A large Danish registry study did find a higher rate of dementia diagnoses among women prescribed combined hormone therapy, including some who started at or before age 55, and the risk rose the longer they used it.

However, the study's own authors say this could reflect an underlying difference in the health of women who need this treatment rather than a direct effect of the hormones themselves. A larger UK study using similar registry methods found no overall increased risk, and even a decreased risk with long-term estrogen-only use. The newest, highest-quality analysis pooling the strongest study designs found no association in either direction. A single registry study is not proof of cause and effect.

What to know about this answer

The real disagreement is not whether the Danish and Finnish numbers are accurate, but whether they reflect a true drug effect or the fact that women prescribed hormone therapy differ in health status from women who are not, in ways that also affect dementia risk.

What the research shows, and which source each finding rests on

A Danish nested case-control study found all-cause dementia hazard ratio 1.24 for combined hormone therapy users, rising to 1.74 for 12 or more years of use, and present even in women who started at or under age 55.

C[1]

A Finnish nationwide case-control study found 9 to 17 percent increased Alzheimer's risk with systemic hormone therapy, with vaginal estradiol showing no association.

C[2]

A UK study of 118,501 dementia cases, larger than the Danish study, found no overall increased dementia risk, a decreased global dementia risk with 10 or more years of estrogen-only use, and only a small Alzheimer's-specific increase with 5 or more years of combined therapy.

D[3]

The December 2025 meta-analysis pooling the highest-quality study designs found no association between hormone therapy and dementia risk in either direction.

A[4]

Where the real disagreement is

Each position is shown with its own sources. A position listed here is not necessarily the accepted view.

there is a real harm signal, even in midlife initiators

the Danish and Finnish registry researchers, and clinicians such as Jen Gunter who treat late-initiation risk as real

These are nationwide, prescription-based studies free of recall bias, with a clear dose-response pattern by duration of use, a pattern that is harder to dismiss as pure chance.

C[1]C[2]

the signal likely reflects confounding by indication, not causation

the Danish study's own authors, and researchers behind the larger UK study

Women with severe menopause symptoms are both more likely to be prescribed hormone therapy and may differ in vascular and sleep profiles relevant to dementia; the larger UK study using similar registry methods found no overall increased risk at all.

D[3]A[4]

What would settle it: Study designs that emulate a randomized trial using an active comparator (hormone therapy versus non-hormonal menopause treatment), plus techniques that can detect and quantify confounding by underlying health status.

What is really going on here

  1. If HT caused dementia in midlife users, why do the biggest UK data and the newest meta-analysis show nothing?
  2. Is this an association, or has causation been shown?

Questions for your doctor

  1. How do you weigh registry studies like this one against randomized trial evidence?
  2. Does my personal health profile put me closer to the population these studies describe?
  3. What symptoms or health factors are driving the recommendation for or against hormone therapy in my case?

Questions with no answer yet

Is the dementia signal in the national prescription registries a real drug effect, or does it reflect that women who need hormone therapy for severe symptoms already differ, in ways relevant to dementia, from women who do not need it?

Partly studied

What is known: The registry authors themselves say further research is needed to determine whether their finding represents an actual effect or an underlying predisposition in women who need treatment. A larger registry study using similar methods found no overall increased risk, and the newest meta-analysis pooling the strongest designs found nothing in either direction.

What would test it: Target-trial-emulation analyses using an active comparator, such as hormone therapy against a non-hormonal menopause treatment, plus sibling designs and negative-control outcomes built specifically to detect and quantify confounding by indication.

Would the size or direction of the registry finding change if a similar analysis were run in a country with different prescribing patterns?

Partly studied

What is known: Two national registries show a modest risk increase; a larger registry in a different country finds no overall risk increase and even a decreased risk with long-term estrogen-only use, so registries relying on similar prescription-based methods have not converged on the same answer.

What would test it: A pooled, harmonized reanalysis across multiple national registries using the same case definitions and confounder adjustment, to see whether the disagreement is a true population difference or a methods difference.

Sources

  1. 1.Menopausal hormone therapy and dementia: nationwide, nested case-control study.Pourhadi N et al., BMJ 2023 · Observational study
  2. 2.Use of postmenopausal hormone therapy and risk of Alzheimer's disease in Finland: nationwide case-control study.Savolainen-Peltonen H et al., BMJ 2019 · Observational study
  3. 3.Use of menopausal hormone therapy and risk of dementia: nested case-control studies using QResearch and CPRD databases.Vinogradova Y et al., BMJ 2021 · Comparative or small study
  4. 4.Menopause hormone therapy and risk of mild cognitive impairment or dementia: a systematic review and meta-analysis.Melville M et al., Lancet Healthy Longev 2025 · Systematic review or guideline

The letter grades the type of study behind a claim, not the answer itself.

This page is general information based on published research and public sources. It is not medical advice and not a personal assessment, and it does not replace a conversation with your doctor. Peritale is a wellness product, not a medical device, and is not intended to diagnose, treat, cure or prevent any disease. For any medical question, speak with a licensed clinician.

You are not imagining it

In about seven minutes on your phone, Peritale reads five signal systems and maps your symptoms, so you can see that what you feel is real, follow your hormonal transition over time, and walk into your doctor ready to ask for the care you deserve. Your first check is free, no credit card.

Start Free Mapping

Related questions