How to approach COVID-19 respiratory failure: A pandemic is not the time to abandon scientific principles

The COVID pandemic has elicited many different takes on the disease entity and not least on how to manage ventilatory support. We bring this position from a group of Scandinavian research intensivists as a guest post. – Ed MA/SV

When faced with a surge of patients with a new disease clinicians seek to increase their understanding of the disease by looking for clinical patterns. With the ambition to improve patient care, these observations are used to nuance current management patterns.

During the current COVID-19 pandemic a plethora of treatment regimes have been presented – awake prone positioning, early vs late vs no intubation, NIV vs no NIV, HFNO vs no HFNO, split ventilators, steroids, hydroxychloroquine, tocilizumab, remdesivir and higher than normal dose DVT/LE prohyylaxis just to mention a few examples. Due to the overwhelming clinical burden during the pandemic it can be tempting and intriguing to try new and untested treatments and regimes.

We feel, however, that some attemps to describe and interpret a signal in often very limited data has reached a point where it’s not entirely unlike trying to find a hidden meaning in Rorschach ink blots.

We should remember that during a pandemic we have both an opportunity to help our patients with well tested therapies, but can also harm an unprecedented number with new, untested and potentially dangerous treatments. If a drug kills a virus in a petri dish, it does not necessarily improve patient centered outcomes. We need well designed randomized controlled trials (RCTs) to determine that. Ideally, the large number of patients presenting during a pandemic should provide ample opportunities for clinical research.

In the following text we will discuss a recent trending topic in the global intensive care community and how the discourse relates to present evidence based clinical principles and opportunities for learning during the pandemic.

It has recently been put forward that the respiratory failure of COVID-19 is not ARDS. This notion has been circulating widely on the social media as illustrated in the examples below:

These ideas are supported by two recent letters to the editor plus a JAMA Clinical Update by renowned ARDS researcher Gattinoni and his group[1-3]. It is put forward that COVID-19 can be divided into two distinct phenotypes, L and H, where traditional ARDS-management should only be applied in type H whereas it can be detrimental in type L. The conclusion of these papers is that COVID-19 patients do not generally have ARDS and thus, as a whole should be managed differently [1-3]

While it may be true that COVID-19 pneumonia is a special case of respiratory failure characterised by different subtypes that warrant special treatment, there is currently no clinical data available to support the claim, nor any trials to support the suggested management.

We believe it is questionable and potentially dangerous to change the current patient management of ARDS which is based on well-made RCTs including several hundreds of patients because of case series and clinical observations. We should keep in mind that many elegant, detailed and thought-through physiological reasonings never survived the merciless assessment in the form of an RCT.

Below we will try to address some of the most widespread claims used to support the notion that evidence from trials done on ARDS-patients do not apply to patients with COVID-19.

1) “COVID-19 is a distinct new disease entity, different from ARDS”

We disagree. ARDS is, by definition, a heterogeneous syndrome caused by different etiologies. It is obvious for any experienced clinician that a patient with staphylococcus aureus pneumonia differs from a patient with severe pancreatitis. Nevertheless, if they fulfill ARDS-criteria we still manage both as ARDS patients. That is because we know both patients would have been included in e.g. the PROSEVA [4] the ARMA trial [5], two trials that helped to form various international guidelines on ARDS management.

The heterogeneous nature of ARDS is further underpinned by the fact that only 45% of patients with clinical ARDS show the hallmark histopathological finding of diffuse alveolar damage on lung biopsy ​[6].

During the COVID-19 pandemic, doctors face large numbers of patients with a specific cause of ARDS that may behave differently from patients with other causes of ARDS. Still these patients would have been included in the ARDS-trials mentioned above. Given this, we believe that these patients shall be treated using established evidence based principle unless evidence from clinical trials provides new insights.

While doing so, we should strive to collect systematic data on as many patients as possible to make sure we enhance our understanding of the clinical phenotype we are treating, and we should test new strategies in well designed RCTs.

