Age Group Comparability of Raw Accelerometer Output from Wrist- and Hip-Worn Monitors

TLDR

The study compared raw triaxial accelerometer outputs from ActiGraph GT3X+ and GENEActiv worn on the hip and wrist and developed regression equations to estimate energy expenditure. Thirty children (7–11 yr) and 30 adults (18–65 yr) performed eight activities while wearing both monitors on the hip and wrist, with oxygen consumption measured by indirect calorimetry and data analyzed via ANOVA, intraclass correlation, and linear regression to establish intensity thresholds. Wrist placement produced higher acceleration values than hip, with no brand effect in adults or children but strong agreement (ICC 0.96–0.99); however, children showed a systematic brand–placement interaction, and both brands correlated well with VO₂, while intensity classification accuracy was higher in adults and for sedentary/light and vigorous activities than moderate.

Abstract

The study aims were to compare raw triaxial accelerometer output from ActiGraph GT3X+ (AG) and GENEActiv (GA) placed on the hip and the wrist and to develop regression equations for estimating energy expenditure.Thirty children (7-11 yr) and 30 adults (18-65 yr) completed eight activities (ranging from lying to running) while wearing one AG and one GA on the hip and the wrist. Oxygen consumption (V˙O2) was measured with indirect calorimetry. Analysis involved the use of ANOVA to examine the effect of activity, brand, and placement on the acceleration values, intraclass correlation coefficient to evaluate the agreement between the two brands and placements, and linear regression to establish intensity thresholds.A significant difference in acceleration values between the hip and the wrist placement was found (P < 0.001). The output from the wrist placement was, in general, higher compared with that from the hip. There was no main effect of monitor brand in adults (P < 0.12) and children (P < 0.73), and the intraclass correlation coefficient showed a strong agreement (0.96-0.99). However, a three-way interaction and systematic error between the brands was found in children. Acceleration from both brands and placements showed a strong correlation with V˙O2. The intensity classification accuracy of the developed thresholds for both brands and placements was, in general, higher for adults compared with that for children and was greater for sedentary/light (93%-97%), and vigorous activities (68%-92%) than that for moderate activities (33%-59%).Accelerometer outputs from AG and GA seem comparable when attached to the same body location in adults, whereas inconsistent differences are apparent between the two brands and placements in children, hence limiting the comparability between brands in this age group.