Development and Validation of the Sequential Organ Failure Assessment (SOFA)-2 Score.

Summary

SOFA-2 updates variables and thresholds across six organ systems and demonstrates improved AUROC (0.79 vs 0.77) for ICU mortality using data from 3.34 million ICU encounters across nine countries. Predictive validity persists from ICU day 1 to day 7, though gastrointestinal and immune domains were not incorporated due to insufficient data.

Key Findings

• Across 3.34 million ICU encounters in 9 countries, ICU mortality was 8.1% (270,108 deaths).
• SOFA-2 improved predictive validity for ICU mortality vs. original SOFA (AUROC 0.79 [95% CI 0.76–0.81] vs 0.77 [95% CI 0.74–0.81]).
• Sequential evaluation from ICU day 1 to day 7 maintained predictive performance; gastrointestinal and immune scores were excluded due to insufficient data and content validity.

Clinical Implications

SOFA-2 may enhance risk stratification, endpoint definition, and enrollment criteria in ARDS and sepsis trials; implementation will require EHR mapping and clinician education.

Limitations

• Retrospective EHR-based analysis may entail ascertainment and measurement biases.
• Exclusion of gastrointestinal and immune domains may limit comprehensiveness; AUROC gain is modest.

Future Directions

Prospective validation in diverse ICU settings, calibration across subpopulations (e.g., ARDS phenotypes), and assessment of clinical impact on decision-making and trial outcomes.