A-155 Unlocking the Complete Blood Count: Derivation of a Single-Panel Laboratory Test That Includes Monocyte Distribution Width to Create a Universal Sepsis Screening Tool

Abstract

Abstract Background Sepsis is the leading cause of in-hospital death. Outcome improvement requires rapid detection and early antimicrobial therapy. The complete blood count (CBC) is ordered for more than two-thirds of all patients in the emergency department (ED). Individual CBC components have limited utility for sepsis screening due to low specificity and sensitivity. We sought to create a reliable sepsis screening tool by combining components of the CBC differential. Methods In this single-center retrospective cohort study, all adult patients who had a CBC with differential performed within 6 h of ED arrival during a 20-month period were eligible for inclusion. Repeat ED visits by the same patient were excluded. Our primary outcome was manifestation of sepsis within 72 h of ED arrival, detected from timestamped electronic health records using a previously validated electronic phenotype based on Sepsis-3 consensus criteria. Multivariable logistic regression was performed to identify individual components of the CBC differential that were independently associated with our outcome and to identify optimal cutoffs for discretization. Odds ratios were used to assign weights to each component and to derive a 10-point system for scoring likelihood of sepsis. Results 34 864 patients were included; 1306 (3.7%) met sepsis outcome criteria within 72 h of ED arrival and 181 (0.5%) had septic shock. White blood cell (WBC) count, monocyte distribution width (MDW), and the neutrophil to lymphocyte count ratio were found to predict sepsis. As shown in the figure, there was a linear relationship between our CBC infection severity index and likelihood of developing sepsis and septic shock. Patients with a high risk score (8 or above) had a greater than one in four chance of developing sepsis within 72 h. Conclusion A simple ten-point index derived from the most commonly used laboratory panel in medicine reliably estimated likelihood of developing sepsis.