A comparison between restricted maximum likelihood method and method of moments for evaluation of individual bioequivalence

• Main Authors: Feng Shih-Ping, 馮詩蘋
• Other Authors: Liu, J.P.
• Format: Others
• Language: en_US
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/83299548167365923623

碩士 === 國立成功大學 === 統計學系 === 88 === Abstract The aim of bioequivalence trials is to demonstrate two formulations (test and reference) of a drug have similar bioavailability. In 1997, the U.S. Food and Drug Administration (FDA) proposed an aggregate criterion, called CR3, to assess individual bioequivalence and CR3 can be divided into the squared difference in population averages, variance of the subject-by-formulation interaction, and the difference in intra-subject variability between the test and reference formulations, all scaled by the reference intra-subject variability. To have the estimators of CR3, FDA (1998) proposed Moments method and Restricted Maximum Likelihood method (REML) to estimate individual parameter. So in our thesis, the purpose is to compare the size, power, bias and relative bias between Moments method and REML method for evaluation of individual bioequivalence. On the other hand, because CR3 is second moments and would be influenced by the underlying distribution. Based on this reason, we also considered our simulation under four different distributions, which are Normal, Uniform, Exponential and Cauchy distributions. Key words : Individual bioequivalence, bioavailability, subject-by-formulation interaction, intra-subject variability, Moments method, Restricted Maximum Likelihood method (REML), size, power, bias and relative bias