Impact of a Nurse-Driven Sepsis Bundle on Early Recognition and Severity Assessment in Neonatal Sepsis

Background: Late-onset neonatal sepsis (LOS) requires early recognition to reduce mortality and morbidity, but clinical signs are often nonspecific. Purpose: We introduced a nurse-driven red-flag–based sepsis bundle in our Level IV NICU in 2021. We assessed its performance in detecting culture-positive LOS by stratifying the severity of illness. Methods: This retrospective observational study included infants 3 days and older who were evaluated for sepsis. Cases were stratified into Green (lowest severity), Amber, Yellow, or Red (highest severity) categories according to clinical and laboratory criteria. We assessed bundle performance by calculating sensitivity, specificity, predictive values, and the area under the receiver operating characteristic curve (AUROC) relative to the Green reference group. Results: Of the 142 evaluated sepsis episodes, the bundle was used in 123 (86.6%). Sensitivity was robust across the severity groups: 85.1% for Amber, 73.7% for Yellow, and 90.0% for Red. However, the specificity was uniformly poor at 13.0%, and the AUROC values remained <0.52 across all categories. Although the bundle consistently identified high-severity cases, it frequently over-identified low-risk infants. Implications for Practice and Research: The nurse-driven sepsis bundle demonstrated high sensitivity for severe LOS, ensuring no critical cases were missed, but lacked discriminative ability due to poor specificity. Integrating additional objective markers and validated diagnostic tools may enhance the specificity and clinical utility of LOS evaluations while supporting nursing autonomy in sepsis recognition. A video abstract summarizing this study is available as supplemental digital content.