Accuracy of Wristband Activity Monitors during Ambulation and Activities

Purpose The main purpose of this study was to examine the accuracy of wristband activity monitors on measuring step counts at prescribed speeds on a treadmill and under short bouts of common daily activities. Methods Thirty healthy young adults wore three wristband activity monitors on both wrists while walking or jogging on a treadmill at different speeds (54, 80, 107, and 134 m·min^-1) and performing six different common daily activities for 5 min each. The monitors included the Fitbit Flex, the Garmin Vivofit, and the Jawbone UP. The common daily activity conditions included two sitting activities (playing a tablet computer game and folding laundry), two walking activities (pushing a stroller, carrying a bag), and two stair climbing activities (down and up). Absolute percentage error (APE) scores were computed to examine the accuracy between actual observed steps and monitor-detected steps. Results Under the treadmill condition, the APE ranged between 1.5% and 9.6%. Accuracy was improved at faster speeds (134 m·min^-1) for all the monitors (APE < 2.5%). In the common daily activity conditions, substantial step counts were registered when folding laundry. All monitors significantly underestimated actual steps (all APE >33%) when pushing a stroller. Higher APE was observed when worn on the dominant wrist under the common daily activity conditions. Conclusions The wristband activity monitors examined were more accurate for measuring step counts between 80 and 134 m·min^-1 as compared with a slower speed. Accuracy under each common daily activity condition ranged widely between monitors and activity, with less error when worn on the nondominant wrist. These results will help to inform researchers on the use and accuracy of wristband activity monitors for future studies.