A paper published in PNAS is titled "Menopause accelerates biological aging." Inside it, the single analysis that compares women of the same age on either side of menopause found no detectable difference.
There is a claim that has become close to furniture in longevity writing: "menopause accelerates biological aging". It is repeated with the confidence of something measured. It is worth asking what was actually measured, because the answer changes what you should do with the claim.
This is the kind of question women's health research exists to answer, and it is also a good test of whether the field is answering it.
Why This Matters
Half the population goes through menopause. If the transition is an aging accelerator in the strict sense, that is a specific and actionable fact: there is a window, and interventions inside it should behave differently than interventions outside it. If instead menopause is a visible marker of an aging process already underway, the advice is different and the window framing is misleading.
Those two stories make identical predictions in a snapshot. They only separate when you follow the same women through.
How To Read The Numbers In This Article
This piece turns on a handful of statistics, and they are the kind that get misreported constantly. Four minutes here will make the rest of it readable.
A hazard ratio compares how fast events piled up in two groups. A hazard ratio of 1.29 means that at any point during the study, the treated group was having events about 1.29 times as often as the comparison group. It is a ratio of rates, not a count of people. On its own it tells you nothing about how many extra people that is, so where the trial published absolute figures they are given alongside the ratio, in events per 10,000 person-years. A ratio with no absolute figure beside it should not move a decision on its own.
1.0 is the no-difference point for a ratio. A ratio above 1.0 means more events in the treated group, below 1.0 means fewer. When a number is a difference rather than a ratio, in years or millimetres or percent per year, the no-difference point is zero instead.
The bracketed range after a ratio is the set of values compatible with the data. When you see "1.29 (1.02 to 1.63)", that range is where the true value plausibly sits, assuming the study itself is sound. If the range includes 1.0, then "no difference" is still one of the values that fits. The strictly correct meaning is a property of the method rather than of this one range: repeat the whole study many times and about 95 percent of the ranges built this way would contain the true value. What that phrasing gives up is that the middle of a range fits far better than its edges, and that no range can detect bias.
A p-value is the chance of seeing a gap at least this large if there were genuinely no effect. A p of .02 means a gap this big would turn up about 2 percent of the time in a world where nothing was going on. It is not the probability the finding is real, not the probability it was a fluke, and it says nothing about how big the effect is. Several p-values below sit just either side of the usual cutoffs, which is exactly why the cutoffs deserve so little weight.
"No detectable difference" is not "no difference." A study that finds nothing may have been too small to find something. Where that matters below, the size of the study is given so you can judge how much could have been hiding inside a null. The reverse also holds: when a range is narrow and sits on 1.0, that genuinely rules out a large effect, which is a real result and not a shrug.
An epigenetic clock is a laboratory score, not a measurement of aging. It reads chemical tags on DNA and returns an estimated age in years. The first clocks were built to track calendar age; later ones, including GrimAge and PhenoAge, were trained to predict illness and death, and they do predict it. What nobody has shown, for any of them, is that moving the number changes how long or how well anyone lives. Every time a clock number appears below, it is a score.
What A Biological Clock Actually Measures
An epigenetic clock reads methylation marks at a few hundred sites on the genome and returns a number in years. Subtract chronological age and you get "epigenetic age acceleration": a positive number means the tissue looks older than the birthday.
The word "accelerates" in a title is doing specific work. It asserts a rate of change. To measure a rate you need at least two measurements of the same thing at different times. This distinction is the whole article, so it is worth being precise about three separate claims that get collapsed into one:
(a) Postmenopausal women have older clocks than premenopausal women of the same chronological age. (b) The same women, measured before and after, show their clocks speed up through the transition. (c) Menopause causes that acceleration, rather than sharing a common cause with it.
Only (b) is a measured rate. Only (c) supports the interventional framing. Here is where each one actually stands.
(a) The Snapshot, And The Null Inside The Famous Paper
Levine and colleagues published the 2016 PNAS paper that anchors this literature. It pooled four cohorts: 1,864 women from the Women's Health Initiative, 200 from InCHIANTI, 256 from PEG, and 790 from the UK National Survey of Health and Development. Every one of them had DNA methylation measured at a single timepoint. The authors say so plainly: "our cross-sectional data make it difficult to dissect causal relationships."
