Abstract: 

In designing a group-randomized trial (GRT), an important step is estimating the number of groups required and group allocation to guarantee the planned study has sufficient power to test the hypothesis of interest; superiority, non-inferiority, or bioequivalence over a competitor(s). If results and information from prior studies are accessible and can be utilized with a new trial to resolve research questions, then the utility of prior investments in research is increased. Currently, the approach to determining the sample size and treatment allocation ratio assumes no other data are available to answer the question. In this paper, an approach is proposed to increase the power of a subsequent group-randomized trial if the design and analysis integrate information from a network meta-analysis (NMA). We illustrate the results from a network meta-analysis, which includes the reference drug product and the placebo to figure out the optimal group allocation strategy for a new group randomized clinical endpoint study. The optimal group allocation strategy minimizes the standard error of the comparative efficacy of the test to the reference compared with other sample allocation strategies. Using simulations, we verify that information can be borrowed from a network meta-analysis to modify the group allocation ratio over treatments and improve the power of the new trial given a fixed total number of groups or reduce the total number of groups required to achieve the desired power.