Medical Research Council Integrative Epidemiology Unit at the University of Bristol, Bristol, UK.
Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, UK.
BMJ. 2024 Oct 14;387:e078093. doi: 10.1136/bmj-2023-078093.
Instrumental variable analysis uses naturally occurring variation to estimate the causal effects of treatments, interventions, and risk factors on outcomes in the population from observational data. Under specific assumptions, instrumental variable methods can provide unbiased estimates of causal effects. This article explains these assumptions and the information and tests typically reported in instrumental variable studies, which can assess the credibility of the findings of instrumental variable studies.
工具变量分析利用自然发生的变异，从观察性数据中估计处理、干预和风险因素对人群结局的因果效应。在特定假设下，工具变量方法可以提供因果效应的无偏估计。本文解释了这些假设以及工具变量研究中通常报告的信息和检验，可以评估工具变量研究结果的可信度。
