Speaker: Lauren McKeen, PhD Candidate, Department of Statistics, Iowa State University

Title: Leveraging Information from Network Meta-Analyses to Plan a New Multi-Center Trial with a Novel Treatment

Abstract: Pairwise and network meta-analysis can be used to plan new trials. However, there is no existing literature on planning a new multi-center trial with a novel treatment while incorporating information from prior meta-analyses. Specifically, we consider the case when researchers need to combine a new multi-center trial with an existing network of trials and analyze those data using a random effects model for NMA. Here, we develop an approach that allows researchers to maximize the value of an available evidence base when planning a new multi-center trial with a novel treatment. We derive formulas for estimation of variance and power of a comparative effect of interest when researchers wish to design a multi-center trial that will be analyzed with a random effects model. We also optimize sample size allocation in the new multi-center trial such that the variance of a comparative effect of interest is minimized. Using a real network of trials, we illustrate our methods. That is, we consider how to allocate subjects to a three-arm multi-center trial under different settings of sample size and number of trial locations. We validate the results using a simulation study. When leveraging information from an existing network of trials, equal allocation of subjects is not always the most powerful approach. The methods proposed here can increase the power of a planned multi-center trial, which can both decrease cost of trials and maximize the information from an existing evidence base.