Application value of constructing a prediction model based on nutritional status combined with clinical data for infectious complications after radical gastrectomy for gastric cancer

Objective To explore the application value of predictive model based on nutritional status combined with clinical data in infectious complications after radical gastrectomy for gastric cancer. Methods The clinical data on 394 gastric cancer patients who underwent radical gastrectomy at Suzhou Ninth Hospital Affiliated to Soochow University from January 2017 to July 2024 were retrospectively collected. According to whether the patients developed infection after surgery, they were divided into infection group (n=73) and non-infection group (n=321) based on whether they developed postoperative infections or not. The least absolute shrinkage and selection operator (LASSO) method was used to obtain the optimal related features associated with infectious complications after radical gastrectomy for gastric cancer. Multivariate Logistic regression analysis was conducted to screen for risk factors of infectious complications after radical gastrectomy. R software was used to randomly select 70% of data as training set (for establishing the nomogram model) and 30% of data as test set. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve, Bootstrap sampling validation, and decision curve analysis (DCA) were used to assess the value of the predictive model. Results of the 394 gastric cancer patients, 73 developed postoperative infectious complications, including 53.42% with lung infections, 17.81% with intra-abdominal infections, 16.44% with urinary tract infections, and 12.33% with wound infections. Age, diabetes mellitus, preoperative nutritional risk, high preoperative PIV, combined organ resection, and prolonged operation time were identified as risk factors for infectious complications after radical gastrectomy (P<0.05). In the training set, the AUC of the nomogram model was 0.898, and the AUC validated in the test set was 0.891. Bootstrap sampling validation with 1, 000 iterations showed that the average errors between the predicted probabilities and actual values in the training and test sets were 0.019 and 0.024, respectively. DCA results indicated that the model had clinical utility. Conclusion Infectious complications after gastrectomy for gastric cancer are common, and the predictive model based on nutritional assessment combined with clinical data has high value in predicting the risk of infectious complications after gastrectomy.