Abstract: Objective In this study, we built a blood transfusion prediction model for perioperative patients with esophageal cancer to provide reference for patient blood management. Methods We retrospectively collected the data of patients esophageal cancer resection at Fujian Cancer Hospital from January 2008 to December 2020. The data was divided into a modeling set and a validation set in a 7∶3 ratio. Univariable and multivariable logistic regression analysis were performed to screen variables, and a nomogram prediction model was constructed. Receiver operating characteristic (ROC) and calibration curve were used to evaluate the performance of the model. Finally, the clinical utility of the model was confirmed through the decision curve. Results A total of 980 patients were enrolled in the study. Multivariable logistic regression analysis showed that preoperative albumin, hemoglobin, surgical method, and the amount of pleural drainage fluid were independent risk factors for perioperative blood transfusion in patients with esophageal cancer. The nomogram model is established based on risk factors. The area under the ROC of the modeling set and the validation set is 0.712 and 0.706, respectively. The calibration curves of both groups fit well with the standard curve, and the model has good predictive accuracy. Finally, decision curve analysis confirms that the model has good clinical utility. Conclusion We constructed a nomogram prediction model by screening risk factors. It can effectively predict the high-risk population for perioperative blood transfusion in patients with esophageal cancer, provided reference for patient blood management, and improved the rationality, safety, and accuracy of clinical blood use.

Key words: Perioperative blood transfusion, Esophageal cancer, Prediction model, Patient blood management