Bioequivalence and Bioavailability Forum 10:17 CET

Main page Policy/Terms of Use Abbreviations Latest Posts

 Log in |  Register |  Search

Mohamed Yehia
Junior

Egypt,
2018-10-02 21:25

Posting: # 19395
Views: 1,210
 

 Subject exclusion [Outliers]

Hi All,

Concerning what stated in EMEA guidelines, if auc of reference cmax of certain subject is less than 5% of the geometric mean, the subject is excluded.

My question is: when we calculate the geometric mean of the reference auc, shall we include the subject we want to inspect in the calculation of the geometric mean or we shall remove it first before making the calculation?

Thanks.


Edit: Category changed; see also this post #1[Helmut]
Ohlbe
Hero

France,
2018-10-02 21:30

@ Mohamed Yehia
Posting: # 19396
Views: 1,146
 

 Subject exclusion

Dear Mohamed,

» My question is: when we calculate the geometric mean of the reference auc, shall we include the subject we want to inspect in the calculation of the geometric mean or we shall remove it first before making the calculation?

The guideline is clear on this point:
A subject is considered to have very low plasma concentrations if its AUC is less than 5% of reference medicinal product geometric mean AUC (which should be calculated without inclusion of data from the outlying subject).

Regards
Ohlbe
Mohamed Yehia
Junior

Egypt,
2018-10-02 21:38

@ Ohlbe
Posting: # 19397
Views: 1,137
 

 Subject exclusion

Dear Ohlbe,

Thanks for your prompt reply. However, my question is will that subject be included in the outlier calculation or I shall remove it first , then calculate outlier of the rest of volunteers and check if there is outlier or not and then calculate geometric mean?


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post! [Ohlbe]
Ohlbe
Hero

France,
2018-10-02 21:46

@ Mohamed Yehia
Posting: # 19399
Views: 1,134
 

 Subject exclusion

Dear Mohamed,

» [...] will that subject be included in the outlier calculation or I shall remove it first , then calculate outlier of the rest of volunteers and check if there is outlier or not and then calculate geometric mean?

The guideline does not ask for any statistical outlier test to be performed. If you have a subject with very low concentrations, calculate the geometric mean of the AUC of the reference without that subject. If the AUC of the subject for the reference product is lower than 5 % of that mean you have a possibility to exclude him. That's all.

Regards
Ohlbe
Mohamed Yehia
Junior

Egypt,
2018-10-02 22:18

@ Ohlbe
Posting: # 19400
Views: 1,120
 

 Subject exclusion

Dear Ohlbe,

» The guideline does not ask for any statistical outlier test to be performed. If you have a subject with very low concentrations, calculate the geometric mean of the AUC of the reference without that subject. If the AUC of the subject for the reference product is lower than 5 % of that mean you have a possibility to exclude him. That's all.

It is stated that we have to remove outlier subjects for reference auc before making th geometric mean calculation. So I think that we have to include all the subjects inckuding the one we want to check and check for the outlier first and in case there is no outlier we then calculate the geometric mean and check for 5 %.


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post! [Ohlbe]
Ohlbe
Hero

France,
2018-10-02 22:35

@ Mohamed Yehia
Posting: # 19401
Views: 1,103
 

 Subject exclusion

Dear Mohamed,

» It is stated that we have to remove outlier subjects for reference auc before making the geometric mean calculation. So I think that we have to include all the subjects including the one we want to check and check for the outlier first and in case there is no outlier we then calculate the geometric mean and check for 5 %.

I have a different understanding. Read the guideline again. The word "outlier" is not used anywhere in this section (and only once in the guideline, on a totally different topic). We're not looking for statistical outliers here, but for subjects who did not swallow the tablet. This 5% thing is just an arbitrary threshold that EMA is ready to accept to assume that the subject spat the tablet out.

Regards
Ohlbe
Mohamed Yehia
Junior

Egypt,
2018-10-02 22:45

@ Ohlbe
Posting: # 19402
Views: 1,115
 

 Subject exclusion

Dear Ohlbe,

» I have a different understanding. Read the guideline again. The word "outlier" is not used anywhere in this section (and only once in the guideline, on a totally different topic). We're not looking for statistical outliers here, but for subjects who did not swallow the tablet. This 5% thing is just an arbitrary threshold that EMA is ready to accept to assume that the subject spat the tablet out.

a subject with lack of any measurable concentrations or only very low plasma concentrations for reference medicinal product. A subject is considered to have very low plasma concentration if its AUC is less than 5% of reference medicinal product geometric mean AUC (which should be calculated without inclusion of data from the outlying subjects). The exclusion of data due to this reason will only be accepted in exceptional cases and may question the validity of the trial.

