Establishing causality is an essential step towards developing interventions for psychiatric disorders, substance use and many other conditions. While randomized controlled trials (RCTs) are considered the gold standard for causal inference, they are unethical in many scenarios. Mendelian randomization (MR) can be used in such cases, but importantly both RCTs and MR assume unidirectional causality. In this paper, we developed a new model, MRDoC2, that can be used to identify bidirectional causation in the presence of confounding due to both familial and non-familial sources. Our model extends the MRDoC model (Minică et al. in Behav Genet 48:337–349, https://doi.org/10.1007/s10519-018-9904-4, 2018), by simultaneously including risk scores for each trait. Furthermore, the power to detect causal effects in MRDoC2 does not require the phenotypes to have different additive genetic or shared environmental sources of variance, as is the case in the direction of causation twin model (Heath et al. in Behav Genet 23:29–50, https://doi.org/10.1007/BF01067552, 1993).

确定因果关系是针对精神疾病、物质使用和许多其他情况制定干预措施的重要一步。虽然随机对照试验 (RCT) 被认为是因果推理的黄金标准，但它们在许多情况下都是不道德的。在这种情况下可以使用孟德尔随机化 (MR)，但重要的是 RCT 和 MR 都假设单向因果关系。在本文中，我们开发了一种新模型 MRDoC2，可用于在存在因家族性和非家族性来源混杂的情况下识别双向因果关系。我们的模型扩展了 MRDoC 模型（Minică 等人在 Behav Genet 48:337–349, https://doi.org/10.1007/s10519-018-9904-4, 2018），同时包括每个特征的风险评分。此外，