Nutritional support is an integral part of intensive care. In practice, caloric needs are estimated using both measured energy expenditure (MEE) and a fixed sum of calories based on predictive equations.^[1–4] Energy expenditure (EE) during the stay in the ICU is known to be dynamic, influenced by body temperature^[5], nutritional support^[6], presence of sepsis^[7], level of sedation^[8], and physiotherapy^[9]. The only measurement of EE using techniques such as indirect calorimetry (IC) can accurately estimate the energy requirements of critically ill patients (CIP).^[10] Measurement of EE contributes to the prevention of overfeeding during the early phase of intensive care and underfeeding during the late phase, both of which are associated with increased mortality.^[11–13]

EE can be accurately calculated with IC which measures oxygen consumption (VO₂) and carbon dioxide production (VCO₂) from the respiratory mixture.^[14,15] EE can then be calculated using Weir’s abbreviated formula^[16]:

ЕE kcal/day = 3.941×VO₂ (L/min) + 1.11×VCO₂ (L/min)×1440

Although IC is the reference method for EE assessment^[17], metabolic monitoring is often unavailable, implementation of IC is time- and resource-consuming, and even in a prospective clinical trial study, IC is performed effectively in only 40% of patients^[18]. Therefore, daily assessment of EE by IC is difficult to implement but may be important, as EE is known to vary significantly over time because of changes in metabolic rate.^[19–21]

In CIPs, the EE calculated by predictive equations shows a significant difference from MEE measured by IC.^[22] More than 200 predictive equations have been developed, and there is no consensus on which of them should be used in routine practice. It is best to use specific equations in patients like the reference population from which the equation was derived. Equations that only consider static variables such as height, age, weight, and gender and do not account for metabolic changes are proverbially inaccurate in estimating the caloric needs of critically ill patients. If equations are to be used to calculate the EE in critically ill patients, the Penn State University equations are the best choice because they include some dynamic parameters, such as body temperature and minute ventilation.^[23,24]

An alternative method is the assessment of EE based on ventilator-derived VCO₂. Modern mechanical ventilators can measure VCO₂ continuously, making the method practical and allowing long-term monitoring.^[25–27] Since the respirator cannot measure VO₂, the Weir’s equation is adjusted using a respiratory quotient (RQ) to calculate EE as follows:

ЕE kcal/day = (3.941×VCO₂/RQ + 1.11×VCO₂)×1440

To date, several studies have examined the EE–VCO₂-ventilator with mixed results.^{[25-27,31-33]}

The aim of the present study was to compare the EE–VCO₂-ventilator in mechanically ventilated patients calculated by the formula (8.19×VCO₂)‌^[2,25], as well as EE from six predictive equations with a reference method MEE-IC.

This is a prospective, non-randomized study conducted over a period of one month. It included six mechanically ventilated patients who were hospitalized and treated in the ICU of the Clinic of Anesthesiology and Intensive Care at St George University Hospital in Plovdiv. Thirty 30-minute measurements were taken using a Cosmed Q-NRG+ metabolic monitor. Three to seven measurements per patient were performed on different days. The sample of patients was random. They were between 52 and 62 years old, two men and four women. Two patients with polytrauma, one of them with a dominant thoracic trauma, with a thoracic drain placed, but with a reported leak from mechanical ventilation of less than 8%. The remaining patients had subdural hematoma or intracerebral hemorrhage. In 8 of the measurements, the patients were conscious and evaluated by GCS, with an average score of 12 points. In the rest of the measurements, the patients were sedated and assessed according to the Ramsay sedation scale, with an average score of 3-4 points. One of the patients was connected to the ventilator with a tracheostomy cannula, the others were connected with endotracheal tubes. In 14 of the measurements, the patients were on pressure support ventilation, in the remaining measurements, they were in a combined mode - controlled plus supported ventilation, the average measured minute ventilation was 8 liters, and the average leak from mechanical ventilation was 4%. Only in three measurements did the patients receive vasopressors but continued to have mean arterial pressures greater than 65 mmHg; otherwise, all patients had stable hemodynamics. The calorimeter was calibrated according to the manufacturer’s recommendations with a routine Pneumotach test before each measurement, as well as a monthly Gas Analyzer test and a Blower test. Patients’ inclusion criteria were stable condition at least 30 minutes before measurement, normocapnia, ventilation with FiO₂ <60% and PEEP <10 mbar. The average CO₂ production from the Drager Evita Infinity V500 respirator (VʹCO₂, ml/min) was tracked for the same period. The МEE-IC was compared with EE–VCO₂-ventilator calculated by this formula - 8.19×VCO₂^[2,25] as well as with EE from six predictive equations: the Harris-Benedict^[28], the Mifflin-St. Jeor^[29], the equations of Penn State University^[30], from the proposed body weight formulas of ASPEN^[4] 12–25 kcal/kg/d and ESPEN^[2] 20–25 kcal/kg/d, 20 kcal/kg/d, and 25 kcal/kg/d were included in the comparison. The average minute ventilation for the calorimetry period as well as the highest body temperature recorded during the last 24 hours was taken to calculate the Penn State University equations. The mean EE of the different methods was calculated. The MEE by IC is the method used as a reference to which the results achieved by all other prediction methods are compared. The mean difference and the standard deviation between each of the calculation methods and the reference one were presented as ΔcEE-mEE. Student’s t-test was applied to compare paired data. Correlations were calculated using the Pearson’s test and the results were presented as r.

