| dc.description.abstract | Background: Liver cirrhosis remains a major global health problem with high in-
hospital mortality, especially in Southeast Asia. Prognostic models such as Child–Pugh,
MELD, and ALBI have variable accuracy, necessitating simpler and more practical
tools. This study aimed to identify independent predictors of in-hospital mortality
among cirrhotic patients in Indonesia and to propose a novel prognostic function.
Methods: A retrospective cohort study was conducted among 190 cirrhotic patients
admitted to Adam Malik General Hospital, Medan, Indonesia, between 2023 and 2024.
Demographic, clinical, and laboratory variables were collected. Bivariate and
multivariate logistic regression analyses were performed to determine predictors of
mortality. A novel prognostic model was derived from significant factors, and its
performance was evaluated using AUROC and Hosmer–Lemeshow calibration.
Results: Of 190 patients, 79 (41.6%) died during hospitalization. Independent
predictors of in-hospital mortality were grade III–IV hepatic encephalopathy (OR 30.12,
95% CI 10.71–84.67, p<0.001), serum creatinine (ln-transformed, OR 3.23, 95% CI
1.80–5.79, p<0.001), and INR (ln-transformed, OR 17.93, 95% CI 3.54–90.83,
p<0.001). The final prognostic model demonstrated excellent discrimination (AUROC
0.906) and good calibration (Hosmer–Lemeshow p=0.587).
Conclusion: Severe hepatic encephalopathy, renal dysfunction, and coagulopathy were
the strongest predictors of in-hospital mortality among cirrhotic patients. The novel
prognostic function derived from these factors showed high predictive accuracy and
may support risk stratification in clinical practice, particularly in resource-limited
settings.
Keywords: liver cirrhosis, mortality, prognostic model, hepatic encephalopathy,
creatinine, INR | en_US |
