老年高血压患者并发体位性低血压预测模型构建及验证  

Construction and validation of a prediction model for orthostatic hypotension in elderly hypertensive patients

作　　者：袁娟 林晓红[1] 殷露露 吴丹[1] YUAN Juan;LIN Xiaohong;YIN Lulu;WU Dan(Department of Cardiology,Affiliated Brain Hospital of Nanjing Medical University,Nanjing 210046,China)

机构地区：[1]南京医科大学附属脑科医院心内科,江苏南京210046

出　　处：《东南大学学报(医学版)》2025年第2期243-250,共8页Journal of Southeast University(Medical Science Edition)

基　　金：南京医科大学校级科研项目(NMUB20230207)。

摘　　要：目的:探讨老年高血压患者并发体位性低血压(OH)的相关因素,建立预测模型并验证。方法:回顾性分析2022年10月至2024年10月本院收治的老年高血压患者的临床资料,从中纳入102例合并OH患者为OH组,并随机选择204例未合并OH患者为非OH组进行横断面研究。采集对比两组基线资料、疾病史、治疗用药等资料,单因素分析后应用最小绝对值收敛和选择算子(LASSO)回归进行变量筛选,多因素Logistic回归模型明确独立预测指标,建立Nomogram预测模型,并通过受试者工作特征(ROC)曲线、H-L检验与校准曲线、决策曲线与临床影响曲线评估模型区分度、校准度、临床受益。结果:OH组年龄、高血压病程及糖尿病、脑卒中、存在营养不良风险、应用α受体阻滞剂、利尿剂占比高于非OH组,体重指数低于非OH组(P<0.05)。LASSO回归变量筛选获得年龄、体重指数、饮酒史、吸烟史、糖尿病、营养不良风险、应用α受体阻滞剂、高血压病程8个非零系数变量,多因素Logistic回归分析显示,年龄、营养不良风险、高血压病程为老年高血压合并OH的独立危险因素(P<0.05)。ROC曲线分析显示,Nomogram预测模型预测老年高血压合并OH患者的曲线下面积(AUC)为0.748,截断阈值为0.294时,最佳预测敏感性为0.775,特异性为0.623,准确性为0.673。H-L检验显示,χ^(2)=5.212、P=0.735,校准曲线中平均绝对误差为0.014。决策曲线显示,阈值概率在0.10~0.67范围,Nomogram预测模型临床收益较高;临床影响曲线进一步发现,阈值概率>0.5时预测高风险患者例数与实际高风险患者例数一致性较高。结论:老年高血压合并OH的影响因素包括年龄、营养不良风险、高血压病程,且据此构建的Nomogram预测模型预测价值较好。Objective:To investigate the risk factors for orthostatic hypotension(OH)in elderly hypertensive patients and to construct and validate a prediction model.Methods:Clinical data of elderly hypertensive patients from October 2022 to October 2024 were retrospectively analyzed.A cross-sectional study was conducted with 102 patients with OH(OH group)and 204 randomly selected patients without OH(non-OH group).Baseline characteristics,disease history,and medication information were collected.After univariate analysis,least absolute shrinkage and selection operator(LASSO)regression was used for variable selection,and multivariate Logistic regression identified independent predictors.A Nomogram prediction model was constructed and evaluated using receiver operating characteristic(ROC)curve,Hosmer-Lemeshow(H-L)test,calibration curve,decision curve,and clinical impact curve for discrimination,calibration,and clinical utility.Results:The OH group showed higher proportions of advanced age,diabetes,hypertension duration,stroke history,risk of malnutrition,use of alpha-receptor blockers and diuretics,and lower body mass index compared to the non-OH group(P<0.05).Through LASSO regression for variable screening,eight variables with non-zero coefficients were obtained:age,body mass index,drinking history,smoking history,diabetes,malnutrition risk,use ofα-receptor blockers,and duration of hypertension.Multivariate Logistic regression confirmed age,risk of malnutrition,hypertension duration as independent risk factors for OH in elderly hypertensive patients(P<0.05).ROC curve analysis showed that the Nomogram prediction model had the area under the curve(AUC)of 0.748 for predicting elderly hypertensive patients with OH.At a cut-off value of 0.294,the optimal predictive sensitivity was 0.775,specificity was 0.623,and accuracy was 0.673.H-L test showedχ^(2)=5.212,P=0.735,with mean absolute error of 0.014 on the calibration curve.Decision curve analysis indicated high clinical benefit within threshold probabilities of 0.10-0.67,and cl

关 键 词：老年高血压 体位性低血压 预测模型 

分 类 号：R544.1[医药卫生—心血管疾病]