Conceptually, ARDS consists in an acute respiratory failure caused by inflammatory pulmonary edema, characterized by increased vascular permeability with extravasation of fluids into the interstitial space with consequent flooding of the alveolar spaces.1 The loss of aerated lung tissue, due to atelectasis in gravity-dependent areas produced by the increased weight of the overlying lung tissue, leads to profound oxygenation impairment secondary to increased intrapulmonary shunt and alveolar dead space, along with a marked decrease in respiratory system compliance. ARDS is also characterized by the presence of lung infiltrates on CXR (CXR) and computed tomography (CT) scans (Fig. 1A and B). Anatomically and pathologically, the main feature of ARDS is an histological pattern known as diffuse alveolar damage (DAD), which includes the presence of hyaline membranes, edema, type I and II alveolar cell necrosis, and hemorrhage.2,3 This description constitutes the "conceptual model of ARDS" and reflects how clinicians "perceive" the syndrome.4

Figure 1.

Images with bilateral infiltrates.

A) CXR frontal view. B) Cross-sectional computed axial tomography.

(0.37MB).

ARDS can develop secondary to a multitude of risk factors of pulmonary (direct) and extrapulmonary (indirect) cause. This etiological heterogeneity likely reflects the activation of different mechanisms of injury (Table 1).5

Another crucial aspect is the lack of gold standard for defining ARDS. Even DAD is not pathognomonic since it might not be identified in all clinically diagnosed ARDS cases.6 Given the pathophysiological complexity of ARDS and the diverse causes that generate it -although all ultimately converge in the activation of proinflammatory mechanisms- a definition to standardize diagnosis, clinical management, and use of various therapeutic approaches was required. Additionally, the definition of a disease promotes the development of related scientific research.

The original definition of ARDS dates to 1967 and was issued by Petty and Ashbaugh, who described a group of 12 patients with acute respiratory failure, profound hypoxemia secondary mainly to intrapulmonary shunt, bilateral infiltrates on CXR and decreased thoraco-pulmonary compliance, with no history of chronic respiratory failure, and in absence of left ventricular failure.6 This condition occurred after exposure to what they called a "catastrophic event" (which are now the risk factors for ARDS), of pulmonary or extrapulmonary origin (Table 2). Because of its clinical and radiological similarity to respiratory distress syndrome secondary to surfactant deficiency in newborns, the authors named this entity "Adult Respiratory Distress Syndrome". Petty and Ashbaugh's definition identified what is currently known as severe ARDS, since from the point of view of oxygenation impairment it referred to patients with PaO2 < 50 mmHg with an inspired oxygen fraction (FiO2) > 0.6; that is, with deep hypoxemia, evidenced by a PaO2/FiO2 ratio of approximately 80 mmHg.7

Since this initial approach, several modifications of the definition have been made. In 1988, Murray et al. proposed the Lung Injury Score (LIS) with the intention of quantifying the severity of the syndrome. The LIS was the average value of 4 variables, expressed as a score from 0 to 4 reflecting increasing severity: hypoxemia (defined as the PaO2/FiO2 ratio); extent of pulmonary infiltrates on chest radiograph (in quadrants), thoraco-pulmonary compliance (ml/cmH2O) and use of positive end-expiratory pressure (PEEP, cmH2O). ARDS was defined by an LIS > 2.5.8 The LIS was adopted for use, but the definition remained a construction of few experts.

In 1994, the American-European Consensus Conference (AECC) definition was published, in which the syndrome known until then as Adult Respiratory Distress Syndrome was considered as compounded by two conditions of progressive severity: acute lung injury (ALI) and ARDS (Acute Respiratory Distress Syndrome), defined by the compromise of oxygenation: PaO2/FiO2 ≤ 300 and ≤200, respectively (Table 3).9 The acute characteristics of the syndrome and the exclusion of cardiovascular causes for pulmonary edema development were maintained.

The AECC definition, like the LIS score, suffered from several criticisms, centered mainly on two aspects. First, experts highlighted the diagnostic difficulties for the evaluation of CXR infiltrates on CXR, due to the great intra- and interobserver variation. In addition, the decision to exclude any standardized level of PEEP for oxygenation assessment produced further variability, secondary to PEEP great impact on the definition of ARDS. Patients could quickly move from one category of hypoxemia to another, without implying a real change in the underlying disease and in its severity.10–12

The publication of the Berlin definition published in 2012 produced other significant changes (Table 3).4 The timing of onset of the acute respiratory failure was incorporated to the preexistent AECC definition: ARDS had to appear within 1 week of exposure to a risk factor. In addition, clarification was added for the origin of edema and of lung images, allowing definition by CT. Three mutually exclusive categories of ARDS severity were established based on the PaO2/FiO2 ratio, evaluated with a minimum PEEP level of 5 cmH2O. In cases of mild ARDS the possibility of considering ARDS in patients who met the diagnostic criteria and were receiving noninvasive ventilation (NIV) was acknowledged. Up till then, ARDS could only be diagnosed in patients undergoing invasive mechanical ventilation.

