Rahel ArgawTsegaab Fikre2026-11-132026-11-132026-05-18https://etd.aau.edu.et/handle/123456789/9581Background: Neonatal mortality remains a major public health challenge in Ethiopia despite substantial improvements in maternal and child health services. Early identification of neonates at high risk of death is essential for timely intervention and optimal allocation of limited neonatal intensive care resources. However, standardized and locally validated prognostic tools for predicting neonatal mortality are scarce in Ethiopian neonatal intensive care units (NICUs). Objective: To develop and internally validate a clinical risk assessment score for early prediction of neonatal mortality among neonates admitted to the NICUs of Tikur Anbessa Specialized Hospital (TASH) and Gandhi Memorial Hospital (GMH), Addis Ababa, Ethiopia. Methods: A hospital-based retrospective case–control study was conducted among 570 neonates admitted to the NICUs of TASH and GMH between March 2025 and January 2026. Cases were neonates who died during NICU admission (n = 189), while controls were neonates discharged alive (n = 381). Data were extracted from medical records using a structured data abstraction tool. Bivariable and multivariable logistic regression analyses were performed to identify independent predictors of neonatal mortality. Variables with p < 0.25 in bivariable analysis were entered into the multivariable model. Model performance was assessed using calibration and goodness-of-fit statistics. A simplified clinical risk score was developed from the regression coefficients of significant predictors. Results: Of the 570 neonates included in the study, 33.2% died during NICU admission. Multivariable logistic regression identified meconium aspiration syndrome (MAS) (AOR = 4.10; 95% CI: 1.61–10.40), gestational age of 28–32 weeks (AOR = 5.41; 95% CI: 1.57–18.66), birth weight of 1000–1499 g (AOR = 6.94; 95% CI: 1.88–25.64), and need for bag-and-mask resuscitation at birth (AOR = 4.83; 95% CI: 1.50–15.54) as independent predictors of neonatal mortality. A risk score ranging from 0 to 7 points was developed based on these predictors. The model demonstrated good calibration (Hosmer–Lemeshow χ² = 6.70, p = 0.570) and acceptable predictive performance, explaining 53% of the variability in neonatal mortality (Nagelkerke R² = 0.53). Conclusion: Meconium aspiration syndrome, prematurity, low birth weight, and the requirement for bag-and-mask resuscitation at birth were significant predictors of neonatal mortality among NICU admissions. The developed clinical risk score is simple, practical, and based on readily available clinical variables, making it suitable for bedside risk stratification in resource-limited settings. External validation in diverse healthcare settings is recommended before widespread implementation.enNeonatal mortalityRisk prediction modelClinical risk scorePrematurityLow birth weightMeconium aspiration syndromeNICUEthiopia.Thesis