Assessment Of Daily Physical Energy Expenditure By Accelerometer Body Placement Using Doubly Labeled Water: Does Placement Matter?

Abstract

PURPOSE: To assess the contribution of accelerometer-measured physical activity by body placement (left wrist, right wrist, or trunk) to daily activity energy expenditure using doubly labeled water. METHODS: This study enrolled 49 adults from the Baltimore Longitudinal Study of Aging that wore ActiGraph accelerometers on both wrists and the chest while their energy expenditure was measured using doubly labeled water (DLW) over 7 consecutive days. Total activity counts (TAC) from the 3 accelerometers were calculated (scaled to 1000 counts/day) and served as the daily physical activity measure. On the day participants began to wear the accelerometers, they received doses of approximately 2.5 g/kg of 10% enriched H218O and 0.12 g/kg of 99.9% enriched 2H2O. Urine samples were collected before and after DLW dosing and after 7 days of physical activity monitoring to determine mean activity energy expenditure (AEE). Sample characteristics were summarized using medians (IQR) and frequencies. Linear regression models with TAC alone and with TAC, age, sex, body composition and height were used to predict AEE for each placement location. R-square (R2) increases due to TAC were used to evaluate the influence of accelerometer placement with respect to variance explained in AEE. RESULTS: Participants (55% women, median age 74 years) recorded a median of 2396 (IQR: 2099, 2679) activity counts/day on the dominant wrist, 2069 (1815, 2322) counts/day on the non-dominant wrist, 37 (25, 43) counts/day on the chest and generated AEE of 725 (538, 979) kCal/day. In the TAC only models, TAC was not associated with AEE and explained minimal variance (R2 = .03-.04; all p > .05). In the multi-predictor models TAC measured through dominant wrist worn and non-dominant wrist worn were associated with AEE (β = 0.19, SE = 0.09; p = .033 and β = 0.22, SE = 0.10; p = .039), but chest worn was not (p = .25). The addition of TAC measured through dominant and non-dominant wrist to the multi-predictor models explained 7% and 8% more variance predicting AEE, respectively (p < .05). CONCLUSIONS: Wrist-worn accelerometers are more closely associated with AEE than chest-worn. Accelerometer placement and participant characteristics are important considerations when estimating daily activity energy expenditure.