Ventilation management in acute respiratory failure related to COVID-19 versus ARDS from another origin - a descriptive narrative review.

INTRODUCTION: It is uncertain whether ventilation in patients with acute respiratory failure related to coronavirus disease 2019 (COVID-19) differs from that in patients with acute respiratory distress syndrome (ARDS) from another origin. AREAS COVERED: We undertook two literature searches in PubMed to identify observational studies reporting on ventilation management--one in patients with acute respiratory failure related to COVID-19, and one in patients with ARDS from another origin. The searches identified 14 studies in patients with acute respiratory failure related to COVID-19, and 8 studies in patients with ARDS from another origin. EXPERT OPINION: In patients with acute respiratory failure related to COVID-19, ventilation management seems to be similar to that of patients with ARDS from another origin. The future lies in studies focused on personalized treatment of ARDS of all origins, including COVID-19.

Introduction

Acute respiratory distress syndrome (ARDS) is a life-threatening complication in critically ill patients with high mortality and morbidity [1]. Sepsis, systemic inflammation, and pneumonia are the most common causes of ARDS [2]. Since December 2020, ICUs worldwide are overwhelmed with patients with acute respiratory failure related to coronavirus disease 2019 (COVID-19). Initially, COVID-19 was seen as a ‘novel’ disease and the acute respiratory failure it can cause as a ‘new form’ of ARDS [3]. It is increasingly understood, however, that acute respiratory failure related to COVID-19 has many issues in common with ‘normal’ ARDS [4].

In ARDS, so–called lung-protective ventilation is strongly advocated to limit lung injury caused by the ventilator [5]. Key components of lung-protective ventilation include the limitation of volumes and pressures, prevention of atelectasis while avoiding overdistension, proper use of prone positioning, and possibly also the use of conservative oxygenation targets. In patients that remain severely hypoxic, additive rescue therapies for hypoxemia can be applied, including muscle paralysis through continuous infusion of neuromuscular blocking agents (NMBAs), inhaled nitric oxide (iNO), or extra–corporeal membrane oxygenation (ECMO). It is very likely that lung–protection and possibly also the use of the abovementioned additive rescue therapies are also important in patients with acute respiratory failure related to COVID-19.

We here describe and compare ventilation management in patients with acute respiratory failure related to COVID-19 to that in patients with ARDS from another origin. Following a succinct overview of the evidence for benefit of the four abovementioned key components of lung-protective ventilation and additive rescue therapies, we summarize ventilation management as reported in several cohorts of patients with acute respiratory failure related to COVID-19, and in preceding international or multicenter cohorts of patients with ARDS from another origin.

Lung-protective ventilation

Smaller volumes and lower pressures

Ventilation with a higher tidal volume (VT) [10 to 15 ml/kg predicted body weight (PBW)] was commonly used because use of a higher VT improved oxygenation. However, animal studies showed that ventilation with a higher VT causes disruption of lung structures and induced a local release of inflammatory mediators––a phenomenon frequently referred to as ‘volutrauma’.

The first randomized clinical trials comparing ventilation with a higher VT to ventilation with a lower VT failed to show profit of this strategy in patients with ARDS [6-8]. One landmark randomized clinical trial from the United States, named ‘ARMA’, unmistakably demonstrated benefit from ventilation with a lower VT [6 to 8 ml/kg PBW] when compared to ventilation with a higher VT [12 ml/kg PBW]––ventilation with a lower VT improved survival and shortened duration of ventilation in patients with ARDS [9]. These findings, confirmed in a metaanalysis that also included the preceding trials [10], brought about a revolutionary change in ventilation management worldwide.

Ventilation with a lower VT comes with lower airway pressures. Consequently, the abovementioned trials compared ventilation at different airway pressures. The extent to which the airway pressure needs to be reduced to prevent injury remains a matter of debate. Animal studies showed that high plateau pressure (Pplat) can cause lung injury, and a posthoc analysis of the abovementioned ARMA study suggested ventilation with a Pplat < 30 cm H2O to be safe [11].

Ventilation with a lower tidal volume is now considered standard of care, and VT of up to 8 ml/kg can be accepted in cases of severe dyspnea, e.g. in patients with a metabolic acidosis. It remains uncertain whether a further reduction of the Pplat, e.g. by an additional reduction of VT below 6 to 8 ml/kg PBW, can improve outcome of ARDS patients.

