Establishment and evaluation of a model for predicting 3-month mortality in Chinese patients with hepatic...

Area of Science:

• Hepatology
• Internal Medicine
• Clinical Prediction Modeling

Background:

• Hepatic encephalopathy (HE) is a severe complication of liver disease, significantly impacting patient prognosis.
• Accurate prediction of mortality in HE patients is crucial for timely intervention and resource allocation.

Purpose of the Study:

• To develop and validate a predictive model for 3-month mortality in Chinese patients diagnosed with hepatic encephalopathy.
• To compare the performance of the new model against established scoring systems like MELD and CTP.

Main Methods:

• A retrospective study involving 609 HE patients, divided into modeling (n=409) and validation (n=200) groups.
• Logistic regression analysis was used to establish the predictive model based on demographic, clinical, and laboratory data.
• Model validity was assessed using Receiver Operating Characteristic (ROC) curve analysis and Youden index.

Main Results:

• The developed model identified red blood cell count (RBC), total bilirubin (TBIL), HE stage, hepatorenal syndrome (HRS), and upper gastrointestinal bleeding as independent predictors of mortality.
• The model achieved an Area Under the ROC Curve (AUC) of 0.931 in the validation group.
• The new model demonstrated superior performance with a higher Youden index, sensitivity (83.1%), and specificity (93.4%) compared to MELD and CTP scores.

Conclusions:

• A novel predictive model incorporating RBC, TBIL, HE stage, HRS, and upper gastrointestinal bleeding accurately predicts 3-month mortality in Chinese HE patients.
• This model offers superior predictive accuracy for HE-related mortality compared to the MELD and CTP scores.
• The findings support the clinical utility of this new model for risk stratification and management of liver disease complications.