HRT is the realistic pharmacotherapy comparator for the menopause stratum — the drug against which the lifestyle levers (-> Menopause and the Shifting Levers) are judged. The wiki appraises what it does; it does not prescribe it. The whole page turns on one distinction the popular is HRT safe? question hides: the answer depends on which outcome, which formulation, and when it is started — and two gold guideline bodies reading a largely shared evidence base disagree on the last.
The reframing that reorganizes everything. The 2002 WHI (the trial that made HRT is risky common knowledge) enrolled women of mean age 63.3, mostly asymptomatic, testing one oral regimen, and its own verdict was scoped to primary prevention of chronic disease:
«this regimen should not be initiated or continued for the primary prevention of CHD.» (for the Women’s Health Initiative Investigators, 2002)
It explicitly did not test the use HRT is now primarily indicated for:
«This trial did not address the short-term risks and benefits of hormones given for the treatment of menopausal symptoms.» (for the Women’s Health Initiative Investigators, 2002)
So WHI showed HRT is harmful is an over-generalization from an older, asymptomatic population and one formulation to a younger, symptomatic one and all formulations. NAMS names exactly this gap: WHI had «limited enrollment of women with bothersome vasomotor symptoms … who were aged younger than 60 years or who were fewer than 10 years from menopause onset—the group of women for whom hormone therapy is currently primarily indicated.» (2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022)
The hard-outcome base — WHI, in absolute terms
(for the Women’s Health Initiative Investigators, 2002) The one large hard-outcome RCT. Continuous oral CEE 0.625 mg + MPA 2.5 mg vs placebo, 16,608 women 50-79, stopped at mean 5.2 y. State it in absolute terms — the decision unit (-> Baseline Risk and the Relative-Absolute Split):
«Absolute excess risks per 10000 person-years attributable to estrogen plus progestin were 7 more CHD events, 8 more strokes, 8 more PEs, and 8 more invasive breast cancers, while absolute risk reductions per 10000 person-years were 6 fewer colorectal cancers and 5 fewer hip fractures. The absolute excess risk of events included in the global index was 19 per 10000 person-years.» (for the Women’s Health Initiative Investigators, 2002)
Relative: CHD HR 1.29 (1.02-1.63), stroke 1.41 (1.07-1.85), PE 2.13 (1.39-3.25), breast cancer 1.26 (1.00-1.59), hip fracture 0.66 (0.45-0.98); total mortality 0.98 (0.82-1.18) — unaffected. (for the Women’s Health Initiative Investigators, 2002) The excesses are rare in absolute terms (all <10/10,000 py individually), and mostly appeared or grew over the trial (CHD/PE early, stroke from year 2, breast cancer after ~4 y).
HRT is two different exposures — E+P vs E-alone (parameter table)
The single most decision-relevant distinction, hidden by the umbrella term. Estrogen-progestin (EPT, for women with a uterus) and estrogen-alone (ET, after hysterectomy) diverge on the outcome that drove the WHI alarm — breast cancer runs in opposite directions.
| Parameter (breast cancer) | Estrogen + progestin (EPT) | Estrogen-alone (ET) | Same quantity? |
|---|---|---|---|
| WHI intervention-phase effect | increased, +9 cases/10,000 py | reduced, 7 fewer/10,000 py; HR 0.79 (0.61-1.02) | No — opposite sign |
| At ~20 y cumulative follow-up | HR 1.28 (1.13-1.45), sustained | HR 0.78 (0.65-0.93), significant reduction | No |
| Breast-cancer mortality (20 y) | HR 1.35 (0.94-1.95), NS | HR 0.60 (0.37-0.97), reduced | No |
| Endometrial protection | the reason progestogen is added | N/A (no uterus) | different indication |
(2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022) Source: «After 20 years of median cumulative follow-up, CEE alone was associated with significantly lower breast cancer incidence (HR, 0.78; 95% CI, 0.65-0.93) and breast cancer mortality (HR, 0.60; 95% CI, 0.37-0.97) … In contrast, CEE plus MPA was associated with significantly higher breast cancer incidence (HR, 1.28 …) but no significant difference in breast cancer mortality» (2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022). So the progestogen, not the estrogen, carries the breast-cancer signal — a claim that reorganizes the risk conversation and that the undifferentiated HRT hides. NICE concurs on direction: breast-cancer incidence «consistently greater with combined HRT than with oestrogen-only HRT» (National Institute for Health and Care Excellence, 2024) (shared evidence -> F, not independent E).
