Validation of three alternative methods to measure total energy expenditure against the doubly labeled water method for older Japanese men

Abstract

In a previous study using young Japanese men as subjects, Ebine et al. found that accelerometer (AC) represents a promising technique for measuring free-living total energy expenditure (TEE) when compared to activity records (AR) and heart rate monitoring (HR). Thus, the present study was designed to validate the use of an AC and to determine whether or not the previous findings regarding the three alternative field methods (AC, AR, and HR) could be extended to older Japanese men (n = 24: mean ± SD age 48 ± 10 y, body mass index 23.1 ± 2.7 kg/m2 and body fat 18.7 ± 4.8%). TEE values obtained over a 3 d period by AR, HR, and AC (3dAC), and AC over a 14 d period (14dAC) were simultaneously validated against TEE measured by the doubly labeled water (DLW) method applied within a 14 d period. TEE values obtained by AR, HR, 3dAC, and 14dAC ranged from 1,750 to 3,447 kcal/d, 1,691 to 5,286 kcal/d, 1,716 to 2,765 kcal/d, and 1,700 to 2,855 kcal/d, respectively. Expenditures obtained by HR were similar to those obtained using the DLW method, with a mean difference of 57 ± 603 kcal/d (2%), but those obtained using AR, 3dAC, and 14dAC differed substantially from the DLW method, with mean differences of -335 ± 289 kcal/d (12%), -542 ± 249 kcal/d (-19%), and -566 ± 223 kcal/d (-20%), respectively. AR, HR, 3dAC, and 14dAC were significantly correlated with the DLW method, with r values of 0.76 (p<0.0001), 0.67 (p<0.001), 0.78 (p<0,0001), and 0.83 (p<0.0001), respectively. Intra-individual variation indicated by the coefficient of variation (CV) was significantly higher for HR (15 ± 11%, p<0,001) than for AR (7 ± 4%), 3dAC (7 ± 5%), and 14dAC (8 ± 3%). The same findings were obtained using Bland and Altman plots at the population level. Interestingly, 3dAC and 14dAC were significantly correlated with r = 0,97 (p< 0.0001), with a lower mean difference of 24 kcal/d. These results suggest that, same as the previous study, AC is superior to HR in estimating TEE, and seems to be satisfactory for estimation at both group and individual levels, particularly for large-scale studies of older individuals when compared to the DLW method. However, some modifications of the AC method may be needed to compensate for the underestimation of TEE.