Construction Of Risk Prediction Model Of Surgical Site Infection After Liver And Gallbladder Surgery

Objectives: The purpose of this study was to establish the prediction model of surgical site infection risk after hepatobiliary operation,and to provide technical support for screening high risk population of surgical site infection.Methods: In this study,we selected 1145 cases of hepatobiliary surgery in 6 hospitals in China from Jan.2013 to Dec.2015.Five aspects of information were collected including demographic characteristics of patients,combined with basic diseases,operation-related information,use of antimicrobial agents and postoperative infection information.All cases were randomly divided into modeling samples and validation samples in a 7: 3 ratio,the modeling group builded the model and the validation group was used to validate the model.In the modeling group,all variables were analyzed by univariate analysis and statistically significant variables were included in the multivariate Logistic regression analysis and the Back-LR method was used to establish the risk prediction model.In the validation samples,the Hosmer-Lemeshow test was used to assess the compliance of the model.The receiver operating characteristic curve(ROC)and area under the curve(AUC)were used to test the discrimination effect performance.To take the maximum value of Youden index in ROC curve for the best threshold,and the model was validated again..According to the established risk prediction model,the patients were risk scored and risk stratified.The risk stratification of patients: ≥5 points in the high risk group,3-4 points in the medium risk group,0-2 points in the low risk group.Results: 1145 patients were included in the study,143 cases occurred surgical site infection,the total infection rate was 12.49%.Among them,89 cases(7.77%)were superficial incision infection,17 cases(1.48%)were deep incision infection and 37 cases(3.23%)were organ space infection.By modeling samples,8 variables were included in the risk prediction model:hypoproteinemia,hypertension,preoperative biliary tract infection,preoperative abdominal surgery,surgical incision classification,duration of surgery,ASA classification and anastomotic fistula.The model was validated by the validation data and overall sample data.The results of the validation samples: ROC curve analysis showed that the AUC of the risk prediction model was 0.851,which was better than NNIS risk score(AUC was 0.731).The optimal threshold of the ROC curve was 0.139,and the results of the validation samples were as follows: the model sensitivity was 68.29%,the specificity was 86.01%,the positive predictive value was 41.18%,the negative predictive value was 94.98%,the total accuracy rate was 83.79%.which indicated that the model had good judgment ability.According to the model to establish the risk score,surgical incision classification assigned 2 points,anastomotic fistula assigned 3 points,the remaining six variables assigned 1 point.the proportion of patients with high/medium and low risk group were3.42%,13.45%,83.13%,the infection rates were 85.71%,30.91%,6.47%.Conclusions:Hypoproteinemia,hypertension,preoperative biliary tract infection,preoperative abdominal surgery,surgical incision classification,duration of surgery,ASA classification and anastomotic fistula are independent risk factors for surgical site infection after hepatobiliary surgery.The risk prediction model has good reliability and validity established by hepatobiliary surgery clinical data of patients with surgical site infection and has certain application value in surgical site infection risk evaluation and population monitoring.