Development of an early sepsis treatment-decision algorithm in children and adolescents with cancer in a middle-income country: Results from a multinational modified Delphi consensus.

Abstract

1540 Background: Despite global efforts, striking childhood cancer survival gaps between low- and middle-income countries (LMIC) and high-income countries persist. Based on data from the Colombian childhood cancer clinical outcomes surveillance system, VIGICANCER, sepsis accounts for approximately 90% of all preventable deaths. In response, we developed a consensus-based Treatment-Decision Algorithm (TDA) for early sepsis detection and treatment, adapted to the local context and balancing optimal clinical management and resource utilization. Methods: We used RAND/UCLA Delphi method (preparatory phase, literature review, rating) to consult experts on the appropriateness of the proposed risk/alert definitions, critical steps, evidence-based interventions, and decision-making trees in the adapted TDA. Consensus involved: a) Pilot anonymous voting on a 22-statement online survey (5-point Likert scale, open-ended questions); b) Reading/rating in a hybrid meeting (in-person/virtual); and c) Voting on statements where consensus was not reached. Consensus was defined as: ≥70%, strong; 51%-69%, moderate; ≤50%, no consensus. Results: Preparatory phase and literature review: A multinational (Colombia, Mexico, US), interprofessional panel of 19 members, including from pediatric cancer centers and academic societies (oncology, emergency medicine, hospital medicine, critical care, infectious diseases, nursing) ensured geographic and resource representation. A Colombian internal expert taskforce (n = 6) completed a comprehensive literature review, met biweekly, and developed the Colombian Protocol for Early Sepsis Detection in Children with Cancer and accompanying TDA with 3 domains: sepsis screening, sepsis huddle, and early treatment. Consensus results: a)Pilot(n = 19 members), strong agreement was obtained for 81% of statements in one round; b) Reading/rating (n = 8 members), strong agreement was reached in 90% of statements after two rounds; and c) Voting on statements without consensus (n = 11 members, online), moderate agreement was reached in 100% of statements (2) in one round. Conclusions: Through collaborative consensus, we successfully developed an evidence-based, user-friendly TDA for early sepsis detection and treatment, tailored to resource-constrained settings. The diverse, interprofessional panel facilitated contextual adaptations of the TDA. The proposed TDA provides clinicians serving children with cancer in Colombia with an easy-to-follow TDA that is clear, exhaustive, and suitable for adaptation to individual local settings. Next steps involve applying improvement science methodology to implement the TDA in Colombia and Mexico and evaluating its predictive value for prompt sepsis detection in children with cancer, contributing to reduction of survival gaps in LMIC.