Relationship between anion gap and in-hospital mortality in intensive care patients with liver failure: a retrospective propensity score matching analysis

Objectives: To elucidate the association between anion gap (AG) and in-hospital mortality in intensive care patients with liver failure. Methods: Demographic and clinical characteristics of intensive care patients with liver failure in the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database were collected, and binomial logistic and Cox regression was conducted to investigate the association between AG and in-hospital mortality. The area under the receiver operating characteristic (ROC) curve (AUC) was conducted to characterize the performance of AG in predicting in-hospital mortality, and was compared with the albumin corrected anion gap (ACAG) and the End-Stage Liver Disease (MELD) score. The Kaplan-Meier curve was plotted for in-hospital survival analysis of AG and patients with liver failure. The propensity score matching (PSM) analysis was performed to mitigate selection bias. Results: AG was an independent risk factor for in-hospital mortality in intensive care patients with liver failure. Before PSM, the AUCs of AG, ACAG, and MELD were 0.666, 0.682, and 0.653, respectively. After PSM, the AUCs of AG, ACAG, and MELD scores were 0.645, 0.657, and 0.645, respectively, and there is no difference in the predictive performance of the three indicators upon comparison. Compared with the low-AG (<= 20 mmol/L) group, the hazard ratio (HR) for in-hospital death of the high-AG (>20 mmol/L) group was determined to be 2.1472 (before PSM)/1.8890 (after PSM). Conclusions: AG is associated with in-hospital mortality in intensive care patients with liver failure and demonstrates a moderate predictive value, which is comparable to the predictive power of the MELD score. AG may serve as an indirect marker of in-hospital mortality of patients with liver failure by reflecting the degree of metabolic acidosis.