In terms of materialscience, mechanical power refers to the ability of a material or component to withstand loads and transmit energy without undergoing deformation.1 In mechanical ventilation, the term Mechanical Power (MP) defines the amount of energy delivered to the respiratory system (rather than the lung) in a minute (rather than 1 cycle).2 Gattinoni proposed a formula for calculating MP obtained by geometric transformation of the pressure-volume loop and based on the main ventilator parameters in volume-controlled (VC) mode: tidal volume (Vt), driving pressure (DP), flow, respiratory rate (RR), and positive end-expiratory pressure (PEEP).3 Several simplified versions of the same formula have been developed.4–6 Becher and van der Meijden have also proposed formulas for calculating MP in pressure control (PC).7,8

Several prospective studies have linked MP to developing ventilator-induced lung injury9 (VILI), and even death10–12 of mechanically ventilated patients.13 Recent clinical studies have suggested that MP levels above 18−20J/min are associated with an increased risk of death in patients undergoing mechanical ventilation.14,15

The primary objective of this study is to determine the prevalence of elevated MP values (>17J/min) utilized in routine clinical practice. This will be achieved through the application of established formulas described in existing literature, specifically focusing on patients undergoing controlled ventilation. As a secondary objective, the study aims to investigate the weight of distinct components of mechanical power in both volume-controlled (VCV) and pressure-controlled (PCV) ventilation modes.

This observational, descriptive, cross-sectional, analytical, multi-centre, and international study was conducted on November 21st, 2019, from 8 a.m. to 3pm (Supplementary material 1). Eligible patients were all those who on November 21, 2019, during the specified time, met the inclusion criteria: being admitted to a critical care area and receiving mechanical ventilation for any cause.

Exclusion criteria included patients admitted to critical care areas who were not receiving invasive mechanical ventilation at the time during the time specified, receiving invasive mechanical ventilation for less than 6h, tracheotomized patients, patients ventilated with APRV (Airway Pressure Release Ventilation) y BiLEVEL (BiLevel Positive Airway Pressure) modes, non-invasive ventilation, selective ventilation, ventilation through nasotracheal intubation, and patients undergoing extracorporeal circulatory and respiratory support (ECMO-VV, ECMO-VA, and external carbon dioxide removal devices).

Clinical-demographic patient variables, variables related to mechanical ventilation, and analytical variables were recorded (Supplementary material 1).

The formula used to calculate MP in VC modes was the "simplified" formula proposed by Gattinoni et al.: MP=0.098 × Vt × RR × PIP − [(Pplat − PEEP)/2]. In the PC mode, the "simplified" formula proposed by Becher et al. was: MP=0.098 × Vt × RR × (ΔPins+PEEP). Based on recent publications, a threshold of high mechanical power was established at 17J/min.14–16

The study received approval from the Clinical Research Ethics Committee of Marqués de Valdecilla University Hospital, Santander, Spain. Informed consent was required. Subsequently, the necessary documentation was provided to each participating centre to obtain approval from their respective Clinical Research Ethics Committess, ensuring that all participating centres had the necessary approvals before starting the study.

The study was registered in the ClinicalTrials.gov database: NCT03936231.

A total of 664 critically ill patients from 133 intensive care units in 15 countries (Argentina, Brazil, Chile, China, Colombia, Ecuador, Spain, Honduras, Italy, Morocco, Mexico, Peru, Poland, Portugal, and Uruguay) were registered. Fifty-seven records were in pediatric patients, 196 in non-ventilated patients in controlled mode. Out of the 411 patients enrolled in controlled mode, 36 were excluded due to errors in the record or having incomplete records (Flowchart in Fig. 1).

Figure 1.

Flowchart of the study patients.

(0.17MB).

Spain led the participation with 188 patients, followed by Ecuador with 54 and Argentina with 34. Colombia and Mexico also made significant contributions with 18 and 30 patients, respectively. Regarding the collaboration of ICUs, Spain led once again with 56 participating ICUs. Ecuador contributed 8 ICUs, while Argentina and Colombia provided 10 and 4 ICUs, respectively. Other countries like Italy, Chile, and Uruguay also participated in the study, standing out for the number of patients and ICUs involved (Fig. 2).

Figure 2.

Distribution of patients and participation of ICUs in the study by country.

(0.16MB).

