Validation of a Rapid Host-Protein Score to Discriminate Bacterial from Viral Infections in Hospitalized Febrile...

Background: The MeMed BV^® BV score is a novel, promising host-protein score, differentiating bacterial from viral infections, that combines the expression levels of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), interferon gamma-induced protein-10 (IP-10), and C-reactive protein (CRP). The aim of our study was to determine its diagnostic accuracy in hospitalized febrile children. Methods: A prospective study was performed from December 2023 to April 2024 in two pediatric clinics at "Aghia Sophia" Children's Hospital, Athens, Greece. Patients > 3 months old, presenting with fever, clinical suspicion of acute infection, and symptoms onset up to 7 days prior were considered eligible. Patients with cancer, Human Immunodeficiency Virus (HIV), Hepatitis B Virus (HBV), Hepatitis C Virus (HCV), Tuberculosis (TB), and immunodeficiency were excluded. Two pediatricians reviewed the clinical, laboratory, microbiological, and radiological data and assigned the final diagnosis. The experts were blinded to the BV scores. Results: One hundred and thirty-five patients were enrolled. The predominant medical condition was respiratory tract infection (59.3% lower, 26.7% upper). Twenty-nine (21.5%) patients were diagnosed with bacterial infections. The BV score demonstrated a sensitivity of 96.2%, specificity of 88.7%, and negative predictive value (NPV) of 98.9% for bacterial infections. Equivocal BV scores were reported in 8.9% of cases and were excluded from calculations. The area under the curve was 0.96 (95% CI: 0.93-0.99). A ROC curve analysis was performed, and the optimal cut-off score was estimated at 60, providing a sensitivity of 93.1%, specificity of 88.7%, and NPV of 97.9%. Conclusions: In our study population, the BV score showed high sensitivity and NPV in bacterial infection diagnosis. Further studies are needed to assess the possibility of replacing the "equivocal" range with a narrower one or a specific cut-off value.