Department of Economics, Business and Statistics, University of Palermo, Viale delle Scienze, Building 13, Palermo, 90128, Italy.
Leibniz-Institut für Präventionsforschung und Epidemiologie - BIPS, Achterstraße 30, Bremen, 28359, Germany.
BMC Med Res Methodol. 2024 Oct 25;24(1):248. doi: 10.1186/s12874-024-02358-4.
Causal mediation analysis is widespread in applied medical research, especially in longitudinal settings. However, estimating natural mediational effects in such contexts is often difficult because of the presence of post-treatment confounding. Moreover, many models frequently used in applied research, like multilevel and latent growth models, present an additional difficulty, i.e. the presence of latent variables. In this paper, we propose a causal interpretation of these two classes of models based on a novel type of causal effects called separable, which overcome some of the issues of natural effects.
We formally derive conditions for the identifiability of separable mediational effects and their analytical expressions based on the g-formula. We carry out a simulation study to investigate how moderate and severe model misspecification, as well as violation of the identfiability assumptions, affect estimates. We also present an application to real data.
The results show how model misspecification impacts the estimates of mediational effects, particularly in the case of severe misspecification, and that the bias worsens over time. The violation of assumptions affects separable effect estimates in a very different way for the mixed effect and the latent growth models.
Our approach allows us to give multilevel and latent growth models an appealing causal interpretation based on separable effects. The simulation study shows that model misspecification can heavily impact effect estimates, highlighting the importance of careful model choice.
因果中介分析在应用医学研究中广泛应用，尤其是在纵向研究中。然而，由于治疗后混杂的存在，在这种情况下估计自然中介效应通常很困难。此外，应用研究中经常使用的许多模型，如多层次模型和潜在增长模型，存在另一个困难，即潜在变量的存在。在本文中，我们基于一种新的因果效应——可分离性，提出了这两类模型的因果解释，该因果效应克服了自然效应的一些问题。
我们根据 g 公式正式推导了可分离中介效应的可识别性条件及其解析表达式。我们进行了一项模拟研究，以调查中度和重度模型误设定以及违反识别假设如何影响估计值。我们还提供了对真实数据的应用。
结果表明，模型误设定如何影响中介效应的估计值，尤其是在严重误设定的情况下，并且随着时间的推移，偏差会恶化。假设的违反对混合效应和潜在增长模型的可分离效应估计值的影响方式非常不同。
我们的方法允许我们基于可分离效应为多层次模型和潜在增长模型提供有吸引力的因果解释。模拟研究表明，模型误设定会严重影响效应估计值，突出了谨慎选择模型的重要性。
