Total energy expenditure of 10- to 12-year-old Japanese children measured using the doubly labeled water method

Background To establish Japanese children's estimated energy requirements, total energy expenditure (TEE) data measured using the doubly labeled water (DLW) method is needed. This study aimed to 1) obtain basic TEE data from Japanese children measured using DLW (TEEDLW), 2) compare TEEDLW with TEE estimated by various estimation formulas to calculate their accuracy, and 3) develop a new equation to estimate TEE using body composition and pedometers. Methods TEE was measured using DLW in 56 10- to 12-year-old Japanese children (33 boys, 23 girls). Physical activity level (PAL) was calculated by dividing TEEDLW by estimated resting energy expenditure. To assess their physical activity, participants wore pedometers during the 7-d DLW period. Total body water was calculated from 2H and 18O; fat-free mass (FFM) and fat mass (FM) were then determined. Results In boys and girls of normal weight, TEEDLW was 2067 ± 230 kcal/d and 1830 ± 262 kcal/d, respectively. Average PAL was 1.58 ± 0.17. FFM was strongly related to TEE (r = 0.702, p < 0.01). After adjusting for FFM and FM, step count was significantly associated with TEE (r = 0.707, p < 0.01). The TEE estimation formula used in the Dietary Reference Intakes (DRI) for the United States and Canada estimated TEEDLW with high accuracy (bias: 2.0%) in both sexes. We developed new equations for TEE consisting of FFM and step count, which accounted for 68% and 65% of TEE variance in boys and girls, respectively: boys, 47.1 × FFM (kg) + 0.0568 × step count (steps/d) - 122, and girls, 55.5 × FFM (kg) + 0.0315 × step count (steps/d) - 117. Conclusions The TEE in 10- to 12-year-old Japanese children measured using DLW was approximately 7% lower for boys and 12% lower for girls compared to the current Japanese DRI. If PAL can be accurately determined, the equation in the DRI for the United States and Canada may be applicable to Japanese children. In addition, TEE could be predicted using FFM and step count.