Comparison of various predictive formulae for the estimation of resting energy expenditure

The accuracy of the Harris-Benedict (HB), Kleiber and Robertson-Reid predictive formulae and the Fleisch tables in estimating resting energy expenditure (REE) has been assessed in 114 cancer patients and 54 patients with nonmalignant illness. The effects of weight status and disease on the predictive ability of the formulae were assessed by comparison of the estimated REE value with that measured by indirect calorimetry. Underestimation of measured REE by greater than 10% occurred in 6-62% of patients, depending on the formula used. Weight loss did not affect predictive ability, whereas the presence of cancer resulted in inaccurate prediction of REE in female patients. Using a multiple comparison procedure, the Fleisch tables were the most accurate in all groups, while the HB formula was least accurate in male patients, irrespective of weight status or disease status. The use of predictive formulae and tables are inappropriate for the accurate estimation of REE in individual patients.