Abstract
PURPOSE: Chronic obstructive pulmonary disease (COPD) patients are at increased risk of sepsis, a condition associated with high mortality. The anion gap (AG) is commonly used to assess acid-base disturbances, but its reliability declines in hypoalbuminemia. The albumin-corrected anion gap (ACAG) may provide greater accuracy, yet its prognostic value in COPD patients with sepsis, defined according to Sepsis-3 criteria, remains unclear. PATIENTS AND METHODS: This retrospective cohort study analyzed 2072 ICU patients with COPD and sepsis from the Medical Information Mart for Intensive Care IV (MIMIC-IV). Cox regression models evaluated the association between ACAG and mortality, Kaplan-Meier curves illustrated survival differences, restricted cubic splines examined nonlinear relationships, and subgroup analyses assessed consistency across strata. Receiver operating characteristic (ROC) curves compared the predictive performance of ACAG, AG, and serum albumin. RESULTS: Elevated ACAG was independently associated with both short- and long-term mortality. In fully adjusted models, each 1 mmol/L increase in ACAG was linked to higher risk of 28-day mortality (HR 1.064, 95% CI 1.042-1.086, P < 0.001) and 365-day mortality (HR 1.065, 95% CI 1.043-1.087, P < 0.001). A threshold effect was observed at ≥19.25 mmol/L, above which mortality risk increased markedly (28-day HR 1.336, 95% CI 1.126-1.586, P = 0.001; 365-day HR 1.429, 95% CI 1.208-1.691, P < 0.001). Kaplan-Meier survival analysis confirmed significant differences (log-rank P < 0.0001), and ROC analysis demonstrated that ACAG provided superior discrimination compared with AG and albumin for both 28-day (AUC = 0.734) and 365-day mortality (AUC = 0.696). Associations were consistent across clinical subgroups without significant interactions. CONCLUSION: Elevated ACAG was an independent predictor of 28-day and 365-day all-cause mortality in critically ill patients with COPD and sepsis. An inflection point of approximately 19.25 mmol/L identified a clinically meaningful threshold for risk stratification. As a simple and widely accessible parameter, ACAG may facilitate threshold-based triage and individualized management in this high-risk population, though external validation in multicenter prospective cohorts is warranted.