RFENDTC is less than RFSTDTC

Hi Venkata,   Yes, you should

Hi Venkata,

Yes, you should explain to FDA all "not-fixable" data management issues in a special document- "Reviewers Guide".

Another option could be to re-evaluate your derivation algorithms for RFSTDTC and RFENDTC variables. However they are expected to be clear and consistent across all your studies within a submission.

I would recommend discussing with your team what is it the best option for your particular case?

Regards,

Sergiy Sirichenko

OpenCDISC

Thanks, Sergiy

Normal 0 false false false EN-US X-NONE X-NONE

In my perspective, capturing screen failure data always a tricky because we don’t have full control on real time concerns ( things may not happen the way it kept on guidelines and what we intended) in our case – I don’t see anything wrong in the database – site was mistakenly made a call to IVRS system and IVRS system did its job (as expected) and CRF clearly stated 4 were screen failure subjects. We cannot do anything more here unfortunately.

As I said we have below options to correct in SDTM domains:

Screen Failures: Change RFSTDTC and RFENDTC derivation for this project – RFSTDTC and RFENDTC should be null if subject is screen failure – don’t look for any dates in EX and IVRS datasets.

Not collected visit time: Add time value as “23:59:59” as a default value for all subjects who randomized – it will take care of the subjects who discontinued on same day of randomization.

Above steps will satisfy the Open CDISC validation checks but my concern is does it acceptable practice to do it?

I was thinking on below options as well –

1.Acknowledge the issues in SDTM domains and document it.

2.Correct the RFSTDTC and RFENDTC @ ADSL level ( we have more flexibility at ADaM level than SDTM level) – that will take care of analysis level issues if any.

Here I am looking for industry experts opinion - which is the best way to move forward – I greatly appreciate your swift suggestions and assistance.

Thanks,

Venkata Maguluri.

“RFSTDTC and RFENDTC should

“RFSTDTC and RFENDTC should be null if subject is screen failure – don’t look for any dates
in EX and IVRS datasets.“

In general, this is not good approach, because you are losing information about Randomization, which very important for review. FDA statisticians have special procedures around analysis of Randomization data. You have to submit Randomization results and additional information for study protocol violations in DV domain.

For subjects who failed screening criteria, but were actually randomized by mistake, ARM variable values are “Treatment X”, “Placebo”, etc., but not “Screen Failure”.

ARM/ARMCD values are defined by Randomization process. Additional values like “Screen Failures” and “Not Assigned” can be used only for subjects who were not randomized.

SD1002 (RFSTDTC is after RFENDTC) check should handle a case when both time points are on the same day and at least one variable has only date part (missing time part). You do not need to populate time as “23:59:59”, which is a kind of imputations to avoid in SDTM data.

Regards, 

Sergiy

Thanks alot Sergiy,

Thanks lot for swift response - it helpfull info:).

Venakta Maguluri.