Predicting New-onset Diabetes Following Acute Pancreatitis: The CAPS Score
Аннотация
OBJECTIVES: Postpancreatitis diabetes mellitus (PPDM-A) is a significant endocrine complication of acute pancreatitis, often with a severe disease course. Using the Los Angeles General Hospital acute pancreatitis (AP) cohort, we aimed to identify risk factors for PPDM-A and create a model to predict its occurrence. MATERIALS AND METHODS: This prospectively ascertained cohort included consecutive adults admitted with AP between 2015 and 2022 with no prior history of diabetes mellitus, chronic pancreatitis, or pancreas resection. The primary outcome was PPDM-A, defined as a hemoglobin A1c value ≥6.5% or documented diagnosis of diabetes following index admission for AP. The cohort was divided into training and validation (80:20) samples. In the training cohort, logistic regression with best subset selection was used to construct a series of models to predict PPDM-A. RESULTS: Among 891 patients with AP, 16.5% of whom had moderately severe or severe disease, PPDM-A developed in 62 (7.0%), with a median of 27.8 months (IQR: 15.0-49.3) after index admission. The optimal model included 4 predictors: history of cirrhosis, serum albumin, history of prediabetes, and history of moderate to severe pancreatitis, summarized by the acronym CAPS. It achieved an AUROC of 0.631 (0.454-0.808) in the validation cohort. A score >6 achieved 50% sensitivity and 74% specificity in the validation cohort. CONCLUSIONS: A score using readily available clinical information at index admission predicts diabetes mellitus following acute pancreatitis. This CAPS tool will need to be validated and refined in large cohorts such as the type 1 diabetes in acute pancreatitis consortium (T1DAPC).
Ҳали таржима қилинмаган