I am looking at the bioequivalence calculations and I am doing fine. One aspect that does trouble me is the “Final Fixed Parameters”. I am having difficulty interpreting what it is telling me. I am using AUC data and the 3 way cross-over design with M and F subjects included. I am looking at the output. Gender is included in the model Sequence+Period+Treatment+gender+Treatment*gender There is an output column entitled Effect Level and the treatments A, B and C are given there. The periods are given there i.e. Period 1, Period 2 and Period 3. Gender is there F or M. Also Treatment :A Gender :F and Treatment: A: Gender: M e.tc. Some of the time the p value is less than 0.05 so this is significant. For example I see Gender :F the p value is 0.04. So does the mean value for the Female Gender group is statistically different? If yes different from what? {I see that the Gender: M group could not be estimated.) Please can you comment on this to me in basic terms to explain the output. Angus
Angus, can you post the project so we can better see the data and results you’re seeing? Simon.
No: not at this time. It is confidential information. What I do have and am going to look at is an old example of a 3 way cross- over BE design. I am going to run this again and look at it from the point of view of understanding the basic output. Now that example I can post and you may wish to use that in the company as a good example. Now this example is a balanced design and has been run by me in SAS, Kinetica and WinNonlin (and now Phoenix 6.3) and gives identical results in all three programs. At an earlier date I was studying ANOVA and the output. Also I have NCSS another good high level stat program with ANOVA. I want to understand in basic terms the output we get and what it means. So I will let you know what I can deduce. The “final fixed parameters” may be the model fitted parameters from the ANOVA statistical model? So I will see what I get when I run this old example. I will look at the output to see what I get. I will let you know what I see; hopefully all will be revealed (or at least something). Thanks , ANGUS
Simon: here is a good example of BE. It is a Williams type balanced study design with 6 sequences {3 periods and 3 treatments for the 3 way cross over study}. The parameter is C max. For Treatment A the Cmax is much higher than reference (Treatment B), but Treatments B,C are BE. You will see that if you look at “Final Fixed Parameters” tab the p value for Treatment A is extremely low (highly significant), whereas for Treatment B it is 0.27 {not significant}. I believe that the p value indicates that a test has been made between mean values of Treatment A and "something else’. The same goes for Treatment B. But I do not know what the comparison is being made to. An estimate could not be obtained for Treatment C. The essence of my question is I do not understand the output in this Tab. I attach the file. Now this is a very good example of BE. Angus [file name=BIOEQUIVALENCE_3_PERIODS_AND_6_SEQUENCES.phxproj size=177821]Certara | Drug Development Solutions (174 KB)
These worksheet results are the equivalent of the ‘SOLUTION’ and ‘CL’ options on the MODEL statement in SAS PROC MIXED. There are descriptions of these options here: SAS/STAT(R) 9.2 User's Guide, Second Edition Also see page 415 on the PHX Users Guide.
Thank you… I will look at these references to see how much I can follow. It looks difficult Angus