Dynamic Nomogram for predicting the severity of acute pancreatitis in adults and Decision Curve analysis: a single-center retrospective study

Objective Acute pancreatitis (AP) is a common and unpredictable disease. Severity stratification and prognostic prediction play an important role in reducing the mortality of AP and improving its prognosis in the early stage. We aimed to establish a clinical prediction model by logistic regression analysis visualized by nomogram. Methods A total of 497 patients with AP in the Nanfang Hospital, Southern Medical University between January 2015 and January 2018 were retrospectively collected. 497 patients were randomly (ratio = 7:3) divided into the training cohort (N = 347) and the validation cohort (N = 147), and both cohorts were divided into non-severe acute pancreatitis (non-SAP) and SAP groups. Univariate and multivariate logistic regression analysis were used to identify the factors associated with the severity of AP. Based on multivariate logistic regression analysis, nomogram was established and validated. The performance, discrimination, and calibration of nomogram were conducted. Decision curve analysis was used to evaluate the net benefits and clinical usefulness of the prediction model. Result Multivariate logistic regression analysis showed that alcohol use, hyperlipidemia, gender, hypertension, ionized serum calcium and serum albumin were independent risk factors for SAP. The individualized nomogram showed good discrimination both in the training cohort (area under the receiver operating characteristic curve [AUC], 0.782) and in the validation cohort (AUC, 0.764) with good calibration. Decision curve analysis demonstrated that in terms of clinical usefulness, the nomogram was found to have some clinical values and can be used in clinical practice. Conclusion The proposed nomogram based on easy-to-obtain features are of high efficacy in predicting SAP patients, which may facilitate clinical decision-making.