Analysis of risk factors of pancreatic portal hypertension and establishment of nomogram model

Abstract

Objective To explore the related risk factors of pancreatic portal hypertension(PPH) and then establish a risk prediction model. Methods The clinical data of 185 patients with pancreatic diseases treated in the First Affiliated Hospital of Army Medical University from January 2004 to January 2022 were collected and analyzed retrospectively. They were divided into study group(n=75) and control group(n=110) according to the occurrence of portal hypertension. The risk factors of PPH were screened by univariate, correlation and multivariate analysis. Based on these variables, a risk prediction model was established by Nomogram. Results In variable screening, the results showed that smoking, history of diabetes and causes were risk factors for PPH(OR=5.008, 3.913, 3.690), and red blood cell count was a protective factor(OR=0.362). A nomogram model of PPH was constructed based on the results of variable screening. The area under the prediction curve of the model was 0.789, and the calibration curve had a good degree of fit, indicating that the model has a good prediction effect. Decision curve analysis showed that a better net gain value was obtained when the risk threshold is between 15% and 80%, and the model had high clinical applicability. Conclusion Our nomogram risk prediction model can effectively identify the high risk factors of PPH and early intervention is of great significance for disease prevention.

Keywords

• pancreatic portal hypertension
• risk factors
• logistic regression analysis
• nomogram model