Of the 372 patients who were ventilated in controlled the main cause of orotracheal intubation was respiratory failure (33%; 95%CI: 12–63), followed by neurological impairment (31%; 95%CI: 11–60). Table 1 shows the main clinical-demographic, gasometric, and ventilatory variables analyzed in the total sample and the main differences between the compared groups are shown in Table 2.

The MP value in the patients under VC ventilation sample was 16 (7) J/min. For patients under PC ventilation, it was 19 (8) J/min. The difference in MP values was 3J/min (95%CI: 1–5; p<0.001) higher in the pressure-controlled group. In 38% of the analyzed subjects, the mechanical power value was greater than 17J/min. The prevalence of high MP (number of patients with MP>17J/min / number of ventilated patients) was 38%. The values of the different variables of the MP formula in VC and PC are shown in Table 3; no significant difference was found in the prevalence of mechanical power exceeding 17J/min between the two groups (p=0.382).

The Table 4 presents the information for nine different models of patients under VC, numbered from 1 to 9. The "best" model is Model 7 (DP+PEEP+StrainRate.Subr+RR), as it has the lowest AICc value. The other models are compared to this best model. The Delta_AICc column shows the difference between the AICc value of each model and the AICc value of Model 7. The data less well support models with larger Delta_AICc values. The Akaike weight of evidence (AICcWt) column shows the weight of evidence in favor of each model, with higher values indicating stronger support. The weight of evidence in deciBans, calculated as 10*log10(AICcWt/max(AICcWt) (WOEagaBM) column presents the weight of evidence in deciBans, with positive values indicating support for a given model and negative values indicating evidence against it.

Based on Table 5, the model with the highest WOEagaBM in patients under pressure control ventilation is Model 16 (Strain_subr_PC+eVT+PEEP+StrainRate_subr_PC+RR), with a WOEagaBM value of 0. Model 16 has a K value of 7, an AICc value of 1652, a Delta_AICc of 0, a ModelLik value of 1e+0, an AICcWt value of 4.55e−1, an LL value of −819, and a Cum.Wt value of 0.455. The WOEagaBM values for the other models range from −975 to 0, with the models having higher values being less supported by the evidence. Therefore, Model 16 is the best supported model according to the weight of evidence analysis provided.

The algorithmic selection of models for each ventilatory mode is detailed in Supplementary Material 3 and 4.

In the studied population, nearly two-fifths of subjects undergoing controlled mechanical ventilation exceeded the 17J/min mechanical power threshold. Despite variations in mechanical power values between patients undergoing volume-controlled (VC) and pressure-controlled (PC) ventilation, no significant differences were observed in the prevalence of elevated mechanical power values.

The calculation of MP in mechanical ventilation is a complex topic that is still evolving. There are various formulas and methods for calculating MP, and often a large amount of detailed information and precise measurements are required for its application.23,24 There are numerous formulas and approaches to calculate MP in mechanical ventilation, and MP values can vary significantly. Simplified formulas – validated in previous studies – such as those proposed by Gattinoni et al.25 and Becher et al.26 are simple and practical approaches for MP calculation in VC and PC modes, respectively.

In this context, the specific MP values described in the literature for VC and PC are heterogeneous. Protti et al.27 found that MP values in VC varied between 4 and 43J/min, depending on the formula used for MP calculation. In a study by Amato et al.28 MP values in VC varied between 10.6 and 16.4J/min, depending on the formula used. Regarding MP values in PC, a study by Costa et al.29 found that MP values varied between 8.8 and 17.3J/min, depending on the formula used. In another study by Bellani et al.,30 MP values in PC varied between 8.7 and 13.2J/min, depending on the formula used.

Our results reveal a significant difference in MP values between patients ventilated in VC and PC modes. This finding is consistent with the data obtained by Rietveld et al.,31 who found that in 46 mechanically ventilated patients without spontaneous breathing, the mean MP in VC was significantly lower than in PC (absolute difference of 1.26J/min; SD 0.14J/min; p<0.00001).These results contradict the hypothesis that PC modes produce less MP than VC, based on the the idea that the maximum peak pressure is equivalent to Pplat.32

In general, the use of surrogates is a commonly employed strategy in modelling complex systems, including the human respiratory system. Surrogates represent lung tissue's response to different ventilation modes, allowing for the evaluation of how mechanical ventilation affects lung tissue. In the literature, several studies have utilized surrogates for respiratory system evaluation. For example, in a 2016 study, Bellani et al.33 used strain and strain rate surrogates to evaluate the relationship between MP and clinical outcomes in acute respiratory distress syndrome (ARDS) patients treated with mechanical ventilation. In another 2018 study, Sinderby et al.34 used volume and pressure surrogates to evaluate different ventilation modes in patients with acute respiratory failure.

