PHYSICAL ACTIVITY AND DEMENTIA IN THE NATIONAL HEALTH AND AGING TRENDS STUDY (NHATS)

Abstract Individuals living with dementia exhibit disrupted sleep/wake rhythms; however, whether daily physical activity volume and patterns are worse among those living with dementia is not well established. We examined the cross-sectional association between dementia classification and accelerometry-derived patterns of daily physical activity in NHATS. Wrist-worn accelerometry was collected using the Actigraph Centrepoint Insight Watch in a subset of Round 11 (2021) participants. Dementia status, including possible and probable dementia, was determined using the NHATS dementia algorithm. We characterized daily physical activity patterns using: most active 10-hours/day (M10), time of M10, least active 5-hours/day (L5), time of L5, relative amplitude (RA), intra-daily variability (IV), inter-daily stability (IS), and activity fragmentation using the active-to-sedentary transition probability (ASTP). We used separate linear regression models to study the association between dementia and each actigraphy feature, adjusting for relevant demographic features including age, gender, race, BMI, and education. 721 participants had ≥3 days of valid accelerometry data, dementia status, and complete covariate data (40% aged ≥80 years, 54% women). Compared to cognitively normal participants, those living with dementia averaged fewer activity counts/day (beta: -382K counts/day p