Bioequivalence and Bioavailability Forum

Dear Ratnakar!

❝ My concern is current EU guideline specifically mention about the decesion to withdraw a subject should be taken before analysis & this should not creat any issues later on, as I got a study where withdrwing a subject at statistical analysis stage reverse the results of the study?

I think you go in circles. The GL is pretty clear (as is your original post). Reasons for exclusions have to be stated in the protocol, and if one of the conditions are applicable, the subject has to be exluded before statistics (see the quote in this post). As Ohlbe said, the subject has to be discontinued from the study (no more samples drawn after the event), but samples analyzed and presented (GL 4.1.8 Evaluation / Presentation of data):

Drop-out and withdrawal of subjects should be fully documented. If available, concentration data and pharmacokinetic parameters from such subjects should be presented in the individual listings, but should not be included in the summary statistics.

If the subject erroneously was kept in the study (noncompliance of the investigator – I’ve seen that), report it as a protocol deviation. According to the protocol the primary analysis is without the subject – you may present the full data set as a sensitivity analysis. The GL is pretty clear about exclusion of outliers – not based on statistics alone.
I would consider it bad practice to

• run the BE analysis on the full data set,
  
• detect an outlier,
  
• check to CRFs or ask the clinical staff just to find out that the protocol was violated,
  
• exclude the subject.

If the protocol was violated and results in the sensitivity analysis are contradictory (without the subject BE, full data set not BE) you are in trouble. It smells of fraud (did the subject really vomit – or only in the CRF?). Expect an inspection. Remember that GxP is opposite to the justice system (‘innocent unless proven guilty’).