MGR
★    

India,
2006-12-16 09:36
(6313 d 06:46 ago)

Posting: # 400
Views: 8,146
 

 Statistical analysis for a parallel study [General Sta­tis­tics]

Dear All,

I have got a pivotal parallel study (two treatment) for statistical analysis using SAS software. Since I have not dealt with this type of design in SAS soft ware, could anybody suggest me how to proceed with the statistical analysis.

Or else please tell me is it ANOVA one-way procedure used for a parallel study (two treatment) in SAS soft ware,In ANOVA one way could anybody tell me which test to be applied (Bartlette, Duncan.....and so on).

And also please tell me what factors/effects (treatment or subject or both) should be taken for statistical analysis in ANOVA one-way procedure

Please tell me any guideline or books or any websites which gives a clear picture regarding the statistical analysis of a parallel study.

Note: Kindly treat this as high priority and reply me at the earliest as it is very urgent.

Regards,
MGR
H_Rotter
★    

Germany,
2006-12-16 14:48
(6313 d 01:33 ago)

(edited by H_Rotter on 2006-12-16 15:25)
@ MGR
Posting: # 402
Views: 6,666
 

 Statistical analysis for a parallel study

Dear MGR!

❝ Or else please tell me is it ANOVA one-way procedure used for a parallel study (two treatment) in SAS soft ware...


Yes, it's a one-way ANOVA (independent samples, unequal variances).

❝ ...In ANOVA one way could anybody tell me which test to be applied (Bartlette, Duncan.....and so on).


Bartlett tests for equal variances (which is not recommended by the FDA).
If you are dealing with one reference: Dunnett (not only for two treatments); Duncan is a stepwise test not suitable for BE...

❝ And also please tell me what factors/effects (treatment or subject or both) should be taken for statistical analysis in ANOVA one-way procedure


Treatment only.

Regards,
Hermann
MGR
★    

India,
2006-12-27 07:54
(6302 d 08:28 ago)

@ H_Rotter
Posting: # 420
Views: 6,733
 

 Statistical analysis for a parallel study

Hi Hermann,

Thanks for your reply.

And a small doubt, you told that bartlett test is not recommended by FDA. So which test is suitable for the Analysis if it is "An open label, balanced, randomized, single treatment, single period, single dose, parallel, bioequivalence study". Since it is a pivotal study, which test is recommended by the FDA for this type of study.

Iam very grateful :-) to you if you reply me about this clarification.

Regards,
MGR
H_Rotter
★    

Germany,
2006-12-27 14:00
(6302 d 02:21 ago)

@ MGR
Posting: # 421
Views: 6,526
 

 Statistical analysis for a parallel study

Dear MGR!

❝ Thanks for your reply.


You are welcome!

❝ And a small doubt, you told that bartlett test is not recommended by FDA.


OK, maybe my wording was ambigious (English is not my native language)...
Maybe better would have been:
'Bartlett's Test' tests for equal variances (which is not recommended by the FDA).
Have a look at FDA's Guideline on http://www.fda.gov/cder/guidance/3616fnl.pdf Statistical Approaches Establishing Bioequivalence (VI. STATISTICAL ANALYSIS, B. Data Analysis, 1. Average Bioequivalence, d. Parallel Designs):
'For parallel designs, the confidence interval for the difference of means in the log scale can be computed using the total between-subject variance. As in the analysis for replicated designs (section VI. B.1.b), equal variances should not be assumed.'
In other words you should not test the assumption of equal variances - what Bartlett would do - but simply not assume equal variances.

❝ So which test is suitable for the Analysis if it is "An open label, balanced, randomized, single treatment, single period, single dose, parallel, bioequivalence study".


One-way ANOVA on log-transformed data, confidence intervals for treatment difference(s) (see above).
From the title it's still not clear to me how many treatments you will have in your study. Only if you will have more than two treatments (e.g. 2 tests and 1 reference), you should go with Duncan's Test (multiple comparisons).

❝ Since it is a pivotal study, which test is recommended by the FDA for this type of study.


If you go with the FDA, they will have to approve your protocol anyway ;-)

Good luck,
Hermann


Edit: Link corrected for FDA’s new site. [Helmut]
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
100 visitors (0 registered, 100 guests [including 9 identified bots]).
Forum time: 16:22 CET (Europe/Vienna)

With four parameters I can fit an elephant,
and with five I can make him wiggle his trunk.    John von Neumann

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