Association of the endothelial activation and stress index with concurrent early sepsis-associated liver injury and subsequent mortality: a retrospective study using the MIMIC database

Sepsis-associated liver injury (SALI) is a common and serious complication of sepsis, but auxiliary markers remain limited. The endothelial activation and stress index (EASIX), a marker of systemic endothelial dysfunction, has not been fully evaluated in SALI. This study examined the association of EASIX with concurrent early SALI status and subsequent mortality. This retrospective study used the MIMIC-IV database and included septic patients meeting Sepsis-3 criteria. Multivariable logistic regression was used to assess the association between EASIX and concurrent SALI status within the first 24 h of ICU admission. Among patients with SALI who were alive and still in the ICU at a 24-hour landmark, Kaplan-Meier analysis and Cox regression were performed to evaluate subsequent all-cause mortality (ACM) from the landmark through day 30 after ICU admission (primary prognostic endpoint) and one year after ICU admission (secondary endpoint). Regression estimates were pooled across 20 imputations. Machine learning (ML) analyses for 30-day ACM were exploratory and internally evaluated. The status analysis included 7,699 septic patients, of whom 1,264 (16.4%) had concurrent early SALI. Each one-unit increase in ln(EASIX) was associated with higher SALI odds (adjusted OR = 1.327, 95% CI: 1.247–1.412). In 1,081 landmark-eligible SALI patients, conventional adjusted HRs were 1.359 (95% CI: 1.246–1.481) for 30-day ACM and 1.269 (95% CI: 1.176–1.369) for one-year ACM. Proportional-hazards violations were detected at both horizons; time-varying analyses showed a stronger association earlier in follow-up. EASIX alone had AUCs of 0.649 and 0.605 for 30-day ACM and one-year ACM. Adding EASIX to a non-circular clinical baseline increased concurrent-SALI test AUC from 0.810 to 0.826. AUCs from exploratory ML tests for mortality ranged from 0.694 to 0.718, without clear incremental benefit from EASIX in paired ablation analyses. EASIX was associated with concurrent early SALI status and subsequent mortality, with time-dependent associations with ACM. Its limited stand-alone discrimination for mortality and modest AUC improvement in concurrent SALI support further evaluation of EASIX as an auxiliary marker within multivariable assessment. ML findings remain exploratory, and external validation is required before clinical use. Not applicable.