Evaluation of SOFA-2 Score Performance Across Demographic Subgroups: An External Validation Study Using MIMIC-IV.

Area of Science:

• Critical Care Medicine
• Health Services Research
• Biostatistics

Background:

• The Sequential Organ Failure Assessment (SOFA)-2 score is a validated tool for predicting intensive care unit (ICU) mortality.
• Previous validation did not assess performance across diverse demographic subgroups.
• Evaluating performance across subgroups is crucial for equitable clinical decision-making.

Purpose of the Study:

• To assess the discrimination and calibration of the SOFA-2 score for ICU mortality prediction across demographic subgroups.
• To identify variations in SOFA-2 performance based on age, sex, race/ethnicity, primary language, and insurance status.

Main Methods:

• Retrospective cohort study using MIMIC-IV database (2008-2022).
• Included adult patients' first ICU admission (n=64,015).
• Calculated first-day SOFA-2 scores and assessed discrimination (AUROC) and calibration across subgroups.

Main Results:

• Overall AUROC for ICU mortality was 0.77.
• Discrimination significantly declined with age (AUROC 0.85 for 18-44 vs. 0.72 for 75+).
• Mortality was underpredicted in older patients; discrimination was lower in non-English speakers.

Conclusions:

• SOFA-2 shows good overall ICU mortality prediction but exhibits significant performance variations across demographic subgroups.
• A notable decline in discrimination with advancing age and poorer performance in non-English speakers were observed.
• Routine equity evaluation of clinical prediction tools is essential before widespread implementation to ensure fairness.