![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/04/1647-locandina-big.jpg?resize=1000%2C864&ssl=1)
In the past few weeks, there has been considerable debate regarding whether or not early COVID respiratory failure is ARDS. This is a linguistic mess, but it has real implications for treatment. So, it’s worth trying to sort it out…
definition(s) of ARDS
When clinicians use the term “ARDS” they are referring to a variety of different things. For the sake of maximal clarity, it may be possible to parse out seven different entities. Let’s walk through them…
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/04/color1.jpg?resize=660%2C495&ssl=1)
(1) Berlin definition of ARDS
This is the technically correct definition of ARDS, as follows:1
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2018/01/berlin.gif?resize=600%2C337&ssl=1)
In reality, the Berlin definition is horrifically broad and clinically useless.2 Nearly all intubated patients with acute bilateral parenchymal lung disease will meet this definition. For example, imagine that you just intubated a patient with alcohol intoxication who had been aspirating a bit. The patient has mild bilateral infiltrates with an oxygen saturation of 90% on 25% FiO2. Since an oxygen saturation of 90% correlates roughly with a pO2 of ~60 mm, this patient will have a P/F ratio (PaO2/FiO2 ratio) of ~240 – so they technically will have ARDS. However, I don’t think that most experienced clinicians would believe that a patient on 25% oxygen has “ARDS.”
Further evidence that the Berlin definition of ARDS is clinically meaningless is that it has never been used as a primary inclusion criteria for any clinical trial. Clinical trialists have been far more thoughtful and selective about who to include in their trials. If a trial were designed based on the Berlin definition, it would be hopelessly heterogeneous and doomed to failure.
So, intubated patients with COVID will meet the Berlin definition of ARDS. Absolutely. Unequivocally. But that has zero meaningful implications regarding how they should be treated.
(3) Lung Injury
Patients with a P/F ratio >150 have lung injury, but this isn’t generally an indication for more aggressive ARDS therapies (e.g. proning). Historically, acute lung injury was defined based on a P/F ratio of 200-300. However, more recent studies suggest that a cutoff of 150 may be more actionable. Regardless of the precise cutoff, patients at higher P/F ratios aren’t severely hypoxemic and don’t require more aggressive ARDS-specific management. Such patients should receive lung-protective ventilation and treatment of their underlying disease (but this is good management for any patient).
(2) Possible ARDS
Possible ARDS refers to patients who were recently intubated and now have a P/F ratio below 150. Some of these patients will improve within 12-24 hours with aggressive optimization. Others won't. Treatment of these patients should focus on management and exclusion of confounding features (e.g. pleural effusions, volume overload, atelectasis). Only after taking these items off the table, will it become clear what is going on.
(4) PseudoARDS (a.k.a. Rapidly Improving ARDS)
These are patients who improve quickly within 12-24 hours of intubation. The most common causes of pseudoARDS are derecruitment or heart failure, which may resolve merely with the institution of positive pressure ventilation (especially if higher PEEP or APRV is utilized). For example, Villar et al. found that among 170 patients who initially met the definition of moderate ARDS (P/F < 200), only 58% still met this definition after a day of ventilation on >10 cm PEEP:3
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2018/01/pfrat.gif?resize=550%2C661&ssl=1)
Identification of pseudoARDS is important, because these patients are unlikely to benefit from interventions such as proning or paralysis. The clinical course of pseudoARDS is often relatively benign.4
(5) PROSEVA definition of ARDS
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2017/11/guerrin.gif?resize=630%2C332&ssl=1)
Patients whose P/F ratio remains <150 despite 12-24 hours of ventilator optimization have more significant disease. This was the clinical entry criteria for the PROSEVA trial on proning.5 As such, the PROSEVA definition of ARDS may be the most evidence-based definition to use when selecting candidates for proning.
(6) DAD-ARDS (Diffuse Alveolar Damage ARDS)
These patients have histology showing diffuse alveolar damage (DAD) with hyaline membranes. This is the traditional pathological definition of ARDS that appears widely in textbooks. These patients have reduced compliance and are often difficult to wean from ventilation.
