A retrospective cohort study included 262 emergency department (ED) patients admitted to the intensive care unit (ICU), divided into non-survivors (n=83, 28-day mortality 31.68%) and survivors. Multivariable Cox regression analysis confirmed that lactate (Lac) and base excess (BE) were independent risk factors for 28-day mortality (both p < 0.05). The combined Lac + BE + MEWS model achieved the highest area under the ROC curve (AUC 0.819, 95% CI: 0.760–0.870), which was significantly better than other models. The BE + MEWS model had an AUC of 0.805 (95% CI: 0.749–0.864) with no statistically significant difference from the Lac + BE + MEWS model (p=0.098). NRI and IDI analyzes showed significant improvement in predictive performance for the BE + MEWS (NRI 85.9%, IDI 0.249) and Lac + BE + MEWS (NRI 83.1%, IDI 0.255; both p < 0.001) models, with BE adding more value to MEWS than Lac. Stratified validation confirmed the best stability and risk stratification of the Lac + BE + MEWS model, especially in intermediate risk patients (MEWS 5–8, AUC=0.812).