Validity of a multi-sensor armband in estimating rest and exercise energy expenditure.

Abstract

The SenseWear Armband (SWA; BodyMedia, Inc.), using multiple sensors, was designed to estimate energy expenditure (EE) in free-living individuals. To examine the reliability and validity of the SWA during rest and exercise compared with indirect calorimetry (IC). EE was assessed with SWA and IC in 13 males during two resting and one cycle ergometry (40 min at 60% VO2peak) sessions. In a second experiment, 20 adults walked on a treadmill for 30 min at three intensities (80.5 m x min, 0% grade; 107.3 m x min, 0% grade; 107.3 m x min, 5% grade) while IC and SWA measured EE. At rest, no significant differences were found between EE measurements from the SWA (1.3 +/- 0.1 kcal x min) and IC (1.3 +/- 0.1 kcal x min), and the two methods were highly correlated (r = 0.76; P < 0.004). The SWA EE estimation was reliable when comparing the two resting visits (r = 0.93; P < 0.001). For the ergometer protocol, no significant differences were found between the SWA and IC measurements of EE early, mid, or late in exercise or for the total bout, although the measurements were poorly correlated (r = 0.03-0.12). The SWA EE estimate of walking increased with treadmill speed but not with incline. The SWA significantly overestimated (13-27%) the EE of walking with no grade (P < 0.02) and significantly underestimated (22%) EE on the 5% grade (P < 0.002). The SWA estimation of EE correlated moderately with IC (r = 0.47-0.69). The SWA provided valid and reliable estimates of EE at rest and generated similar mean estimates of EE as IC on the ergometer; however, individual error was large. The SWA overestimated the EE of flat walking and underestimated inclined walking EE.