Fewer atelectasis and less overdistension

After its first use in clinical practice, positive end–expiratory pressure (PEEP) was mainly used to correct hypoxemia that was refractory to large increases in the fraction of inspired oxygen (FiO2). Animal studies showed that ventilation with higher PEEP can prevent or reduce repeated opening and closing of lung tissue––a phenomenon frequently referred to as ‘atelectrauma’. However, animal studies also showed that ventilation with higher PEEP can cause overdistension of healthy lung parts, which is increasingly recognized as harmful.

Three well–designed and sufficiently powered randomized clinical trials, named ‘ALVEOLI’, ‘LOVES’ and ‘EXPRESS’, failed to show benefit of ventilation with higher PEEP [13 to 15 cm H2O] compared to ventilation with lower PEEP [7 to 10 cm H2O] in patients with ARDS [12-14]. An individual patient data metaanalysis of these studies suggested benefit from ventilation with higher PEEP [15]––it was associated with a lower mortality in patients with moderate to severe ARDS. However, that same metaanalysis suggested harm from ventilation with higher PEEP in patients with mild ARDS, and one more recent randomized clinical trial, named ‘ART’, showed ventilation with higher PEEP combined with lung recruitment maneuvers to increase mortality in patients with severe ARDS, despite improvements in oxygenation and lung compliance [16]. One explanation for these findings could be that in some patients, benefit of ventilation with higher PEEP with respect to less atelectasis is nullified by coinciding overdistension.

Taken together, these findings suggest there is close balance between the positive and negative effects of ventilation with higher PEEP, and it remains debated whether PEEP should be titrated according to a higher PEEP–lower FiO2 table or a lower PEEP–higher FiO2 table in patients with ARDS.

Prone positioning

There are several reasons for why prone position can improve oxygenation, including a better match of ventilation with pulmonary blood flow, enhanced clearance of airway secretions, and recruitment of lung tissue due to a shifting weight of the mediastinum and the diaphragm. Experimental studies showed that prone positioning improves gas exchange, lung compliance and alveolar recruitment, and prevents lung injury.

Initial randomized clinical trials confirmed the improvement in oxygenation, but failed to show a better outcome with prone positioning [17-20]. Two metaanalyses, though, suggested an improved outcome with use of prone positioning in patients with severe ARDS [21,22]. One landmark randomized clinical trial from France, named ‘PROSEVA’, indisputably showed benefit from prone positioning, if applied early and for sufficiently long periods [23]––prone positioning improved survival in patients with ARDS.

Consequently, prone positioning is now part of standard care in patients with ARDS with severe hypoxemia refractory to high FiO2.

Oxygen targets

ARDS patients are by definition hypoxemic, requiring higher FiO2 than other ventilated critically ill patients. However, prolonged hyperoxia is also known to be toxic, causing pulmonary histopathological changes comparable to those seen in ARDS [24]. Two systematic reviews of retrospective studies performed in heterogenic cohorts of ventilated critically ill patients showed an association between hyperoxia and increased mortality [25,26]. A fairly large but nevertheless underpowered randomized clinical trial from Italy, named ‘OXYGEN–ICU’, showed a decreased mortality and a greater number of ventilator–free days in patients receiving conservative oxygenation therapy (arterial oxygen tension [PaO2] 9.3 to 13.3 kPa), compared to a liberal approach (PaO2 up to 20 kPa) [27]. These findings were supported by two metaanalyses of studies comparing various oxygenation strategies [28,29]. One randomized clinical trial from Australia, named ‘ICU–ROX’, comparing a strategy using a conservative oxygenation target (pulse oximetry [SpO2] < 90%) to one that uses a conventional target (SpO2 ≥ 97%) showed no difference in duration of ventilation or mortality [30]. It should be noted, though, that the conventional group was already treated with a low FiO2, maybe explaining the lack of difference in outcomes in this study. A recent randomized clinical trial from France, named ‘LOCO2ʹ was prematurely aborted due a higher mortality in the conservative group [31]. In this study, the oxygen targets were SpO2 88% to 92% versus > 96%.

Although the effects of an intermediate approach remain unclear, it seems justified to avoid liberal use of oxygen, i.e. to use conservative oxygenation targets during ventilation in patients with ARDS.

Additive rescue therapies

Muscle paralysis through continuous infusion of NMBAs can reduce oxygen consumption. This strategy may also prevent collapse of alveoli by improving patient–ventilator synchrony [32]. However, it may induce neuromuscular weakness [33] and requires deep sedation, which could also worsen outcome [34]. Of note, one recent RCT in patients with ARDS from another origin, named ‘ROSE’, failed to show benefit of early and continuous infusion of neuromuscular blockers [35].

