Acute respiratory distress syndrome (ARDS) is a heterogeneous syndrome characterized by an acute inflammation, increased permeability, and edema of the lung tissue, with a typical histological hallmark of diffuse alveolar damage, leading to alveolar collapse and a severe impairment in oxygen diffusion.1 ARDS accounts for 10% of intensive care unit (ICU) admissions and 23% of patients receiving mechanical ventilation, with a mortality rate that ranges from 35% in mild cases to 46% in severe cases.2 According to the Berlin definition, severity of ARDS is classified into 3 categories (mild, moderate, and severe) based on the arterial partial pressure of oxygen to fraction of inspired oxygen (PaO2/FiO2) ratio, with a minimum positive end-expiratory pressure (PEEP) level of 5cm H2O.3 These criteria have been useful for designing clinical trials and observational studies, as well as for improving clinical decision-makings and therapeutic interventions.4

However, the PaO2/FiO2 ratio has some limitations for assessing ARDS severity. First, PaO2/FiO2 ratio is affected by ventilatory setting such as PEEP and inspiratory/expiratory time ratio. Villar et al. improved ARDS risk stratification using standard ventilatory settings (PEEP≥10cmH2O and FiO2≥0.5).5,6 Second, PaO2/FiO2 ratio may change over time. A number of authors found a better mortality prediction by patients’ reclassification ≥24h after ARDS onset.5–7 And third, PaO2/FiO2 ratio is not always linked to mortality in patients with ARDS.8,9

Recently, Sayed and coworkers proposed a novel criterion to address Berlin definition gap by including PEEP in the new index named P/FPE, defined as PaO2/(FiO2×PEEP). The increase the PEEP level with the same FiO2 yields different degree of blood oxygenation. By using machine learning approaches, the authors demonstrated that the P/FPE index after onset and at third day is markedly better than the current PaO2/FiO2 ratio to assess ARDS severity.10 The present study was aimed to evaluate the impact of the P/FPE index on mortality of patients with ARDS.

The present study found that 31.8% of patients with ARDS on day 1 and 77.0% on day 3 were reclassified into a different category of ARDS severity using the P/FPE index. The performance for predicting hospital mortality increased with the P/FPE index, compared to the PaO2/FiO2 ratio. The P/FPE index and ventilatory settings, such as tidal volume and driving pressure, were independently related to increased mortality in multivariate analysis. ARDS severity stratification improved on third day after ARDS diagnosis. Mean PEEP level was lower when ARDS severity was categorized according to the P/FPE index rather than the PaO2/FiO2 ratio.

In order to overcome the gap of the Berlin criteria, Sayed et al. recently proposed the PaO2 to (FiO2×PEEP) ratio, named P/FPE index, as a novel criterion to reclassify ARDS patients in terms of severity.10 Improvement of ARDS severity classification is essential for current critical care medicine since misclassification may lead to errors in clinical judgment and decision-making. For instance, we found that a number of patients were reclassified in a higher category of ARDS severity using the P/FPE index, whereas 72.0% of patients without ARDS on day 3 according to the PaO2/FiO2 ratio truly had ARDS by the P/FPE index. Recently, Palanidurai et al. observed that more than half of the patients were reclassified into a different severity category of ARDS by the P/FPE index, compared to the PaO2/FiO2 ratio. These similar results indicate that changes in severity classification with the P/FPE index reflect the true severity of ARDS and the applied PEEP strategy.19

Under-recognition of ARDS is a common and serious problem with important clinical consequences, particularly in terms of therapeutic options not considered.20 The LUNG SAFE study demonstrated that the diagnosis of ARDS is delayed or missed in 40% of patients.2 P/FPE index is able to identify patients with ARDS who would not be classified with ARDS according to the PaO2/FiO2 ratio, and is also a better predictor of mortality.19 Therefore, by using the P/FPE index clinicians may implement strategies to improve mortality such as optimal PEEP, lower FiO2, prone positioning and the use of neuromuscular blocking agents.

PEEP is a confounder in clinical practice since optimal PEEP is difficult to obtain through several methods,21 and clinicians commonly prescribes high PEEP in patients with adequate oxygenation goals.6,22 Clinical interpretation of the PaO2/FiO2 ratio may be biased by nonpulmonary factors such as hemoglobin concentration and arterial-venous oxygen content difference.23 Consequently, oxygen toxicity with adverse impact on outcomes may be developed due to excess FiO2.24 P/FPE index is attractive because physicians can achieve the best combination of FiO2 and PEEP to reach adequate oxygenation. In our cohort, the frequency of prone positioning was similar to that reported in epidemiologic studies while infusion of neuromuscular blocking agents was lower.2 However, the use of these drugs is controversial because of its side effects and unclear benefits25; currently they are indicated only for the treatment of ventilatory asynchrony, supporting pronation, or assuring protective ventilation goals.21

Our findings demonstrated the clinical validity of P/FPE index. The better performance, compared to PaO2/FiO2 ratio, shows the usefulness of P/FPE index in predicting mortality. Palanidurai et al. also found that P/FPE index has a greater predictive validity for predicting hospital mortality in ARDS patients than the PaO2/FiO2 ratio.19 In fact, the AROC for P/FPE index (0.71 vs. 0.72) and PaO2/FiO2 ratio (0.66 vs. 0.62) was similar to the present study, which supports external validity of our results.