(7) Non-DAD ARDS
ARDS is a syndrome, not a specific diagnosis. As such, we are left here with a wastebasket full of all types of non-vascular lung disease with all sorts of histological and radiographic patterns (e.g. consolidative pneumonia, vasculitis, various interstitial lung diseases).
definition of recruitment
Recruitment is most generally the concept that positive pressure can be used to open up areas of the lung which have become collapsed. This is an extremely important concept, but unfortunately, it’s defined in a variety of different ways.
Different time frames may be utilized:
- Immediate recruitment: Increasing PEEP or performing a recruitment maneuver (i.e. transient use of very high pulmonary pressures) causes an immediate improvement in lung aeration within minutes.
- Gradual recruitment: Manipulation of PEEP or airway pressures causes an improvement over several hours.
Different endpoints may be utilized:
- Improvement in lung mechanics (e.g. calculated compliance based on plateau pressure, PEEP, and tidal volume). This can be performed only in a passive patient (e.g. deeply sedated or paralyzed) – which can limit its applicability to many patients.
- Improvement in oxygenation (e.g. P/F ratio).
My opinion is that any determination of recruitment based on short-term improvements is flawed (consistent with the repeated clinical failure of short-term recruitment maneuvers).6 Patients with severe atelectasis (e.g. lobar collapse) won’t improve in a few minutes. Several hours are generally required to gradually re-inflate the lung. Likewise, practitioners experienced with APRV will observe that improvements in oxygenation often occur over a timeframe of hours (not minutes).
The most evidence-based and actionable definition of recruitment might (once again) be hidden within the PROSEVA trial. Patients were selected for this trial only if they remained hypoxemic despite 12-24 hours of optimization on mechanical ventilation. This might be the most practically useful definition of recruitment: the ability of positive pressure ventilation to cause improvements in oxygenation over several hours.
Gattinoni et al. have reported that early COVID is not recruitable.7 This is based upon the following evidence, which may be incorrect:
- (1) Short-term measurements of lung recruitment in one study found low recruitability. These investigators evaluated the ability of the lung to recruit over a single breath – hardly enough time to truly see how the patient would respond to higher mean airway pressures.8 This may technically be accurate, but it overlooks the possibility that gradual recruitment may occur over longer time frames.
- (2) In a subsequent paper, Gattinoni et al. wrote “the amount of non-aerated tissue is very low, consequently the recruitability is low.”9 This seems to be based upon defining recruitability in terms of the volume of aerated lung tissue, rather than lung function. However, if areas of non-aerated tissue are functioning as physiological right-to-left shunts, then opening up these small lung areas could have a substantive effect on oxygenation. Thus, recruitability shouldn't be assessed from a single, static CT scan.10
Using a PROSEVA definition of recruitability (gradual improvement in oxygenation with increased airway pressure), I think that early COVID is actually recruitable:
- Continuous positive airway pressure (CPAP) via helmet interface has been generally regarded to be successful for COVID. The way CPAP works is largely via lung recruitment. If early COVID were truly not recruitable, then CPAP would fail clinically.
- My (albeit limited) experience with intubated COVID patients is that when placed on early APRV, they recruit very readily with drops in FiO2 within 6-12 hours. Anecdotally, I’ve heard from others about good responses to APRV. If early COVID weren't recruitable, then APRV wouldn’t work very well (since APRV is, fundamentally, a prolonged recruitment maneuver).
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/04/APRVcxr-scaled.jpg?resize=587%2C325&ssl=1)
so, where does COVID fit into this?
It was immediately evident to everyone managing these patients that early COVID doesn't cause textbook DAD-ARDS (since these patients have a fairly preserved compliance). This has caused confusion, since there is often a tendency to conflate all ARDS with DAD-ARDS.
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/04/color2.jpg?resize=626%2C467&ssl=1)
Early COVID seems to be pseudo-ARDS:
- Early application of high levels of mean airway pressure are often successful at increasing the P/F ratio to above 150.
- Lung compliance is normal (arguing against, for example, DAD-ARDS).
