Construction of a Predictive Model for Unplanned Withdrawal in Severe Pancreatitis Patients Undergoing CRRT Treatment

LI Linli; YANG Jiahui

doi:10.3639/j.issn.2095-3097.2025.03.004

Abstract：Objective To explore the influencing factors of unplanned withdrawal in patients with severe pancreatitis undergoing continuous renal replacement therapy(CRRT),and to construct a nomogram prediction model for the risk of un-planned withdrawal in patients,providing a basis for the development of intervention measures.Method A retrospective col-lection of clinical data was conducted on 85 patients with severe pancreatitis who received CRRT treatment at Jiangsu Provincial People's Hospital from January 2021 to January 2024.Patients were divided into unplanned and planned groups based on the occurrence of unplanned episodes(n=32 and n=53,respectively).Univariate and multivariate Logistic regression models were used to analyze the influencing factors of unplanned termination of CRRT in patients with severe pancreatitis,and a nomogram risk prediction model was constructed.The receiver operating characteristic(ROC)curve was drawn to verify its predictive efficacy.Results The proportion of patients using heparin anticoagulation,experiencing vascular access dysfunction,having mechanical ventilation,and having a higher hematocrit was all higher in unplanned group patients than in planned group patients.Blood flow velocity and prothrombin activity were lower in unplanned group patients than in planned group patients,with statistical significant differences(P<0.05).The results of multiple Logistic regression analysis showed that the use of heparin anticoagulation,vascular access dysfunction,mechanical ventilation,and high hematocrit were independent risk factors for unplanned abortion in patients with severe pancreatitis undergoing CRRT treatment(OR>1,P<0.05).High blood flow velocity and prothrombin activity were protective factors for unplanned abortion in patients with severe pancreatitis undergoing CRRT treatment(OR<1,P<0.05).The nomogram risk prediction model was constructed,and the verification model showed that the C-index value was 0.968,with good discrimination.The standard curve showed that the calibration curve was similar to the Y-X line,and the model accuracy was good.The predictive efficacy of the risk model was verified.The results showed that the AUC of the nomogram risk prediction model for evaluating unplanned CRRT in patients with severe pancreatitis was 0.968,the 95%confidence interval of AUC was 0.937-0.999,P<0.05,the sensitivity was 0.969,the specificity was 0.906,and the Youden index was 0.875,which had good predictive efficacy.Conclusion The use of heparin anticoagulation,the occurrence of vascular access dysfunction,mechanical ventilation,blood flow velocity,prothrombin activity,and hematocrit are all influencing factors that affect the unplanned occur-rence of CRRT treatment in patients with severe pancreatitis.The nomogram risk model constructed based on these factors can predict the risk of unplanned withdrawal of CRRT treatment in patients.