Lack of admission biomarkers' clinical utility in outcomes prediction in patients suspected with infection in the emergency department.

Introduction: Initial procalcitonin (PCT) levels may fail in mortality and septic shock prediction and raise cost-effectiveness issues. Since measurement of lactate, C-reactive protein (CRP), white blood cells and neutrophils is common in the emergency department (ED), we compared prediction abilities of these biomarkers to PCT.

Methods: From January 1st to December 31st, 2018, an observational, single center, retrospective study was conducted in the adult ED of the Reims University Hospital (France). Endpoints were bacteremia, septic shock, and in-hospital mortality, related to the same ED visit.

Results: Over one year, 459 patients suspected with infection were included, of mean age 60.4 years (SD: 22.0), with 50.8% male, and 364 (79.3%) were hospitalized following ED visit. Overall, 45 (9.8%) patients had a bacteremia, 39 (8.5%) a septic shock and 54 (11.8%) died during their hospitalization. PCT and CRP showed the best discrimination for bacteremia, with an area under curve (AUC) of 0.68 for PCT and 0.65 for CRP. PCT and lactate showed similar good discriminative power for septic shock, with an AUC of 0.78 for both, and poor discrimination for in-hospital mortality, with an AUC of 0.62 for PCT and 0.69 for lactate. Systolic blood pressure and pulse oximetry showed similar discrimination for septic shock as PCT or lactate, while they showed higher discrimination for in-hospital mortality than PCT.

Conclusion: Usual admission biomarkers lack clinical utility in predicting septic shock or in-hospital mortality. CRP and PCT are poorly efficient in predicting bacteremia.