DOI: 10.5937/jomb0-70872 ISSN: 1452-8258

Prognostic value of inflammatory and coagulation-related laboratory indicators in patients with sepsis receiving extracorporeal membrane oxygenation: A propensity score-matched study

Leping Ning, Fangyan Gan, Jingshu Fang, Shuang Deng

Background: Laboratory indicators reflecting systemic inflammation and coagulation dysfunction may help identify patients with sepsis at increased risk of death during extracorporeal membrane oxygenation (ECMO). This study investigated prognostic laboratory indicators and developed a nomogram for estimating intensive care unit (ICU) mortality in patients with sepsis receiving ECMO support. Methods: Clinical and laboratory data were retrospectively collected from 257 adults treated with ECMO between January 2025 and March 2026, including 200 patients with sepsis. Propensity score matching was performed at a 1:1 ratio using ECMO type, International Society on Thrombosis and Haemostasis disseminated intravascular coagulation score, Sequential Organ Failure Assessment score, coagulation indicators, bilirubin levels, massive hemorrhage, and ECMO duration. The matched patients with sepsis were classified according to ICU survival status. PSM (Propensity Score Matching) was performed to control for ECMO-related confounders, isolating the effect of sepsis within the matched cohort. Prognostic indicators were selected using univariate analysis and least absolute shrinkage and selection operator regression and subsequently entered into multivariable logistic regression. A nomogram was constructed, and its performance was assessed using receiver operating characteristic, calibration, and decision curve analyses. Results: Propensity score matching generated 47 matched pairs with no significant between-group differences in the included characteristics (all P>0.05). Among the 47 matched patients with sepsis, 16 (34.04%) died in the ICU. International normalized ratio, antithrombin, interleukin-6, alanine aminotransferase, aspartate aminotransferase, lactic acid, and massive hemorrhage differed significantly between nonsurvivors and survivors (all P<0.05). Least absolute shrinkage and selection operator regression retained antithrombin, IL-6, and massive hemorrhage. Multivariable analysis identified elevated IL-6 levels (OR=1.001, 95% CI: 1.000–1.001; P=0.003) and massive hemorrhage (OR=7.725, 95% CI: 1.082– 55.150; P=0.041) as independent predictors of ICU mortality. The resulting nomogram achieved an area under the curve of 0.898 (95% CI: 0.806–0.990), with a sensitivity of 0.812 and specificity of 0.903. Calibration and decision curve analyses indicated satisfactory agreement and potential clinical utility. Conclusions: IL-6 levels and massive hemorrhage were independently associated with ICU mortality in patients with sepsis receiving ECMO support. The nomogram incorporating these indicators showed promising prognostic performance, although further validation in larger, independent cohorts is required.

More from our Archive