Researchers following adults aged 90 and older have found that the sex, racial, and ethnic differences in dementia risk documented in younger age groups do not disappear in the tenth decade of life.
The finding comes from LifeAfter90, a study run by UC Davis Health and Kaiser Permanente that has tracked participants 90 and older since 2018. The analysis of more than 800 participants, median age 92, was published in The Lancet Healthy Longevity. Women 90 and older had roughly twice the dementia risk of men. Black participants had a 75 percent greater risk than Asian participants, and Black and Hispanic participants had significantly higher dementia incidence rates than white and Asian participants.
The result is worth reading twice, because the more consequential fact is structural. This was an open question until now, not because it was scientifically difficult, but because the studies capable of answering it had rarely been built. Disparities can only be measured when enough participants from each group are enrolled to measure them.
The Oldest Old Are the Fastest Growing and Least Studied Group
Two facts sit awkwardly together.
People aged 90 and older are among the fastest growing segments of the population, with projections pointing toward roughly 230 million people worldwide in that age band by 2100. And until recently, researchers had limited evidence about how advanced age affects the brain, because most aging cohorts were assembled around people 65 and older and thinned out sharply above 85.
Rachel Whitmer, a UC Davis Health professor of public health sciences and neurology, chief of epidemiology and senior author on the study, framed the gap plainly: work in people 65 and older showed differences in dementia rates and higher risk in women, but nobody knew whether that held after 90.
Studying this group is genuinely harder. Participants must be recruited before dementia onset, retained through frequent assessment, and followed while competing mortality removes them from the cohort. LifeAfter90 enrolled Kaiser Permanente members 90 or older with no signs of dementia and examines them roughly every six months. Because participants are long-term members, the team has access to decades of health records, in some cases reaching back to the 1960s.
Diversity in a Cohort Is a Design Decision, Not an Accident
The second structural point is about who gets recruited.
UC Davis describes this as the first effort to investigate dementia after 90 in a highly diverse cohort. That is why it can report differences at all. A study that enrolls one group cannot detect disparities between groups, and its findings cannot be assumed to generalize.
That matters for how families should read aging research generally. When a study of dementia risk, or of a drug, or of a screening test, is conducted in a population that does not include people like your parent, the finding may still be true for them, but nobody has checked.
Hilary Colbeth, a UC Davis postdoctoral scholar in public health sciences and first author, described it as striking that disparities observed in younger adults continue into the tenth decade of life.
The study does not explain the disparities themselves. Documented contributors in dementia research include differences in education access, cumulative cardiovascular risk, quality and timing of medical care, neighborhood conditions, and lifetime stress exposure. Reporting a gap is not the same as explaining it, and this study did not attempt to do so.
The Genetics Behaved Differently at This Age
The study also examined APOE, the main genetic risk gene for Alzheimer's disease, and found a pattern that does not simply extend what is known from younger groups.
APOE2, the protective variant, continued to lower risk past age 90, reducing it by 60 percent. APOE4, the variant that raises Alzheimer's risk substantially in younger populations, did not appreciably increase dementia incidence in the group as a whole. But breakdowns told a different story: it raised risk among men, and it essentially doubled risk among Black participants.
That last detail is the one most easily lost, and it cuts against a reassuring reading. An average that shows no effect can conceal substantial effects in subgroups moving in different directions. Colbeth said the sex difference has prompted the team to look more closely at how APOE genotypes affect mortality among people with and without dementia by age 90.
Nothing here means APOE4 is harmless, and nothing here changes whether genetic testing is worthwhile. Genetic testing for APOE remains a decision to make with a clinician or genetic counselor, since the result changes risk estimates without predicting whether an individual will develop dementia and can carry insurance and psychological consequences.
The researchers also flagged an unresolved puzzle. Some participants had hypertension, high cholesterol and other risk factors in their sixties and seventies yet stayed cognitively sound, as did some carrying APOE4. Understanding what protected them is the next question.
Why Representation Is a Funding and Recruitment Question
The practical fix for research gaps is not mysterious, and it is expensive.
Recruiting and retaining underrepresented participants requires community partnership, multilingual staff, transportation support, flexible scheduling and long-term relationships built before enrollment opens. Those costs sit inside grant budgets that are competitive and, in the current environment, uncertain. Cohorts that skip that investment produce cheaper studies with narrower conclusions. This work was supported by the National Institute on Aging.
Whitmer drew the clinical implication directly, saying doctors need to know which groups are at higher or lower risk and that reaching 90 without dementia should not be treated as being in the clear. Risk reduction, she said, is a conversation worth having with everyone.
For readers, there are two reasonable takeaways. The first is that participation in research is what makes findings about your own community possible; anyone interested can ask a clinician or a local academic medical center about registries. The second is to ask, when reading any health finding, who was studied.
Nothing here changes clinical care today. Adults concerned about memory in a very old relative should seek evaluation rather than assuming decline is inevitable at that age, since treatable causes including medication effects, thyroid disease, depression, hearing loss and vitamin deficiency are common and frequently missed in the oldest patients.
MedicalDaily will report further LifeAfter90 findings as the cohort matures.
Frequently Asked Questions
What did the study find? Dementia risk disparities by sex, race and ethnicity persist after age 90. Women had roughly twice the risk of men, and Black participants had 75 percent greater risk than Asian participants.
How many people were studied? More than 800 participants, median age 92.
Why was this unknown before? Most aging research enrolled people 65 and older, included few participants past 90, and often lacked racial and ethnic diversity.
Who ran the study? UC Davis Health and Kaiser Permanente, through the LifeAfter90 cohort, published in The Lancet Healthy Longevity.
What did it show about APOE? APOE2 remained protective, reducing risk by 60 percent. APOE4 did not appreciably raise risk overall but did among men, and roughly doubled it among Black participants.
Does this explain why disparities exist? No. It documents them. Causes are generally attributed to structural factors including access to care and education.
What can I do? Ask who was studied when reading health findings, and seek evaluation for memory changes in older relatives rather than assuming decline is normal.
