Traditional scoring systems are challenging for early diagnosis and intervention in liver failure. While albumin-corrected anion gap (ACAG) predicts adverse outcomes in various conditions, its effectiveness in liver failure remains unclear. This study analyzed hepatic failure patients from MIMIC-IV, stratified into Q1-Q4 by ACAG quartiles. Kaplan-Meier survival curves examined mortality differences. The connection of ACAG with mortality was examined using Cox regression and restricted cubic spline (RCS) analyses. Additionally, we compared the predictive performance of ACAG, MELD, and their combination for mortality using receiver operating characteristic (ROC) curve analysis, calculated the area under the curve (AUC), and compared the AUC values using the DeLong test. Finally, subgroup analyses were conducted to evaluate the relationship between ACAG and prognosis in different types of liver failure. Current study involving 2,016 patients with liver failure revealed 30- and 90-day mortality rates of 27.23% and 30.01%, respectively. Kaplan-Meier survival curves demonstrated mortality increased with ACAG levels. Multivariate Cox regression confirmed ACAG’s positive association with mortality HR 1.032 (30-day), 1.031 (90-day); both P < 0.001, and the RCS showed a linear relationship. The AUCs for 30-day mortality were 0.641 (ACAG), 0.634 (MELD), and 0.668 (combined); the corresponding values for 90-day mortality were 0.630, 0.621, and 0.654. The AUC of the combined model was significantly higher than that of ACAG alone (P = 0.0009 and 0.0016) and MELD alone (P = 0.0004 and 0.0002) for 30- and 90-day mortality, respectively. Although ACAG yielded a marginally higher AUC than MELD, DeLong’s test indicated no significant difference (P = 0.7035 and 0.5387 for 30-and 90-day mortality, respectively). Subgroup analyses observed interactions in the age subgroups and CRRT subgroups; ACAG was consistently elevated in non-survivors across all liver-failure subtypes. Our study demonstrates that the ACAG index is significantly associated with mortality in patients with hepatic failure; its integration with the MELD score further enhances predictive accuracy.
Wang et al. (Sat,) studied this question.