Matching design for augmenting the control arm of a randomized controlled trial using real-world data

Area of Science:

• Clinical Trials Methodology
• Real-World Data Analysis
• Drug Development

Background:

• Randomized clinical trials (RCTs) are the gold standard but face limitations like small patient populations and ethical issues in rare diseases.
• Real-world data (RWD) and historical data offer potential solutions to augment or replace control arms in current trials.
• Challenges include balancing covariates and ensuring comparability between historical and concurrent controls when using RWD.

Purpose of the Study:

• To propose and evaluate two novel matching methods for incorporating historical control data into clinical trials.
• To enhance statistical power by pooling matched historical and concurrent control subjects.
• To mitigate bias and ensure the comparability of responses between control groups.

Main Methods:

• Development of two matching techniques to balance observed baseline covariates between historical and concurrent control groups.
• Ensuring comparability of response variables between historical and concurrent controls to reduce Type I error.
• Utilizing simulation studies to assess the performance of the proposed matching methods regarding Type I error rate and statistical power.

Main Results:

• The proposed matching methods demonstrate the ability to balance key baseline covariates.
• Ensuring response comparability between control groups helps reduce Type I error inflation.
• Simulation results indicate the effectiveness of the methods in boosting statistical power.

Conclusions:

• The proposed matching methods offer a viable strategy for leveraging historical control data in clinical trials.
• These methods can enhance statistical power and provide protection against unmeasured confounding bias.
• The approach is particularly relevant for rare disease studies where traditional RCTs are challenging.