Accurately estimating resting energy expenditure (REE) in critically ill obese patients remains a significant clinical challenge, as predictive equations are consistently inadequate. Metabolic heterogeneity across obesity classes and the role of substrate utilization are insufficiently characterized. Objective: To evaluate the impact of different weight-normalization methods on the interpretation of REE and to identify independent metabolic determinants of weight-adjusted energy expenditure in critically ill patients with obesity. Methods: Bicentric cross-sectional study of 148 critically ill adults with obesity undergoing indirect calorimetry. REE normalized by actual body weight (REE/kg), ideal body weight (REE/IBW), and adjusted body weight (REE/AdjBW) was calculated. Multivariable models with robust standard errors (HC3), stratified analyses by obesity class (I–III) with a Chow test, and internal validation were performed using 10-fold cross-validation and bootstrap resampling (1000 iterations). Results: Absolute REE did not differ significantly between BMI categories (p = 0.679), while REE/kg progressively decreased from normal weight (27.8 kcal/kg/day) to class III obesity (16.9 kcal/kg/day; p < 0.001). The respiratory quotient (RQ) emerged as the most robust independent correlate of adjusted REE (β = −13 to −15 kcal·kg−1·day−1; p < 0.001), whereas clinical severity scores (SOFA, APACHE II) and comorbidity (Charlson) did not show significant associations. Stratified analyses revealed significant structural heterogeneity between obesity classes (F = 4.545, p = 0.0001), with no significant predictors identified in class III obesity, likely reflecting limited statistical power in this subgroup. Conclusions: Normalizing REE using different weight indices fundamentally alters its metabolic interpretation. RQ surpasses traditional clinical scores as a correlate of adjusted REE, consistent with a phenotype of metabolic inflexibility. The heterogeneity between obesity classes underscores the need for individualized indirect calorimetry rather than reliance on predictive equations.
