lechia
☆    

C of U,
2017-05-18 23:47

Posting: # 17370
Views: 2,218
 

 Handling samples with unknown sample collection time [PK / PD]

Hi all,

I'm curious about your approach to handling samples that were collected, but the time of collection is uncertain. Let me explain by some examples:

Example 1: Time of one blood draw was not recorded at all. Only scheduled time is known.

Example 2:

Scheduled timepoint X: 20:05
Recorded time (time recorded by the person collecting the blood): 20:05

Scheduled timepoint X+1: 24:05
Recorded time: 20:07

Scheduled timepoint X+2: 06:05
Recorded time: 06:05

Example 3:

Scheduled and recorded timepoint X: 10:12

Scheduled timepoint X+1: 11:12
Recorded timepoint X+1: 11:21

Scheduled and recorded timepoint X+2: 12:12


In all cases, the person collecting the blood did not record that the deviation was out of window where it had to be if we take the time as written. All other evidence points to the fact that most likely the wrong time was recorded, eg. in example 3, this person would have been busy with another subject at 11:20 and 11:22, so it was virtually impossible for her to have actually collected this blood at 11:21. Most likely 12 was incorrectly written as 21.

I believe that in all cases, the most stringent approach should be taken and the sample should be considered as if it was missing. Seems like the easiest approach to keep agencies happy that no manipulation has taken place. Do you agree or do you then use the nominal time or do you keep the time as recorded even if it's again hard to fathom that a person would collect blood draws scheduled so far away only two minutes apart (example 2)?
ElMaestro
★★★

Belgium?,
2017-05-19 00:32

@ lechia
Posting: # 17371
Views: 2,065
 

 Handling samples with unknown sample collection time

Hi Lechia,

difficult. You should treat these phenomena as per your SOP.
I guess there is none that deals with those scenarios since you ask? I have an opinion, but it is not a very qualified one and I would like to hear other people's views as well.

» Example 1: Time of one blood draw was not recorded at all. Only scheduled time is known.

For PK you are supposed to use the actual time, which is the recorded time. Since the actual time is unknown this sample does not contribute.

» Example 2:
»
» Scheduled timepoint X: 20:05
» Recorded time (time recorded by the person collecting the blood): 20:05
»
» Scheduled timepoint X+1: 24:05
» Recorded time: 20:07
»
» Scheduled timepoint X+2: 06:05
» Recorded time: 06:05

That's annoying. If no SOP deals with this I think you'd need to use the actual (recorded) time point.



» Example 3:
»
» Scheduled and recorded timepoint X: 10:12
»
» Scheduled timepoint X+1: 11:12
» Recorded timepoint X+1: 11:21
»
» Scheduled and recorded timepoint X+2: 12:12
»

Use recorded time.

» In all cases, the person collecting the blood did not record that the deviation was out of window where it had to be if we take the time as written. All other evidence points to the fact that most likely the wrong time was recorded, eg. in example 3, this person would have been busy with another subject at 11:20 and 11:22, so it was virtually impossible for her to have actually collected this blood at 11:21. Most likely 12 was incorrectly written as 21.

Overriding a record on basis of "Most likely" is an invitation for a 483 or even a warning letter. Most likely, the Sponsor has an interesting and creative opinion of what is most likely, especially if the study failed narrowly.:-D


» I believe that in all cases, the most stringent approach should be taken and the sample should be considered as if it was missing. Seems like the easiest approach to keep agencies happy that no manipulation has taken place. Do you agree or do you then use the nominal time or do you keep the time as recorded even if it's again hard to fathom that a person would collect blood draws scheduled so far away only two minutes apart (example 2)?

In the absence of SOPs, the latter.

I guess I am saying generally, that if no recording of the activity exists treat things as absent data (much like an inactive audit trail in chromatography; this in itself can be grounds for data dismissal - check carbamazepine in the FDA database for a very good example). If records exist, be faithful to them. Override only if you have a manifest root cause.

You might also need to think about Root Causes and CAPAs here. Clearly something did go terribly wrong if such events have occurred, and you'd not like to see it happening again. Revision of SOPs, discussion with staff, re-training come to mind.

Le tits now.

Best regards,
ElMaestro
lechia
☆    

C of U,
2017-05-19 22:01

@ ElMaestro
Posting: # 17374
Views: 1,843
 

 Handling samples with unknown sample collection time

» You might also need to think about Root Causes and CAPAs here. Clearly something did go terribly wrong if such events have occurred, and you'd not like to see it happening again. Revision of SOPs, discussion with staff, re-training come to mind.

Thank you ElMaestro. I am asking because I am considering amending the SOP and I am trying to come up with various possible situations and best approaches. Looking forward to input from others.

Thanks.
Activity
 Admin contact
20,250 posts in 4,262 threads, 1,397 registered users;
online 16 (0 registered, 16 guests [including 7 identified bots]).
Forum time (Europe/Vienna): 18:49 CET

Many people tend to look at programming styles and languages like religions:
if you belong to one, you cannot belong to others.
But this analogy is another fallacy.    Niklaus Wirth

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