Causal Inference in Clinical Research: A Primer

Area of Science:

• Epidemiology and Biostatistics
• Causal Inference Methodology
• Periodontology Research Applications

Background:

• Scientific inquiry fundamentally relies on understanding causal relationships, particularly in disease etiology and treatment.
• Applied researchers need accessible methods to distinguish causal effects from mere associations.

Purpose of the Study:

• To provide an overview of causal inference for applied researchers.
• To emphasize intuition and practical examples from periodontology.
• To introduce tools and frameworks for addressing causal questions.

Main Methods:

• Introduction to directed acyclic graphs (DAGs) for visualizing causal structures and identifying threats like confounding.
• Discussion of randomized controlled trials (RCTs) as the gold standard.
• Presentation of a three-step framework: causal question, causal effect identification, and analysis.
• Outline of design-based and adjustment-based identification strategies.

Main Results:

• Causal reasoning focuses on the impact of changing an exposure on an outcome.
• DAGs help explicate assumptions and identify potential biases.
• Both RCTs and observational studies are discussed, with a focus on assumptions for causal identification from observational data.
• The credibility of causal claims hinges on the plausibility of underlying assumptions, not solely on statistical methods.

Conclusions:

• Causal inference is assumption-driven, requiring careful consideration of the scientific context.
• Applied researchers can utilize frameworks and tools like DAGs to strengthen causal claims.
• Understanding and making assumptions explicit is crucial for valid causal inference, especially when using observational data.