Abstract:

Objective To construct and verify a predictive model to assess the risk of extracorporeal circuit clotting during continuous renal replacement therapy（CRRT）. Methods A total of 320 patients with CRRT were enrolled from April 2021 to January 2022，and risk factors between a blockage group（222 cases） and a non-blockage group（98 cases） were compared using logistic regression for model construction. The goodness of fit of the model was verified by Hosmer-Lemeshow test. The predictive validity of the model was evaluated by the area under the ROC curve. From February to June 2022，160 patients were recruited for application of the model. Results The incidence of coagulation in this study was 30.6%. The factors that ultimately entered the predictive model were CVVHD mode of treatment（OR=2.482），CVVH mode of treatment（OR=2.724），citrate anticoagulation（OR=3.425），argatroban（OR=3.150），without anticoagulation（OR=9.103），pump interruption（OR=4.114），access outflow dysfunction（OR=2.769），the platelet count（OR=1.005），and APTT（OR=0.859）. The model formula was P=1/[1+exp（-0.866+0.909 × CVVHD+1.002 × CVVH+1.231×citrate anticoagulation+1.147 × argatroban+2.209 × without anticoagulation+1.415 × pump interruption+1.018 × access outflow dysfunction+0.005 × the platelet count-0.152 × APTT）]. The area under the ROC curve was 0.865，while the Youden index was 0.584，with sensitivity of 75.50% and specificity of 82.90%. The model verification results showed the sensitivity of 84.10%，the specificity of 79.31%，and the accuracy of 80.63%. Conclusion The risk prediction model has satisfactory prediction effects. The risk prediction can be completed at the beginning of CRRT，which can provide references for preventative treatment and nursing measures for high-risk patients.

Key words: Continuous Renal Replacement Therapy, Clotting, Risk Prediction Model, Nursing Care

摘要：

目的 建立预测连续性肾脏替代治疗（continuous renal replacement therapy，CRRT）体外循环装置凝血风险的模型，并验证模型的应用效果。 方法 采用便利抽样法，收集2021年4月—2022年1月在常州市某三级甲等医院行CRRT的患者320例，按照是否发生体外循环装置凝血分为凝血组（n=98）和非凝血组（n=222），将两组的各项指标进行对比并应用二元Logistic回归构建预测模型，采用Hosmer-Lemeshow判断模型的拟合度，采用受试者操作特征曲线（receiver operating characteristic curve，ROC）的曲线下面积检验模型预测效果。将2022年2月—6月行CRRT的160例患者作为验证组，验证模型的临床应用效果。 结果 体外循环装置凝血发生率为30.6%，最终纳入模型的预测变量为连续性静脉-静脉血液透析（continuous venovenous hemodialysis，CVVHD）（OR=2.482）、连续性静脉-静脉血液滤过（continuous venovenous hemofiltration，CVVH）（OR=2.724）、枸橼酸抗凝（OR=3.425）、阿加曲班抗凝（OR=3.150）、无肝素抗凝（OR=9.103）、血泵停泵（OR=4.114）、血流量不足（OR=2.769）、血小板计数（OR=1.005）和活化部分凝血活酶时间（OR=0.859）。风险预测模型公式为P=1/[1+exp（-0.866+0.909 × CVVHD+1.002 × CVVH+1.231 × 枸橼酸抗凝+1.147 × 阿加曲班抗凝+2.209 × 无肝素抗凝+1.415 × 血泵停泵+1.018 × 血流量不足+0.005 × 血小板计数-0.152 × 活化部分凝血活酶时间）]。该模型ROC曲线下面积为0.865，最大约登指数为0.584，灵敏度为75.50%，特异度为82.90%。模型验证结果：灵敏度为84.10%，特异度为79.31%，准确率为80.63%。 结论 该研究构建的模型预测效果较好，可为临床识别CRRT患者凝血的发生及实施预防性护理提供借鉴。

关键词: 连续性肾脏替代治疗, 凝血, 风险预测模型, 护理