So I think that we have to check for outlying subjects first then remove those subjects if found then check for the geometric mean . That's why I think that if the subject is not an outlying subject we will include it in the calculation


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post! Ohlbe]
Ohlbe
Hero

France,
2018-10-02 22:51

@ Mohamed Yehia
Posting: # 19403
Views: 1,096
 

 Subject exclusion

Dear Mohamed,

» [...] which should be calculated without inclusion of data from the outlying subjects).

Apologies. I'm the one who read this paragraph too fast :-(

» So I think that we have to check for outlying subjects first then remove those subjects if found then check for the geometric mean. That's why I think that if the subject is not an outlying subject we will include it in the calculation

I stick to my opinion, let's wait for others.

Regards
Ohlbe
Mohamed Yehia
Junior

Egypt,
2018-10-02 22:55

@ Ohlbe
Posting: # 19404
Views: 1,101
 

 Subject exclusion

» I stick to my opinion, let's wait for others.

Dear Ohlbe,

;-) ok we will wait

Thanks


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post #5[Helmut]
Helmut
Hero
avatar
Homepage
Vienna, Austria,
2018-10-02 23:27

@ Mohamed Yehia
Posting: # 19406
Views: 1,112
 

 Subject exclusion

Salam Mohamed,

first of all: Please follow the Forum’s Policy about avoiding TOFU. THX.

» ;-) ok we will wait

In principle I agree with Ohlbe. But IMHO, “the subject didn’t swallow the drug” is not the only reason behind this statement in the GL. IIRC, this topic appeared for the first time in Rev.1 of the PKWP’s Q&A document in a section about omeprazole (updated in Rev.2, deleted in Rev.12). Esp. old – monolithic – gastric resistant PPI-formulations had problems with the coating. If you dive deeper into this issue you will face further questions, e.g.,
  • Do you start with the most obvious one and calculate the geometric mean of the others? Then you may find another one and repeat the procedure. Where will you end?
  • Note that the GL mentions “outlying subjects” (plural!). How can we classify AUCs as potential outliers without a test? Eye-ball PK?
  • Replicate designs. One low value and one “normal” value. Geometric mean or least squares mean?

Cheers,
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
Mohamed Yehia
Junior

Egypt,
2018-10-02 23:51

@ Helmut
Posting: # 19407
Views: 1,107
 

 Subject exclusion

Salam Helmut

» first of all: Please follow the Forum’s Policy about avoiding TOFU. THX.

Ok I will

» In principle I agree with Ohlbe.

May be but for me we have to include all numbers to check for the IQR and check for outliers


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post #5! You posted TOFU in 85% of your replies and were already warned here. Hence, this is the second warning. If you continue to ignore the Forum’s Policy, you will be blocked for three weeks without further notice. [Helmut]
Helmut
Hero
avatar
Homepage
Vienna, Austria,
2018-10-03 23:47

@ Mohamed Yehia
Posting: # 19408
Views: 994
 

 Subject exclusion

Hi Mohamed,

» » first of all: Please follow the Forum’s Policy about avoiding TOFU. THX.
»
» Ok I will…

… but didn’t do. Why? :confused:

» May be but for me we have to include all numbers to check for the IQR and check for outliers

The IQR includes 50% of the data. So what? If you want to go the nonparametric route (which the EMA hates), consider screening for extreme outliers (construct whiskers with 2× or even 3× the IQR). With the default in most software (1.5× for moderate outliers) you may “detect” a lot.

Cheers,
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
Activity
 Thread view
Bioequivalence and Bioavailability Forum |  Admin contact
18,929 posts in 4,039 threads, 1,286 registered users;
online 5 (1 registered, 4 guests [including 2 identified bots]).

As soon as we abandon our own reason, and are content
to rely upon authority, there is no end to our troubles.    Bertrand Russell

The BIOEQUIVALENCE / BIOAVAILABILITY FORUM is hosted by
BEBAC Ing. Helmut Schütz
HTML5 RSS Feed