Mon-Fri 9am-6pm PST
+1 (800) 686-5898
support@longevitybased.com
Mon-Fri: 9am-6pm PST
Does Collagen Really Work? Debunking the Latest Review
I couldn’t agree more about the importance of reading the entire article before drawing conclusions. I’ve been taking collagen for years and was initially thrilled to read that an analysis suggested I didn’t need to continue spending my hard-earned cash on it. However, upon delving deeper into the full paper, I realized that the findings were far removed from the initial claims. This is a perfect illustration of why relying solely on abstracts can be deceiving.
Why the New Findings Are So Jarring
It’s surprising for me — and I’m sure many others — that such a wide-ranging body of previous research was seemingly contradicted. For instance, in 2023, a massive meta-analysis analyzed 26 RCTs. In that analysis, researchers discovered collagen peptides dramatically improved skin hydration and elasticity. These two markers tend to diminish over time, so discovering a substance that reverses this trend is exciting.
Although the new analysis initially produced the same findings as the 2023 meta-analysis — namely, that collagen peptide supplementation produced improvements in hydration, elasticity, and wrinkles — the researchers conducted some subgroup analyses.
The Subgroup Analyses Were Shocking
Subgroup analyses are used to isolate subsets of data. In this case, researchers conducted the following subgroup analyses:
- Studies not funded by supplement or pharmaceutical companies
- Studies ranked as high quality
- Studies ranked as high quality and not funded by industry
Using these subgroup analyses, the pooled results of each subgroup were no longer statistically significant for hydration, elasticity, and wrinkles. At face value, this appears to be a powerful indicator that the presence of industry funding bias and/or poor quality of studies contributed to the previously reported positive results.
However, if that were true, you would expect most of the high-quality studies within the subgroups to report no effect. That wasn't the case.
What the Individual High Quality Studies Really Found
The meta-analysis identified nine studies that were ranked high quality based on two standard quality metrics.
If the pooled subgroup result indicated "no benefit," you would expect most of the nine high-quality studies to indicate no effect. However, that is not what occurred.
Of the nine high-quality studies,
- One study reported no benefit
- One study reported a mixed result
- Seven studies reported positive effects regarding measures of skin aging.
Then the researchers narrowed their focus to the highest quality studies that were not funded by industry. There were only five of those studies.
Four out of five studies still reported positive results.
Therefore, the vast majority of the best studies demonstrated beneficial effects, yet the pooled subgroup analysis concluded there was no benefit.
That is the central puzzle.
How Statistical Issues Can Produce a Misleading Pooled Analysis
1. Different studies measure skin in different ways
One study may assess hydration through transepithelial water loss. Another study may measure hydration using electrical conductivity. While both methods are acceptable, they represent two different metrics.
Meta-analyses will typically use a process called standardization to combine different measurements into a single scale. While this provides the ability to place disparate measurements on one scale, it comes at the cost of stripping context from the measurements. Additionally, the process of standardizing effects can make it difficult to understand the nature of the differences observed among studies.
2. Confidence intervals can eliminate statistical significance when the effect sizes of the studies are largely positive.
When studies demonstrate a range of effects, the average effect size of the studies is less precise. Therefore, the confidence intervals of the pooled estimates expand.
This creates a strange outcome; most studies can produce positive results. Yet, the pooled confidence intervals can be narrow enough to barely include zero. This produces the outcome of "not statistically significant" even when most of the plausible values in the interval are positive.
This is exactly what occurred in the paper under discussion.
For example, in one subgroup, the effect size was positive, but the confidence interval ranged from a negative number to a positive number. The reason the authors characterize the finding as "no evidence of benefit" is that the confidence interval includes zero. Statistically speaking, this is true. Practically speaking, however, it can be misleading. If most of the plausible values in the confidence interval are positive, and most of the individual studies demonstrate a positive effect, it may be misleading to conclude that the effect is zero.
This represents a classic scenario where "not statistically significant" is interpreted as "has no effect," despite the fact that the data are consistent with a meaningful positive effect.
Additional Mechanistic Evidence Strengthens the Case for Real Effects
While the clinical trial results provide evidence that collagen peptides can positively affect skin health, additional mechanistic evidence exists that supports the possibility that collagen peptides can produce real effects.
Laboratory studies have shown that fibroblasts treated with collagen peptides experience an increase in the production of collagen, elastin, and proteoglycans, which play a role in maintaining hydration.
Additionally, human trials have demonstrated that collagen peptides stimulate the release of relevant peptides into the bloodstream post-ingestion. Data from animals and humans also support the idea that these peptides can enter the skin.
Mechanistic evidence alone does not validate clinical effectiveness. However, when combined with repeated positive trial results, it increases our confidence that the observed effects are real.
The One Obvious Limitation of Collagen Research
One obvious limitation of the current state of collagen research relates to the issue of comparison. Most collagen trials have compared collagen peptides to a placebo. Comparing collagen peptides to a matched protein control has rarely been done.
Without controlling for protein intake, it is difficult to determine if the positive results obtained are unique to collagen peptides or simply due to the added protein.
There is at least one study demonstrating that collagen peptides produced more rapid wound healing than protein in burn patients. Therefore, it appears that the peptides themselves may possess some mechanism of action independent of standard protein.
Nevertheless, we require a clean trial demonstrating that collagen peptides are superior to a matched protein control for skin aging outcomes.
Such a study would allow us to resolve the debate definitively.
The practical takeaways
There are three key takeaways based on this case study.
- Don't Rely On Abstracts
Abstracts tend to minimize complexities of the original research and may create conclusions which do not accurately reflect the full scope of the study. - Pooling Statistics Can Be Misleading When Studies Are Heterogeneous
If virtually all of the high-quality individual studies reveal benefits; however, the pooled subgroup statistic barely meets or fails to cross the statistical threshold; the correct interpretation should be to recognize the uncertainty as opposed to the dismissal of the treatment. - Overall Body Of Evidence Supports Benefit With Collagen
Most of the studies examined demonstrate improvements in skin hydration, elasticity, and wrinkling; as well as many which were high-quality and not funded by an industry entity. Additionally, the mechanistic data supporting collagen peptide supplementation demonstrates plausibility.
Collagen Peptide Supplements Also Show A Strong Safety Profile.
In Summary:
This new analysis did not provide sufficient proof to conclude that collagen has no value. Instead, it demonstrated that when the large dataset of studies is sliced into smaller subgroups and the various measures are pooled together, statistically significant results are lost, although most individual studies demonstrated positive outcomes.
Better Conclusion
Collagen peptide supplements very likely improve skin hydration, elasticity, and wrinkles, but further, high-quality research, particularly studies that compare collagen peptide supplementation to a matched control group (i.e., protein), will enhance the confidence level of the effect size and make the true effects of collagen peptide supplementation clearer.
For a larger overview of the scientific literature regarding collagen peptide supplementation, please refer to the following article: What Science Really Says About Collagen Peptides.
If you are supplementing with collagen, the existing evidence suggests that it can be a relatively safe and useful adjunct to promote healthy skin, especially if you incorporate fundamental science-based skincare practices (e.g., sunscreen/UV protection, adequate sleep, regular resistance training).