The Berlin definition was a great advance. One of its strengths was its empirical validation in 3670 patients, unlike previous consensus definitions which only involved agreement between experts. An attempt was also been made to make the new definition compatible with previous ones, especially with the AECC. Another positive feature is Berlin definition’s predictive validity for mortality. Thus, as severity of ARDS increases, an increase in mortality and comorbidities was observed. The experts sought to ensure that the variables defining ARDS were easily measurable, that is to say, that the application of the definition was feasible. For example, the increase in extravascular lung water was considered the variable that best reflected ARDS, but its incorporation into the definition was discarded due to the technical difficulties and expensive technology involved. In addition, the experts established a "conceptual model" of ARDS, which was mentioned at the beginning of this review.

Berlin's definition, however, had limitations. The first is that it requires the use of a minimum level of PEEP for diagnosis, either with invasive or noninvasive ventilation―in the latter case in mild ARDS only. The assessment of bilateral infiltrates on CXR continues to lack intra- and interobserver reproducibility (reliability), and the consideration of a 7-day interval within which the syndrome should develop after exposure to a risk factor is completely arbitrary.13

Another historical debate, which is reinforced after the presentation of each new definition, is whether the syndrome called ARDS really exists or is simply a compilation of multiple, very heterogeneous diseases causing acute hypoxemic respiratory failure. Thus, all definitions of ARDS would necessarily be "unsatisfactory and superficial".13–16

Despite these criticisms, the utilization of AECC definition and the homogenization it implied helped to establish crucial therapeutic achievements which decreased mortality, such as protective ventilation. In addition, the use of high PEEP, compared to the conventional approach of intermediate PEEP, was shown to have no benefit.17,18 The Berlin definition was also widely adopted.

In 2016, in a study conducted in Kigali, Rwanda, the researchers noted that with the previous definitions no patient with acute hypoxemic respiratory failure could be diagnosed as having ARDS, because blood gas measurements were unavailable; it was therefore impossible to know the PaO2/FiO2.19 Only a single daily measurement of peripheral oxygen saturation measured with pulse oximetry (SpO2) was available. The typical images of ARDS could also not be considered for diagnosis since less than 50% of patients with hypoxemia had access to CXR. However, lung ultrasound was available. Furthermore, less than 30% of the hypoxemic patients could receive mechanical ventilation, due to the lack of ventilators; therefore, the PEEP ≥ 5 cmH2O criteria could not be applied either. Finally, due to lack of beds, only 30% of patients with hypoxemic acute respiratory failure could be admitted ICU; thus, patients with less severe ARDS were possibly underdiagnosed. These profound deficiencies in critical care provision are also present in other low-resource areas.

The researchers proposed the Kigali definition of ARDS as a modification of the Berlin definition (Table 3):

• •

  The PaO2/FiO2 ≤ 300 required for the diagnosis of ARDS was replaced by SpO2/FiO2 ≤ 315, based on the acceptable linear correlation between PaO2 and SpO2, provided that SpO2 is ≤ 97%, there are no hemoglobin abnormalities, and that peripheral perfusion is adequate. The Rice equation reflects this relationship:

  SpO2 /FiO2 = 64 + 0–84 (PaO2/FiO2)20

  However, above the 97% threshold, changes in PaO2 generate minimal impact on SpO2, due to the shape of the oxyhemoglobin dissociation curve; therefore, in the flat part of the curve the correlation between both methods of oxygenation assessment is lost.

• •

  Lung ultrasound was added as diagnostic method for diagnosis of the syndrome, in the absence of access to CXR or lung CT. Even though these variables were not incorporated at that time into the Berlin definition as it would have been appropriate, what occurred shows the importance of constituting panels of experts with members from all regions of the world. The Berlin definition of ARDS, like the previous AECC definition, had been designed by experts from high-income countries (mostly North America and Europe) without the involvement of researchers from middle- and low-income countries.21

The Kigali definition also reflected the recent expansion of ultrasound as a diagnostic method in the ICU, which might be more reliable than CXR for evaluation of lung infiltrates when trained operators are involved.22,23

The SARS-CoV-2 pandemic that devastated the world and generated maximum stress on health care systems, particularly on the ICUs and emergency departments, prompted a reevaluation of the definition of ARDS. Thousands of patients with acute hypoxemic respiratory failure secondary to COVID-19 were simultaneously admitted to hospitals worldwide; about 15–20 % presented severe disease and 5% required admission to the ICU.24 This trigged the need of continuous monitoring of oxygenation in patients who could rapidly worsen and require some type of respiratory support, from oxygen therapy to invasive mechanical ventilation. Thus, the usual PaO2/FiO2 monitoring with arterial blood gases, which is intermittent, laborious, invasive and resource-intensive, was replaced by the continuous, noninvasive and less costly SpO2 measurement, capable of detecting changes rapidly.

