ElMag
☆    

2011-03-03 14:10
(4773 d 19:46 ago)

Posting: # 6694
Views: 7,326
 

 Statistical test for detection of Outliers [Outliers]

Dear Helmut,

I hope that you are fine!
Can you help me out with the following:
The CPMP/QWP/EWP/1401/98 Rev. 1 states regarding the outliers: "The applicant should justify that the calculated intra-subject variability is a reliable estimate and that it is not the result of outliers".
Do have any statistical test in mind (in WinNolin or in SPSS or in another program) by which I can detect the outliers (based on the statistical value of intra-subject variability)? Do the Authorities accept such tests?

I would appreciate your prompt reply!

Thank you!!

ElMag
ElMaestro
★★★

Denmark,
2011-03-03 14:19
(4773 d 19:38 ago)

@ ElMag
Posting: # 6695
Views: 6,440
 

 Statistical test for detection of Outliers

Hi ElMag,

❝ Do have any statistical test in mind (in WinNolin or in SPSS or in another

❝ program) by which I can detect the outliers (based on the statistical value

❝ of intra-subject variability)?


This has been discussed before, e.g. here.

❝ Do the Authorities accept such tests?


Not likely. Many have tried and failed, beacuse "Exclusion of data cannot be accepted on the basis of statistical analysis or for pharmacokinetic reasons alone, because it is impossible to distinguish the formulation effects from other effects influencing the pharmacokinetics."

Pass or fail!
ElMaestro
ElMag
☆    

2011-03-03 15:12
(4773 d 18:45 ago)

@ ElMaestro
Posting: # 6697
Views: 6,336
 

 Statistical test for detection of Outliers

Let me provide some further clarity to my original question.
As per the EMA Bioequivalence guideline, under section 4.1.10 on Highly Variable Drugs, it is depicted that the applicant should justify that the calculated intra-subject variability for the Cmax of the reference product is a reliable estimate and that it is not the result of outliers.
What we would therefore like to enquire is what statistical test could be used (and is perhaps accepted from the authorities as per your experience?) in order to demonstrate that the calculation of the intra-subject variability is in fact a reliable estimate and that outliers have not been included in the aforementioned calculation.
Would it be possible to use WinNonlin for this purpose? Or perhaps SPSS or SAS?
Thank you in advance for your response!
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2011-03-03 15:55
(4773 d 18:02 ago)

@ ElMag
Posting: # 6699
Views: 6,360
 

 Statistical test for detection of Outliers

Dear ElMag!

❝ As per the EMA Bioequivalence guideline, under section 4.1.10 on Highly Variable Drugs, it is depicted that the applicant should justify that the calculated intra-subject variability for the Cmax of the reference product is a reliable estimate and that it is not the result of outliers.

❝ What we would therefore like to enquire is what statistical test could be used (and is perhaps accepted from the authorities as per your experience?) in order to demonstrate that the calculation of the intra-subject variability is in fact a reliable estimate and that outliers have not been included in the aforementioned calculation.


EMA was asked this question last June to no avail. We are still waiting for the clarification document. I spoke to members of the PK-group last week in Barcelona: The document is already finalized and has to go through the mills of EMA's bureaucracy in order to get published.

❝ Would it be possible to use WinNonlin for this purpose?


WinNonlin (up to the last version 5.3) - no.
In Phoenix/WinNonlin 6.1 at least you get the model's studentized residuals. But actually we are interested in the intra-subject residuals, which are not available. :-(
Let's see what the next release (6.2) announced by the end of March will have to offer…

❝ Or perhaps SPSS or SAS?


SPSS no idea. For SAS see the reference in this post. For a reference about outliers in 2×2 cross-overs see this post. Hotelling's is implemented in package bear for R.

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
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,640 registered users;
73 visitors (0 registered, 73 guests [including 4 identified bots]).
Forum time: 09:57 CET (Europe/Vienna)

Nothing shows a lack of mathematical education more
than an overly precise calculation.    Carl Friedrich Gauß

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