Does Menopause Really Take Thirty Percent of Your Collagen?

The number is everywhere. Women lose about 30 percent of their skin collagen in the first five years after menopause, and roughly 2 percent a year after that. It appears in dermatology explainers, supplement copy, morning television segments, clinic handouts and the science pages of skincare brands. It is usually stated flat, with no citation, or with a citation that turns out to be another article stating it flat.

An accountant reads a number like that and wants to know who measured it, in how many people, and by what method. So I went looking for the study.

Where does the thirty percent figure actually come from?

The paper most often credited is Brincat and colleagues, published in Obstetrics and Gynecology in December 1987, under the title "A study of the decrease of skin collagen content, skin thickness, and bone mass in the postmenopausal woman."¹ Its abstract does not contain the figure. What it reports is that skin collagen content, skin thickness, metacarpal index and forearm bone mineral content all declined by between 1 and 2 percent per year after menopause, and that the four measures correlated with each other.¹ One to two percent a year is not 30 percent in five years. It is closer to 5 to 10 percent in five years.

The paper is not open access, so I cannot rule out that a larger figure sits somewhere in the body text. I can say that the abstract of the study everybody points to reports a different, smaller number than the one being attributed to it.

The bigger number enters somewhere in the reviews, so that is where I went next.

What did the papers underneath it actually measure?

The most influential modern statement of the claim is a 2013 narrative review in Dermato-Endocrinology by Julie Thornton, which reads: "Type I and III skin collagen is thought to decrease by as much as 30% in the first five years after menopause."² The hedge is doing real work there. The sentence carries two reference numbers, and I pulled both.

The first is a 1985 paper in the British Journal of Obstetrics and Gynaecology on the long-term effects of menopause and sex hormones on skin thickness.³ It compared postmenopausal women treated with sex hormone implants against untreated women. Its abstract reports that collagen content and thickness were both significantly greater in the treated group, and that in untreated women collagen declined in relation to menopausal age rather than chronological age. It states no percentage. Its abstract does not report a sample size.

The second is Affinito and colleagues in Maturitas in 1999.⁴ That study had 32 women. Fourteen premenopausal, eighteen postmenopausal. Skin biopsies taken during open abdominal surgery, with collagen type I and type III measured by immunohistochemistry and computerised image analysis. It reports correlation coefficients between collagen percentages and years since menopause. It does not report a 30 percent five-year loss.

Two references. Neither contains the number they are cited for.

Do other sources point anywhere better?

They point at each other. A 2006 review in Clinical Interventions in Aging carries the sentence in almost its final form: approximately 30 percent of skin collagen lost in the first five years after menopause, with an average decline of 2.1 percent per postmenopausal year over 20 years.⁵ That sentence has no citation attached to it at all. The nearest reference belongs to the following sentence and is a review article on human skin as an endocrine organ, which is not a study of collagen loss over time.⁵ The same reference list dates a foundational skin collagen paper to 1985 when the digital object identifier embedded in the entry shows 1975.⁵ That is a small error, and it is the kind of small error that tells you a citation chain is being copied rather than read.

Elsewhere the figure is attributed to a 2005 review in Climacteric by Brincat, Baron and Galea.⁶ That paper describes its own method plainly: a Medline and PubMed search on the keywords collagen, elastin, estrogen, hormone replacement therapy, skin and aging.⁶ It is a literature review, not a measurement. So the chain runs review to review, and where it finally touches primary data, the primary data says something else.

I cannot trace the 30 percent figure to an original measurement. If it exists in a primary source, it is not in any of the papers the claim is routinely hung on.

Is there any real measurement of collagen loss with age?

Yes, and it is old. In 1975, Shuster, Black and McVitie measured forearm skin collagen, dermal thickness and collagen density in what they described as a large number of normal subjects, explicitly to establish a reference standard.⁷ Skin collagen decreased with age and was lower in women at all ages. They also warned that collagen density varies in disease, which limits using dermal thickness as a proxy for collagen content.⁷ That caveat matters, because a great deal of the menopause skin literature measures thickness and then talks about collagen.

What survives all of this is modest and defensible. Skin collagen declines with age and declines faster in women, and several small studies find the decline tracks years since menopause better than birthday age. The size of the drop in the first five years is not established.

What does estrogen actually do to skin in a trial?

The randomized evidence is thinner than the enthusiasm around it, and it is old. The best-known trial is Maheux and colleagues in the American Journal of Obstetrics and Gynecology in 1994: 60 postmenopausal women, randomized double blind to conjugated estrogens or placebo for 12 months, with skin thickness measured by ultrasound and dermal changes assessed on biopsy.⁸ Skin thickness and dermal thickness both increased significantly at the right great trochanter in the treated group and not in the control group.⁸

The endpoint was thickness of hip skin, measured by ultrasound and biopsy, in 60 women, thirty years ago. That is a real finding, and it is a different finding from a face that looks different in a mirror.

