新疆哈萨克族人群原发性高血压发病风险预测模型构建  

Construction of predictive model for precipitation risk of primary hypertension in Kazak population in Xinjiang

作　　者：徐月贞[1,2] 王家威 刘金宝 樊琼玲[1] 罗园园 詹怀峰 王红军 陈蕊 陶宁 由淑萍[1] XU Yue-zhen;WANG Jia-wei;LIU Jin-bao;FAN Qiong-ling;LUO Yuan-yuan;ZHAN Huai-feng;WANG Hong-jun;CHEN Rui;TAO Ning;YOU Shu-ping(School of Nursing,Xinjiang Medical University,Urumqi,Xinjiang 830011,China)

机构地区：[1]新疆医科大学护理学院,新疆乌鲁木齐830011 [2]新疆医科大学公共卫生学院,新疆乌鲁木齐830011 [3]新疆乌鲁木齐县水西沟卫生院,新疆乌鲁木齐830000 [4]新疆乌鲁木齐县小渠子乡卫生院,新疆乌鲁木齐830000

出　　处：《现代预防医学》2019年第16期2885-2889,2894,共6页Modern Preventive Medicine

基　　金：2017年度国家教育部人文社会科学研究(项目编号：17YJCZH230);新疆护理学会项目(2018XH42)

摘　　要：目的构建适合于新疆哈萨克族牧民的高血压发病风险预测模型,为牧区哈萨克族牧民高血压的预防筛选提供工具。方法采用分层整群随机抽样的方法,抽取18岁以上哈萨克族进行队列研究,参加每2年1次随访的5 327人为研究对象,随机分为(60%)建模队列和(40%)验证队列,采用Weibull回归方法构建南山模型,并将南山模型与同类高血压风险评估模型(即模型I和模型II)进行比较,模型验证和比较分析采用ROC曲线下面积和HosmerLemeshowχ^2拟合度,采用净重分组提高指数(NRI)分析新变量对南山模型的预测能力影响。结果南山模型未来4年和2年高血压发病预测ROC曲线下面积分别为0.901(95%CI:0.892~0.909,P<0.05)和0.863(95%CI:0.846~0.880,P<0.05);Hosmer-Lemeshow检验值分别为14.289(P=0.115)和18.124(P=0.086);南山模型的区分能力和标定能力均优于模型I和模型II;NRI分析南山模型4年和2年预测概率分别提高了17.45%(Z=5.892,P<0.05)和16.34%(Z=3.67,P<0.05)。结论构建的南山模型具有很好的预测能力,可以对新疆哈萨克族牧民进行高血压发病风险预测。Objective To construct a predictive model of hypertension risk for Kazak herdsmen in Xinjiang and provide a tool for the prevention and screening of hypertension in Kazak herdsmen. Methods A stratified group random sampling method was used to take a cohort study of ethnic Kazakh over the age of 18. A total of 5327 people were followed every two years as subjects, who were randomly divided into modeling queues and(60%) validation queues(40%). The Nanshan model was constructed by using the Weibull regression method, and Nanshan model and similar hypertension risk assessment model(model I and model II) comparative analysis. The model verification and comparative analysis used ROC curve and HosmerLemeshow fitting degree, and the net reclassification improvement(NRI) was used to increase the index to analyze the new variable. Predictive ability effect on Nanshan model. Results The area under the ROC curve(AUC) of the Nanshan model for the next 4 years and 2 years of hypertension prediction was 0.901(95 % CI: 0.892-0.909, P<0.05) and 0.863(95% CI: 0.846-0.880,P<0.05). The test values of Hosmer-Lemeshow were 14.289(P=0.115) and 18.124(P=0.086), respectively. The difference ability and calibration ability of Nanshan model were better than those of Model I and Model II. The NRI analysis of Nanshan model increased the predicted probability of 4 years and 2 years by 17.45%(Z=5.892, P<0.05) and 16.34%(Z=3.67, P<0.05). Conclusion Nanshan model has a good ability to predict hypertension risk among Kazak herdsmen in Nanshan pastoral area of Xinjiang.

关 键 词：哈萨克族 高血压 模型构建 风险预测 

分 类 号：R195[医药卫生—卫生统计学]