Game changer or gimmick: inflammatory markers to guide antibiotic treatment decisions in neonatal early-onset sepsis

Background The diagnosis of neonatal early-onset sepsis (EOS) is challenging, and inflammatory markers are widely used to guide decision-making and therapies. Objectives This narrative review presents the current state of knowledge regarding the diagnostic value and potential pitfalls in the interpretation of inflammatory markers for EOS. Sources PubMed until October 2022 and searched references in identified articles using the search terms: neonatal EOS, biomarker or inflammatory marker, and antibiotic therapy or antibiotic stewardship. Content In situations with a high or low probability of sepsis, the measurements of inflammatory markers have no impact on the decision to start or stop antibiotics and are just gimmick, whereas they may be a game changer for neonates with intermediate risk and therefore an unclear situation. There is no single or combination of inflammatory markers that can predict EOS with high probability, allowing us to make decisions regarding the start of antibiotics based only on inflammatory markers. The main reason for the limited accuracy is most probably the numerous noninfectious conditions that influence the levels of inflammatory markers. However, there is evidence that C-reactive protein and procalcitonin have good negative predictive accuracy to rule out sepsis within 24 to 48 hours. Nevertheless, several publications have reported more investigations and prolonged antibiotic treatments with the use of inflammatory markers. Given the limitations of current strategies, using an algorithm with only moderate diagnostic accuracy may have a positive impact, as reported for the EOS calculator and the NeoPInS algorithm. Implications As the decision regarding the start of antibiotic therapy is different from the process of stopping antibiotics, the accuracy of inflammatory markers needs to be evaluated separately. Novel machine learning-based algorithms are required to improve accuracy in the diagnosis of EOS. In the future, inflammatory markers included in algorithms may be a game changer reducing bias and noise in the decision-making process.