Liver Failure: Cause Analysis And Logistic Regression Analysis For The Factors Affecting Its Prognosis

Objective Genesis describes the etiology of the different types of liver failure, and liver failure, some of the risk factors for prognosis were summarized. Fitting Regression Analyzing the prognosis of patients with liver failure, clinical treatment, the choice of the best treatment options and timing of treatment, in order to improve the prognosis.Methods: In order to ensure the accuracy of the research results, select the clinical data received in December 2009 to April 2015 during liver failure patients, a total of 425 cases as sample data for this study, the use of binary Logistic regression method to draw ROC curve, Research on a number of factors and combination of factors for the prognosis of the dangers of liver failure, and the establishment of prognostic models in order to find the optimal threshold. The clinical data of 48 liver failure reservation, will bring the prognostic value of the regression model, the test reliability prognosis model.Results: Univariate logistic regression analysis showed that the risks affecting the prognosis of different liver failure were involved in patient’s age, level of total bilirubin(TBIL) and direct bilirubin(DBIL), ratio of glutamic oxalacetic transaminase to glutamic-pyruvic transaminase(AST/ALT), content of albumin, blood ammonia and prealbumin, prothrombin(PT), prothrombin activity(PTA), count of white blood cell(WBC) and hemoglobin(HGB), level of creatinine, urea nitrogen and serum sodium as well as concomitant hepatic encephalopathy, upper gastrointestinal hemorrhage, hepatorenal syndrome, spontaneous peritonitis and fluid and electrolyte imbalance.Among them, prealbumin and prothrombin activity elevated liver failure prognosis protective factors than the risk factor. Among the various prognostic factors included in the study does not have multicollinearity. Multivariate Logistic regression analysis of total bilirubin(OR = 0.997, P = 0.004), prealbumin(OR = 1.015, P = 0.000), prothrombin activity(OR = 1.109, P = 0.000), associated with hepatorenal syndrome(OR = 0.008, P = 0.000), water and electrolyte disorders(OR = 0.494, P = 0.007) five factors used to fit the regression model. Fitted regression model P=1/1+e-y,Y=-2.636-0.003X1+0.015X2+0.103X3-4.842X4-0.706X5. The area under the ROC curve 0.863,95% CI(0.828-0.898), the optimal threshold is 0.595, corresponding Y value 0.385. Test the effectiveness of the regression model to predict the patient’s prognosis and compared with the actual prognosis, its credibility was 70.8%.Conclusion Liver failure more common in men, liver and chronic hepatitis B, based on the most common failure. Liver failure have a poor prognosis and high mortality. The prognosis of patients affected by many factors, such indicators can be used for clinical evaluation of the patient’s condition and prognosis. This study using regression model high credibility, the applicable assessment of liver failure in patients with clinical prognosis in the region.