Mortality Prediction Model in Sepsis Emergency: Combination of Biomarkers and Clinical Parameters at Grandmed Hospital Lubuk Pakam in 2023

Abstract

Background: Sepsis is a complex clinical syndrome and a life-threatening medical emergency with a high global mortality rate. Worldwide, sepsis accounts for an estimated 11 million deaths annually, equivalent to nearly 20% of all global deaths. In Indonesia, mortality rates in intensive care units remain between 30–60%. One of the major challenges in sepsis management is delayed diagnosis and the difficulty in identifying patients at high risk of death. Conventional prognostic tools such as SOFA or qSOFA scores are widely used, but their predictive accuracy is limited. Biomarkers such as procalcitonin (PCT), C-reactive protein (CRP), and lactate have been shown to provide diagnostic and prognostic information, yet their utility as single predictors remains insufficient. Objective: This study aimed to develop a mortality prediction model for septic patients by combining biomarkers (PCT, CRP, lactate) and clinical parameters (SOFA score) at RS Grandmed Lubuk Pakam in 2023. Methods: An observational analytic study with a prospective cohort design was conducted on 110 adult patients diagnosed with sepsis based on Sepsis-3 criteria. Demographic, clinical, and biomarker data were collected within the first 24 hours of admission. Statistical analysis included bivariate testing and multivariate logistic regression. Model performance was assessed using ROC curves, AUC, sensitivity, specificity, PPV, and NPV.

Results: The overall sepsis mortality rate was 59.1%. Multivariate analysis identified SOFA ≥9 (OR=3.74; p=0.006), PCT ≥10 ng/mL (OR=2.91; p=0.028), and lactate ≥4 mmol/L (OR=4.56; p=0.001) as independent predictors of mortality. The combined model of SOFA, PCT, and lactate demonstrated the highest accuracy with an AUC of 0.91, sensitivity 85%, and specificity 83%, outperforming any single predictor. Conclusion: The integration of SOFA score, procalcitonin, and lactate substantially improves the predictive accuracy of mortality in septic patients compared to conventional approaches. This model may serve as a valuable clinical decision support tool for early risk stratification, although external validation in larger cohorts is required before routine clinical implementation.