Validity of a multisensor armband in estimating 24-h energy expenditure in children

Physical activity (PA) and energy expenditure (EE) in children are frequently assessed using portable activity monitors. Algorithms used to predict EE by these monitors are often based on adult populations and may not be accurate for children. Purpose: To evaluate the accuracy of the SenseWear Pro Armband (SWA) for assessing EE in African American children during treadmill exercise, sedentary activities, rest, sleep, and total 24-h EE, using indirect room calorimetry (IRC) as a reference standard. Methods: Participants were healthy African American children (10 boys, 11 girls; age: 11.6 +/- 0.9 yr; weight: 47.3 +/- 13.0 kg; height: 151.6 +/- 8.8 cm; BMI: 20.4 +/- 4.8 kg.m(-2)). EE was measured simultaneously using IRC and SWA during a 24-h stay in the IRC. Recorded activities included sedentary behaviors, treadmill exercise, rest periods, and sleep. Results from both methods were matched minute-by-minute and compared by Bland-Altman plot. Multiple linear regression analysis was used to describe the relationship between EE assessed by both methods and children's descriptive characteristics. Results: SWA overestimated EE compared with IRC during all activities and time periods, ranging from 116% during sleep to 143 % during rest after treadmill exercise. The SWA-predicted EE was improved by using linear regression modeling. Simple equations for sedentary activities and treadmill exercise were EE [kcal.min(-1)] = 0.462EE (SWA) [kcal.min(-1)] + 0.015 x body weight [kg], and EE [kcal.min(-1)] = 0.637EE (SWA) [kcal.min(-1)]+ 0.034 x body weight [kg], respectively. The prediction equation for RMR was RMR [kcal.min(-1)] = 0.453EE (SWA) [kcal.min(-1)] + 0.011 x body weight [kg]. Conclusion: EE estimated using SWA was significantly higher than EE measured using IRC in African American children ages 10-14 yr. Bias in individual EE estimated using SWA could be improved by an adjustment for the body weight of a child. The SWA manufacturer should work with researchers on improving existing algorithms for children.