Host biomarkers to predict the severity of acute febrile illness: A scoping review

BackgroundAcute febrile illness (AFI) ranges from mild to mortal, yet severity is difficult to assess. Host biomarkers may identify patients with AFI who require a higher level of care; choosing appropriate biomarkers for this role among an expanding pool of candidates is challenging. We performed a scoping review to evaluate the performance of novel host biomarkers to predict AFI severity. MethodsPubMed was systematically searched for manuscripts published January 1, 2013 to February 10, 2018 for studies reporting the association of host biomarker levels and a measure of disease severity among patients with a suspected or diagnosed cause of AFI. Identified abstracts and full text manuscripts were reviewed for eligibility by 2 reviewers. Biomarker performance was evaluated primarily by the area under the curve (AUC) of the receiver operator characteristic to distinguish severe disease. We aggregated data describing biomarker performance by AUC using weighted mean, fixed effects meta-analyses, and random effects meta-analyses. ResultsAmong 2,303 manuscripts identified, 281 manuscripts met criteria for analysis. Data was extracted for 278 biomarkers evaluated in 45,737 participants. Among 89 biomarkers evaluated by [≥]2 studies, there were 6 biomarkers (proadrenomedullin, copeptin, pro-atrial natriuretic peptide, serum triggering receptor expressed on myeloid cells-1, chitinase-3-like protein-1, and the pediatric sepsis biomarker risk model), that showed a weighted mean AUC >0.75 (range 0.75-0.84) in >500 patients over >2 studies. ConclusionsAlthough several biomarkers show promise in predicting AFI severity across multiple studies, their test characteristics do not suggest that they may be used alone to determine AFI prognosis. SummaryA systematic review and limited metanalysis of 89 host biomarkers revealed that most individual biomarkers offer modest performance in predicting the severity of acute febrile illness; several however have performance characteristics which have shown promise in multiple studies.