The potent selective pulmonary vasodilator, iNO, can be used to improve oxygenation by redirecting blood flow toward well–aerated regions of the lungs, reducing ventilation–perfusion mismatch [36]. Additionally, this may also reduce pulmonary hypertension [37]. It should be noted, though, that one systematic review of 13 RCTs in patients with ARDS from another origin showed no survival benefit despite transient improvements in oxygenation [38].

ECMO provides oxygenation and CO2 removal. Disadvantages include the risk of hemorrhage due to systemic anticoagulation and complexity of care, making it less available in non–specialized centers. In one RCT in patients with ARDS from another origin, named ‘CESAR’, referral to centers specialized in ECMO (not necessarily meaning that all patients continued with ECMO) compared to continuation of conventional ventilation in a non–specialized center resulted in lower mortality rates [39]. In one more recent RCT in patients with ARDS form another origin, named ‘EOLIA’, ECMO did not increase survival [40], albeit that this study was terminated early for futility, and a Bayesian analysis of this study showed ECMO to have a strong potential to increase survival [41].

Methods

Search strategy

To compile evidence on current ventilation management in patients with acute respiratory failure related to COVID-19, and in patients with ARDS from another origin, we performed two comprehensive literature searches in PubMed.

For the search focusing on ventilation in patients with ARDS from another origin, we used keywords and MeSH terms for ‘epidemiology’ AND ‘intensive care’ AND ‘mechanical ventilation’ AND ‘acute respiratory distress syndrome’ AND ‘tidal volume’ OR ‘plateau pressure’ OR ‘positive end expiratory pressure’ OR ‘prone positioning’ OR ‘oxygenation’ OR ‘neuromuscular blockage’ OR ‘inhaled nitric oxide’ OR ‘ECMO’––selection of articles was restricted to those published after 2012, as the Berlin definition for ARDS had not been defined before then [42]. For the search focusing on ventilation in patients with acute respiratory failure related to COVID-19, keywords and MeSH terms for ‘acute respiratory distress syndrome’ were interchanged with those for ‘coronavirus disease 2019ʹ––selection of articles of this search was restricted to those published after March 2020.

Screening of abstracts and titles identified by our search methods were performed independently by two authors. If a study was considered potentially eligible, full text was obtained. The full text was reviewed for selection according to inclusion and exclusion criteria. Inclusion criteria for both searches were: (1) reporting on invasively ventilated critically ill patients with ARDS, or acute respiratory failure related to COVID-19; (2) reporting on at least one of the four key components of lung-protective ventilation, i.e. VT or Pplat, PEEP, prone positioning or FiO2, or additive rescue therapies, i.e. use of NMBA, iNO or ECMO.

We excluded randomized clinical trials, reviews, research letters, and case reports. In addition, studies not using the Berlin definition, and reports on highly specific populations, such as pediatric patients, one–lung ventilation, etc. were excluded. We restricted this review to studies with at least 500 invasively ventilated ARDS patients, and to studies with at least 100 invasively ventilated patients with acute respiratory failure related to COVID-19.

Reporting

Data on key components of lung-protective ventilation and additive rescue therapies were extracted per study. Findings were reported for each study and, depending on reporting and completeness, existing of proportions of patients receiving the lung-protective intervention, defined as ventilation using lower VT or airway pressures, and median or mean VT and Pplat, the median or mean PEEP, the proportion of patients receiving prone positioning, the median or mean FiO2, and the proportion of patients receiving NMBAs, iNO or ECMO, if reported.

Medians were reported with interquartile ranges, means with 95%–confidence intervals or standard deviations in the tables; in the text we reported medians or means where appropriate. We neither metaanalysed the data nor did we use statistical tests to compare ventilation management in patients with acute respiratory failure related to COVID-19 versus in patients with ARDS from another origin.

Results

Search results

The two searches identified 826 articles on ventilation management in patients with ARDS from another origin, and 216 articles on ventilation management in patients with acute respiratory failure related to COVID-19. Reasons for exclusion of articles are shown in Figure 1. We included 8 articles of ventilation in patients with ARDS from another origin [43-50], and 14 articles of ventilation in patients with acute respiratory failure related to COVID-19 [51-64]. Characteristics of the reported studies and the described cohorts are provided in Table 1 and Table 2. Articles of patients with ARDS from another origin were comprised of cohorts from > 50 countries––articles of patients with acute respiratory failure related to COVID-19 included national cohorts from 8 different countries.

Figure 1.

Search results

Table 1.

