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

Area of Science:

• Neonatal Medicine
• Clinical Research
• Biomarker Discovery

Background:

• Neonatal sepsis presents a significant clinical challenge requiring prompt intervention.
• Current prognostic tools for neonatal sepsis outcomes are limited.
• There is a need for reliable predictive models using accessible data.

Purpose of the Study:

• To develop a predictive model for 30-day mortality in neonatal sepsis.
• To utilize readily available laboratory data for risk assessment.
• To enhance clinical decision-making in managing neonatal sepsis.

Main Methods:

• Retrospective analysis of 195 neonates with sepsis (January 2019 - December 2022).
• Identification of independent risk factors using univariate and multivariate analyses.
• Evaluation of the predictive model's performance using receiver operating characteristic (ROC) curve analysis.

Main Results:

• Hemoglobin levels >133 g/L and prothrombin time >16.6 seconds were identified as independent risk markers for 30-day mortality.
• Hemoglobin >133 g/L showed a protective effect (hazard ratio: 0.351, p=0.042).
• Prothrombin time >16.6 s indicated increased mortality risk (hazard ratio: 4.140, p=0.005).
• A predictive model achieved a high area under the curve (0.756).

Conclusions:

• A novel predictive model for 30-day neonatal sepsis mortality was successfully established.
• The model offers objective and accurate risk prediction for individual patients.
• This tool can guide clinical treatment, decision-making, and follow-up strategies.