Maura
☆    

2010-09-02 19:09
(4773 d 03:39 ago)

Posting: # 5876
Views: 5,577
 

 Cross-over 3x6 WNL x SAS [Software]

I'm trying to get the same results with SAS, for the WinNonlin output, for a cross-over design 3x6 (Chow & Liu, page 321, PURICH's data), without success.
Someone tell me how to get the same results with SAS code?
Following the data, the SAS program and output, and the WinNonlin output.

Subj    Per     SEQ     Droga   asc     Carry   lnasc
1       1       RTS     1       5.68    1       1.736951
2       1       RTS     1       3.6     1       1.280934
3       1       SRT     2       3.55    1       1.266948
4       1       SRT     2       7.31    1       1.989243
5       1       TSR     3       6.59    1       1.885553
6       1       TSR     3       9.68    1       2.270062
7       1       STR     2       4.63    1       1.532557
8       1       STR     2       8.75    1       2.169054
9       1       TRS     3       7.25    1       1.981001
10      1       TRS     3       5       1       1.609438
11      1       RST     1       4.63    1       1.532557
12      1       RST     1       3.87    1       1.353255
1       2       RTS     3       4.21    1       1.437463
2       2       RTS     3       5.01    1       1.611436
3       2       SRT     1       5.07    2       1.623341
4       2       SRT     1       7.42    2       2.004179
5       2       TSR     2       7.72    3       2.043814
6       2       TSR     2       8.91    3       2.187174
7       2       STR     3       7.23    2       1.978239
8       2       STR     3       7.59    2       2.026832
9       2       TRS     1       7.88    3       2.064328
10      2       TRS     1       7.84    3       2.059239
11      2       RST     2       6.77    1       1.912501
12      2       RST     2       7.62    1       2.030776
1       3       RTS     2       6.83    3       1.921325
2       3       RTS     2       5.78    3       1.754404
3       3       SRT     3       4.49    1       1.501853
4       3       SRT     3       7.86    1       2.061787
5       3       TSR     1       7.26    2       1.98238
6       3       TSR     1       9.04    2       2.201659
7       3       STR     1       5.06    3       1.621366
8       3       STR     1       4.82    3       1.572774
9       3       TRS     2       9.02    1       2.199444
10      3       TRS     2       7.79    1       2.052841
11      3       RST     3       5.72    2       1.743969
12      3       RST     3       6.74    2       1.90806



WinNonlin output:
Partial Tests of Model Effects
              Hypothesis    Numer_DF    Denom_DF      F_stat     P_value
------------------------------------------------------------------------
                int            1           6       335.277        0.0000
                seq            5           6.27      1.66382      0.2711
                droga          2          18         3.69067      0.0454
                Per            2          18         3.28399      0.0608
                Carry          2          18         1.21341      0.3204

SAS Program:
PROC GLM DATA=purich;
CLASS seq subj per droga carry;
MODEL lnauc =  droga per seq subj(seq) carry  /SS3;
RANDOM subj(seq)/test;
RUN;
QUIT;


SAS Output:
The GLM Procedure
Tests of Hypotheses for Mixed Model Analysis of Variance

Dependent Variable: lnauc

Source                      DF     Type III SS     Mean Square    F Value    Pr > F

Droga                        2        0.230110        0.115055       3.44    0.0543
Per                          2        0.085202        0.042601       1.27    0.3037
Subj(SEQ)                    6        0.702524        0.117087       3.50    0.0179
Carrytest                    2        0.081128        0.040564       1.21    0.3204

Error: MS(Error)            18        0.601883        0.033438


Source                      DF     Type III SS     Mean Square    F Value    Pr > F

SEQ                          5        0.874041        0.174808       1.58    0.2896

Error                   6.3004        0.695535        0.110395
Error: 0.92*MS(Subj(SEQ)) + 0.08*MS(Error)


Mainly the period effect is very different!
Thanks and sorry about my English!!


Edit: Category changed. Please don't use tabulators in you post (see here on how to use BBCodes) - use the Preview before posting. [Helmut]
d_labes
★★★

Berlin, Germany,
2010-09-03 10:58
(4772 d 11:50 ago)

@ Maura
Posting: # 5879
Views: 4,587
 

 WNL partial F vs. SAS type III F

Dear Maura,

I don't own WNL therefore I can't verify your WNL results.
But from the form of output I guess that you have done a mixed model analysis in WNL?

