RESUMO
PURPOSE: Measured resting metabolic rate (RMR) was compared with predicted RMR in a sample of young, non-obese women. METHODS: In 52 women aged 19 to 30 with a body mass index of 16 to 29 kg/m2, RMR was measured with a MedGem indirect calorimeter and predicted with five commonly used equations: the Harris-Benedict (1919), Mifflin (1989), Owen (1985), Schofield (weight) (1985), and Schofield (weight and height) (1985) equations. Measured RMR and predicted RMR were compared through the use of various measures. RESULTS: In comparison with the measured RMR, the RMR predicted with four of the five equations was significantly higher (by 16 to 225 kcal/day, p < 0.001). At the group level, the Owen equation performed best and captured the greatest proportion of individuals (65%) for whom predicted RMR differed from measured RMR by less than 10%. With the other four equations, residuals exceeded 10% for more than two-thirds of participants. For the Harris-Benedict, Mifflin, and Owen equations, every 100 kcal/day increase in measured RMR was associated with a 6% to 8% decrease in error. The optimal prediction range (within 10% of the measured RMR) was different for each: Owen equation 1105 to 1400 kcal/day, Mifflin equation 1280 to 1595 kcal/day, and Harris-Benedict equation 1345 to 1630 kcal/day. CONCLUSIONS: Prediction equations should be modified according to the amount of corresponding percentage error. Where possible, RMR should be measured. Barring this, the Owen equation should be used for young, non-obese women.