P/FP ratio: incorporation of PEEP into the PaO(2)/FiO(2) ratio for prognostication and classification of acute respiratory distress syndrome

BACKGROUND: The current Berlin definition of acute respiratory distress syndrome (ARDS) uses the PaO(2)/FiO(2) (P/F) ratio to classify severity. However, for the same P/F ratio, a patient on a higher positive end-expiratory pressure (PEEP) may have more severe lung injury than one on a lower PEEP. OBJECTIVES: We designed a new formula, the P/FP ratio, incorporating PEEP into the P/F ratio and multiplying with a correction factor of 10 [(PaO(2)*10)/(FiO(2)*PEEP)], to evaluate if it better predicts hospital mortality compared to the P/F ratio post-intubation and to assess the resultant changes in severity classification of ARDS. METHODS: We categorized patients from a dataset of seven ARDS network trials using the thresholds of ≤ 100 (severe), 101–200 (moderate), and 201–300 (mild) for both P/F (mmHg) and P/FP (mmHg/cmH(2)O) ratios and evaluated hospital mortality using areas under the receiver operating characteristic curves (AUC). RESULTS: Out of 3,442 patients, 1,057 (30.7%) died. The AUC for mortality was higher for the P/FP ratio than the P/F ratio for PEEP levels > 5 cmH(2)O: 0.710 (95% CI 0.691–0.730) versus 0.659 (95% CI 0.637–0.681), P < 0.001. Improved AUC was seen with increasing PEEP levels; for PEEP ≥ 18 cmH(2)O: 0.963 (95% CI 0.947–0.978) versus 0.828 (95% CI 0.765–0.891), P < 0.001. When the P/FP ratio was used instead of the P/F ratio, 12.5% and 15% of patients with moderate and mild ARDS, respectively, were moved to more severe categories, while 13.9% and 33.6% of patients with severe and moderate ARDS, respectively, were moved to milder categories. The median PEEP and FiO(2) were 14 cmH(2)O and 0.70 for patients reclassified to severe ARDS, and 5 cmH(2)O and 0.40 for patients reclassified to mild ARDS. CONCLUSIONS: The multifactorial P/FP ratio has a greater predictive validity for hospital mortality in ARDS than the P/F ratio. Changes in severity classification with the P/FP ratio reflect both true illness severity and the applied PEEP strategy. Trial registration: ClinialTrials.gov–NCT03946150. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13613-021-00908-3.