When an implementer is working how to represent data for a study, they first consult the SDTMIG and SDTM to find the appropriate standard variable. If no appropriate standard variable is found, the SDTMIG directs them to create a supplemental qualifier, also called a non-standard variable.
Examples published in the SDTMIG or in Therapeutic Area User Guides (TAUGs) illustrate data represented with non-standard variables. If a implementer working in a particular therapeutic area consults a relevant TAUG, they might find a published example that provides a non-standard variable, which fits their data. CDISC has published over 40 TAUGs and it would be an onerous task to search through each TAUG to find an appropriate non-standard variable. The Non-Standard Variable Registry (NSV Registry) comprises a curated list of non-standard variables from published examples to give an implementer one place to find what they're looking for. Leveraging the NSV Registry allows implementers to create SDTM-based datasets with NSVs that are more consistent across studies and submissions than if each implementer created their own non-standard variables.
If the implementer can't find a variable in the NSV Registry, they will have to make up their own. The rules used to create variables in the NSV Registry and the Fragments Database can help to create a well-formed non-standard variable.
Why are these NSVs published in a separate registry? Why not just make these standard variables? Most NSVs are used in fairly narrow contexts, and making all of them standard variables would make it difficult to find the most-used variables among the seldom-used variables. The variables in the SDTM are the ones determined to be most commonly used, while the non-standard variables are considered to be less-used.