冠心病心力衰竭并发肾衰竭的风险预测

Study on risk prediction of coronary heart disease with heart failure complicated with renal failure

目的探讨冠心病心力衰竭并发肾衰竭的风险因素,据此构建多重logistic回归模型。方法回顾性分析2018年9月至2022年7月河南省直第三人民医院收治的478例冠心病心力衰竭病人的临床资料,按照7∶3的比例将其分为建模组(n=335)与验证组(n=143)。根据肾衰竭并发情况又将建模组病人进一步分为并发组与未并发组,采用单因素及多因素logistic回归分析法分析建模组冠心病心力衰竭病人并发肾衰竭的风险因素,并建立多重logistic回归模型。采用受试者操作特征曲线(ROC曲线)、Hosmer-Lemeshow拟合优度检验、决策曲线(DCA)对模型进行内外部检验。结果478例冠心病心力衰竭病人中有113例并发肾衰竭(23.64%),建模组335例病人中有86例并发肾衰竭(24.48%),验证组143例病人中有31例并发肾衰竭(21.68%);并发组年龄≥60岁、美国纽约心脏病协会(NYHA)心功能分级Ⅳ级、原发性高血压史、糖尿病史、心房颤动史、利尿剂服用、左室后壁厚度(LVPWT)>11 mm、左室射血分数(LVEF)≤45%、血尿素氮(BUN)>6 mmol/L、血肌酐(Scr)>133μmol/L、血尿酸>420μmol/L、胱抑素C(CysC)>1.10 mg/L、N末端B型利钠肽前体(NT-proBNP)>3000 ng/L、血红蛋白(Hb)<110 g/L的占比均高于未并发组(P<0.05);共线性诊断显示LVPWT、LVEF、BUN、血尿酸、NT-proBNP存在共线性问题,在此基础上行多因素logistic回归分析,结果显示年龄≥60岁、NYHA心功能分级Ⅳ级、原发性高血压史、糖尿病史、心房颤动史、利尿剂服用史、LVEF≤45%、BUN>6 mmol/L、Scr>133μmol/L、CysC>1.10 mg/L、Hb<110 g/L均是冠心病心力衰竭病人并发肾衰竭的危险因素(P<0.05);根据logistic回归分析结果构建的风险预测logistic回归模型,经内部检验,ROC曲线下面积为0.84[95%CI:(0.78,0.89)],灵敏度、特异度分别为81.82%、77.42%,而Hosmer-Lemeshow拟合优度检验显示P=0.787;利用验证组数据实施外部检验,ROC曲线下面积为0.83[95%CI:(0.76,0.89)],灵敏度、特异度分别为80.00%、77.00%,而Hosmer-Lemeshow拟合优度检验显示P=0.776;DCA曲线显示,当阈值概率为0~0.9时,logistic回归模型对冠心病心力衰竭病人并发肾衰竭风险预测的净收益更高。结论老年、NYHA心功能分级Ⅳ级、原发性高血压史、糖尿病史、心房颤动史、利尿剂服用史、LVEF≤45%、BUN>6 mmol/L、Scr>133μmol/L、CysC>1.10 mg/L、Hb<110 g/L均是冠心病心力衰竭病人并发肾衰竭的危险因素,根据此结果而建立的风险预测logistic回归模型具有较好的预测价值,能够帮助临床工作者早期筛选并发肾衰竭的高风险人群,以便及时予以干预,降低肾衰竭发生率。

Objective To explore the risk factors of coronary heart disease with heart failure complicated with renal failure,and to construct a multiple logistic regression model bases on this.Methods The clinical data of 478 patients with coronary heart disease and heart failure admitted to The Third People's Hospital Directly Under He'nan Province from September 2018 to July 2022 were analyzed retrospectivel,and they were randomly assigned into modeling group(n=335)and verification group(n=143)according to the ratio of 7∶3.According to the complication of renal failure,the patients in the modeling group were further divided into concurrent group and non concurrent group.The risk factors of coronary heart disease with heart failure patients complicated with renal failure in the modeling group were analyzed by single factor and multiple factor logistic regression analysis,and a multiple logistic regression model was established.The receiver operating characteristic(ROC)curve,the Hosmer-Lemeshow goodness of fit test and decision curve method(DCA)were used to internal and external test the model.Results Among 478 patients with coronary heart disease and heart failure,113 patients were complicated with renal failure(23.64%),and 86 cases in 335 patients in the modeling group were complicated with renal failure(24.48%),and 31 cases in 143 patients in the validation group were complicated with renal failure(21.68%).The proportions of age≥60 years old,New York Heart Association(NYHA)cardiac function classⅣ,primary hypertension,diabetes,atrial fibrillation,diuretics,left ventricular posterior wall thickness(LVPWT)>11 mm,left ventricular ejection fraction(LVEF)≤45%,blood urea nitrogen(BUN)>6 mmol/L,serum creatinine(Scr)>133μmol/L,serum uric acid>420μmol/L,cystatinC(CysC)>1.10 mg/L,N-terminal pro B-type natriuretic peptide(NT-proBNP)>3000 ng/L and hemoglobin(Hb)<110 g/L in the concurrent group were higher than those in the non concurrent group(P<0.05).The collinearity diagnosis showed that LVPWT,LVEF,BUN,blood uric acid and NT-proBNP had collinearity problems.On this basis,multivariate logistic regression analysis was performed,and the results showed that the age≥60 years old,NYHA cardiac function classⅣ,history of primary hypertension,history of diabetes,history of atrial fibrillation,history of diuretics,LVEF≤45%,BUN>6 mmol/L,Scr>133μmol/L,CysC>1.10 mg/L,Hb<110 g/L were all risk factors for coronary heart disease with heart failure patients complicated with renal failure(P<0.05).According to the results of logistic regression analysis,the logistic regression model for risk prediction was constructed.Through internal test,the area under the ROC curve was 0.84[(95%CI:0.78,0.89)],and the sensitivity and specificity were 81.82%and 77.42%respectively,while the Hosmer-Lemeshow goodness of fit test showed P=0.787.External test was performed using the validation group data,and the area under the ROC curve was 0.83[95%CI:(0.76,0.89)],and the sensitivity and specificity were 80.00%and 77.00%respectively,while the Hosmer-Lemeshow goodness of fit test showed P=0.776.DCA curve showed that when the threshold probability was 0~0.9,the net benefit of logistic regression model for predicting the risk of coronary heart failure complicated with renal failure was higher.Conclusions Old age,NYHA heart function classⅣ,history of primary hypertension,history of diabetes,history of atrial fibrillation,history of diuretics,LVEF≤45%,BUN>6 mmol/L,Scr>133μmol/L,CysC>1.10 mg/L,Hb<110 g/L are all risk factors for coronary heart disease and heart failure patients complicated with renal failure.The risk prediction logistic regression model established based on this result has good predictive value,and it can help clinical workers to screen high-risk groups complicated with renal failure early,so as to intervene in time and reduce the incidence of renal failure.