RESUMO
BACKGROUND: Acute respiratory tract infections (RTIs) are the most common reason to seek medical care, with many patients receiving inappropriate antibiotics. Novel testing approaches to identify aetiology at the point-of-care are required to accurately guide antibiotic treatment. OBJECTIVE: To assess the diagnostic accuracy of biomarker combinations to rapidly differentiate between acute bacterial or viral RTI aetiology. DATA SOURCES: MEDLINE, Embase and Web of Science databases were searched to February 2021. STUDY ELIGIBILITY CRITERIA: Diagnostic accuracy studies comparing accuracy of point-of-care and rapid diagnostic tests in primary or secondary care, consisting of biomarker combinations, to identify bacterial or viral aetiology of RTI. METHODS: Risk of bias was assessed using the QUADAS-2 tool. Sensitivity and specificity of tests reported by more than one study were meta-analysed using a random effects model. RESULTS: Twenty observational studies (3514 patients) were identified. Eighteen were judged at high risk of bias. For bacterial aetiologies, sensitivity ranged from 61% to 100% and specificity from 18% to 96%. For viral aetiologies, sensitivity ranged from 59% to 97% and specificity from 74% to 100%. Studies evaluating two commercial tests were meta-analysed. For ImmunoXpert, the summary sensitivity and specificity were 85% (95% CI 75%-91%, k = 4) and 86% (95% CI 73%-93%, k = 4) for bacterial infections, and 90% (95% CI 79%-96%, k = 3) and 92% (95% CI 83%-96%, k = 3) for viral infections, respectively. FebriDx had pooled sensitivity and specificity of 84% (95% CI 75%-90%, k = 4) and 93% (95% CI 90%-95%, k = 4) for bacterial infections, and 87% (95% CI 72%-95%; k = 4) and 82% (95% CI 66%-86%, k = 4) for viral infections, respectively. CONCLUSION: Combinations of biomarkers show potential clinical utility in discriminating the aetiology of RTIs. However, the limitations in the evidence base, due to a high proportion of studies with high risk of bias, preclude firm conclusions. Future research should be in primary care and evaluate patient outcomes and cost-effectiveness with experimental study designs. CLINICAL TRIAL: PROSPERO registration number: CRD42020178973.