1333. Utility of Admission Procalcitonin Level in Patients Presenting to the Hospital with Bloodstream Infection: Real-World Evidence from 250 US Hospitals

Background Serum procalcitonin (PCT) may aid in early detection and treatment of bacterial bloodstream infections (BSI), yet evidence for this indication is inconclusive. We leveraged real-world data to examine biological variability in PCT across host and pathogen factors and its utility for ruling out BSI on admission. Methods PCT measurements within 24 hours of admission were examined in patients presenting with monomicrobial BSI to 250 hospitals in the Cerner Healthfacts Database. The reliability of admission PCT for ruling out BSI at hospital presentation was assessed using two different thresholds (<0.5 and <0.25ng/mL) and then stratifying results by presence vs. absence of sepsis (using CDC Adult Sepsis Event criteria), fever or hypothermia vs. normothermia, various presumed sources of BSI, and organism taxon. Results Between 2007 and 2017, PCT was measured on admission in 4,358/42,465 (10.3%) adults with BSI present on admission at 60 hospitals. Of these, 870 (20%) met CDC surveillance criteria for sepsis. The median admission PCT was 4.89 [0.93, 23.98] and varied by taxon, BSI source, patient temperature, and the presence and severity of sepsis; acute illness severity was the greatest driver of high PCT levels (Fig 1). Using a threshold of ≥ 0.50 ng/mL, the sensitivity of PCT for detection of BSI was 84% for all patients. Notably, BSI without sepsis was 4-fold more likely to yield a false negative PCT (<0.5ng/mL) than bacteremic sepsis. Sensitivity ranged from 77% with normothermia to 83% with fever/hypothermia (P = 0.06), between 81 and 88% across sources of BSI (P = 0.13) and more widely between 64 and 91% across taxa (P = 0.02). Enterococcal BSI was 2- and 4-fold more likely to have a falsely negative PCT than S. aureus or S. pneumoniae BSIs, whereas non-glucose fermenters other than P. aeruginosa had a 2 and 3-fold higher likelihood of being missed compared with P. aeruginosa and Enterobacteriaceae BSIs respectively (Fig 2). Pathogen-level variation in PCT sensitivity was also observed for BSI without sepsis (62–90%; P = 0.02) and upon using a stricter rule-out threshold of <0.25 ng/mL (P = 0.01). Conclusion PCT levels and the reliability of this test for ruling out bacteremia at hospital presentation varies by pathogen, presenting signs, and presence vs. absence of sepsis. Disclosures All authors: No reported disclosures.