•  
  •  
 

EVALUATING THE ACCURACY AND PRECISION OF COMMERCIALLY AVAILABLE ACTIVITY TRACKERS IN SCHOOL-AGED CHILDREN

Abstract

Alex Atwood1, Hassan-Galaydh Farah2, Michael Teater2, Julia Brisbane2, Robin Queen, FACSM2, Stephanie DeLuca3. 1Virginia Tech Carilion, Roanoke, VA. 2Virginia Tech, Blacksburg, VA. 3Fralin Biomedical Research Institute, Roanoke, VA.

BACKGROUND: Physical activity (PA) has been linked to many health benefits and plays a crucial role in the prevention of obesity (Janssen & LeBlanc, 2010; Strong et al., 2005). Quantifying PA is difficult because it encompasses a wide array of constructs. One of the most common measures of PA in children is self-reported questionnaires. Given the subjectivity of self-reported measures, activity trackers have garnered increased interest to obtain objective data. In addition, their portability allows for measurement across differing environments. However, there is a gap in knowledge regarding the accuracy and precision of activity trackers within pediatric populations. Studies have demonstrated the accuracy and reliability of activity trackers in recording step counts in adults (Paul et al., 2015; Takacs et al., 2014), but those investigating accuracy in pediatric populations are limited (Godino, Job G., et al., 2020). We sought to address this gap by determining the accuracy and precision of two activity trackers (Fitbit ACE 2 & Garmin Vivofit Jr. 2). METHODS: Thirty ambulatory children (8 - 14 years old) were recruited from nearby communities. The study involved participants walking on a treadmill (TM) and completing an obstacle course (OC) while wearing the activity trackers. Steps recorded by activity trackers were compared to actual step count (obtained via motion tracking software) to determine device accuracy and precision. Percent error (PE)in step count was used to determine accuracy, while precision was determined using linear correlations. RESULTS: The average PE for the Fitbit during TM and OC walking were -4.94% and -1.32%, respectively, while the Garmin had -1.32% during TM walking and 3.35% during the OC. Both devices had a strong positive correlation with actual step count during TM walking (Fitbit: r = 0.84, Garmin: r = 0.89). During OC walking, the Fitbit demonstrated a moderately strong positive correlation with actual step count (r = 0.57), while the Garmin Vivofit Jr. 2 demonstrated a negligible correlation with the actual step count (r = 0.20). CONCLUSION: This data suggests that both devices can be used to accurately monitor step counts in school-aged children. It also suggests that both devices demonstrate high precision in step counting when walking at a set pace. Precision, however, appears to be impacted by more diverse environments with reduced precision during OC walking.

This document is currently not available here.

Share

COinS