The correlation between PEEP and FRC in the context of mechanical ventilation is grounded in the principles of alveolar structure and lung elastic properties. The application of PEEP in mechanical ventilation involves maintaining a constant positive pressure in the airways, with effects on maintaining alveolar patency. Consequently, this resembles the condition of the FRC (the amount of air that remains in the lungs after a normal expiration, preventing collapse of smaller alveoli and providing a foundation of air for the next inspiration). Since applying PEEP during mechanical ventilation has a similar effect to maintaining the FRC, it can be postulated that PEEP could serve as an indirect indicator of FRC in patients under ventilatory support. However, it is important to recognize that PEEP and FRC are not equivalent, and there are physiological and clinical differences between them. PEEP can influence alveolar distension and transpulmonary pressures, whereas FRC is related to functional reserve and preventing alveolar collapse.35

The presence of "strain" and "strain rate" surrogates in the AIC-established models of MP in our results may be related to the viscoelastic behaviour of the lung.36 Lung tissue has viscoelastic behaviour, which means it can deform and recover its original shape in response to applied forces. Still, the deformation and recovery may have different speeds and characteristics depending on the viscoelastic properties of the tissue.37 In a study by Gattinoni et al., it was demonstrated that using Driving Pressure (as a surrogate for strain) and Flow/PEEP (as a surrogate for strain rate) in patients on controlled ventilation was associated with a reduction in mortality compared to using plateau pressure as an adjustment parameter.38 A study by Talmor et al. found that using strain and strain rate surrogates for adjusting mechanical ventilation parameters in patients with ALI was associated with improved arterial oxygenation and reduced mechanical ventilation time.39

We consider that the main strengths of the study encompass its multicenter and international design, the inclusion of a diverse sample of critically ill patients, the utilization of rigorous statistical methodologies, and the incorporation of sophisticated analyses and measures for model selection and comparison. Furthermore, the study's transparency and detailed presentation of the employed procedures contribute to its robustness.

As for the weaknesses of the study, first: the analysis of variables with "significance" which are included in the calculi for the mechanical power has potential statistical collinearity, as a circular argument. Second: it is an observational study, which means that a controlled and randomized intervention was not performed. Consequently, causal relationships between mechanical ventilation modes and measured outcomes cannot be established. Additionally, although multivariable regression models were adjusted using the AIC, and the fit of each model was evaluated with the analysis of their standardized residuals and with QQ plots for normality, it is possible that other fitting models could have been used, which could affect the results. In other hand, the patient sample per hospital and the single-day period may raise concerns regarding the representativeness and robustness of the results.

With the presented data, we can conclude that a substantial portion of patients undergoing mechanical ventilation may be at risk of experiencing elevated mechanical power levels, potentially contributing to complications associated with ventilator-induced lung injury. Despite the observed differences in mechanical power values between VC and PC ventilation, no statistically significant difference was found in the prevalence of mechanical power exceeding 17J/min between the two groups. This indicates that while mechanical power levels differed, they did not result in a significant disparity in the prevalence of elevated mechanical power values.

Alejandro González-Castro: Initial study design, study promotion, data refinement, manuscript writing.

Patricia Escudero-Acha: Initial study design, study promotion, data refinement.

Alberto Medina Villanueva: Initial study design, manuscript writing and translation correction.

Aurio Fajardo Campoverdi: Corrección de la versión final del manuscrito, Promoción del estudio, Análisis estadístico.

Federico Gordo Vidal: Initial study design, study promotion, final manuscript version correction.

Ignacio Martin-Loeches: Initial study design, study promotion, final manuscript version correction.

Angelo Roncalli Rocha: Initial study design, study promotion, final manuscript version correction.

Marta Costa Romero: Initial study design, study promotion, final manuscript version correction.

Marianela Hernández López: Initial study design, study promotion, final manuscript version correction.

Carlos Ferrando: Initial study design, study promotion, final manuscript version correction.

Alessandro Protti: Initial study design, study promotion, final manuscript version correction.

Vicent Modesto i Alapont: Initial study design, study promotion, final manuscript version correction, statistical analysis.

Mechanical Power Day Group*: Variable collection.