A Host Score Differentiates Bacterial from Viral Infection in Emergency Department Febrile Patients

• Published in: The Journal of emergency medicine 2023-03, Vol.64 (3), p.421-421
• Main Authors: Klein, Adi, Weissman, Alexandra, Arias, Cesar, Ryan, Leticia, Bachur, Richard, Rothman, Richard, Halabi, Salim, Motov, Sergey M., Kaplan, Sheldon
• Format: Article
• Language: English
• ISSN: 0736-4679; 2352-5029
• DOI: 10.1016/j.jemermed.2023.03.020

To study the performance of a host-protein score (BV) for differentiating bacterial from viral infection in US emergency departments (EDs). The distinction between bacterial and viral etiologies when patients present to the ED with acute febrile illness is challenging. A score (0-100) indicative of bacterial versus viral infection comprising TNF-related apoptosis-induced ligand (TRAIL), interferon gamma-induced protein-10 (IP-10), and C-reactive protein (CRP) has been developed, as well as a platform to measure the score in 15 minutes. Patients were recruited prospectively at 7 US EDs and 2 in Israel (Apollo, NCT04690569). Reference standard etiology was adjudicated by an expert panel provided with complete patient characterization. For primary performance analysis, experts blinded to one another, to BV and to procalcitonin (PCT) and CRP, classified every case as bacterial or viral. For secondary performance analysis, experts blinded to one another and to BV classified cases as bacterial, viral or indeterminate; indeterminates were removed from the secondary analysis. The diagnostic accuracy of BV and PCT was assessed in comparison to the reference standards. Study cohort (n=290) comprised 153 adults (median age: 37.7 yrs; interquartile range, IQR 27.1) and 137 children (median age: 4.9 yrs; IQR 6.7). 130 (44.8%) patients presented within 2 days of symptom onset and 63 (21.7%) were hospitalized for a median of 4 days. In the primary cohort, 49 cases (16.9%) were adjudicated bacterial; 241 (83.1%) viral. In the secondary cohort, 35 (12.1%) cases were adjudicated bacterial; 219 (75.5%) viral; 36 (12.4%) indeterminates. Predominant syndrome was upper respiratory tract infection (43.1% in the primary cohort; 43.7% in the secondary cohort). BV attained sensitivity and specificity 83.7% (69.7-92.2%) and 87.7% (82.6-91.4%), respectively, in the primary analysis (n=290; 9.7% equivocal); 91.2% (81.1-10.0%) and 91.0% (87.1-95.0%), respectively, in the secondary analysis (n=254; 7.5% equivocal); and significantly outperformed PCT across both cohorts (p < 0.001; [figure1]). Increasing score was significantly associated with increasing likelihood of bacterial infection for both cohorts (p < 0.001; [table1]). BV accurately distinguishes bacterial from viral infection in ED patients with acute febrile illness, and, using the rapid platform, has potential to facilitate ED treatment and improve antibiotic stewardship.