工具变量法在环境流行病学因果推断中的应用与进展

Applications and advancements of instrumental variable approach in causal inference in environmental epidemiology

环境流行病学研究中环境暴露广泛而非随机、混杂偏倚较为复杂,对其因果推断带来了巨大的挑战。近年来,随着因果推断方法引入观察性研究中,其为环境流行病学因果推断研究提供了更多统计方法的选择。工具变量法(Ⅳ)作为一种能够很好控制未测量混杂因素的因果推断方法,逐渐应用在环境流行病学研究领域中。本研究介绍了Ⅳ的基本原理,归纳了目前应用Ⅳ进行环境流行病学因果推断的研究进展与局限性。当前环境流行病领域应用Ⅳ方法进行因果推断研究尚处于初级阶段,合理使用Ⅳ并与其他因果推断方法的有效整合将成为环境流行病学因果推断的发展重点。本研究旨在为我国环境暴露的人群健康效应的因果推断研究提供方法学上的参考和依据。

In environmental epidemiological research,extensive non-random environmental exposures and complex confounding biases pose significant challenges when attempting causal inference.In recent years,the introduction of causal inference methods into observational studies has provided a broader range of statistical tools for causal inference research in environmental epidemiology.The instrumental variable(IV)approach,as a causal inference technique for effectively controlling unmeasured confounding factors,has gradually found application in the field of environmental epidemiological research.This article reviewed the basic principles of IV and summarized the current research progress and limitations of applying IV for causal inference in environmental epidemiology.IV application in the field of environmental epidemiology is still in the initial stage.Rational use of IV and effective integration with other causal inference methods will become the focus of the development of causal inference in environmental epidemiology.The aim of this paper is to provide a methodological reference and basis for future studies involving causal inference to target population health effects of environmental exposures in China.