Regression Discontinuity Designs in Epidemiology: A Practical Guide.

Area of Science:

• Clinical Epidemiology
• Econometrics
• Health Services Research

Background:

• Quasi-experimental methods are crucial for estimating causal effects from observational data in clinical research.
• Regression discontinuity design (RDD) is a powerful quasi-experimental method for causal inference.
• The application of RDD in clinical epidemiology is growing due to electronic health records and rule-based interventions.

Purpose of the Study:

• To provide an overview of the Regression Discontinuity Design (RDD) methodology.
• To describe the key assumptions enabling RDD's use with observational clinical data.
• To demonstrate RDD application in estimating treatment effects, using a statin prescription example.

Main Methods:

• Overview of continuity-based and local randomization RDD approaches.
• Demonstration of RDD implementation using statistical software (R and Stata).
• Worked example: Estimating statin treatment effect on LDL cholesterol using a cardiovascular disease risk score rule.

Main Results:

• The study outlines the principles and assumptions of RDD for causal inference.
• It demonstrates the practical application of RDD in a clinical context.
• The worked example illustrates how RDD can assess treatment effects based on defined thresholds.

Conclusions:

• Regression discontinuity design is a valuable tool for causal inference in clinical epidemiology.
• RDD facilitates the assessment of clinical decision-making effectiveness using observational data.
• The method is particularly useful when interventions are assigned based on specific thresholds or rules.