The blood findings are real but they are not what most people think. In the WHI subsample, each additional year of age at menopause was associated with 0.063 years lower epigenetic age acceleration. Time since menopause went the other way, at 0.038 years per year. Both of those are comparisons among women who were already postmenopausal. Neither is a comparison of a postmenopausal woman with a premenopausal woman.
The paper does contain exactly one analysis of that second kind, and it is the cleanest natural experiment in the whole dataset. In the NSHD sample, all 790 women were sampled at exactly 53 years old. Roughly 469 were postmenopausal and 321 were not. Same chronological age, different menopausal status, one clock.
The result, verbatim: "no significant association was found between AgeAccel and menopausal status at age 53 y."
The paper also reports that age at natural or surgical menopause "did not correlate with epigenetic AgeAccel" across those postmenopausal buccal samples.
That null sits inside the paper whose title is the claim. It is not hidden, it is not a footnote, and it is almost never quoted.
(b) The Longitudinal Measurement Does Not Exist
This is the load-bearing finding, and it is a negative one.
A complete search of the indexed literature, intersecting epigenetic clock terms (epigenetic age, epigenetic clock, GrimAge, PhenoAge, DunedinPACE) with menopause terms (menopause, oophorectomy, ovarian insufficiency), returns 35 papers. Not one of them measures DNA methylation repeatedly in the same women spanning premenopause to postmenopause.
SWAN, the Study of Women's Health Across the Nation, is the obvious place this should have been done. It has followed thousands of women through the transition since the 1990s with repeated blood draws. It has no published epigenetic clock analysis at all. A search returns eight hits and all eight are false positives: the "SWAN" array normalization algorithm, and the unrelated Swan 71 cell line.
There is one genuine within-woman longitudinal result, and it does not use an epigenetic clock. Xiang and colleagues, in BMC Medicine in 2025, tracked women who crossed from pre- to postmenopausal during follow-up and found greater increases in Klemera-Doubal biological age, which is computed from ordinary clinical blood chemistry. In the CMEC cohort women who crossed the transition gained 1.33 extra years of Klemera-Doubal biological age (95% CI 0.89 to 1.76 years) across 684 transitioners over a median 2.01 years; in UK Biobank the gain was 2.60 years (1.91 to 3.30) across 223 transitioners over 4.50 years. These are differences in years rather than ratios, so here the no-difference point is zero, not 1.0, and both ranges sit clear of it.
That is real evidence, and it points the way the popular claim does. It is also a different measurement than the one people cite, on short follow-up, in modest numbers.
(c) The Causal Claim Is Thinner Now Than In 2016
Levine's causal argument rested on a Mendelian randomization using two genetic variants. One reached P = 0.031. The other gave P = 0.763. There was no inverse-variance-weighted analysis, no MR-Egger, no multi-instrument approach. A single nominally significant result out of two tests would not survive correction for those two tests.
The paper's other genetic support is a correlation between age at menopause and epigenetic age acceleration of rG = -0.256, which the authors themselves describe as "marginally significant" at a one-sided P of 0.054. Two-sided that is roughly 0.108. The abstract's phrasing, "we find evidence of coheritability," is a stronger sentence than the number underneath it.
Since then the question has been asked properly. Wang and colleagues ran a dedicated bidirectional Mendelian randomization with full instrument sets and instrument F-statistics from 23.99 to 651.62. They found no evidence of a causal effect of any of six DNA methylation aging measures on age at menopause. In the reverse direction only granulocyte proportion reached significance, at a coefficient of 0.0010.
The best prospective test of the reverse direction comes from CARDIA. Appiah and colleagues measured 583 premenopausal women in 2000 and 2001, mean age 41.2, and followed them to menopause through 2021. Baseline GrimAge acceleration predicted menopause about 0.12 years earlier per year of acceleration, roughly six weeks, and the association was not significant after adjustment.
Meanwhile the genetics of age at menopause is dominated by DNA damage response biology. Ruth and colleagues identified 290 loci in about 200,000 women, implicating DNA repair machinery. That is a mechanism for ovarian aging being downstream of general cellular maintenance, not upstream of it.
What Is Measured, Beyond Argument
None of the above says the transition is uneventful. The physiology has been tracked longitudinally, against each woman's own final menstrual period, and it is dramatic. This is where the confident language belongs.
