Construction Of A Nomograph Prediction Model For Anastomotic Fistula Risk After Esophageal Cancer

Objective To investigate the independent risk factors of anastomotic fistula after radical resection of esophageal cancer,construct and verify the linear graph prediction model to predict the risk of anastomotic fistula after radical resection of esophageal cancer.Methods Collecting clinical data of patients undergoing radical resection of esophageal cancer.Screening cases according to inclusion and exclusion criteria,and the data of the finally included cases are randomly dividing into training set and test set in a ratio of 7:3.Using the training set to build the prediction model,and the test set to verify the model efficacy externally.Single factor analysis is performed on the training set data to screen the selected variables,single factor logistic regression and multiple factor stepwise logistic regression based on AIC criterion are performed on the selected variables,and Nomogram line diagram analysis is performed on the remaining variables to construct a risk prediction model.Finally,evaluating the performance of the rosette model on the training set and test set by ROC curve and calibration line.Results 817 cases were included,and the incidence of anastomotic fistula was 10.6%（n=87）.Data were randomly assigned to the training set and the test set,and there was no significant statistical difference between the two groups.Multivariate Logistic regression analysis showed that the history of diabetes,Serum sodium,urea nitrogen,the length of ventilation in one lung,intraoperative colloid amount,endoscopic surgery,postoperative albumin,postoperative hemoglobin,postoperative pleural effusion,postoperative pulmonary infection were independent risk factors for anastomotic fistula after radical resection of esophageal cancer.The prediction model was constructed based on the above indicators,and the AUC in the training set and test set were 0.845（95%CI: 0.796～0.894）and 0.788（95%CI: 0.693～0.882）,respectively.the calibration curves of the training set and test set fit well with the standard curve,and the differentiation and consistency of the model were verified internally and externally.Conclusions Multivariable logistic regression successful screening independent risk factors for esophageal cancer effect a radical cure postoperative anastomotic fistula,build the nomogram of risk prediction model by the internal and external validation efficiency is good,help clinical provide individualized prediction for postoperative anastomotic fistula,early adopt effective intervention measures,so as to reduce the incidence of postoperative anastomotic fistula.