Biases in Sepsis Surveillance and Quality Assurance

This research article investigates the biases in sepsis surveillance and quality assurance stemming from inaccurate coding in inpatient administrative health data (IAHD). The study aims to validate the coding accuracy of sepsis and its associated mortality risk factors through a retrospective analysis of 10,334 inpatient cases from ten German hospitals, treated between 2015 and 2017. The findings reveal that while the ICD-coding of sepsis shows a high positive predictive value, it suffers from low sensitivity, leading to an underestimation of sepsis incidence. The variability in coding accuracy across hospitals is significant, with sensitivity rates ranging from 29% to 71.7%. The study concludes that due to the under-coding of sepsis, previous epidemiological studies have likely underestimated the burden of sepsis in Germany. The research emphasizes that IAHD alone is insufficient for assessing the quality of sepsis care, highlighting the need for improved coding practices and quality indicators.