Performance of presepsin and procalcitonin predicting culture-proven bacterial infection and 28-day mortality: A cross sectional study.

Presepsin is a highly specific biomarker for diagnosing bacterial infections, but its clinical usefulness is not well validated. A retrospective cross-sectional study was conducted. Among the patients suspected bacterial infection or fulfilled the criteria of systemic inflammatory response syndrome (SIRS) and patients who underwent blood culture, presepsin, procalcitonin (PCT), and C-reactive protein (CRP) at the same time were included. Receiver operating characteristic (ROC) curve analysis and logistic regression were used to compare performance of three biomarkers. A total of 757 patients were enrolled, including 256 patients (33.8%) with culture-proven bacterial infection and 109 patients (14.4%) with bacteremia. The 28-day mortality rate was 8.6%. ROC curve analysis revealed that the area under the curve (AUC) of PCT was higher than that of presepsin for both culture-proven bacterial infection (0.665 and 0.596, respectively; = 0.003) and bacteremia (0.791 and 0.685; < 0.001). In contrast, AUC of PCT for 28-day mortality was slower than presepsin (0.593 and 0.720; = 0.002). In multivariable logistic regression analysis, PCT showed the highest ORs for culture-proven bacterial infection (OR 2.23, 95% CI 1.55-3.19; < 0.001) and for bacteremia (OR 5.18, 95% CI 3.13-8.56; < 0.001), while presepsin showed the highest OR for 28-day mortality (OR 3.31, 95% CI 1.67-6.54; < 0.001). CRP did not show better performance than PCT or presepsin in any of the analyses. PCT showed the best performance predicting culture-proven bacterial infection and bacteremia, while presepsin would rather be useful as a prognostic marker.