Mechanical Power in Prone Position Intubated Patients with COVID-19-Related ARDS: A Cohort Study

Background Respiratory monitoring of mechanical ventilation (MV) is relevant and challenging in COVID-19. Mechanical power (MP) is a novel and promising monitoring tool in acute distress respiratory syndrome (ARDS), representing the amount of energy transferred from the ventilator to the patient. It encompasses several setting parameters and patient-dependent variables that could cause lung injury. MP can therefore be an additional tool in the assessment of these patients. Objective This study aims to evaluate respiratory monitoring through MP and its relationship with mortality in patients with COVID-19-related ARDS (CARDS) under mechanical ventilation (MV) and prone position (PP) strategies. Methods Retrospective, unicentric, and cohort studies. We included patients with CARDS under invasive MV and PP strategies. Information regarding MP, ventilation, and gas exchange was collected at 3 moments: (1) prior to the first PP, (2) during the first PP, and (3) during the last PP. We tested the relationship between MP and VR with in-hospital mortality. Results We included 91 patients. There was a statistically significant difference in MP measurements between survivors and nonsurvivors only in the last prone position (p < 0.001). This is due to the significant increase in MP measurements in nonsurvivors (difference from the baseline: 3.63 J/min; 95% CI: 0.31 to 6.94), which was not observed in the group that survived (difference from the baseline: 0.02 J/min; 95% CI: −2.66 to 2.70). In multivariate analysis, MP (p=0.009) was associated with hospital death when corrected for confounder variables (SAPS 3 score, mechanical ventilation time, age, and number of prone sessions). Conclusions MP is an independent predictor of mortality in PP patients with CARDS.


Introduction
Te prone position (PP) is a cornerstone of invasive ventilatory supply in moderate to severe acute distress respiratory syndrome (ARDS) and has been a current recommendation in COVID-19-related ARDS (CARDS) [1,2]. Tis maneuver is associated with a reduction in mortality in patients with severe ARDS compared to the supine position, as well as an improvement in oxygenation measured by the change in the PaO 2 /FiO 2 ratio [2].
Te patient's response to PP is defned primarily as an improvement in oxygenation, especially through the PaO 2 / FiO 2 ratio. It should be noted that the main reason for reducing mortality in PP is less overdistension in nondependent lung regions and less cyclical opening and closing in dependent lung regions [1,3]. Determining the response to PP using a single oxygenation index can limit the possible benefts of this therapy. Possibly, the concomitant evaluation of changes in other variables of respiratory monitoring may provide relevant information for better ventilatory support [4]. Due to its combined efects of recruiting vertebral parts of the lung, making the distribution of ventilation more homogeneous, and reducing intracycle recruitment/derecruitment, PP dampens lung stress and strain resulting from mechanical ventilation (MV) [3], decreasing the risk of ventilator-induced lung injury (VILI) [5].
More recently, the degree of VILI has been associated with the energy load transferred from the mechanical ventilator to the patient's respiratory system, converting the mechanical stimulus into intracellular biochemical and molecular signals. Some studies have shown that manipulation of individual ventilatory parameters does not add beneft in terms of reducing VILI if it does not cause a concomitant reduction in dynamic strain and energy/ power load [6,7]. Mechanical power (MP) represents the amount of energy transferred from the ventilator to the patient per unit of time. Tis index includes several setting parameters and patient-dependent variables that could cause VILI in a single measurement [8]. Recent studies have suggested that monitoring ventilatory mechanics performed by driving pressure (ΔP) and MP may be more appropriate when a customized ventilation strategy is aimed [6,7]. PP have a well-known efect on new open pulmonary units and improve the mechanical characteristics of already opened units that reach a more favorable position on the volumepressure curve [1], associated with a more homogeneous ventilation distribution [3]. Tese phenomena may theoretically have an impact on the energy load delivered to the lungs [8], refecting on MP. In a previous study, PP associated with a PEEP titration strategy minimized the parameters associated with VILI, such as MP [9].
As a result of this theoretical efect of PP on a lower and more homogeneous distribution of energy in the lung parenchyma [8], the objective of our study is to evaluate MP as a monitoring tool for ventilatory mechanics, the longitudinal trend throughout hospitalization, and prone-induced changes in intubated patients with moderate-to-severe CARDS.

