High anion gap and albumin-adjusted anion gap are associated with hospital mortality in intensive care unit patients with liver cirrhosis: A retrospective cohort

Data of patients with liver cirrhosis (LC) were collected from the Medical Information Mart for Intensive Care III database to explore whether anion gap (AG) and albumin-adjusted AG (AA-AG) values were associated with outcomes in patients with LC. We retrospectively analyzed data of adult patients with LC. Based on the AG and AA-AG level, patients were then divided into groups according to third percentile. Lowess smoothing was first applied to visualize the crude relationship between AG or AA-AG and inhospital mortality. Survival curves were generated with the Kaplan–Meier and compared by log-rank test. Multivariable logistic regression was constructed to quantify the independent effect of elevated or AA-AG on hospital mortality after adjustment multiple confounding factors. Model discrimination was assessed with area under the receiver operating characteristic curve (AUC) and 95% confidence intervals (CI). Lowess Smoothing technique showed that AG and AA-AG were associated with hospital mortality for patients with LC. Crude outcomes and Kaplan–Meier survival curve analysis revealed that hospital survival rates of patients with high AG and AA-AG values were significantly lower ( P < .001) compared to those with lower values. After adjusting for multiple confounding factors, analysis revealed that elevated AG (>19 mmol/L) was an independent risk factor for increased inhospital mortality in patients with LC (odds ratio: 1.887 [95% CI: 1.208–2.95]; P < .05), and elevated AA-AG (>21.5 mmol/L) was an independent risk factor for increased inhospital mortality in patients with LC (odds ratio: 1.892 [95% CI: 1.229–2.912]; P < .05). Specifically, the AG demonstrated an AUC of 0.6704 (95% CI: 0.63–0.71) in predicting hospital mortality. The Model for End-Stage Liver Disease (MELD), on the other hand, exhibited a higher predictive accuracy with an AUC of 0.7186 (95% CI: 0.68–0.76). When AG and MELD were combined, the predictive performance further improved, yielding an AUC of 0.7302 (95% CI: 0.69–0.77). Similarly, the AA-AG showed an AUC of 0.684 (95% CI: 0.64–0.73) in predicting hospital mortality, and when combined with the MELD, the AUC increased to 0.7376 (95% CI: 0.70–0.78). Elevated serum AG (≥19 mmol/L) and AA-AG (≥21.5 mmol/L) were risk factors for inhospital mortality among critically ill patients with LC.