Body composition affects the accuracy of predictive equations to estimate resting energy expenditure in older adults:...

To investigate the accuracy of ten different predictive equations to estimate resting energy expenditure (REE) in a sample of Brazilian older adults and develop a predictive equation for estimating REE based on body composition data.

A cross-sectional study with thirty-eight Brazilian older adults aged 60-84 years, who had their REE measured by indirect calorimetry (IC) and BC assessed by dual-energy x-ray absorptiometry (DXA). REE was compared to the estimation of ten predictive equations, and the differences between BC and anthropometric-based equations were investigated using Bland-Altman plots and Lin's concordance correlation. Accuracy was evaluated considering ±10% of the ratio between estimated and measured REE.

The sample was composed of 57.9% men, with a mean age of 68.1 (5.8) years, and a mean REE by IC of 1528 (451) kcal. The highest accuracy was 47.4% obtained by Luhrmann and Fredrix equations, and the lowest accuracy was 13.2% reached by Weigle equation. In general, the proportion of underestimation was higher than overestimation. All anthropometric-based equations presented a good agreement with REE from IC. For those equations derived from BC, however, three of them reached only a moderate agreement. In terms of accuracy, all equations presented lower than 50% of accurate prediction of REE.

In this sample of older adults, previous predictive equations to estimate REE did not show good accuracy, and those based on BC presented even worse results, showing that changes in BC related to aging could impact the accuracy of these equations.