Agreement Between Fitbit And Actigraph Estimates Of Physical Activity In Young Children

Abstract

Commercial wearables are used to track physical activity (PA) levels in children and as tools for increasing PA participation in youth. However, few studies have examined the agreement between commercial and research grade devices in assessing PA in children. PURPOSE: To compare estimates from a wrist-worn Fitbit Flex 2 to a waist-worn Actigraph GT9X Link in elementary school children. METHODS: Forty children aged 6-10 years wore a Fitbit Flex 2 (on non-dominant wrist) and an Actigraph GT9X Link (on waist) for up to two weeks while school was in session. Children were instructed to wear the Fitbit continuously and to wear the Actigraph during all waking hours. Parents were given wear reminders via text each morning. For each device, an adapted Choi algorithm was used to flag non-wear periods (defined as ≥90 minutes of continuous 0 step values). Steps and intensity-specific estimates of daily PA from each device were averaged across time periods when both devices were worn simultaneously, and estimates were compared using days with 10+ h of concordant wear and using all available data. RESULTS: Across 91 days with concordant device wear of 10+ h, Fitbit estimates were 25% higher for steps (Mean±SD = 10318±3846 steps) and 44% higher for vigorous intensity PA (16.2±22.1 min) compared to the Actigraph GT9X Link (8260+3614 steps; 11.3±15.9 min). In contrast, Fitbit estimates of moderate intensity PA (26.8±21.1 min) were 20% lower than Actigraph estimates (33.3±24.3 min). Pearson correlations between device estimates were higher for steps (r=0.62) than for moderate (r=0.55) or vigorous (r=0.28) intensity PA. Similar patterns were observed when all available concordant data were analyzed (n=377 days), with the Fitbit recording 20% more steps, 28% less moderate PA, and 36% more vigorous PA than the Actigraph. CONCLUSIONS: Wrist-worn consumer wearables may produce higher estimates of steps and vigorous intensity PA, and lower estimates of moderate intensity PA, in elementary school children. Absent additional evidence, consumers and researchers should be cautious when using wrist-worn consumer devices to assess absolute levels of PA in youth. Supported by a grant from NHLBI (R01HL135359).