2) “COVID-19 patients have normal compliance and therefore do not have ARDS”

We disagree. The claim stems from the above mentioned letter to the editor, where Gattinoni et al reports a compliance of around 50 ml/cmH2O from measurements on 16 patients with COVID-19 [1]​. Even if we disregard the limitations of the low number of included patients in the letter, there are several issues with this claim that need to be addressed.

First, compliance measurements are not a part of the Berlin definition and were actually removed from the draft because they did not add any prognostic value to the present criteria constituting the ARDS definition ​[7].

Second, is a compliance of 50 ml/cmH2O in patients with ARDS abnormally high? The answer is no, not necessarily. Gattinoni himself reported in 2006 a compliance of 44 +/- 17 ml/cmH2O in an ARDS/ALI population [8].

The two previously mentioned large RCTs reported lower compliances. In PROSEVA, the static compliance was reported to be 36 +/- 23 ml/cmH2O in the intervention group at inclusion ​[4]. In ARMA, the median compliance was 0.5 ml/cmH2O/kg predicted body weight which corresponds to a compliance of 40 ml/cmH2O in an 80 kg male [5]. Though the numbers from these two trials indicate that there might be a difference between the trial populations and the COVID-19 patients Gattinoni et al. reports on, a further look at the results from the trials can cast some light on whether compliance is important for treatment outcome or not.

3) “Compliance is a determinant of response to standard ARDS treatment”

In PROSEVA, the application of prone positioning improved oxygenation and survival, but did not improve compliance as seen in the supplementary appendix, table S3 [4]. Therefore there is no evidence to support the claim that ‘Prone positioning of patients with relatively high compliance results in a modest benefit’ [1]. On the contrary, the evidence supports the use of prone positioning in patients fulfilling inclusion criteria from the PROSEVA trial.

In ARMA the application of low tidal volume ventilation on top of a fixed PEEP to FiO2 scale resulted in a reduction in mortality. From figure 2 in the original paper it can be seen that the effect of the intervention is greatest in the 3 lower quartiles of compliance whereas there is no difference in the highest quartile that corresponds to a compliance of 50-120 ml/cmH2O in a person with a predicted body weight of 80 ​[5].

This suggests that in patients with normal to high compliance, strict adherence to ARMA settings in terms of low tidal volume ventilation might not be mandatory, but without harm.

However there is no evidence to support the rather unclear statement that ‘High PEEP in a poorly recruitable lung tends to result in severe hemodynamic impairment and fluid retention’ ​[1], at least not if high PEEP is defined according to the traditional ARMA PEEP-to-FiO2 scale. If high PEEP is defined as the settings suggested in the open lung approach as tested in the ART trial ​[9] or the higher PEEP settings from ALVEOLI [10] there is nothing new to the fact that these interventions are at best minimally effective and at worst plainly harmful.

4) “Patient self-inflicted lung injury (P-SILI) not only exists, it is a driving factor in the disease progression and should influence our clinical decisions”

We believe this is unclear. In Gattinonis papers on COVID-19 the notion of P-SILI holds a central role in their recommendations for clinical management. P-SILI is suggested as a mechanism for the disease progression in COVID-19 pneumonia [1-3].

The authors mention negative inspiratory thoracic pressure as the leading cause of P-SILI, referencing a paper by Brochard et al. [11]. In the referenced paper, mainly experimental animal data and tangential evidence is cited. Furthermore, Brochard et al., although proponents of the P-SILI theory, explicitly state in this paper that the very existence of the phenomenon is still a hypothesis.

In line with the P-SILI hypothesis, early intubation, deep sedation and neuromuscular blockade is suggested as a key strategy to avoid lung injury and improve patient outcomes [3].

The theory of P-SILI is intriguing, but clinical trials supporting it are lacking. In the recent ROSE trial by the Prevention and Early Treatment of Acute Lung Injury (PETAL) Network, neuromuscular blockade in mechanically ventilated patients with ARDS was not superior to usual care [12]. The results of the ROSE trial increases the burden of proof on the proponents of the P-SILI hypothesis.