Hormones. In 1,215 SWAN women with 9,435 hormone measurements, FSH starts rising 6.10 years before the final period, accelerates 2.05 years before it, and stabilizes 2.00 years after. Estradiol has a much narrower window: it does not change until 2.03 years before the final period, falls fastest at the final period itself, and is stable 2.17 years after. Population mean estradiol goes from 54.08 pg/mL two years before to 30.01 at the final period to 18.35 two years after.
So the endocrine transition runs about eight years for FSH and about four for estradiol. It is not an event.
Bone. Spine and femoral neck density are flat from five years before the final period to one year before: lumbar spine loss over that interval is -0.02 percent per year, with a confidence interval spanning zero. Then it falls off a cliff. Through the transmenopause, defined as one year before to two years after the final period, lumbar spine density falls 2.46 percent per year and femoral neck 1.76 percent per year. Cumulative ten-year loss is 10.6 percent at the spine, and 7.38 of those 10.6 percentage points happen inside that three-year window: about seven-tenths of the decade's loss in under a third of the time.
Body composition. The rate of fat gain rises about 70 percent, from 1.0 percent per year before the transition to 1.7 percent during it, or 0.25 kg a year to 0.45 kg. Lean mass reverses, from gaining 0.2 percent per year to losing 0.2 percent per year. (SWAN's own abstract says the rate of fat gain "doubled". Its own numbers say 70 percent. That is a small thing, and it is the same species of overstatement this article is about, so it is worth naming rather than repeating.) And a detail that contradicts the usual telling: weight itself does not accelerate. SWAN's own wording is that weight "climbed linearly during premenopause without acceleration at the MT." What changes is the composition, not the number on the scale.
Lipids. Total cholesterol rises 2.98 mg/dL per year before the final period, then 6.47 mg/dL per year in the twelve months around it, then flattens to -0.16. LDL goes 1.57, then 5.20, then 0.14. Apolipoprotein B goes 0.67, then 3.24, then -0.40. Each of those middle segments is significantly steeper than both of its neighbours.
HDL cholesterol is the interesting one, because it goes the wrong way for the usual story. It rises fastest before the final period, slows during it, and declines afterwards. Its cardioprotective function deteriorates even while the number looks acceptable, which is a reminder that a lipid panel is a proxy and not the thing itself.
Triglycerides, glucose, fibrinogen and blood pressure fit a straight line through the whole period. Those track chronological aging, not the transition. Separating the two is the same discipline that distinguishes insulin sensitivity from longevity: a marker that moves with age is not thereby a marker of the thing you care about.
So the honest summary is that the menopause transition has a sharp, measurable, roughly three-year signature in bone, body composition and atherogenic lipids, and no distinct signature in several other risk factors. Whether an epigenetic clock registers that signature is, at the moment, unmeasured.
The Timing Hypothesis, Read From The Trials
The other half of this topic is hormone therapy, and specifically the idea that the WHI was misread and that starting early changes the answer. Some of that is right. The confident version is not.
Start with what the WHI found. In the estrogen-plus-progestin arm, 16,608 women with a mean age of 63.3 at entry, coronary heart disease came out at a hazard ratio of 1.29 with a nominal confidence interval of 1.02 to 1.63. Adjusted for multiple comparisons, that interval is 0.85 to 1.97 and crosses one. Invasive breast cancer was 1.26, nominal 1.00 to 1.59, adjusted 0.83 to 1.92. Venous thromboembolism was 2.11 and stayed significant after adjustment, at 1.26 to 3.55. Fractures went down: total fractures 0.76, adjusted 0.63 to 0.92. In the trial's own absolute terms, estrogen plus progestin produced 7 more coronary events, 8 more strokes, 8 more pulmonary embolisms and 8 more invasive breast cancers per 10,000 person-years, against 6 fewer colorectal cancers and 5 fewer hip fractures. Doubling a rare event leaves it rare.
All-cause mortality was 0.98, and the paper's own conclusion is that "all-cause mortality was not affected during the trial."
The estrogen-alone arm gave different answers, and this is the single most inverted fact on this beat. The two arms are not interchangeable, and collapsing them into "hormone therapy causes breast cancer" or "hormone therapy does not" is wrong in both directions.
Now the reanalysis. Rossouw and colleagues in 2007 stratified by age and by years since menopause. Coronary heart disease by years since menopause gave 0.76, then 1.10, then 1.28, with P for trend of .02.
