Body weight, creatinine and age predict energy expenditure in critically ill AKI patients

Critically ill patients with acute kidney injury (AKI) show considerable variation in how much energy their bodies actually expend, and a new study reports that a simple equation combining body weight, serum creatinine and age can estimate that expenditure from a bedside indirect calorimetry measurement better than several conventional weight-based or formula-based approaches, while still falling short of the precision needed to replace direct measurement in individual patients.

The study, published in BMC Nephrology, was conducted as a prospective cross-sectional investigation carried out alongside the ENGINE randomized controlled trial at two medical intensive care units between March and November 2023. The researchers enrolled 80 critically ill patients with AKI who were not receiving renal replacement therapy (RRT), a group in which accurate assessment of energy requirements is particularly important because AKI alters metabolic balance and because nutritional support in these patients must be tailored carefully.

Indirect calorimetry (IC) is regarded as the gold standard for measuring energy expenditure, because it quantifies energy use directly from respiratory gas exchange. Its clinical use, however, is limited by availability of the equipment and the practical demands of the measurement. That gap has motivated numerous attempts to predict energy expenditure from routinely collected clinical variables, and the team behind the new study set out to identify the factors associated with IC-derived energy expenditure (ICEE), to develop and internally validate an exploratory prediction equation, and to categorize patients into hypometabolic, normometabolic and hypermetabolic groups.

In univariable analyses, eight variables were associated with ICEE: age, body weight, body mass index, current creatinine, AKI stage, serum lactate, PaO₂ and PaCO₂. After reducing overlapping predictors, six variables were considered for multivariable model development. The resulting equation, named TEE-cAKI, takes a strikingly compact form: predicted ICEE (kcal) equals 1275.385 plus 9.517 times body weight, plus 71.257 times current creatinine, minus 6.694 times age. In other words, heavier patients and those with higher creatinine values were predicted to expend more energy, while older age was associated with lower predicted expenditure.

The model was internally validated using repeated 10-fold cross-validation, and the investigators evaluated a battery of performance metrics: cross-validated R², root mean square error (RMSE), mean absolute error, mean absolute percentage error, calibration intercept and slope, and the proportions of predictions falling within ±10% and ±15% of the measured IC-derived value. Internal validation of this kind tests how the equation performs within the same dataset rather than in a new population, a limitation the authors acknowledge in describing the work as exploratory.

When patients were categorized by metabolic status, 45% were hypermetabolic, 36.25% normometabolic and 18.75% hypometabolic. This distribution underscores the heterogeneity of energy expenditure in this population: nearly half of the patients burned more energy than would be expected, while almost one in five burned less, meaning that a single fixed caloric target is unlikely to suit everyone.

To place TEE-cAKI in context, the researchers compared it against weight-based methods of 20, 25 and 30 kcal/kg/day, the Harris–Benedict equation, and the Harris–Benedict equation multiplied by a stress factor of 1.3, using Bland–Altman analyses against measured ICEE. The TEE-cAKI equation showed the smallest mean bias of all the evaluated methods. However, all predictive methods exhibited wide limits of agreement with measured ICEE, indicating substantial variability at the individual-patient level. A wide limit of agreement means that for any given patient, the predicted value could differ considerably from the measured one, even if the average bias across the group is small.

The authors’ conclusion reflects this tension. They observed considerable heterogeneity in energy expenditure among critically ill patients with AKI not receiving RRT, and they report that TEE-cAKI showed the smallest mean bias and lowest RMSE among the evaluated methods, but that its explanatory and individual-level predictive performance were modest. On that basis, they state that indirect calorimetry remains the preferred method whenever it is available.

For clinicians, the practical message is twofold. First, routinely available variables — body weight, current creatinine and age — carry real signal about energy expenditure in this population, and creatinine’s appearance in the equation is consistent with the metabolic derangement that AKI produces. Second, no equation evaluated in the study, including the new one, achieved the individual-level agreement that would allow it to substitute confidently for a direct measurement. The findings therefore support a layered approach: equations may offer a reasonable starting estimate when indirect calorimetry is unavailable, but the gold standard measurement retains its place at the bedside.

The study’s scope is also worth keeping in view. It involved 80 patients at two medical intensive care units over roughly nine months, all with AKI and none receiving RRT, and the equation was internally, not externally, validated. Whether TEE-cAKI performs as well in other centers, in patients on renal replacement therapy, or in larger and more diverse cohorts remains to be tested. As the authors note, the equation’s performance was modest at the level of the individual patient, and the wide limits of agreement observed across all predictive methods serve as a caution against over-reliance on any single formula when nutritional decisions carry clinical weight.

Subject of Research: Prediction of indirect calorimetry-derived energy expenditure in critically ill patients with acute kidney injury not receiving renal replacement therapy

Article Title: Factors associated with indirect calorimetry-derived energy expenditure in critically ill patients with acute kidney injury not receiving renal replacement therapy

Article References: Jeerangsapasuk, W., Hongmeng, W., Tantiyavarong, P., Khantason, A., & Chatkrailert, A. (2026). Factors associated with indirect calorimetry-derived energy expenditure in critically ill patients with acute kidney injury not receiving renal replacement therapy. BMC Nephrology. https://doi.org/10.1186/s12882-026-05456-1

Image Credits: AI Generated

DOI: 10.1186/s12882-026-05456-1

Keywords: acute kidney injury, indirect calorimetry, energy expenditure, critical care, nutrition, prediction equation, intensive care unit, metabolism, Bland-Altman, TEE-cAKI, Factors, associated