A practical predictive model to predict 30-day mortality in neonatal sepsis

Abstract

OBJECTIVE: Neonatal sepsis is a serious disease that needs timely and immediate medical attention. So far, there is no specific prognostic biomarkers or model for dependable predict outcomes in neonatal sepsis. The aim of this study was to establish a predictive model based on readily available laboratory data to assess 30-day mortality in neonatal sepsis. METHODS: Neonates with sepsis were recruited between January 2019 and December 2022. The admission information was obtained from the medical record retrospectively. Univariate or multivariate analysis was utilized to identify independent risk factors. The receiver operating characteristic curve was drawn to check the performance of the predictive model. RESULTS: A total of 195 patients were recruited. There was a big difference between the two groups in the levels of hemoglobin and prothrombin time. Multivariate analysis confirmed that hemoglobin>133 g/L (hazard ratio: 0.351, p=0.042) and prothrombin time >16.6 s (hazard ratio: 4.140, p=0.005) were independent risk markers of 30-day mortality. Based on these results, a predictive model with the highest area under the curve (0.756) was built. CONCLUSION: We established a predictive model that can objectively and accurately predict individualized risk of 30-day mortality. The predictive model should help clinicians to improve individual treatment, make clinical decisions, and guide follow-up management strategies.