Reproducibility of Energy Intake Calculated from Energy Balance Equation in Overweight Adults

Background: Energy intake (Ein) calculated from the energy balance equation has been suggested as a more precise method than self-reported Ein. Therefore, this study aimed to determine the reproducibility of Ein from the energy balance equation comparing two assessment methods of body energy stores (BES) in overweight adults. Methods: In 39 overweight adults (aged 30 ± 5y, BMI 28 ± 1kg/ m2; 51% female), the body composition (dual X-ray absorptiometry [DXA]) and total energy expenditure (TEE; doubly labeled water [DLW]) were measured for two consecutive 2-week periods without diet or exercise prescriptions. Ein was calculated as TEE + BES. The BES was computed either by individual changes in DXA fat mass (FM, 9.5kcal/g) and fat-free mass (FFM, 1.02kcal/g) or using changes in body weight and average body composition (from linear regression of changes in FM/FFM vs. body weight) for males and females. The intra-class correlation coefficients (ICC) and coefficient of variations (CV) were compared between duplicate measurements. Results: Energy intake calculated using changes in BES assessed by individual DXA showed a poor test-retest agreement (ICC = 0.34) and high within-subject variability (CV = 34%). In contrast, the Ein calculated using weight change and average body composition showed excellent test-retest agreement (ICC = 0.89) and low withinsubject variability (CV = 7.9%). There was a better agreement (ICC = 0.94 vs 0.67) and lesser variability (CV = 4.6 vs 11.4%) for males than females using the body weight-based method. Conclusions: In overweight adults, the reproducibility of Ein calculated from the energy balance equation was limited by individual variation in body composition by DXA. The use of body weight calibrated for average DXA changes showed better reproducibility.