Menopausal Hormone Therapy and Dementia: What Changes Risk

Menopausal hormone therapy (MHT) does not have one dementia-risk profile. The answer changes with when treatment begins, whether it contains a progestogen, and how long it continues. That is why one study can find an apparent benefit while another finds harm without either result settling the question for every woman.

The clearest current pattern is cautious rather than glamorous: starting MHT near menopause for symptom treatment appears neurologically safer than starting it later in life, while longer use of combined estrogen-progestogen therapy carries a small increased association with Alzheimer's disease in registry data. MHT should not be started to prevent dementia. Decisions still belong in the larger calculation of symptoms, uterine status, cardiovascular risk, breast-cancer risk, and personal preferences.

Why the studies have disagreed

A 2024 meta-analysis by Kuck et al. pooled five national registries with 912,157 women, including 278,495 who developed Alzheimer's disease. Its central finding was an interaction: formulation, age at initiation, and duration changed the observed association. Treating any one of those variables as the whole story produces the familiar contradictory headlines.

For combined MHT, estrogen plus a progestogen, Alzheimer's risk was modestly but consistently higher after 5–10 years of use regardless of starting age. With more than 10 years of use, the elevated risk clustered among women who started before age 60. For estrogen-only therapy, women who started before 60 had no association with Alzheimer's disease at any studied duration. But women who started estrogen alone after 60 had higher risk even during the first five years.

The analysis cannot establish that hormones caused those differences. Registry studies compare people who made different treatment choices, and those choices track with health, education, healthcare use, menopausal symptoms, and other factors that prescription data cannot fully capture. Still, its size and subgroup detail explain why a yes-or-no answer to "Does MHT cause dementia?" is the wrong question.

Two 2025 systematic reviews use the same broader frame. Song et al. assessed duration, timing, route, and formulation together, while Melville et al. evaluated mild cognitive impairment and dementia as separate outcomes. The available abstracts do not provide scenario-specific hazard ratios and confidence intervals, so readers should resist precision that the published summaries do not support. They reinforce the interaction model; they do not supply a universal risk number.

Early treatment is safer than late treatment—but not a free pass for decades

The "critical window" idea has a reasonable clinical basis. Estrogen may act differently when started close to the menopausal transition than when it is introduced after years of low estrogen exposure and aging-related vascular change. A mechanistic review by Bortz et al. proposes that estrogen helps maintain choline acetyltransferase activity and basal-forebrain cholinergic neurons. If that system has already declined past a recoverable threshold, later estrogen may not restore it. This is a hypothesis, not proof from a dementia-prevention trial.

The strongest direct evidence on early initiation comes from the 2026 KEEPS Continuation randomized trial. Women ages 42–58 received four years of oral conjugated equine estrogen or transdermal 17β-estradiol with micronized progesterone, compared with placebo. On long-term follow-up, neither treatment adversely affected cognitive function or Alzheimer's biomarkers measured by amyloid PET and MRI.

That result supports the safety of short-term treatment begun early for the population studied. It does not show that MHT prevents Alzheimer's disease: amyloid imaging and MRI are biomarkers, not clinical dementia outcomes, and the trial was underpowered to detect Alzheimer's disease itself.

It also helps define what the Women's Health Initiative Memory Study (WHIMS) did—and did not—test. WHIMS enrolled women age 65 or older using oral conjugated equine estrogen plus medroxyprogesterone acetate (MPA) and found more probable dementia with no cognitive benefit. That result remains highly relevant to late initiation with that regimen. It cannot automatically answer the question for a woman beginning therapy near menopause with transdermal estradiol or micronized progesterone.

Starting MHT near menopause is not evidence that indefinite combined treatment protects the brain.

A 2023 Danish nationwide nested case-control study complicates any simple "early equals safe" message. It found a positive association between long-term MHT and dementia among women who began before 60, including users of contemporary estradiol-based products. The finding argues that duration may matter independently of age at initiation. The Kuck meta-analysis points in the same direction for long-duration combined treatment, but neither study identifies a precise biological cutoff where risk begins.

The progestogen may be part of the difference

Estrogen-only therapy is generally an option only for women without a uterus, because unopposed systemic estrogen can stimulate the endometrium. For women who need endometrial protection, the type of progestogen may matter as much as its presence.

A 2025 mechanistic review by Ylikorkala et al. describes ways synthetic progestogens, especially MPA and norethisterone acetate, could counter estrogen's neural effects. Proposed pathways include glucocorticoid-receptor activity and inflammatory cytokine signaling. Micronized progesterone may be more neutral by comparison, but that remains a mechanistic inference. Large registry datasets often combine several progestogens, so they cannot yet provide trustworthy, agent-by-agent dementia-risk estimates.

Route also remains unresolved. Oral and transdermal products differ in metabolism and in some non-neurological risks, but the available evidence here does not justify calling one route dementia-protective. Song et al. included route in its review; the accessible summary does not provide enough quantitative detail to rank routes for Alzheimer's prevention.

Why observational results still need skepticism

A 2021 UK nested case-control analysis by Vinogradova et al. also found risk signals under particular treatment conditions and documented discordant findings across Finnish registry studies. Differences in cohort age, follow-up, product classification, and outcome definitions can change the result. So can bias: women who use MHT may differ systematically from nonusers, and early cognitive changes may lead some women to stop treatment before a dementia diagnosis. That can make short-duration use look safer than it is or alter apparent risk among long-term users.

Diagnostic coding adds another limitation. Registries often use clinical dementia diagnoses rather than neuropathological confirmation, which can blur Alzheimer's disease and vascular dementia. A South Korean retrospective cohort study extends the evidence beyond Nordic and UK populations, but it shares the basic strengths and limits of administrative-database research: broad coverage, many participants, and imperfect control of confounding.

What this means in practice

For a woman considering MHT for bothersome menopausal symptoms, timing and duration deserve an explicit conversation. Starting near menopause for a limited period has more reassuring cognitive evidence than initiating treatment after 60 or 65. Combined therapy deserves reassessment as treatment extends beyond five years, because the registry signal becomes less favorable with longer exposure. Women without a uterus have a different formulation choice: estrogen alone showed no Alzheimer's association among early initiators in the Kuck analysis, but late initiation still showed higher risk.

Dementia is only one outcome. A 2024 cardiovascular meta-analysis found that MHT's cardiovascular balance depends on timing, duration, type, and baseline cardiovascular health. A 2026 breast-cancer meta-analysis found an overall association with incidence of odds ratio 1.15 (95% confidence interval 1.09–1.22), with substantial heterogeneity and regimen type as an important modifier. The 2025 Cochrane review places cognition alongside mortality, cardiovascular disease, cancer, gallbladder disease, and fractures.

The sensible conclusion is narrower than a longevity slogan: use MHT when its established benefits for menopausal symptoms justify it, begin near menopause when appropriate, use the lowest effective dose for the needed duration, and revisit long-term combined therapy. Do not use late-start MHT as a brain-protection strategy.