Designing a response-over-continuous-intervention (ROCI) randomised trial: Implementation in the Phase 2C part...

Treatments for tuberculosis (TB) are often long and complicated. Standard 2-arm non-inferiority trials have been used to evaluate shorter durations of treatment regimens. The new response-over-continuous-intervention (ROCI) trial design has recently been proposed as a practical alternative for optimising some continuous aspect (e.g. treatment duration) of treatment administration.

We demonstrate the use of simulation for designing a ROCI trial in the TB setting. We use the Phase 2C part (duration ranging) of the PARADIGM4TB trial as a case study to illustrate the simulation procedure and the important design considerations to be explored in simulation. Phase 2C of PARADIGM4TB aims to optimise durations of novel treatment regimens, compared to a 6-month standard-of-care treatment regimen, with the aim to support advancement to Phase 3 trials.

A ROCI design randomising 200 patients to 5 equally spaced duration arms of the novel treatment regimen (with an additional 40 patients randomised to the standard-of-care treatment regimen) is sufficient to achieve reasonable power to identify the optimal duration in a range of scenarios. Modelling the duration-response curve with a fractional polynomial model of degree 1 improves power to select shorter durations compared with pairwise comparisons. A design with 5 durations of the novel regimen is preferred to a design with 3 durations, because of the improved operating characteristics in scenarios where the duration-response curve is not flat.

The ROCI design is an appealing design option for TB treatment trials. The design of ROCI trials can be done by conducting simulation studies to explore key design considerations.