Step count accuracy and reliability of two activity tracking devices in people after stroke

Background/Purpose: The increasing popularity of activity tracking devices presents an opportunity to monitor physical activity in patients after stroke. We sought to determine the reliability and accuracy of the Garmin Vivofit and Fitbit Zip for adults after stroke. Methods: Twenty-four participants with stroke-induced hemiparesis wore a Fitbit Zip on the nonparetic hip and Garrnin Vivoflits on both wrists during a 6-minute walk test to determine the accuracy of the devices against video-determined step counts. Participants also wore the devices during two trials of exactly 50 steps to determine the reliability of the devices. Results: Fitbit Zip showed excellent reliability (ICC2,1 = 0.974) and accuracy (4.2% error) for participants who walked faster than 0.35 m/s. Garmin Vivofit (nonparetic side) had excellent reliability (lCC(2,1) = 0.964) but poor accuracy (<=-16.0%) for all participants. Garmin Vivofit (paretic side) had excellent reliability (ICC2,1 = 0.858) and accuracy (-40% error) for faster walkers (>0.48 m/s) but poor accuracy (-68.2%) for slower walkers. Conclusion: Fitbit Zip was more accurate and reliable for persons with stroke than Garmin Vivofit, but slower walking speeds were associated with greater undercounting of steps for both devices. The Fitbit Zip is appropriate for counting steps in adults poststroke who range from household to community ambulators.