Absolute size, in perspective (EPT, mean age 63): «less than one additional case of breast cancer diagnosed per 1,000 users annually … similar to the risk reported with obesity and low physical activity» (2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022) — and duration-dependent (EPT HR 1.66 at 1-4 y rising to 2.68 at 15+ y). NICE adds it «started increasing in the first year of use» and persists >=10 y after stopping. (National Institute for Health and Care Excellence, 2024)
The four evidence-states, by outcome
(2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022; inferred from for the Women’s Health Initiative Investigators, 2002; National Institute for Health and Care Excellence, 2024) Kept distinct: benefit · harm · no meaningful effect · insufficient/contested. This is the honest menu the safe or not? question collapses.
| Outcome | State | Evidence |
|---|---|---|
| Vasomotor symptoms | BENEFIT (large) | HT «reduce[s] weekly symptom frequency by 75%» (NAMS); NICE «Offer HRT to people with vasomotor symptoms» — the gold-standard indication |
| Bone / fracture | BENEFIT | HT reduces fracture in healthy postmenopausal women (NAMS Level I); WHI 6 fewer hip + 6 fewer vertebral/10,000; abates within 1-2 y of stopping |
| Genitourinary symptoms | BENEFIT (low-dose vaginal) | vaginal ET effective, minimal systemic absorption (NAMS) |
| Breast cancer | HARM (EPT) / benefit-or-null (ET) | see the E+P vs ET table above — opposite signs by formulation |
| Stroke / VTE | HARM, route- and timing-modified | WHI stroke HR 1.41, PE 2.13; rare in <60 initiators; NICE: combined HRT raises stroke with oral but not transdermal oestrogen |
| CHD | CONTESTED (see Does Timing Modify HRT Cardiovascular and Mortality Effect) | WHI +7/10,000 (older, asymptomatic); NAMS: reduced in <60 initiators; NICE: null, no clear timing effect |
| All-cause mortality / life expectancy | NO MEANINGFUL EFFECT overall | WHI total mortality HR 0.98; NICE 1.6.1 «unlikely to affect life expectancy»; a 50-59 subgroup reduction is disputed (Does Timing Modify HRT Cardiovascular and Mortality Effect) |
| Cognition / dementia | NO benefit; HARM if started >65 | neutral early; WHIMS +23 dementia cases/10,000 when started >65 (NAMS); NICE: «might» increase if started later |
| Type 2 diabetes | BENEFIT (not an indication) | EPT 16 fewer T2D cases/10,000 py (NAMS) — noted, not government-approved for it |
The tension — does timing modify the cardiovascular effect? (type-D, two gold bodies)
Filed as a standalone tension -> Does Timing Modify HRT Cardiovascular and Mortality Effect (this section is the nucleus’s summary of it).
(inferred from 2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022; National Institute for Health and Care Excellence, 2024) This is a genuine joined-issue: the same question — does age / time-since-menopause modify HRT’s effect on CHD and mortality? — answered oppositely by two gold guideline families reading a largely shared trial base. It is the guidance-null defeated by showing why they disagree, not by picking a side.
| NAMS 2022 (endorses the window) | NICE NG23 2024 (declines it) | |
|---|---|---|
| CHD, <60 / <10-y initiators | reduced — 2015 Cochrane RR 0.52 (0.29-0.96); 2019 meta-regression OR 0.61 | «did not have a clear tendency to decrease or increase with … the age … or the time between … menopause and when they start HRT» -> research recommendation |
| All-cause mortality, 50-59 | reduced — «significantly reduced in the pooled trials … (HR, 0.69; 95% CI, 0.51-0.94)» (age trend p=.01) | the same subgroup: «In isolation, this was a statistically significant figure. But in a wider context, it cannot be interpreted as good evidence of a different effect» |
| Evidence admitted | observational + reanalyses + meta-regression + WHI subgroups | RCTs only for mortality (confounding); subgroup differences judged non-significant / chance |
| Verdict | benefit-risk favorable if started <60 / <10 y | HRT «unlikely to affect life expectancy» either arm; no timing recommendation |
Counter-passage — is this joined, or are they answering different questions? Checked NAMS’s own text: the window rests substantially on evidence NICE excludes, and NAMS itself records the fragility — «the results for CHD, cardiac mortality, and all-cause mortality were all attenuated after excluding open-label trials» (2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022). So the divergence is partly a different evidence base (guidance-difference reason 2: NICE restricts to RCTs; NAMS admits observational + open-label) and partly a genuine disagreement on the same WHI subgroups (reason 3: significant-in- isolation vs chance). It is not a fake tension of different scopes — both are answering “does starting earlier change the CV/mortality outcome?”
