Performance Evaluation of Prehospital Sepsis Prediction Models.

Area of Science:

• Emergency medicine
• Critical care
• Predictive modeling

Background:

• Sepsis is a life-threatening condition requiring prompt identification and treatment.
• Prehospital sepsis detection is challenging but crucial for timely intervention.
• Existing prediction models need evaluation for real-world prehospital use.

Purpose of the Study:

• To evaluate the effectiveness of various prediction models for identifying sepsis in prehospital settings.
• To compare model performance against unaided clinical judgment by emergency medical services (EMS) personnel.
• To identify optimal tools for early sepsis detection during prehospital care.

Main Methods:

• A nested case-control study was conducted across four emergency departments (EDs) in Utah.
• Adult, nontrauma patients transported by ambulance with available prehospital records were analyzed.
• Model discrimination was assessed using the area under the precision-recall curve (AUPRC).

Main Results:

• Of 16,620 patients, 1,037 (6.2%) met Sepsis-3 criteria in the ED.
• The Prehospital Early Sepsis Detection (PRESEP) model achieved an AUPRC of 0.33, outperforming unaided EMS assessment (AUPRC, 0.17).
• PRESEP also outperformed the quick Sequential Organ Failure Assessment score and other evaluated predictors.

Conclusions:

• PRESEP was the only model demonstrating superior discrimination for identifying sepsis in prehospital patients compared to unaided EMS assessment.
• The findings suggest PRESEP's potential utility in improving prehospital sepsis identification.
• Further research may refine PRESEP or develop similar tools for enhanced prehospital sepsis care.