WIP 82 - Reducing Diagnostic Error in Pediatric Sepsis

Publication Number: WIP 82.7431

Taylor Patterson, Seattle Children's, Seattle, WA, United States; Lori Rutman, Seattle Children's, Seattle, WA, United States; Yasaman Fatemi, University of Washington School of Medicine, Seattle, WA, United States

Background: Pediatric sepsis is a leading cause of death in children. Early diagnosis of sepsis can be challenging as the signs and symptoms can be non-specific. In 2024, the Society of Critical Care Medicine Pediatric Sepsis Definition Task Force developed the Phoenix criteria which can be applied retrospectively to more accurately identify true sepsis cases. Sepsis is one of the top 3 areas of diagnostic error (DE) and there is a critical need to decrease DE in sepsis.

Objective: We aim to retrospectively assess hospitalized patients with sepsis (per Phoenix criteria definition) for potential DE and categorize the types of DE identified. This study will serve to better understand the diagnostic process surrounding sepsis recognition with the goal of identifying potential interventions to improve sepsis diagnosis, management, and outcomes. Understanding where DEs occur can illuminate potential improvement points in the diagnostic process such as electronic health record tools or clinical pathways.

Design/Methods: We will retrospectively identify cases that met Phoenix sepsis criteria at a quaternary care children’s hospital over a 12-month period (excluding ED and ICU settings). The study will undergo IRB review. We will use the modified SaferDx instrument to assess cases for potential DE. Sepsis episodes will be reviewed 72-hours prior to the earliest point of sepsis recognition as defined by the Improving Pediatric Sepsis Outcomes (IPSO) collaborative. Two study team members will review cases using the SaferDx instrument, discordant scores will be resolved via consensus, and inter-rater reliability will be calculated. This review will be completed by January 2025. Sepsis cases deemed to have DE will be categorized using Diagnostic Error Evaluation and Research (DEER) taxonomy to localize DEs to specific points of the diagnostic process. Statistical analysis will compare cases with and without DE to compare outcomes (in-hospital mortality, post-sepsis diagnosis length of stay, and ICU transfer) with multivariable regression.