Traditionally, the key analyses in clinical trials are analyzed independently of wider evidence applying a frequentist hypothesis testing approach. However, in situations of high-unmet medical need, which often occur in pediatrics and rare diseases, it is impossible to conduct a full set of phase III clinical studies. In these cases, ignoring previous information may result in less accurate estimates. To address this issue, Bayesian methods that borrow strength from previous information have been proposed. This talk will share the application of a Bayesian method in phases III clinical trials to extrapolate efficacy from source to target population. We used a weakly informative prior for the model parameters in the source population and an informative prior for the parameters in the target population patients. The informative prior is a mixture of the posterior distribution obtained from the source population and of a weakly informative prior. The inclusion of the weakly informative mixture component provides robustness in case of prior-data conflict. A posterior predictive model diagnostic check was used to check the compatibility between the source and target data.