Custom Domain Naming

Custom domain naming

The "convention" is that when a dataset name starts with two characters of an existing domain, it is considered as a "splitted off dataset".
Essentially however, the software should look into what is in the define.xml, to the "Domain" attribute of "ItemGroupDef". If e.g. the value of the "Domain" attribute is "LB" and the "Name" is not "LB", then the software should know that it is a "splitted off" dataset of LB. When the value of "Domain" does not correspond to a published domain, it should know that it is a custom domain.
However, I do not think the Pinnacle21 validator is using this mechanism and is reading the define.xml for this rule.
Another "convention" is that custom domain names start with "X", "Y" or "Z". But conventions are conventions and should not be rules.
 

Custom Domains

Jozef, It is not a convention, that custom domains have to start with X, Y, or Z.

This is a quote from the SDTM 3.2 IG:

To eliminate any risk of a sponsor using a name that CDISC later determines to have a different meaning, domain codes beginning with the 
letters X, Y, or Z have been reserved for the creation of custom domains. Any letter or number may be used in the second position. Note the use of codes beginning with X, Y, or Z is optional, and not required for custom domains.

Custom domains

Thanks Lex for correcting me / refining my answer!
I did indeed not mean that there is an obligation that custom domains must start with X,Y, or Z. Personally, I believe that what Sai is doing is fully legal, or is there a rule that custom domain names may not be longer than 2 characters?

Custom domains

From the same SDTM 3.2 IG:

"Determine the domain code. Check the CDISC Controlled Terminology [see Appendix C – Controlled Terminology] for reserved two-character domain identifiers or abbreviations. If one has not been assigned by CDISC, then the sponsor may select the unique two-character domain code to be used consistently throughout the submission. Apply the two-character domain code to the appropriate variables in the domain."

This strongly suggests 2 character domain names for custom domains.

DOMAIN is defined as a two-character attribute

Per the Model Document, Table 2.2.4:

Domain: Two-character abbreviation for the domain most relevant to the observation. The Domain abbreviation is also used as a prefix for variables to ensure uniqueness when datasets are merged.

Per the SDTM IG, Section 2.2: Datasets and Domains

Each domain dataset is distinguished by a unique, two-character code that should be used consistently throughout
the submission.

This is to accommodate other naming conventions, such as split domains, SUPPxx domains, AP domains, etc. that all have to work within the 8-character limit, not to mention the same limit for variables.  For example, what is the --ORRES variable name in the dataset, LBUSORRE? or LBORRES, indicating it belongs to the LB domain?

(And before this leads to comments on how XPT needs to be replaced with XML, etc., however valid that argument may be, today's framework requires an 8-character limit.)

To show additional data that is conceptually aligned to labs, instead of LBUS, use something like LU, LX, etc.  The custom domain description and other metadata in the Define-XML, along with details in the Annotated CRF (acrf.pdf) and the Clinical Study Data Reviewer's Guide (csdrg.pdf) will make clear the context.

Regards,

Carlo