This study aims to assess the performance of biomarkers used for the prediction of bacterial, viral, and fungal infection in immunocompromised children upon presentation with fever.
We performed a literature search using PubMed and MEDLINE and In-Process & Other Non-indexed Citations databases. Cohort and case–control studies assessing biomarkers for the prediction of bacterial, viral, or fungal infection in immunocompromised children vs. conventional microbiological investigations were eligible. Studies including adult patients were eligible if pediatric data were separately assessable. Data on definitions used for infections, fever, and neutropenia and predictive values were collected. Risk of bias was assessed with the Quality Assessment of Diagnostic Accuracy Studies-2 tool.
Fifty-two studies involving 13,939 febrile episodes in 7,059 children were included. In total, 92.2% were in cancer patients (
There is great heterogeneity in the biomarkers studied and cutoff values and definitions used, thus complicating the analysis. Literature for immunocompromised children with non-malignant disease and for non-bacterial infection is sparse. Literature on novel diagnostics was not available. We illustrated the challenges of diagnosing fever adequately in this study population and the need for improved biomarkers and clinical decision-making tools.