What Statistical Programmers Need to Know About the Biggest SDTM Update in Years
CDISC is preparing to release the most significant update to its foundational tabulation standards since the model was first published over two decades ago. SDTM v3.0 and SDTMIG v4.0 have completed internal review and are expected to enter public review in early 2026, with final publication planned alongside SENDIG v4.0 in 2026. The SDTMIG for Medical Devices (SDTMIG-MD v2.0) will follow shortly after.
For the statistical programming community, these updates bring changes that will reshape daily workflows: supplemental qualifier datasets are being replaced, new domains are being introduced, several familiar variables are being deprecated, and the model itself is being enriched with metadata structures designed to support machine-readable standards and automation.
This article summarizes the key changes across both documents and highlights what programmers should start preparing for now.
SDTM v3.0 and SDTMIG v4.0 were developed in tandem to ensure alignment. Unlike previous minor releases (SDTM v2.1 was a narrow update for the Tobacco Implementation Guide), v3.0 is a full major version. Key milestones include:
| Milestone | Status / Timing |
| Internal review | Completed (SDTM 3.0, SDTMIG 4.0, and SENDIG 4.0 reviewed together) |
| Public review | Expected late 2025 |
| Final publication | Planned 2026 (SDTM 3.0, SDTMIG 4.0, SENDIG 4.0 released together) |
| SDTMIG-MD v2.0 | To follow shortly after the main release |
| Note: The public review period is your opportunity to provide feedback. CDISC posts all comments and resolutions publicly to show how feedback shaped the final standard. |
The single most impactful change for programmers is the replacement of Supplemental Qualifier datasets (SUPP--) with Nonstandard Variables datasets (NS--). This is not a cosmetic rename. It is a fundamental structural change:
| SUPP-- (Previous) | NS-- (SDTM v3.0+) |
| Vertical: one row per nonstandard variable value | Horizontal: one row per parent record with any NS variable populated |
| QVAL is always character | NS columns can be character or numeric |
| IDVARVAL is character (requires char-to-num conversion) | IDVARVLN is numeric; merge without conversion |
| Requires transposing before merge | Direct merge with parent dataset |
| Value-level metadata in Define-XML for QVAL | Standard variable-level metadata in Define-XML |
Practical impact: Every program that currently builds, reads, or merges SUPP-- datasets will require updates. The good news is that the NS-- structure is simpler, and the merge logic that programmers have complained about for years becomes straightforward.
In the Role column of variable metadata tables, a single "Qualifier" designation now replaces the separate Grouping, Result, Synonym, Variable, and Record Qualifier roles. The former "Variables Qualified" column has been removed entirely. These sub-types were always informative and are not part of submission metadata, so this change reduces complexity without affecting data content.
SDTM v3.0 introduces two new structural concepts to the model:
All root variable definitions have been completed (v2.0 definitions were incomplete). Definitions are now managed as controlled terminology, and every root variable carries a C-code. This means LBDTC, AEDTC, and EGDTC are all formally recognized as instances of the root variable --DTC, with a shared definition and code.
| Code | Domain Name | Purpose |
| DC | Demographics for Multiple Participations | Captures demographic data when a subject participates in a trial more than once. Required when at least one subject has multiple participations; prohibited otherwise. |
| EA | Event Adjudication | Standardizes the submission of adjudicated event data, such as cardiovascular endpoint committees or independent review boards. |
| GI | Gastrointestinal System Findings | A new Findings domain for gastrointestinal assessments and endoscopic results. |
The DC domain is particularly noteworthy. It includes FOCID (supporting study designs with distinct treatments to multiple body parts), DCSEQ (aligning with other special-purpose domains that allow multiple records per subject), and collected race/ethnicity variables (CRACE, CETHNIC). The DM domain itself remains largely unchanged but also gains CRACE and CETHNIC.
The DV domain gains a new classification variable (--CLASSI) to distinguish categories such as important vs. non-important protocol deviations. This update was driven by the need to better support Bioresearch Monitoring (BIMO) activities.
| Scope | Change | Detail |
| All Datasets | SUBJID added to identifiers | Supports multiple subject participations across all classes |
| DM / DC | AGELO and AGEHI replace AGETXT | Separate low/high age variables replace the free-text age range |
| DM / DC | CRACE, CETHNIC added | Collected race/ethnicity values as originally reported by the subject |
| Events | --CLASSI added | Classification of Protocol Deviation (restricted to DV domain) |
| Interventions | --TRTCD added | Standardized Intervention Code associated with --DECOD |
| Findings | --CBRFL added | Conditionally Branched Item Flag (restricted to QRS domains) |
| Findings | --RESCNT added | Result Count (restricted to EG domain) |
| Findings | --BLFL, --MODIFY, --BODSYS removed | Deprecated from the Findings observation class |
| Findings | --PTFL, --PDUR reclassified | Moved from Timing to Findings class variables |
| Trial Design | TIRL removed from TI | Trial Inclusion/Exclusion Rule removed from the TI domain |
| The removal of --BLFL from Findings is significant. Baseline flags in SDTM have long been a source of confusion. Programmers should watch the public review materials closely for guidance on how baseline identification will be handled going forward. |
Beyond data-level changes, the SDTMIG v4.0 document itself has been significantly reorganized. Introductory sections on SDTM fundamentals, conformance, and general assumptions have been updated and restructured. Metadata tables now include variable definitions and standardized documentation of variable relationships, aligning with the model-level enhancements in SDTM v3.0.
The introduction of variable groups and variable relationships in SDTM v3.0 is a clear signal that CDISC is building toward machine-readable, automation-friendly standards. Combined with the USDM and 360i initiatives, the trajectory points toward a future where trial design metadata flows directly into SDTM specifications with minimal manual intervention. Programmers who understand these structural changes at the model level will be better positioned as the industry moves toward that vision.
When the public review opens, CDISC will post the draft documents on the CDISC Wiki. Reviewers submit comments via JIRA (the Wiki and JIRA share the same login credentials, which are separate from your cdisc.org account). All comments and resolutions are published afterward.
Given the scale of these changes, the statistical programming community should engage actively. This is particularly important for the SUPP-to-NS transition, variable deprecations, as these decisions will directly affect submission timelines and program libraries.
| Resource | URL |
| SDTM Standards Page | cdisc.org/standards/foundational/sdtm |
| SDTMIG Standards Page | cdisc.org/standards/foundational/sdtmig |
| CDISC Standards Timeline | cdisc.org/standards/timeline |
| Standards in Development | cdisc.org/standards/in-development |
| CDISC Wiki (for Public Review) | wiki.cdisc.org |
| 2025 US Interchange Program | cdisc.org/events/interchange/2025-cdisc-tmf-us-interchange/program |
Bottom line: SDTM v3.0 and SDTMIG v4.0 represent a generational update to the standards that define our daily work. The transition from SUPP to NS alone justifies early preparation. Start reviewing your program libraries, engage in the public review, and plan your migration strategy now, well before the FDA Data Standards Catalog is updated to reference these new versions.
No comments yet. Be the first!