By Vibhu Sharma, MD, MS

Assistant Professor of Medicine, University of Colorado, Denver

SYNOPSIS: In this retrospective cohort study, improvement in the PaO2/FiO2 (P/F) ratio by 54% was the optimal cutoff to predict those more likely to be alive at 28 days.

SOURCE: Lee HY, Cho J, Kwak N, et al. Improved oxygenation after prone positioning may be a predictor of survival in patients with acute respiratory distress syndrome. Crit Care Med 2020;48:1729-1736.

This was a retrospective cohort study of patients admitted to a single center in Seoul, South Korea. The cohort included was admitted to the intensive care unit (ICU) between 2014 and 2020, prior to the onset of the COVID-19 pandemic. The authors reviewed charts of patients who were proned for at least 12 hours. Patients were proned if their PaO2/FiO2 (P/F) ratio was < 150 with a positive end expiratory pressure (PEEP)  5 and FiO2 0.6, consistent with inclusion criteria in the PROSEVA trial.1 The authors reviewed the charts of 223 patients with moderate to severe acute respiratory distress syndrome (ARDS) by the Berlin criteria. Of these, 142 were proned; 26 of these were excluded because they were proned for less than 12 hours. Thus, 116 patients formed the study cohort. Almost all (96%) received neuromuscular blockade.

The primary outcomes were ICU and 28-day mortality. Arterial blood gases (ABGs) were collected prior to proning (time 0), at 4 hours (time P1), at 4-8 hours (time P2), at 8-12 hours (time P3), and at some point between 12 hours and the end of prone positioning (time P4). The P/F ratio also was calculated within four hours of supination (time S1) and at 4-12 hours after supination (time S2). The authors assessed the predictive utility of percent change in P/F ratio and the change in minute ventilation (Ve) to PaCO2 ratio (Ve:PaCO2) at the specified times after the first prone positioning session during the ICU stay. The latter variable allows for an indirect measurement of improvement in dead space. Since (almost) all of the patients were paralyzed, assuming Ve remained constant with controlled mechanical ventilation, PaCO2 would be expected to decrease (“PaCO2 responders”) and the ratio to increase as ventilation perfusion (VQ) mismatch improved.

The median duration of the first prone session was 17.8 hours. ICU survivors were more likely to be proned early (1.2 vs. 2.7 days, P < 0.006). Patients who survived were more likely to have a larger change in their P/F ratio at all time points after proning. A change in P/F ratio of 53.5% from time 0 to time P3 optimally predicted improved mortality. Patients with at least this percent change in P/F ratio were deemed “prone responders.” The optimal cutoff for the percent change in the P/F ratio from time 0 to time S2 (> 4 hours after supination) was 10.4%. A change of > 41.7% in the Ve:PaCO2 ratio from time 0 to time P3 (8-12 hours from time proned) also predicted survival at 28 days.


The authors of this study concluded that among patients with moderate to severe ARDS, prone responders were more likely to survive their ICU stay. The PROSEVA trial demonstrated that prone positioning of patients with ARDS for at least 16 hours a day reduced 28- and 90-day mortality when adjusted for Sequential Organ Failure Assessment (SOFA) scores, vasopressor use, and use of neuromuscular blockade. In the PROSEVA trial, responders were defined as those with an increase in P/F ratio of at least 20 mmHg. Changes in the P/F ratio by this amount did not predict mortality. ABGs were not assessed at fixed time points in the PROSEVA trial; however, a retrospective analysis of the PROSEVA cohort found that ABGs at any time did not predict mortality.2 In addition, there were no differences in survival when ABGs were assessed at the completion of the first session of prone ventilation.

Other studies have also shown that the P/F ratio is not an independent predictor of mortality in multivariate analyses that control for other severity of illness variables.3-4 With respect to PaCO2, the large LUNG-SAFE study showed that sustained hypercapnia (over days 1-2) was not harmful, using a 28-day mortality endpoint, and was seen mostly in patients with severe ARDS.5 However, hypocapnia was harmful in mild to moderate ARDS. The LUNG-SAFE study did not assess changes in PaCO2 over time; however, there is some literature to suggest that improvements in PaCO2 with proning may predict a better prognosis.6

Another notable critique of this study was that delivery of mechanical ventilation in the pressure-controlled mode resulted in a mean tidal volume of 7 ± 1.3 mL/kg ideal body weight, with some patients therefore receiving more than the upper limit of recommended tidal volume per ARDSNet criteria. Second, this was a single-center study, and the majority of the patients were immunosuppressed related to a malignant process and therapeutics, with 90% in each group receiving corticosteroids, thereby limiting generalizability. It is unclear what, if anything, clinicians should take to the bedside with respect to the results of this retrospective single-center study given the disappointing associations of oxygenation with mortality. A recent study assessed the predictive value of the oxygen saturation index (OSI).7 This index incorporates the stiffness of the lung and is calculated as follows:

(FiO2 × mean airway pressure × 100)/SpO2.

In a cohort of 326 adult patients with ARDS in diverse clinical settings (trauma/medical/surgical ICUs), the OSI was found to predict mortality when measured on the first day of diagnosis of ARDS, with the worst value for SpO2 and the highest values for FiO2 and mean airway pressure incorporated.7 When available, measurement of the OSI on day 1 of diagnosis of ARDS predicts mortality, and if available, the percent change in P/F ratio and, in my opinion, more importantly, the Ve:PaCO2 ratio may provide supportive evidence for a better prognosis in the narrow spectrum of patients described by Lee et al.


  1. Guérin C, et al. Prone positioning in severe acute respiratory distress syndrome. N Engl J Med 2013;368:2159-2168.
  2. Albert RK, et al; Proseva Investigators. Prone position-induced improvement in gas exchange does not predict improved survival in the acute respiratory distress syndrome. Am J Respir Crit Care Med 2014;189:494-496.
  3. Ware LB. Prognostic determinants of acute respiratory distress syndrome in adults: Impact on clinical trial design. Crit Care Med 2005;33:S217-S222.
  4. Seeley E, et al. Predictors of mortality in acute lung injury during the era of lung protective ventilation. Thorax 2008;63:994-998.
  5. Madotto F, et al. Patterns and impact of arterial CO(2) management in patients with acute respiratory distress syndrome: Insights from the LUNG SAFE study. Chest 2020;158:1967-1982.
  6. Gattinoni L, et al. Decrease in PaCO2 with prone position is predictive of improved outcome in acute respiratory distress syndrome. Crit Care Med 2003;31:2727-2733.
  7. DesPrez K, et al. Oxygenation saturation index predicts clinical outcomes in ARDS. Chest 2017;152:1151-1158.