Establishment of a model for predicting delayed post-polypectomy bleeding: A real-world retrospective study.

Background: Delayed post-polypectomy bleeding (DPPB) is the most common complication which occurs within 30 days after polypectomy, it has become rather common with the widespread of colorectal cancer screening. It is important to clarified predictors of DPPB and identify patients at high risk. Materials and methods: This was a real-world retrospective study based on medical records from The First Affiliated Hospital of Nanjing Medical University. Cases of patients who underwent colonoscopic polypectomy between January 2016 and December 2020 were reviewed to identify risk factors of DPPB. We use the LASSO-Logistic regression analysis model to identify independent predictors and create a predictive model. The model finally got visualized by developing a nomogram. Results: Colonoscopic polypectomy was done on 16,925 patients in our study. DPPB occurred in 125 (0.74%) of these instances. In multivariate analysis, age, sex, hypertension, polyp location, polyp size, and operative modality were found to be independent risk factors and were integrated for the construction of a nomogram. The model's C-index is 0.801 (95%CI: 0.761-0.846). We also found polyps located at the right semicolon and polyp ≥ 1 cm associated with active bleeding under the therapeutic colonoscopy. Conclusion: Young age, male, hypertension, polyp ≥ 1 cm, proximal colon location and operative modality were finally identified as significant predictors of DPPB. We developed and validated a nomogram which performs well in predicting the incidence of DPPB, the model we established can be used as a valuable screening tool to identify patients who are at high risk of bleeding.