2 Motivation for mediation analysis
Basic, clinical and epidemiological research are often interested in whether a specific intervention has an effect on a disease. Once this is established, the natural subsequent question is why this intervention had an effect? (1).
The subsequent and most relevant question is HOW?
Mediation analysis answers the question regarding HOW a treatment/exposure/intervention impacts an outcome of interest by uncovering the underlying pathways, and mechanisms.
2.1 Motivation for mediation analysis
2.1.1 Explanation and understanding
2.1.2 Confirmation and refutation of theory
Let’s assume we estimate a familial risk of diabetes and that risk estimates follow a normal distribution at the population level. Then we select the extremes of the distribution (2.5%) and assess the relative contribution of the genetic makeup, and risk phenotypes, among individuals with a strong/weak familial risk of type 2 diabetes.
2.1.3 To refine interventions
Let’s consider the evidence from the landmark prevention trials. The Diabetes Prevention Programme (2), The Diabetes Prevention Study (3), and The Da Quin study (4). These studies randomized high risk individuals to either metformin or a lifestyle intervention consisting of a physical activity or physical activity + diet. The main finding was a ~58% lower risk of incidence type 2 diabetes in the lifestyle intervention group compared to the metformin group
We might be interested in further refining the intervention so as to increase the magnitude of the effect. This might be done by altering or improving components of the intervention that target a particular mechanism for the outcome and discarding those components of which perhaps are not relevant for the outcome
When implementing the same lifestyle interventions in real world settings, lifestyle interventions show smaller effect and the long-term adoption and sustainability over time were some of the challenges to the effectiveness of these prevention efforts
2.2 Motivations for running this course
2.2.1 Methods for conducting mediation analysis are used incorrectly
A recent systematic review of the reporting quality of studies applying mediation analysis showed that most of the studies had incomplete reporting (5). This might impact both reproducibility and translation into practice.
2.2.2 The methodological framework for conducting mediation analysis evolved rapidly in recent years
Therefore, most applied examples are outdated in mediation analysis methods. A scoping review found that between 2015 to 2019 only 29.3% of studies using mediation analysis used modern mediation analysis methods (6).
Before moving on to the first exercise
Spend 5 minutes talking to your neighbour and discuss your own motivations for participating in this course. Consider the different motivations behind conducting mediation analysis.
Exercise: “Understanding different study designs”. Duration: 30 minutes
You will be divided into small groups of four people. Your assignment consist on reading the selected mediation analyses sections of the assigned articles and briefly present in plenum the methodological approach of your paper, and share the more relevant points discussed in your group in 3 minutes. Each paper has a different study design and you will have 15 min to complete the reading task described below.
2.2.3 Group 1: Intervention study a
Paper: Shoer et al. 2023
- read the abstract
- read section ‘Mediation analysis’ under ‘Methods’
- read section ‘The microbiome mediates the diet’s effect’ under ‘Results’
- try to understand Fig. 4.
Task: explain the overall study and the mediation analysis. What does the results imply?
2.2.4 Group 2: Intervention study b
Paper: Johansen et al. 2020
- read the abstract
- read from “Single mediation analysis” under ‘Statistical analysis’
- try to understand table 4
- read from “To explore the role of weight loss in relation to beta cell function we performed single mediation analysis” under ‘Discussion’
Task: explain the overall study and the mediation analysis. What does the results imply? How did the authors interpret the results? How do you interpret the results?
2.2.5 Group 3: Cohort study
Paper: Wang et al. 2022
- read the abstract
- read from “To test whether an association between the metabolite profile score and risk of type 2 diabetes…” under ‘Statistical analysis’
- try to understand Figure 4
- read the conclusion
Task: explain the overall study and the mediation analysis. What does the results imply? How did the authors interpret the results? How do you interpret the results?
2.2.6 Group 4: Cross-sectional study
Paper: Shi et al. 2022
- read the abstract
- they used a specific algorithm to identify mediation. You do not have to understand how it works. Simply note whether they adjusted for specific variables in their analysis (Hint: look in the methods section)
- Read section ‘Mediation by the gut microbial species Dialister invisus’ under ‘Results’
Task: explain the overall study design, what exposure, mediator and outcomes were they interested in? Are their results causal?