Abstract
While the albumin-corrected anion gap (ACAG) has established prognostic utility across multiple disease states, its association with both short- and long-term mortality in patients with gastrointestinal bleeding (GIB) remains indeterminate. This study seeks to systematically evaluate the relationship between ACAG and mortality outcomes in this patient population. We performed a retrospective cohort study using data from the Medical Information Mart for Intensive Care IV (version 3.1, which includes emergency department and ICU admissions at Beth Israel Deaconess Medical Center from 2008 to 2022). Critically ill patients with GIB were included to evaluate the prognostic value of ACAG for mortality risk. Feature selection was conducted using the Boruta algorithm, followed by risk stratification with X-tile analysis. To assess associations between ACAG and patient outcomes, we performed multivariable Cox proportional hazards regression for short- and long-term mortality. Survival trends were compared across groups using Kaplan-Meier analysis. The discriminatory ability of ACAG was further examined via receiver operating characteristic curve analysis. Nonlinear relationships were tested with restricted cubic splines, and subgroup analyses were conducted to explore potential interaction effects. Finally, we constructed a nomogram based on the predictive factors. This retrospective study analyzed 1624 eligible patients. Boruta analysis confirmed ACAG's prognostic significance, with X-tile identifying ≥ 20 as the optimal cutoff (28-day mortality-based). Adjusted Cox models showed ACAG ≥ 20 independently predicted higher all-cause mortality (all P < .001), supported by Kaplan-Meier analysis (log-rank P < .001). Receiver operating characteristic analyses demonstrated moderate discrimination, while restricted cubic splines indicated linear mortality risk association. Subgroup analyses revealed consistent prognostic value. The ACAG-based nomogram achieved 30-/90-/180-/365-day mortality AUCs of 0.806 (0.779-0.833), 0.805 (0.782-0.829), 0.797 (0.773-0.821), and 0.798 (0.774-0.821). The ACAG demonstrates a significant association with both short- and long-term mortality in critically ill patients with GIB.