A generation was told to stop. That is not what the study said.

In July 2002 a single trial ended hormone therapy for millions of American women almost overnight. In November 2025 the FDA removed the warnings that trial produced. The twenty-three years in between are one of the most instructive episodes in modern medicine — not because the science was wrong, but because of how completely it was compressed.

Fifteen million women, and a consensus

By 2001, roughly fifteen million American women were taking prescription estrogen, with or without a progestin. Something close to 42 percent of women aged 50 to 74 had hormone exposure. It was not a fringe therapy. It was close to standard care for the menopausal transition, and it was supported by a large body of observational research — most famously the Nurses’ Health Study — suggesting that women who took hormones had substantially less heart disease than women who did not.

That observational signal turned out to be the problem. Women who chose hormone therapy in the 1980s and 1990s also tended to be leaner, better educated, more physically active and more likely to see a doctor at all. Epidemiologists call this the healthy-user effect, and it is very difficult to adjust away. The only way to settle the question was a randomized trial. So the National Institutes of Health ran one.

July 2002

The Women’s Health Initiative randomized 16,608 postmenopausal women to conjugated equine estrogens plus medroxyprogesterone acetate, or to placebo. In May 2002 the safety monitoring board recommended stopping it. On July 9 the participants were told; on July 17 the results were published.

The number that travelled was a 26 percent increase in invasive breast cancer. Alongside it: a 29 percent increase in coronary heart disease, a 41 percent increase in stroke, and a doubling of pulmonary embolism. The coverage was immediate and it was total. Within a year, hormone therapy prescriptions in the United States had fallen by about 38 percent. Prescriptions for the specific drug used in the trial fell by two thirds.

Women stopped mid-course, often without a conversation. And an entire cohort of physicians trained after 2002 learned menopause management as something you talk a patient out of.

What the headline could not carry

None of the findings above were fabricated, and it is worth saying clearly that they have never been retracted. What the headline could not carry was the shape of them.

The absolute numbers were small. A 26 percent relative increase in breast cancer worked out to eight additional cases per 10,000 women per year — a figure the trial’s own authors reported, and one that falls into the category epidemiologists classify as rare. Relative risk makes a headline. Absolute risk is what a woman actually faces.

The women studied were not the women who stopped. The average participant was 63 years old, and two thirds were 60 or older. Very few were in the menopausal transition itself. The trial was designed to ask whether hormones prevent chronic disease in older women — not whether they safely relieve symptoms in a 51-year-old who cannot sleep. It answered the first question. It was read as though it had answered the second.

The drugs were of their era. The trial used oral conjugated equine estrogens with a synthetic progestin. That is not estradiol, it is not micronized progesterone, and it does not describe how most of this is prescribed now. Route matters especially: estrogen swallowed in a tablet passes through the liver first, and that first pass is where a substantial part of the clotting signal originates.

One more point of precision, because it matters and it is frequently overstated in the other direction. When the trial’s findings were adjusted for the number of outcomes being tested at once, no individual result — breast cancer included — remained statistically significant. That is a legitimate observation, and the trial’s own paper reports it. It is not a license to say the trial was debunked. The overall index of harm remained significant, and the WHI investigators continue to defend the breast cancer finding to this day. The honest summary is narrower and more useful: a real result, in an older population, on one specific formulation, generalized far beyond what it could support.

The correction took twenty years, and it was quiet

The reversal did not arrive as a single headline. It accumulated.

2004 — the estrogen-only arm. The WHI also ran a second trial: women who had undergone hysterectomy took estrogen alone, without a progestin. It was stopped in 2004 over a stroke signal. But followed out past twenty years, that group showed something nobody expected in 2002 — a 22 percent lower incidence of breast cancer, and a 40 percent lower rate of death from it. The progestin, not the estrogen, appears to have carried the breast signal. This finding applies specifically to women without a uterus taking estrogen alone; it does not transfer to combined therapy, and anyone who quotes it without that sentence attached is selling something.

2007 — the timing question. Reanalysis by years since menopause found that coronary risk tracked with how long a woman had been postmenopausal when she started. Women within ten years of menopause trended toward benefit; women twenty years out carried clear excess risk. The trend across those groups was statistically significant.

