Accuracy of predictive equations for the measurement of resting energy expenditure in older subjects

Background and aims: The measurement of resting energy expenditure (REE) is important to assess individual total energy requirements in older subjects. The validity of REE prediction equations in this population has not been thoroughly evaluated and therefore the main aim of this analysis was to assess the accuracy of REE prediction equations in older subjects. Methods: Weight, height and body mass index (BMI) were measured. REE was measured by indirect calorimetry (IC) in 68 older subjects (age: 60-94 years, M/F: 13/55, BMI: 26.3±5.0kg/m2). Measured REE was compared to 14 equations for the calculation of REE estimates. In addition, two novel approaches (Aggregate model and meta-regression equations) for the prediction of REE were evaluated. Paired t test and Bland-Altman method were used to assess the agreement of the equations. Results: The average measured REE was 1298±264kcal/day. The equation with the smallest bias was proposed by Muller (Bias±2SD=+3±294kcal/day) whereas the Mifflin equation was associated with the largest error (Bias±2SD=-172±282kcal/day). The Aggregate, Muller, Harris-Benedict and Fredrix equations were characterised by a prediction within ±10% of measured REE in more than 60% of subjects. Of the four algorithms, only the Aggregate equation did not show a significant association of the measurement bias with age, BMI and gender. Conclusions: The Aggregate algorithm was characterised by a higher, overall accuracy for the prediction of REE in older subjects and its use should be advocated in older subjects. However, due to the large variability of the estimates, the measurement of REE by IC is still recommended for an accurate assessment of individual REE.