Association between anion gap and mortality in critically ill patients with gastrointestinal perforation: Evidence from the American MIMIC-IV and the Chinese Central-III Study.

Critically ill patients with gastrointestinal (GI) perforation present with complex, rapidly progressing conditions and high mortality. The impact of anion gap (AG) on mortality in these patients remains understudied, particularly across distinct geographic populations. This retrospective, multicenter cohort study analyzed 1167 critically ill patients with GI perforation from the American MIMIC-IV database (n = 552) and the Chinese Central-III cohort (n = 615). AG was measured using the first value obtained within 24 hours of intensive care unit admission. Patients were stratified by AG quartiles. Multivariable Cox regression assessed mortality risk, restricted cubic spline analysis evaluated nonlinearity, and receiver operating characteristic analysis identified optimal AG thresholds and quantified prognostic performance. Comprehensive subgroup analyses were performed. Mortality rates were comparable between cohorts (MIMIC-IV: 11.8% in-hospital, 24.6% 28-day; Central-III: 16.3% in-hospital, 24.8% 28-day). In both cohorts, the highest AG quartile showed significantly elevated mortality risk versus the lowest quartile (MIMIC-IV: aHR = 5.97 for in-hospital, aHR = 5.26 for 28-day; Central-III: aHR = 3.35 for in-hospital, aHR = 4.18 for 28-day). Restricted cubic spline revealed a consistent nonlinear AG-mortality relationship. Receiver operating characteristic analysis confirmed AG's significant discriminatory power and identified optimal cutoff values in both populations. This first binational (US-China) multicenter analysis confirms that elevated AG levels independently predict increased mortality in critically ill patients with gastrointestinal perforation. The consistent findings across Western and Asian intensive care unit populations underscore AG's robustness as a readily available prognostic marker in this high-risk patient group.