Abstract

The method used to select representative gas exchange data from large datasets influences the resting metabolic rate (RMR) returned. This study determines which of three methods yields the lowest RMR (as recommended for use in human energy balance studies), and in which method the greatest variance in RMR is explained by classical determinants of this variable. A total of 107 young and 74 middle-aged adults underwent a 30 min RMR examination using a breath-by-breath metabolic cart. Three gas exchange data selection methods were used: (i) steady state (SSt) for 3, 4, 5, or 10 min, (ii) a pre-defined time interval (TI), i.e., 6-10, 11-15, 16-20, 21-25, 26-30, 6-25, or 6-30 min, and (iii) "filtering", setting thresholds depending on the mean RMR value obtained. In both cohorts, the RMRs yielded by the SSt and filtering methods were significantly lower (p < 0.021) than those yielded by the TI method. No differences in RMR were seen under the different conditions of the SSt method, or of the filtering method. No differences were seen between the methods in terms of the variance in RMR explained by its classical determinants. In conclusion, the SSt and filtering methods return the lowest RMRs and intra-measurement coefficients of variation when using breath-by-breath metabolic carts.