Three things about that number are almost never said together.
First, the paper set its own significance threshold at P less than .01, "to partially account for multiple testing issues and the post hoc nature of some of the tests." The abstract's conclusion states that the trend "did not meet our criterion for statistical significance."
Second, the paper ran 137 statistical tests. Two were significant. One or two were expected by chance.
Third, and most important, the age stratification was prespecified in the WHI protocol and the years-since-menopause stratification was not. The variable that generates the timing hypothesis is the one that was chosen after the fact.
Then look at which end of the gradient is doing the work. The nearest stratum, under ten years since menopause, has a confidence interval of 0.50 to 1.16, which includes one. The most distant stratum, twenty or more years, has 1.03 to 1.58, which does not. The gradient is produced by the excess of events far from menopause, not by protection near it, and even that far stratum clears only the ordinary one-in-twenty cutoff, not the stricter one-in-a-hundred bar the paper set for itself. "Coronary harm accrues with distance" and "coronary benefit accrues with proximity" are different claims, and only the first one has statistical support here. The nearest stratum still points toward benefit, at 0.76, but a range running from 0.50 to 1.16 covers everything from a halving of risk to a modest harm, which is too wide to call either way.
Stroke makes this sharper. Hormone therapy raised stroke risk overall, hazard ratio 1.32 (1.12 to 1.56), and the risk did not vary significantly by age (P for trend .97) or by years since menopause (P for trend .36). The point estimate was highest in the women closest to menopause, 1.77 versus 1.23 and 1.26. With P for trend at .36 and 137 tests in the paper, that is not evidence of an early-window stroke penalty either. What it does rule out is the claim that early initiation abolishes stroke risk.
For all-cause mortality, the asymmetry is complete. By age, P for trend was .06. By years since menopause, it was .51. The timing hypothesis, stated as a claim about years since menopause, has no mortality support in that paper at all.
The 13-year and 18-year follow-ups did not rescue it. In Manson 2013, coronary heart disease, the trial's primary cardiovascular endpoint, had no significant timing interaction in either arm, at P for trend 0.08 both times. In Manson 2017, across 27,347 women and 7,489 deaths over 18 years, all-cause mortality was 27.1 percent on hormone therapy and 27.6 percent on placebo. Hazard ratio 0.99, confidence interval 0.94 to 1.03. Cardiovascular, cancer and other-cause mortality were all null.
There is one genuine signal in favour of an age effect, and it should be reported precisely: in the estrogen-alone trial, the global index of monitored events was more favourable in younger women, and produced 19 fewer adverse events per 10,000 person-years at ages 50 to 59. In the combined-therapy arm, which is what a woman with a uterus would be prescribed, no age gradient was detected at all (P for trend above .99), and net harm was positive at every age band.
The Two Trials Built To Test This
ELITE and KEEPS were designed specifically to test early initiation, and both used imaging surrogates rather than clinical events.
ELITE found a significant interaction by stratum for carotid intima-media thickness, at P = 0.007. The effect size was 0.0034 mm per year. Over five years that is roughly 17 micrometres of avoided arterial wall thickening, against a baseline mean thickness of 0.75 mm. The trial's sample size was computed for that interaction, not for events, and its only clinical event data are adverse-event tallies of three versus one myocardial infarctions and two versus three clots, none significantly different.
Inside the same trial, the coronary CT endpoints, calcium score, calcium presence, stenosis and plaque, were null in both strata, with interaction P values of 0.36, 0.29, 0.83 and 0.35. In the early stratum they pointed numerically the wrong way.
KEEPS was null on its primary endpoint.
A change in carotid wall thickness is not a prevented heart attack, and no trial has shown that it converts into one. This is the most common inferential overreach on this topic, and it survives because the alternative sentence is less satisfying.
What This Leaves You With
The transition is real, measurable and worth taking seriously, especially for bone. Spine density falling 2.46 percent per year for three years is not a subtle effect, and it is the strongest argument in this whole area for acting rather than waiting.
Hormone therapy has clear indications. The 2022 North American Menopause Society position is that for women under 60 or within ten years of menopause onset, without contraindications, "the benefit-risk ratio is favorable for treatment of bothersome vasomotor symptoms and prevention of bone loss," and less favourable beyond that window. Contraindications include unexplained vaginal bleeding, liver disease, prior estrogen-sensitive cancer, and prior cardiovascular events or clots. For premature ovarian insufficiency the guidance is different and stronger: therapy is recommended at least until the average age of menopause, and the WHI results in older women do not apply.
