E3 PreliminaryModerate confidencePEM unclearMethods-PaperPeer-reviewedMachine draft
Tutorial: The practical application of longitudinal structural equation mediation models in clinical trials.
Goldsmith, Kimberley A, MacKinnon, David P, Chalder, Trudie et al. · Psychological methods · 2018 · DOI
Quick Summary
This tutorial paper explains how researchers can better understand how treatments work by measuring patients multiple times throughout a study, rather than just at the beginning and end. It uses the PACE trial (a study about treatments for ME/CFS) as an example to teach researchers different statistical methods for analyzing whether a treatment works by changing something specific in a patient's body or behavior.
Why It Matters
Understanding how treatments work—not just whether they work—is essential for improving ME/CFS care and developing better therapeutic approaches. This tutorial equips researchers with rigorous statistical methods to identify the actual mechanisms of benefit in clinical trials, which can guide treatment refinement and personalization for ME/CFS patients.
Observed Findings
- No empirical findings are reported; this is a methods paper using simulated data
- Provided worked examples of three longitudinal mediation model types applicable to repeated-measures clinical trial data
- Demonstrated how measurement error and unmeasured confounding can be addressed in longitudinal mediation frameworks
- Illustrated interpretation of direct and indirect effects over time using PACE trial-like data structure
Inferred Conclusions
- Longitudinal mediation models are more appropriate than cross-sectional analyses for understanding treatment mechanisms in clinical trials because they respect temporal ordering
- When clear a priori theory is absent, fitting multiple plausible models provides a sensitivity analysis that strengthens confidence in mediation hypotheses
- Flexible structural equation approaches can incorporate realistic features of clinical data (measurement error, repeated assessment) often ignored in simpler analyses
Remaining Questions
- How do researchers determine which among multiple plausible mediation models should be favored when statistical evidence is mixed?
- Do results from longitudinal mediation analyses of the PACE trial itself support or refute specific mechanistic hypotheses about how cognitive behavioral therapy and graded activity affect functional outcomes in ME/CFS?
What This Study Does Not Prove
This is a methodological tutorial rather than an empirical study of treatment efficacy or mechanisms; it does not prove that any particular treatment for ME/CFS is effective or demonstrate what the true mechanisms of the PACE trial are. The simulated data used are illustrative examples and do not represent actual trial results or validate any specific treatment approach.
Tags
EXPLORATORYPEM UNCLEAR
Metadata
- DOI
- 10.1037/met0000154
- PMID
- 29283590
- Review status
- Machine draft
- Evidence level
- Early hypothesis, preprint, editorial, or weak support
- Last updated
- 10 April 2026
About the PEM badge: “PEM required” means post-exertional malaise was an explicit required diagnostic criterion for participant inclusion in this study — not that PEM was studied, observed, or discussed. Studies using criteria that do not require PEM (e.g. Fukuda, Oxford) are tagged “PEM not required”. How the atlas works →
Spotted an error in this entry? Report it →