Characteristics of ‘ARDS of another origin’ studies

Name/study acronym	Ref.	Design	Location	N	Berlin classification	Ventilation characteristics	Reported as
APRONET	43	Pro	Belgium, France and Switzerland	735	25.4%, 54.3% and 20.3% with mild, moderate or severe ARDS	VT, Pplat, PEEP, FiO2, prone positioning, duration	median (IQR), %
Gupta et al.	44	Pro	Peru	514	not reported	VT, Pplat, PEEP, FiO2	mean (SD), proportions ≤ 6 ml/kg PBW
Lanspa et al.	45	Retro	US	1385	24.5%, 48.7% and 26.8% with mild, moderate or severe ARDS	VT, Pplat, PEEP, FiO2	median (IQR)
Leite et al.	46	Retro	US	1142	not reported	VT, Pplat, PEEP	median (IQR), mean (SD)
LOTUS–FRUIT	47	Pro	US	895	not reported	VT	mean (95%CI), proportions ≤ 8 ml/kg PBW
LUNG–SAFE	48	Pro	worldwide study	2377	30.0%, 46.6% and 23.4% with mild, moderate or severe ARDS	VT, Pplat, PEEP, FiO2, prone positioning	mean (95%CI), proportions ≤ 8 ml/kg PBW
Parhar et al.	49	Pro	Canada	633	31.6%, 54.2% and 14.2% with mild, moderate or severe ARDS	VT, Pplat, PEEP	median (IQR), proportions ≤ 8 ml/kg PBW
Schmidt et al.	50	Retro	US	543	not reported	VT, Pplat, PEEP	median (IQR)

Abbreviations: ARDS = acute respiratory distress syndrome; COVID-19 = coronavirus disease 2019; CI = confidence interval; FiO2 = fraction of inspired oxygen; IQR = interquartile range; PBW = predicted body weight; Pplat = Plateau pressure; PEEP = Positive end-expiratory pressure; Pro = prospective; Retro = retrospective, SD = standard deviation; UK = United Kingdom; US = United States; VT = Tidal volume

Table 2.

Characteristics of ‘acute respiratory failure in COVID-19ʹ studies

Name/study acronym	Ref.	Design	Location	N	Berlin classification	Ventilation characteristics	Reported as
Camporota et al.	51	Retro	UK	213	10.8%, 57.3% and 31.9% with mild, moderate or severe ARDS	VT, PEEP	mean (95%CI)
Chand et al.	52	NA	US	274	4.3%, 35.3% and 52.0% with mild, moderate or severe ARDS	prone positioning	%
COVID–ICU Network	53	Pro	France	2635	24%, 52% and 24% with mild, moderate or severe ARDS	VT, Pplat, PEEP, prone positioning	median (IQR), %
Cummings et al.	54	Pro	US	203	Berlin definition not used	VT, Pplat, PEEP, FiO2, prone positioning	median (IQR), %
Ferrando et al.	55	Pro	Spain	742	17.1%, 44.6% and 38.1% with mild, moderate or severe ARDS	VT, Pplat, PEEP, FiO2, prone positioning	median (IQR), proportions ≤ 6 ml/kg PBW, %
Fusina et al.	56	Retro	Italy	187	not reported	VT, PEEP, FiO2	median (IQR), mean (SD)
Grasselli et al.	57	Retro	Italy	1150	Berlin definition not used	PEEP, FiO2, prone positioning	median (IQR), %
Grasselli et al.	58	Retro	Italy	2929	Berlin definition not used	PEEP, FiO2	median (IQR)
McWilliams et al.	59	Pro	UK	110	Berlin definition not used	prone positioning	%
Primmaz et al.	60	Pro	Switzerland	124	Berlin definition not used	PEEP, FiO2, prone positioning	median (IQR), %
PRoVENT–COVID	61	Retro	Netherlands	553	24.9%, 66.5% and 8.6% with mild, moderate or severe ARDS	VT, PEEP, FiO2, prone positioning, duration	median (IQR), proportion ≤ 6 ml/kg PBW, %
Roedl et al.	62	Retro	Germany	167	4%, 42% and 49% with mild, moderate or severe ARDS	PEEP, prone positioning	median (IQR), %
Schenck et al.	63	NA	US	267	not reported	VT, Pplat, PEEP, prone positioning	median (IQR), %
Thomson et al.	64	Pro	UK	136	Berlin definition not used	PEEP, prone positioning	median (IQR), %

Abbreviations: ARDS = acute respiratory distress syndrome; COVID-19 = coronavirus disease 2019; CI = confidence interval; FiO2 = fraction of inspired oxygen; IQR = interquartile range; PBW = predicted body weight; Pplat = Plateau pressure; PEEP = Positive end-expiratory pressure; Pro = prospective; Retro = retrospective, SD = standard deviation; UK = United Kingdom; US = United States; VT = Tidal volume

