Longitudinal mediation analysis with multilevel and latent growth models: a separable effects causal approach.

Publisher: BioMed Central Country of Publication: England NLM ID: 100968545 Publication Model: Electronic Cited Medium: Internet ISSN: 1471-2288 (Electronic) Linking ISSN: 14712288 NLM ISO Abbreviation: BMC Med Res Methodol Subsets: MEDLINE

Original Publication: London : BioMed Central, [2001-

Mediation Analysis* ; Causality*; Humans ; Longitudinal Studies ; Models, Statistical ; Multilevel Analysis/methods ; Computer Simulation ; Algorithms

Background: 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.
Methods: 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.
Results: 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.
Conclusion: 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.
(© 2024. The Author(s).)

J Causal Inference. 2017 Sep;5(2):. (PMID: 29387520)
Annu Rev Psychol. 2006;57:505-28. (PMID: 16318605)
Lifetime Data Anal. 2019 Oct;25(4):593-610. (PMID: 30218418)
Biom J. 2019 Sep;61(5):1270-1289. (PMID: 30306605)
Dev Cogn Neurosci. 2023 Oct;63:101281. (PMID: 37536082)
Biometrics. 1982 Dec;38(4):963-74. (PMID: 7168798)
Stat Med. 2015 Dec 20;34(29):3866-87. (PMID: 26278111)
Lifetime Data Anal. 2006 Jun;12(2):143-67. (PMID: 16817006)
Am J Med. 2009 Jul;122(7):656-63.e1. (PMID: 19559168)
Stat Med. 2017 Nov 20;36(26):4153-4166. (PMID: 28809051)
Epidemiology. 2021 Mar 1;32(2):209-219. (PMID: 33512846)
Epidemiology. 1992 Mar;3(2):143-55. (PMID: 1576220)
Struct Equ Modeling. 2003 Apr 1;10(2):238. (PMID: 20157639)
Epidemiology. 2017 Mar;28(2):266-274. (PMID: 27984420)
Res Nurs Health. 2012 Dec;35(6):647-58. (PMID: 22911221)
Cogn Sci. 2013 Aug;37(6):1011-35. (PMID: 23899340)
Behav Res Methods. 2018 Aug;50(4):1398-1414. (PMID: 29067672)
Patterns (N Y). 2020 Jun 12;1(3):. (PMID: 32656541)
J Public Health (Oxf). 2017 Dec 1;39(4):e152-e160. (PMID: 27613768)
Biom J. 2020 May;62(3):532-549. (PMID: 30779372)
Epidemiology. 2014 Mar;25(2):300-6. (PMID: 24487213)
Biostatistics. 2016 Jan;17(1):122-34. (PMID: 26272993)
Stat Med. 2019 Oct 30;38(24):4828-4840. (PMID: 31411779)
J R Stat Soc Series B Stat Methodol. 2017 Jun;79(3):917-938. (PMID: 28824285)
Psychol Methods. 2012 Mar;17(1):15-30. (PMID: 22251268)
Stat Med. 2019 Sep 30;38(22):4334-4347. (PMID: 31286536)
Psychol Methods. 2006 Jun;11(2):142-63. (PMID: 16784335)
Lifetime Data Anal. 2021 Oct;27(4):588-631. (PMID: 34468923)

Keywords: Causal effects; Latent growth models; Longitudinal mediation analysis; Multilevel models; Separable effects

Date Created: 20241025 Date Completed: 20241026 Latest Revision: 20241030

20241031

PMC11515317

10.1186/s12874-024-02358-4

39455967