efficacy.
Subjective assessment by Biohacking Kompakt based on published scoring rules – not a scientific rating and not a medical recommendation. Rules and all ratings (German)

Frequently asked questions about collagen peptides

Does collagen really work against skin aging?

According to the best analysis currently available, no. A meta-analysis of 23 studies with 1,474 participants separated the studies by funding source: in the studies without manufacturer or pharmaceutical funding, there was no effect on skin moisture, elasticity or wrinkles. The study’s conclusion is that there is currently no clinical evidence for its use against skin aging.

Why do the studies on collagen contradict each other?

Because the decisive separation is usually missing. An older meta-analysis of 26 studies found clear effects but did not analyze by funding. The 2025 analysis uses largely the same studies and sorts them by origin and quality. After that, nothing of the positive overall picture remains.

Does it matter who paid for a study?

In this case, it decides the result. Positive findings on skin were found exclusively in industry-funded studies; independently funded ones showed no effect. Sorting by methodological quality produced the same pattern, with only low-quality studies finding an elasticity gain. Two independent criteria point in the same direction.

Does collagen help with joint pain?

There is a pain signal on a weak basis. Four randomized studies with 507 people with knee osteoarthritis yielded a standardized mean difference of minus 0.58 for pain, but with strong variation between the studies and a high risk of bias in all four. There are no data on cartilage structure or preserved function.

How large is the effect on wrinkles?

For collagen specifically, this has not been quantified. A meta-analysis of 19 studies with 1,341 participants on oral and topical peptides in general, not only collagen, found an overall mean difference of 0.27 for wrinkles at p equal to 0.04, driven by oral polypeptides. That is barely statistically notable and clinically marginal. A statistically detectable difference is not the same as a visible one.

Is collagen harmful?

Tolerability is considered good; no serious problems were reported in the studies analyzed. In the joint meta-analysis, the odds ratio for adverse events was 1.66, with a confidence interval of 0.99 to 2.78 and very low certainty of evidence. If you are allergic to fish or beef, pay attention to the source of the raw material.

The podcast episode (in German)

Episode 60