Prediction of Non-Resting Energy Expenditure using Accelerometry

The accurate measurement of total energy expenditure is a cornerstone of metabolic research. However, there is a lack of measurement methods that are valid, objective, inexpensive, and easy to use. Accelerometry, along with validated prediction equations for resting energy requirements, may provide an opportunity to fill this void. Twenty weight stable adults (12 female, 8 male) who recently participated in a controlled feeding study comprised the study sample. Total energy requirements were assessed from the controlled feeding period in which weight stability was achieved using the intake-balance method. Resting energy expenditure was assessed using the Mifflin-St. Jeor equation. Participants wore accelerometers to objectively assess habitual physical activity. The accelerometer data obtained along with subjects' demographic and biometric data were used to predict non-resting energy expenditure (NREE) using step-wise linear regression in JMP. Bland-Altman plots and Spearman's Rho correlations were used to determine the validity of the total energy requirements obtained from the sum of the predicted non-resting energy expenditure. Estimated resting energy expenditure was compared with the total energy requirements assessed using the intake-balance method from the controlled feeding period. The resulting prediction equation is as follows: 480.93 – 180.69(sex) + 0.21(Accelerometer kcals) + 617.98(BF%) = AEE. The sex was coded as 1 for females and 0 for males. This prediction model has a coefficient of determination of 0.74 (0.70 adjusted). On average, the model overestimates AEE by 76 kcals. This new model could be the key to accurately, inexpensively and objectively measuring total energy requirements.