In January 2025 a number arrived that was made for headlines: in a long-running US study, about 4 in 10 people who reached 55 without dementia went on to develop it. That was far higher than the figures that had circulated for years.
Over the same years, a different body of research kept reporting something that sounds like the opposite. The chance that a person of a given age develops dementia over the next few years has been dropping in the US and Europe for decades.
Both are true. They answer different questions, and mixing them up leaves people frightened or falsely reassured. The same confusion surrounds APOE, the best-known Alzheimer’s gene, where a test result can sound like a verdict and isn’t one.
Where the 42% figure comes from
The number comes from the Atherosclerosis Risk in Communities study, known as ARIC, which has followed adults in four US communities since 1987: Forsyth County, North Carolina; Jackson, Mississippi; Minneapolis; and Washington County, Maryland. In Nature Medicine in 2025, researchers analyzed 15,043 participants free of dementia at 55, 26.9% of them Black and 55.1% women, over a median of 23 years.
Their estimate of lifetime risk, from 55 to 95, was 42%. The calculation treats death as a competing event: someone who dies of a heart attack at 70 never gets the chance to develop dementia, and the math allows for that instead of pretending everyone lives to 95.
Timing matters. Only 3.9% of participants developed dementia between 55 and 75; after that the curve climbs steeply, and the median age at diagnosis was 81. The 42% mostly describes people who live into their 80s and 90s.
Why is it so much higher than earlier US estimates of 11% to 14% for men and 19% to 23% for women? The authors point mainly to how hard they looked. Earlier studies relied on study visits plus medical and death records, and people whose thinking is slipping are less likely to attend visits. ARIC added regular phone interviews with participants and with informants such as relatives, and those calls found about half of all cases. The cohort is also more racially and economically diverse. Counting phone-detected cases only when hospital or death records confirmed them, lifetime risk was 36%, a reminder that the figure depends on how dementia is counted.
A risk percentage is not your forecast
A lifetime risk is the share of a group that developed dementia before dying or turning 95. It is not a prediction for any one member of that group, and it shifts as you age. For someone dementia-free at 65, the remaining lifetime risk was 45%. At 75 it was 52%, and at 85, 56%. That sounds backward, but the likely reason is simple. People who reach 85 without dementia have already outlived many of the heart attacks, cancers and strokes that take others first, so more of them live through the years when dementia is common.
The claim that nearly half of dementia could be prevented is another population ceiling, unpacked in our piece on the risk factors people rarely guess.
Why the risk at each age has been falling
The second number asks something narrower: among people of a given age, how many develop dementia over a set period, and has that changed?
The Framingham Heart Study in Massachusetts has watched its participants for dementia since 1975. In a 2016 analysis in the New England Journal of Medicine covering 5,205 people 60 and older, the five-year rate of new dementia, adjusted for age and sex, fell from 3.6 per 100 people in the late 1970s and early 1980s to 2.0 per 100 in the late 2000s and early 2010s, a 44% decline. It appeared only among people with at least a high school diploma. Falling vascular risk did not fully explain it, and obesity and diabetes rose over the same years.
A 2020 analysis in Neurology pooled seven cohorts in the US and Europe, 49,202 people over 65 followed between 1988 and 2015, and found incidence falling 13% per decade, consistently across the studies.
National surveys agree. The Health and Retirement Study, which tests a nationally representative sample of older Americans, found that the share of people 65 and older with dementia dropped from 11.6% in 2000 to 8.8% in 2012. A 2022 analysis of the same survey, covering 2000 to 2016, put the age-adjusted decline at 30%, and estimated that rising schooling accounted, statistically, for about 40% of the drop in men and 20% in women. A 2025 study in JAMA Network Open found that in the US, Europe and England, later-born generations had lower dementia prevalence at the same ages.
Nobody fully knows why. The National Institute on Aging says education and healthier behavior may contribute, but cause and effect remain uncertain. Nor is the trend universal: a 2025 review in Current Opinion in Psychiatry noted rising incidence in Japan, Taiwan and South Korea.
