Menu Back to Poster-Presentations-Details

P130: Harnessing the Synergistic Effect of Commercial Software and R to Build and Power a Larger Clinical Design Space Exploration

Poster Presenter

Sydney Ringold

Customer Success Manager
Cytel, Inc.
United States

Objectives

The objective is to illustrate the synergistic effect of R and commercial software to compare treatment selection options, including Bayesian options. The combination of R and commercial software allows a statistician to focus more on clinical trial design strategy and less on software development.

Method

Within an oncology study, options to advance treatment arms consisted of a Bayesian rule with advancement of any treatment arm greater than a historical response rate threshold, those greater than both that threshold and control, and comparison of arms to historical data instead of a fixed value.

Results

Commercial software allows for confident and quick design through validated workflows and pre-coded and verified design types, but with this speed and confidence there is also a degree of inflexibility in terms of the methods imbedded into the software. Coding in R allows for almost limitless flexibility in terms of methods, but is dependent on the user’s coding ability, requires time for writing and validation, and demands additional resources for communicating results and design selection. In this oncology study, augmenting commercial software with custom R code enabled easy comparison of various treatment selection approaches beyond what was readily available within the commercial software. Rather than needing a considerable time investment for software development, the integration of two key tools allowed for rapid and efficient design exploration.

Conclusion

The ability to build upon commercial software enables robust, efficient design exploration incorporating elements not available within software. Combining commercial software and custom R code allows the statistician to focus on providing the clinical development team with important strategic input rather than focusing on software development tasks. Furthermore, a thoughtful combination of these different approaches can offer added flexibility and confidence in design simulation, enabling the statistician to design beyond the capabilities of commercial software without sacrificing the time needed to validate and test custom code. Utilizing this combined approach in an oncology study, the statistician was able to evaluate more design options than were available in commercial software and in less time than it would have taken to develop an entire simulation package from scratch. This allowed the team to focus on understanding the benefits and risks of diverse options and select the design that provided the highest likelihood of a successful trial. Integrating custom R code with commercial software can allow statisticians to shift their focus to clinical trial strategies rather than monopolizing their time on software development.