Hormone Therapy: The WHI Story, Corrected

In July 2002, the largest randomized trial of hormone therapy ever run stopped early, a press conference announced the results to the world, and within a few years millions of women had thrown away their prescriptions. Two decades later, the corrected reading of that trial tells a more useful story — one in which formulation, timing, and individual risk matter more than the headlines ever did. This page explains the evidence. It never prescribes.

July 2002: What Actually Happened

The Women's Health Initiative enrolled 16,608 postmenopausal women with an intact uterus and randomized them to a single oral pill — 0.625 mg of conjugated equine estrogens plus 2.5 mg of medroxyprogesterone acetate — or placebo. The plan was 8.5 years; the data and safety board stopped the arm after 5.2 because breast cancer crossed a pre-set boundary. A press conference followed, the phrase "26% increased risk" went around the world, and within three years hormone therapy prescriptions in the US had fallen by roughly two-thirds. What almost nobody in the room that day conveyed was who the trial had actually enrolled: women with a mean age of 63, most more than a decade past their final period, many with cardiovascular risk factors, and — because symptoms would have unblinded the study — virtually none of the severely symptomatic women who actually seek this treatment. The results of a trial of 70-year-olds were applied to healthy 52-year-olds, and a generation of symptom relief and bone protection was lost in the transaction.

Three Things That Went Wrong

• 👵 The wrong population. The trial asked whether hormones help women who started them 13-plus years after menopause. It could not answer the question women were actually asking — whether hormones help now, in their early fifties, when symptoms are worst. The average participant was 63; fewer than 4% were in their early 50s.
• 💊 The wrong regimen, generalized. Oral conjugated equine estrogens pass through the liver and shift clotting and inflammatory proteins; medroxyprogesterone acetate, the progestin used, carries its own signals for breast and vascular tissue. Modern alternatives — transdermal estradiol, micronized progesterone — have more favorable biological profiles, though the direct comparative trial evidence is thinner than anyone would like.
• 🧮 The wrong arithmetic. "26% increase in breast cancer" is a relative risk. The absolute version: about 8 extra breast cancers per 10,000 women per year — 38 versus 30. Both statements describe the same data. Only one of them made headlines, and only one of them tells you what a decision actually costs.

The Absolute Numbers, Laid Out

Read this table slowly, because it is the whole story. The harms are real but small in absolute terms, and they have shape: the blood-clot risk is front-loaded — highest in the first year — while the breast cancer increase accrues with years of use. The benefits are real too: hip fractures fell by a third, all fractures combined by a quarter (152 versus 199 per 10,000 women per year). Whether those columns net out positive or negative for a given woman depends almost entirely on who she is — which is the argument the next two sections make.

The Estrogen-Only Arm: The Twist Nobody Expected

Women who had had a hysterectomy — and therefore needed no progestin to protect the uterine lining — were in a separate arm: 10,739 women on conjugated equine estrogen alone. When it was stopped in 2004, the finding that shocked the field: breast cancer was 23% lower in the estrogen group — 26 versus 38 cases per 10,000 women per year. Stroke and blood clots were still increased, hip fractures still fewer, but the breast cancer signal ran in the opposite direction from the combination arm. The thirteen-year cumulative follow-up published in 2013 sharpened the picture further: women who started in their fifties had lower all-cause mortality than placebo, and a later meta-analysis of all the world's observational data (the 2019 Collaborative Group review of over 100,000 breast cancer cases) showed that the excess risk concentrates in current users of combination therapy, grows with duration beyond about five years, and is small or absent for short-term estrogen use. The lesson that survived: formulation and duration are not details — they are the story.

The Corrected Picture: The Timing Hypothesis

Re-analyzing WHI by age changed the cardiovascular column. Among women who started within ten years of menopause or under age 60, there was no coronary excess — and in some analyses a signal of protection. The ELITE trial tested the idea directly in 643 women: oral estradiol slowed carotid-artery thickening in women within six years of menopause, but had no effect in women ten-plus years out — the arteries, it seems, close the window. An open-label Danish trial in women averaging 50 years old reported fewer deaths, heart attacks, and heart failure admissions on therapy. This is the "timing hypothesis," and it has a proper skeptical summary: the supportive evidence comes from surrogate endpoints like arterial thickness, subgroup reanalyses, and open-label trials — not from a large blinded event trial. The defensible position is the one mainstream societies take: cardiovascular effects in younger initiators are probably neutral to favorable, but hormone therapy is not prescribed to prevent heart disease. For what actually prevents heart disease, the Blood Pressure pillar is the better place to spend your attention.

What Hormone Therapy Is Genuinely For Today

The pattern is clean: symptom relief is the strongest indication, by a wide margin. Prevention of bone loss is a real, proven second job — but for women whose only issue is bone, the risk-benefit arithmetic usually favors the bone-specific medications covered in the Bone Health topic plus the non-hormonal levers in the Strength pillar. What hormone therapy is not: a cardiovascular preventive, a cognitive enhancer, or an anti-aging treatment. Anyone selling it as those is selling something the evidence does not stock.

A Decision Framework, Not a Prescription

The modern guidance from the North American Menopause Society and comparable bodies reads like this: for symptomatic women under 60 or within ten years of their final period, the benefits of hormone therapy usually outweigh the risks; for treatment begun later or continued into the late sixties and beyond, the balance shifts. The inputs to any individual decision are few and knowable: age and years since the final period, symptom severity, personal and family history of breast cancer and blood clots, and the formulation used. The operating principles: start at the lowest effective dose, use the shortest duration that meets the goal, and reassess with a clinician at least yearly — because the balance itself moves as you age. The same habit that the Quarterly Audit builds for numbers applies to this decision: review it on purpose, not by default.

Go Deeper: Subtopics

• 🔎 The 2002 misreporting, forensically — how relative risks became headlines, and the absolute numbers that were lost. Read it →
• 🔎 Estrogen-only vs combined — the two arms' different risk profiles (breast, stroke, heart) explained separately. Read it →
• 🔎 The timing hypothesis, deep — the under-60 window evidence (ELITE, Danish trial) and its skeptical summary. Read it →
• 🔎 Formulations & routes — oral vs transdermal vs vaginal: the risk differences that matter. Read it →
• 🔎 The decision framework — symptoms, age, family history: how an individual actually weighs it (⚠️ clinician conversation). Read it →

Related Topics