中国组织工程研究 ›› 2023, Vol. 27 ›› Issue (5): 657-662.doi: 10.12307/2022.1005

• 组织构建临床实践 clinical practice in tissue construction •    下一篇

中老年人代谢综合征危险因素分析及列线图预测模型的构建

沈炼伟,朱洪柳,王  维   

  1. 锦州医科大学附属第一医院,辽宁省锦州市  121000
  • 收稿日期:2022-01-07 接受日期:2022-02-16 出版日期:2023-02-18 发布日期:2022-07-23
  • 通讯作者: 王维,硕士,副主任医师,锦州医科大学附属第一医院,辽宁省锦州市 121000
  • 作者简介:沈炼伟,男,1995年生,广东省潮安县人,汉族,锦州医科大学在读硕士,主要从事肌骨康复方面的研究。

Risk factor analysis of metabolic syndrome and construction of a nomogram prediction model in middle-aged and elderly people

Shen Lianwei, Zhu Hongliu, Wang Wei   

  1. First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China
  • Received:2022-01-07 Accepted:2022-02-16 Online:2023-02-18 Published:2022-07-23
  • Contact: Wang Wei, Master, Associate chief physician, First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China
  • About author:Shen Lianwei, Master candidate, First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China

摘要:

文题释义:
倾向性评分匹配法:是一种统计学方法,用于处理观察研究的数据。在观察研究中,由于种种原因,数据偏差和混杂变量较多,倾向评分匹配的方法正是为了减少这些偏差和混杂变量的影响,以便对实验组和对照组进行更合理的比较。
列线图:又称诺莫图(Nomogram图),它是建立在多因素回归分析的基础上,将多个预测指标进行整合,然后采用带有刻度的线段,按照一定的比例绘制在同一平面上,从而用以表达预测模型中各个变量之间的相互关系。

背景:随着国内人口老龄化的加剧及生活方式的改变,中老年人代谢综合征的患病率不断升高,筛选该病的危险因素及构建该病的预测模型很有必要。
目的:探讨国内中老年人代谢综合征的危险因素并构建代谢综合征列线图预测模型,为中老年代谢综合征的防控工作提供参考。 
方法:研究数据来源于中国健康与养老追踪调查2015年随访数据,以是否患有代谢综合征为因变量,纳入左手肌力、右手肌力、站立时间、步行速度、从椅子上起立时间、性别、年龄、吸烟、饮酒共9个变量探讨中老年人代谢综合征的相关因素,在SPSS上进行倾向性评分与描述性分析,在Rstudio上将原始数据集分为训练集与验证集,训练集进行单因素、多因素回归分析、列线图模型构建及内部验证,验证集进行外部验证。 
结果与结论:①共筛选中老年人7 384人,其中检出有代谢综合征518人(7%),无代谢综合征6 865人(93%);②倾向性评分以1∶1从无代谢综合征的6 865人中选出518人作为对照组,代谢综合征患者作为患病组,构建原始数据集,以7∶3将原始数据集随机分为训练集(728人)与验证集(308人);③模型的构建:根据训练集(建模队列)的二元logistic回归分析结果,选取出右手肌力、步行速度、性别、饮酒、站立时间等5个变量构建模型;模型的评价:模型受试者工作特征曲线曲线下面积为0.877(95%CI:0.851-0.902),表明模型具有较高的区分度,校准曲线拟合良好说明模型具有较高的校准度;④模型的内部验证:采用Bootstrop法,经过1 000次有放回的重复抽样生成新样本,生成的校准曲线拟合良好;⑤模型的外部验证:由验证集(验证队列)检验,所得的受试者工作特征曲线曲线下面积为0.89(95%CI:0.854-0.926),表明模型具有较高的区分度,校准曲线拟合良好说明模型具有较高的校准度,提示模型在验证队列也具有良好的效能;⑥构建的中老年人代谢综合征预测模型具有较好的可信度,由此得出的列线图,可根据右手肌力、步行速度、性别、饮酒、年龄、站立时间情况预测中老年人患有代谢综合征的概率,有利于中老年人代谢综合征的早发现、早诊断、早治疗,可在临床上推广。 
缩略语:中国健康与养老追踪调查:China Health and Retirement Longitudinal Study,CHARLS

https://orcid.org/0000-0002-7658-9152 (沈炼伟)

中国组织工程研究杂志出版内容重点:组织构建;骨细胞;软骨细胞;细胞培养;成纤维细胞;血管内皮细胞;骨质疏松;组织工程

关键词: 中老年人, 代谢综合征, 列线图预测模型, 危险因素, 回归分析

Abstract: BACKGROUND: With population aging and life-style variation in China, the prevalence of metabolic syndrome in the middle-aged and elderly is increasing. Therefore, it is necessary to screen the risk factors of the disease and construct a prediction model of the disease.
OBJECTIVE: To investigate the risk factors of metabolic syndrome in middle-aged and elderly people in China and to construct the nomogram prediction model of metabolic syndrome, so as to provide reference for the prevention and control of metabolic syndrome in such populations. 
METHODS: The data were from the 2015 follow-up data of the China Health and Retirement Longitudinal Study (CHARLS), with metabolic syndrome as a dependent variable. Nine variables including left hand muscle strength, right hand muscle strength, standing time, walking speed, time to get up from the chair, sex, age, smoking and drinking were involved to explore the related factors of metabolic syndrome in middle-aged and elderly people. Propensity score matching and descriptive analysis were performed on SPSS. The original data set was divided into training set and verification set with Rstudio. The training set was subject to univariate and multivariate regression analysis, nomogram model construction and internal verification, whereas the verification set was subject to external verification. 
RESULTS AND CONCLUSION: (1) A total of 7 384 middle-aged and elderly people were screened, including 518 (7%) with metabolic syndrome and 6 865 (93%) without metabolic syndrome. (2) 6 865 subjects without metabolic syndrome was paired by 1:1 with propensity score matching to screen out 518 subjects as the control group and those with metabolic syndrome acted as the case group. The control group and the case group constituted the original data set. The original data set was randomly divided into the training set (728 subjects) and the validation set (308 subjects) at 7:3. (3) Construction of the model: according to the binary logistic regression analysis results of the training set (modeling cohort), five variables such as right hand muscle strength, walking speed, sex, drinking and standing time were selected to build the model. Evaluation of the model: the area under the receiver operator characteristic curve of the model was 0.877 (95% confidence interval: 0.851-0.902), indicating that the model has high discrimination and good fitting of the calibration curve, indicating that the model has a high calibration degree. (4) Internal verification of the model: the bootstrap method was used to generate new samples after 1000 repeated sampling with return, and the generated calibration curve fitted well. (5) External verification of the model: the model was tested by the verification set (verification queue), the area under the receiver operator characteristic curve was 0.89 (95% confidence interval: 0.854-0.926), indicating that the model has high discrimination and good fitting of the calibration curve, that is, the model has high calibration degree and good efficiency in the verification queue. (6) The prediction model of metabolic syndrome in the middle-aged and elderly constructed in this study has good reliability. The nomogram obtained from this can predict the probability of metabolic syndrome in the middle-aged and elderly according to the right hand muscle strength, walking speed, sex, drinking, age and standing time, which is conducive to the early detection, diagnosis and treatment of metabolic syndrome in the middle-aged and elderly and can be popularized in clinical practice. 

Key words: middle-aged and elderly people, metabolic syndrome, nomogram prediction model, risk factor, regression analysis

中图分类号: