Bioequivalence and Bioavailability Forum

Dear All,

in a 2x2-crossover BE-Study with 80 subject the subjects are divided in two groups with 40 Subject due to clinical capacity.

Treatment date of group 1: 1.9 (treatment 1) and 8.9 (treatment 2)
Treatment date of group 1: 4.9 (treatment 1) and 11.9 (treatment 2)

With is the most appropriate approach (from statistical and from guideline point of view):

a)
usual 2x2-ANOVA model with Terms SUBJECT PERIOD FORM with 2 Periods
(treatment 1 --> period 1, treatment 2 --> period 2)

proc GLM data=work.dummi;
   class SUBJECT PERIOD FORM;
   model LogAUCT LogCmax = SUBJECT PERIOD FORM/ SS3;
   lsmeans FORM / pdiff CL alpha=0.10;

b)
usual 2x2-ANOVA model (see code above) with Terms SUBJECT PERIOD FORM with 4 Periods
(treatment 1, group 1 --> period 1, treatment 1, group 2 --> period 2, treatment 2, group 1 --> period 3, treatment 2, group 2 --> period 4)

c)
including a TERM 'GROUP' in the ANOVA-Model. And which is the best way ? 2 stage design way without alpha-adjustment ?

d)
way proposed form a non-statistican (of our sponsor)

proc GLM data=work.dummi;
   class SUBJECT GROUP PERIOD FORM;
   model LogAUCT LogCmax = SUBJECT GROUP PERIOD FORM/ SS3;
   lsmeans FORM / pdiff CL alpha=0.10;

approach does not work because of zero degrees of freedom for the factor group

Best regards and thank you in advance

BEman