Evaluating sepsis definitions for clinical decision support against a definition for epidemiological disease surveillance
Epidemiological definitions for sepsis have been used to improve disease surveillance and assist in national assessment of sepsis detection and treatment. However, epidemiological definitions are often optimized for retrospective review and alignment of epidemiological and real-time clinical decision support definitions is unknown. This 15-month retrospective observational study involving 43,046 adult patient encounters and 30,973 unique patients compared cohorts captured by the EHR-based Sepsis-1, Sepsis-3, and Duke Adult Sepsis definitions. We assessed hospital length of stay, in-hospital mortality, admission to the intensive care unit, and sensitivity and specificity of these definitions compared to the Centers for Disease Control and Prevention epidemiological gold-standard definition. Both Duke Adult Sepsis and Sepsis-3 maintained a high specificity, but the former had higher sensitivity. By directly comparing definitions designed for clinical decision support with an epidemiological disease surveillance tool, this work provides guidance to health systems seeking to implement clinical decision support to improve sepsis management.