Epidemiology Of Physical Activity Patterns From Wrist-worn Accelerometry In The Baltimore Longitudinal Study Of Aging

Abstract

PURPOSE: To describe the epidemiology of objectively measured physical activity among older adults participating in the Baltimore Longitudinal Study of Aging (BLSA), America’s longest running study of human aging, which switched to wrist-worn accelerometry collected since 2015. METHODS: Data were from 658 participants (mean age 70 years, 55% women) enrolled in the BLSA who were instructed to wear a wrist-worn accelerometer for 7 consecutive days and were seen in clinic between 2015-2019. Accelerometer data were processed into activity counts (unitless quantities of movement) and summarized into three continuous metrics: 1) total activity counts/d; 2) active minutes/d using a threshold of ≥1,853 counts/min; and 3) activity fragmentation defined as the reciprocal of the mean active bout length (consecutively occurring active minutes) x 100 (%). Since the BLSA visit schedule is age-dependent (i.e., older persons have more repeated visits), longitudinal analyses were conducted in a restricted sample aged ≥65 years old (n = 225). Mixed effects models were constructed to estimate longitudinal changes (baseline age x years since baseline) in each accelerometer metric. Years since baseline was treated as a random effect. Covariates included sex, race, education, employment, body mass index, chronic conditions, usual gait speed, and wear days. RESULTS: Participants, aged 22-97 years, had a mean total 2,209,488 activity counts/d, 430 active minutes/d and 24% activity fragmentation at baseline. Mean wear time was high (1440 min/d) and mean wear days was 6 days. The longitudinal sample, aged 65-97 years, descriptively engaged in less physical activity at baseline, with a mean of 2,060,886 activity counts/d, 414 active minutes/d and activity fragmentation of 25%. Mixed effects modeling estimates showed that, per year, total activity counts decreased by 4894 (SE = 1379, p < 0.001) counts/d, activity minutes decreased by 1 (SE = 0.28, p = 0.001) min/d, and activity fragmentation increased by 0.06 (SE = 0.02, p = 0.0001) %. CONCLUSIONS: Findings update objectively measured physical activity patterns previously measured in the BLSA. With more studies switching to wrist accelerometry collection, these results help define and characterize physical activity patterns and longitudinal change, particularly later in life.