Integrated arterial blood gas and clinical risk model for mortality prediction in the pediatric intensive care unit
Raid M. R. Umran, Husam N. JawadAbstract
Background:
Early risk stratification of critically ill children remains challenging in pediatric intensive care units (PICUs). Arterial blood gas (ABG) analysis provides rapid bedside assessment of acid–base status, ventilation, oxygen transport, and tissue perfusion. This study evaluated the prognostic value of early ABG abnormalities and developed an integrated physiological-clinical risk (IPCR) model to predict PICU mortality.
Subjects and Methods:
This retrospective cohort study included 432 children aged 1 day–14 years who were admitted to a mixed medical–surgical tertiary PICU between June 2023 and June 2025. ABG parameters obtained within 60 min of admission were analyzed. Multivariable logistic regression was used to develop the IPCR model, and discrimination, calibration, and internal validation were assessed.
Results:
The overall PICU mortality rate was 64.8% (280/432). Nonsurvivors had higher lactate and PaCO 2 levels and lower arterial pH and ctO 2 values than survivors. In the adjusted model, lactate, ctO 2 , sepsis, invasive mechanical ventilation, and vasoactive drug use were associated with mortality. The model showed excellent discrimination, with an apparent AUC of 0.908 (95% confidence interval: 0.875−0.935) and optimism-corrected AUC of 0.90. At the optimal cutoff of 0.691, sensitivity was 81.4% and specificity was 88.8%.
Conclusions:
The Integrated ABG and Clinical Risk Model demonstrate strong predictive performance for PICU mortality. This integrated model may support early risk stratification in high-acuity PICU settings. External validation is required before routine clinical implementation.