吉林大学学报(医学版) ›› 2023, Vol. 49 ›› Issue (2): 395-401.doi: 10.13481/j.1671-587X.20230216

• 临床研究 • 上一篇    下一篇

不同MMSE评分下阿尔兹海默病发病风险因素的贝叶斯分位数回归联合模型分析

王廉源1,杨毅1,丛慧文1,王浩桦1,包绮晗1,李承圣1,周立雯1,丁子琛1,李艳丽2,石福艳1(),王素珍1   

  1. 1.潍坊医学院公共卫生学院卫生统计学教研室,山东 潍坊 261053
    2.潍坊医学院图书馆,山东 潍坊 261053
  • 收稿日期:2022-08-11 出版日期:2023-03-28 发布日期:2023-04-24
  • 通讯作者: 石福艳 E-mail:shifuyan@126.com
  • 作者简介:王廉源(1997-),男,山东省潍坊市人,在读硕士研究生,主要从事健康测量与健康统计方面的研究。
  • 基金资助:
    国家自然科学基金项目(81872719);国家统计局一般项目(2018LY79);山东省卫健委自然科学基金项目(ZR2019MH034);山东省教育厅高等学校青创人才引育计划项目(2019-6-156);山东省卫健委医药卫生科技发展计划项目(2018WS066)

Bayesian quantile regression joint model analysis on risk factors of Alzheimer’s disease in people with different MMSE scores

Lianyuan WANG1,Yi YANG1,Huiwen CONG1,Haohua WANG1,Qihan BAO1,Chengsheng LI1,Liwen ZHOU1,Zichen DING1,Yanli LI2,Fuyan SHI1(),Suzhen WANG1   

  1. 1.Department of Health Statistics,School of Public Health,Weifang Medical University,Weifang 261053,China
    2.Library of Weifang Medical University,Weifang 261053,China
  • Received:2022-08-11 Online:2023-03-28 Published:2023-04-24
  • Contact: Fuyan SHI E-mail:shifuyan@126.com

摘要:

目的 探讨校正简易精神状态检查(MMSE)评分轨迹后的阿尔兹海默病(AD)发病风险影响因素,阐明不同MMSE评分人群AD发病的风险因素。 方法 基于AD神经成像计划数据库收集2005—2016年的随访数据,经筛选后最终纳入425名随访者的随访数据,采用LASSO方法对变量进行筛选;采用贝叶斯分位数回归模型分析MMSE评分在不同分位数上的影响因素,采用Cox模型和贝叶斯分位数回归联合模型方法分析影响AD发病的主要风险因素。 结果 经筛选后,纳入的变量包括白蛋白、总胆固醇和血糖浓度等10个变量。贝叶斯分位数回归联合模型的纵向子模型分析,在MMSE评分的不同分位数处,影响MMSE评分轨迹变化的因素相同,均为白蛋白、血糖浓度、年龄、性别、载脂蛋白E4(APOE4)基因、种族和教育评分。联合模型的Cox回归子模型分析,种族和APOE4基因在所有分位数上均对AD发病有影响,其中APOE4基因在4个分位数上的风险比分别为2.188(95%CI:1.775,2.620)、1.751(95%CI:1.422,2.042)、1.706(95%CI:1.391,2.102)和2.056(95%CI:1.439,3.206)。总胆固醇水平和家族史仅在部分分位数上对AD发病有影响。 结论 不同MMSE评分分布的人群,AD发病的风险因素不同,影响程度也有差异。有APOE4基因和白种人在不同分位数上均是AD发病的风险因素,总胆固醇水平和家族史仅在部分分位数上是AD发病的风险因素。

关键词: 贝叶斯分位数回归联合模型, 分位数回归模型, Cox模型, 阿尔兹海默病, 简易精神状态检查量表, 风险因素

Abstract:

Objective To discuss the influencing factors for the risk of Alzheimer’s disease (AD) after correction the Mini-Mental State Examination (MMSE) score trajectory, and to clarify the risk factors for of AD in the people with different MMSE scores. Methods The follow-up data from 2005-2016 were collected based on the AD Neuroimaging Program Database. After screening, the follow-up data of 425 people were finally included, and the variables were screened by LASSO method; the influencing factors of MMSE scores at different quantiles were analyzed by Bayesian quantile regression model, and COX model and Bayesian quantile regression joint model were used to analyze the main risk factors of AD. Results After screening, 10 variables were included, including albumin, total cholesterol, and blood glucose concentration and so on. The results of longitudinal sub-model analysis of the Bayesian quantile regression joint model showed that the influencing factors which affecting the change of MMSE score trajectory at different quartiles were the same, such as albumin, blood glucose concentration, age, gender,apolipoprotein E(APOE4) gene, race, and educational score. The Cox regression sub-model anslysis of the joint model results showed that race and APOE4 gene had the effects on the AD at all quartiles, and the risk ratios of APOE4 gene were 2.188((95%CI:1.775,2.620), 1.751(95%CI:1.422,2.042), 1.706(95%CI:1.391,2.102), and 2.056(95%CI:1.439,3.206).The total cholesterol level and history had the effects on the AD only at parts of quartiles. Conclusion The risk factors of AD are different in the people with different MMSE score distributions, and the degree of influencing are also different. Both APOE4 gene and whiteness are the risk factors of AD at different quartiles, and the total cholesterol level and history are the risk factors of AD only at parts of quartiles.

Key words: Bayesian quantile regression joint model, Quantile regression model, Cox model, Alzheimer’s disease, Mini-Mental State Examination, Risk factor

中图分类号: 

  • R749.16