Comparison of Outcomes Derived from the ActiGraph GT3X+ and the Axivity AX3 Accelerometer to Objectively Measure 24-Hour Movement Behaviors in Adults: A Cross-Sectional Study.

Marieke De Craemer,Vera Verbestel

doi:10.3390/ijerph19010271

Abstract

The aim of this cross-sectional study is to investigate convergent validity of outcomes derived from the ActiGraph GT3X and Axivity accelerometer and to investigate feasibility and acceptability of using outcomes derived from these devices to measure 24-h movement behaviors (i.e., sleep, sedentary behavior, and physical activity) in adults during free-living conditions. Twenty-four adults (33.3% male; 39.79 ± 13.09 years old) simultaneously wore a hip-mounted ActiGraph and thigh-mounted Axivity during 48 consecutive hours. The ActiGraph was switched from hip to wrist during the night. To assess convergent validity between the two devices, Paired sample t-tests, linear regressions, Bland–Altman plots and κ statistic were conducted. Feasibility and acceptability of the devices was self-reported on a five-point scale. Daily time spent in 24-h movement behaviors differed between both devices. Except for sleep, the mean differences in daily time spent in the behaviors were clinically relevant and the limits of agreement were wide. For all 24-h movement behaviors, except for sleep and vigorous physical activity, proportional errors were observed. κ statistic showed substantial agreement for total physical activity and outstanding agreement for sedentary behavior and sleep. Adults perceived the Axivity as more practical than the ActiGraph to wear it for more than two days whereas the feasibility to wear it for two days was comparable. Depending on the measure, the observed biases were clinically relevant, proportional to the size of the measure and/or erratically variable. When taking into account the data processing approaches applied in this study, the ActiGraph and the Axivity cannot be used interchangeably to measure 24-h movement behaviors since the bias between both devices is clinically relevant and unpredictable.

Highlights

A shift in research occurred going from studying movement behaviors in isolation (e.g., physical activity (PA)) towards an integrated approach in which all behaviors conducted within a 24-h time span are distributed across a movement continuum going from no movement (sleep, sedentary behavior (SB)) to movement (categorized from low movement towards moderate PA (MPA) and vigorous PA
Paired sample t-tests showed a significant difference in daily time spent in SB (t = 8.42, p < 0.001), sleep (t = −3.42, p = 0.001), Light intensity physical activity (LIPA) (t = 12.57, p < 0.001), Moderate intensity physical activity (MPA) (t = −11.39, p < 0.001), Vigorous intensity physical activity (VPA) (t = −4.30, p < 0.001), moderate-to-vigorous PA (MVPA) (t = −13.79, p < 0.001), and total PA (t = −6.89, p < 0.001) between both devices (Table 1)
We found no studies of which the results were generalizable to this study, our results are consistent with the major finding of Rosenberger and colleagues’ (2016) review that several wearable devices are not able to accurately measure MVPA

Summary

Introduction

A shift in research occurred going from studying movement behaviors in isolation (e.g., physical activity (PA)) towards an integrated approach in which all behaviors conducted within a 24-h time span are distributed across a movement continuum going from no movement (sleep, sedentary behavior (SB)) to movement (categorized from low movement (light-intensity PA or LIPA) towards moderate PA (MPA) and vigorous PA (VPA)). These behaviors interact, which means that time spent on one or more of these behaviors (e.g., PA) has a consequence on the time that can be spend on the other behaviors (i.e., SB and sleep) [1].

Objectives

Methods

Results

Conclusion