基于健康管理队列的高血压风险预测模型  

A hypertension risk prediction model based on health management cohort

作　　者：于涛[1,2] 刘焕乐[3] 冯新[4] 徐付印[4] 陈亚飞[1,2] 薛付忠[1,2] 张成琪[5] 

机构地区：[1]山东大学公共卫生学院生物统计学系,山东济南250012 [2]山东大学齐鲁生物医学大数据研究中心,山东济南250012 [3]胜利油田中心医院保健科,山东东营257000 [4]胜利油田中心医院健康管理中心,山东东营257000 [5]山东大学附属千佛山医院健康管理中心,山东济南250014

出　　处：《山东大学学报（医学版）》2017年第6期61-65,共5页Journal of Shandong University:Health Sciences

基　　金：国家自然科学基金(81273082)

摘　　要：目的基于健康管理队列,构建高血压风险预测模型。方法依托山东多中心健康管理纵向观察队列,排除基线高血压、心脑血管疾病、血肌酐>177μmol/L、年龄<20岁者,构建高血压研究队列(共22 177人,其中男12 044人,女10 133人),分性别采用Cox回归建立高血压预测模型,并评价模型预测效果。结果观察期间新发高血压4 571例,发病密度为62.84/1 000人年。最终男性模型中的变量包括年龄、体质量指数、收缩压、舒张压、空腹血糖和红细胞压积,ROC曲线下面积(AUC)为0.821(95%CI:0.812~0.830);女性模型中的变量包括年龄、体质量指数、收缩压、红细胞计数和高密度脂蛋白,AUC为0.818(95%CI:0.806~0.828)。十折交叉验证结果显示,男女AUC分别为0.819(95%CI:0.810~0.828)、0.814(95%CI:0.803~0.825)。结论该模型具有较好的预测能力,可用来识别高血压高危个体。Objective To establish a prediction model for the risk of hypertension based on health management cohort. Methods Based on Shandong Multi-center Longitudinal Cohort for Health Management and after exclusion of partici- pants with baseline hypertension, cardiovascular and cerebrovascular diseases and other related diseases, a cohort with 22 177 subjects, including 12 044 males and 10 133 females, was established. Cox regression was used to construct pre- diction models for hypertension in males and females. The effect of model prediction was evaluated. Results A total of 4 571 new hypertention cases were observed, resulting in a cumulative incidence of 62.84/1 000 person years. The risk factors of the model for males included age, body mass index, systolic pressure, diastolic pressure, fasting blood- glucose and hematocrit. The estimated AUC of the model was 0. 821 （95% CI： 0. 812-0. 830）. In the model for fe- males, the risk factors included age, body mass index, systolic pressure, red blood cell and high density lipoprotein cholesterol. The estimated AUC of the model was 0. 818 （95% CI：0. 806-0. 828 ）. Ten-fold cross validation showed that the estimated AUC of the model for males was 0. 819（95% CI： 0. 810-0. 828） and 0. 814（95% CI： 0. 803-0. 825）for females. Conclusion The model has good prediction ability and could be used to identify individuals with high risk of hypertension.

关 键 词：高血压 健康管理队列 预测模型 

分 类 号：R544.1[医药卫生—心血管疾病]