Defining Posttraumatic Sepsis for Population-Level Research.

Area of Science:

• Critical care medicine
• Trauma surgery
• Health services research

Background:

• Sepsis identification in trauma patients is crucial for outcomes.
• Existing administrative methods for sepsis classification lack comparison with clinical criteria in trauma populations.

Purpose of the Study:

• To compare agreement between 3 sepsis classification methods in critically ill trauma patients.
• To assess the sepsis-associated risk of adverse outcomes using each classification method.

Main Methods:

• Retrospective cohort study of adult trauma patients requiring mechanical ventilation for ≥3 days.
• Compared an automated clinical method (EHR data) with the National Trauma Data Bank (NTDB) and medical billing codes for sepsis classification.
• Analyzed chronic critical illness and in-hospital mortality as primary outcomes.

Main Results:

• Sepsis was identified in 23% by automated criteria, 4% by NTDB, and 17% by billing codes.
• Agreement between the three methods was poor (κ=0.16).
• Automated criteria showed higher adjusted relative risks for chronic critical illness (9.9) and in-hospital mortality (1.3) compared to NTDB and billing codes.

Conclusions:

• Administrative sepsis classification methods misclassify and underestimate sepsis in critically ill trauma patients.
• An automated clinical approach aligns with Sepsis-3 criteria and is feasible.
• Automated sepsis classification may enhance health services and population-based research in trauma.