What happened when someone ran a big trial on wrinkles?

Somebody did run that trial, and the result is the part of this literature that rarely gets quoted. In 2008, Phillips and colleagues published a 48-week randomized, double blind, double dummy, placebo controlled multicenter study in the Journal of the American Academy of Dermatology.⁹ It enrolled 485 postmenopausal women, randomized to placebo, to norethindrone acetate with 5 micrograms of ethinyl estradiol, or to the same with 10 micrograms.⁹ The primary endpoints were investigator global assessment of coarse and fine facial wrinkling at week 48 and the women's own assessment of change.

There were no statistically significant differences between either hormone arm and placebo on the primary endpoints.⁹ The authors noted the limitation themselves: this tested low-dose estrogen for 48 weeks in women who were on average five years past menopause, and higher doses, longer treatment or perimenopausal women cannot be extrapolated from it.⁹

So the strongest single trial with an appearance endpoint, and by a wide margin the largest, was null. That is the trial that should be quoted next to the 30 percent number, and it almost never is.

Where does hormone therapy officially stand on skin?

The Menopause Society's 2022 hormone therapy position statement, still the current one, includes a section on skin, hair and special senses. It says estrogen therapy may benefit wound healing by modifying inflammation, stimulating granulation tissue formation and accelerating re-epithelialization, and that estrogen therapy increased epidermal and dermal thickness, increased collagen and elastin content and improved skin moisture, with fewer wrinkles.¹⁰ Its formal key point is graded Level II: estrogen therapy appears to have beneficial effects on skin thickness and elasticity and collagen when given at menopause.¹⁰

That entire passage carries a single reference. Reference 144 is the 2013 Thornton narrative review.¹⁰ The same review that says the 30 percent figure is "thought to" be true and cites two papers that do not report it. A professional society's Level II statement on skin is resting on the same citation chain.

The regulatory position is unambiguous by comparison. The FDA prescribing information for oral conjugated estrogens lists six indications: moderate to severe vasomotor symptoms due to menopause, moderate to severe vulvar and vaginal atrophy due to menopause, hypoestrogenism due to hypogonadism, castration or primary ovarian failure, palliative treatment of metastatic breast cancer in selected patients, palliative treatment of advanced androgen-dependent prostate cancer, and prevention of postmenopausal osteoporosis.¹¹ Skin is not on that list. The words wrinkle, collagen and skin aging do not appear anywhere in the label.¹¹

The label also contraindicates the drug in anyone with breast cancer or a history of breast cancer, except in selected patients being treated for metastatic disease.¹¹ I am a breast cancer survivor, so I will state that part without decoration: for many women with a breast cancer history, systemic hormone therapy is contraindicated or complicated, which closes this conversation before it opens, and the skin question then has to be answered some other way.

The regulatory picture is also moving. On 10 November 2025 the FDA informed manufacturers of menopausal hormone therapies that it was requesting labeling changes, including proposed removal of risk statements about cardiovascular disease, breast cancer and probable dementia from the boxed warnings.¹² As of the prescribing information version dated June 2026, the boxed warning on oral conjugated estrogens still names endometrial cancer, cardiovascular disorders, breast cancer and probable dementia.¹¹ An announced label change and a changed label are not the same object. Neither of them makes skin an indication.

What changes at menopause that is not collagen?

Quite a lot, and the non-collagen changes are better characterized than the famous one. A 2022 review in Climacteric by Zouboulis and colleagues sets out the list: dryness and pruritus, thinning and atrophy, wrinkles and sagging, poor wound healing and reduced vascularity in the skin and mucosa.¹³ On the scalp, reduced hair growth and density, presenting as diffuse effluvium from follicular rarefication or as female pattern hair loss, plus altered hair quality and structure and increased unwanted hair growth on the face.¹³

The same review makes a claim that cuts against most of the marketing built on this subject. Skin premalignant and malignant lesions and skin aging signs are, in its words, almost exclusively caused by environmental factors, especially solar radiation.¹³ And its position on treatment is explicit: hormone replacement therapy is not indicated for skin and hair symptoms alone, given the risk and benefit balance.¹³ The Menopause Society reaches the same place on hair from the other direction, noting that hair density and female pattern hair loss worsen after menopause but that research on whether hormone therapy mitigates it is lacking.¹⁰

How much of a changing face is even skin?

