Conclusions

CTS can be viewed as an extension of conventional statistical power analysis for selecting and adequately sizing a design. CTS allows us to consider a richer set of drug development decisions and design optimization criteria than simply "is there evidence for efficacy?" A full-scale simulation project, such as the one we describe, requires a larger investment of time than a simple power analysis, so it will not be the best choice for all drug development programs. The analysis methods used for the simulated trial data were candidates for the actual trial's formal analysis plan. Further, the more difficult task of integrating the analysis results into an evaluation framework for drug development decisions was attempted. The simulations helped set expectations about the likelihood of success for these more complex decisions.

Despite efforts to be as realistic as possible in the models and criteria, this work just provides an approximation to reality. Inevitably, design questions will be asked that stretch the limits of the models' relevance. For instance, when design factors have opposing impacts (as with the total sample size and number of groups), the models might predict a specific configuration to be optimum. However, different model assumptions might move this optimal performance to another configuration, begging the question of how literally the simulation results should be taken. Thus the simulation results must be integrated with everything else known about the design and the scientific setting to develop a trial plan that has the best chances for success.

Both the interdisciplinary planning and the review of the simulation results contributed to the progress of the clinical program development. This helped to focus discussion upon the specific trial goals and decisions to be made, while also helping us to incorporate, in a coherent manner, the knowledge available from prior clinical and nonclinical sources.

Was this article helpful?

0 0

Post a comment