This comparative study assessed resting energy expenditure (REE) measured by indirect calorimetry (m-REE) against estimates from seven predictive equations in critically ill patients. The cohort was stratified by illness severity using APACHE II scores to examine whether the relationship between measured and predicted REE varied with severity. The central aim was to evaluate agreement between methods and explore associations of m-REE with markers of inflammation and clinical prognosis.
REE was measured using indirect calorimetry; the study compared m-REE with REE values calculated by seven predictive equations: Harris-Benedict (H-B), adjusted H-B, Mifflin-St Jeor, WHO/FAO/UNU, Owen, Henry, and a weight-per-kilogram equation. Patients in the intensive care unit were categorized according to APACHE II scores (threshold at 15) to create higher-severity and lower-severity subgroups.
Inflammatory status was assessed using serum procalcitonin (PCT) and high-sensitivity C-reactive protein (hsCRP). Clinical prognosis measures included ICU length of stay (ICU staying days) and overall hospital length of stay. The abstract reports statistical comparisons between m-REE and predicted values and correlation analyses between m-REE and the inflammatory and prognostic variables. Specific sample size, measurement timing, calorimetry protocol, and statistical methods beyond p-value thresholds were not provided in the abstract and would require review of the full text for full methodological detail.
In patients with APACHE II ≥15, the study found that m-REE measured by indirect calorimetry was significantly higher than the REE values predicted by most equations (reported p < 0.01). However, three predictive methods—adjusted Harris-Benedict, Owen, and the weight-kilogram equation—did not differ significantly from m-REE in this higher-severity subgroup. This suggests that in more severely ill patients, many standard predictive formulas may underestimate energy expenditure, while a subset of equations produced estimates closer to measured values.
When APACHE II <15, the pattern reversed: m-REE tended to be lower than the predicted values from the equations. The abstract reports that m-REE was generally lower than predictive estimates in this lower-severity group, though specific p-values for each comparison are not listed in the abstract. This indicates that predictive equations may overestimate energy needs for less severely ill ICU patients.
The study reports a positive correlation between m-REE and serum levels of hsCRP and PCT in the subgroup with APACHE II ≥15. In contrast, no association between m-REE and these inflammatory markers was observed when APACHE II <15. These findings indicate that elevated measured energy expenditure in more severely ill patients aligns with markers of systemic inflammation, suggesting a pathophysiologic link between inflammatory response and increased metabolic demand.
Measured REE was also positively correlated with both ICU staying days and hospital length of stay in the APACHE II ≥15 group. No associations between m-REE and length of stay were found in patients with APACHE II <15. This pattern implies that higher metabolic rates measured early or during critical illness may be associated with longer resource use and prolonged hospitalization among more severely ill patients.
The authors conclude that when indirect calorimetry is not available, the adjusted Harris-Benedict, Owen, and weight-per-kilogram predictive equations may be acceptable surrogates for estimating REE in patients with APACHE II ≥15. The study further suggests that m-REE may have utility as a marker of inflammatory status and a potential predictor of clinical prognosis in critically ill patients, given its correlations with hsCRP, PCT, and length of stay in the higher-severity group.
From a clinical nutrition perspective, these results highlight the potential for both underfeeding and overfeeding depending on illness severity and choice of predictive equation. For severely ill patients, many standard equations may underestimate needs, which could lead to inadequate caloric provision if caloric targets rely solely on those formulas. Conversely, for less severely ill patients, predictive equations may overestimate needs, risking overfeeding if used without confirmation.
The abstract reports key group-level findings but does not include several important methodological details in the summary provided: exact sample size, patient demographic characteristics, timing and duration of indirect calorimetry measurements, definitions used for nutritional targets, numerical values or effect sizes of the differences, and precise correlation coefficients. Those details are necessary to appraise the magnitude and clinical relevance of the reported differences and would require consultation of the full-text article. The authors reported no conflicts of interest.
Overall, this study supports the use of indirect calorimetry as the reference standard for assessing resting energy expenditure in critical care when available, identifies a subset of predictive equations that align with measured values in more severely ill patients, and links higher m-REE to inflammation and longer hospital courses within the higher APACHE II subgroup.