Why it doesn’t feel like dementia is declining
If age-specific risk is falling, why does dementia seem to be everywhere? First, more people are reaching their 80s and 90s, where risk is highest. The ARIC team projected new US cases rising from about 514,000 a year in 2020 to about 1 million in 2060, though that projection assumed incidence would stay flat.
Second, diagnoses are recorded more often. A 2022 study in JAMA Health Forum found that the share of Medicare decedents 67 and older with a dementia diagnosis in their last two years of life rose from 34.7% in 2004 to 47.2% in 2017. The curve bent upward between 2011 and 2013, when Medicare claims began carrying more diagnosis codes, yet the patients looked similar across those years. More diagnoses on paper is not more disease.
Five ways of counting the same disease
| Measure | The question it answers | Recent US finding | Direction |
|---|---|---|---|
| Lifetime risk | Of people dementia-free at 55, what share develop it by 95, allowing for deaths from other causes? | 42% (ARIC, 2025) | A single estimate, not a trend |
| Age-specific incidence | Among people of a given age, how many develop it in a set period? | 44% lower in Framingham across three decades | Falling |
| Prevalence | What share of older adults have it right now? | 11.6% in 2000, 8.8% in 2012 (ages 65 and up) | Falling |
| New cases per year | How many people develop it each year? | About 514,000 in 2020, possibly 1 million by 2060 | Rising, mostly from aging |
| Recorded diagnoses | How often does it appear in Medicare records near death? | 34.7% to 47.2% of decedents, 2004 to 2017 | Rising, partly from coding |
When you meet a dementia statistic, find its row before deciding what it means.
Women, race and unequal risk
Women. In ARIC, lifetime risk was 48% for women and 35% for men, and the Alzheimer’s Association reports that almost two-thirds of Americans with Alzheimer’s are women. The main documented driver is survival: ARIC men were nearly twice as likely as women to die without dementia, so fewer reached the ages when it strikes. A 2014 Framingham analysis put lifetime risk at 45 at about 1 in 5 for women and 1 in 10 for men, yet found similar cumulative incidence at the same ages. Men’s cardiovascular death rate at 45 to 54 was about six times women’s, and the authors suggested that the selective survival of men with healthier hearts might partly explain the gap. Other contributors are not ruled out, but longevity accounts for much of it.
Race and ethnicity. A 2016 study of 274,283 Kaiser Permanente members 64 and older found yearly dementia incidence of 26.6 per 1,000 among African American members and 22.2 among American Indian and Alaska Native members, compared with 19.3 among white members and 15.2 among Asian American members. In ARIC, lifetime risk was 44% for Black participants and 41% for white participants, with the gap opening by about 75, and median age at diagnosis was 79 versus 82.
The ARIC authors suggest these disparities may reflect the cumulative effects of structural racism and inequality: unequal access to education and nutrition, more vascular risk in midlife and less access to care. That is their interpretation of the data. None of these numbers describes an individual, and none is fixed.
What APOE adds, and what it doesn’t
APOE is a gene everyone carries in some combination of three common variants, e2, e3 and e4. The e4 variant is considered the strongest genetic risk factor for the common late-onset form of Alzheimer’s disease. The Alzheimer’s Association estimates that 20% to 30% of people in the US carry one or two copies, about 2% carry two, and 40% to 65% of people diagnosed with Alzheimer’s have the variant.
In ARIC, lifetime dementia risk was 39% with no e4 copies, 48% with one copy and 59% with two. Carrying no e4 is no guarantee: nearly 4 in 10 noncarriers developed dementia. And about 4 in 10 people with two copies did not, before death or age 95.
The effect also differs by group. In ARIC, two copies went with a lifetime risk of 64% among white women and 67% among Black women, but 39% among Black men, an estimate with a wide margin of uncertainty. A 1997 meta-analysis in JAMA, pooling data from 40 research teams, found the e4 association weaker on average in African American and Hispanic samples, stronger in Japanese samples, and diminishing after age 70.
