In this second post of the “Good Data Submission Doctor” ( read my first post The Master Recipe: Quality and Attention to Detail Matter here) I would like to go through some of my favorite SDTM Frequently Asked Questions. These are questions I regularly receive in my capacity as a CDISC Subject Matter Expert, either from my colleagues or from the sponsor. Let’s start by taking a look at five of the most recent.
1. What are the rules for naming sponsor domains?
If you are absolutely sure that the data can not be mapped to any existing standard domain, first check if there is a reserved name in the CDISC Controlled Terminology version you are using, otherwise the convention is to create a domain starting with X if it is a domain using the Findings observation class, Z if Events and Y if Interventions e.g. YA.
2. Code-list for RACE is non-extensible. Could you please tell me why “MULTIPLE” is acceptable?
“MULTIPLE” is allowed as per SDTM Implementation Guidance (Ig), but the standard validation checks are not always fully aligned with the SDTM Ig. This is for many other instances where MULTIPLE is allowed by SDTM Ig (i.g. action taken in AE). Similarly for “UNKNOWN” for example when a subject refuses to provide race information.
The reviewer’s guide in such situations should contain a justification for this issue.
3. Is it ok to flag --BLFL for Unscheduled VISITS in SDTM. I mean I have two records on or before the Treatment in which Unscheduled is last record.
Yes you can. There are no specific rules on how to derive --BLFL, but usually it is the last observation prior to study drug administration, regardless of type of visit.
4. For the –SEQ variable, do we need to restart at 1 for a new subject or can we go from 1 and increment until the end of the dataset (--seq = _n_)?
Although in the SDTM Ig there is no obligation to start from 1, it makes sense to reset –SEQ for each subject and as such having –SEQ starting from 1 for each subject.
5. I am getting the following message: “Missing End Time-Point value”. My study did not collect whether not the AE was ‘Ongoing’. In the case where AEENDTC is missing could I derive AEENRF and set it to “ONGOING”?
No, the recommendation from the SDTM Ig 3.2 section 220.127.116.11 is to not derive the variable it if is not collected in the CRF. The warning/error from P21 can be justified / documented in the reviewer’s guide.
Cytel offers support for CDISC standards governance, and migrations, as well as consultancy on your CDISC strategy. Click the button below to arrange a discussion about your next project.