Application of continuous renal replacement therapy (CRRT) in patients with severe acute pancreatitis: an analytical...

To Explore the correlation between Continuous Renal Replacement Therapy (CRRT) and the prognostic outcomes of patients with Severe Acute Pancreatitis (SAP); to analyze the impact of CRRT initiation time on the final prognosis of SAP patients; to evaluate factors affecting the therapeutic effect of CRRT in SAP patients without absolute indications for CRRT, and to develop a multi-factorial predictive model for the efficacy of CRRT in treating SAP.

This retrospective cohort study analyzed clinical data from SAP patients admitted to The Affiliated LiHuili Hospital of Ningbo University (2015-2024), collecting baseline characteristics (demographics, disease severity scores), CRRT parameters (initiation timing, laboratory values), and clinical outcomes. We stratified patients by prognosis and CRRT status, and performed subgroup analyses for CRRT-treated cases. After excluding patients with definitive CRRT indications, we randomly allocated cases to training (80%) and validation (20%) sets. Using logistic regression, we identified CRRT failure predictors, evaluated their predictive value via ROC analysis, and developed/validated a nomogram prediction model.

Among 563 initially screened SAP patients, 282 were included after exclusions. (1) Prognosis analysis revealed significant differences between improved and poor outcome groups in age, pancreatitis type, CRRT use, APACHE II and Marshall scores (all P < 0.05). Multivariate analysis identified CRRT as an independent protective factor and Marshall score as a risk factor. Compared to non-CRRT patients, CRRT-treated patients showed significantly shorter hospitalization and vasopressor duration (P < 0.05), with comparable costs. (2) In CRRT-treated patients (with/without absolute indications), earlier CRRT initiation (< 36 h) correlated with better outcomes (P < 0.05). Multivariate analysis confirmed CRRT initiation time as an independent prognostic factor (optimal cutoff: 36 h). (3) Among 114 CRRT-treated patients without absolute indications (91 in training set), significant differences existed in pancreatitis type, APACHE II, Marshall score, lactate, calcium, albumin, PT and PCT (P < 0.05). Multivariate analysis identified APACHE II, PCT and lactate as independent risk factors, and calcium/albumin as protective factors for CRRT failure. The combined model showed excellent predictive value (AUC = 0.912, 95%CI:0.841-0.982). The nomogram demonstrated good calibration in both training and test sets.

CRRT serves as an independent protective factor against poor outcomes (discharge against medical advice (death within 24 h post-discharge (against medical advice)) /death) in SAP patients, while simultaneously reducing hospitalization duration and vasopressor requirements without increasing financial burden. Early application (within 36 h) demonstrates greater therapeutic benefit. The developed nomogram prediction model, incorporating key prognostic factors, exhibits excellent clinical applicability and provides an objective basis for evaluating treatment timing in patients without strong CRRT indications.