The 2024 argument about two copies
In May 2024 a Barcelona-based team, with Juan Fortea of the Sant Pau Memory Unit as first author, argued in Nature Medicine that carrying two copies of APOE e4 amounts to a distinct, genetically determined form of Alzheimer’s disease. Drawing on 3,297 brain donors and 10,039 people in five biomarker studies, they found Alzheimer’s brain changes in almost every carrier of two copies. By 65, nearly all had abnormal amyloid in their spinal fluid and 75% had positive amyloid scans. Symptoms began at about 65, roughly seven to 10 years earlier than in people with two copies of the common e3 variant.
What the study measured matters. It showed near-universal Alzheimer’s biology, meaning brain changes, in convenience samples that were predominantly white and drawn from the US and Europe. The authors write that they could not calculate how often people with two copies go on to develop dementia, and that deaths from other causes can mask the disease. Even the range in which they expected 95% of symptom onsets to fall spanned about 32 years. A strong genetic push is not a date on a calendar.
Family history and the rare inherited forms
A joint practice document from the American College of Medical Genetics and the National Society of Genetic Counselors estimates that parents, siblings and children of a person with Alzheimer’s face a lifetime risk of 15% to 39% without any gene test, two to four times the general population’s.
Truly inherited Alzheimer’s, caused by a single faulty gene passed from parent to child, is rare. The Alzheimer’s Association puts these deterministic genes at 1% or less of cases, with symptoms usually starting from the early 40s to the mid-50s. The genetics document describes the telltale pattern: at least three affected people across two or more generations, two of them parents, siblings or children of the third. The Association adds that a family history is not needed to develop the disease, and the 2024 study notes that parents can pass on APOE e4 without ever showing symptoms.
People with Down syndrome face especially high and early risk. Their extra copy of chromosome 21 carries a gene involved in making amyloid. According to the Alzheimer’s Association, citing the National Down Syndrome Society, about 30% of people with Down syndrome in their 50s and about 50% in their 60s have Alzheimer’s dementia. Early signs are often changes in personality, behavior or daily function rather than memory, and seizures that begin in adulthood can be one of them.
When the question belongs with a clinician
- A change in memory, thinking or behavior, noticed by you or someone close to you. That calls for an evaluation, not a risk calculator; our walk-through of how memory evaluations now work covers the steps, and the gray zone short of dementia is the subject of our article on memory changes.
- A family pattern that fits the inherited form, such as dementia beginning before about 60 to 65, or several affected relatives across generations. Whether genetic testing makes sense for that family is a decision for a genetics professional.
- An adult with Down syndrome who becomes less sociable, loses everyday skills or has a first seizure. Experts cited by the Alzheimer’s Association recommend documenting adult abilities by 35 so changes stand out.
- Confusion that comes on suddenly, or new difficulty speaking or following speech. The CDC treats these as possible stroke signs: call 911 rather than driving to the hospital.
Before an APOE test: what it can and cannot tell you
APOE results are easy to obtain, sometimes without a clinician. US genetics bodies remain cautious. The 2011 ACMG and NSGC document says testing for susceptibility genes such as APOE “is not clinically recommended due to limited clinical utility and poor predictive value.” It says Alzheimer’s genetic testing should happen only alongside genetic counseling, that children should not be tested, and that direct-to-consumer testing is not advised. Both groups have since reclassified it as a practice resource rather than an evidence-based guideline, NSGC in 2016 and ACMG when it reaffirmed the document in 2018. The Alzheimer’s Association’s consumer page takes no position for or against testing but urges counseling before a test is ordered and again when results arrive.
Disclosure research shows why counseling matters. In the REVEAL trials, telling adult children of people with Alzheimer’s their APOE status caused no significant short-term psychological harm in a selected, highly educated group. Yet about half of participants who correctly recalled their result believed their risk differed significantly from what they had been told. E4 carriers were more likely to start dietary supplements, which the genetics document flags as a vulnerability to marketing, and to change long-term care insurance.
