In observational studies, instrumental variable (IV) methods are commonly applied when there exists some unmeasured covariates. In Mendelian Randomization (MR), constructing an allele score by using many single nucleotide polymorphisms (SNPs) is often implemented; however, there are risks estimating biased causal effects by including some invalid IVs. Invalid IVs are candidates of IVs associated with some unobserved variables. To solve this problem, we propose a novel strategy in this paper: using Negative Control Outcomes (NCOs) as auxiliary variables. By using NCOs, we can essentialy select only valid IVs and exclude invalid IVs without any information of IV candidates. We also propose the new two-step estimating procedure and prove the semiparametric efficiency. We demonstrate the superior performance of the proposed estimator compared with existing estimators via simulation studies.
翻译:在观察研究中,当存在一些未测的共变因素时,通常会采用工具变量(IV)方法。在门德尔兰随机化(MR)中,通过使用许多单核酸多形态(SNPs)来构建异差分,这往往得到实施;然而,通过纳入一些无效的四体,有可能估计有偏向的因果关系。无效四体是与一些未观测的变量相关的四体的候选物。为了解决这个问题,我们在本文件中提出了一个新颖的战略:使用负控制结果作为辅助变量。通过使用 NCOs,我们可以在不向IV 候选人提供任何信息的情况下,只选择有效的四体,排除无效四体。我们还提出了新的两步估计程序,并证明半对准效率。我们通过模拟研究,展示了拟议天花与现有天花相比的优异性表现。