Every SDTM programmer knows the SUPP-- dataset: the place where non-standard variables go, linked back to the parent domain by IDVAR and IDVARVAL. It works, but it is awkward to build, awkward to review and easy to get wrong. The next major version of the standard changes that.
What is changing
CDISC is finalising SDTM v3.0 and SDTMIG v4.0, projected for 2027 after public review. The headline change for implementers is that SUPP-- datasets are replaced by NS-- datasets for non-standard variables. A further review of the expanded scope of SDTMIG v4.0 is open until 12 December 2026.
What is not changing yet
FDA's Data Standards Catalog lists SDTMIG 3.4 as required for studies starting from March 2025, and SDTMIG 3.3 remains supported. Nothing in v4.0 is required until FDA adds it to the catalog and announces dates, which typically comes with a long lead time. Studies running today will finish on 3.3 or 3.4.
Conformance rules are getting an open engine
In November 2026 CDISC released version 1.0 of CORE, its open conformance rules engine, covering SDTMIG 3.2 to 3.4 and FDA's SDTM business rules. Rules for ADaM and SEND are still in development. Running CORE alongside your usual validation tool is a cheap way to catch issues earlier.
How to get ready
- Keep non-standard variables clearly identified in your mapping specifications, with a stated origin and purpose. That makes any later move to NS-- datasets mechanical.
- Avoid custom workarounds that depend on the SUPP-- structure itself.
- Track the CDISC "in development" page and the FDA catalog, not vendor blogs, for dates.
We will update this article when SDTMIG v4.0 is published. If you are planning a standards update across a programme, we can help you work out what to change and when.