To compare measured resting energy expenditure (REE) with that predicted by formulas derived from populations with normal weight or obesity and from women with severe and morbid obesity.
66 women (aged 35.6 +/- 10.3 y and BMI of 44.7 +/- 4.9 kg/m2) were evaluated by indirect calorimetry with a metabolic monitor Deltatrac (Datex Inst., Finland), before undergoing gastric bypass. REE was calculated with the following equations: Harris-Benedict's with both actual and adjusted weight, Ireton-Jones', Mifflin's, and Carrasco's Fast Estimation, which corresponds to 16.2 kcal x kg actual weight.
(mean +/- sd). Measured REE was 1797 +/- 239 kcal/day. All formulas, except Harris-Benedict's with adjusted weight, overestimated REE. The Ireton-Jones' equation presented the greater overestimation (689 +/- 329 kcal/day), whereas Mifflin's equation overestimated REE only by 6 +/- 202 kcal/day. No significant differences were detected between measured and calculated REE by Mifflin's and Carrasco's Fast Estimation. Accuracy (defined as difference between calculated and measured REE within +/- 10%) was greater with Mifflin's equation (68%), followed by Harris-Benedict's with actual weight (64%) and Carrasco's Fast Estimation (61%). By using the Bland-Altman analysis, significant correlations were observed between calculated-measured REE and mean REE (calculated + measured/2) with all equations except Carrasco's Fast Estimation. This means that all but one formula underestimate or overestimate REE depending on the level of measured REE.
In severe and morbid obese women, Mifflin's and Carrasco's Fast Estimation equations provided the best performance to estimate REE. Before recommending an equation in an a subset of individuals it is necessary to make previous validation studies to determine that equation with the best predictive power for this particular group of patients.