Risk Factors Of Gastrointestinal Fistula After Radical Gastrectomy In Elderly Patients And The Establishment Of A Nomogram Prediction Model

Objective: To investigate the risk factors for the occurrence of gastrointestinal fistula in elderly patients after radical gastric cancer surgery and to establish a visual prediction model,so as to provide a theoretical basis for early prevention of gastrointestinal fistula and formulation of individualized measures by clinical health care personnel.Methods: The clinicopathological data of 843 patients who underwent radical surgery for gastric cancer in Yiji Shan Hospital,were collected between 2018.01 and 2020.10,divided into two groups according to whether or not gastrointestinal fistula occurred at the outcome.Single-factor and multi-factor logistic regression analyses were performed to find independent risk factors,and regression equations and column line graph models were constructed.The Receiver operating characteristic(ROC)curves and calibration curves were plotted to assess the discrimination and calibration of the columnar graph model.The clinical benefit of the model was examined using decision curves.Results: 1.The probability of postoperative gastrointestinal fistula in this study was4.2%.2.Univariate analysis showed that Geriatric nutritional risk index(GNRI)<92,duration of surgery ≥4H,American society of anesthesiology(ASA)≥3,diabetes,history of laparotomy,perioperative Transfusion,and neoadjuvant chemotherapy were statistically different between the gastrointestinal fistula and non-gastrointestinal fistula groups(P<0.05);3.The results of the multifactorial regression analysis were:diabetes(OR=2.871,95% CI: 1.222-6.744,P=0.015),history of abdominal surgery(OR=2.637,95% CI: 1.258-5.529,P =0.010),GNRI ≥92(OR=2.334,95% CI:1.087-5.013,P=0.030),and duration of surgery ≥4H(OR=3.645,95% CI: 1.790-7.420,P<0.001);4.Successful construction of logistic regression prediction models and visual presentation of risk column line graph models;5.Evaluation of the line graph model:the area under the receiver operating characteristic curve(AUC)calculated based on the four predictors was the largest and had the best prediction performance,and the differentiation of the prediction model was: AUC=0.752(95% CI: 0.762-0.833),which were better than each independent predictor;the calibration graph curve was plotted and it was found that the simulated curve fitted well with the actual curve,indicating that the column line graph There was good agreement.Clinical validity: The decision curves based on all predictors of the line plot in this study showed that the net benefit for patients with postoperative gastrointestinal fistula was consistently higher than the other two extreme curves within a threshold probability range of 1% to 47%,indicating the clinical value of the established risk line plot model.Conclusion: In patients aged 60 years and older with gastric cancer,diabetes mellitus,a high risk of geriatric nutritional risk index,longer operative time,and a history of laparotomy may be key factors in causing postoperative gastrointestinal fistula.The clinical risk prediction model established by combining risk factors can help clinical identification and prevention of postoperative gastrointestinal fistula after radical gastric cancer surgery in the elderly.Adequate preoperative preparations including blood glucose adjustment,improved nutrition,and shorter operative time during the treatment process can minimize the occurrence of postoperative gastrointestinal fistula.