Using Directed Acyclic Graphs (DAGs) to Determine if the Total Causal Effect of an Individual Randomized Physical...

Area of Science:

• Public Health
• Behavioral Science
• Epidemiology

Background:

• Physical activity offers significant public health benefits, including improved well-being and disease prevention.
• Individual-level behavioral interventions effectively increase physical activity in adults, but their causal impact is often unclear.
• Directed acyclic graphs (DAGs) can clarify causal inference but are rarely used for physical activity interventions.

Purpose of the Study:

• To demonstrate the application of DAGs in determining the identifiability of causal effects for individual-level physical activity interventions.
• To illustrate how DAGs can isolate the true impact of interventions from spurious associations.
• To provide a methodological example using real-world study data.

Main Methods:

• Utilized a directed acyclic graph (DAG) approach to analyze causal pathways.
• Applied DAGitty and Mplus software for the analysis.
• Based the demonstration on data from the Well-Being and Physical Activity study (ClinicalTrials.gov, identifier: NCT03194854).

Main Results:

• The study successfully demonstrated how a DAG can be used to assess the identifiability of the total causal effect of a physical activity intervention.
• The methodology allows for clearer articulation of causal inference from behavioral interventions.
• Provided annotated files for reproducibility and further application.

Conclusions:

• DAGs are a valuable tool for rigorously evaluating the causal impact of physical activity promotion interventions.
• Employing DAGs enhances the clarity and reliability of findings in public health research.
• This approach can strengthen the evidence base for effective physical activity strategies.