中老年人高同型半胱氨酸血症风险列线图预测模型的构建与验证

Construction of a Nomogram Risk Prediction Model for Hyperhomocysteinemia Among the Middle-Aged and Elderly and Its Validation Test

目的探讨中老年人发生高同型半胱氨酸血症(hyperhomocysteinemia,HHcy)的危险因素并构建列线图风险预测模型。方法便利抽样法选取唐山市某三级甲等综合医院体检健康管理中心进行健康体检的1180例中老年人为研究对象,按照7∶3的比例划分为训练集(n=826)和验证集(n=354)。根据是否发生HHcy将训练集的中老年人分为病例组(n=376)和对照组(n=450)。通过单因素和多因素Logistic回归分析筛选中老年发生HHcy的独立危险因素并构建列线图预测模型。采用受试者工作特征曲线、校准曲线和决策曲线来评估模型的预测效能。结果年龄、性别、吸烟、饮酒、心理状态、体质量指数、总胆固醇、甘油三酯、高密度脂蛋白胆固醇、低密度脂蛋白胆固醇和尿酸为中老年发生HHcy的独立危险因素(均P<0.05)。构建的预测模型曲线下面积为0.860(95%CI=0.836~0.884),Hosmer-Lemeshow检验χ^(2)=11.228(P=0.189),一致性指数为0.860。外部验证结果显示曲线下面积为0.849(95%CI=0.810~0.888),Hosmer-Lemeshow检验χ^(2)=10.109(P=0.258),一致性指数为0.863。临床决策曲线的阈值概率在0.01~0.92时,临床净受益较高。结论构建的中老年人HHcy风险预测模型有较好的预测效率和临床适用性,有助于医护人员筛查高风险人群,并据此制订个性化的预防措施。

Objective To explore the risk factors of hyperhomocysteinemia(HHcy)among middle-aged and elderly people,and construct a nomogram risk prediction model.Methods Convenient sampling was used to select 1180 middle-aged and elderly citizens who underwent physical examination in the center of physical examination and health management of a tertiary A hospital in Tangshan City.The participants were divided into the training set(n=826)and the validation set(n=354)at a ratio of 7∶3.The participants in the training set were then divided into the case group(n=376)and the control group(n=450)according to the occurrence of HHcy.Univariate analysis and multivariate logistic regression analysis were used to screen the independent risk factors of HHcy among middle-aged and elderly people,and construct the nomogram risk prediction model.The receiver operating characteristic curve,calibration curve and decision curve were adopted to evaluate the prediction efficiency of the model.Results Age,gender,smoking,drinking,psychological status,body mass index,total cholesterol,triglyceride,high density lipoprotein cholesterol,low density lipoprotein cholesterol and uric acid were independent risk factors of HHcy among the middle-aged and elderly(all P<0.05).The area under curve of the prediction model was 0.860(95%CI=0.836-0.884),the Hosmer-Lemeshow testχ^(2)was 11.228(P=0.189),and the concordance index was 0.860.The external validation showed the area under curve was 0.849(95%CI=0.810-0.888),the Hosmer-Lemeshow testχ^(2)was 10.109(P=0.258),and the concordance index was 0.863.When the threshold probability of clinical decision curve was from 0.01 to 0.92,the clinical net benefit was higher.Conclusions The HHcy risk prediction model for the middle-aged and elderly has good prediction efficiency and clinical applicability,which is helpful for medical staff to screen high-risk groups and formulate individualized prevention measures accordingly.