Related Experiment Video
Updated: Jun 27, 2026

Observational Study Protocol for Repeated Clinical Examination and Critical Care Ultrasonography Within the Simple Intensive Care Studies
Published on: January 16, 2019
A Simple Admission-Based Score for Early Mortality Risk Stratification in Non-Shock Sepsis: A Pilot Study
Simona Maria Borta1,2, Romana Olivia Popetiu1,2, Larisa Alexandra Rus1,2
1Department of Internal Medicine, Faculty of Medicine, "Vasile Goldiș" Western University of Arad, Bulevardul Revoluției 94, 310025 Arad, Romania.
None:
Background/Objectives: Early risk stratification in non-shock sepsis remains challenging, as patients may appear clinically stable despite ongoing deterioration. We used key variables across seven pathophysiological domains (cardiovascular, hematological, metabolic, hepatic, renal, immune, and comorbidity burden) to identify admission-based predictors of in-hospital mortality for these patients and derive a simple, clinically applicable triage score. Methods: This prospective pilot study included 182 adult non-shock sepsis patients transferred from the Emergency Department to the internal medicine ward in a tertiary hospital in Arad (Romania). Markers of cardiovascular (atrial fibrillation), renal (creatinine, urea), immune-inflammatory (IL-6, CRP, ESR, procalcitonin, presepsin), metabolic (FBG), hepatic (AST, ALT), and comorbidity burden (CCI) domains were analyzed using logistic and LASSO regression with bootstrap validation. Results: Non-survivors exhibited a significantly higher prevalence of atrial fibrillation (p = 0.012), as well as significantly higher creatinine, urea, IL-6, CRP, ESR, CCI, and presepsin (p ≤ 0.042). Although most variables, (except IL-6) were significant in univariate analysis (p ≤ 0.047), only atrial fibrillation (AOR = 2.31, 95% CI: 1.09-2.84, p = 0.021) and creatinine (AOR = 1.40, 95% CI: 1.00-1.95, p = 0.048) remained independent predictors in the multivariate model. LASSO regression (1000 bootstrap iterations, selection frequency ≥ 80%) confirmed their robustness. Three-parameter models combining atrial fibrillation, creatinine, and inflammatory biomarkers showed good discrimination, with the presepsin-based model achieving an AUC of 0.843, 80.0% sensitivity, 82.9% specificity, and NPV > 95% at the optimal cut-offs (creatinine > 1.49 mg/dL; presepsin > 1446 pg/mL). The Sepsis CORE score demonstrated progressive risk stratification, with mortality rising from 0% to 60.0% across score categories. Conclusions: The proposed score integrates cardiovascular, renal, and immune domains into a simple admission-based tool with high negative predictive value. It may support early triage and risk stratification in non-shock sepsis patients admitted to general wards, although external validation is required.
