The Timing Hypothesis, Deep

The timing hypothesis is the most important idea to come out of the WHI correction, and the easiest to overstate. It says the cardiovascular effects of estrogen depend on when a woman starts it: healthy arteries respond well early, diseased arteries do not — the window, if it exists, sits somewhere under 60 or within ten years of the final period. This page walks the evidence for that window piece by piece, including the trials that failed to find it, and ends with the skeptical summary the data actually support.

The Hypothesis in One Paragraph

Estrogen acts on blood vessels through receptors that exist only while the vessels are healthy enough to respond. Start therapy while arteries are still supple — early in menopause, before plaques are established — and estrogen appears to slow the disease process. Start it a decade later, when plaques already exist, and the same hormone can destabilize what is already there, while its clotting effects push in the wrong direction. The WHI's average participant was 63, most more than a decade past menopause, many with cardiovascular risk factors — which is why, under the timing hypothesis, the trial's cardiovascular harm may describe late starts only. The implication that matters: the 2002 cardiovascular scare, like the breast cancer scare, may not transfer to symptomatic women in their early fifties.

Where It Came From: Monkeys and Observation

The hypothesis has two roots, one experimental and one observational. The experimental root is primate research led by Thomas Clarkson: in cynomolgus monkeys, estrogen begun immediately after surgical menopause suppressed coronary atherosclerosis dramatically, while estrogen begun after plaques had formed had little or no effect — the same hormone, opposite results, separated only by timing (Clarkson, Menopause, 2007). The observational root is the Nurses' Health Study, which reported substantially fewer coronary events in women who started hormone therapy near menopause (Grodstein et al., Annals of Internal Medicine, 2000). That observational result carries a famous caveat: women who take hormones early are systematically healthier, wealthier, and better cared for — the "healthy user" bias — so observational support can overstate the effect. The primate data, which cannot be healthy-user biased, is what gave the hypothesis its credibility.

The WHI Reanalysis by Age

The first human test was internal: re-slicing the WHI itself by age and by years since menopause. In the pooled analysis of both arms, coronary risk varied systematically: for women within ten years of their final period, the hazard ratio was 0.76 — below 1, meaning fewer events than placebo, though the confidence interval crossed 1.0 and the finding was not statistically significant. For women ten to nineteen years out, the ratio was 1.10; for those twenty-plus years out, 1.28 (Rossouw et al., JAMA, 2007). The gradient ran the right direction, but it was a subgroup pattern in a trial not designed for it — suggestive, not decisive.

The Calcium Score Insight

A WHI substudy added an imaging layer. Among roughly 1,000 women who had cardiac CT scans years after the trial ended, those assigned to estrogen alone in their fifties had meaningfully lower coronary artery calcium scores than placebo — the artery walls themselves carried less accumulated plaque (Manson et al., NEJM, 2007). Coronary calcium is a strong predictor of future events, so this is not a trivial surrogate; but it is still a surrogate, and surrogates have misled this field before. The finding matched the age-gradient pattern: the benefit appeared in the youngest group, consistent with the window — and nowhere near the women who started late.

The Direct Tests: KEEPS and ELITE

Two trials then tested the hypothesis head-on, and they disagreed. KEEPS randomized 727 women who were within three years of their final period to oral conjugated estrogens, transdermal estradiol, or placebo, and followed carotid thickness and coronary calcium for four years. Result: no significant difference on either measure (Harman et al., Annals of Internal Medicine, 2014). ELITE randomized 643 women to oral estradiol or placebo, split into two groups by time since menopause — under six years and ten or more. In the early group, carotid wall thickening progressed measurably slower on estradiol than placebo; in the late group, there was no difference at all (Hodis et al., NEJM, 2016). The window hypothesis survived its cleanest test — barely. Four years of KEEPS may simply have been too short and too early to see a difference, but the honest reading is that the direct trial evidence is thinner and more mixed than the hypothesis's popularity suggests.

The Danish Trial: Events, Not Surrogates

The strongest — and most contested — human evidence comes from Denmark. The Danish Osteoporosis Prevention Study randomized about 1,000 recently postmenopausal women, average age around 50, to hormone therapy or no treatment, and followed them for ten years. The primary composite of death, heart attack, or heart failure hospitalization occurred in 16 treated women versus 33 untreated — a hazard ratio near 0.48 (Schierbeck et al., BMJ, 2012). Extended follow-up published in 2016 found the mortality difference persisted. The contest comes from the design: the trial was open-label, so both women and clinicians knew the treatment, and the numbers, while striking, are small — 16 versus 33 events is a difference of 17 women. Open-label trials are vulnerable to subtle selection in who stays, who leaves, and who gets other care. The Danish trial is the reason the timing hypothesis is taken seriously; it is not the reason it is settled.

What the Window Does Not Cover

The timing story applies to cardiovascular outcomes, and even there it is incomplete. It does not apply to stroke, which rose in both WHI arms across age groups — starting early did not erase the stroke signal, and stroke risk rises steeply with age and with hypertension, which is why the cardiovascular risk topic treats blood pressure as the real lever. And the window has a dark side: cognition. The WHIMS substudy found that women who started combined therapy after 65 had roughly double the rate of probable dementia (hazard ratio 2.05), and estrogen alone showed a smaller but same-direction signal (Shumaker et al., JAMA, 2003). Whatever the early window is, it does not extend into the late sixties — late starts look harmful on cognition, not merely neutral.

The Skeptical Summary

• 🔬 The strongest support is not human. The primate evidence is clean and cannot be healthy-user biased; the human evidence is a mosaic of subgroups, surrogates, and open-label trials.
• 📉 The gradient is real but non-significant. The WHI age reanalysis shows the right direction with wide confidence intervals — a pattern, not a proof.
• 🧪 The direct trials split. KEEPS found nothing; ELITE found the window on one surrogate measure. That is an honest mixed result.
• 🚧 The decisive trial does not exist. A large, blinded, event-driven trial in symptomatic women starting under 60 would settle the question; it has never been run, and the societies' guidance — neutral to favorable early, not a prevention drug — is the honest codification of that absence.

Related Topics