Hormone therapy is not recommended at any age to prevent cognitive decline. In the WHI Memory Study, in women 65 and older, probable dementia went up, not down: 40 cases versus 21, hazard ratio 2.05 (1.21 to 3.48).
What the evidence does not support is the sentence that started this article. Menopause has not been shown to accelerate an epigenetic clock, because nobody has run one across the transition. The claim is an inference from snapshots, and the one snapshot designed to test it directly came back null.
That is not a reason to dismiss the transition. It is a reason to be precise about which parts of it are measured and which are asserted, because those two categories get different amounts of your attention. The same discipline applies to sex differences in drug response, where a confident mechanism turned out to be a search summary pointing the wrong way, and to biological age testing generally, where a number in years invites more confidence than the method has earned.
Frequently Asked Questions
Does menopause make you age faster?
The physiology changes sharply and measurably: bone, body composition and atherogenic lipids all shift on a roughly three-year schedule around the final period. Whether an epigenetic clock speeds up through the transition has not been measured, because no study has run one on the same women before and after. The claim that it does is an inference from snapshots, and the one snapshot designed to test it directly returned a null.
Is an epigenetic age test worth taking around menopause?
There is no evidence base for using one to make a menopause decision. Every clock finding in this area is cross-sectional, the causal direction is unresolved, and a purpose-built Mendelian randomization found no evidence of a causal effect in either direction. A biomarker panel with an established action attached to it, such as a lipid panel or a DXA scan, will change what you do. A number in years currently will not.
Was the Women's Health Initiative wrong about hormone therapy?
It was widely over-read in both directions. The trial's own findings stand: clots roughly doubled off a low base, an excess of about 8 pulmonary embolisms per 10,000 person-years; fractures fell; and at 18 years deaths ran at 27.1 percent on hormone therapy versus 27.6 percent on placebo, a compatible range narrow enough to rule out any large effect on survival in either direction. What is not supported is the popular correction, that early initiation reverses the picture. The years-since-menopause analysis behind that idea was not prespecified, missed the paper's own significance threshold, and is driven by harm far from menopause rather than benefit near it.
Should I take hormone therapy for prevention rather than symptoms?
Current society guidance ties it to indications: bothersome vasomotor symptoms, genitourinary symptoms, and prevention of bone loss, for women under 60 or within ten years of menopause onset and without contraindications. It is not recommended at any age to prevent cognitive decline, where the randomized evidence in older women points the other way. The picture is different and stronger for premature ovarian insufficiency.
What actually matters most in the transition?
Bone, on the current evidence. Spine density is flat until a year before the final period and then falls 2.46 percent per year for about three years, which is when most of the decade's loss happens. That is a narrow window with a measurable outcome, which is more than can be said for most of what gets discussed under this heading. Resistance training and protein intake are the levers that do not require a prescription.
Funding Transparency
This article received no funding. Longevity Science Daily has no commercial relationship with any manufacturer of hormone therapy, epigenetic age testing, or supplements referenced here. No sources were compensated. The author has no financial interest in any product or company mentioned.
Every figure above was read in the primary publication or its abstract on the journal or PubMed page. Nine claims were dropped during research because their primary source could not be resolved or because the number appeared only in secondary reporting, and ten more were removed after an adversarial verification pass. Those are recorded with reasons in the research notes for this article.
Related Reading
- Why Women's Health Is Longevity Science
- The Diabetes Drug That Extended Lifespan in Males and Did Nothing in Females
- Healthspan vs Lifespan
- Omega-3s and Biological Aging
- Skin Aging Science
- Your Body's Biomarkers
- The Hallmarks of Aging
Written with the help of AI tools, shaped and verified by humans. Every number in this article was checked against the primary publication rather than a summary of it.
This article is for information only and is not medical advice. Decisions about hormone therapy depend on your own history, symptoms and risk factors, and should be made with a clinician who knows them.
