Body weight definitions for estimating energy expenditure in intensive care, a prospective monocentric observational study

BACKGROUND AND AIMS: Accurate estimation of energy expenditure (EE) in critically ill patients is essential to avoid underfeeding and overfeeding. Indirect calorimetry (IC) is the reference method but remains underused in routine practice. Consequently, clinicians rely on predictive equations such as the Harris-Benedict (H-B) equation with and without stress correction or simplified weight-based prescriptions (25 kcal/kg/day), despite ongoing uncertainty regarding their accuracy and the optimal body weight (BW) definition (actual, ideal, or adjusted). We aimed to compare EE measured by IC with EE estimated using these approaches across different BW definitions. METHODS: In this prospective, observational, single-center study conducted in a medical ICU (Clermont-Ferrand, France) between March 2023 and April 2025, adult patients with at least one IC measurement before day 5 were included. IC measurements were performed on days 3, 5, 7, 10, and 14 when feasible. Predicted EE was calculated using the H-B equation (with and without 1.3 stress correction) and the 25 kcal/kg/day formula using multiple BW definitions (admission, daily, ideal [Lorentz or Robinson], and adjusted BW). Agreement between measured and predicted EE was assessed using repeated-measures Bland-Altman analysis based on linear mixed-effects models, accounting for within-patient correlation. RESULTS: , ICU mortality 28%) contributed 225 IC measurements. Median EE remained relatively stable from day 3 (1667 kcal/day) to day 14 (1716 kcal/day). Predictive equations demonstrated systematic bias with wide limits of agreement. Unadjusted H-B underestimated EE, whereas 1.3 stress-corrected H-B and weight-based approaches tended to overestimate it. For the 25 kcal/kg/day formula, bias varied according to BW definition (-262 to +189 kcal/day). Adjusted BW provided the lowest mean bias (-5 to -10 kcal/day), but limits of agreement remained wide (approximately ±1000 kcal/day). Larger discrepancies were observed in patients with higher severity and invasive mechanical ventilation. CONCLUSION: Predictive equations show poor agreement with IC at the individual level in critically ill patients. Although adjusted BW reduces systematic bias, variability remains clinically unacceptable. EE appeared relatively stable between days 3 and 14. IC should remain the reference method to guide nutritional therapy; when unavailable, predictive approaches should be used with caution.