To evaluate the non-linear association and threshold effect between serum anion gap (AG) levels on the first day of ICU admission and the risk of delirium during the ICU stay in acute pancreatitis (AP) patients.

This retrospective study utilized the medical information mart for intensive care-Ⅳ (MIMIC-Ⅳ) database, including adult AP patients admitted to the ICU for the first time between 2008 and 2019. The primary exposure variable was the AG value first measured within 24 hours of ICU admission and the primary outcome was the occurrence of delirium during the ICU stay. Multivariate logistic regression models, adjusting for demographic characteristics, disease severity, and other key clinical variables, were used to assess the independent association between AG and delirium. Generalized additive models (GAM) and two-piecewise linear regression models were employed to explore its non-linear relationship and potential threshold.

A total of 890 AP patients were included, of whom 237 (26.6%) experienced delirium during their ICU stay. After multivariate logistic regression models adjustment, the overall AG was not significantly associated with the delirium risk [odds ratio (OR) = 1.028, 95% confidence interval (CI) (0.993, 1.064), P = 0.117]. However, GAM and two-piecewise linear regression analysis revealed a significant non-linear relationship between AG and delirium, with a threshold of 14.0 mmol/L. Specifically, when AG was < 14.0 mmol/L, an increase of 1.0 mmol/L in AG was associated with a 15.5% increased risk of delirium [OR = 1.155, 95%CI (1.018, 1.310), P = 0.025]; when AG was ≥ 14.0 mmol/L, the association was not statistically significant [OR = 1.005, 95%CI (0.963, 1.048), P = 0.827].

Serum AG levels on the first day of ICU admission in AP patients are non-linearly associated with the risk of delirium during ICU stay, with a positive correlation only present when AG is below approximately 14.0 mmol/L. AG, as a routine, convenient, and cost-effective metabolic indicator, has potential value in early delirium risk stratification for AP patients, offering a reference for clinical warning and targeted interventions.