摘要
不良饮食是慢性非传染性疾病最重要的可控危险因素之一,但通过随机对照试验定量阐明具体饮食因素与健康结局的因果关联面临很多困难。近年来,因果推断的迅速发展为充分利用和发掘观察性研究数据,产生高质量的营养流行病学研究证据提供了有力的理论和方法工具。其中,因果图模型通过整合大量先验知识将复杂的因果关系系统可视化,提供了识别混杂和确定因果效应估计策略的基础框架,基于不同的因果图,可选择调整混杂、工具变量或中介分析等不同的分析策略。本文对因果图模型的思想和各种分析策略的特点及其在营养流行病学研究中的应用进行介绍,旨在促进因果图模型在营养领域的应用,为后续研究提供参考和建议。
Suboptimal diet is one of the most important controllable risk factors for non-communicable diseases.However,randomized controlled trials make it difficult to quantify the causal association between specific dietary factors and health outcomes.In recent years,the rapid development of causal inference has provided a robust theoretical and methodological tool for making full use of observational research data and producing high-quality nutritional epidemiologic research evidence.The causal graph model visualizes the complex causal relationship system by integrating a large amount of prior knowledge and provides a basic framework for identifying confounding and determining causal effect estimation strategies.Different analysis strategies such as adjusting confounders,instrumental variables,or mediation analysis can be created based on other causal graphs.This paper introduces the idea of the causal graph model and the characteristics of various analysis strategies and their application in nutritional epidemiology research,aiming to promote the application of the causal graph model in nutrition and provide references and suggestions for the follow-up research.
作者
唐丹
肖雄
杨帆
胡逸凡
殷建忠
赵星
Tang Dan;Xiao Xiong;Yang Fan;Hu Yifan;Yin Jianzhong;Zhao Xing(West China School of Public Health and West China Fourth Hospital,Sichuan University,Chengdu 610041,China;School of Public Health,Kunming Medical University,Kunming 650500,China;Baoshan College of Traditional Chinese Medicine,Baoshan 678000,China)
出处
《中华流行病学杂志》
CAS
CSCD
北大核心
2021年第10期1882-1888,共7页
Chinese Journal of Epidemiology
基金
国家重点研发计划(2017YFC0907305,2017YFC0907300)
国家自然科学基金(81973151,81773548)
四川省科技厅重点研发(2020YFS0215)
中国营养学会——振东国人体质与健康研究基金(CNS-ZD2020-149)。
关键词
因果推断
因果图模型
营养流行病学
Causal inference
Causal graph model
Nutritional epidemiology
