124 - Performance of an Epic Pediatric Emergency Department Sepsis Screening Tool

Poster Number: 124
Publication Number: 124.1617

Background: Sepsis is a leading cause of pediatric mortality. Many pediatric emergency departments (PEDs) use sepsis screening tools, but there is limited data on how well they perform in real-world settings.

Objective: The objective of our project was to assess the accuracy of the Epic PED Sepsis Scoring System red alert.

Design/Methods: We conducted a retrospective study of all patients under age 18 who were seen in the PED at a medium-sized academic children’s hospital between May 11, 2022 and May 11, 2023. At our institution, the Epic PED Sepsis Scoring System continuously monitors all patients’ charts and assigns points based on sepsis risk factors. These factors include abnormal vital signs, perfusion, mental status, or the presence of high-risk conditions (Table 1). When this score surpasses a red (high-risk) threshold of 6, a Best Practice Alert (BPA) fires. The BPA notifies the care team after which they can initiate a bedside sepsis huddle. We subsequently recorded if the patient was admitted to the PICU or categorized as presumed sepsis or critical sepsis, as defined by the Improving Pediatric Sepsis Outcomes Collaborative (IPSO Sepsis or IPSO Critical Sepsis, Figure 1). We calculated test characteristics of the screening tool against outcomes of PICU admission and IPSO Sepsis or IPSO Critical Sepsis designations.

Results: There were 22,839 pediatric ED patients seen with 397 distinct red alerts triggered, showing an alert rate of 1.7%. Sensitivity, specificity, positive and negative predictive values were reported against PICU admission, IPSO Sepsis, and IPSO Critical Sepsis designations (Table 2). The Epic PED Scoring System red alert was notable for low to moderate sensitivity (ranging from 15.0% for PICU admission to 69.2% for IPSO Critical Sepsis), but high negative predictive value (ranging from 98.5% for PICU admission to 99.9% for IPSO Critical Sepsis).

Conclusion(s): The Epic PED Sepsis Scoring System at the red alert threshold produces a low alert burden but has inadequate sensitivity as a screening tool when used in isolation. Consequently, our PED also incorporates two other pathways to a sepsis huddle: a triage sepsis screen for early detection, as well as ongoing screening at the lower “yellow” Epic sepsis score threshold followed by a nurse-driven secondary screen. Future research is needed to assess the performance of all three sepsis screening pathways, combined with a bedside sepsis huddle, in the early detection of sepsis in the PED.