Conditional borrowing external data to establish a hybrid control arm in randomized clinical trials.

Area of Science:

• Pharmaceutical Sciences
• Biostatistics
• Clinical Trial Design

Background:

• External real-world data, including historical clinical trials, is increasingly used in drug development.
• Current methods for incorporating external data often rely on matching baseline characteristics, which may not fully account for outcome differences.
• Ensuring outcome comparability is crucial for effectively leveraging external data in clinical trials.

Purpose of the Study:

• To propose a novel resampling method for augmenting control arms in randomized controlled trials (RCTs) within a conditional borrowing framework.
• To establish a similarity measure for determining the extent of borrowing external data.
• To evaluate the performance of the proposed hybrid control arm method.

Main Methods:

• A resampling method is proposed to augment control arms in randomized controlled trials (RCTs).
• The method establishes empirical distributions for hazard ratios between external and trial control subjects.
• Borrowing decisions and weights are determined by a similarity measure, incorporated into a weighted partial likelihood for treatment effect evaluation.

Main Results:

• The operating characteristics of the hybrid control arm were evaluated under both conditional and unconditional borrowing frameworks.
• Simulation studies assessed Type I error, bias, and statistical power.
• An illustrative example using simulated data was presented to demonstrate the method's application.

Conclusions:

• The proposed resampling method offers a statistically sound approach for integrating external data into RCT control arms.
• Conditional borrowing, guided by outcome similarity, can enhance the efficiency and reliability of hybrid trial designs.
• This methodology holds potential for optimizing drug development by leveraging real-world evidence more effectively.