- There appears to be a substantial component of atelectasis. This may cause rapid deterioration (due to progressive atelectasis), but may also allow rapid recovery (when positive pressure is used to recruit collapsed alveoli). Atelectasis is often a component of the physiology of pseudoARDS.
Over time on a ventilator, ventilator-induced lung injury may lead to diffuse alveolar damage (DAD). Thus, some days or weeks into their disease course, patients may transition to a phenotype of DAD-ARDS (with poor compliance and more “typical” findings of ARDS).
practical implications for ventilating the COVID patient
OK, enough wrestling with definitions, what does this actually imply for clinical management of these patients?
COVID patients can be treated with a standard PROSEVA-based approach (as could any patients with possible ARDS)
Initial patient management can proceed similarly to that of any patient with possible ARDS, as shown below. My preference is often to use early APRV to definitively determine the amount of recruitable lung tissue. However, low tidal volume ventilation with a generous amount of PEEP will achieve the same thing (even the standard ARDSnet PEEP table may provide sufficient PEEP for most COVID patients).
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/04/approards.jpg?resize=536%2C394&ssl=1)
It seems that most early COVID patients will respond to initial optimization with an increase in P/F ratio >150. Such patients may be maintained on ventilation with an adequate amount of mean airway pressure to maintain recruitment. Once the disease has started resolving (e.g. with falling inflammatory markers, after a couple days have passed), more aggressive ventilator weaning can occur.
Low PEEP & high FiO2 strategies should probably be avoided
The concept of keeping PEEP values as low as possible has become popular, while using high FiO2 to maintain adequate oxygenation. However, there is little evidentiary or theoretical support for this concept. Gattinoni based this recommendation upon the concept that COVID lung isn’t recruitable, but this seems to be misguided. The early COVID lung can be recruited, if one is patient and willing to wait for 6-12 hours.
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/03/peep.gif?resize=395%2C407&ssl=1)
The use of very low PEEP levels is not generally supported by the ARDS literature. There is lively debate between standard PEEP vs. high PEEP. Likewise, there is debate between conventional ventilation versus APRV (with one RCT favoring APRV).11 All of these approaches seem to be reasonable, without definitive evidence of the superiority of any single technique. In contrast, there is little evidentiary support for a minimal-PEEP/maximal-FiO2 strategy.
Minimizing the PEEP and using high FiO2 could be dangerous:
- Using low PEEP would leave many parts of the lung partially atelectatic. This could lead to atelectotrauma as alveoli repeatedly pop open and closed during each respiratory cycle.
- High FiO2 can lead to oxygen toxicity. The precise cutoff when this occurs is unclear, but excessive FiO2 for prolonged time periods could cause harm.
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2015/12/bullet21.gif?resize=750%2C137&ssl=1)
- Like the term “sepsis,” different clinicians use the term “ARDS” to refer to different things. This is a source of ongoing confusion and consternation.
- The Berlin definition of ARDS is hopelessly broad and clinically almost useless. Whether or not a patient meets this definition has no direct implications for their clinical management.
- Based largely on the approach taken by the PROSEVA trial, seven sub-groups of ARDS are described (figure below). Most patients with early COVID appear to fall into the PseudoARDS category. Over time, ventilator-induced lung injury may cause a transition to DAD-ARDS phenotype.
- Clinicians define recruitability in different ways. Gattinoni et al. have popularized the concept that early COVID isn't recruitable, but this applies to short-term recruitment. When exposed to high mean airway pressures over a period of hours, patients with COVID seem to be recruitable.
- Currently, it may be safest to ventilate patients with COVID using traditional strategies used with ARDS (e.g. low tidal volume ventilation with standard or high PEEP, or APRV; all of which have demonstrated success in RCTs). The use of unusually low PEEP has not been validated and may increase risks of atelectotrauma.
