Measuring free-living energy expenditure and physical activity with triaxial accelerometry

OBJECTIVE: To investigate the ability of a newly developed triaxial accelerometer to predict total energy expenditure (EE) (TEE) and activity-related EE (AEE) in free-living conditions. RESEARCH METHODS AND PROCEDURES: Subjects were 29 healthy subjects between the ages of 18 and 40. The Triaxial Accelerometer for Movement Registration (Tracmor) was worn for 15 consecutive days. Tracmor output was defined as activity counts per day (ACD) for the sum of all three axes or each axis separately (ACD-X, ACD-Y, ACD-Z). TEE was measured with the doubly labeled water technique. Sleeping metabolic rate (SMR) was measured during an overnight stay in a respiration chamber. The physical activity level was calculated as TEE x SMR(-1), and AEE was calculated as [(0.9 x TEE) - SMR]. Body composition was calculated from body weight, body volume, and total body water using Siri's three-compartment model. RESULTS: Age, height, body mass, and ACD explained 83% of the variation in TEE [standard error of estimate (SEE) = 1.00 MJ/d] and 81% of the variation in AEE (SEE = 0.70 MJ/d). The partial correlations for ACD were 0.73 (p < 0.001) and 0.79 (p < 0.001) with TEE and AEE, respectively. When data on SMR or body composition were used with ACD, the explained variation in TEE was 90% (SEE = 0.74 and 0.77 MJ/d, respectively). The increase in the explained variation using three axes instead of one axis (vertical) was 5% (p < 0.05). DISCUSSION: The correlations between Tracmor output and EE measures are the highest reported so far. To measure daily life activities, the use of triaxial accelerometry seems beneficial to uniaxial