Bioequivalence and Bioavailability Forum • FDA HVD SAS Code

jag009
★★★

NJ,
2013-04-30 19:11
(4396 d 22:16 ago)

Posting: # 10510
Views: 5,261

FDA HVD SAS Code - Average BE using Proc Mixed [RSABE / ABEL]

Hi everyone,

Question regarding the ABE SAS code from FDA's progesterone guidance when SCABE is not applicable due to within subject SD being < 0.294 (Yes I am back with this. Helmut feel free to link this message to the previous one I posted a few days ago). For a three way partial replicate study (Test, Reference x2)

PROC MIXED

data=pk;

CLASSES SEQ SUBJ PER TRT;

MODEL LAUCT = SEQ PER TRT/ DDFM=SATTERTH;

RANDOM TRT/TYPE=FA0(2) SUB=SUBJ G;

REPEATED/GRP=TRT SUB=SUJ;

ESTIMATE 'T vs. R' TRT 1 -1/CL ALPHA=0.1;

ods output Estimates=unsc1;

title1 'unscaled BE 90% CI - guidance version';

title2 'AUCt';

run; data unsc1;

set unsc1;

unscabe_lower=exp(lower);

unscabe_upper=exp(upper);

run;

Proc Mixed output two residuals, a reference and test. Which one does SAS use to compute the 90%CI? My guess is the test? (Yes Helmut, I read your slides but I have doubts snice this is a test vs Refx2 Partial Rep study).

Example:

Covariance Parameter Estimates

Cov Parm  Subjec   Group             Estimate

FA(1,1)   subject                    1.2543

FA(2,1)   subject                    1.3108

FA(2,2)   subject                    0.4234

Residual  subject  formulation Ref   0.3115

Residual  subject  formulation Test  0.3192

T/R Ratio (%)  Lower 90% CI  Upper 90% CI

89.9586        74.5097       108.611

Thanks
John

Helmut
★★★

Vienna, Austria,
2013-04-30 19:44
(4396 d 21:43 ago)

@ jag009
Posting: # 10511
Views: 4,230

Data set?

Post reply

Hi John,

before I try to come up with an answer: Which data set did you use?
BTW, please try to avoid tabs (by copypasting from somewhere else) in posts. The output in (X)HTML looks messy (I edited your post; hopefully I got the columns right). ;-)

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

jag009
★★★

NJ,
2013-04-30 22:06
(4396 d 19:21 ago)

@ Helmut
Posting: # 10513
Views: 4,197

Data set?

Post reply

Thanks Helmut,

Yes the saga continues, same beast... The question is when we run Proc GLM, we use the MSE (there is only 1 which is a composite of both test and reference since it's a T vs R study) to get the intrasubject CV and eventually work out the 90% CI goal post. With Proc Mixed in a partial rep study (T vs R vs R), as shown in my first post with the Covariance Matrix, there are two residuals, one for test and one for reference. Is the test residual similar to the total MSE we get from a T vs R study (Meaning the value is attributed to both products)?

I will post the same dataset (more subjects since I have 100% data). It's even stranger now... Stay tune :-)

PS. Please see my original post (the one which I posted the data).

Thanks
John