Lactate and pH as Predictors of In-Hospital Mortality in Extracorporeal Cardiopulmonary Resuscitation: A Single-Center Retrospective Study

Objectives Extracorporeal cardiopulmonary resuscitation (ECPR) is increasingly employed for patients with refractory cardiac arrest, yet optimal criteria for patient selection remain poorly understood. Pre-cannulation lactate and pH values are prognostic markers, however, given the wide ranges in existing literature, there is no consensus on threshold values that optimally stratify risk. This study aims to evaluate individual and combined prognostic value of pre-cannulation lactate and pH in predicting hospital mortality. Design Retrospective cohort study. Setting Single, quaternary academic center. Participants Adult patients undergoing ECPR between September 2011 and April 2025 (n = 484). Interventions None. Measurements and Main Results Pre- and post-cannulation lactate and pH values were retrospectively identified. Logistic regression assessed associations between lactate and pH and hospital mortality, receiver operating characteristic (ROC) curves and area under curve values were calculated. Optimal cut-off values for lactate and pH were calculated using Youden-index in ROC analyses. Median lactate was 8.7 mmol/L (IQR 4.1-12.3), median pH was 7.23 (IQR 7.09-7.34). In-hospital mortality was 62.4%. Non-survivors had significantly higher median lactate (9.3 vs. 7.3, P<0.001) and lower median pH (7.2 vs 7.27, P<0.001). In multivariable logistic regressions, both lactate (OR 1.13 per mmol/L, 95% CI 1.05-1.22) and pH (OR 0.61 per 0.1-unit, 95% CI 0.47-0.76) were each independently associated with mortality. However, in models including both variables, pH remained a significant predictor of mortality whereas lactate did not. Optimal cutpoints were 10.85 mmol/L for lactate and 7.2 for pH. Conclusions These findings support the inclusion of lactate and pH in a cumulative, multifactorial risk stratification framework to inform real-time prognostication and patient selection for ECPR on an individual programmatic basis.