ARDS patients with CRS less than 45 cm/H2O and small body sizes may not be fit for high-PEEP levels: A secondary analysis of a randomized controlled trial 

Abstract

Object: Identification of respiratory mechanics phenotypes could provide novel information for the selection of suitable positive end-expiratory pressure (PEEP) in specific ARDS patients. Methods: Patient information data and other data in the current study were downloaded from a randomized controlled trial (low-PEEP vs. high-PEEP for ARDS) on the ARDS network public database (http://www.ardsnet.org/). Based on respiratory mechanics characteristics, phenotypes were identified through an unsupervised clustering algorithm applied to routinely available respiratory mechanics features within 24 hours of hospital presentation before the patients were treated with lower-PEEP or higher-PEEP. The respiratory mechanics characteristics included direct or indirect lung injury, respiratory rate, mean airway pressure, inhaled oxygen concentration, arterial partial pressure of oxygen, arterial partial pressure of carbon dioxide, respiratory system compliance (CRS) and plateau pressure. Statistical analysis of 60-day mortality was utilized to identify specific patients with ARDS who could benefit (or could be harmed) from a lower versus a higher PEEP, with a P<0.05 cut-off. Results: A total of 549 patients with ARDS were enrolled in the current secondary analysis, of which 273 patients had a low-PEEP (8.3±3.2 cm/H2O), and 276 patients had a high-PEEP (13.2±3.5 cm/H2O). Three respiratory mechanics phenotypes were identified, and phenotype I was classified as the “higher-PEEP not fit for” phenotype since higher-PEEP resulted in a significant increase in 60-day mortality compared with low-PEEP (39.0% vs 23.1%). Patients with phenotype I showed a worse CRS (29.8±16.1 ml/cm/H2O), were more commonly female (67.3%) and were more commonly characterised by a lower height (164.5±7.5 cm) and/or a decreased weight (76.8±20.3 kg). Statistical analysis of the mortality indicated that there were no significant differences between the high-PEEP groups and low-PEEP groups in the cohorts with other phenotypes. Conclusion: ARDS patients with CRS less than 45 cm/H2O and a small body size may not be good candidates for a high-PEEP strategy.