Hidden insight: the critical-window hypothesis is carried by the weaker evidence in the stack (observational, open-label, non-significant RCT subgroup trends). A body that weights RCT evidence hardest will not endorse it; a body that admits the wider evidence will. This is exactly the telos’s case where guidance families disagree because the evidence does not determine the answer — and the U/J-shaped-artifact discipline applies to a subgroup signal too: a 50-59 mortality benefit «significant in isolation» that shows no age pattern is a candidate artifact, not an established effect modifier -> The U-Shaped Association Artifact, Certainty of Evidence vs Strength of Recommendation.
Where they AGREE (shared evidence -> F): HRT does not raise all-cause mortality overall; EPT raises breast cancer and ET does not; oral oestrogen raises stroke and transdermal does not; HRT is not indicated for CVD or dementia prevention. Agreement across two families on shared trials strengthens the null on those points but is not independent corroboration (both rest on WHI/Cochrane).
The surrogate note — estrogen’s BMD->fracture chain HOLDS (contrast with testosterone)
(inferred from 2022 Hormone Therapy Position Statement Advisory Panel of The North American Menopause Society, 2022) A clean contrast with the sibling page. In Testosterone Adiposity and Muscle, the credentialed surrogate inverted: testosterone raised bone density but increased fractures (TRAVERSE HR 1.43). With estrogen, the BMD->fracture chain transmits correctly — HRT raises BMD and reduces fractures in the same trials (WHI: 5 fewer hip fractures/10,000 py; NAMS Level I). So the lesson is not BMD is always a bad surrogate but a surrogate earns a target only if its transmission to the outcome is itself evidenced -> Surrogate Outcomes: estrogen passes that test on fracture, testosterone fails it. The BMD benefit reverses within 1-2 y of stopping HRT — a durable target requires continued exposure.
Guideline vs underlying evidence (the R15 lens)
NAMS and NICE are guidelines resting on trials, not independent evidence. For an effect size, cite the trial (WHI) not the guideline; the guidelines’ value here is (a) the E+P/ET disaggregation, (b) the timing appraisal, and (c) the disagreement itself. NICE makes its own appraisal method explicit (RCT-restriction for mortality, route-stratified stroke, research recommendations where power is lacking) — a worked Certainty of Evidence vs Strength of Recommendation instance. Which non-health objective moved either body’s framing is not visible here -> Which Objective Moved This Recommendation (not charged, only noted).
Decision relevance
- The question is never is HRT safe? but for which outcome, which formulation, started when? For a symptomatic woman <60 or <10 y from menopause with no contraindication, both bodies support HRT for VMS and bone; the disputed part is a bonus CV/mortality benefit (NAMS) vs no effect (NICE) — and the no-mortality-effect is agreed, so the decision rests on symptom relief and bone against a rare breast- cancer/stroke/VTE harm, not on a life-expectancy gain.
- Formulation is a real lever, not a detail. ET (no uterus) avoids the breast-cancer signal EPT carries; transdermal oestrogen avoids the oral-oestrogen stroke signal. These change the harm side materially.
- Age/timing shifts the harm side regardless of the CV dispute: starting >60 / >10 y raises absolute CHD/stroke/VTE/dementia risk (agreed) — so late initiation is worse whether or not early initiation helps.
- HRT is a symptom + bone treatment, not a preventive or anti-ageing drug — not indicated for CVD, dementia, or muscle -> Menopause and the Shifting Levers (the lifestyle levers).
- Out of scope (appraise, do not prescribe): individual candidacy, contraindication screening (prior breast/estrogen-sensitive cancer, VTE, liver disease), dose, route selection, monitoring — prescriber acts needing this person’s history and labs.
Limits and gaps
- One hard-outcome RCT, one regimen. WHI tested only oral CEE+MPA; transdermal/estradiol/progesterone
hard-outcome RCT data are largely absent — most route/formulation claims are observational (
type-G; NICE itself flags research needs on progestogen type, mode of administration, and timing). - The timing hypothesis is unresolved, not settled either way — held as a live tension, not a finding.
A dedicated RCT of early-initiation hard outcomes would resolve it;
AWAITSsuch a source. - Trajectory/quality-of-life under-measured relative to event counts (-> Surrogate Outcomes).
- POI / premature menopause is a distinct stratum (NAMS: WHI does not apply; HT recommended at least to ~age 52) — noted, not developed here.
- The loop is open. This grades coherence and source-fidelity, never validity; no operation here checks a recommendation against a realized outcome.