Study Design.
In this retrospective study, we collected data from patients admitted between 5th May and 9th September 2020 to a tertiary intensive care unit (ICU) in Brazil. Te local ethics committee approved the study (Grupo Hospitalar Conceição Ethics Comitee-Plataforma Brasil number CAAE 51855421600005530-approval date: October 20 th , 2021). Te informed consent requirement was waived due to the retrospective and noninterventionist nature of the study. Tis study was performed in accordance with the ethical standards of the responsible committee on human experimentation and with the Helsinki Declaration of 1975. Te inclusion criteria for the analysis were patients with laboratory-confrmed SARS-CoV-2 infection (i.e., a positive result of a real-time reverse transcriptase-polymerase chain reaction assay of nasal and pharyngeal swabs), who had moderate to severe ARDS according to the Berlin criteria [10] and who required intervention therapy with mechanical ventilation and prone position. Te application of prone position therapy was carried out according to the Proseva trial criteria [2], with prone sessions of at least 16 hours. MV management was carried out with strategies that limit tidal volumes (4-8 ml/ kg of predicted body weight) and inspiratory pressures (plateau pressure <30 cm H 2 O) [11]. PEEP values are titrated with the aim, primarily, of a plateau pressure <30 cm H 2 O and, ideally, also a ΔP < 15 cm H 2 O. Recruitment maneuver was not performed in each session of prone position. In the population included in this study, prone sessions interrupted by hemodynamic instability were not identifed, and patients with severe hemodynamic instability were not included. All patients had a continuous measurement of invasive blood pressure using an arterial catheter. Patients were preferentially ventilated in volume-controlled ventilation (VCV), and all patients received continuous infusions of neuromuscular blocker drugs during the protocol. Te infusion rate of drugs for sedation and analgesia in these patients was titrated through bispectral index (BIS) monitoring.
Multiple information was collected from the electronic medical record, including admission data (demographic and anthropometric data and comorbidities) and data on clinical evolution during the ICU stay period, including ventilatory settings parameters (i. e. positive end expiratory pressure (PEEP), tidal volume (Vt), respiratory rate (RR), FiO 2 ), ventilatory monitoring (i.e. peak pressure (Ppeak), plateau pressure (Pplat), ΔP, respiratory system compliance (Crs), minute ventilation (VE)) and monitoring pulmonary gas exchange (PaCO 2 and PaO 2 /FiO 2 ratio, and ventilatory ratio [VR]). Vt was reported in ml/kg of predicted body weight (PBW). ΔP was defned as the diference between Pplat and PEEP. Static compliance (Cst) was calculated as Vt/ (Pplat − PEEP). VR was calculated from the following equation: VR � [VE (ml/min) × PaCO 2 (mmHg)]/ (PBW × 100 × 37.5). MP was expressed in J/min and calculated as previously proposed by Chiumello et al. [12] from the following equations: when in volume-controlled ventilation and MP (J/min) � 0.098 × RR × Vt × (ΔPinsp + PEEP) in pressure-controlled ventilation. Te ventilatory setting parameters and respiratory variables monitoring data were collected at three moments: (1) before the frst prone position, (2) during the frst prone position (after at least 6 hours), and (3) during the last prone position. Te primary outcome was hospital mortality.

Statistical Analysis.
Continuous variables that were normally distributed were presented as mean and standard deviations (SD). Nonnormally distributed continuous variables were represented by medians and interquartile ranges (IQR). Categorical data are presented as absolute numbers (n) and percentages (%). Te Mann-Whitney test was used to compare continuous nonparametric variables between study groups. Pearson's chi-square test was used for categorical variables. A generalized estimating equation was performed to explore the interaction between the diferent variables measured (MP, VR, PaO 2 /FiO 2 ratio, driving pressure, PEEP, and tidal volume) and the efect of time of mensuration (immediately before the frst prone position, during the frst session of the prone position, and in the last prone position), as well as its interaction with survival at ICU discharge. Tis modeling also allows us to evaluate the interaction in both time and survival status with each variable (interaction survival × time) and also to evaluate the impact of MP in survival status when corrected for confounding variables (SAPS III, number of days on MV, number of prone maneuvers, and age). We performed an exploratory analysis evaluating the diference between MPs at the three time points: MP in the frst prone minus MP preprone (delta [∆]MP 1), MP in the last prone minus MP preprone (∆ MP 2), and the MP in the last prone minus MP in the frst prone (∆ MP 3). Statistical signifcance was defned as p < 0.05. Statistical analysis was performed with SPSS software 21.0 (SPSS, IBM-Chicago, Illinois, USA) and jamovi 2.3.18.

Results
During the study period, data were collected from a total of 91 COVID-19 ARDS intubated patients in a prone position. Te ICU and the hospital mortality rate were 49% (n � 45). Most of the patients were men (63.7%); the mean age was 60.2 ± 12.8 years; and the median body mass index was 30 (26.8-34.6) kg/m 2 . Te mean simplifed acute physiology score (SAPS) 3 on admission to the ICU was 68.6 ± 15.5 points, and the mean sequential acute organ failure assessment (SOFA) score was 7 ± 2.3 points. Te median number of prone maneuvers in the overall cohort was 2 sessions (IQR 1-4). Table 1 summarizes the clinical characteristics of the patients and their clinical outcomes.    need for renal replacement therapy was signifcantly higher in nonsurvivors (57.8% vs. 26.1%; p � 0.004). Tere was no statistically signifcant diference in the incidence of venous thromboembolism.
Te ventilatory setting and monitoring parameter data collected in the preprone position, during the frst prone position, and in the last prone position are presented in Table 2. Tere was a statistically signifcant diference in MP measurements between survivors and nonsurvivors only in the last prone position (p < 0.001). Tis is due to the signifcant increase in nonsurvivors in the last prone position (diference from baseline: 3.63 J/min; 95% CI: 0.31 to 6.94), which was not observed in the group that survived (difference from baseline: 0.02 J/min; 95% CI: −2.66 to 2.70). Tere was a statistically signifcant interaction between survival status and time of measurement (pre, frst, and last prone positions) in the parameters MP (p � 0.009), VR (p � 0.009), RR (p < 0.001), and PaO 2 /FiO 2 ratio (p � 0.001) in a model adjusted for potential confounders. Tere was an interaction between ΔP and PEEP measurements only with respect to the moment of measurement and not with the survival status.