![](https://i0.wp.com/emcrit.org/wp-content/uploads/2020/04/color1.jpg?resize=632%2C474&ssl=1)
related
- ARDS vs. pseudoARDS – Failure of the Berlin definition (PulmCrit, 2018)
references
- 1.ARDS Definition Task Force., Ranieri V, Rubenfeld G, et al. Acute respiratory distress syndrome: the Berlin Definition. JAMA. 2012;307(23):2526-2533. doi:10.1001/jama.2012.5669
- 2.Villar J, Pérez-Méndez L, Kacmarek R. The Berlin definition met our needs: no. Intensive Care Med. 2016;42(5):648-650. doi:10.1007/s00134-016-4242-6
- 3.Villar J, Pérez-Méndez L, López J, et al. An early PEEP/FIO2 trial identifies different degrees of lung injury in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med. 2007;176(8):795-804. doi:10.1164/rccm.200610-1534OC
- 4.Schenck E, Oromendia C, Torres L, Berlin D, Choi A, Siempos I. Rapidly Improving ARDS in Therapeutic Randomized Controlled Trials. Chest. 2019;155(3):474-482. doi:10.1016/j.chest.2018.09.031
- 5.Guérin C, Reignier J, Richard J-C, et al. Prone Positioning in Severe Acute Respiratory Distress Syndrome. N Engl J Med. June 2013:2159-2168. doi:10.1056/nejmoa1214103
- 6.Cavalcanti AB, Suzumura ÉA, 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. October 2017:1335. doi:10.1001/jama.2017.14171
- 7.Gattinoni L, Coppola S, Cressoni M, Busana M, Rossi S, 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
- 8.Pan C, Chen L, Lu C, et al. Lung Recruitability in SARS-CoV-2 Associated Acute Respiratory Distress Syndrome: A Single-center, Observational Study. Am J Respir Crit Care Med. March 2020. doi:10.1164/rccm.202003-0527LE
- 9.Gattinoni L, Chiumello D, Caironi P, et al. COVID-19 pneumonia: different respiratory treatments for different phenotypes? Intensive Care Med. April 2020. doi:10.1007/s00134-020-06033-2
- 10.Amato M, Santiago R. The Recruitability Paradox. Am J Respir Crit Care Med. 2016;193(11):1192-1195. doi:10.1164/rccm.201601-0178ED
- 11.Zhou Y, Jin X, Lv Y, et al. Early application of airway pressure release ventilation may reduce the duration of mechanical ventilation in acute respiratory distress syndrome. Intensive Care Med. 2017;43(11):1648-1659. doi:10.1007/s00134-017-4912-z
- PulmCrit Wee: Rational selection of infusion rate based on loading dose - June 25, 2024
- PulmCrit: PPIs are safe and effective for GI prophylaxis… the end. - June 18, 2024
- PulmCrit: Bilevel Sequence Intubation (BSI) – The new standard - June 17, 2024
Very nicely done examination of ARDS in the age of COVID-19, or as I like to call it, new and improved SARS.2 . I like having residents read your articles for clarity of description of APRV and COVID in general, despite your final clarifying statement after the graphic, “for the sake of maximum clarity, ARDS can be described 7 different ways”. But how did you resist labelling the chapter, “COVID-19: Not your DAD’s ARDS” ? All that aside, when trying to simplify ARDS for rural care with limited transfer, I have thought that the PROSEVA definition really helps, and advocated… Read more »
Sounds great. But unfortunately oxygenation is a very unreliabel marker of recruitment. Especially if you are using high airway pressures (e.g. APRV). Those high pressures tend to reduce cardiac output and therefore reduce right-left shunt which improves oxygenation (even without recruitment of a single alveoli).
Actually the proseva example for the effects of recruitment is a bad one – in this trial improvement of oxygention was not associated with improvment in survivial (https://www.atsjournals.org/doi/pdf/10.1164/rccm.201311-2056LE).
So I believe you can not use it to explain the beneficial effects of “lung recruitment” and improved oxygenation.