This is the finding that reframed the subject for me. In 2019, Windhager and colleagues published a geometric morphometric study in the American Journal of Physical Anthropology using three-dimensional surface scans of 88 faces, aged 26 to 90, from Split in Croatia and neighbouring islands, with 585 measurement points per face.¹⁴

Facial shape change followed a similar path in both sexes until around age 50, at which point the female trajectory turned sharply, and the rate of change was higher in women, especially in early postmenopause.¹⁴ In postmenopausal women, facial aging was best predicted by years since last menstruation, and was mainly attributable to bone resorption in the mandible.¹⁴

Bone, not collagen. The jaw resorbing changes the scaffold the face is draped over, which shows up as a softened jawline and deepened folds. Eighty-eight people in one region is a small, geographically narrow sample and I would not build a doctrine on it. But it is a direct three-dimensional measurement, and it suggests that a meaningful share of what gets sold as a collagen problem is a skeletal one, which no cream and no supplement addresses.

What actually has trial evidence behind it?

Two things, neither of them hormonal, and both of them boring.

• Daily sunscreen. In a randomized community trial in Nambour, Australia, 903 adults under 55 were assigned to daily broad spectrum sunscreen or discretionary use, and after 4.5 years the daily group showed no detectable increase in skin aging measured by microtopography of skin casts, with 24 percent less aging than the discretionary group, an odds ratio of 0.76 with a 95 percent confidence interval of 0.59 to 0.98.¹⁵
• Topical tretinoin. In a 1993 New England Journal of Medicine study, collagen I formation was 56 percent lower in photodamaged forearm skin than in sun-protected buttock skin from the same subjects, and 10 to 12 months of daily 0.1 percent tretinoin cream produced an 80 percent increase in collagen I formation against a 14 percent decrease with vehicle.¹⁶

Both have limits worth stating. The sunscreen trial was conducted at 26 degrees south in a high ultraviolet environment, its participants were under 55, and its own authors flagged missing outcome data and modest power.¹⁵ The tretinoin study is a mechanistic trial with 15 subjects on active treatment and 14 on vehicle, measuring collagen formation immunohistologically, not a study of how anyone looked.¹⁶ Neither was run in postmenopausal women specifically. What they share is something the hormone literature mostly lacks: a randomized design with a physical measurement and an effect size you can write down.

Do oral collagen supplements do anything?

The funding question earns its keep here. In 2025, Myung and Park published a systematic review and meta-analysis in the American Journal of Medicine covering 23 randomized controlled trials with 1,474 participants, and they did something most reviews of this literature do not: they split the results by funding source and by study quality.¹⁷

Pooled across all 23 trials, collagen supplements significantly improved skin hydration, elasticity and wrinkles. In the subgroup of trials not funded by pharmaceutical companies, there was no effect on hydration, elasticity or wrinkles. In the industry-funded subgroup, the effects were significant. High quality studies showed no significant effect in any category. Low quality studies showed improvement in elasticity.¹⁷ The authors' conclusion is one sentence: there is currently no clinical evidence to support the use of collagen supplements to prevent or treat skin aging.¹⁷

It is worth seeing what a typical positive trial looks like up close. A widely cited 2014 study in Skin Pharmacology and Physiology randomized 69 women aged 35 to 55 into three arms of 23 each, giving 2.5 grams or 5.0 grams of collagen hydrolysate or placebo daily for 8 weeks.¹⁸ The primary endpoint was instrument-measured skin elasticity, which improved significantly against placebo in both dose groups. Skin moisture and transepidermal water loss moved in a favourable direction in a subgroup analysis but did not reach statistical significance.¹⁸

Twenty-three women per arm. Eight weeks. A cutometer reading rather than an appearance endpoint. A subgroup analysis carrying the hydration story. That is not fraud, it is just a small study with a soft endpoint, and a market has been built on a stack of them.

So what is the honest version?

Skin collagen declines with age, and it declines faster in women, and several small studies suggest the decline tracks years since menopause more closely than chronological age. The specific claim that 30 percent of it goes in the first five years has no traceable primary measurement behind it. The study most often credited reports 1 to 2 percent per year in its own abstract. The reviews that popularised the bigger figure cite each other, cite reviews, or cite papers that measured something else in fewer than three dozen women.

Estrogen does act on skin. The Menopause Society grades that at Level II on the strength of a narrative review, and the FDA has never approved a hormone product for it. The one large randomized trial with wrinkles as a primary endpoint found nothing. Whether to take hormone therapy is a decision between a woman and her clinician, made on symptoms and risk, and skin is not a reason to start it or a reason to stop it.

What is left is unglamorous and cheap. Ultraviolet exposure is the dominant driver of the visible changes people attribute to hormones. Sunscreen and retinoids have randomized trials with measurable effects. Oral collagen has an effect that disappears when you remove the industry-funded studies. And a good share of what people see in the mirror after 50 may be the jaw, not the dermis.

The 30 percent figure survives because it is useful. It converts a slow, poorly characterized biological process into a countdown with a deadline, and a countdown sells product. The number does not have to be true to do that job, only repeated, and it has been repeated for forty years without anyone stopping to ask where it came from.