Matching within a hybrid RCT/RWD: framework on associated causal estimands.

Area of Science:

• Clinical Trials
• Real-World Evidence
• Biostatistics

Background:

• Regulatory bodies increasingly accept real-world evidence (RWE) in drug development.
• Hybrid clinical trials, incorporating external control data, are gaining traction.
• Lack of clarity exists on matching strategies and estimands in external data borrowing.

Purpose of the Study:

• To delineate matching schemes and their corresponding estimands for external control data in hybrid trials.
• To provide a framework for understanding what different matching methods aim to estimate.
• To evaluate the performance of various matching schemes and estimation methods.

Main Methods:

• Detailed formulation of estimands for four distinct matching schemes.
• Conducting simulation studies to assess performance characteristics.
• Evaluating impact of different estimation methods, effect sizes, and confounder missingness.

Main Results:

• The study formulates estimands for four matching schemes: intersection, union, concurrent control, and investigational treatment matching.
• Simulation results provide insights into the performance of these schemes under various conditions.
• Understanding estimands is crucial for appropriate application of external control data.

Conclusions:

• Clear definition of estimands is essential for the valid use of external control data in hybrid trials.
• The choice of matching scheme directly influences the research question that can be answered.
• This work supports the informed integration of real-world data in clinical trial design and analysis.