Statistics and Its Interface
Volume 11 (2018)
Robust estimate of regional treatment effect in multi-regional randomized clinical trial in global drug development
Pages: 129 – 139
With the globalization of drug development, and thus clinical trials over multiple regions, determining and inferring the regional effects of a treatment under study are of increased interests in the global drug development, and is becoming a new research field. Existing methods mostly use subjectively specified models, and will be more or less deviated from the true one. In practice, we often have some prior knowledge of the model, but are not sure how well it will fit the data. To address this problem, we propose a semiparametric model, which is a mixture with a known parametric and an unknown nonparametric component. The parametric component represents our prior knowledge about the model, and the nonparametric part reflects our uncertainty. In this way, the prior knowledge is effectively incorporated into the robust model, due to the nonparametric component. The model parameters are estimated by maximizing the corresponding profile likelihood, and the null hypothesis of no regional effect is tested using the profile likelihood ratio statistic. We derive the asymptotic properties of the estimators. Simulation studies are then conducted to evaluate the performance of the model, and results show the clear advantages of the proposed method over existing parametric model. Then model is then used to analyze a real multi-regional clinical trial data as an illustration.
clinical trial, hypothesis test, multi-regional effects, profile likelihood, semiparametric model
The research of Tan is partially supported by R01CA164717.
Paper received on 14 July 2016.