The validity and accuracy of wrist-worn activity monitors in lower-limb prosthesis users

Purpose: To investigate the accuracy of consumer-grade wrist-worn activity monitors during over ground walking in persons using lower-limb prosthetics. Method: Thirty-two participants using lower-limb prosthetics (age = 49.7 +/- 14.0 yrs, height = 176.1 +/- 11.6 cm, weight = 87.8 +/- 21.1 kg) were fitted with a Polar Loop, Fitbit Flex, MOVEBAND, Garmin Vivofit, and a Fitbit Charge on the right and left wrists as well as an Omron HJ-113 pedometer on the right and left hip, then walked 140 m at a self-selected pace on an indoor flat surface. Results: There were no significant differences between any of the respective right and left monitors, p > 0.05. When comparing step counts with actual step counts, Polar Loop (p = 0.001), Fitbit Flex (p = 0.001), and MOVEBAND (p = 0.001) were significantly lower than actual step counts. No significant differences existed between the remaining monitors and actual step counts (p > 0.05). Omron incurred the least error (0.6%), followed by Garmin Vivofit (1.3%) and Fitbit Charge (3.6%), with greatest error in the MOVEBAND (21.4%) and Polar Loop (13.1%). Bland-Altman plots suggest Garmin Vivofit to have the least error along with tightest agreement among the wrist-worn activity monitors Conclusion: When considering the use of consumer-grade wrist-worn activity monitors for assessing step counts in persons using lower-limb prostheses, the Garmin Vivofit seems to be the best option followed by Fitbit Charge.