Estimated plasma volume for predicting sepsis-associated ARDS risk: a multicentre retrospective cohort study

Sepsis-associated acute respiratory distress syndrome (SA-ARDS) is a lethal complication demanding early predictors. This study assessed estimated plasma volume status (ePVS) for predicting SA-ARDS development.

This multicentre cohort study analysed 3854 adult patients with sepsis 3.0 from China. The primary outcome was the incidence of SA-ARDS. Multivariate logistic regression, propensity score matching (PSM) and inverse probability weighting (IPW) were used to assess associations between ePVS and SA-ARDS.

Elevated ePVS was independently associated with SA-ARDS incidence across all models (multivariate OR: 1.56, 95% CI 1.50 to 1.63; PSM OR: 1.72, 1.63 to 1.81; IPW OR: 1.58, 1.49 to 1.69; p<0.001). Receiver operating characteristic analysis demonstrated a strong discriminative ability of ePVS for SA-ARDS incidence (area under the curve (AUC): 0.772, 95% CI 0.757 to 0.788). SA-ARDS patients exhibited higher mortality (31.8% vs 23.1%, p<0.001), mechanical ventilation (MV) use (82.3% vs 48.4%, p<0.001) and continuous renal replacement therapy (CRRT) requirements (7.7% vs 4.7%, p<0.001). Generalised additive model analysis revealed a significant linear association between ePVS >8.0 dL/g and SA-ARDS mortality (p<0.001). Although ePVS >8.0 dL/g alone showed limited predictive value for in-hospital mortality in SA-ARDS patients (AUC: 0.549), its combination with acute physiology and chronic health evaluation II significantly improved discrimination (AUC: 0.823). This synergistic effect persisted even after excluding early deaths (≤72 hours), reinforcing ePVS's complementary role alongside established severity scores in mortality risk stratification. Stratified analysis revealed ePVS levels strongly correlated with acute respiratory distress syndrome (ARDS) severity (p<0.001), particularly distinguishing mild from moderate cases. Patients with pulmonary infections exhibited higher ePVS than non-pulmonary sources (p<0.001).

These findings highlight ePVS is strongly associated with ARDS onset, disease severity and mortality risk, supporting its utility as a prognostic marker in risk stratification.