Title: Use of external data in randomized clinical trials
Presenters: Heinz Schmidli and Marius Thomas, Novartis
Abstract: Use of relevant external data can make clinical trials more efficient, by reducing the number of patients required for a new study. Naive direct use of external data is not valid due to differences in patient characteristics and other confounding factors. Instead, we suggest the application of robust Bayesian meta-analytic-predictive methods to use external data in a principled way. To illustrate the framework, we describe in detail the ongoing NEOS trial in children with multiple sclerosis, which uses a Bayesian design to borrow strength from historical trials in adults and children. The design was extensively discussed with the FDA as part of their complex innovative design (CID) pilot program.
Heinz Schmidli, PhD, is working as an Executive Director in the Statistical Methodology group at Novartis, Basel, Switzerland. Since he joined the group in 2007, he contributes to the development of innovative approaches for the design and analysis of clinical trials, with a focus on Bayesian methods.
Marius Thomas, PhD is currently working as a Senior Principal Biostatistician in the Neuroscience development unit of Novartis. Since joining Novartis, he has worked on various projects in the area of multiple sclerosis. He is the trial statistician for the complex innovative NEOS trial in children with multiple sclerosis.