Association of Physical Activity Level With Risk of Dementia in a Nationwide Cohort in Korea

Importance: Current guidelines recommend 500 to 999 metabolic equivalent (MET)-minutes per week (MET-min/wk) of regular physical activity. However, evidence regarding the association between light-intensity physical activity (LPA) and dementia in older adults is inconsistent. Objective: To assess the association between physical activity and new-onset dementia, focusing on the dose-response association between physical activity and dementia and the association of LPA with the incidence of dementia. Design, Setting, and Participants: For this nationwide retrospective cohort study, we analyzed 62 286 participants aged 65 years or older without preexisting dementia who had available health checkup data from the Korean National Health Insurance Service database from January 2009 to December 2012. Participants were followed up until December 31, 2013, and data analysis was performed from July 2020 to January 2021. Exposures: Physical activity level was assessed using a standardized, self-reported questionnaire at baseline. Physical activity-related energy expenditure (in MET-min/wk) was calculated by summing the product of frequency, intensity, and duration. Main Outcomes and Measures: Incidence of dementia. Incidence rates were calculated by dividing the number of events by the person-time at risk (presented as the incidence per 1000 person-years). Hazard ratios (HRs) and 95% CIs for dementia were analyzed according to physical activity level. Competing risk regression was performed by using the Fine-Gray subdistribution hazard model, with mortality as the competing risk for dementia events. Multivariable regression models were constructed with adjustment for various patient characteristics. Incident dementia occurring 2 years after enrollment was assessed, and separate analyses included all follow-up periods. Restricted cubic spline curves were used to examine the association of continuous values of physical activity with dementia. Results: Among 62 286 participants, 60.4% were women, and the mean (SD) age was 73.2 (5.3) years. During a median follow-up of 42 months, 3757 participants (6.0%) developed dementia, and the overall incidence was 21.6 per 1000 person-years. Compared with inactive individuals (0 MET-min/wk), insufficiently active (1-499 MET-min/wk; mean, 284 MET-min/wk), active (500-999 MET-min/wk; mean, 675 MET-min/wk), and highly active participants (≥1000 MET-min/wk; mean, 1627 MET-min/wk) showed 10% (adjusted hazard ratio HR, 0.90; 95% CI, 0.81-0.99), 20% (adjusted HR, 0.80; 95% CI, 0.71-0.92), and 28% (adjusted HR, 0.72; 95% CI, 0.60-0.83) reduced dementia risk, respectively. Thus, a progressive decrease in the adjusted HR of dementia was associated with increasing physical activity level, and a restricted cubic spline curve showed that this association started with a low amount of total physical activity. This association was consistent regardless of age, sex, and other comorbidities or after censoring for stroke. Compared with total sedentary behavior, even a low amount of LPA (1-299 MET-min/wk; mean, 189 MET-min/wk) was associated with reduced dementia risk (adjusted HR, 0.86; 95% CI, 0.74-0.99). Conclusions and Relevance: In older adults, an increased physical activity level, including a low amount of LPA, was associated with a reduced risk of dementia. Promotion of LPA might reduce the risk of dementia in older adults.