2017 — the mortality question, answered. Eighteen years of cumulative follow-up across both arms, 27,347 women. All-cause mortality: no difference. Cardiovascular mortality: no difference. Cancer mortality: no difference. Whatever the trial had found in 1998, women who had taken hormone therapy were not dying at higher rates two decades later. For a therapy the public had come to regard as dangerous, neutrality was the most powerful result available.

Where it is genuinely unsettled

We would rather tell you this part than have you find it elsewhere.

The timing hypothesis — start early and the profile is favorable — has good support and real gaps. The ELITE trial found that estradiol slowed the thickening of carotid artery walls in women within six years of menopause, and did nothing for women ten years out. But that was a surrogate measurement of arterial anatomy, and the same trial found no difference in coronary calcium or plaque. KEEPS, which tested oral and transdermal estrogen against placebo in recently menopausal women, did not meet its primary endpoint at all. And in the 2007 reanalysis, stroke risk did not fall away in younger women the way coronary risk did.

The WHI investigators themselves, writing to the FDA in September 2025, put it bluntly: the claim that starting young preserves a lower risk profile into later decades “remains untested and cannot be proven… it reflects hope rather than solid scientific evidence.” They also state, and we agree, that hormone therapy should not be prescribed to prevent heart disease or to extend life.

So: hormone therapy is the most effective treatment available for hot flashes, night sweats and the genitourinary symptoms of menopause, and it is approved for the prevention of postmenopausal bone loss. It is not a cardiac drug, it is not a dementia drug, and it is not a longevity drug. Any clinic telling you otherwise is ahead of its evidence.

November 10, 2025

On that date the FDA asked manufacturers to remove the boxed warning — the black box, the strongest warning the agency issues — from menopausal hormone therapy products. Out went the boxed language on cardiovascular disease, breast cancer and probable dementia, along with the instruction to use the lowest dose for the shortest possible time. In came a line telling prescribers to consider hormone therapy for moderate to severe vasomotor symptoms in women under 60 or within ten years of menopause.

FDA Commissioner Marty Makary described the previous two decades as a case of “a medical dogma rooted in a distortion of risk.” On February 12, 2026 the agency approved the first batch of revised labels; twenty-nine manufacturers had filed proposed changes.

What did not change

This is the part we think is missing from most of the coverage, and it is the part a careful reader should want.

• The endometrial cancer boxed warning remains on systemic estrogen-alone products. If you have a uterus, unopposed estrogen is still unsafe, and progesterone alongside it is not optional.
• The cardiovascular and breast cancer information was moved, not deleted. It now sits under Warnings and Precautions rather than in the box. It is still on the label and it still informs the decision.
• No new indication was approved. The approved uses are what they were: vasomotor symptoms, genitourinary syndrome of menopause, and prevention of postmenopausal osteoporosis. The FDA did not approve hormone therapy for prevention of anything else.
• None of it applies to compounded products, which sit outside the approved-labeling system entirely. Compounded preparations — including estradiol pellets, of which there is no FDA-approved version in the United States — are a legitimate clinical tool, and we use them. They are simply not what the FDA reviewed.
• The consensus was already there. The Menopause Society concluded in 2022 that benefits outweigh risks for most healthy symptomatic women under 60 or within ten years of menopause. The label finally caught up with the guideline; the guideline did not change.

Where that leaves you

If you are in your late forties or fifties, sleeping badly, thinking less clearly than you used to and carrying weight that will not respond to what always used to work, the twenty-year detour is not an abstraction. It is the reason nobody offered you anything.

The current position is not that hormone therapy is safe for everyone. It is that the decision is individual, it turns on who you are and when you start, and it deserves an hour and a full panel rather than a shrug. A history of breast cancer, unexplained bleeding, prior clot or stroke, or a start many years past menopause all genuinely change the calculation — and establishing that properly is exactly what a baseline evaluation is for. Where systemic therapy is not right, there is nearly always something worth doing instead.

That is the whole argument for doing this with a physician who measures rather than assumes. You can read how we approach hormone therapy for women, or see what a baseline panel actually includes.

Written by Athlos Health, Torrance, California. This article is general education and is not medical advice, a diagnosis, or a treatment recommendation. Hormone therapy is prescribed only after individual evaluation by a physician, and the full risk discussion happens in consultation and in writing before anything is prescribed. Every figure above is linked to its source below.