Serum anion gap predicts lactate poorly, but may be used to identify sepsis patients at risk for death: A cohort study

Purpose: (1) To test whether serum bicarbonate or anion gap can be used to predict elevated lactate or mortality in emergency department (ED) patients with sepsis, and (2) to define thresholds that may predict elevated lactate and mortality. Methods: Retrospective diagnostic-validation study of adults with sepsis treated in a 60,000-visit Midwestern university ED (2010–2015). In the derivation sample, 8 experts selected thresholds based on objective measures to optimize clinical utility. Test performance was reported using likelihood ratios (LR +/−) in the validation cohort. Results: We included 4159 patients. Anion gap predicted lactate > 2 better than bicarbonate [ROC AUC 0.680 vs. 0.609], and anion gap predicted lactate > 4 better than lactate > 2 [ROC AUC 0.816 vs. 0.680]. In the validation cohort, anion gap ≥ 20 mEq/L had LR + for lactate > 2 of 3.670 (2.630–5.122), lactate > 4 of 7.019 (5.310–9.278), and mortality of 2.768 (1.922–3.986). Anion gap predicted mortality similar to lactate > 2 [LR + 2.768 vs. LR + 2.09; LR − 0.823 vs. 0.447]. Conclusions: Anion gap and serum bicarbonate poorly predict changes in lactate and mortality. In resource-limited settings where lactate is unavailable, anion gap ≥ 20 mEq/L may be used to further risk-stratify patients for ongoing sepsis care, but lactate remains a preferred biomarker.