Discussion
In this study, we propose that, in patients with CARDS submitted to PP, the variability of MP presents diferent dynamics in survivors and in nonsurvivors, and these results persist even after adjustment for potentially confounding variables. Te hospital mortality rate in this cohort was consistent with that presented in other studies for the same patients with moderate-to-severe CARDS under MV [11,12].
Te association between MP and mortality in acute respiratory failure due to COVID-19 was also observed in a secondary analysis of the PRoVENT-COVID study, where MP was independently associated with mortality at 28 days [13]. Te mean MP in our study is comparable to another study in the feld [14], and, despite the lack of a universally accepted MP threshold to guide the proper use of MV, lower levels than those found in our study are already consistently associated with increased mortality [15,16]. Furthermore, MP values tend to be higher in CARDS than in ARDS due to other etiologies [17], possibly because patients with CARDS request high ventilatory demands to maintain acceptable PaCO 2 and pH, which require higher Vt and RR. Te physiological impact generated by PP may suggest an association between the maneuver and its impact on MP. Its efect on the recruitment of vertebral parts of the lung, making the ventilation distribution more homogeneous [3], may dampen lung stress and strain due to an increase in the surface of the lung that is capable of accommodating energy transfer [8]. However, few data explored the efect of PP on MP in patients with CARDS, with variability trends similar to those of our study [18]. We also observed a diference in variability between the three moments of measurement in the PaO 2 /FiO 2 ratio in survivors and nonsurvivors. Tere was also a trend of longitudinal increase in the MP during the PP strategy when comparing the values obtained during the frst and last prone positions in relation to the baseline value, with a greater increase detected in nonsurvivors. Tis fnding may suggest that not only the level of MP should have an impact on the outcome but also that the duration of parenchymal exposure to it can cause additional lung injury [19]. Although the results of our work may demonstrate that the magnitude of the improvement in oxygenation due to PP may be related to better results, we consider that the association between the improvement in the PaO 2 /FiO 2 ratio and improved outcomes is an issue that needs to be clarifed in further studies. In ARDS patients, while Gattinoni et al. observed that the "PaO 2 responders" (those who increased PaO 2 /FiO 2 by 20 mmHg) had an outcome similar to that of nonresponders [20], Scaramuzzo et al. demonstrate that a sustained improvement in oxygenation of PP after resupination would be associated with improved clinical outcomes [5]. Our study points to an association between high MP values and hospital mortality in severe forms of COVID-19. However, patients with COVID-19 have a high incidence of long-term pulmonary alterations in survivors, with a relevant impact on the quality of life of this population [21]. In further studies, it will be relevant to evaluate the impact of diferent ventilatory mechanical variables on the long-term outcomes of this population.
Tere are several limitations to our study. Due to the observational nature of the study, therapeutic assistance cannot be standardized. Te patients were preferentially ventilated in VCV mode. All patients received continuous infusions of neuromuscular blockade drugs, and the infusions of sedative and analgesic drugs were titrated by BIS monitoring. Tere has not been a uniform PEEP titration strategy during the prone maneuver, despite the fact that PEEP has variable responses in patients with COVID-19. In a previous study, the shunt fraction, alveolar dead space, and ventilation/perfusion matching were not afected by PEEP [22]. However, a potential impact of PEEP titration on MP cannot be assessed, as suggested in a previous study [9]. Our PEEP settings, however, are similar to other work in this feld [23,24]. Te observed PBW values for Vt/kg were slightly higher than the recommended 6 ml/kg [25]. However, we do not believe that this fact has an important infuence on the results, considering that (1) there were no statistically signifcant diferences between survivors and nonsurvivors; (2) other LPV measurements were respected, including an average ΔP < 15 cm H 2 O in both groups; and (3) as already described above, patients with CARDS tend to experience relatively good lung compliance and, according to some references, larger tidal volumes (7-8 ml/kg of PBW) without worsening the risk of VILI [26].

Conclusions
MP appears to comprise adequate respiratory monitoring to provide more personalized and adaptive ventilatory support. More prospective trials are needed to test whether this strategy is capable of improving the outcome of mortality.

Data Availability
Te data supporting the fndings of the current study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.