Tidal volumes and plateau pressures

In studies in patients with ARDS from another origin, median VT varied from 6.7 to 8.4 ml/kg PBW––mean VT from 7.1 to 8.7 ml/kg PBW (Table 1 and Figure 2). In the worldwide ‘LUNG SAFE’ study, almost two–thirds of patients received lung-protective ventilation with VT of ≤ 8 ml/kg PBW [48]. In one study from the United States, a vast majority of patients received lung-protective ventilation with a VT ≤ 8 ml/kg PBW [47]. In one study from Canada, less than half of patients received a VT ≤ 8 ml/kg PBW [49]. In one study from Peru, only a fraction of patients received ventilation with a VT of ≤ 6 ml/kg PBW [44]. In studies in patients with acute respiratory failure related to COVID-19, median VT varied from 5.8 to 7.0 ml/kg PBW––mean VT was ~7 ml/kg PBW in one study from the United Kingdom [51] (Table 2 and Figure 2). In one study from the Netherlands, named ‘PRoVENT–COVID’, more than half of patients received ventilation with a VT of ≤ 8 ml/kg PBW [61]. In the study from Spain, more than three–quarters of patients received ventilation with a VT of ≤ 8 ml/kg PBW [55].

Figure 2.

Summary of ventilation settings in patients with acute respiratory failure related to COVID-19 versus in patients with ARDS from another origin. references are provided between square brackets. green fonts: median or mean VT ≤ 8 ml/kg PBW, PEEP ≤ 10 cm H2O, FiO2 ≤ 60%, and prone ventilation use ≥ 15% – red fonts: median or mean VT > 8 ml/kg PBW, PEEP > 10 cm H2O, FiO2 > 60%, and prone ventilation use < 15% (cutoffs are arbitrarily chosen)

In studies in patients with ARDS from another origin, median Pplat varied from 19 to 26 cm H2O––mean Pplat from 23 to 27 cm H2O (Table 1). In the worldwide ‘LUNG SAFE’ study, two–thirds of patients received ventilation with Pplat ≤ 30 cm H2O in combination with a lower VT [48]. In another study, four–fifths of patients received ventilation with Pplat ≤ 30 cm H2O [49]. In studies in patients with acute respiratory failure related to COVID-19, median Pplat varied from 24 to 27 cm H2O (Table 2). In one study that reported Pplat per oxygenation category, Pplat was slightly higher in patients with the most severe oxygenation problems [53].

Positive end–expiratory pressure

In studies in patients with ARDS from another origin, median PEEP varied from 7.5 to 10.0 cm H2O––mean PEEP from 7.6 to 10.1 cm H2O (Table 1 and Figure 2). In the ‘LUNG SAFE’ study, a vast majority of patients received a PEEP < 12 cm H2O [48]. In studies in patients with acute respiratory failure related to COVID-19, median PEEP was from 10 to 15 cm H2O, mean from 10 to 11 cm H2O (Table 2 and Figure 2). In one study from Spain, 6.4% of patients received ventilated with PEEP > 12 cm H2O [55], while in one study from France, this proportion was 54% [53]. In the ‘PRoVENT–COVID’ study from the Netherlands [61] and one study from Italy [57], PEEP was reported to vary substantially between patients––from 5 and 20 cm H2O, and from 4 and 22 cm H2O, respectively.

In studies in patients with ARDS from another origin, prone positioning was used from 7.9% to 13.7% (Table 1 and Figure 2). In the ‘LUNG SAFE’ study, prone positioning was used in 1%, 5.5% and 16.3%, in patients with mild, moderate and severe ARDS, respectively, [48]. Duration of proning in patients with ARDS, thus far only reported in one study, named ‘APRONET’, was median 8 hours [43]. In studies in patients with acute respiratory failure related to COVID-19, prone positioning was used much more often, in up to three–quarters of patients (Table 2 and Figure 2). Clear differences in use of prone positioning between ARDS severity groups was seen in studies from Spain and from France [53,55], with five in every six patients receiving this intervention when oxygenation problems were severe [55]. In one study that reported on duration of prone positioning, sessions lasted median 13 hours [61].

Oxygen

In studies in patients with ARDS from another origin, median FiO2 varied from 45% to 50%, mean from 54% to 65% (Table 1 and Figure 2). Much higher FiO2 was used in studies in patients with acute respiratory failure related to COVID-19, median varying from 60 to 80%, mean 60% (Table 2 and Figure 2). In one study from France, up to three–quarters of patients needed FiO2 ≥ 50% [53]. In one Italian study, a vast majority of patients received ventilation with FiO2 ≥ 50% [57] and in one study for the United States, reported median of highest FiO2 was even 100% in the first day of ventilation [54].

