Abstract
The two most familiar scores used for prognostication of liver cirrhosis are the Model for End-stage Liver Disease (MELD) and Child-Turcotte-Pugh (CTP), while the Glasgow-Blatchford (GB) score is used for sorting non-variceal upper gastrointestinal hemorrhage into high- or low-risk categories. This study evaluates the validity of the CTP, MELD, and GB scoring systems in prognosticating the in-hospital outcome of bleeding portal hypertensive patients. In this study, the ROC curve and Younden index determine the efficacy of three scoring systems. The results indicate that CTP was the most efficient score as the predictor of outcome (AUC = 0.9, cut-off value > 7); followed by MELD (AUC = 0.8, cut-off value > 18) and then the GB score (AUC = 0.64, cut-off value > 14) (p < 0.05). In pair-wise comparison, the difference between CTP and MELD was insignificant (p > 0.05). Patients with a CTP score of >7 had notably higher in-hospital mortality (19.8% vs. 0.9%, p < 0.0001). Similarly, mortality with a MELD score > 18 was significant (14.8% vs. 5.9% (p < 0.0001). The GB score was not a good indicator of the outcome. Platelets, albumin, CTP, and MELD scores were the independent contributors to mortality. Thus, as liver cirrhosis prognosticators, CTP and MELD scores can also both be used as predictive scores of the in-hospital outcomes of bleeding patients due to portal hypertension. Compared to the GB score, CTP and MELD scores are fairly efficient predictors in these patients.