Surviving to 90 without dementia is not, by itself, a sign that the risk has passed. A prospective cohort study called LifeAfter90, published September 1 in The Lancet Healthy Longevity, followed more than 800 Kaiser Permanente members aged 90 and older who were cognitively healthy at enrollment, tracking them with cognitive exams every six months since 2018. Researchers found that dementia risk in this "oldest-old" group remained strongly shaped by sex, race and a well-known Alzheimer's-linked gene — disparities seen earlier in life that do not disappear once someone reaches 90.
Sharp disparities by sex and race
Women in the study faced roughly twice the dementia risk of men after 90. Racial and ethnic disparities were also pronounced: Black participants had a 75% higher incidence rate than Asian participants, with elevated risk among Black and Hispanic participants compared with white and Asian participants. The APOE gene, long linked to Alzheimer's risk, showed a more complicated pattern at this advanced age: the protective APOE2 variant was associated with a 60% lower dementia risk, while APOE4 — usually a strong risk factor — showed no clear increase in risk across the population as a whole, but nearly doubled risk specifically among Black participants and was elevated in men.
"We can't just assume that someone from a high-risk group makes it to 90 without dementia and they're in the clear," said Rachel Whitmer, one of the study's researchers, according to ScienceDaily's coverage of the findings.
We can't just assume that someone from a high-risk group makes it to 90 without dementia and they're in the clear.
— Rachel Whitmer, UC Davis
The study is observational, not a controlled trial, so it can identify associations but cannot prove that sex, race or genotype directly cause the differences in risk. Researchers caution that racial disparities in dementia incidence plausibly reflect a lifetime of unequal access to healthcare, education and economic opportunity as much as biology — factors a cohort study of this kind cannot fully separate from race itself. The sample, while notably diverse for research on people over 90, is still just over 800 people drawn from a single health system, meaning the racial and genetic subgroup estimates carry wider uncertainty than the topline figures suggest, and may not generalize to all populations.
One further finding complicates the picture: some participants who had classic midlife risk factors, including hypertension and high cholesterol in their 60s and 70s, nonetheless remained cognitively healthy deep into their 90s — evidence, the researchers say, that unidentified protective mechanisms are also at work and that risk in the oldest-old is not simply a matter of accumulated damage.
