Validity of Consumer-Based Physical Activity Monitors Under Free-Living

Abstract

Accelerometry-based physical activity monitors are commonly used in research but recently many consumer-based monitors have emerged in the market. The devices are marketed to provide indicators of activity and energy expenditure (EE) but no information is available to substantiate their validity under free-living conditions. PURPOSE: The present study examines the validity of EE estimates from a variety of consumer-based physical activity monitors under free-living conditions. METHODS: Fifty one (25.2 ± 4.8 years) healthy males (n=26) and females (n=25) wore a total of ten monitors simultaneously while completing a 71 minute protocol, involving a diverse array of both free-living and structured activities including (but not limited to) stationary biking, walking up stairs, resting, carrying a backpack, and riding in a car. The monitors included the BodyMedia FIT armband worn on the right arm (R) and left arm (L), DirectLife monitors around the neck (N) and on the belt (B), Fitbit, Gruve, Omron pedometer, and, KAM (Kinetic Activity Monitor) worn on the belt, as well as a Garmin GPS monitor and Polar heart rate monitor. Estimates of EE from each monitor were compared with criterion values from a portable metabolic system (Oxycon Mobile, IC) using standard measurement agreement methods. RESULTS: The correlation coefficients (r) between indirect calorimetry and consumer activity monitors were.81,.78,.46,.47,.70,.65,.35, 32,.41and.66 for BodyMedia FIT(R), BodyMedia FIT(L), DirectLife(N), and DirectLife(B), Fitbit, Gruve, Omron, KAM, Polar, and Garmin (respectively). The corresponding absolute error rates (computed as average absolute value of the individuals' errors) were 13 ± 7%, 13 ± 8%, 27 ± 10%, 19 ± 18%, 14 ± 8%, 48 ± 9%, 45 ± 14%, 45 ± 14%, 27 ± 20%, and 45 ± 9%, respectively. ANOVA and Post hoc analyses with Tukey procedures revealed non-significant differences (p>.05) between the Oxycon Mobile and BodyMedia FIT(R) and BodyMedia FIT(L) in both males and females. The Fitbit yielded non-significant differences for males but not females; while all other devices significantly underestimated the observed EE. CONCLUSION: The Bodymedia FIT armband monitor provided the most accurate estimate of EE of the 8 consumer-based physical activity monitor tested.