Albumin-corrected anion gap and short-term mortality in patients with sepsis and heart failure: A retrospective...

The conventional anion gap (AG) is widely used to evaluate metabolic acidosis, but its accuracy is limited in hypoalbuminemia, a common condition among critically ill patients. Albumin-corrected anion gap (ACAG) has been proposed to overcome this limitation, yet its prognostic value in patients with sepsis complicated by heart failure (HF) remains unclear. We retrospectively analyzed adult patients with sepsis and HF from the Medical Information Mart for Intensive Care IV database. ACAG was calculated as (44 - serum albumin [g/L]) × 0.25 + AG, and patients were stratified into high (≥20 mmol/L) and low (<20 mmol/L) groups using X-tile-derived thresholds. The primary outcomes were 28-day and in-hospital mortality. Survival was assessed using Kaplan-Meier curves with log-rank tests. Independent associations were examined with multivariable Cox regression. Restricted cubic splines were applied to evaluate nonlinear trends. Variable importance was assessed using the Boruta algorithm. Discrimination between ACAG and AG was compared using receiver operating characteristic curves and DeLong's test. Subgroup analyses explored consistency across clinical strata. A total of 715 patients were included, of whom 366 were classified into the high-ACAG group. High ACAG was associated with significantly higher 28-day and in-hospital mortality (log-rank P < .001). Multivariable Cox regression confirmed ACAG as an independent risk factor for poor prognosis (28-day mortality: hazard ratio = 1.80, 95% confidence interval: 1.27-2.56; in-hospital mortality: hazard ratio = 1.75, 95% confidence interval: 1.20-2.54). Restricted Cubic Splines showed a nonlinear association with 28-day mortality and a near-linear relationship with in-hospital mortality. Boruta analysis ranked ACAG higher than AG and several conventional indicators. Receiver operating characteristic analysis demonstrated that ACAG provided better discrimination than AG for both outcomes, with the combined Sequential Organ Failure Assessment + ACAG model achieving the best performance (P < .01). Associations remained consistent across subgroups without significant interactions. ACAG is a robust independent prognostic marker in septic patients with heart failure, outperforming conventional AG in short-term outcome prediction. Integration of ACAG with Sequential Organ Failure Assessment may enhance early risk stratification and guide clinical decision-making.