Risk factors for post-acute pancreatitis diabetes mellitus and construction of a nomogram prediction model
- VernacularTitle:急性胰腺炎新发糖尿病的危险因素分析及列线图模型构建
- Author:
Fujun LI
1
;
Rong ZHANG
1
;
Kun FANG
1
;
Juan CHEN
1
;
Tianshi ZHUANG
1
;
Chao WANG
1
Author Information
- Publication Type:Journal Article
- Keywords: Pancreatitis; Diabetes Mellitus; Risk Factors; Nomogram
- From: Journal of Clinical Hepatology 2026;42(8):1908-1916
- CountryChina
- Language:Chinese
- Abstract: ObjectiveTo investigate the risk factors for post-acute pancreatitis diabetes mellitus （PPDM-A） in patients with acute pancreatitis （AP）， to construct a nomogram prediction model， and to provide a reference for the development of individualized treatment regimens. MethodsA total of 351 patients with AP who were admitted to Xuzhou Municipal Hospital Affiliated to Xuzhou Medical University from June 2021 to January 2025 were prospectively enrolled， and they were randomly divided into modeling group with 246 patients and validation group with 105 patients at a ratio of 7∶3. According to the presence or absence of PPDM-A in the patients with AP， the modeling group was further divided into PPDM-A group with 86 patients and non-PPDM-A group with 160 patients. Clinical data were collected from all patients. The least absolute shrinkage and selection operator （LASSO） regression analysis was used to determine independent variables， and a Logistic regression analysis was used to investigate the influencing factors for PPDM-A. R software was used to construct a nomogram model. The receiver operating characteristic curve was used to assess the discriminatory ability of the model， and the Hosmer-Lemeshow test was used to test the model fitting degree， and the calibration curve was used to evaluate the model consistency； and decision curve analysis （DCA） was used to assess its clinical application value. The independent-samples t test was used for comparison of continuous data between two groups， and the chi-square test was used for comparison of categorical data between two groups. ResultsAmong the 246 patients， 86 developed PPDM-A， resulting in an incidence rate of 34.96%. There were significant differences between the PPDM-A group and the non-PPDM-A group in the proportion of patients with an age of ≥60 years （65.12% vs 40.62%， P<0.05）， male sex （75.58% vs 55.63%， P<0.05）， a body mass index （BMI） of ≥24 kg/m2 （63.95% vs 37.50%， P<0.05）， alcoholic AP （56.98% vs 36.87%， P<0.05）， moderate-to-severe AP （54.65% vs 35.00%， P<0.05）， or a computed tomography severity index （CTSI） score of ≥4 points （48.84% vs 28.75%， P<0.05）， as well as significant differences in the levels of blood calcium （1.46±0.35 mmol/L vs 1.89±0.37 mmol/L， P<0.05） and random blood glucose （Glu） （17.68±4.12 mmol/L vs 11.68±4.27 mmol/L， P<0.05）. The LASSO regression analysis obtained 8 independent variables. The Logistic regression analysis showed that age， sex， BMI， alcoholic AP， moderate-to-severe AP， CTSI score， and Glu were risk factors for PPDM-A （all P<0.05）， while blood calcium was a protective factor （P<0.05）. The model had an area under the ROC curve （AUC） of 0.932 （95% confidence interval ［CI］： 0.903 — 0.962） in the modeling group， and the Hosmer-Lemeshow goodness-of-fit test yielded χ2=7.346 （P=0.728）， the accuracy of model fitting was good； the calibration curve showed that the predicted probability was consistent with the actual probability， indicating that the consistency was good. The model had an AUC of 0.835 （95%CI： 0.753 — 0.917） in the validation group， and the Hosmer-Lemeshow goodness-of-fit test yielded χ2=7.014 （P=0.711）， the accuracy of model fitting was good； the calibration curve showed that the predicted probability was consistent with the actual probability， indicating that the consistency was good. The DCA results of the modeling group showed that the model exhibited a high clinical value in evaluating PPDM-A when the threshold probability was 0.13 — 0.94. ConclusionAge， sex， BMI， alcoholic AP， moderate-to-severe AP， blood calcium， CTSI score， and Glu are influencing factors for PPDM-A. The nomogram model constructed based on the above influencing factors shows good performance in predicting the risk of PPDM-A and can thus provide a reference for developing prevention strategies for PPDM-A in clinical practice.
