Frontiers in Pediatrics (Jun 2014)
Designing a Pediatric Severe Sepsis Screening Tool
Abstract
We sought to create a screening tool with improved predictive value for pediatric severe sepsis and septic shock that can be incorporated into the electronic medical record and actively screen all patients arriving at a pediatric Emergency Department (ED). Gold standard severe sepsis cases were identified using a combination of coded discharge diagnosis and physician chart review from 7,402 children who visited a pediatric ED over two months. The tool’s identification of severe sepsis was initially based on International Consensus Conference on Pediatric Sepsis (ICCPS) parameters that were refined by an iterative, virtual process that allowed us to propose successive changes in sepsis detection parameters in order to optimize the tool’s predictive value based on receiver operating curve (ROC) characteristics. Age-specific normal and abnormal values for heart rate (HR) and respiratory rate (RR) were empirically derived from 143,603 children seen in a second pediatric ED over three years. Univariate analyses were performed for each measure in the tool to assess its association with severe sepsis and to characterize it as an early or late indicator of severe sepsis. A split-sample was used to validate the final, optimized tool. The final tool incorporated age-specific thresholds for abnormal HR and RR and employed a linear temperature correction for each category. The final tool’s positive predictive value was 48.7%, a significant, nearly three-fold improvement over the original ICCPS tool. False positive Systemic Inflammatory Response Syndrome (SIRS) identifications were nearly six-fold lower.
Keywords
