Arterial Blood Gas Parameters as Mortality Predictors in Critically Ill Patients Admitted from the Emergency Department: A Prospective Observational Cohort Study

Abstract

Background: Arterial blood gas (ABG) analysis is rapidly available in emergency and intensive care settings and constitutes a core component of many critical illness scoring systems. Although ABG parameters are incorporated into many multidimensional critical illness scoring systems, their independent contribution to mortality prediction remains uncertain because these variables may change rapidly during the early course of critical illness and resuscitation. This study evaluated the prognostic performance of ABG parameters and validated severity scoring systems for predicting 28-day mortality in critically ill patients admitted from the emergency department (ED) to the intensive care unit (ICU).

Methods: In this prospective observational cohort study, 242 adult patients admitted from the ED to the internal medicine ICU between January 2012 and March 2013 were enrolled. Admission ABG parameters (pH, PaCO₂, PaO₂, and HCO₃⁻) and severity scores (APACHE II, APACHE IV, and SAPS II) were recorded. The primary outcome was 28-day all-cause mortality. Prognostic discrimination was assessed using receiver operating characteristic (ROC) curve analysis.

Results: Overall 28-day mortality was 47.1%. Non-survivors had significantly higher APACHE II, APACHE IV, and SAPS II scores than survivors (all p < 0.001). SAPS II demonstrated the best prognostic performance (AUC 0.720; 95% CI 0.659–0.776), significantly outperforming APACHE II and APACHE IV. In contrast, none of the individual ABG parameters independently predicted 28-day mortality (all p > 0.05).

Conclusions: Multidimensional severity scoring systems, particularly SAPS II, provide superior prognostic discrimination compared with isolated ABG parameters in critically ill ED patients. Although ABG analysis remains essential for acute physiologic assessment, isolated blood gas variables appear insufficient as standalone mortality predictors in heterogeneous ICU populations.