In studies in patients with ARDS from another origin, the proportion of patients receiving continuous infusion of NMBAs ranged from 21.7% to 31.8%, use of iNO ranged from 4.7% to 7.8%, and the incidence of ECMO was between 0.5% and 4.8%. In studies in patients with acute respiratory failure related to COVID-19, the proportion of patients receiving NMBAs varied between 22.2% and 90%, the use of iNO ranged from 10.8% to 19.1%, and the incidence of ECMO was between 0.6% and 16.1%.

Discussion

Summary of findings

The findings of this review can be summarized as follows: (1) lower VT are more rigorously used in patients with acute respiratory failure related to COVID-19 than in patients with ARDS from another origin; (2) PEEP is higher in patients with acute respiratory failure related to COVID-19, and there is substantial variation in PEEP between patients within studies; (3) patients with acute respiratory failure related to COVID-19 receive prone positioning much more often than patients with ARDS from another origin; and (4) FiO2 is higher in patients with acute respiratory failure related to COVID-19 than in patients with ARDS from another origin.

Strengths of this review

This review has several strengths. It used two systematic searches to identify articles reporting on ventilation management. The searches identified articles originating in various countries worldwide, were all of sufficient size. The cohorts reporting on COVID-19 patients covered practice over a relative short period of time. Studies reporting on ventilation management in patients with ARDS from another origin were only included if the Berlin definition for ARDS [42] was used, increasing homogeneity of those cohorts. All studies were rich in detail, especially with regard to the four key components of lung-protective ventilation.

Ventilation with a lower tidal volume

Ventilation with a lower VT is applied more frequent in patients with acute respiratory failure related to COVID-19 than in patients with ARDS from another origin. This could have several reasons. First, it is possible that this component of lung-protective ventilation is more easily applied in acute respiratory failure related to COVID-19. This may not be the most obvious reason, however, since acute respiratory failure related to COVID-19 has much in common with ARDS of another origin [65,66]. Second, it could be that ICUs were well prepared, using local guidelines designed especially for the COVID-19 outbreak. Unfortunately, we had no access to the local guidelines, neither from centers that participated in the various studies of ventilation management in patients with ARDS from another origin, nor from those in studies of ventilation management in acute respiratory failure related to COVID-19. Third, it could be that compliance with existing guidelines was much better in patients with acute respiratory failure related to COVID-19, which can be explained in two ways. It is possible that because care for a surge of patients with COVID-19 had to be provided by hospital personnel who had less experience or confidence with setting a ventilator, the compliance with existing guidelines for the ventilation of patients with ARDS increased; it is also possible that use of ventilation with lower VT improved overall over recent years––new service reviews in patients with ARDS from another origin may reveal a better overall compliance with recommendations on use of a lower VT.

Use of PEEP

The debate on what PEEP to use in patients with ARDS is ongoing, even after publication of studies that failed to show benefit [12-14], or even demonstrated harm from ventilation with higher PEEP [16]. Consequently, it is also uncertain how much PEEP to use in COVID-19 patients. In absence of randomized clinical trial evidence, it has been recommended to use ‘higher’ PEEP, following the higher PEEP/lower FiO2 table of the ARDS Network [67-69]. The substantial variation in PEEP between patients with acute respiratory failure related to COVID-19 within the studies could reflect the uncertainty of physicians in titration of PEEP. Of note, how PEEP was titrated, and whether caregivers indeed used a higher PEEP–lower FiO2 table or a lower PEEP–higher FiO2 table was not collected in the studies. Although it has been suggested that there could be two different phenotypes of acute respiratory failure related to COVID-19 [3], possibly needing different levels of PEEP, several studies failed to confirm this [66,70-72]. The use of high PEEP could also have been caused by a desperate attempt to normalize the at times severe hypoxemia, in a time when mortality from severe COVID-19 was reported to be as high as 80% [73] .

