PCI术后对比剂肾损伤风险预测模型的建立与验证

Establishment and validation of risk prediction model of contrast-induced kidney injury after PCI

目的建立经皮冠状动脉介入(PCI)术后对比剂肾损伤风险预测模型并进行验证。方法选择2022年1月至2023年11月在我院行PCI术的634例患者为研究对象,以7∶3比例分为模型组443例,验证组191例。在电子病历系统中收集可能影响PCI术后对比剂肾损伤的参数,以术后72 h内发生对比剂肾损伤为阳性结局并将患者分为两组,比较两组患者的基本情况、疾病、手术相关指标及住院期间用药类型,采用LASOO回归筛选变量后行多因素Logisitic回归分析,以此建立列线图模型并进行验证。结果本研究模型组443例患者中发生对比剂肾损伤62例(14.0%)。在LASSO回归基础上行多因素Logisitic回归分析结果显示,急诊PCI、支架数量、对比剂使用量、对比剂类型、水化类型、利尿药为PCI术后对比剂肾损伤的独立性影响因素(P<0.05)。以多因素Logistic回归所获得的独立性影响因素构建预测模型,采用列线图形式展示。模型验证结果显示,区分度:模型组的ROC曲线下面积为0.728,95%CI:(0.661~0.795);验证组的ROC曲线下面积为0.719,95%CI:(0.652~0.786)。准确度:模型组与验证组校准曲线斜率为1,截距为0.000,模型曲线与理想模型基本拟合成对角线,H-L拟合优度检验(P>0.05)。临床有效性分析结果显示,当预测概率阈值在0.15~0.75时使用本研究模型预测PCI术后对比剂肾损伤的净获益最高。结论PCI术后对比剂肾损伤受急诊PCI、支架数量、对比剂使用量等因素影响,本研究建立的预测模型可用于预测PCI术后发生对比剂肾损伤的风险。

Objective To develop and validate a predictive model for the risk of contrast kidney injury after PCI.Methods 634 patient cases who underwent PCI in our hospital from January 2022 to November 2023 were selected for the study.They were divided into 443 cases in the model group and 191 cases in the validation group in a 7∶3 ratio.Parameters that may affect contrast renal injury after PCI were collected in the electronic medical record system,and patients were divided into 2 groups based on the positive outcome of contrast renal injury occurring within 72 h after surgery.The basic conditions,diseases,surgery-related indexes,and types of medications used during hospitalization were compared between the 2 groups,and multifactorial Logisitic regression was performed after screening the variables with LASOO regression,so as to establish a nomogram model and validate the model.Results A total of 62(14.0%)of 443 patients in the model group of this study developed contrast renal injury.The results of multifactorial Logisitic regression analysis performed on the basis of LASSO regression showed that emergency PCI,number of stents,amount of contrast used,type of contrast,type of hydration,and diuretics were the independent influencing factors of contrast renal injury after PCI(P<0.05).The predictive model was constructed with the independence influences obtained from multifactor Logistic regression and presented in the form of Nomogram.The results of model validation showed the following:differentiation:the area under the ROC curve in the model group was 0.728,with a 95%CI of(0.661~0.795);the area under the ROC curve in the validation group was 0.719,with a 95%CI of(0.652~0.786).Accuracy:the slopes of the calibration curves of the model and validation groups were 1,the intercept was 0.000,the model curves were basically fitted to the ideal model as a diagonal,and the H-L goodness-of-fit test was performed(P>0.05).The results of clinical validity analysis showed the highest net benefit of using the model of this study to predict post-PCI contrast renal injury when the predictive probability threshold was 0.15 to 0.75.Conclusion Post-PCI contrast renal injury is influenced by factors such as emergency PCI,number of stents,and amount of contrast used,and the model developed in this study can be used to predict post-PCI contrast renal injury.