I’m not sure if APRV with that “high” mean airway pressure reduces the CO and the R-L shunting. Experts in TCAV postulate that the P High is aimed to achieve a lung volume around FRC, where the PVRs are the lowest possible and your right ventricle should do well his job. Moreover, the left heart is enjoying high positive intrapleural pressure by reducing the LV’s transmural pressure. You’ll indeed have to avoid hypovolemia. Thus, I see no reason to be concerned about hemodynamics. https://www.aprvnetwork.org/about/
Well, I think the lung with normal compliance is at FRC at a endexpiratory pressure of 0 (without PEEP ). If you distent the lung (and there is minimal recruitment) with a pressure of 30cmH20, the lung will be at 30mH20 x Compliance above FRC. You will see hemodynamic effects in that case (lung with minimal recruitment) for sure. It is also described in the IBCC chapter APRV. If spontaneous breathing occurs at a p-high of 30cmH20 (in a lung with normal compliance) the result can be dangerously high transpulmonary pressures. Improvement in oxygenation does not equal recruitment does not… Read more »
Thanks Josh for your dissection of ARDS definition and the PROSEVA-based approach.
We initially applied high PEEP with Helmet CPAP in our first COVID patients to obtain SpO2 95%. We noticed that most had a TAPSE reduction and low urine output. We therefore shifted to lower PEEP with a 90% SpO2 target avoiding this side effect. This may justify a lower PEEP initial application in COVID pts in the presumable L stage.
Another great post.
Thought this would be a good place to leave the fact that P/F is a terrible marker for us to be using. May be ok-ish for an initial marker of illness, but nothing after that.
Patient A: pao2 70 on 50% fio2 and peep of 5
Patient B: pao2 140 on 100% fio2 and peep of 15.
Same p/f, but I’m thinking of extubating A while B I think we’d agree is still pretty sick. Really think we can do better.
https://clinicaltrials.gov/ct2/show/NCT03946150 new formula to include peep
I like where you’re heading, but wouldn’t you want to avoid making any decisions based on Patient B’s numbers until you optimize your vent settings and figure out how much FiO2 and PEEP he really needs? Then get another ABG and see what that PF Ratio is??
Outstanding article. Thank you. I listened to a Gaattinoni webinar and have read his articles describing the 2 phenotypes. I think there are several problems with his ideas. 1. He describes the initial hypoxemia seen in COVID as a vascular problem, where local vasodilation causes increased pulmonary R–> hunting. (If the lung looks minimally damaged on CT, then the V/Q mismatch must be due to a blood flow problem not a ventilation problem) 2. This leads to an increased resp drive, increased transpulmonary pressures, leading to “self induced” lung injury. 3. After self induced lung injury you get “classical ARDS”.… Read more »
Risky business bringing up oxygen toxicity, isn’t that a taboo or something? Oxygen toxicity is certainly a thing however much of what I know about it is from the diving literature where they were subjecting people to much higher partial pressures than what we could achieve. I believe there are at least a few papers (from both the medical and diving world) suggesting that 1 bar is not okay after a certain number of hours (that would be a FiO2 of 1 at sea level for anyone who doesn’t like SCUBA diving). I’ve never tried to make sense of these… Read more »
Josh, I appreciated the round table discussion from this past Sunday. I tried to tune in live, but I live in Wisconsin and the timing was off. Listened last night. Very thoughtful exchange of ideas. It’s wonderful to have experts sharing their ideas and concerns. Thanks for helping the nation during this crisis. It is true there are various levels of skill and its great to learn from all of you. Looking forward to studying this article as well.
ARDS has many causes. Pneumonia is a very typical etiology. COVID 19 refers to the whole spectrum of disease. This is a ARDS having a viral pneumonia as primary insult. We treat it like that. Anyway we have not many therapeutic choices unless you have all resources by your hand.
excellent presentation
few points:
1- Berlin Criteria is a clinical criteria, with high sensitivity and low specificity, the septicity is close to 70% if PF ratio less than 150 and Patient on the vent for more than 72 hours.
2- Using lung protective ventilation in non ARDS is not a bad thing.
3- Have you prone someone while the vent more is APRV.
4- I could not find data about thoracentesis and ARDS, not sure even ARDS ca give you p effusion.
I am such a big fan of your posts. Currently getting my butt kicked in the ICU but working hard to correct. This is extremely helpful