Validity of Predictive Equations for Resting Energy Expenditure in Greek Adults

Abstract

Aim: To examine the validity of published resting energy expenditure (REE) equations in Greek adults, and if indicated, develop new cohort-specific predictive REE equations. Methods: Indirect calorimetry and anthropometric data were obtained from 226 adult volunteers of diverse age groups and body mass index ranges (18–60 years, 16.6–67.7 kg·m^–2). Measured REE was compared to preexisting prediction equations via correlation, regression, and Bland-Altman analysis. Then, cohort-specific REE equations were developed using curve estimation and nonlinear regression. To reduce type I error, presently derived equations were validated by splitting the sample into a training and validation group. Results: Preexisting equations over-predicted in-cohort REE. Equations by Livigston and Kohlstadt were most accurate at the individual level (63% accuracy), while formulas by Owen and collaborators elicited highest accuracy at the group level (–1.8% bias). Bland-Altman analysis showed proportional bias for most equations. Currently developed equations showed highest overall accuracy with 70% at the individual and group level (1.0% bias), with small differences between measured and predicted REE values (mean, 95% CI 36 [–15 to 88] kcal·day^–1). Conclusion: Data indicate currently developed equations to be the most accurate and valid for estimating REE in Greek adults. Further studies should examine the developed equations in an independent sample.