Construction and application effect of the risk nomogram model of deep vein thrombosis in patients with continuous renal replacement therap

Guo Miao; Wang Zheng; Li Gang; Feng Xicui; Zhao Jing; Qiaol Li

doi:10.13498/j.cnki.chin.j.ecc.2024.06.11

Abstract：Objective To observe the clinical application effect of a nomogram model for predicting the risk of deep vein thrombosis (DVT) in ICU patients undergoing continuous renal replacement therapy (CRRT). Methods The clinical data of 300 patients who were admitted to the ICU of Peking University International Hospital from April 2022 to April 2024 were retrospective reviewed and randomized into training set (n=225) and validation set (n=75). The training set was further divided into a DVT group (patients who developed DVT) and a non-DVT group. Univariate and binary logistic regression analyses were applied to identify independent risk factors for DVT after (CRRT),and the nomogram risk prediction model was constructed. Then,the nomogram model was validated in the validation set,and the receiver operating characteristic (ROC) curve,goodness of fit test,calibration curve and clinical decision curve of the model were evaluated respectively. Results 56 (24.89％) of DVT occurred in the training set. In the validation set,19 patients (25.33％) developed DVT. Significant differences between the DVT and non-DVT groups in the training set were observed in terms of BMI,D dimer,HGB,Plt,Ca2+,puncture orientation,ultrasound guidance,and Caprini risk assessment model scores(P<0.05). Univariate analysis revealed that D-dimer,hemoglobin concentration and Caprini risk assessment model were risk factors for DVT,while platelet count,blood calcium concentration and ultrasound guidance were independent protective factors. The calibration curve showed that the slopes for both the training set and validation sets were close to 1. The ROC curve results showed that the AUC of the nomogram prediction model in the training and validation sets were 0.99 (95％CI:0.99-1.00) and 0.99 (95％CI:0.98-1.00),respectively,with cut-off values of 0.353 for both. The DCA decision curve of the DVT risk nomogram model after CRRT showed higher high-risk thresholds. Conclusion The nomogram model for predicting the risk of secondary DVT in ICU patients undergoing CRRT is based on the results of D-dimer,hemoglobin concentration,platelet count,blood calcium concentration,ultrasound guidance,and the last Caprini risk assessment before CRRT. Its clinical application can help CRRT patients achieve higher positive net benefits.