The final analytic samples included 1526 early adult, 1943 midlife, and 885 late life participants. All participants provided informed consent, and the study followed established guidelines.
Physical Activity Measurement Using the Activity Index
Physical activity was measured using the Physical Activity Index, which assigns weighted values to hours spent in sleep, sedentary, slight, moderate, and heavy activities. Higher scores indicate greater total activity. Scores were divided into age-specific quintiles for comparison.
Participants were followed until dementia diagnosis, death, or December 31, 2023. Dementia, all-cause or Alzheimer’s disease, was diagnosed through expert consensus using standard diagnostic criteria.
Cox proportional hazards models estimated hazard ratios for dementia risk across physical activity quintiles, adjusting for age, sex, education, body mass index (BMI), smoking, hypertension, diabetes, hyperlipidemia, and apolipoprotein E (APOE) ε4 status.
Additional analyses considered tertiles and quartiles of activity, activity intensity levels, APOE ε4 stratification, and Alzheimer’s dementia specifically.
Dementia Incidence and Early Patterns Across Life Stages
Participants in late life generally had lower education, higher BMI, and more cardiometabolic conditions than younger groups. Throughout follow-up, 567 individuals developed dementia, 4 percent of early adult, 14 percent of midlife, and 26 percent of late life participants.
Across all age groups, participants with lower physical activity had higher mortality. These results emphasize the higher risk of dementia among less active individuals, especially in midlife and late life.
After accounting for demographic and health factors, only higher physical activity during midlife and late life, not early adulthood, was associated with a significantly reduced risk of all-cause dementia.
In both midlife and late life, individuals in the highest activity quintiles had a significantly lower dementia risk than those in the lowest quintile, approximately 40 percent lower, with Q4 and Q5 showing the strongest associations.
Early adult activity showed no significant association with dementia risk, likely reflecting the small number of dementia cases in this age group.
APOE ε4 Genetics Modify Midlife and Late-Life Activity Effects
APOE ε4 status modified these associations. The interaction between APOE ε4 and physical activity was statistically significant for midlife only, indicating that genetic risk influenced the strength of the association during this period.
Among non-carriers, higher midlife activity was linked to substantially lower dementia risk, while late-life activity predicted reduced risk for both carriers and non-carriers. In ε4 carriers, only the highest late-life activity level, Q5, reached statistical significance.
Patterns were similar for Alzheimer’s dementia specifically, with the highest level of midlife and late-life activity associated with lower incidence. However, for Alzheimer’s disease in ε4 carriers, the association did not reach statistical significance, a nuance not attributable to insufficient sample size in the late-life group.
Moderate and heavy midlife activity were particularly protective, whereas slight activity showed no benefit. In late life, benefit was observed regardless of activity intensity.
Strengths and Limitations of the Life-Course Physical Activity Approach
The study shows that midlife and late-life physical activity are key periods during which higher activity levels are associated with reduced risk of all-cause and Alzheimer’s dementia.
Midlife benefits were tied to moderate or heavy activity, whereas late-life benefits occurred regardless of intensity. APOE ε4 status influenced these associations, with stronger effects in non-carriers during midlife and more limited effects in carriers until late life.
These findings reinforce the importance of promoting physical activity across adulthood, particularly during midlife, to support cognitive health.
The life course design, extended follow-up, comprehensive adjudication of dementia, and detailed covariate data are significant strengths of this analysis. However, the predominantly European ancestry sample, limited numbers of dementia cases in early adulthood, reliance on self-reported activity at a single time point, inability to assess earlier-life activity, and potential misclassification or survival bias may limit the generalizability of the results.
Overall, the study identifies midlife and late life as critical windows for physical-activity-based dementia prevention.