Why all this matters

Recent trending claims in the physician community could introduce the risk of unwarranted practice variation. This will in turn lead to missed opportunities to clinically apply sound evidence based practice principles within one of the most well studied fields in intensive care medicine. The importance of this must be emphasized since compliance with established, evidence based guidelines tend to improve outcomes and reduce costs [13,14]. The latter is especially important in the setting of a pandemic as reduced costs is synonymous with decreased resource utilization and consequently increases the opportunity to provide better care to more patients.

We also see a real risk of establishing new ‘standards of care’ without sound scientific evidence. Getting rid of wide-spread treatments without evidence to support them is much harder than introducing the treatments in the first place, as illustrated in the well-known examples of early goal directed therapy for sepsis [15-18] or the use of liberal blood transfusions for upper gastrointestinal bleeding [19].

This in turn risks diverting massive research efforts into disproving unfounded but widespread therapies, efforts that could instead have been used to systematically test promising new treatments in an efficient manner in the first place.

In summary, the idea that COVID-19 is a new disease to which well established treatment regimes for ARDS should not be applied, is a hypothesis that could be tested in RCTs. But the hypothesis should not be considered evidence to change current patient management.

Futher reading


  1. Gattinoni L, Coppola S, Cressoni M, Busana M, Chiumello D. Covid-19 Does Not Lead to a “Typical” Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med . March 2020. doi: 10.1164/rccm.202003-0817le
  2. Gattinoni L, Chiumello D, Caironi P, et al. COVID-19 pneumonia : different respiratory treatment for different phenotypes ? Intensive Care Med . 2020;(2020):1-6. doi: 10.1007/s00134-020-06033-2
  3. Marini JJ, Gattinoni L. Management of COVID-19 Respiratory Distress. 2020. doi: 10.1001/jama.2020.6825
  4. Guérin C, Reignier J, Richard JC, et al. Prone positioning in severe acute respiratory distress syndrome. N Engl J Med. 2013;368(23):2159-2168. doi: 10.1056/NEJMoa1214103
  5. Brower RG, Matthay MA, Morris A, Schoenfeld D, Thompson BT, Wheeler A. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med. 2000;342(18):1301-1308. doi: 10.1056/NEJM200005043421801
  6. Cardinal-Fernández P, Bajwa EK, Dominguez-Calvo A, Menéndez JM, Papazian L, Taylor Thompson B. The presence of diffuse alveolar damage on open lung biopsy is associated with mortality in patients with acute respiratory distress syndrome: A systematic review and meta-analysis. Chest . 2016;149(5):1155-1164. doi: 10.1016/j.chest.2016.02.635
  7. Ranieri VM, Rubenfeld GD, Thompson BT, et al. Acute respiratory distress syndrome: The Berlin definition. JAMA – J Am Med Assoc . 2012;307(23):2526-2533. doi: 10.1001/jama.2012.5669
  8. Gattinoni L, Caironi P, Cressoni M, et al. Lung Recruitment in Patients with the Acute Respiratory Distress Syndrome. N Engl J Med. 2006;354(17):1775-1786. doi: 10.1056/NEJMoa052052
  9. Cavalcanti AB, Suzumura ÉA, Laranjeira LN, et al. Effect of Lung Recruitment and Titrated Positive End-Expiratory Pressure (PEEP) vs Low PEEP on Mortality in Patients With Acute Respiratory Distress Syndrome. JAMA. 2017;318(14):1335. doi: 10.1001/jama.2017.14171
  10. Brower RG, Lanken PN, MacIntyre N, et al. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med. 2004;351(4):327-336+411. doi: 10.1056/NEJMoa032193
  11. Brochard L, Slutsky A, Pesenti A. Mechanical ventilation to minimize progression of lung injury in acute respiratory failure. Am J Respir Crit Care Med. 2017;195(4):438-442. doi: 10.1164/rccm.201605-1081CP
  12. Moss M, Huang DT, Brower RG, et al. Early neuromuscular blockade in the acute respiratory distress syndrome. N Engl J Med. 2019;380(21):1997-2008. doi: 10.1056/NEJMoa1901686
  13. Pun BT, Balas MC, Barnes-Daly MA, et al. Caring for Critically Ill Patients with the ABCDEF Bundle: Results of the ICU Liberation Collaborative in Over 15,000 Adults. Crit Care.ed. 2019;47(1):3-14. doi: 10.1097/CCM.0000000000003482
  14. Gao F, Melody T, Daniels DF, Giles S, Fox S. The impact of compliance with 6-hour and 24-hour sepsis bundles on hospital mortality in patients with severe sepsis: a prospective observational study. Crit Care. 2005;9(6):R764. doi: 10.1186/cc3909
  15. Peake SL, Delaney A, Bailey M, et al. Goal-directed resuscitation for patients with early septic shock. N Engl J Med. 2014;371(16):1496-1506. doi: 10.1056/NEJMoa1404380
  16. Yealy DM, Kellum JA, Huang DT, et al. A Randomized Trial of Protocol-Based Care for Early Septic Shock. N Engl J Med. 2014;370(18):1683-1693. doi: 10.1056/NEJMoa1401602
  17. Mouncey PR, Osborn TM, Power GS, et al. Trial of Early, Goal-Directed Resuscitation for Septic Shock. N Engl J Med. 2015. doi: 10.1056/NEJMoa1500896
  18. Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med. 2001;345(19):1368-1377. doi: 10.1056/NEJMoa010307
  19. Villanueva C, Colomo A, Bosch A, et al. Transfusion Strategies for Acute Upper Gastrointestinal Bleeding. N Engl J Med. 2013;368(1):11-21. doi: 10.1056/NEJMoa1211801
Spread the love
  • 130
Martin Schønemann-Lund
PhD-student on Respiratory Failure in the ICU. Resident in Anesthesiology and Intensive Care.
Living in Greater Copenhagen but forever lost his heart to the great forests of Northern Sweden.
Johan Olsson
Anesthesiology and Intensive Care Medicine resident in Northern Sweden. Passionate about learning how to do good in an intangible and uncertain world.
Johan Olsson on Twitter
Theis Skovsgaard Itenov
Resident anaesthesiologist dedicated to improving the care for the critically ill through evidence-based research including development of new diagnostics, improving understanding of physiology and, not least, through rigorous RCT testing of anything applied to humans.
Theis Skovsgaard Itenov on Twitter
Morten Bestle
Senior consultant intensivist, Ass Prof, PhD, EDIC. Research interests sepsis and ARF.
Notify of

