Abstract / Summary
Background Severe acute pancreatitis (SAP) is characterized by rapid development, complex clinical changes, and a high mortality rate. Early identification of patients with acute pancreatitis that has the potential to become severe is important for optimizing clinical outcomes. However, at present, there is no effective prediction or evaluation tool. This study aimed to determine the ability of the triglyceride–glucose (TyG) index and C-reactive protein-to-albumin ratio (CAR) to predict the risk of progression to SAP and 28-day mortality. Methods This retrospective cohort study included 316 patients with a diagnosis of acute pancreatitis. Information on background characteristics, including sex, age, comorbidities, etiological factors, other relevant variables, and laboratory results was collected. The primary endpoints were progression to SAP and 28-day mortality in patients with SAP. Associations of the TyG index and CAR with the risk of SAP were assessed by logistic regression. Prediction performance was evaluated by receiver-operating characteristic curve analysis and internally validated via 1,000 bootstrap resamples with calibration. Cox regression and Firth’s penalized Cox regression were used to evaluate 28-day mortality. Survival was compared between groups using the Kaplan–Meier method. Results TyG index and CAR values were significantly higher in the SAP group than in the non-SAP group (both p < 0.001). After full adjustment for potential confounders, multivariate logistic regression, the TyG index (odds ratio 2.42, 95% CI 1.36–4.32, p = 0.003) and CAR (odds ratio 1.40, 95% CI 1.12–1.74, p = 0.003) were independently associated with SAP. The apparent areas under the curve for the TyG index and CAR for prediction of SAP were 0.702 and 0.793, respectively; following bootstrap internal validation, the optimism-corrected areas under the curve were 0.701 and 0.793, respectively, with optimism values below 0.05 in both cases and calibration curves demonstrating good agreement between predicted and observed probabilities. Firth’s penalized likelihood Cox regression revealed positive associations of elevated TyG index and CAR values with 28-day mortality in patients with SAP, with respective hazard ratios of 11.90 (95% CI 3.47–50.53) and 10.22 (95% CI 2.31–96.19). Kaplan–Meier survival analysis confirmed that clinical outcomes were significantly worse in patients with SAP and a TyG index ≥12.12 or CAR ≥ 3.72 ( p < 0.001, log-rank test). Conclusion The TyG index and CAR were identified as independent predictors of progression from acute pancreatitis to SAP. High-risk thresholds (TyG index ≥12.12 and CAR ≥ 3.72) showed good prediction performance for 28-day mortality in patients with SAP, highlighting their value for early risk stratification. Nevertheless, these cut-off values require validation in prospective multicenter cohorts.