The Effect of BMI and Walking Speed on Step Count Validity of Wearable Activity Monitors

Abstract

Activity monitors have been widely used in health promotion and weight loss programs to promote increases in physical activity. However, the accuracy of these devices to measure walking behavior in the overweight/obese population has not been clearly established. PURPOSE: To determine if body mass index (BMI) and walking speed affect the step output of commonly used activity monitors while on the treadmill. METHODS: Participants were stratified into BMI categories, normal weight (n=25), overweight (n=15) and obese (n=10). They were asked to walk on a treadmill at three different speeds (40, 67, and 90 m/min) while wearing a multidirectional accelerometer (ActiCalTM) and a piezo-electric pedometer (NL-2000) on the right hip, a uniaxial (ActiGraphTM GT1M) accelerometer on the left hip and a step activity monitor (StepWatchTM) over the lateral malleolus of the left ankle. For the criterion measure, a trained investigator counted steps by hand. The percent of actual steps were used for analysis. RESULTS: Repeated measures ANOVA revealed marginal device x BMI interaction (p=0.053) at the slowest speed (40 m/min), with the StepWatch and NL-2000 recording most actual steps regardless of BMI category (99% and 75%, respectively), while the Actical and ActiGraph recorded 57% and 43% of steps. The accuracy of the Actical and ActiGraph tended to be higher in the obese group (65% and 53%) than the overweight (58% and 48%) and normal weight groups (53% and 37%). At the faster speeds, all devices recorded between 96% and 99% of actual steps. CONCLUSIONS: These preliminary data suggest wearable activity monitors are not affected by BMI at faster walking speeds (>/= 67 m/min). However, at slower speeds this can not be concluded with certainty.