Three-Protein Blood Test Shows High Accuracy in Telling Bacterial From Viral Infections in Children

One of the most persistent dilemmas in pediatric medicine is the febrile child whose cause of fever cannot be determined at the bedside. Bacterial and viral infections can produce nearly identical clinical pictures in young patients, yet they demand entirely different management strategies: antibiotics for the former, supportive care for the latter. Conventional laboratory markers such as C-reactive protein (CRP) and procalcitonin, while widely used, are often non-specific and can be elevated in both types of infection. A new systematic review and meta-analysis published in BMC Pediatrics offers the most comprehensive pediatric-focused evaluation to date of a diagnostic approach designed to resolve this ambiguity, and its findings suggest that a simple blood-based host-protein signature may be able to change antibiotic prescribing practices in children’s hospitals and emergency departments around the world.

The test in question, known as the MeMed BV or MMBV score, measures the relative concentrations of three proteins that the human immune system produces in response to infection: TRAIL (tumor necrosis factor-related apoptosis-inducing ligand), IP-10 (interferon gamma-induced protein 10), and CRP. The underlying biology is elegant. TRAIL and IP-10 are part of the antiviral arm of innate immunity, induced through interferon signaling pathways when the body detects a virus, whereas CRP rises robustly during bacterial challenge as part of the acute-phase response. By integrating the three measurements into a single algorithmic score, the test exploits the distinct immunological fingerprints that bacterial and viral pathogens leave on the host, rather than searching for the pathogens themselves. This host-response approach has a practical advantage over cultures and PCR panels: it does not depend on detecting an organism that may be present in low quantities, may be localized to a difficult-to-sample site, or may take days to grow in the laboratory.

The meta-analysis, conducted by a research team led by Azizah Hendi Aljohany and Muayad Albadrani of Taibah University in Saudi Arabia together with Shyamkumar Sriram of the University of North Texas, is, according to the authors, the first pediatric-focused evaluation of the diagnostic test accuracy of the MMBV score. The team searched PubMed, Scopus, the Cochrane Library, and Web of Science for relevant studies published up to October 2025. After removing 252 duplicates from an initial haul of 407 records and screening the remainder against predefined criteria, the researchers identified 17 entries eligible for inclusion in their review. Two independent reviewers extracted data covering baseline study characteristics, diagnostic accuracy estimates, and antibiotic prescription rates, and all analyses were performed in R version 4.3.3, following the PRISMA reporting framework and the QUADAS-2 quality assessment tool for diagnostic accuracy studies.

The pooled diagnostic performance figures are striking. Across 12 studies that could be combined in a bivariate model, the host-protein signature achieved a summary sensitivity of 89.3 percent, with a 95 percent confidence interval of 82.5 to 93.6 percent, and a summary specificity of 91 percent, with a 95 percent confidence interval of 89.1 to 92.5 percent. In practical terms, the test correctly identified roughly nine out of ten bacterial infections while simultaneously correctly classifying more than nine out of ten viral infections as non-bacterial. The area under the summary receiver operating characteristic curve, a single number that captures the overall discriminative power of a diagnostic test across all possible thresholds, was 0.943, with a confidence interval of 0.88 to 0.96. Values above 0.9 are conventionally regarded as indicating high diagnostic accuracy, and few biomarker-based tests for infection etiology reach this level in pooled pediatric analyses.

For clinicians, the asymmetry between rule-in and rule-out performance matters enormously. A test with excellent sensitivity and a very low negative likelihood ratio can be used to confidently exclude bacterial infection, which is precisely the decision that drives antibiotic prescribing in febrile children. When a clinician can rule out a bacterial cause with high confidence, the rationale for empirical antibiotics largely disappears, and the child can be managed with observation and supportive care instead. The meta-analysis authors emphasize this rule-out capability as the signature’s most clinically valuable property, noting that it is the ability to safely withhold antibiotics, rather than to confirm bacterial disease, that stands to deliver the greatest public health benefit.

