Evaluation Of A Risk Prediction Model For Multi-drug Resistant Organisms Infection After Liver Transplantation

?Objectives?This study aimed to evaluate the risk prediction model for multi-drug resistant organisms infection after liver transplantation?We collected and analyzed prospectively the data of patients who underwent liver transplantation in a liver transplantation center of the first people's hospital and Renji hospital affiliated to Shanghai Jiao Tong University from October 2016 to September 2017.Then,To evaluate the validity and stability of the risk prediction model for multi-drug resistant organisms infection after liver transplantation.To provide an effective prediction tool to prevent and control multiple drug-resistant organisms infections and to reduce the incidence and improve the survival rate of patients after liver transplantation.?Methods?A cohort study was conducted to collect and analyze the clinical data of liver transplant recipients from two Hospital in shanghai from October 2016 to September 2017.First,univariate analysis and multivariate logistic regression analysis were used to determine independent risk factors.Then,according to the model we divided the patients into three groups as into low risk group(0~3),moderate risk group(3.1~6),high risk group(>6).Then according to whether patients with multiple drug-resistant organisms infections during hospitalization,compare the predicted infection rate and the actual infection rate.The Hosmer-Lemeshow test and O / E values were used to evaluate the compliance of the model in predicting the risk of infection.The C-statistics(area under the ROC curve)were used to evaluate the discriminant validity of the model in predicting the risk of infection.?Results? 1.General backgrounds According to the inclusive criteria,totally 114 patients were collected in our study.The rate of multi-drug resistant bacteria infection after liver transplantation was 24.6%(28 cases).Among them,54 patients after liver transplantation in the A Hospital had 33.3%(18 cases)of multidrug-resistant bacteria after liver transplantation;60 patients from B Hospital had multi-drug resistant infection rate after liver transplantation 16.7 %(10 cases).In-hospital mortality was significantly higher in the infected group than in the non-infected group.Among the total patients in the two liver transplantation centers,20 cases(71.4%)were multi-drug resistant Gram-negative bacteria,of which Acinetobacter baumannii was the most common.Multi-drug resistant Gram-positive bacteria was 8 cases(28.6%),Staphylococcus aureus was the most common.The most common site of infection was pulmonary infection,accounting for 57.0%.Followed by bloodstream infections accounted for 25.0%.2.Analysis of risk factors Univariate analysis showed that 10 factors were potential risk factors for multi-drug resistant organisms infection after liver transplantation,including blood loss?3L,erythrocyte input> 8U,ICU indwelling time ?10 days,Mechanical ventilation time > 24 h,indwelling endotracheal intubation time ? 72 h,indwelling catheter time ? 5 days,indwelling gastric tube time ? 4 days,resumption of diet,reoperation and thoracentesis.There was no significant difference in age,gender,ascites,upper gastrointestinal bleeding,diabetes,hypertension,serum creatinine,serum bilirubin,biliary complications and indwelling deep venous catheter for more than 14 days(P> 0.1).Multivariate logistic regression analysis showed the following four factors: prothrombin time prolongation(OR: 3.102,95% CI 0.066-9.024,P = 0.038),ICU indwelling time?10 days(OR: 3.745,95% CI 1.313-10.687,P = 0.014),indwelling endotracheal intubation time ? 72h(OR: 8.972,95% CI 1.983-40.582,P = 0.004)and thoracentesis(OR: 3.564,95% CI 1.223-10.387,P = 0.020)were the independent risk factors for multi-drug resistant organisms infection.3.Evaluating the risk prediction model of infection The actual infection rate of multiple drug-resistant bacteria was 24.56% in 114 liver transplant recipients,while the model predictive infection rate was 20.2%,the O / E value was 0.822,and the P value of Hosmer-Lemeshow good-of-fit test's P value was 0.944.The results suggest that this model has satisfactory prediction ability with multi-drug resistant bacterial infection after liver transplantation.According to the results of risk prediction model,the patients were divided into low-risk group,middle-risk group and high-risk group.The O/E values of each group were 0.510,1.041 and 1.427,respectively.It was suggested that this model has good accuracy in predicting the multi-drug resistant bacterial infection in the middle-risk group after liver transplantation,but overestimates the actual infection rate in the low-risk group and underestimates the actual infection rate in the high-risk group.The C-statistic of this the risk prediction model was 0.757,suggesting that the model performed well in discriminant validity of the predictive power of multidrug-resistant bacterial infections in this study.?Conclusion?Multiple drug-resistant bacterial infections are serious complications that affect the prognosis of patients after liver transplantation.Prothrombin time,ICU indwelling time ? 10 days,tracheal intubation time ? 72 hours and thoracentesis were independent risk factors for MDRO infection after liver transplantation.Although there are deficiencies in the prediction ability of low-risk group and high-risk group.The calibration and discrimination of risk prediction models for multidrug-resistant organisms infection after liver transplantation all performed well.The results show that the model performs well for predicting the accuracy of multiple drug-resistant bacterial infections and provides a scientific basis for clinical work.Therefore,the prediction and prevention of multi-drug resistant bacterial organisms should be carried out.