Revolutionizing Clinical Trials: The Power of Historical Borrowing

Leveraging Historical Data in Clinical Trials for Enhanced Efficiency and Accuracy

Clinical trials are the backbone of medical research, providing the necessary evidence to bring new treatments to the market. However, the traditional approach to these trials, especially in terms of control arms, often overlooks a valuable resource: historical data. Kurt Viele, a director and senior statistical scientist at Barry Consultants, sheds light on the transformative potential of historical borrowing in clinical trials.

The Rationale Behind Historical Borrowing

The control arm of a clinical trial, which serves as a benchmark against the novel treatment being tested, doesn't exist in isolation. It often has a background in numerous prior studies, whether for initial approval or post-approval research. This accumulated data offers a comprehensive understanding of the control arm's performance, including success rates, mean outcomes, and survival times. The key question is: Can we utilize this historical information to augment current studies and thereby conduct more efficient trials?

The advantages of incorporating historical data are clear. If the previous studies align with the current research context, the historical data can provide:

• More precise estimates
• Reduction of possible biases
• Enhanced decision-making on the efficacy of the new treatment

This approach can potentially shorten trials by up to 20% and allow for a greater focus on the treatment arm, improving the assessment of safety and efficacy.

Addressing the Challenges of Historical Borrowing

However, the application of historical data is not without its challenges. Borrowing from an inappropriate source, or a temporal drift in the control arm's behavior, can introduce bias, decreasing the trial's power and increasing the risk of Type 1 errors. To maximize the benefits and minimize the drawbacks of historical borrowing, it's essential to:

• Determine the appropriate weight for historical data
• Use dynamic borrowing based on accruing data to adjust weights

The Methodology of Dynamic Borrowing

Dynamic borrowing involves using hierarchical models to estimate the weight of historical data dynamically, based on the observed drift. This approach allows for the adjustment of the weight given to historical data as the trial progresses, offering a balanced risk and maximizing the potential benefits.

Key Benefits:

• Improved Point Estimates: Leveraging historical data can lead to lower mean square error and reduced Type 1 error, as long as the drift remains small.
• Increased Power: With dynamic borrowing, it's possible to achieve comparable power with a smaller sample size, leading to more efficient trials.
• Flexibility and Robustness: By adjusting the weight of historical data based on real-time data, dynamic borrowing offers a flexible and robust approach to trial design that can accommodate unexpected shifts in data.

Case Study: Antibiotic Trials

A practical example of dynamic borrowing's potential is seen in antibiotic trials. By incorporating historical data, researchers were able to reduce the required sample size by 20%, while ensuring more participants received the treatment under investigation. This not only made the trial more efficient but also provided a stronger basis for evaluating the treatment's efficacy.

Conclusion

Historical borrowing represents a significant advancement in the design and execution of clinical trials. By intelligently leveraging past data, researchers can conduct more efficient and accurate studies, ultimately accelerating the development of new treatments. As the medical field continues to evolve, approaches like dynamic borrowing will be crucial in meeting the demands of both regulatory bodies and patients.

For a deeper understanding of historical borrowing and its implementation in clinical trials, watch Kurt Viele's insightful talk: Revolutionizing Clinical Trials: The Power of Historical Borrowing.