Development and External Validation of a Nomogram for Predicting Upper Gastrointestinal Bleeding in Patients After Percutaneous Coronary Intervention: A Retrospective Multicenter Cohort Study

Jingyi Yang,1,* Qifeng Liu,2,* Songnan Wang,3 Yanling Jv,3 Bing Pan,1 Shengxiang Feng,1 Jiani Li,1 Lida Zhang,1 Xuqi Li,2 Hongwei Yu3 1Department of Cardiovascular Medicine, Graduate Training Base, Jinzhou Central Hospital, Jinzhou Medical University, Jinzhou, People’s Republic of China; 2Department of Cardiology, The First Affiliated Hospital of Jinzhou Medical University, Jinzhou, People’s Republic of China; 3Department of Cardiology, Jinzhou Central Hospital, Jinzhou, People’s Republic of China*These authors contributed equally to this workCorrespondence: Hongwei Yu, Department of Cardiology, Jinzhou Central Hospital, Jinzhou, Liaoning, 121000, People’s Republic of China, Email phyhw@163.comBackground: Dual antiplatelet therapy (DAPT) after percutaneous coronary intervention (PCI) increases the risk of upper gastrointestinal bleeding (UGIB) in patients with the acute coronary syndrome (ACS). As UGIB leads to a poor prognosis, it is essential to predict its occurrence early and effectively.Objective: The study aimed to develop and validate a nomogram for predicting UGIB in patients with ACS undergoing DAPT after PCI.Methods: This study was conducted on 1820 patients with ACS receiving DAPT after PCI in Jinzhou Central Hospital from January 2019 to September 2022. A logistic regression analysis was conducted to identify the risk factors of UGIB, which were utilized to develop a model for predicting the probability of UGIB in patients receiving DAPT. A validation cohort was used for verification. The discrimination, calibration, and clinical practicability of the nomogram were verified using receiver operating characteristic (ROC) curve analysis, Hosmer–Lemeshow (H–L) test, and decision curve analysis (DCA), respectively.Results: Age, history of gastrointestinal ulcer/bleeding, heart failure, drinking status, and creatinine were independent risk factors for UGIB and included in our nomogram. The nomogram demonstrated good discriminative ability, with Area under the curve (AUC) values of 0.829, 0.848, and 0.838, respectively. The calibration curve and H–L test indicate that the model has good consistency (P = 0.948, P = 0.777, and P = 0.913, respectively). The nomograms can be clinically beneficial when the threshold probability is > 0.02 in both the training and validation cohorts.Conclusion: Our prediction model can guides clinical physicians in risk stratification of undergoing DAPT patients after PCI by calculating the probability of UGIB. Our study may help clinicians in the early identification of patients at a high risk of UGIB and in providing a personalized treatment and management strategies to reduce the associated adverse outcomes.Keywords: percutaneous coronary intervention, upper gastrointestinal bleeding, risk factor, prediction model, dual antiplatelet therapy