Performance Evaluation of Prehospital Sepsis Prediction Models
Authors' conclusion
PRESEP was the only evaluated prediction model that demonstrated better discrimination than unaided EMS infection assessment for the identification of ambulance-transported adult patients who met Sepsis-3 criteria in the ED.
Study context
- Population
- Adult nontrauma patients arriving via ambulance with prehospital care records, treated in 4 Utah emergency departments (EDs) during 2018434 cases · 434 controls
- Comparison
- 21 prehospital sepsis prediction models vs unaided emergency medical services (EMS) infection assessment
- Outcome
- Discrimination (AUPRC) for meeting Sepsis-3 criteria in the ED
Methodological detailsmulticentre
- Multicentre
Main finding
As reported in the abstract
| Group | Area under the precision-recall curve (AUPRC) | AUPRC 95% CI |
|---|---|---|
| PRESEP | 0.33 | 0.27-0.41 |
| Unaided EMS infection assessment | 0.17 | 0.13-0.23 |
| Quick Sequential Organ Failure Assessment score | 0.13 | 0.11-0.16 |
Area under the precision-recall curve (AUPRC)
PRESEP0.33
Unaided EMS infection assessment0.17
Quick Sequential Organ Failure Assessment score0.13
AUPRC 95% CI
PRESEP0.27-0.41
Unaided EMS infection assessment0.13-0.23
Quick Sequential Organ Failure Assessment score0.11-0.16
- PRESEP vs unaided EMS infection assessment: P < 0.001 — PRESEP outperformed
- PRESEP vs quick Sequential Organ Failure Assessment score: P < 0.001 — PRESEP outperformed
Observational data: associations, not effects.
Further findings
| Outcome | PRESEP |
|---|---|
| Sensitivity and PPV of 28 dichotomous predictors of ED sepsis | |
| Sensitivity | 60% |
| PPV | 20% |
- Sensitivity and PPV of 28 dichotomous predictors of ED sepsis — Sensitivity range across predictors
- 6% to 91%
- Sensitivity and PPV of 28 dichotomous predictors of ED sepsis — PPV range across predictors
- 8-100%
Prevalence of Sepsis-3 criteria in the ED among ambulance arrivals
- 1,037 of 16,620 met Sepsis-3 criteria
- 6.2%
Original publication
Performance Evaluation of Prehospital Sepsis Prediction Models. Crit Care Med. Published 12 February 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →