Keywords: external control, concurrent control, randomized clinical trial
Randomized clinical study design is commonly used and most desirable to minimize biases in evaluating treatment effect. When it is necessary to minimize the number of patients exposed to the concurrent control (e.g., placebo), an external control (e.g., from natural history) may need to be relied upon and form a hybrid control arm, which is composed of the concurrent control and the external control. The use of such a hybrid control has surfaced in regulatory applications and may provide an opportunity to properly weight the external control in constructing the hybrid, depending on the similarity/dissimilarity between the concurrent control and the external control. Certainly all the issues with use of natural history remain. This talk will focus on statistical issues with use of hybrid control for design and analysis.