Then the appropriate SAS code will be:
Proc mixed data=maura;
  class seq subj per droga carry;
  model lnasc=seq per droga carry/DDFM=satterth;
  random subj(seq);
run;

This gives among the output:
...
                   The Mixed Procedure

                   Covariance Parameter
                       Estimates

                 Cov Parm      Estimate

                  Subj(SEQ)      0.02788
                  Residual       0.03344
...
             Type 3 Tests of Fixed Effects

                   Num     Den
     Effect         DF      DF    F Value    Pr > F

     Droga           2      18       3.44    0.0543
     Per             2      18       1.27    0.3037
     SEQ             5    6.32       1.56    0.2956
     Carry           2      18       1.21    0.3204

This is very close to your results with Proc GLM, as I had expected.

Thus I can't figure out why WNL gives distinct F-tests.
Check your model specifications and data thoroughly.

Maybe that WNL partial F-tests do not correspond to SAS type III tests? Any of the WNL owners out there with an opinion?

❝ Thanks and sorry about my English!!


(emphasis by me) Welcome to the club.:-D

Regards,

Detlew
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2010-09-03 13:13
(4772 d 09:35 ago)

@ d_labes
Posting: # 5880
Views: 4,802
 

 WNL partial F vs. SAS type III F

Dear D. Labes & Maura,

❝ This gives among the [SAS Proc mixed] output:

...

                   The Mixed Procedure


                   Covariance Parameter

                       Estimates


                 Cov Parm      Estimate


                  Subj(SEQ)      0.02788

                  Residual       0.03344

...

             Type 3 Tests of Fixed Effects


                   Num     Den

     Effect         DF      DF    F Value    Pr > F


     Droga           2      18       3.44    0.0543

     Per             2      18       1.27    0.3037

     SEQ             5    6.32       1.56    0.2956

     Carry           2      18       1.21    0.3204

❝ This is very close to your results with Proc GLM, as I had expected.


❝ Thus I can't figure out why WNL gives distinct F-tests.

❝ Check your model specifications and data thoroughly.


OK, here my results (Phoenix/WinNonlin 6.1):
Model Specification and User Settings
       Dependent variable : lnAUC
                Transform : Data already ln-transformed
              Fixed terms : int+Sequence+Droga+Period+Carry
    Random/repeated terms : Sequence*Subject

Final variance parameter estimates:
    Var(Sequence*Subject)    0.0278831
            Var(Residual)    0.0334379
          Intersubject CV    0.168153
          Intrasubject CV    0.1844

     REML log(likelihood)      2.96255
 -2* REML log(likelihood)     -5.92511
 Akaike Information Crit.     22.0749
   Schwarz Bayesian Crit.     38.5676

Partial Tests of Model Effects
              Hypothesis    Numer_DF    Denom_DF      F_stat     P_value
------------------------------------------------------------------------
                     int           1        7.78     920.601      0.0000
                Sequence           5        6.31     1.55832      0.2957
                   Droga           2          18     3.44086      0.0543
                  Period           2          18     1.27404      0.3037
                   Carry           2          18     1.21312      0.3204

Partial Sum of Squares
              Hypothesis          DF          SS          MS      F_stat     P_value
------------------------------------------------------------------------------------
                Sequence           5           2         0.4   0.0852022      1.2740
        Sequence*Subject        6.31     4.04403    0.640933     19.1678      0.0000
                   Droga           2     0.23011    0.115055     3.44086      0.0543
                  Period           2   0.0852022   0.0426011     1.27404      0.3037
                   Carry           2   0.0852022   0.0426011     1.27404      0.3037
                   Error          18    0.601883   0.0334379


❝ Maybe that WNL partial F-tests do not correspond to SAS type III tests?


They are almost (!) the same. ;-)

❝ Any of the WNL owners out there with an opinion?


