The Trial That Stopped Hormone Therapy — and the Eighteen-Year Follow-Up Nobody Quotes

Two later analyses of the Women's Health Initiative are quoted constantly and read carefully far less often. One found that cardiovascular risk varied by how many years had passed since menopause.[1] The other followed 27,347 women for eighteen years and found no difference in all-cause mortality between hormone therapy and placebo.[2] Both are real. Neither says hormone therapy is safe, and neither says it is dangerous.

What the timing analysis found

The WHI randomised two populations: 10,739 women who had had a hysterectomy, to conjugated equine estrogens or placebo, and 16,608 who had not, to those estrogens plus medroxyprogesterone acetate or placebo. A secondary analysis asked whether the effect on coronary heart disease and stroke depended on age, or on years since menopause began.[1]

For coronary heart disease, by years since menopause:

Years since menopause	Hazard ratio for CHD (95% CI)	Absolute excess risk
Fewer than 10	0.76 (0.50–1.16)	−6 per 10,000 person-years
10 to 19	1.10 (0.84–1.45)	+4 per 10,000 person-years
20 or more	1.28 (1.03–1.58)	+17 per 10,000 person-years

The trend across those categories was statistically significant (P for trend = .02).[1] That is the finding the "timing hypothesis" rests on, and on its own terms it holds up.

Three things usually get left out.

First, look at the confidence intervals. Only the 20-or-more row excludes 1. The under-10 group's interval runs from 0.50 to 1.16 — consistent with meaningful benefit, and equally consistent with mild harm.

Second, the equivalent trend across age groups was not significant (P for trend = .16), and the authors' own conclusion is notably cautious: women starting closer to menopause "tended to have reduced CHD risk," but "this trend test did not meet our criterion for statistical significance."[1] The paper is more hedged than its reputation.

Third, and most often dropped: stroke risk was elevated regardless of timing. Hazard ratio 1.32 (95% CI 1.12–1.56), and it "did not vary significantly by age or time since menopause."[1] Starting early does not buy you out of that one.

What the eighteen-year follow-up found

The mortality analysis pooled both WHI trials — 27,347 women, mortality follow-up available for more than 98%, 7,489 deaths over 18 years.[2]

Outcome	Hormone therapy vs placebo	Hazard ratio (95% CI)
All-cause mortality	27.1% vs 27.6%	0.99 (0.94–1.03)
Cardiovascular mortality	8.9% vs 9.0%	1.00 (0.92–1.08)
Total cancer mortality	8.2% vs 8.0%	1.03 (0.95–1.12)

Estrogen plus progestin gave a hazard ratio of 1.02 (0.96–1.08); estrogen alone, 0.94 (0.88–1.01).[2]

This is a genuinely important result, and it is important for a reason that is easy to misstate. It does not show hormone therapy is beneficial. It shows that across nearly three decades of accumulated follow-up, women randomised to hormone therapy died at the same rate as women randomised to placebo. The strong claim it rules out is the one in the other direction — that the WHI regimens carried a mortality cost that would show up given enough time. It did not appear.

There was a signal by age: comparing women aged 50–59 with those aged 70–79, the ratio of nominal hazard ratios for all-cause mortality was 0.61 (0.43–0.87) during the intervention phase, attenuating to 0.87 (0.76–1.00) over the full 18 years — without significant heterogeneity between the trials.[2] Suggestive, consistent with the timing story, and not the same as proof of it.

Why both camps misquote this

The two analyses get deployed as though they settle opposite arguments, and neither does.

"The WHI was wrong and hormone therapy is safe." The mortality data do not say that. Stroke risk was elevated regardless of when treatment started,[1] and equal mortality is not the same as equal harm — a non-fatal stroke does not appear in a death rate.

"The WHI proved hormone therapy is dangerous." Eighteen years of follow-up found no mortality difference, in either trial, on any of the major cause-specific measures.[2] A treatment that carried a large hidden mortality cost would have been very likely to reveal it in that dataset.

The reasonable reading is the one the current guidance takes: the balance is favorable for women under 60 or within 10 years of menopause who have bothersome symptoms, and less favorable further out, because the absolute risks of coronary heart disease, stroke, venous thromboembolism and dementia are larger then.[3] That is a graded, conditional statement, which is what the underlying evidence supports.

What the WHI did not test

This matters for reading anything across from it.

The trials used conjugated equine estrogens, orally, with medroxyprogesterone acetate as the progestin. They did not test transdermal estradiol, and they did not test micronized progesterone. Route changes the clot picture measurably — see oral versus transdermal estrogen — and progestogen type is a separate variable the position statement lists explicitly among the things that change risk.[3] How much it changes risk is itself measurable — see micronized progesterone versus synthetic progestins.

None of that makes the WHI irrelevant. It means the trial answers a question about a specific regimen, and generalising its numbers to a different hormone by a different route is an inference, not a finding.

Frequently asked questions

Did the WHI actually get retracted or overturned? No. The trials stand and their results stand. What changed is the interpretation: later analyses of the same data separated the population by age and time since menopause, and the guidance now reflects that.

Does starting hormone therapy early protect my heart? That is not what these data establish. The CHD hazard ratio for women under 10 years post-menopause was 0.76, with a confidence interval spanning 1 — and hormone therapy is not recommended for cardiovascular prevention.[3]

If mortality was identical, why not just take it? Because mortality is one outcome. Stroke, venous thromboembolism and breast cancer are measured separately and do not disappear because death rates matched. Equal mortality is a reassurance about magnitude, not a clean bill of health.

Do these numbers apply to a transdermal estradiol spray? Not directly. The WHI tested oral conjugated equine estrogens with medroxyprogesterone acetate. Applying its risk estimates to a different molecule by a different route goes beyond what was measured.

The honest summary

The WHI showed real harms in the population it studied. Later analysis of the same trial found those harms concentrated in women further from menopause, with the notable exception of stroke. Eighteen years on, there was no mortality difference at all. All three statements are true simultaneously, and any account that needs one of them to be false is selling something.

Keen connects you with a licensed clinical team for a telehealth consultation and lab review.

References

3 sources

1. 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 · PMID 17405972
2. 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 · PMID 28898378
3. The North American Menopause Society 2022 Hormone Therapy Position Statement Advisory Panel. The 2022 hormone therapy position statement of The North American Menopause Society. Menopause. 2022;29(7):767–794. doi:10.1097/GME.0000000000002028 · PMID 35797481