Dementia is receding in younger generations: what a comparison of the US, Europe and England has shown

Good news from the epidemiology of aging: people born later are less likely to have dementia at the same age than their grandparents. This does not mean that cases will decrease in absolute numbers (the population is aging rapidly), but the age-specific risk is decreasing in almost all regions and is more noticeable in women. This is the conclusion of an international team that analyzed data from the US, Europe, and England in JAMA Network Open.

Background

Over the past three decades, the “aging paradox” has been increasingly observed in rich countries: the population is rapidly aging, the absolute number of dementia cases is growing, but age-specific indicators (risk in people of the same age) are gradually decreasing. The scientific explanation consists of several lines. First, the “cognitive reserve” has increased: education has become longer and better, intellectually and socially active lifestyles are more widespread. Second, the control of vascular risk factors has improved dramatically - arterial hypertension, cholesterol, diabetes, smoking; it is vascular damage to the brain that is important for both Alzheimer's and mixed dementia. Third, the background determinants of health are improving - nutrition, hearing aids, vision, treatment of depression, air quality and medicine in general.

To separate the “luck of a particular era” from a stable trend, epidemiologists compare birth cohorts: what is the proportion of people with dementia at the same age, but born earlier vs. later. This cohort approach allows us to partially separate three effects - age, calendar time (improved diagnostics, access to care) and generational (different childhood, education, habits, medicine throughout life). It is fundamentally important for planning: if the age-related risk decreases, then old forecasts of the burden on health care and long-term care systems may be overstated, despite the “silver wave”.

There are methodological nuances, too. The prevalence of dementia is a function of incidence (how many new cases arise) and survival (how long people live with the diagnosis). It is affected by diagnostic changes (algorithms, scales, criteria), “survivor bias,” migration, and gender differences: women have historically had faster increases in education and vascular risk control, which may give them a greater “cohort gain.” Large international panels with repeated measurements are therefore the best tool to see how risk changes across generations and where “bottlenecks” still remain (obesity and diabetes are “getting younger,” loneliness and depression are scaling up, air pollution is declining unevenly).

Against this backdrop, a new study in JAMA Network Open does exactly what was missing: it compares age-specific prevalence of dementia between earlier and later cohorts in several regions, looks at men and women separately, and uses uniform cognitive classification algorithms. This lens helps both health policy (updating personnel and budget forecasts), prevention (keeping the focus on modifiable factors), and the clinic (better understanding where exactly generational gains are “working” – education, vascular health, hearing, etc.).

How it was studied

• We took three long-term population panels: US Health and Retirement Study (HRS, 1994-2021), Survey of Health, Ageing and Retirement in Europe (SHARE, 2004-2020) and English Longitudinal Study of Ageing (ELSA, 2002-2019). A total of 62,437 people aged 70+.
• Participants were divided into birth cohorts and age-specific prevalence of dementia was compared between earlier and later cohorts at the same age.
• An algorithmic classification of dementia was used (a combination of cognitive tests, functional limitations, etc.), and the models took into account age and calendar period. The results are a cross-sectional comparison of cohorts.

The result is a clear "tilt" in favour of later generations. For example, among people aged 81-85 in the US, the proportion with dementia fell from 25.1% (born 1890-1913) to 15.5% (born 1939-1943); in Europe, from 30.2% (1934-1938) to 15.2% (1939-1943). In England, the trend is milder: 15.9% (1924-1928) versus 14.9% (1934-1938). The greatest decline was observed among women; among men in England, it was statistically inconclusive.

Why this is important now

• Resource planning: Forecasts of care and staffing needs often get stuck on old age factors. Taking cohort changes into account reduces the risk of overestimating the load on the system - and helps to allocate funding more accurately.
• Message to the population: "debut at a later age" is a reality in many countries. This is a window to maintain independence and quality of life longer.
• But the absolute numbers will rise: even with a lower risk at each age, the total number of cases will rise because of the “silver wave.” This double truth is what policy should be guided by.

What could be behind the improvement? The authors did not directly test the causes, but journalistic and university comments point to familiar drivers: better education, control of vascular factors (blood pressure, diabetes, cholesterol), less smoking, access to medical care, possibly better air and hearing aids. At the same time, some of the “major gains” could have already happened in the 20th century, so it might be naive to expect such a sharp decline further down the line.

What does this change in practice and policy?

• Update forecasts: Primary care, neurology, and long-term care caseload models should include cohorts (not just age and gender).
• The focus is on prevention: even with a downward trend, modifiable factors remain - hypertension, obesity, inactivity, depression, isolation, hearing loss, air pollution. Here, cheap measures have a high systemic effect.
• Gender lens: The risk decline is greater for women, likely due to historical increases in education enrollment. This suggests that investments in education and cognitive capital have a long prevention tail.

Important Disclaimers

• Design: Cross-sectional comparison of cohorts rather than following the same individuals throughout life; survivorship effects and "hidden" differences in diagnosis between waves are always possible.
• Algorithmic diagnosis: This standardizes assessment but does not replace clinical diagnosis in each individual.
• Does not look for reasons: the work is descriptive - it does not prove why the risk falls; therefore, any interpretations of factors are hypotheses, not conclusions.

What do science and managers need next?

• Decompose the contribution of factors: education, vascular health, smoking, hearing, air - how much does each of them contribute in different countries and generations?
• Keeping an eye on the 'upstarts': obesity and type II diabetes are 'getting younger' - will they ruin the positive trend among those born after the 1960s?
• Scenario planning: build budgets and care staffing taking into account the decline in age-related risk, but the growth in the absolute number of elderly people - otherwise it is easy to miss in both directions.

Conclusion

All things being equal, your peers from a later cohort have a lower risk of dementia than people of the same age who were born earlier. This is not a reason to relax, but a reason to fine-tune prevention and planning so that the generational gains do not dissolve in the tsunami of an aging world.

Source: Dou X. et al. Generational Differences in Age-Specific Dementia Prevalence Rates. JAMA Network Open, 2 June 2025 (e2513384). Additional context: University of Queensland press release and media coverage. doi:10.1001/jamanetworkopen.2025.13384