The Response Over Continuous Intervention design, or ROCI design, is used in late-phase trials when an aspect of the treatment is continuous, such as duration, dose or frequency. A continuous treatment implies that there are lots of possible values to test, and we need to select an option that is neither too much nor too little.

What is the ROCI design and why is it relevant?

Most late-phase randomised clinical trials involve choosing two treatment options and then comparing the results between the two groups. Very often, the treatment options are something like ‘short duration vs. long duration’ or ‘no treatment vs. a specific treatment dose’. This approach comes with some challenges. What if we choose these options badly? What if a medium duration is as good as a long duration? What if a lower dose is as good as the dose used in the clinical trial?

The ROCI design aims to avoid choosing arbitrary treatment options and, instead, explore several options to identify which one is the best one. This is done by randomising participants to, for example, a specific treatment duration from a range, and estimating how outcomes depend on duration.

The figure below compares the design of two-arm trials vs ROCI trials. The left panel shows three different two-arm trials; one tested two possible doses, one tested two possible frequencies and one tested two possible durations. The dots in green represent the option that resulted in better outcomes (Min for Dose, Max for Frequency and Min for Duration). The right panel shows three ROCI comparisons to identify the acceptable dose, the acceptable frequency or the acceptable duration. The dots in blue represent the option that resulted in better outcomes.

What have we done?

Our initial motivation to develop the ROCI approach came from the need to design a clinical trial comparing different durations of antibiotic treatment for children with community-acquired pneumonia. This trial, called PediCAP, has now been designed using ROCI.

We have subsequently used this design for trials in various settings, for a total of ten clinical trials, including trials in cancer (REFINE, REFINE-Lung), tuberculosis (UNITE4TB) and emergency care (T4P).

To facilitate the implementation of the ROCI design, we have developed estimands, estimation methods and operating characteristics. More recently, we have also provided guidance on how to design ROCI trials.

In the future, we plan to keep extending the design, for example using time-to-event outcomes, developing more efficient sample size calculation strategies, allowing for multiple stages/interim analyses and developing guidelines on how to best use information from baseline covariate data.

How will this make a difference?

In general, finding out the correct dose or frequency for a new treatment is done in early phase trials. However, the ROCI design enables you to do that in late-phase trials by exploring several treatment options to identify which one is the best one.




Recordings from the "Innovative Phase III trial designs Symposium" (June 2023) - YouTube