Secondary Logo

Accuracy of Fitbit Activity Trackers During Walking in a Controlled Setting: 1344 Board #19 June 1 900 AM - 1030 AM

Gamez, Jose L.1; Gonzalez, Jesus1; Leyva, Perla1; Figueroa, Ivan A.1; Lucio, Naomi1; Salazar, Vanessa E.1; Salazar, Cindy2; Garcia, Miriam2; Funk, Merrill D.1

Medicine & Science in Sports & Exercise: May 2017 - Volume 49 - Issue 5S - p 363
doi: 10.1249/01.mss.0000517874.07408.5a
C-31 Free Communication/Poster - Activity Trackers and Smartwatches Thursday, June 1, 2017, 7:30 AM - 12:30 PM Room: Hall F

1University of Texas Rio Grande Valley, Brownsville, TX. 2University of Texas School of Public Health Health Science Center at Houston Brownsville Regional Campus, Brownsville, TX.

(No relationships reported)

BACKGROUND: Activity trackers are widely used to measure daily physical activity. Many devices have been shown to measure steps more accurately at higher intensities, however, it is also important to determine accuracy at a walking pace.

PURPOSE: To assess 6 popular activity trackers at measuring steps while walking on a treadmill.

METHODS: Twenty-six college students (Mean±SD; 22.1±3.7yrs; 25.1±4.0kg/m2; 13 male) walked 500 steps at 3mph on a treadmill while wearing 6 different activity trackers (Pedometer, Fitbit Blaze, Charge HR, Alta, Flex, Zip, One). The Charge HR was placed two fingers above the right wrist while the Flex was next to the wrist bone. The Blaze was placed two fingers above the left wrist while the Alta was next to the wrist bone. The Fitbit Zip and the One were aligned with the hip bone on the left and right waistband respectively. A trained researcher using a hand tally counter counted the steps. Missing values were replaced with the mean value for that device. Step counts were correlated between Fitbit devices and the pedometer and tally counter using Pearson correlations. Significance was set at p<0.05. Mean bias scores were calculated between the step counts for each device and the tally counter. Mean Absolute Percent Error (MAPE) values were also calculated for each device relative to the tally counter.

RESULTS: Fitbit Zip and One were significantly correlated with the tally counter (r=0.50, p<0.05; r=0.68, p<0.01, respectively) while the other devices were not significantly correlated. Mean bias and MAPE values were as follows:



Fitbit Zip and One were within one half percent of actual steps while wrist-worn Fitbits ranged from 7.0-19.7% from actual step counts.

CONCLUSION: Consistent with previous research, activity trackers worn at the waist provide the most accurate step counts compared to wrist-worn models. Differences found in wrist-worn models may result in significant over- or underestimation of activity levels when worn for long periods of time.

© 2017 American College of Sports Medicine