Validity of predictive equations for resting energy expenditure according to the body mass index in a population of 1726 patients followed in a Nutrition Unit.

International audienceThe resting energy expenditure (REE) predictive formulas are often used in clinical practice to adapt the nutritional intake of patients or to compare to REE measured by indirect calorimetry. We aimed to evaluate which predictive equations was the best alternative to REE measurements according to the BMI. 28 REE prediction equations were studied in a population of 1726 patients without acute or chronic high-grade inflammatory diseases followed in a Nutrition Unit for malnutrition, eating disorders or obesity. REE was measured by indirect calorimetry for 30 min after a fasting period of 12 h. Some formulas requiring fat mass and free-fat mass, body composition was measured by bioelectrical impedance analysis. The percentage of accurate prediction (±10%/REE measured) and Pearson r correlations were calculated. Original Harris & Benedict equation provided 73.0% of accurate predictions in normal BMI group but only 39.3% and 62.4% in patients with BMI 40 kg m(-2), respectively. Usual predictive equations of REE are not suitable for predicting REE in patients with extreme BMI, in particularly in patients with BMI