Use of prone positioning

In patients with respiratory failure due to COVID-19, use of prone positioning was remarkably higher than in patients with ARDS from another origin. Alike the better use of a lower VT, this could reflect an improvement in ventilation care over the recent years. Interestingly enough, limited staffing and resources during the pandemic did not have an effect on use of prone positioning, although this was reported as the reason for the somewhat lower use in one study [54]. Probably, the common use of prone positioning is due to the high incidence of severe hypoxemia in patients with COVID-19 [51,57]. Not only has prone positioning been indisputably shown to improve oxygenation and mortality in ARDS [23], it has been suggested to be a safer method of improving oxygenation than use of PEEP [74]. The posthoc analysis of one recent randomized clinical trial from France, named ‘LIVE’, that compared standard of care to personalized ventilation consisting of either lower PEEP and early prone positioning in non–recruitable ARDS to higher PEEP and prone positioning as a rescue intervention in recruitable ARDS, showed that if patients received the ‘wrong’ treatment, i.e. higher PEEP and prone positioning in non–recruitable ARDS or lower PEEP and early prone positioning in recruitable ARDS, outcome worsened. If COVID-19 behaves like non–recruitable ARDS, prone ventilation could be much more effective than use of higher PEEP [75].

Titration of FiO2

Not surprisingly, FiO2 was higher in patients with acute respiratory failure related to COVID-19, most likely reflecting the higher incidence of severe hypoxemia in these patients. The use of high FiO2 supports the idea that COVID-19 patients do not respond well to increases in PEEP and could thus be classified as having non–recruitable ARDS. However, lung infiltrates are not the only contributor to hypoxemia in COVID-19 patients––coagulopathy is frequently observed in these patients [61,64] and could be a major determinant of the severity of hypoxemia. In a study comparing computed tomography pulmonary angiography between non–ventilated and ventilated COVID-19 patients, both with clinical suspicion of pulmonary embolism, ventilated patients showed much larger areas of scattered perfusion defects bilaterally [76]. The pattern in ventilated patients was independent of pulmonary embolism, suggestive of microcirculation dysfunction. Impeded perfusion in normally aerated lung tissue leads to increased dead space ventilation and shunting to diseased areas of the lung, which has deleterious effects. The matter of recruitability may be less relevant in these patients.

The incidence of continuous infusion of NMBAs in the cohorts of COVID-19 patients was up to threefold higher than in cohorts of patients with ARDS from another origin. This could have multiple reasons. First, it is not uncommon for physicians to consider muscle paralyses to be mandatory in patients that are positioned prone, and often this is continued until a patient is turned back to supine––since prone positioning is applied often in COVID-19 patients due to severe hypoxemia, it is not surprising that with this increase, use of muscle paralysis increased as well. Second, NMBAs might have been used to combat the fits of coughing, frequently seen in COVID-19 patients, and at times resulting in deep desaturations. Third, the high incidence of severe hypoxemia could have triggered the use of NMBAs in an attempt to lower oxygen consumption. The severity of hypoxemia, and the urge to act on it, probably also clarifies the high use of iNO and ECMO in COVID-19 patients, which was up to fourfold higher than in patients with ARDS of another origin.

Also, there was remarkable variability in the use of additive therapies in the cohorts of COVID-19 patients. This is not surprising, as recommendations for use of NMBAs and iNO were at best contradictory before the COVID-19 pandemic. Use of these additive rescue therapies is probably at the discretion of attending physicians, and less dictated by local guidelines than the key components of lung-protective ventilation. Indication for use of ECMO before the COVID-19 pandemic may have been better defined, but use of ECMO could depend on available resources, workloads, admission capacities and also transfer possibilities.

Respiratory mechanics in acute respiratory failure in COVID-19 vs. ARDS of another origin

Although respiratory failure in COVID-19 was recognized as, and even named ‘ARDS’ fairly early in the pandemic, it was suggested that ARDS in COVID-19 patients had different characteristics than ARDS from another origin. There is increasing evidence against this. For instance, one study reported on lung physiology and recruitability in 30 invasively ventilated COVID-19 patients compared to 30 matched ARDS patients from an existing database [77]. Although compliance was slightly higher in COVID-19 patients compared to patients with ARDS from another origin, heterogeneity of respiratory mechanics was comparable. COVID-19 patients showed a varied, but overall greater response in oxygenation and respiratory mechanics in response to increase of PEEP than ARDS from another origin, but the response was independent of recruitability, which was similar in the two groups. Although the sample of this study was limited, these results coincide with other studies showing large variability in respiratory mechanics and responses to PEEP [66,78]. These findings support the idea that ventilation management in COVID-19 patients should follow the recommendations for patients with ARDS from another origin.