ARDS is one of the major reasons of ICU admission, and continues to have high mortality rates despite advances in supportive care.2,26 Ventilatory support is the keystone in the management of patients with ARDS, but ventilatory setting may have an impact on outcomes.27 Since tidal volume and driving pressure, along with the P/FPE index, were related to increased mortality in multivariate analysis, our findings suggest that the negative effect of ventilatory variables remains unchanged through the course of the disease, which is in line with recent evidences.28,29 Additionally, the relationship between the P/FPE index with mortality explains ARDS severity, but also the potentially harmful effects of PEEP.

The present study confirmed that ARDS severity stratification is improved on third day after ARDS diagnosis.7,10 Lai et al. demonstrated that the PaO2/FiO2 ratio after a period of stabilization may be a more appropriate predictor of mortality than the initial PaO2/FiO2 ratio at the onset of ARDS.30 Chiu et al. found that patients with resolved or improving ARDS severity on day 3 had lower mortality, whereas patients with the same or worsening ARDS severity on day 3 had higher mortality.7 Apparently, patients need to be exposed to a sufficient period of time for response to medical therapies and adjusted IMV settings before being classified.

We found that a lower mean PEEP was used in all class of ARDS severity when patients were stratified by P/FPE index, which might reduce the risk of excess PEEP. Palanidurai et al. observed that the predictive validity of P/FPE index improved with progressively higher levels of PEEP, indicative of the negative effects of higher PEEP.19 PEEP is not considered for ARDS severity evaluation by the Berlin definition. Adding PEEP to the PaO2/FiO2 ratio takes into consideration the respiratory system compliance and lung recruitment. Furthermore, by using PEEP as a quantitative variable, P/FPE index conserves information and improves accuracy of estimations.31 However, although PEEP increases functional residual capacity and improves blood oxygenation, tissue oxygen delivery decreases because of reduced cardiac output.32 PEEP also increases the risk of volutrauma and ventilator-induced lung injury,33 causing increased mortality when a high-PEEP strategy is used.34

This study has a number of strengths. First, the study was conducted in a center with high standard of health care and in an ICU with qualified intensivists 24h a day, seven days a week. Second, potential sources of bias were reduced, which lend additional strength to our analysis. Third, this is a well-powered study with a representative sample size so estimation errors were minimized. Fourth, by multivariate analysis we were able to control for potential confounders including ventilatory setting, and severity of illness. Fifth, the LUNG SAFE study showed that most ARDS patients are not ventilated using a protective ventilation approach.2 We reached ventilatory goals on day 3, which indicates an improvement in quality of ventilatory management according to current recommendations.18,29

There are several limitations of our study. First, this is an observational study from a single center, thus results may not be representative of other institutions or regions. Second, a mixed cohort of surgical and nonsurgical patients with several clinical and pathophysiological disorders were analyzed, which could have effects on outcomes. Indeed, hospital mortality rate was higher than reported in recent epidemiologic studies.2 Since our hospital is a national referral center, case-mix of patients with more severe diseases were more likely included compared with community-based hospitals. Compared with patients enrolled in the LUNG SAFE study,2 our analyzed patients had more chronic diseases including immunoincompetence, cardiovascular disease, and cancer, and had a higher rate of extrapulmonary sepsis and noncardiogenic shock, all which may explain the higher mortality rate observed in the study. Third, trauma patients were not included in the study so results cannot be applied to this type of patients. Fourth, several phenotypes and subphenotypes have been recognized in ARDS patients with impact on outcomes.35,36 In the present study, analyses were not stratified according to ARDS phenotypes; consequently, further studies are required to define the effects of the interaction or association of P/FPE index-based ARDS severity and ARDS phenotypes, as well as its clinical implications, which opens a new agenda of work for future researches. Finally, we explored ARDS severity within the first 72h after starting IMV, and the clinical course of patients beyond this period of time may affect outcomes.

In conclusion, assessment of ARDS severity based on P/FPE index is better than current PaO2/FiO2 criteria for predicting mortality, especially on third day after starting IMV. P/FPE index is easy to use at the bedside by involving information of the two therapeutic strategies used for managing hypoxemia such as FiO2 and PEEP. We recommend further clinical trials to clarify the advantages of ARDS severity classification based on P/FPE index for clinical decision-making.

FDMB contributed in the concepts, design, definition of intellectual content, literature search, clinical studies, data acquisition, data analysis, statistical analysis, manuscript preparation, manuscript editing, and manuscript review.

REM and VOR contributed in the design, definition of intellectual content, clinical studies, data acquisition, manuscript editing, and manuscript review.

TTS contributed in the literature search, manuscript preparation, manuscript editing, and manuscript review.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors declare that they have no competing interest.