THANG ĐIỂM MELD 3.0 TRONG TIÊN LƯỢNG 90 NGÀY BỆNH NHÂN XƠ GAN MẤT BÙ

Abstract

Cirrhosis remains a major cause of morbidity and mortality worldwide. Early prognostication is essential for clinical decision-making and resource allocation in patients with liver cirrhosis. The MELD 3.0 score is a recent update of the original MELD model, optimized to enhance prognostic accuracy, particularly in predicting 90-day mortality. Methods: A prospective descriptive study was conducted on 100 patients diagnosed with liver cirrhosis at Hanoi Medical University Hospital. Clinical characteristics, laboratory parameters, and MELD-based scores (MELD, MELD-Na, MELD 3.0) were recorded and analyzed. The predictive performance for 90-day mortality was assessed using receiver operating characteristic (ROC) curve analysis. Results: The mean age of the study population was 56.9 ± 10.2 years. The majority of patients were male (86%), with a female-to-male ratio of approximately 1:6. The most common etiology was alcohol-related liver disease (61%), followed by hepatitis B (16%), hepatitis C (3%), dual HBV/HCV infection (1%), and other causes (12%) including autoimmune hepatitis, primary biliary cirrhosis, and cryptogenic cirrhosis. The mean MELD 3.0 score was 18.9 ± 5.5. MELD 3.0 demonstrated good prognostic ability for predicting 90-day mortality, with an area under the ROC curve (AUROC) of 0.747 (p < 0.05). At the optimal cutoff point of 19, MELD 3.0 yielded a sensitivity of 92% and specificity of 52%.The AUROC values for MELD 3.0, MELD-Na, and MELD were 0.747, 0.726, and 0.711, respectively. Although MELD 3.0 showed the highest predictive accuracy, the differences among the models were not statistically significant. Conclusions: The MELD 3.0 score provides acceptable predictive performance for short-term mortality in patients with decompensated cirrhosis. While it demonstrates a slightly better discriminative ability than MELD and MELD-Na, further large-scale studies are warranted to confirm its clinical superiority