Impact of hourly serial SOFA score on signaling emerging sepsis

Abstract

BackgroundSepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection. Under suspicion of infection, sepsis can be identified as an acute increase in Sequential Organ Failure Assessment (SOFA) score of 2 points or more. Professional critical care societies have called for early detection and treatment of sepsis; however, the fundamental tool to address the need remains unmet. ObjectivesThe present study aims at exploring the possibility of a solution to bridging clinical information system with AI medicine and supporting decision-based medical tasks through integrated data-intensive intelligence. Patients and methodsWe extracted data from a well-recognized database and explored the feasibility of a real-time solution to sepsis identification for adult ICU patients under suspicion of infection. To analogize the requirement of a randomized controlled trial, we adopted propensity score matching to tackle the imbalance of baseline covariates between study groups frequently encountered in observational studies. ResultsOur study indicates that the hourly assessment protocol outperforms the 24-h assessment counterpart in terms of the timing of sepsis identification by a median of 14.5 h earlier and the change of total SOFA score by a median of 1.0 point lower. ConclusionsWe conclude that real-time SOFA score to signal emerging sepsis becomes feasible through the introduction of data-intensive intelligence and data processing technologies.