To identify the risk factors of re-hemorrhage in patients with cirrhotic portal hypertension after treatments, and to construct a nomogram prediction model.

Clinical data of143 patients with cirrhotic portal hypertension treated in Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine from January 2015 to January 2020 were retrospectively analyzed. The informed consents of all patients were obtained and the local ethical committee approval was received. Among them, 77 patients were male and 66 female, aged (56±12) years on average. 109 cases were diagnosed with liver cirrhosis after hepatitis B. Baseline data, medical history and laboratory examination results upon admission were collected. Follow-up time was 3 years. Re-hemorrhage after treatments was regarded as the main outcome event. All patients were divided into the re-hemorrhage (n=32) and non-hemorrhage groups (n=111) according to the incidence of re-hemorrhage. The risk factors of re-hemorrhage were identified by Logistic univariate and multivariate regression analyses. Relevant variables were screened by stepwise regression method. Nomogram prediction model was constructed by R software, and internal verification was carried out by Bootstrap method. Finally, the prediction efficiency and clinical effectiveness of this model were analyzed and validated based on the ROC curve, calibration curve and decision curve analysis (DCA).

Logistic univariate and multivariate regression analyses showed that age, weight, Child-Pugh grading above grade A and surgical treatment were the independent influencing factors of re-hemorrhage in patients with cirrhotic portal hypertension after treatment (OR=1.048, 0.947, 5.980, 0.238; P<0.05). Based on 4 independent influencing factors, the nomogram prediction model for re-hemorrhage was constructed. The area under the ROC curve of this prediction model was 0.806 (0.811 after validation), both of which were >0.75, indicating that this prediction model yielded high discrimination. Brier score was calculated as 0.114 (0.102 after validation), equally <0.25, and R² was 0.437 (0.477 after validation), equally >0.4, suggesting that the model had high calibration. The net benefit of DCA curve of this model was higher than those of two extreme curves, indicating that the prediction model had clinical effectiveness.

Surgery is an important approach to prevent and treat re-hemorrhage in patients with cirrhotic portal hypertension. Constructing nomogram prediction model contributes to identifying high-risk post-treatment re-hemorrhage patients with cirrhotic portal hypertension at early clinical stage.