I would say wrong coding. The model should be specified with:
Fixed Effects Sequence+Droga+Period+Carry
Variance Structure / Random 1 Subject(Sequence) Type Variance Components

Chow/Liu reported carryover effects at p=0.32 and further results (3rd ed. 2009, Tables 10.3.15/16, p322):
Carryover   R      T      S
  Yes      5.67   7.24   6.30
  No       6.01   7.06   6.45

Comparison  Carryover       90% CI
T vs R        Yes      107.20%, 134.41%
              No       104.75%, 129.92%
S vs R        Yes       94.47%, 119.68%
              No        92.62%, 119.79%
T vs S        Yes      101.63%, 129.83%
              No        97.54%, 122.72%

I followed Maura's coding here; R=solution (R), T=domestic tablet (T1), S=European tablet (T2). According to footnote (a): Calculations were based upon [...] an estimate of solution formulation mean which is 5.97 (??) in the presence of carryover effects and is 6.01 in the absence of carryover effects.

I got:
Carryover   R      T      S
  Yes      5.77   7.16   6.22
  No       5.77   6.83   6.26

Comparison  Carryover       90% CI
T vs R        Yes      107.35%, 143.40%
              No       103.95%, 134.86%
S vs R        Yes       93.30%, 124.63%
              No        95.29%, 123.62%


No idea whether the LSM of the reference in Chow/Liu contains another typo (5.67 ↔ 5.97) - and which value they have actually used in calculating the 90% CI.

@Maura: Where did you get the coding of carryover from?

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

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

Berlin, Germany,
2010-09-03 14:10
(4772 d 08:38 ago)

@ Helmut
Posting: # 5881
Views: 4,576
 

 WNL almost the same as SAS

Dear Helmut!

❝ OK, here my results (Phoenix/WinNonlin 6.1):

❝ ...

❝ They are almost (!) the same. ;-)


Very fine :ok:!

❝ Chow/Liu reported carryover effects at p=0.32 and further results

❝ (3rd ed. 2009, Tables 10.3.15/16, p322):

Carryover   R      T      S

   Yes     5.67   7.24   6.30

   No      6.01   7.06   6.45

...


From the values and a short look into the book it seems they have analyzed the AUC values un-transformed! Not a very good idea with respect to nearly all guidances around the world :no:.

@Maura: I agree with Helmut. There is a coding error in your carry-over factor. In the first period there is no carry-over possible.
BTW: I would not take a model with carry-over if no very strong arguments are present that such an effect could have occured. An sufficient wash-out period should have avoided that.
If not the simple first order carry-over model is not an efficient tool to deal with.
See
Stephen Senn
"Cross-over Trials in Clinical Research"
Chapter 10
Second edition
Wiley, Chichester, 2002

Regards,

Detlew
Maura
☆    

2010-09-03 16:07
(4772 d 06:41 ago)

@ d_labes
Posting: # 5883
Views: 4,610
 

 WNL almost the same as SAS

Dear all,

First I am very thank you for help!!
Where did I get the coding of carryover from? This was removed from the forum itself. I found the encoding of an example of a book (C5300.zip):

/***SAS code for examples in Chapter 4 of                              ***/
/*** Bioequivalence and Statistics in Clinical Pharmacology,           ***/
/*** by Scott Patterson and Byron Jones,***/
/*** Chapman and Hall/CRC Press: Boca Raton, London and New York, 2005


In example 4.5 the data is arranged in the same way. I create the column “Carry” based on this example.

Those results from WNL was a friend who showed me, because I don't have the software. The results achieved through the SAS I get well.
I am very happy to have achieved these results with the help of you, but what I really need is to estimate the effect of "carryover" with one degree of freedom (ANOVA in Chow & Liu page 366 - second edition), when this same model is used to drug interaction (with x alone), and so I would not know how to create the column "Carry".
Recalling that in Brazil, ANVISA is criticizing this design, so we are trying to estimate the carryover effect and determining its significance, trying to justify the design. ANVISA wants it done a study for bioequivalence between "A + I" and "A" and another for "A + I" and "I".
Does anyone know how to do this parameterization??!

Thank you all.


Edit: Full quote removed. Please delete anything from the text of the original poster which is not necessary in understanding your answer; see also this post! [Helmut]
UA Flag
Activity
 Admin contact
22,761 posts in 4,775 threads, 1,628 registered users;
21 visitors (0 registered, 21 guests [including 6 identified bots]).
Forum time: 22:48 CEST (Europe/Vienna)

Medical researches can be divided into two sorts:
those who think that meta is better and those
who believe that pooling is fooling.    Stephen Senn

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5