Insurance matters too. According to the National Human Genome Research Institute, the Genetic Information Nondiscrimination Act bars health insurers, and employers with 15 or more workers, from using genetic information, but it does not cover long-term care, life or disability insurance. Those gaps get more room in our explainer on what privacy law misses in health apps.
One newer use complicates the picture. The labels of two anti-amyloid antibody drugs for early Alzheimer’s say APOE e4 testing should be performed before treatment starts, because people with two copies, about 15% of Alzheimer’s patients, face a higher risk of brain swelling and bleeding. That is safety screening for people already diagnosed, not a forecast for healthy people; the drugs’ trade-offs belong to the evaluation guide mentioned above. Consumer DNA kits as a category are a broader subject than this one.
Questions to ask before an APOE test
- How would each possible result change what I do? Genetic counselors walk people through both a positive and a negative result before testing for exactly this reason.
- Who will explain the result, before and after? The genetics document calls for counseling by someone with expertise, not an app report read alone.
- Where will the result live, and who can see it? A medical record, a company database or both.
- Could it affect insurance I may want later? GINA’s protections stop short of long-term care, life and disability coverage.
- Who else does it describe? A result carries information about parents, siblings and children who did not choose to learn it.
- Is this about prediction, or about a treatment decision? Those carry different stakes.
The most useful way to hold these numbers
A 42% lifetime risk and a falling age-specific risk are two views of one reality: Americans are living long enough for dementia to become common, while the chance of developing it at any particular age has been dropping. APOE follows the same logic. One e4 copy shifts the odds, two shift them sharply, and no copy count sets a date or a certainty.
Use risk numbers to understand populations, not to grade yourself. What changes your situation is a noticed change in thinking, which deserves an evaluation, or a strong early-onset family pattern, which deserves a genetics conversation. The risks you can change, from blood pressure to hearing, are weighed in our coverage of overlooked risk factors and of multidomain prevention trials.
Keep exploring: social isolation and dementia and study techniques.
Sources
- Nature Medicine. “Fang M, Hu J, Weiss J, et al. "Lifetime risk and projected burden of dementia." Nat Med, 2025;31(3):772-776. PubMed. DOI: 10.1038/s41591-024-03340-9.” pubmed.ncbi.nlm.nih.gov
- New England Journal of Medicine. “Satizabal CL, Beiser AS, Chouraki V, et al. "Incidence of Dementia over Three Decades in the Framingham Heart Study." N Engl J Med, 2016;374(6):523-532. PubMed. DOI: 10.1056/NEJMoa1504327.” pubmed.ncbi.nlm.nih.gov
- Neurology. “Wolters FJ, Chibnik LB, Waziry R, et al. "Twenty-seven-year time trends in dementia incidence in Europe and the United States: The Alzheimer Cohorts Consortium." Neurology, 2020;95(5):e519-e531. PubMed. DOI: 10.1212/WNL.0000000000010022.” pubmed.ncbi.nlm.nih.gov
- JAMA Internal Medicine. “Langa KM, Larson EB, Crimmins EM, et al. "A Comparison of the Prevalence of Dementia in the United States in 2000 and 2012." JAMA Intern Med, 2017;177(1):51-58. PubMed. DOI: 10.1001/jamainternmed.2016.6807.” pubmed.ncbi.nlm.nih.gov
