Cell population data for early detection of sepsis in patients with suspected infection in the emergency department

Abstract Objectives Traditional biomarkers used for sepsis diagnosis have limited sensitivity and specificity and, so far, are not recommended for sepsis diagnosis. We aimed to evaluate diagnostic accuracy of XN-9000 ® hematology analyzer derived cell population data (CPD) for sepsis. Methods We conducted a cross-sectional cohort study on patients admitted to an emergency department (ED) with a suspicion of infection, having a complete blood count with differential (CBC-Diff). CBC-Diff were performed on XN-9000 ® analyzer (Sysmex, Kobe, Japan). CPD were measured routinely for each CBC-Diff ordered by ED physician. They include: neutrophils-related - Neut-GI and Neut-RI; monocytes-related - Mono-X, Mono-Z, Re-Mono and Mono-Y; IG referring to immature granulocytes; and lymphocytes-related - As-lymp and Re-lymp. Intensive care infection (ICIS) and neutrophile and monocyte (NEMO) scores were calculated using several CPD parameters. Diagnostic performance of each biomarker was computed together with receiver operating characteristic curves for sepsis diagnosis (according to Sepsis-3 definition). Results A total of 1,155 patients with a suspicion of infection were included and 230 had sepsis. Median age was 64 years and 49 % were female. Except for lymphocyte count with an area under the receiver operating characteristic (AUROC) of 0.67 (95 % confidential interval 0.63–0.70), the other CPD exhibited modest performances with AUROC under 0.65. The ICIS and NEMO scores had a modest performance with AUROC of 0.56 (0.52–0.61) and 0.55 (0.51–0.59) respectively. Conclusions None of the biomarkers and scores tested demonstrated sufficient diagnostic accuracy to be recommended for routine sepsis screening in the ED.