1377: Developing a Sepsis Decision-Making Tool for Children With Cancer Using a Modified Delphi Consensus

Introduction: Despite initiatives aimed at improving outcomes for children with cancer, the survival gap between low-and middle- income countries (LMIC) and high-income countries remains substantial. In Colombia, a LMIC, sepsis represents 90% of treatment-related mortality. We aimed to reduce sepsis-related mortality in pediatric cancer patients through the development of a consensus-driven tool for early sepsis detection and treatment tailored to local context. Methods: A decision-making tool was adapted through a structured three phase process over a period of approximately eight months.. Initially, a collaborative interprofessional team including professionals from eight Colombian institutions was established. The committee conducted a comprehensive literature review and convened biweekly to develop the National Protocol for Early-Sepsis Detection in Children with Cancer. In a second phase, a standardization of risk/alert concepts and evidence-based essential steps within a decision-making tool were established. A last step was to convene a multinational and international experts consensus from pertinent professional medical societies in Colombia, United States, and Mexico. The team utilized the RAND/UCLA Delphi methodology. Results: A list of recommended clinical phenotypes, risk/alert criteria, essential steps, evidence-based therapies, and decision-making trees were developed. The expert advisory committee initially evaluated the items using an internet poll and subsequently gathered in a hybrid conference (both in-person and virtual) to review and assess 22 items within the decision-making tool. A consensus (≥80% agreement) was achieved on 21 out of 22 items following two rounds of evaluation. This agreement enabled contextual modifications of the decision-making tool, fostering the creation of user-friendly decision trees. Conclusions: By reaching an interprofessional consensus, we adapted an evidence-based, user-friendly tool for early sepsis identification and treatment that is tailored to our local context. Subsequent steps entail employing quality improvement approaches to validate and implement the algorithm, and assess its predictive value. This strategy supports the goal of reducing treatment-related mortality associated with sepsis by enhancing the medical decision-making process.