The implications for antibiotic stewardship are where the analysis becomes most consequential. Multiple studies included in the review indicate that the signature has significant potential to reduce inappropriate antibiotic use. Modeling estimates cited in the analysis suggest that the test could lower antibiotic overuse in patients with viral infections by as much as 88 percent, with relative reductions ranging from 78 percent to 97 percent across different clinical settings. In one specific model, antibiotic overuse among children ultimately diagnosed with viral infections could fall from 29.6 percent to 9.1 percent, a relative reduction of 69.4 percent. Additional analyses set in emergency departments and inpatient hospital environments support the test’s utility, showing potential reductions in both antibiotic prescriptions and diagnostic errors when the score is incorporated into clinical decision-making.

These numbers should be understood against the backdrop of global antimicrobial resistance. A substantial proportion of antibiotics prescribed to children are given for viral respiratory and febrile illnesses against which they have no effect. Every unnecessary course contributes to the selection pressure that drives resistant bacteria, exposes children to side effects ranging from diarrhea to more serious adverse events, and disrupts the developing microbiome. A diagnostic tool that reliably separates bacterial from viral disease at the point of care therefore functions not merely as a laboratory convenience but as an antimicrobial stewardship intervention. If the pooled accuracy estimates translate into routine practice, the MMBV score could give clinicians the confidence to de-escalate or withhold antibiotics in the majority of febrile children who present without an obvious bacterial focus.

The authors are careful about the limitations inherent in any meta-analysis of diagnostic accuracy studies. The included studies varied in setting, patient age distribution, reference standards used to classify infections, and prevalence of bacterial disease, all of which can introduce heterogeneity into pooled estimates. Diagnostic accuracy studies also tend to be conducted under more controlled conditions than routine clinical care, and real-world performance can degrade when the test is applied to ambiguous cases, mixed infections, or children with underlying conditions. The bivariate modeling approach used by the team accounts for the trade-off between sensitivity and specificity, but the confidence intervals around the summary estimates, particularly for sensitivity, indicate some variability across the evidence base. The review nonetheless provides the strongest synthesis available of how the three-protein signature performs specifically in pediatric populations, a group in which adult-derived accuracy data cannot simply be assumed to apply.

The broader significance of the study lies in the maturation of host-response diagnostics as a field. Rather than chasing pathogens with increasingly sensitive molecular assays, researchers have increasingly turned to measuring the host’s own transcriptional and proteomic response to infection, an approach that captures the physiological consequences of disease regardless of which organism triggered it. The TRAIL/IP-10/CRP signature is among the most advanced examples of this strategy to reach clinical evaluation, and the new meta-analysis consolidates the evidence that it performs at a level that could justify wider deployment. The authors conclude that the analysis confirms the high accuracy of the MMBV host-protein signature for differentiating between viral and bacterial febrile illness in children, and that its excellent rule-out capability can reduce unnecessary antibiotic prescriptions. Whether that potential is realized at scale will depend on implementation studies, cost-effectiveness analyses, and the willingness of health systems to embed a rapid host-protein assay into the first hour of a febrile child’s evaluation, but the evidence assembled here makes a compelling case that the technology is ready for that next step.

Subject of Research: Diagnostic accuracy of the TRAIL/IP-10/CRP host-protein signature for differentiating bacterial from viral infections in children

Article Title: The TRAIL/IP-10/CRP host-protein signature (MMBV Test) in differentiating bacterial vs. viral infection in pediatric populations: a systematic review and meta-analysis of diagnostic test accuracy studies

Article References: Aljohany, A. H., Abualfaraj, T., Sriram, S., & Albadrani, M. (2026). The TRAIL/IP-10/CRP host-protein signature (MMBV Test) in differentiating bacterial vs. viral infection in pediatric populations: a systematic review and meta-analysis of diagnostic test accuracy studies. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07725-2

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07725-2

Keywords: host-protein signature, MMBV test, TRAIL, IP-10, C-reactive protein, pediatric fever, diagnostic test accuracy, meta-analysis, antibiotic stewardship, bacterial infection, viral infection, BMC Pediatrics