Mathematical ratios lead to spurious conclusions regarding age- and sex-related differences in resting metabolic rate

Resting metabolic rate (RMR) data have been normalized by dividing RMR by fat-free mass (FFM) (ie, ratio method), or by using a regression-based approach. We compared both data-normalization procedures on age- and sex-related differences in RMR. The ratio method showed no differences in adjusted RMR between older men (0.084 +/- 0.004 kJ.FFM-1.min-1) and younger men (0.082 +/- 0.003 kJ.FFM-1.min-1), whereas analysis of covariance showed a lower (P < 0.01) adjusted RMR in older men (4.81 +/- 0.04 kJ/min) than in younger men (5.14 +/- 0.04 kJ/min). In another example, the ratio method showed that women had a higher (P < 0.05) adjusted RMR (0.10 +/- 0.004 kJ/min) than did men (0.08 +/- 0.003 kJ/min), whereas analysis of covariance showed a lower (P < 0.01) adjusted RMR in women (4.45 +/- 0.03 kJ/min) than in men (4.62 +/- 0.03 kJ/min). The ratio method provides misleading conclusions regarding sex- and age-related differences in RMR when compared with a regression-based approach.