Day-to-day variability of multiple sensor-based physical activity parameters in older persons with dementia

Objective: To investigate the day-to-day variability of various sensor-based physical activity (PA) parameters and to analyze effects of weekdays vs. weekend days as well as the influence of concomitant factors (gender, living conditions, cognitive status, physical performance, and level of PA) in multi-morbid, older persons with mild-to-moderate stage dementia. Methods: In 53 participants, PA was recorded on three consecutive days. Day-to-day variability was estimated by calculating intraclass correlation coefficients (ICCs) for two consecutive days each (Friday-Saturday, Saturday-Sunday). Results: Almost all non-walking parameters (lying time, sitting time, standing time, active time, inactive time, and sit-to-stand transitions) showed a consistently low day-to-day variability for Friday-Saturday as well as Saturday-Sunday (ICCs: .60-.96) and hence remained almost unaffected by specific days of the week and concomitant factors. Only the sub-analysis by level of PA revealed slightly deviating results (ICCs: .38-.89). The walking parameters (walking time, walking episodes, and steps) revealed a higher day-to-day variability for Friday-Saturday (ICCs: .01-.40) and a generally lower variability for Saturday-Sunday (ICCs: - .08 - .88), also depending on the respective concomitant factors. Conclusions: Two consecutive days are adequate to reliably assess non-walking parameters, whereas walking parameters showed higher day-to-day variability with a relevant influence of type of days and concomitant factors.