Bioequivalence and Bioavailability Forum

Hi Helmult/everybody

I would like to have your opinion about SAS code for the analysis of ABE with partial replicate design data and group effects. According to the FDA's guideline on Statistical Approaches to Establishing Bioequivalence (appendix E), the code to be implemented without group variable is:

PROC MIXED;
CLASSES SEQ SUBJ PER TRT;
MODEL Y = SEQ PER TRT/ DDFM=SATTERTH;
RANDOM TRT/TYPE=FA0(2) SUB=SUBJ G;
REPEATED/GRP=TRT SUB=SUBJ;
ESTIMATE 'T vs. R' TRT 1 -1/CL ALPHA=0.1;

In a previous post, you have written that FDA's advice for analysing ABE with multiple group (according to a letter from Barbara Davit) is to use the following variables:
Group - fixed
Sequence - fixed
Treatment - fixed
Subject(Group*Sequence) - random
Period(Group) - fixed
Group*Sequence - fixed
Group*Treatment - fixed

Therefore, and as I'm a total roockie in SAS with FDA's code, what would be the changes in the above SAS code in order to incorporate Group effects?

And for Phoenix, what you would consider for
Variance Structure [Random 1] Random Effects = ?
Variance Blocking Variables (Subject) = ?
Type= Banded No-Diagonal Factor Analytic (f)? No. of factors = 2?

And for Variance Structure Repeated
Repeated Specification = ?
Variance Blocking Variables(Subject) = ?
Group = ?
Type = ?

In the case of FDA's RSABE, what changes would you include in the SAS code published in the Progesterone Guideline regarding intermediate analysis of ilat and dlat (proc glm) in order to consider also group effects?

Many thanks in advance for the help
Rgs
Silva