Limitations

This narrative review has limitations. First, despite the observational character of the included studies, it could be that daily practice was affected by knowledge of data collection in some studies. Second, the studies reporting on ventilation management in patients with acute respiratory failure related to COVID-19 originated in only 8 countries––all of which included centers with no limitations in resources. It remains uncertain if ventilation management in resource-limited settings differs from that in centers in high-income countries. Third, we excluded studies reporting only on ‘special’ cohorts, like patients with previous lung conditions, children, and pregnant patients, and special treatments, like ECMO. Fourth, while use of the Berlin definition for ARDS increased homogeneity in cohorts of patients with ARDS from another origin, we could not use this definition for selection of studies of ventilation management in COVID-19 patients––in fact, most studies reported on patients receiving ventilation for ‘severe COVID-19ʹ or ‘COVID-19 pneumonia’. Nevertheless, in those reports that explicitly mentioned the proportions of COVID-19 patients with ARDS according to the Berlin definition, nearly all patients fulfilled the Berlin definition for ARDS. Fifth, heart–lung interactions, if present, were not reported in the studies identified by the two literature searches. Heart–lung interactions, however, do exist and may affect ventilation management. Absence of information in study reports may reflect how difficult it is to collect data on heart–lung interactions, or worse, that heart–lung interactions are inadequately used in ventilator management. Sixth, in the studies identified by the literature searches, occurrence of barotrauma was not routinely collected and may thus have been underreported. The incidence of barotrauma, including pneumothorax and pneumomediastinum, has declined after the implementation of lung-protective ventilation strategies. If it happens, however, it is an important adverse event of invasive ventilation in ARDS patients [79]. Seventh, the data extracted from the studies was heterogeneous and therefore difficult to compare. We did not perform any analysis for this reason and only described the data, making the findings in this review only suggestive.

Conclusion

In patients with acute respiratory failure related to COVID-19, ventilation with a lower VT and prone positioning is applied more rigorously than in patients with ARDS from another origin. Applied positive end–expiratory pressures and oxygen fractions are higher in patients with acute respiratory failure related to COVID-19. It is likely that lung protection is of equal importance in COVID-19 patients as it is in patients with ARDS from another origin.

Expert opinion

Mechanical ventilation has moved from a ‘simple concept of “in goes the good air, out goes the bad air” into a sophisticated life support system in common use today’, and from a ‘technology to “normalize” physiology as much as possible with little regard for harm beyond overt barotrauma and cardiac compromise’ to one that follows the ‘principles of lung-protective ventilation management as a cornerstone of management’––well described in a recent historic overview of ventilation from the 1970 to 2020 [80]. Indeed, the targets in ventilation in ARDS patients have shifted––oxygenation and decarboxylation are now secondary to preventing ventilation–induced injury. A large number of studies have been dedicated to perfecting lung-protective ventilation, or even bypassing the need for oxygenation from ventilation by techniques such as ECMO. In the end, the ultimate goal is to minimize the intensity of the treatment, i.e. the additional harm caused by ventilation. It is not yet clear how to achieve this exactly and individually, as every patient presents with their own clinical and biological characteristics, making generalization of treatment at times challenging.

When the world was overcome by the COVID-19 pandemic, COVID-19 was initially perceived as a ‘new disease’. This perception could have caused us to abandon current evidence–based ventilation strategies for other approaches, including strategies that have been shown not to benefit patients with ARDS from another origin. The findings of this review suggest, at least in part, that acute respiratory failure related to COVID-19 may not be too different from ARDS from another origin with respect to ventilation management. In addition, this summary suggest there are many similarities in respiratory mechanics between acute respiratory failure related to COVID-19 and ARDS from another origin. The actual question is now––are more studies needed, or do we accept that ventilation management in COVID-19 patients should be as that in patients with other forms of ARDS?

Probably, added information could come from research dedicated to the effects of ‘personalized’ ventilation, in all patients with ARDS, i.e. irrespective of its cause, thus ARDS from another origin or COVID-19. Research points to the implications of certain differences between individual patients for ventilation management, for instance whether or not ARDS has a ‘focal’ or a ‘non–focal’ appearance. The recent seminal randomized clinical trial, named ‘LIVE’, attempted to compare a strategy considered beneficial in patients with ‘focal’ ARDS with one that could be preferred in patients with ‘non–focal’ ARDS. Classification of patients was a challenge, but the study results suggest that if an incorrect strategy was chosen, i.e. a ‘focal’ approach in patients with ‘non–focal’ ARDS, or a ‘non–focal’ approach in patients with ‘focal’ ARDS, outcomes worsened dramatically. The findings of this study guide the way we could go. From a ‘one size fits all’ approach to one that uses certain information to individualize, and thereby improve ventilation management.

One important lesson remains that physiological improvements, like a better oxygenation, do not automatically translate in better outcomes, like duration of ventilation or death. Certainly, this also applies also for any personalized approach.