The average МEЕ–IC was 1650 kcal. The mean measured EЕ–VCO₂-ventilator was 1669 kcal. A statistically nonsignificant difference was found between the two measurements (p=0.84, r=0.98), the mean difference to the reference method and standard deviation (19±68). Of the predictive equations we compared, the lowest difference to the reference method was calculated with the Penn State 3 (30±236), mean EE of 1679 (p=0.81, r=0.78). The obtained results are presented in Table 1.

The present prospective study in mechanically ventilated patients confirms the concept that EE can be accurately estimated by ventilator-derived VCO₂. Moreover, it shows that this method is more accurate than most predictive equations, especially those using only static parameters. However, the Penn State 3 equation is a good choice if predictive equations are to be used because it includes variables (minute ventilation and body temperature).

The results of our study are consistent with those of Stepel et al., who, in a sufficient sample size (84 patients on mechanical ventilation, a heterogeneous group), compared МEE from a 24-h IC, EP–VCO₂-ventilator as well as EE–VCO₂ from a metabolic monitor. They found that the EE–VCO₂-ventilator was acceptably accurate and more precise than predictive equations.^[25] Based on the study, ESPEN^[2] in 2018 recommended that in the absence of IC, measurement of EE should be carried out by using the carbon dioxide production obtained from the ventilator according to the formula proposed by Stepel: EE = VCO₂×8.19.^[25] In another study which makes a comparison of a different method of energy expenditure in COVID-19 mechanically ventilated patients, Saseedharan et al.^[31] concludes that the EE estimated by ventilator-derived carbon dioxide correlated better with IC values than the energy expenditure derived from weight-based calculations.

One of the latest studies, Linder et al.^[32] published on 7 January 2023, shows that median measured resting EE was significantly higher in the critically ill (1457 kcal/d) versus the healthy cohort (1351 kcal/d), with low predictive equations accuracy rates (21% to 49%), showing again the need to be more accurate.

Rousing et al.^[26] confirmed the lack of accuracy of predictive equations and indicated the use of VCO₂ as a more accurate alternative for EE estimation. The study confirms that using only VCO₂ without VO₂ is a sensitive method for determining EE. This study’s shortcoming is using VCO₂ only from the calorimeter and not from the ventilator. Koekkoek et al.^[27] came out with a controversial opinion: EE–VCO₂-ventilator compared to EE–IC overestimates actual energy expenditure, and predictive equations, although inaccurate, may even predict EE better compared to the VCO₂ method. In this study, in contrast to Stepel’s study^[25], a greater difference in reported VCO₂ from the ventilator compared with VCO₂ from the metabolic monitor was noted, which appears to be the cause of an overestimation of EE–VCO₂-ventilator versus EE–IC. In the study by Koekkoek et al.^[27], the authors themselves indicated that the significant bias and low levels of accuracy in the study could be attributed to inaccuracy in VCO₂ measurement from the ventilator (calibration error, patient-ventilator desynchronization), or inaccuracy of the metabolic monitor (error during calibration, large variability (>10%) of VCO₂ and VO₂ during measurement. Another reason for the difference between the two studies was a difference in the duration of the measurements, which was 24 hours in the study by Stapel et al.^[25] and 6 times per day for 10 minutes each in the study by Koekkoek et al.^[27]

Briassoulis et al.^[33] conclude that VCO₂-derived predicted EE cannot be recommended as an alternative to EE measured by IC in mechanically ventilated children. They stated that a new generation of user-friendly, cost-effective calorimeters incorporated into the ventilators’ hardware and software is a one-way street to overcome the current limitations in reliably measuring real-time EE in an intensive care setting. To some extent, we agree with that. IC remains the gold standard method to measure EE. Maybe, this new generation of calorimeters is the future, but they are not widespread in the world, also results of measurements with them will be discussed first. So, while IC is still a time- and resource-consuming process, we need an alternative. In this study, calculated EE by predictive equations in children is not included.

A disadvantage of our study is the small number of patients, 6 patients who had a total of 30 measurements on different days, which can be a potential source of bias in the obtained results. Moreover, factors such as gender and age are not taken into consideration.

In critically ill patients on invasive mechanical ventilation, assessment of EЕ by analysis of CO₂ production by the ventilator is a reliable alternative. EE–VCO₂-ventilator is more accurate than most predictive equations. Unlike indirect calorimetry, the method is easy to apply, convenient for long-term monitoring, does not take additional time and resources, and is not associated with additional disconnection of the patient from the respiratory circuit. The results of our study coincide with those of Stapel et al. Further research is needed to determine the applicability of the method in routine practice. If, however, it is necessary to use predictive equations in mechanically ventilated patients, the Penn State 3 equation is a good choice.