Arterial Blood Gas-Derived Risk Models for 28-Day Mortality in Critically Ill Adults: A Retrospective Cohort Study

Area of Science:

• Critical care medicine
• Physiology
• Biochemistry

Background:

• Acid-base disturbances are frequent in critically ill patients.
• These disturbances offer crucial physiological insights and prognostic information.
• Early identification aids in patient management and outcome prediction.

Purpose of the Study:

• To investigate the epidemiology of specific acid-base patterns in critically ill adults.
• To develop predictive models for 28-day mortality using arterial blood gas (ABG) data.
• To assess the prognostic value of ABG-derived parameters and acid-base status.

Main Methods:

• Retrospective analysis of 1150 critically ill adults with acid-base disorders.
• Utilized first arterial blood gas (ABG) results post-ICU admission.
• Developed three multivariable logistic regression models: continuous, threshold-based, and ABG-only, evaluating discrimination (AUC) and calibration.

Main Results:

• Mixed acid-base disorders (35.7%) and respiratory alkalosis were associated with high mortality (73.2% and 69.6%, respectively).
• Non-survivors often presented with a normal/higher pH despite hypocapnia and elevated lactate.
• The ABG-only model demonstrated high discrimination (AUC 0.92), comparable to models including physiological variables (AUC 0.94).

Conclusions:

• Routine arterial blood gas (ABG) analysis and basic clinical data are effective for early mortality risk stratification in ICU patients.
• An apparently normal pH can mask significant metabolic stress, highlighting the importance of integrated acid-base assessment.
• Integrated acid-base assessment is vital for accurate prognostication in critical care settings.