Association between the endothelial activation and stress index and all-cause mortality in patients receiving extracorporeal membrane oxygenation: a retrospective cohort study based on the MIMIC-IV database

Abstract Objective To evaluate the association between the endothelial activation and stress index (EASIX) within 24 h of ICU admission and 28-day all-cause mortality in patients receiving extracorporeal membrane oxygenation (ECMO) support, and to examine whether this association can be used for early risk stratification. Methods A retrospective cohort study was conducted based on the MIMIC-IV v3.1 (2008–2022) database. Adult critically ill patients receiving ECMO support were included, the ECMO stay was designated as the index stay for each patient, and follow-up was counted from the time point of ICU admission of this stay. The exposure was the worst EASIX within 24 h of ICU admission, taken as the base-2 logarithm (each 1-unit increase meaning a doubling of EASIX). The primary endpoint was 28-day all-cause mortality, and the secondary endpoints were 90-day and 365-day all-cause mortality. Results A total of 155 patients were included in the analysis; 80 died at 28 days (51.6%), 90 at 90 days (58.1%) and 94 at 365 days (60.6%). The median EASIX was 12.14 (IQR 4.04–55.94), range 0.48–2589.5. After adjustment, each doubling of EASIX was associated with a 17% increase in the risk of 28-day all-cause mortality (HR 1.169, 95% CI 1.070–1.277, P < 0.001); the direction and magnitude of the results at 90 days and 365 days were consistent (HR 1.180, 95% CI 1.084–1.284; 1.182, 95% CI 1.087–1.285; both P < 0.001). Restricted cubic spline found no evidence of departure from linearity (P for non-linearity = 0.745), and the association rose monotonically over the interval of log₂EASIX from 0 to 12. When grouped by cohort tertiles, 28-day mortality rose with increasing tertile (32.7%, 54.9%, 67.3%; log-rank χ² = 16.07, P < 0.001); with T1 as the reference, the HR was 2.044 (1.080–3.871) for T2 and 3.347 (1.750–6.403) for T3, and the HR for trend per one-tertile increase was 1.805 (1.319–2.471). Within each subgroup of age, sex, ECMO mode, cardiogenic shock and SOFA the direction of the association was consistent, and none of the interaction tests was statistically significant (all P > 0.05). Among the eight sensitivity analyses, six had an HR still above 1 and statistically significant; only after excluding those with unknown ECMO mode (n = 74) and in those analysed for the VA mode only (n = 45) did the point estimate remain unchanged (1.163 and 1.201) while the confidence interval crossed 1. Conclusions In adult critically ill patients receiving ECMO support, a raised EASIX within 24 h of ICU admission was independently associated with 28-day all-cause mortality, and this association persisted at 90 days and 365 days and was consistent across multiple subgroups. EASIX is composed of three routine bedside indices at zero additional cost, and can serve as an auxiliary tool for early risk stratification of ECMO patients.