This site uses Akismet to reduce spam. Learn how your comment data is processed.

newest most voted
Inline Feedbacks
View all comments
Frank Hansen

In ROSE trial the control had deeper sedation then NMB, id i revalidering it right, so i dont think P-SILI hypothesis is dead yet

Frank Hansen

supplemental material to rose study shows a deeper achieved level of sedation in the control group (average RASS of -2.3 to -2.7 over the first 48 hours) despite a goal level of sedation defined as 0 to -1, and 17% of control group patients required open label paralysis within the first 48 hours

Johan Olsson

In the ROSE trial, the individuals in the intervention group were more deeply sedated than the control subjects. A large proportion of patients in the control group were lightly sedated. As seen in the ROSE supplementary material, table S4, the intervention group was sedated to an average RASS of -4.8 during the first day, as compared to a RASS of -2.7 in the control group. Looking at the following days, the intervention group is more heavily sedated up until day 7. By day 2 the average RASS in the control group is closer to -2. Also, around 40% of the… Read more »

frank hansen

agree, and what i was saying, i just saying the control was in a relative deep sedation(or 60% of them) and they had NMB. don’t think the P-SILI debate is al don there. since sedation and NMB is the tx for dyssynchrony and p-SILI. But is a small step and give us bette way for more inside. all in all there no good good evidens on, in sedation in ARDS. the problem is we know LTVV gives more dyssynchrony, so sedation is can be need. ROSE just give me evidens that i don’t all ways need NMB. it also a… Read more »