SIMPLE—A modular tool for simulating complex platform trials

Abstract

Platform trials are becoming increasingly popular within drug development, attracting interest by patients, clinicians, regulatory agencies and statisticians. More often than not, these platform trial designs are highly complex and involve many weakly predictable events (e.g. number of investigational treatments that will enter over time) to determine the impact of relevant design parameters (e.g. decision rules, sharing of information across cohorts and allocation ratios) on the operating characteristics with high confidence. Simulations may address these uncertainties at the design stage. However, the number and combination of design elements for potential implementation in real platform trials is immense. As a result, simulation software which is developed based on specific project needs is typically limited in the variety of available design options for comparison, as such software is developed for a particular need, not for researching all potential new approaches to clinical research and statistical science. On the other hand, software solutions which allow for a wide range of design options may easily overload the user with requirements for design specifications. We have developed an R software package (“SIMPLE”), which is modular in the sense that if users want to simulate the most common platform designs, minimal effort and understanding of the package is needed, but it allows the users to take control of different parts of the simulation (e.g. patient accrual, outcome simulation, etc.) step-by-step, thereby facilitating the simulation of arbitrarily complex platform trials. We will give an overview of this software package alongside some examples on how to simulate common platform trial designs and derive their operating characteristics.