Additionally, during the pandemic, the use of high-flow nasal cannula (HFNC), which was increasingly used in acute hypoxemic respiratory failure,25 became widespread.

As a result of these changes, in 2021, 54 years after Petty and Ashbaugh's publication, an update of the definition of ARDS that could also be universally applicable was proposed. An evolution from the "expert consensus" to "a scientific system of categorization" using approaches adopted in other fields of knowledge to build definitions of syndromes (also called "constructs") for which there are no gold standards, was deemed crucial.26 Lack of a gold standard for definition not only occurs in ARDS but also in other well-known syndromes such as fragility, heart failure, and irritable bowel syndrome, among others.27

For these reasons, the construction of the New Global definition followed a rigorous methodology: description of the clinical phenomenon (or syndrome, or "construct") to be "captured"; justification of the changes proposed for the new definition; explicit criteria for the selection of experts for the panel, which should reflect not only expertise but gender, cultural, geographic, socioeconomic, and ethnic diversity; and specification of the method by which agreements would be reached. Usually, an agreement implies a majority of at least 70% when voting to accept or discard a variable or a statement.

Another objective was to reevaluate the conceptual model of ARDS recommended by the Berlin definition and to develop a New Global Definition of ARDS (Table 4).28 The feasibility of the New Global Definition was also evaluated, i.e., whether its components are easily measurable, both in clinical situations and in research, and can be measured worldwide. The reliability (or reproducibility), namely, the ability of the definition to diagnose the same patient equally in different scenarios and when applied by different professionals, was also analyzed. Finally, the validity of the new definition was explored, that is to say, its ability to reflect what clinicians really want to identify. These concepts include surface validity (the ability to identify the characteristics that are obviously part of ARDS and together distinguish patients with the condition from those without it); and predictive validity (whether the definition predicts outcomes, such as mortality, which should be more frequent in patients with the syndrome compared to those without it).29

In addition, it was intended that the New Global Definition of ARDS should be supported by different critical care societies worldwide. After a meticulous review process, the following updates were incorporated into the Berlin consensus definition:

There are several topics that might have been considered in this new global definition. For example, the discussion about prognosis enrichment, which might help to optimize validity and applicability of the results obtained in clinical trials on ARDS. As such, the consideration of stricter inclusion criteria, such as limiting participation to intubated patients with moderate to severe ARDS, or to those meeting the definition of ARDS with a programmed PEEP value of 10 cmH2O, could identify more homogeneous subgroups of patients. In turn, this would improve the understanding of clinical variability and facilitate translation to clinical practice of the findings.

Another aspect that was not addressed is how to assess unilateral infiltrates on CXR. In clinical practice, they are considered an exclusion criterion for ARDS, which may preclude the application of protective ventilation strategies.14 However, it is known that CXR is susceptible to biases and limitations, as bilateral infiltrates may go unrecognized especially when taken with the equipment commonly used in ICUs, and in patients with pre-existing lung disease.41 Another common misconception is that patients with unilateral infiltrates on mechanical ventilation have better oxygenation and lower mortality, compared to those with bilateral infiltrates and ARDS.42 Yet data from the LUNG SAFE study reveal that while patients with unilateral infiltrates usually have lower initial severity, mortality does not differ significantly from that of patients with ARDS. Furthermore, in patients with unilateral infiltrates, ventilatory monitoring variables such as driving pressure are important risk factors for developing ARDS, highlighting the importance of implementing protective ventilatory strategies also in this population.43

On the other hand, there is still debate over some issues. Researchers could choose to focus on subjects with ARDS who continue to meet the definition beyond 24 h if they wish to exclude those who improve rapidly or introduce a stabilization period as proposed by Guerin in the PROSEVA study, therefore achieving more homogeneous subgroups to evaluate a response to a therapeutic maneuver.44–46 Other subcategories of ARDS (subgroups and subphenotypes) that were not included in the current definition have been proposed to assess possible different responses to treatment. These subcategories might be defined by biomarker levels, types of images (focal or diffuse lung infiltrates), recruitment potential, or physiological variables such as inspiratory pressure, elasticity behavior, or the ventilatory ratio.47–53

The New Global Definition of ARDS is an extension of the Berlin definition in matters of diagnosis and monitoring of the disease. The category of intubated patients preserves the previously described PaO2/FiO2 severity cut-off points but allows the use of equivalent SpO2/FiO2 cutoff values in situations where arterial blood gases are not available. The two new categories added: ARDS in non-intubated patients, and ARDS in patients assisted in regions with low health resources, constitute a relevant innovation, but it is especially highlighted that the diagnostic modifications proposed for these regions should be considered only in them. It is recommended that researchers continue to report their data based on the Berlin definition when possible. The epidemiological impact of the modified definition for low-resource areas is still uncertain. Looking ahead, there is great hope that the identification of ARDS subphenotypes may facilitate the assessment of response to different treatments and the development of personalized treatments.

GAP has received funding for teaching programs from Medtronic LATAM and Vapotherm Inc. USA.