Responsiveness of Patient-Reported and Device-based Physical Activity Measures: Secondary Analysis of Four Randomised Trials

Abstract

ABSTRACT Purpose This study aimed to compare the responsiveness of patient-reported and device-based instruments within four physical activity trials. Methods This was a secondary analysis of four randomised trials that used both a patient-reported outcome measure (the Incidental and Planned Exercise Questionnaire, IPEQ) and a device-based instrument (ActiGraph or ActivPAL) to measure physical activity. The four trials included were (i) Activity and MObility UsiNg Technology (AMOUNT): digitally-enabled exercises in those undertaking aged care and neurological rehabilitation; (ii) Balance Exercise Strength Training (BEST) at Home: home-based balance and strength exercises in community-dwelling people aged ≥65 years; (iii) Coaching for Healthy AGEing (CHAnGE): physical activity coaching and fall prevention intervention in community-dwelling people aged ≥60 years; and (iv) Fitbit trial: fall prevention and physical activity promotion with health coaching and activity monitor in community-dwelling people aged ≥60 years. We estimated treatment effects for all variables within each physical activity instrument using regression analyses and expressed results as effect sizes (ES). Results Overall, device-based instruments were more responsive among healthy older adults (ES range: 0.01 to 0.32), whereas the IPEQ was more responsive among adults requiring rehabilitation (ES range: -0.06 to 0.35). Both the IPEQ and device-based instruments were more responsive in trials that promoted walking via coaching participants to increase their daily steps (AMOUNT [ES range: -0.06 to 0.35], CHAnGE [ES range: -0.24 to 0.22] and Fitbit trial [ES range: -0.23 to 0.32]). Individual variables within the IPEQ and device-based instruments varied in their responsiveness (ES range: -0.13 to 0.20). Conclusions Both the IPEQ and device-based instruments are able to detect small changes in physical activity levels. However, responsiveness varies across different interventions and populations. Our findings provide guidance for researchers and clinicians in selecting an appropriate instrument to measure changes in physical activity.