Validating the Fitbit Charge 4© wearable activity monitor for use in physical activity interventions

Abstract

Objectives: Commercially available wearable activity monitors can promote physical activity behaviour. Clinical trials typically quantify physical activity with research grade activity monitors prior to testing interventions utilising commercially available wearable activity monitors aimed at increasing step count. Therefore, it is important to test the agreement of these two types of activity monitors.

Objectives: Observational.

Methods: Thirty adults (20-65 years, n = 19 females) were provided a Fitbit Charge 4©. To determine reliability using an intraclass correlation coefficient, two, one-minute bouts of treadmill walking were performed at a self-selected pace. Subsequently, participants wore both an ActiGraph wGT3X-BT and the Fitbit for seven days. To determine agreement, statistical equivalence and the mean absolute percentage error were calculated and represented graphically with a Bland-Altman plot. Ordinary least products regression was performed to identify fixed or proportional bias.

Results: The Fitbit showed 'good' step count reliability on the treadmill (intraclass correlation coefficient = 0.75, 95 % CI = 0.53-0.87, p < 0.001). In free-living however, it overestimated step count when compared to the ActiGraph wGT3X-BT (mean absolute percentage error = 26.02 % ± 14.63). Measurements did not fall within the ± 10 % equivalence region and proportional bias was apparent (slope 95 % CI = 1.09-1.35).

Conclusions: The Fitbit Charge 4© is reliable when measuring step count on a treadmill. However, there is an overestimation of daily steps in free-living environments which may falsely indicate compliance with physical activity recommendations.

Keywords: Accelerometer; ActiGraph; Agreement; Reliability; Step count.