Application of Bayesian Borrowing Methods in Clinical Trials for Children with Type II Diabetes Mellitus

Abstract

Clinical trials for children with type II diabetes mellitus (T2DM) pose challenges often due to recruitment issues. The variability in the treatment effect for pediatrics with T2DM tends to be much larger than that for adults, therefore, a larger pediatric study is needed to independently detect a similar treatment effect. If leveraging adult information to a pediatric population can be appropriately justified, and scientific rational has been given for the relevancy of the adult information, then Bayesian borrowing methods can aid in reducing the number of patients needed for a pediatric study and therefore increase feasibility and efficiency. We introduce Bayesian borrowing methods to obtain scientifically sound and conclusive results with adequate study power in anti-diabetic products development for children with T2DM. To apply Bayesian borrowing methods, it is important to (1) identify the external data that can be leveraged, (2) pre-specify model parameters, (3) assess operating characteristics, and (4) pre-specify weights and the maximum amount of borrowing needed to achieve a study win. To protect against prior-data conflicts that may exist due to differences in T2DM between adults and pediatrics, we select a mixture prior to take an advantage that they adjust the degree of information borrowed based on the similarity between adults and pediatrics. We applied these methods to an anti-diabetic product. In conclusion, the outcomes of our step-by-step demonstration of the application of Bayesian borrowing methods provides a guide on how to pre-specify parameters and considerations that should be made when planning to implement said methods.

Keywords: Bayesian borrowing methods; Effective sample size (ESS); Operating characteristics; Type 2 diabetes mellitus in pediatrics.