Blog – We’re Still Not Counting LGBT People in Dementia Research

Pride season is coming to an end. I have wanted to write this blog for a while, but I kept stalling because I wasn’t sure I would find enough evidence in LGBT dementia research to fill it. That turned out to be the point. The question I started with seems simple enough: do LGBT people face a higher risk of dementia?

I work in dementia epidemiology, I’m part of this community, and I assumed a reasonable answer would be sitting in a paper somewhere. It mostly isn’t. Once you understand why, it is hard to look at our data infrastructure in quite the same way again.

Why we might expect a difference

Minority stress theory suggests that living with stigma, discrimination and constant low-grade vigilance takes a cumulative toll. Chronic stress has measurable effects on the ageing brain.

LGBT people also experience higher rates of several risk factors we already know matter, including depression, smoking and cardiovascular disease.

The hypothesis is sound. Testing it is the hard part.

What the LGBT dementia research shows

The largest study so far appeared in Neurology in 2024. Huo and colleagues analysed nearly 400,000 adults enrolled in the US All of Us Research Program, roughly one in ten of whom belonged to a sexual or gender minority group.

They found 15% higher odds of a composite brain health outcome covering stroke, dementia and late-life depression. For dementia alone, the odds ratio was 1.14, with a confidence interval scraping 1.00. It is a possible signal, but a marginal one.

The subgroup findings were more striking. Gender-diverse participants had roughly double the odds of dementia, while transgender women had elevated odds of stroke. Both estimates were based on small numbers and wide confidence intervals, so I would treat them as leads worth following rather than settled findings.

A second study used data from the Nurses’ Health Study II, which included more than 70,000 women. Sexual minority women reported 29% more symptoms of subjective cognitive decline than completely heterosexual women, with the largest disparity found among bisexual women.

Then there is the counterweight. Perales-Puchalt and colleagues compared older adults in same-sex and opposite-sex relationships and found no difference in dementia or mild cognitive impairment. A UK study complicated the picture further, suggesting that any excess risk may be concentrated among people under the age of 55.

The picture is mixed. Mixed evidence is normal, and I do not find that troubling in itself. What concerns me is the reason behind it.

Reading the methods carefully

The All of Us study was cross-sectional. It captured identity and diagnosis at a single point in time, so it could not establish which came first. Causal claims are therefore off the table.

It is also a volunteer cohort. The people who enrol differ systematically from the wider population, which limits how far the estimates can travel.

The Nurses’ Health Study II has the opposite profile: a strong longitudinal design and well-characterised participants, but an outcome based on subjective cognitive decline rather than diagnosed dementia. Subjective decline predicts dementia imperfectly.

There is also a question I have not seen anyone resolve. If you experience greater psychological distress, are you more likely to notice and report memory lapses, independently of any underlying pathology? If so, that could inflate the disparity without there being an equivalent difference in brain health.

The cohort is also made up almost entirely of white female nurses, which tells us little about gay men or transgender people.

The Perales-Puchalt study inferred sexual minority status from participants being in a same-sex relationship. Consider who that misses: single people, widowed people and anyone who is not openly partnered.

The underlying problem

Most health datasets never asked. That single fact shapes almost everything about LGBT dementia research.

The UK Census included a question about sexual orientation for the first time in 2021. Decades of otherwise excellent cohort data are silent on this, and you cannot retrospectively add the question to a cohort recruited 20 years ago.

The consequences run deep. We have no population-level neuropathology data for LGBT people, so we do not know whether the underlying brain changes differ. We have almost nothing on cognitive ageing among transgender people, despite the All of Us signal and genuine unanswered questions about long-term hormone therapy.

Nearly every quantitative study comes from the US, UK or Canada and uses predominantly white samples. Intersections with race, class and disability have barely been examined. LGBT people living with dementia are also largely absent from the literature in their own voices.

I do not think any of this warrants pessimism. The remedy is unglamorous but clear.

Ask about sexual orientation and gender identity in censuses, cohort studies and clinical records. Fund longitudinal research designed around these populations instead of adapting existing studies to include them as an afterthought. Recruit LGBT people inclusively into brain donation programmes.

We did not understand vascular dementia until we built studies capable of seeing it. The same holds here.

As Pride season ends, the most useful thing our field can do is start counting properly.

Dr Connor Richardson

Author

Dr Connor Richardson is a Neuro-epidemiology Research Associate at The University of Edinburgh. His research interests lie in using advanced statistical modelling and machine learning to measure dementia risk. Connor blogs about his research, Equality, Diversity and Inclusion and sometimes his Pomapoo’s.

Follow @connorrichards2