The association between anion gap and in-hospital mortality of post-cardiac arrest patients: a retrospective study

We aimed to determine the association between anion gap and in-hospital mortality in post-cardiac arrest (CA) patients. Extracted the data of patients diagnosed with CA from MIMIC-IV database. Generalized additive model (GAM), Cox regression and Kaplan–Meier survival analysis were used to demonstrate the association between AG levels and in-hospital mortality. ROC curve analysis for assessing the discrimination of AG for predicting in-hospital mortality. Totally, 1724 eligible subjects were included in our study finally. 936 patients (551 males and 385 females) died in hospital, with the prevalence of in-hospital mortality was 54.3%. The result of the Kaplan–Meier analysis showed that the higher value of AG had significant lower survival possibility during the hospitalization compared with the lower-value of AG patients. In the crude Cox regression model, high-level of AG subjects was associated with significant higher HR compared with low-level of AG subjects. After adjusted the vital signs data, laboratory data, and treatment, high-level of AG (group Q3 and group Q4) were also associated with increased risk of in-hospital mortality compared with low-level of AG group, 1.52 (95% Cl 1.17–1.85; P < 0.001), 1.64 (95% Cl 1.21–2.08; P < 0.001), respectively. The ROC curve indicated that AG has acceptable discrimination for predicting in-hospital mortality. The AUC value was found to be 0.671 (95% CI 0.646–0.698). Higher AG levels was associated with poor prognosis in post-CA patients. AG is a predictor for predicting in-hospital mortality of CA, and could help refine risk stratification.