New Tools Take Diagnosis of Pediatric Sepsis to Next Level

3 million children worldwide each year. Around 85% of these deaths occur in children under the age of five, and many survivors face long-term complications. Over one-third (34%) of children who survive sepsis exhibit cognitive impairments lasting at least 28 days after hospital discharge, and nearly half require readmission. This overwhelming reaction to infection disproportionately affects lower-resource clinics. In January, for the first time in 20 years, the diagnostic criteria for pediatric sepsis were updated by an international research team. The new system, based on numeric scores of organ dysfunction, demonstrated better diagnostic accuracy than the previous criteria, which focused on inflammatory response. Now, researchers have created tools to apply these updated sepsis criteria consistently and accurately to electronic health records.

To create the diagnostic scoring system called the Phoenix criteria, the research team led by University of Colorado Anschutz Medical Campus (Aurora, CO, USA), gathered data from 10 hospitals across the U.S., Bangladesh, China, Colombia, and Kenya. The Phoenix criteria represent a significant shift in pediatric sepsis diagnosis, aligning with the adult definition centered on life-threatening organ dysfunction and applicable across diverse healthcare settings worldwide. The goal was to create a global diagnostic standard for pediatric sepsis. The next step was to equip researchers with tools to apply the Phoenix criteria to electronic health records without having to manually implement them. The use of the new tools – an R package and Python module – developed by the CU Anschutz team will allow researchers to accurately and consistently calculate pediatric sepsis scores.

Collecting pediatric-specific data was crucial for the project, as sepsis, septic shock, and other conditions often present differently in children than in adults. To create the Phoenix criteria, the team needed an extensive dataset, ultimately collecting over three million patient encounters from contributors around the world. These freely available tools, detailed in JAMIA Open, will reduce errors in research and provide a standardized approach for different research groups using various data formats and programming languages. While these tools are aimed at researchers, the team is also collaborating with a Canadian group to develop a bedside diagnostic tool for clinicians, which would allow for rapid diagnosis of sepsis in children at high risk, a critical next step according to the researchers.

Peter DeWitt, PhD, assistant research professor in the Department of Biomedical Informatics at CU Anschutz said he created the tools “to provide us a way to make sure that if somebody says, ‘Based on this patient’s record, they have sepsis’ it shows that, yes, they do. We can trust that this diagnostic is accurate as opposed to having the question, ‘Are you sure?’ … We know that if this tool is used, then we can trust those conclusions in the research. It’ll give us a good foundation for comparing results, not only within published papers, but across published papers.”