OBJECTIVE: To evaluate the accuracy of seven predictive equations, including the Harris-Benedict and the Mifflin equations, against measured resting energy expenditure (REE) in hospitalized patients, including patients with obesity and critical illness. DESIGN: A retrospective evaluation using the nutrition support service database of a patient cohort from a similar timeframe as those used to develop the Mifflin equations. SUBJECTS/ SETTING: All patients with an ordered nutrition assessment who underwent indirect calorimetry at our institution over a 1-year period were included. INTERVENTION: Available data was applied to REE predictive equations, and results were compared to REE measurements. MAIN OUTCOME MEASURES: Accuracy was defined as predictions within 90% to 110% of the measured REE. Differences >10% or 250 kcal from REE were considered clinically unacceptable. STATISTICAL ANALYSES PERFORMED: Regression analysis was performed to identify variables that may predict accuracy. Limits-of-agreement analysis was carried out to describe the level of bias for each equation. RESULTS: A total of 395 patients, mostly white (61%) and African American (36%), were included in this analysis. Mean age+/-standard deviation was 56+/-18 years (range 16 to 92 years) in this group, and mean body mass index was 24+/-5.6 (range 13 to 53). Measured REE was 1,617+/-355 kcal/day for the entire group, 1,790+/-397 kcal/day in the obese group (n=51), and 1,730+/-402 kcal/day in the critically ill group (n=141). The most accurate prediction was the Harris-Benedict equation when a factor of 1.1 was multiplied to the equation (Harris-Benedict 1.1), but only in 61% of all the patients, with significant under- and over-predictions. In the patients with obesity, the Harris-Benedict equation using actual weight was most accurate, but only in 62% of patients; and in the critically ill patients the Harris-Benedict 1.1 was most accurate, but only in 55% of patients. The bias was also lowest with Harris-Benedict 1.1 (mean error -9 kcal/day, range +403 to -421 kcal/day); but errors across all equations were clinically unacceptable. CONCLUSIONS: No equation accurately predicted REE in most hospitalized patients. Without a reliable predictive equation, only indirect calorimetry will provide accurate assessment of energy needs. Although indirect calorimetry is considered the standard for assessing REE in hospitalized patients, several predictive equations are commonly used in practice. Their accuracy in hospitalized patients has been questioned. This study evaluated several of these equations, and found that even the most accurate equation (the Harris-Benedict 1.1) was inaccurate in 39% of patients and had an unacceptably high error. Without knowing which patient's REE is being accurately predicted, indirect calorimetry may still be necessary in difficult to manage hospitalized patients.
OBJECTIVE: To evaluate the accuracy of seven predictive equations, including the Harris-Benedict and the Mifflin equations, against measured resting energy expenditure (REE) in hospitalized patients, including patients with obesity and critical illness. DESIGN: A retrospective evaluation using the nutrition support service database of a patient cohort from a similar timeframe as those used to develop the Mifflin equations. SUBJECTS/ SETTING: All patients with an ordered nutrition assessment who underwent indirect calorimetry at our institution over a 1-year period were included. INTERVENTION: Available data was applied to REE predictive equations, and results were compared to REE measurements. MAIN OUTCOME MEASURES: Accuracy was defined as predictions within 90% to 110% of the measured REE. Differences >10% or 250 kcal from REE were considered clinically unacceptable. STATISTICAL ANALYSES PERFORMED: Regression analysis was performed to identify variables that may predict accuracy. Limits-of-agreement analysis was carried out to describe the level of bias for each equation. RESULTS: A total of 395 patients, mostly white (61%) and African American (36%), were included in this analysis. Mean age+/-standard deviation was 56+/-18 years (range 16 to 92 years) in this group, and mean body mass index was 24+/-5.6 (range 13 to 53). Measured REE was 1,617+/-355 kcal/day for the entire group, 1,790+/-397 kcal/day in the obese group (n=51), and 1,730+/-402 kcal/day in the critically ill group (n=141). The most accurate prediction was the Harris-Benedict equation when a factor of 1.1 was multiplied to the equation (Harris-Benedict 1.1), but only in 61% of all the patients, with significant under- and over-predictions. In the patients with obesity, the Harris-Benedict equation using actual weight was most accurate, but only in 62% of patients; and in the critically ill patients the Harris-Benedict 1.1 was most accurate, but only in 55% of patients. The bias was also lowest with Harris-Benedict 1.1 (mean error -9 kcal/day, range +403 to -421 kcal/day); but errors across all equations were clinically unacceptable. CONCLUSIONS: No equation accurately predicted REE in most hospitalized patients. Without a reliable predictive equation, only indirect calorimetry will provide accurate assessment of energy needs. Although indirect calorimetry is considered the standard for assessing REE in hospitalized patients, several predictive equations are commonly used in practice. Their accuracy in hospitalized patients has been questioned. This study evaluated several of these equations, and found that even the most accurate equation (the Harris-Benedict 1.1) was inaccurate in 39% of patients and had an unacceptably high error. Without knowing which patient's REE is being accurately predicted, indirect calorimetry may still be necessary in difficult to manage hospitalized patients.
Authors: Natalie S Hauser; Irini Manoli; Jennifer C Graf; Jennifer Sloan; Charles P Venditti Journal: Am J Clin Nutr Date: 2010-11-03 Impact factor: 7.045
Authors: R Wirth; C Smoliner; C Spamer; C Marburger; F S Schreiber; H P Willschrei; R Lenzen-Großimlinghaus; R Schäfer; D Volkert Journal: Eur J Clin Nutr Date: 2014-05-21 Impact factor: 4.016
Authors: Rebecca B Little; Robert A Oster; Betty E Darnell; Wendy Demark-Wahnefried; L Burt Nabors Journal: Nutr Cancer Date: 2016-06-24 Impact factor: 2.900