Sources
- Levine ME, Lu AT, Chen BH, et al. Menopause accelerates biological aging. Proc Natl Acad Sci USA. 2016;113(33):9327-9332. DOI: 10.1073/pnas.1604558113
- Writing Group for the Women's Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women. JAMA. 2002;288(3):321-333. DOI: 10.1001/jama.288.3.321
- Anderson GL, Limacher M, Assaf AR, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy. JAMA. 2004;291(14):1701-1712. DOI: 10.1001/jama.291.14.1701
- Rossouw JE, Prentice RL, Manson JE, et al. Postmenopausal hormone therapy and risk of cardiovascular disease by age and years since menopause. JAMA. 2007;297(13):1465-1477. DOI: 10.1001/jama.297.13.1465
- Manson JE, Chlebowski RT, Stefanick ML, et al. Menopausal hormone therapy and health outcomes during the intervention and extended poststopping phases of the Women's Health Initiative randomized trials. JAMA. 2013;310(13):1353-1368. DOI: 10.1001/jama.2013.278040
- Manson JE, Aragaki AK, Rossouw JE, et al. Menopausal hormone therapy and long-term all-cause and cause-specific mortality: the Women's Health Initiative randomized trials. JAMA. 2017;318(10):927-938. DOI: 10.1001/jama.2017.11217
- Hodis HN, Mack WJ, Henderson VW, et al. Vascular effects of early versus late postmenopausal treatment with estradiol (ELITE). N Engl J Med. 2016;374(13):1221-1231. DOI: 10.1056/NEJMoa1505241
- Harman SM, Black DM, Naftolin F, et al. Arterial imaging outcomes and cardiovascular risk factors in recently menopausal women (KEEPS). Ann Intern Med. 2014;161(4):249-260. DOI: 10.7326/M14-0353
- Shumaker SA, Legault C, Rapp SR, et al. Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in postmenopausal women: WHIMS. JAMA. 2003;289(20):2651-2662. DOI: 10.1001/jama.289.20.2651
- Randolph JF Jr, Zheng H, Sowers MR, et al. Change in follicle-stimulating hormone and estradiol across the menopausal transition. J Clin Endocrinol Metab. 2011;96(3):746-754. DOI: 10.1210/jc.2010-1746
- Greendale GA, Sowers M, Han W, et al. Bone mineral density loss in relation to the final menstrual period in a multi-ethnic cohort: SWAN. J Bone Miner Res. 2012;27(1):111-118. DOI: 10.1002/jbmr.534
- Greendale GA, Sternfeld B, Huang M, et al. Changes in body composition and weight during the menopause transition. JCI Insight. 2019;4(5):e124865. DOI: 10.1172/jci.insight.124865
- Matthews KA, Crawford SL, Chae CU, et al. Are changes in cardiovascular disease risk factors in midlife women due to chronological aging or to the menopausal transition? J Am Coll Cardiol. 2009;54(25):2366-2373. DOI: 10.1016/j.jacc.2009.10.009
- Harlow SD, Gass M, Hall JE, et al. Executive summary of the Stages of Reproductive Aging Workshop + 10. J Clin Endocrinol Metab. 2012;97(4):1159-1168. DOI: 10.1210/jc.2011-3362
- Xiang Y, Meng Q, Huang Z, et al. Menopausal status, transition, and age at menopause with accelerated biological aging across multiple organ systems: findings from two cohort studies. BMC Med. 2025;23(1):461. DOI: 10.1186/s12916-025-04223-7
- Appiah D, Wellons MF, Schreiner PJ, et al. The prospective association of cellular markers of biological aging with menopause in the Coronary Artery Risk Development in Young Adults Study. Menopause. 2025;32(1):91-94. DOI: 10.1097/GME.0000000000002457
- Wang L, Xu S, Chen R, et al. Exploring the causal association between epigenetic clocks and menopause age: insights from a bidirectional Mendelian randomization study. Front Endocrinol (Lausanne). 2024;15:1429514. DOI: 10.3389/fendo.2024.1429514
- Ruth KS, Day FR, Hussain J, et al. Genetic insights into biological mechanisms governing human ovarian ageing. Nature. 2021;596(7872):393-397. DOI: 10.1038/s41586-021-03779-7
- The 2022 hormone therapy position statement of The North American Menopause Society. Menopause. 2022;29(7):767-794. DOI: 10.1097/GME.0000000000002028
- National Institute for Health and Care Excellence. Menopause: identification and management. NICE guideline NG23, 2015, updated 2024.