Moderate and vigorous leisure time physical activity in older adults and Alzheimer's disease-related mortality in the USA: a dose-response, population-based study.

Background: Data for the dose-response associations of moderate physical activity (MPA) and vigorous physical activity (VPA) with Alzheimer's disease-related mortality are scarce. We aimed to examine the prospective associations of such activity with Alzheimer's disease-related mortality.

Methods: In this dose-response, population-based study, we pooled data from 22 consecutive waves of the US National Health Interview Survey (from 1997 to 2018). Participants aged 68 years or older were included in the study if they had full data for physical or muscle-strengthening activity, chronic conditions, functional limitations, marital status, education level, alcohol consumption, smoking status, and BMI, and follow-up time after study entry. Participants were linked to the National Death Index until Dec 31, 2019. We defined Alzheimer's disease-related mortality as a leading cause by using the G-30 code of the International Statistical Classification of Diseases and Related Health Problems tenth revision. Information on MPA and VPA was self-reported, with participants asked to answer four questions on the frequency and duration of both types of physical activity. We calculated hazard ratios (HRs) and used restricted cubic splines models to assess dose-response associations, and we estimated the annual number of Alzheimer's disease-related deaths that could be prevented through physical activity using adjusted population attributable fractions.

Findings: After excluding 21 377 participants, 91 298 adults were included in the analysis. The weighted mean participant age was 75·8 years (SE 0·0); 55 658 (56·7%) were female and 76 796 (87·6%) were White (weighted percentages). The median follow-up was 6·5 years (IQR 3·6-10·7), corresponding to 703 393 person-years. 2176 (2·4%) participants died due to Alzheimer's disease as the leading cause. For MPA, we did not find a significant dose-response association with Alzheimer's disease-related mortality, whereas for VPA, we observed a significant L-shaped association between 20 weekly min and 190 weekly min. For VPA, we identified a minimal amount (ie, 50% of the optimal amount) at 40 min/week (HR 0·91, 95% CI 0·84-0·95) and an optimal amount (ie, the nadir of the curve) at 140 min/week (0·79, 0·66-0·95) for reducing Alzheimer's disease-related mortality. For the USA, we estimated that 40 weekly min of VPA would prevent 12 238 deaths per year (95% CI 89-23 172) and 140 weekly min of VPA would prevent 37 710 deaths per year (311-63 567), compared with a scenario in which US adults did not do any VPA.

Interpretation: These findings might inform future guidelines for preventing Alzheimer's disease-related mortality by emphasising the importance of VPA over MPA and providing specific VPA targets.

Funding: None.