Analysis Of Prognostic Factors And Constr Uction Of A Predictive Model For Prognosis In The Elderly Liver Failure Patients

ObjectiveLiver failure has a high fatality rate,poor prognosis and no definitive treatment at present.The incidence and mortality of liver failure in the elderly are increasing in recent years.Thus,it is very important to evaluate the prognosis of elderly patients with liver failure to improve the survival rates.This study analyzed the independent risk factors affecting prognosis through relevant clinical data of elderly patients with liver failure,and established a prognostic model of elderly patients with liver failure.This can provide an effective assessment tool for elderly liver failure patients and scientific basis of preventing and controlling patient’s condition.MethodsThe clinical data of the 201 elderly patients with liver failure（age≥60 years）in the infectious diseases hospital in Zhenjiang from January 2012 to July 2020 was retrospectively collected according to inclusion criteria and exclusion criteria.The etiology of elderly patients with liver failure was analyzed.The elderly patients with liver failure were divided into improved group（109 cases）and deteriorated group（92 cases）depending on the prognosis.Measurement data are depicted as the mean standard deviation and were compared between two groups using the t-test.Count data are described as percentages,and theχ~2test or Fisher s exact probability test were applied for group comparisons.General data,clinical indicators and complications of the two groups were compared to pre-screen the prognostic risk factors in elderly patients with liver failure.And then,The risk factors with statistically significant differences in the results were further analyzed by binary Logistic regression analysis to obtain the independent risk factors affecting the prognosis of elderly patients with liver failure.The regression coefficient and constant term were determined to establish the prognostic model of elderly patients with liver failure combined with independent risk factors.The receiver operating characteristic curve（ROC curve）was drawn,and the evaluation effect of the constructed prognostic prediction model was evaluated by comparing with the parameters of the traditional prognostic model MELD.Results1.General dataThe results of this study reveal that age and gender between the improvement group and the deterioration group have no statistically significant difference（P>0.05）.Among 201 elderly patients with liver failure,there were 21 cases of acute liver failure,112 cases of acute-on-chronic liver failure,24 cases of chronic liver failure and 44 cases of subacute liver failure,and there was no statistically significant difference（P>0.05）in the types of liver failure.The etiology of elderly liver failure patients is mainly caused by HBV infection（107,53.23%）,followed by drug-induced liver damage（27,13.43%）and alcoholic liver disease（23,11.44%）,and other reasons are relatively few.2.Constructing the prognostic model for elderly liver failure patients（1）The comparison between the improvement and deterioration group showed that the WBC,NH₃,TBIL,ALT,PT,infection,hepatic encephalopathy,upper gastrointestinal bleeding and hepatorenal syndrome were statistically significant differences（P<0.05）.Meanwhile,other indicators were no significant differences（P>0.05）.（2）Multivariate logistic regression analysis showed that TBIL（OR=1.005,P<0.001）,PT（OR=1.185,P<0.001）,hepatic encephalopathy（OR=8.359,P<0.001）and infection（OR=8.344,P<0.005）were independent risk factors of the prognosis of elderly patients with liver failure,and TBIL,PT,infection and hepatic encephalopathy were combined to establish a prediction model.3.Comparison of the new established prediction model and MELD scoreROC results showed that the area under the ROC curve of the prediction model was 0.923（95%CI:0.885～0.961,P<0.001）,the maximum value of Yoden index was 0.736,the sensitivity was 90.1%,the specificity was 83.5%,the diagnostic cut-off value was 0.384,the positive likelihood ratio was 3.77,the negative likelihood ratio was 0.1.By contrast,the area under ROC curve of the traditional MELD scoring model was 0.689（95%CI:0.617-0.761,P<0.001）,yuden index was 0.285,sensitivity was 58.56%,specificity was 70.00%,accuracy was 63.68,diagnostic cut-off value was 11.05,positive likelihood ratio was 1.41,negative likelihood ratio was 0.55.ConclusionsThe etiology of elderly patients with liver failure was mainly HBV infection,and the most patients were chronic acute liver failure.The increase of PT and TBIL and the occurrence of infection and hepatic encephalopathy predicted the poor prognosis of elderly patients with liver failure.By comparison,the established prediction model of elderly liver failure patients in this study has significantly better predictive efficacy than MELD scoring.The prediction effect of the model of elderly liver failure patients is good,and it can be used for the prevention and clinical treatment of elderly patients with liver failure.