24-h Energy Expenditure Estimation From Accelerometry Assessments: A Cross-validation At Low Levels Of Physical Activity

Estimation of 24-h energy expenditure (24-hEE) is challenging due to the difficulty of estimating its most variable component- physical activity EE (PAEE). PURPOSE: We utilized whole room indirect calorimeter (WRIC) technology to cross-validate estimations of 24-hEE from models that utilize the sum of PAEE from accelerometry, resting metabolic rate (RMR), and thermic effect of food (TEF). METHODS: Sixty-two assessments from 46 participants (men n = 24 and women n = 22; age = 29.5 ± 8.0 yrs., BMI = 26.1 ± 4.2 kg/m2) were included in this analysis. 24-h EE and its components (RMR, TEF and PAEE) were measured in WRIC under a tightly scripted activity protocol. Participants wore an accelerometer around their dominant wrist to measure PA, which was quantified with Euclidian Norm Minus One (ENMO) units. We utilized 2 previously validated models to estimate 24-hEE: the first was calculated by predicted RMR from anthropometric variables (weight, height, age and sex), TEF was assumed to be 10% of 24-hEE and PAEE was predicted from a linear regression model with ENMO units as an independent variable (24-h EE-L); in the second PAEE was obtained from a quadratic model (24-hEE-Q). Physical activity level was confirm to be low (PAL = 1.30 ± 0.11). Agreement between measured and predicted 24-EE was analyzed by concordance correlation coefficient (ρc) and Bland and Altman plots. RESULTS: Both measured 24-h EE and PAEE were well correlated with estimated values (24-hEE-L r = 0.903, 24-hEE-Q r = 0.864, PAEE-L r = 0.551 and PAEE-Q r = 0.497; for all P < 0.001). Linear and quadratic models had moderate concordance with measured values of 24-hEE (ρc = 0.726 and ρc = 0.621 respectively). A systematic bias for 24-EE was observed in both models, which tended to overestimate 24-hEE (linear = 329 kcal/day and quadratic = 353 kcal/day); this discrepancy is partially explained by the large difference between estimated and measured PAEE (287 kcal/day on average). Nevertheless, these differences were strongly driven by participants with the largest 24-hEE. CONCLUSIONS: Our results suggest that the new models provide moderate validity to estimate 24-EE at least for people with low levels of physical activity. ENMO units provide a promising platform to generate improved regression models to estimate 24-hEE in free-living conditions. National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under Award Number R01DK105829 and 1DP3DK113358-01