- Proceedings of the National Academy of Sciences. “Hudomiet P, Hurd MD, Rohwedder S. "Trends in inequalities in the prevalence of dementia in the United States." Proc Natl Acad Sci U S A, 2022;119(46):e2212205119. PubMed. DOI: 10.1073/pnas.2212205119.” pubmed.ncbi.nlm.nih.gov
- JAMA Network Open. “Dou X, Lenzen S, Connelly LB, Lin R. "Generational Differences in Age-Specific Dementia Prevalence Rates." JAMA Netw Open, 2025;8(6):e2513384. PubMed. DOI: 10.1001/jamanetworkopen.2025.13384.” pubmed.ncbi.nlm.nih.gov
- Current Opinion in Psychiatry. “Vella AS, Roehr S, Sachdev PS. "The incidence of dementia in populations around the globe." Curr Opin Psychiatry, 2026;39(2):168-174 (online October 2025). PubMed. DOI: 10.1097/YCO.0000000000001053.” pubmed.ncbi.nlm.nih.gov
- JAMA Health Forum. “Davis MA, Chang CH, Simonton S, Bynum JPW. "Trends in US Medicare Decedents' Diagnosis of Dementia From 2004 to 2017." JAMA Health Forum, 2022;3(4):e220346. PubMed. DOI: 10.1001/jamahealthforum.2022.0346.” pubmed.ncbi.nlm.nih.gov
- Alzheimer's & Dementia. “Chêne G, Beiser A, Au R, et al. "Gender and incidence of dementia in the Framingham Heart Study from mid-adult life." Alzheimers Dement, 2015;11(3):310-320 (online January 2014). PubMed. DOI: 10.1016/j.jalz.2013.10.005.” pubmed.ncbi.nlm.nih.gov
- Alzheimer's & Dementia. “Mayeda ER, Glymour MM, Quesenberry CP, Whitmer RA. "Inequalities in dementia incidence between six racial and ethnic groups over 14 years." Alzheimers Dement, 2016;12(3):216-224. PubMed. DOI: 10.1016/j.jalz.2015.12.007.” pubmed.ncbi.nlm.nih.gov
- JAMA. “Farrer LA, Cupples LA, Haines JL, et al. "Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis." JAMA, 1997;278(16):1349-1356. PubMed. PMID: 9343467 (no DOI in record).” pubmed.ncbi.nlm.nih.gov
- Nature Medicine. “Fortea J, Pegueroles J, Alcolea D, et al. "APOE4 homozygosity represents a distinct genetic form of Alzheimer's disease." Nat Med, 2024;30(5):1284-1291. PubMed. DOI: 10.1038/s41591-024-02931-w (Publisher Correction: Nat Med 2024;30(7):2093, DOI: 10.1038/s41591-024-03127-y).” pubmed.ncbi.nlm.nih.gov
- Genetics in Medicine (American College of Medical Genetics and National Society of Genetic Counselors). “Goldman JS, Hahn SE, Catania JW, et al. "Genetic counseling and testing for Alzheimer disease: joint practice guidelines of the American College of Medical Genetics and the National Society of Genetic Counselors." Genet Med, 2011;13(6):597-605. PubMed. DOI: 10.1097/GIM.0b013e31821d69b8.” pubmed.ncbi.nlm.nih.gov
- American College of Medical Genetics and Genomics. “ADDENDUM: Genetic counseling and testing for Alzheimer disease: joint practice guidelines (reaffirmed 25 June 2018; reclassified as a Practice Resource).” acmg.net
- National Institute on Aging (Alzheimers.gov). “Can I Prevent Dementia? (updated June 16, 2026).” alzheimers.gov
- Alzheimer's Association. “Alzheimer's Disease Facts and Figures (2026 report page).” alz.org
- Alzheimer's Association. “Is Alzheimer's Genetic?.” alz.org
- Alzheimer's Association. “Down Syndrome and Alzheimer's Disease.” alz.org
- National Human Genome Research Institute. “Genetic Discrimination (updated January 6, 2022).” genome.gov
- National Library of Medicine (DailyMed), FDA-approved labeling. “Lecanemab-irmb injection label, boxed warning (updated July 21, 2026).” dailymed.nlm.nih.gov
- National Library of Medicine (DailyMed), FDA-approved labeling. “Donanemab-azbt injection label, boxed warning (updated December 30, 2025).” dailymed.nlm.nih.gov
- Centers for Disease Control and Prevention. “Signs and Symptoms of Stroke (reviewed May 19, 2026).” cdc.gov