|
Abstract:
|
Pediatric trials are an important component of drug development and are typically conducted when the efficacy and safety has been established in adult population. Many disease areas are presented with inherent difficulties in conducting large pediatrics trials making such development highly challenging. When pediatric trials are conducted after adult trials, it allows for utilization of prior adult information to help enhance efficiencies for scenarios that justify borrowing prior information. Innovative clinical trials using Bayesian framework to leverage historical information have been discussed extensively in recent years. This framework helps alleviate logistical and ethical challenges in conducting large clinical trials in children by providing avenues for designing pediatric trials with feasible sample size which in turn facilitate timely decision making and treatment access to children. We consider three frequently used Bayesian frameworks including hierarchical model, power prior, mixture prior, and explore their relationship with one another. Chen & Ibrahim (2006) established an 1-1 correspondence between power prior and hierarchical model under certain settings. In this paper, we further established relationship between power prior and mixture prior using their operating characteristics. This newly established relationship allows unifying the three commonly used Bayesian frameworks: hierarchical model, power prior, and mixture priors. We demonstrate the correspondence across three methods using the key parameters that controls the amount of borrowing of historical information via simulations. Although discussed in the context of pediatrics development, the application is this work is applicable for any other scenario of borrowing prior information.
|
