All Times EDT
Keywords: adaptive design; real world evidence; sample size re-estimation; historical control
Sample size re-estimation has been a useful tool in study design to evaluate if the trial sample size needs to be increased to achieve enough power to detect an expected treatment effect during an interim look. For the clinical studies which patient population is small, to have a fully powered randomized controlled trails is very challenging. In such situation, to borrowing or extrapolate information from real world evidence or real-world data has been well applied in trial design. Depending on the disease indications, real-world data can be sourced from historical trials, natural history studies, and other registries. High quality real-world data, if leveraged properly, has the potential to generate real-world evidence to assist interim decision-making, lower enrollment burden, and reduce study timeline and costs. In this research work, we focus on how sample size re-estimation approach can be applied with the use of historical control to boost trial efficiency. With proper borrowing from historical control, some of the challenges in these high unmet medical need studies could be resolved considerably, e.g. decrease in trial sample size, shorter trial duration, patient being protected from further exposure in potential futile treatment. We examine the strategy in pediatric Type II diabetes trials where recruitment has been challenging and the completion is hardly on time. Simulations under various scenarios, such as borrowing ratio or matching method, are conducted to assess the borrowing strategies, in combination of sample size re-estimation. The type I error for each strategy is reported and compared to demonstrate how the type I error has been controlled. Comparison of power and reductions in sample size are reported to demonstrate the advantages of proposed methods.