Latent class analysis to predict intensive care outcomes in Acute Respiratory Distress Syndrome: a proposal of two pulmonary phenotypes

Abstract Background Acute respiratory distress syndrome remains a heterogeneous for clinicians and researchers difficulting successful tailoring of interventions trials. To this moment, phenotyping has been approached by means inflammatory laboratory panels. Nevertheless, the systemic expression acute might not reflect its mechanics gas exchange. Methods Retrospective analysis prospective cohort two hundred thirty-eight patients consecutively admitted under mechanical ventilation presenting with syndrome. All received standardized monitoring clinical variables, computed tomography scans at predefined PEEP levels. Employing latent class analysis, an unsupervised structural equation modelling method, on mechanics, gas-exchange tomography-derived gas- tissue-volumes level 5cmH 2 O, distinct pulmonary phenotypes were identified. Results Latent was applied to 54 CT-derived tissue-volume two-class model identified as best fitting. Phenotype 1 ( non-recruitable ) presented lower system elastance, alveolar dead space amount potentially recruitable lung volume than phenotype ). responded increase in ventilated tissue, compliance PaO /FiO ratio p < 0.001), addition decrease recruitment manoeuvre. Patients belonging higher intensive care mortality (hazard 2.9, 95% confidence interval 1.7–2.7, = 0.001). Conclusions The present study identifies ARDS based tissue-volumes. These are characterized distinctly diverse responses manoeuvre diverging mortality. Given multicentre validation, simple rapid identification these could facilitate enrichment future trials addressing strategies ARDS.