|
Constructing nomogram model to evaluate the risk of anticoagulation failure in continuous renal replacement therapy
LIU Qian, ZHANG Ming, SUN Shao-ting, LIU Yi-kun
2025, 24 (04):
278-282.
doi: 10.3969/j.issn.1671-4091.2025.04.004
Objective To explore the nomogram model based on LASSO-logistic regression analysis to evaluate the risk of anticoagulation failure in continuous renal replacement therapy (CRRT). Methods The clinical data of 168 patients who underwent CRRT in our hospital from March 2021 to March 2024 were retrospectively collected by convenient sampling method. According to whether the patients had coagulation during cardiopulmonary bypass within 24 hours, they were divided into coagulation group and non-coagulation group. The risk factors were selected by single factor analysis, then all variables were screened by LASSO regression, and the common variables were combined. Logistic regression analysis was used to analyze the influencing factors, and then nomogram model was constructed, and the performance of nomogram model is verified by using receiver operating characteristic (ROC) curve, calibration curve and Bootstrap method. The decision curve was used to evaluate its clinical utility. Results 50 patients with CRRT had cardiopulmonary bypass coagulation within 24 hours, and the incidence rate was 29.76%. Logistic regression results showed that high platelet count (PLT), high hematocrit (HCT), shortened activated partial prothrombin time (APTT), increased transmembrane pressure (TMP), slow blood flow and pump failure with blood were the risk factors for coagulation during 24h of CRRT cardiopulmonary bypass (OR=2.672, 3.040, 0.435, 2.568, 0.471, 2.662, 95% CI: 1.436~4.974, 1.507~6.133, 0.220~0.861, 1.288~5.119, 0.253~0.874, 1.284~5.519, P=0.002, 0.002, 0.017, 0.007, 0.017, 0.008,respectively). To construct a nomogram model of the risk of coagulation during 24h of CRRT cardiopulmonary bypass. The results of Bootstrap method and ROC curve show that the model has a good discrimination, and the consistency index (C-index) was 0.936 (95% CI: 0.844~0.963), and the area under the curve (AUC) was 0.978 (95% CI: 0.958~0.997). Hosmer-Lemeshoe goodness-of-fit test results show that the model has good fitting effect and accuracy (χ2=2.922, P=0.939), and the calibration curve was close to the ideal curve. Conclusions Coagulation during 24h of CRRT cardiopulmonary bypass is closely related to PLT, HCT, APTT, TMP, slow blood flow and pump stop, and the nomogram model has good predictive value and clinical benefits.
|