Nomogram for the prediction of infected pancreatic necrosis in moderately severe and severe acute pancreatitis.

OBJECTIVES

As a serious complication of moderately severe acute pancreatitis (MSAP) and severe acute pancreatitis (SAP), infected pancreatic necrosis (IPN) can lead to a prolonged course of interventional therapy. Most predictive models designed to identify such patients are complex or lack validation. The aim of this study was to develop a predictive model for the early detection of IPN in MSAP and SAP.

METHODS

A total of 594 patients with MSAP or SAP were included in the study. To reduce dimensionality, least absolute shrinkage and selection operator regression analysis was used to screen potential predictive variables, a nomogram was then constructed using logistic regression analysis. The receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA) were used to evaluate the discrimination, accuracy, and clinical efficacy of the model. External data were also obtained to further validate the constructed model.

RESULTS

There were 476, 118, and 82 patients in the training, internal validation, and external validation cohorts, respectively. Platelet count, hematocrit, albumin/globulin, severity of acute pancreatitis, and modified computed tomography severity index score were independent factors for predicting IPN in MSAP and SAP. The area under the ROC curves were 0.923, 0.940, and 0.817, respectively, in the three groups. There was a good consistency between the actual probabilities and the predicted probabilities. DCA revealed excellent clinical utility.

CONCLUSION

The constructed nomogram is a simple and feasible model that has good clinical predictive value and efficacy in clinical decision-making for IPN in MSAP and SAP.