Establishment and verification a nomogram for predicting portal vein thrombosis presence among admitted cirrhotic patients

Background: Portal vein thrombosis (PVT) is an increasingly recognized complication of cirrhosis and possibly associated with mortality. This study aims to evaluate provoking factors for PVT, then establish a concise and efficient nomogram for predicting PVT presence among admitted cirrhotic patients.

Materials and methods: All cirrhotic patients admitted in Hunan Provincial People's Hospital between January 2010 and September 2020 were retrospectively reviewed, the clinical and laboratory data were collected. Multivariate logistic regression analysis and the least absolute shrinkage and selection operator regression method were used for screening the independent predictors and constructing the nomogram. The calibration curve was plotted to evaluate the consistent degree between observed outcomes and predicted probabilities. The area under the receiver operating characteristics curve was used to assess the discriminant performance. The decision curve analysis (DCA) was carried out to evaluate the benefits of nomogram.

Results: A total of 4,479 patients with cirrhosis were enrolled and 281 patients were identified with PVT. Smoking history, splenomegaly, esophagogastric varices, surgical history, red blood cell transfusion, and D-dimer were independent risk factors for PVT in cirrhosis. A nomogram was established with a good discrimination capacity and predictive efficiency with an the area under the curve (AUC) of 0.704 (95% CI: 0.664-0.745) in the training set and 0.685 (95% CI: 0.615-0.754) in the validation set. DCA suggested the net benefit of nomogram had a superior risk threshold probability.

Conclusion: A concise and efficient nomogram was established with good performance, which may aid clinical decision making and guide best treatment measures.