Estimating physical activity and sedentary behaviour in a free-living environment: A comparative study between Fitbit Charge 2 and Actigraph GT3X
Open Access
- 11 June 2020
- journal article
- research article
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 15 (6) , e0234426
Abstract
Activity trackers such as the Fitbit Charge 2 enable users and researchers to monitor physical activity in daily life, which could be beneficial for changing behaviour. However, the accuracy of the Fitbit Charge 2 in a free-living environment is largely unknown. To investigate the agreement between Fitbit Charge 2 and ActiGraph GT3X for the estimation of steps, energy expenditure, time in sedentary behaviour, and light and moderate-to-vigorous physical activity under free-living conditions, and further examine to what extent placing the ActiGraph on the wrist as opposed to the hip would affect the findings. 41 adults (n = 10 males, n = 31 females) were asked to wear a Fitbit Charge 2 device and two ActiGraph GT3X devices (one on the hip and one on the wrist) for seven consecutive days and fill out a log of wear times. Agreement was assessed through Bland-Altman plots combined with multilevel analysis. The Fitbit measured 1,492 steps/day more than the hip-worn ActiGraph (limits of agreement [LoA] = -2,250; 5,234), while for sedentary time, it measured 25 min/day less (LoA = -137; 87). Both Bland-Altman plots showed fixed bias. For time in light physical activity, the Fitbit measured 59 min/day more (LoA = -52;169). For time in moderate-to-vigorous physical activity, the Fitbit measured 31 min/day less (LoA = -132; 71) and for activity energy expenditure it measured 408 kcal/day more than the hip-worn ActiGraph (LoA = -385; 1,200). For the two latter outputs, the plots indicated proportional bias. Similar or more pronounced discrepancies, mostly in opposite direction, appeared when comparing to the wrist-worn ActiGraph. Moderate to substantial differences between devices were found for most outputs, which could be due to differences in algorithms. Caution should be taken if replacing one device with another and when comparing results.Keywords
Funding Information
- Horizon 2020 (643309)
- Oak Foundation (OCAY-18-774-OFIL)
This publication has 54 references indexed in Scilit:
- Assessment of Wear/Nonwear Time Classification Algorithms for Triaxial AccelerometerMedicine & Science in Sports & Exercise, 2012
- The Relationship of Actigraph Accelerometer Cut-Points for Estimating Physical Activity With Selected Health OutcomesResearch Quarterly for Exercise and Sport, 2012
- The Relationship of Actigraph Accelerometer Cut-Points for Estimating Physical Activity With Selected Health Outcomes: Results From NHANES 2003–06Research Quarterly for Exercise and Sport, 2012
- Technical variability of the GT3X accelerometerMedical Engineering & Physics, 2012
- Validation and comparison of ActiGraph activity monitorsJournal of Science and Medicine in Sport, 2011
- An artificial neural network to estimate physical activity energy expenditure and identify physical activity type from an accelerometerJournal of Applied Physiology, 2009
- Review of physical activity measurement using accelerometers in older adults: Considerations for research design and conductPreventive Medicine, 2009
- Effective techniques in healthy eating and physical activity interventions: A meta-regression.Health Psychology, 2009
- I. Using the Bland–Altman method to measure agreement with repeated measuresBritish Journal of Anaesthesia, 2007
- STATISTICAL METHODS FOR ASSESSING AGREEMENT BETWEEN TWO METHODS OF CLINICAL MEASUREMENTThe Lancet, 1986