Is HRT safe? The honest answer has three variables in it

What the position statement actually says

Quoted rather than summarised, because the two halves are usually reported apart and each one is misleading on its own.

The 2022 hormone therapy position statement of The North American Menopause Society. VMS is vasomotor symptoms, meaning hot flashes and night sweats.

The absolute numbers behind the 2002 headline

The Women's Health Initiative stopped its conjugated equine estrogens 0.625 mg/d plus medroxyprogesterone acetate 2.5 mg/d arm after a mean 5.2 years because invasive breast cancer crossed the stopping boundary. Both columns below come from that paper. The left is the number that travelled. The right is the size of it.

Excess events on the trial's combined global index came to 19 per 10,000 person-years. Total mortality was unaffected during the trial, hazard ratio 0.98. Read the excess figures as what they are: 7 to 8 additional events per 10,000 women per year for each of the four harms, in a cohort averaging 63.4 years old.

Estrogen alone and estrogen plus a progestogen reached opposite conclusions on breast cancer

This is the split that the phrase "HRT causes breast cancer" erases. The WHI ran two trials. Women with a uterus received conjugated equine estrogens plus medroxyprogesterone acetate. Women who had had a hysterectomy received estrogen alone. After more than 20 years of follow-up:

Two consequences follow, and both cut against the shorthand. A woman who has had a hysterectomy and takes estrogen alone is being quoted a risk from a regimen she is not on. And the progestogen in the trial was medroxyprogesterone acetate, a synthetic, rather than the micronized progesterone most US prescribers now use, so the combined figure is from a regimen that has itself moved on. What the micronized progesterone label does and does not show.

How much risk, at your age

The integrated 2013 overview reports the combined global index in the units a person can actually use: excess events per 10,000 women annually, by age band. The regimen changes the sign of the answer at the younger end.

excess global-index events per 10,000 women annually, from Manson JE et al, JAMA 2013. A woman of 55 on estrogen alone sat at 19 FEWER global-index events per 10,000 annually than placebo. A woman of 75 on the same drug sat at 51 more. Same molecule, opposite sign, twenty years apart. Neither regimen affected all-cause mortality.

The timing hypothesis, with its own caveat attached

The 2007 re-analysis is where the "window of opportunity" idea comes from. It holds for coronary heart disease by years since menopause, and the paper is careful about how far it holds.

Two things the enthusiastic version leaves out. The age-band trend did not meet the authors' significance criterion, at P .16, and they describe the coronary finding as a tendency rather than a proof. And stroke rose regardless: hazard ratio 1.32 (1.12 to 1.56), and risk did not vary significantly by age or by time since menopause. Starting early moves the coronary arithmetic. It does not move the stroke arithmetic, and any page telling you the window makes HRT risk-free is selling you the half of the paper it liked.

Route is the third variable, and it is the one you can change today

Age and timing are fixed by the time you are reading this. Route is a prescription decision. Estradiol swallowed as a pill passes through the liver before reaching general circulation and estradiol absorbed through the skin does not, which is the mechanism behind the difference below.

Venous thromboembolism against no exposure, from Vinogradova Y et al, BMJ 2019, 5,795 cases and 21,670controls. The authors' own conclusion: "Transdermal treatment was the safest type of hormone replacement therapy when risk of venous thromboembolism was assessed." This is an observational study, so the odds ratios are associations rather than randomised effects, and it is the largest direct comparison of the two routes that exists.

Practically: the two lowest rows in that table are a transdermal product and an oral estradiol paired with dydrogesterone, and the top row is the exact regimen the WHI tested. Of the 13 women's-wing platforms whose prices we verify, the cheapest published route and the lowest-risk route are different routes, which is why the price-per-form table carries the same warning.

Eighteen years later, mortality was flat

The outcome most people assume 2002 settled was never what 2002 measured. Across 18 years of cumulative follow-up and 7,489 deaths, all-cause mortality was 27.1 percent on hormone therapy against 27.6 percent on placebo, hazard ratio 0.99 (0.94 to 1.03). By trial: 1.02 for estrogen plus progestogen and 0.94 for estrogen alone. Manson JE et al, JAMA 2017.

Who should not take systemic estrogen

Quoted from The Menopause Society's patient guidance, read 2026-08-15, because paraphrasing a contraindication list is rewriting a clinical boundary:

"In general, women who have breast cancer, uterine cancer, unexplained uterine bleeding, liver disease, a history of blood clots, and cardiovascular disease should not use hormone therapy."

Several of those entries turn on detail only a prescriber holding your history can weigh. A clot ten years ago on a plane and a clot last month on an oral contraceptive are the same three words on a list and a different conversation. Treat it as the reason to bring your history to an intake rather than as a self-assessment you can complete alone.

For women in that group with genuinely bothersome symptoms there is a route that does not involve systemic estrogen. ACOG names antidepressants, gabapentin and clonidine as prescription options for hot flashes, and a non-hormonal neurokinin antagonist is approved for the same indication. What the non-hormonal route costs and who prescribes it. For genitourinary symptoms alone, the 2022 position statement treats low-dose vaginal estrogen separately from systemic therapy, and its label rates sit on our route-by-route side effects page.

Where to go from here

The regulatory half of this question, what the FDA announced in November 2025 and approved in February 2026, is covered in the black box explainer, and our count of which posted labels have actually dropped the warning is on the Safety Center. The narrative version of who the treatment suits is in Is HRT safe in 2026. Dementia gets its own page because the evidence there is weaker than either camp claims: HRT and Alzheimer's risk.

Sources

• The 2022 hormone therapy position statement of The North American Menopause Society (PMID 35797481)
• Writing Group for the Women's Health Initiative Investigators, JAMA 2002, risks and benefits of estrogen plus progestin (PMID 12117397)
• Anderson GL et al, JAMA 2004, effects of conjugated equine estrogen in women with hysterectomy (PMID 15082697)
• Rossouw JE et al, JAMA 2007, cardiovascular disease by age and years since menopause (PMID 17405972)
• Manson JE et al, JAMA 2013, intervention and extended poststopping phases (PMID 24084921)
• Manson JE et al, JAMA 2017, long-term all-cause and cause-specific mortality (PMID 28898378)
• Chlebowski RT et al, JAMA 2020, breast cancer incidence and mortality (PMID 32721007)
• Vinogradova Y et al, BMJ 2019, HRT and risk of venous thromboembolism (PMID 30626577)
• The Menopause Society, patient education on hormone therapy (read 2026-08-15)
• ACOG, The Menopause Years

This is general information, not medical advice. Every figure above is a population average and none of them is a prediction about you. Whether hormone therapy is right for you, at which dose and by which route, is a decision for you and a licensed clinician. Written and reviewed by Iacob Pastina.