Randomization in phase II trials: No exemption based on sample size

Conducting randomized clinical trials (RCTs) is challenging. For this reason, in earlier phases of drug development and rare diseases with limited sample sizes, there is a tendency to omit control groups and to replace them with historical information for perceived feasibility arguments. This has significant implications for interpretation of trial results. We present headline results from a trial in stem cell transplantation, where fludarabine and clofarabine were investigated for induction chemotherapy in high-risk acute myeloid leukaemia or advanced myelodysplastic syndrome. The trial was initially designed as single-arm trial (SAT) and is discussed here from the planning and assessment perspective. Changes in the treatment setting or patient selection may lead to 'promising results' in SATs, which are difficult to identify and may raise unwarranted hopes which cannot be confirmed in a subsequent randomized trial. RCTs should be used to achieve a straightforward interpretation of the results, even if the sample size is small. Interpreting SATs is only straightforward in very rare instances where the underlying treatment effect is large and the outcome in each individual patient surpasses all previous clinical experience.