Validation of the Fitbit Charge 2 compared to the ActiGraph GT3X+ in older adults with knee osteoarthritis in free-living conditions

Objective: To evaluate physical activity (PA) and sedentary time in subjects with knee osteoarthritis (OA) measured by the Fitbit Charge 2 (Fitbit) and a wrist-worn ActiGraph GT3X+ (AGW) compared to the hip-worn ActiGraph (AGH). Design: We recruited a cohort of subjects with knee OA from rheumatology clinics. Subjects wore the AGH for four weeks, AGW for two weeks, and Fitbit for two weeks over a four-week study period. We collected accelerometer counts (ActiGraphs) and steps (ActiGraphs, Fitbit) and calculated time spent in sedentary, light, and moderate-to-vigorous activity. We used triaxial PA intensity count cut-points from the literature for ActiGraph and a stride length-based cadence algorithm to categorize Fitbit PA. We compared Fitbit wear times calculated from a step-based algorithm and a novel algorithm that incorporates steps and heart rate (HR). Results: We enrolled 15 subjects (67% female, mean age 68 years). Relative to AGH, Fitbit, on average, overestimated steps by 39% and sedentary time by 37% and underestimated MVPA by 5 minutes. Relative to AGH, AGW overestimated steps 116%, underestimated sedentary time by 66%, and captured 281 additional MVPA minutes. The step-based wear time Fitbit algorithm captured 14% less wear time than the HR-based algorithm. Conclusions: Fitbit overestimates steps and underestimates MVPA in knee OA subjects. Cut-offs validated for AGW should be developed to support the use of AGW for PA assessment. The HR-based Fitbit algorithm captured more wear time than the step-based algorithm. These data provide critical insight for researchers planning to use commercially-available accelerometers in pragmatic studies.