Statistical Modeling of Clinical Trials (Recruitment and Randomization)

Abstract

This article is devoted to developing further a statistical technique for modeling patient recruitment together with randomization process in multicentre clinical trials. The analytic technique for predicting the number of patients recruited in different centers/regions for ongoing trials accounting for possible delays and closure of some centers is developed. The asymptotic properties of the recruitment in particular regions are investigated and the analysis of recruitment performance in centers/regions is provided. The approximations and predictive confidence bounds for the number of randomized patients in some region using center-stratified randomization are derived. These results are used for creating software tools for predictive patient recruitment and drug supply modeling in GlaxoSmithKline.

Keywords:

• Estimation
• Multicentre clinical trials
• Patient recruitment
• Prediction
• Randomization

Mathematics Subject Classification:

• 60J
• 60K
• 62M05
• 62M20

Acknowledgments

The author would like to thank Dr. Valerii Fedorov and the Drug Development Sciences, Global Clinical Operations and Supply Operations teams at GlaxoSmithKline for continuous discussions. The author is also grateful to Prof. Stephen Senn for helpful comments.