Implementation of Response-Adaptive Randomisation in a Rare Disease Setting

Well-designed randomised controlled trials (RCTs) are recognised as the gold standard for conducting evidence-based clinical research to assess the efficacy of interventions. Yet, standard RCTs can require substantial time and resources, and therefore can become impractical in cases such as rare diseases where recruitment process is slow and limited in size. Response-adaptive randomisation (RAR) designs aim to address some of these issues by increasing the likelihood of allocation to the most promising treatment arm based on accumulating responses, while maintaining randomisation, which is particularly appealing in rare diseases where patient populations are small. However, implementing RAR in such settings introduces practical challenges, including undesirable treatment allocations.

Motivated by StratosPHere 2, an ongoing trial in a rare disease setting, we propose Mapping as a decision strategy that converts the continuous randomisation probabilities produced at the interim stage into a target vector of discrete allocation ratios. This approach helps to avoid undesirable treatment allocations while remaining faithful to the principles of RAR. Comparing the performance of mapped designs with the original trial design, we find that implementing Mapping improves efficiency in terms of both statistical performance and patient benefit. Overall, this work demonstrates how adaptive methods can be made operationally feasible without compromising methodological rigour, and highlights the importance of bridging methodological innovation with practical implementation to support the use of RAR in clinical practice.

Dr Rajenki Das (MRC Biostatistics Unit, University of Cambridge)

Contact Lexy Sorrell and Victoria Volodina for South West Local Group.