Dataset-JSON, USDM and IDMP arrive in the same quarter

Three modernization tracks that biometrics teams have been told to ignore as “still pilot” landed within weeks of each other. CDISC published Dataset-JSON v1.1 with a paired API v1.0 Standard and a Compressed variant, alongside Define-XML v2.1.11, Glossary v20.0 and ADaM Oncology Examples v1.0. Per FDA’s CBER-CDER Data Standards Program Action Plan, Dataset-JSON testing is in progress, with implementation stages still pending — but the explicit framing is “potential replacement for XPT v5” (see FDA’s Study Data Standards Resources). After thirty years, the SAS v5 transport format finally has a designated successor with regulator engagement behind it.

That alone would be the story. It isn’t.

The protocol becomes a dataset

ICH M11 CeSHarP was adopted at Step 4 in November 2025 and finalized by FDA in the Federal Register on May 22, 2026 (Docket FDA-2022-D-3054), with a guidance, template and technical specification for machine-readable protocols. That moves the CDISC-co-developed Unified Study Definitions Model from industry experiment to regulatory anchor. TransCelerate’s Digital Data Flow — which targets the documented four-month protocol-to-startup lag and the manual SDTM conversion bottleneck — now has a finalized ICH guideline pointing at the same artifact. OpenStudyBuilder, the Novo Nordisk–owned COSA project built on Neo4j with USDM and DDF API support, stops being a curiosity and starts being a reference implementation. Whether anyone outside Novo Nordisk has put it through an end-to-end submission is still the right question to ask vendors.

Product identifiers, observational data, and the rest of the stack

EMA has published Chapter 3 of its PMS guidance on implementing the ISO IDMP suite (11615, 11616, 11238, 11239, 11240) for European MAHs. That feeds OMS/SMS reference data and, downstream, the controlled terminology your MP and EX domains are supposed to be using. The guidance is operational — how product data must be structured, submitted and maintained — not aspirational. The binding compliance deadline isn’t stated in the chapter itself, which is the part to chase before your next EDC build for an EU study.

On the real-world data side, JSCDM has published a practical implementation paper on CDISC’s 2024 “Considerations for SDTM in Observational Studies,” cataloguing exactly where the framework breaks: heterogeneous collection, no protocol-defined visits, Trial Summary domain representation, non-standard variable proliferation when registry or claims data refuse to fit the interventional-trial domain shapes they were designed around. There is no FDA endorsement of the CDISC OS framework yet — it remains CDISC-only guidance against rising agency expectations for CDISC-compliant RWD submissions. Worth pre-specifying in any RWE protocol that mapping decisions will be documented against this framework, on the assumption reviewers will start asking.

Surrounding all of this: CDISC and SCDM have formalized a Vision 2030 partnership covering co-developed education, a Japan Conference, and joint thought leadership on terminology and digital data. No binding deliverable, no certification. Treat it as a content alliance until something concrete ships. Three CDISC standards remain open for public review — ADaMIG Anti-Drug Antibody Data v1, DHT1 Draft v1.0, and SDS & ADaM QRS Supplements — which is where teams with actual implementation experience can change wording before it hardens.

What it means for the pipeline

The unwelcome operational reality is that XPT, the document protocol and free-text product strings are all being replaced on overlapping but unsynchronized timelines, by different bodies, with different conformance regimes. Pinnacle 21 will need to validate Dataset-JSON v1.1 and Define-XML v2.1.11. EDC libraries will need IDMP-aligned product coding. Study setup tooling will need a USDM-aware path or a manual bridge. The sponsors who quietly moved off pure-SAS pipelines — see J&J’s hybrid R/SAS submission case study and the R Submissions Working Group’s 2026 plans — are better positioned than those still treating XPT as eternal.

Protocol read: The “pilot forever” assumption is no longer safe; Dataset-JSON, USDM-based protocols and IDMP product data are now three converging streams a 2027 submission may need to handle simultaneously, and the validation toolchain has not caught up.

What to do now:

• Inventory which submission-bound studies still assume XPT-only transport, and ask your validation vendor for a written Dataset-JSON v1.1 and Define-XML v2.1.11 support date.
• For EU-bound MAHs, map current MP/EX coding against the IDMP OMS/SMS reference data flagged in EMA PMS Chapter 3 before the next EDC build.
• File public-review comments on ADaMIG ADA, DHT1 and the QRS Supplements while the windows are open; post-publication objections cost more.
• Treat USDM/OpenStudyBuilder as a vendor diligence question now that M11 is final — ask for named non-Novo production deployments, not roadmaps.