Bioequivalence and Bioavailability Forum

Dear Helmut,

Greetings,

We are planning for a three period three sequnce two treatment patial replicate crossover design, where referece will be replicated. BE decision will be made based on Scaled average BE criteria. Here we will be able to estimate intra subject variability of reference only. Since we are applying reference scaling, variability of reference only required not test.

Sequences which will be used in the study are,
         P1  P2  P3
Seq1     T   R   R
Seq2     R   T   R
Seq3     R   R   T

For Estimating the variance components which method wil be appropriate, REML or method of moments. Please suggest me?

The SAS code for repeated mixed effects anova model, which wiil be used for analysis is given below,
PROC MIXED METHOD= REML;
CLASSES SEQ SUBJ PER TRT;
MODEL CMAX = 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;
run;

I think this code will be sufficient for the analysis, where estimation will be using REML method. Is there any modification should be made in the above code? Or any other methods available for estimation? Please suggest me?

Thanks in Advance.

Regards,

Bijin