Welcome to Chinese Journal of Clinical Pharmacology and Therapeutics,Today is Chinese

Chinese Journal of Clinical Pharmacology and Therapeutics ›› 2019, Vol. 24 ›› Issue (2): 164-170.doi: 10.12092/j.issn.1009-2501.2019.02.008

Previous Articles     Next Articles

Sample size estimation for average bioequivalence for highly variable drugs using two replicated crossover designs

WANG Ao, CAI Jingjing, BAI Jianling, LIU Yue, YU Xuanxuan, CHEN Feng, ZHAO Yang, YU Hao   

  1. Department of Biostatistics, School of Public Health, Nanjing Medical University, Nanjing 211166, Jiangsu, China
  • Received:2018-09-16 Revised:2018-12-24 Online:2019-02-26 Published:2019-03-04

Abstract:

AIM: To study the sample size estimation for average bioequivalence for highly variable drug using two replicate crossover designs recommended by FDA and EMA. METHODS: The variance of two replicated crossover designs is derived under the framework of statistical model, and the method of Reference-Scaled Average Bioequivalence recommended by FDA and EMA is combined to estimate the sample size by the relationship between power and sample size. SAS program is compiled to facilitate the application in research for researchers. RESULTS:The variance of two replicated crossover designs is equal, at the same time, combining the Reference Scaled Average Bioequivalence recommended by FDA and EMA, it is found that the sample size required under EMA is larger than the sample size of FDA under the same parameter configuration. When CV equals 30%, sample size estimation of EMA is continuous, while the sample size of FDA is discontinuous. When CV is greater than 50%, because EMA uses a fixed equivalent boundary, the sample size begins to increase, and FDA is still scaled equivalent boundary, so the sample size is decreasing or remains unchanged. CONCLUSION: The sample size estimation based on the variances of two replicated crossover designs using best linear unbiased estimator and Reference-Scaled Average Bioequivalence has the rigorous mathematical statistics. It is hoped to provide a help for sample size estimation for highly variable drugs bioequivalence studies.

Key words: highly variable drug, average bioequivalence, replicated design, sample size

CLC Number: