The TMF Reference Model as a Tailoring Exercise: Mapping DIA/CDISC Zones to ICH E6(R3) Essential Records
The DIA TMF Reference Model is a community-built taxonomy that organizes the essential documents of a clinical trial into a consistent Zone > Section > Artifact hierarchy (commonly cited as 11 zones, roughly 48 sections, and around 250 artifacts, with a sub-artifact layer added in v3.2). It originated under the Drug Information Association (DIA) and is now stewarded by CDISC. It is widely adopted, vendor-neutral, and useful. It is also voluntary.
Aileen
Aileen writes practical guidance for clinical trial teams at GCP Blog.
On this page · 12 sections
- 01 At a glance
- 02 What the TMF Reference Model is, and what it is not
- 03 The Zone > Section > Artifact grid, and how essentiality is actually decided
- · Zone-by-zone reference (illustrative)
- 04 How the model maps to ICH E6(R3) essential records, and whose duty it is
- · ICH E6(R3) essential record to Reference-Model-artifact mapping (illustrative)
- 05 Tailoring: keep, drop, add, with documented rationale
- 06 Versions and migration without orphaning documents
- 07 Proving completeness, contemporaneity, and inspection-readiness
- 08 Where teams get it wrong
- 09 Getting the assets and operationalizing them
- 10 Sources
At a glance
- The DIA TMF Reference Model (now stewarded by CDISC) is industry best practice, not a regulation. Adopting it wholesale does not make a trial compliant; the sponsor and investigator stay responsible for what their trial actually requires.
- The regulatory obligation is to maintain a trial master file (TMF) of essential records that is complete, contemporaneous, and inspection-ready. ICH E6(R3) and the EMA TMF guideline set that obligation; the Reference Model is one structure for meeting it.
- Use the model’s Zone > Section > Artifact grid as a tailoring scaffold: keep, drop, or add artifacts against your protocol, and document the rationale where you reduce content.
- ICH E6(R3) Appendix C lets you assess essentiality against stated criteria and use a structured content list to prospectively identify essential records. That assessment maps cleanly onto the Reference Model grid.
- Version migration (3.1 to 3.2 to 3.2.1 / 3.3.1 to v4) is a metadata exercise, not a refile: preserve filing history, version control, and traceability so no already-filed document is orphaned.
- Completeness and contemporaneity are provable, not aspirational, when an expected-document list plus risk-based QC sits on top of the grid.
What the TMF Reference Model is, and what it is not
The DIA TMF Reference Model is a community-built taxonomy that organizes the essential documents of a clinical trial into a consistent Zone > Section > Artifact hierarchy (commonly cited as 11 zones, roughly 48 sections, and around 250 artifacts, with a sub-artifact layer added in v3.2). It originated under the Drug Information Association (DIA) and is now stewarded by CDISC. It is widely adopted, vendor-neutral, and useful. It is also voluntary.
That distinction is the whole point. The Reference Model is not law and confers no compliance on its own. What the regulations require is a TMF, not a particular taxonomy. ICH E6(R3) §C.2.3 states that essential records should be maintained in, or referred to from, repositories held by the sponsor and by the investigator/institution, and that these repositories may be referred to as a trial master file. The EMA TMF guideline likewise defines a TMF as the collection of essential documents used by sponsors, CROs, and investigators to manage the trial and by monitors, auditors, and inspectors to verify that it was conducted in line with GCP and applicable regulatory requirements. Neither names the Reference Model. The model is a way to satisfy these obligations; it is not the obligation.
So treat the 250-artifact list as a menu, not a mandate. The regulatory work is deciding which artifacts your trial actually generates, filing them on time, and being able to prove it.
The Zone > Section > Artifact grid, and how essentiality is actually decided
The grid gives you a shared vocabulary: zones group broad domains (trial management, central trial documents, regulatory, IRB/IEC and other approvals, site management, IP and trial supplies, safety reporting, central testing facilities, third parties, data management, statistics), sections subdivide each zone, and artifacts are the individual document types. v3.2 introduced sub-artifacts, a finer layer that lets one artifact hold related document variants without inventing new artifact rows.
The regulatory backbone for what belongs in that grid is ICH E6(R3) Appendix C. It does not hand you a fixed checklist. ICH E6(R3) §C.3.1 sets out criteria for whether a record is essential and states that a structured content list for storage repositories may be used to prospectively identify essential records. That is precisely what the Reference Model’s artifact list is: a structured content list. ICH E6(R3) §C.1.1 makes the determinative point that the nature and extent of records generated and maintained depend on the trial design, its conduct, the application of risk-proportionate approaches, and the relevance of each record to the trial. In other words, the guideline expects the essential-record set to vary by trial.
The Essential Records Table in ICH E6(R3) Appendix C is also explicitly non-exhaustive: §C.3.2 states the listed records are not an exhaustive list and that other trial records may be considered essential by the sponsor or investigator. The EMA TMF guideline reaches the same conclusion from the other direction: it states the ICH GCP essential-documents list should not be used as a definitive checklist for TMF content, that it is not exhaustive, and that many trials require additional documents, so the sponsor and investigator/institution should include any documentation that facilitates reconstructing and evaluating trial conduct. Two in-scope sources, one message: the list is a floor and a prompt, not a ceiling.
Zone-by-zone reference (illustrative)
The Reference Model’s zone names and counts are industry conventions, not regulatory text, so treat this table as orientation rather than a citable requirement.
| Zone | Name (industry convention) | Purpose | Example artifacts |
|---|---|---|---|
| 1 | Trial Management | Oversight, planning, governance | Trial management plan, recruitment plan, meeting minutes |
| 2 | Central Trial Documents | Core trial-wide documents | Protocol and amendments, Investigator’s Brochure |
| 3 | Regulatory | Authority submissions and approvals | Regulatory authority authorizations, notifications |
| 4 | IRB/IEC and Other Approvals | Ethics oversight | IRB/IEC approval, composition records |
| 5 | Site Management | Site selection, oversight | Monitoring reports, site qualification |
| 6 | IP and Trial Supplies | Investigational product lifecycle | Accountability, shipping, storage, destruction records |
| 7 | Safety Reporting | Pharmacovigilance | SAE/SUSAR notifications, safety updates |
| 8 | Central Trial Documents (testing facilities) | Labs and central facilities | Normal ranges, accreditation, validation |
| 9 | Third Parties | Vendors and service providers | Agreements, oversight and selection records |
| 10 | Data Management | Data handling | Data management plan, validation, query resolution |
| 11 | Statistics | Analysis | SAP, randomization list, analysis datasets |
The example artifacts above trace to record types that ICH E6(R3) Appendix C and the EMA guideline treat as essential; the zone numbering and naming come from the Reference Model and may differ across versions.
How the model maps to ICH E6(R3) essential records, and whose duty it is
The mapping discipline is simple to state: for each Reference Model artifact, ask which ICH E6(R3) essentiality criterion or Essential Records Table entry it satisfies, and which party owns it. Artifacts that map to nothing are candidates to drop; obligations that map to no artifact are gaps to add.
The duty is shared and non-delegable in substance. ICH E6(R3) §C.2.6 requires the sponsor and investigator/institution to retain essential records so that they remain complete, readable, and readily available, directly accessible upon request by regulatory authorities, monitors, and auditors, with any alteration traceable. ICH E6(R3) §C.2.4 requires both parties to maintain a record of where essential records are located, including source records, and to use a storage system that provides for identification, version history, search, and retrieval. Filing is not optional housekeeping: ICH E6(R3) §C.2.5 requires the sponsor and investigator/institution to ensure essential records are collected and filed in a timely manner.
The EMA TMF guideline raises the bar from “have the documents” to “tell the story.” It states the TMF documentation should be sufficient to adequately reconstruct the activities undertaken in conducting the trial, along with the decisions and justifications made, and that documents in the TMF should collectively permit confirmation of compliance with the protocol and GCP and the integrity of data collected without the need for additional explanation from sponsor, CRO, or site staff. The reconstruction test is the one inspectors apply. A grid that files documents but cannot reconstruct the trial fails it.
ICH E6(R3) essential record to Reference-Model-artifact mapping (illustrative)
| ICH E6(R3) essential record (Appendix C) | Maps to Reference Model zone/area | Owning party |
|---|---|---|
| Signed protocol and amendments | Central trial documents | Sponsor (filed at site too) |
| IRB/IEC approval and composition | IRB/IEC and other approvals | Both |
| Regulatory authority authorizations | Regulatory | Sponsor |
| Signed informed consent forms; participant ID code list | Site management / investigator-controlled | Investigator/institution |
| SAE/SUSAR notifications and safety information | Safety reporting | Both |
| IP accountability, shipping, storage, destruction | IP and trial supplies | Both, as applicable |
| Site monitoring reports (selection, initiation, routine, close-out) | Site management | Sponsor |
| Data management and statistics documentation; randomization list | Data management / statistics | Sponsor |
| Computerised system validation / fitness-for-purpose evidence | Trial management / data management | Sponsor |
Note one division the EMA guideline is explicit about: certain investigator-held records containing direct participant identifiers, such as the subject identification code list, subject-related source documents, and signed consent forms, should remain under the sole control of the investigator/institution. Your mapping must respect that boundary rather than push everything into a sponsor eTMF.
Tailoring: keep, drop, add, with documented rationale
Tailoring is where the model earns its keep, and where the regulations give you both permission and a guardrail.
Permission to reduce: the EMA TMF guideline states that, because the TMF content takes into account all characteristics of the trial including whether it is low-intervention, some documentation specified in the ICH GCP guideline may not be necessary under a risk-proportionate approach. But here is the guardrail, in the same source: the justification for reducing documentation should be documented in the TMF. That single sentence converts tailoring from a quiet omission into a recorded decision. Drop an artifact silently and an inspector sees a gap; drop it with a logged rationale and they see a judgment.
On the essentiality side, ICH E6(R3) §C.3.1 notes that the assessment of whether a record is essential, while important, is not required to be documented, and it is here that the two in-scope sources pull in different directions. ICH E6(R3) treats the essentiality assessment itself as something you need not write down, whereas the EMA TMF guideline requires that any reduction of the ICH-listed documentation be justified in the TMF. The practical reading is to hold both: you may keep the general essentiality reasoning informal, but where you reduce below what ICH GCP lists, record why. Do not reconcile this into “documentation is optional”; under EMA, for in-scope EU trials, the reduction rationale is not.
A workable tailoring checklist, one row per Reference Model artifact:
| Field | What to capture |
|---|---|
| Artifact / sub-artifact | Reference Model identifier and name |
| Mapped obligation | ICH E6(R3) Appendix C criterion or Essential Records Table entry, or “none” |
| Decision | Keep / Drop / Add |
| Owning party | Sponsor / Investigator/institution / Both |
| Rationale | Why kept, dropped, or added; required when reducing ICH-listed content |
| Expected? | Whether this artifact is on the expected-document list for this trial |
Adds are as important as drops. Both in-scope sources anticipate documents the ICH list never names: ICH E6(R3) §C.3.2 says other records may be essential, and the EMA guideline gives concrete examples (quality-system forms and checklists, IMP qualified-person certification, assay validation reports, eCRF/IRT validation, data management and statistics documentation, the investigator delegation log). If your trial generates them, file them.
Versions and migration without orphaning documents
Versions of the Reference Model (3.1, 3.2 with sub-artifacts, 3.2.1, 3.3.1, and v4) refine zone/section/artifact definitions, but the model is a structure, not a regulation, so a version change carries no regulatory force by itself. What does carry force is the obligation that filed documents stay traceable and retrievable across the change.
The constraint comes from the record-management duties. ICH E6(R3) §C.2.1 requires that records be identifiable and version controlled where appropriate, and ICH E6(R3) §C.2.4 requires that the storage system provide identification, version history, search, and retrieval. The EMA TMF guideline adds that any migration of data and documents to new media or a new format should be verified to ensure long-term readability and to maintain integrity, and that any transfer or migration needs to be validated to ensure migrated data carry the same information as the original, including metadata.
Translated to a Reference Model upgrade: do not refile. Re-map. Maintain a crosswalk from old artifact identifiers to new ones, preserve original filing dates and version history as metadata rather than overwriting them, validate the remapping, and confirm that every previously filed document remains searchable and retrievable under its new coordinates. An in-flight TMF mid-migration must remain complete throughout, because the completeness duty does not pause for an upgrade.
| Version | What changed (industry convention) | Migration impact |
|---|---|---|
| 3.1 | Baseline zone/section/artifact grid | Establishes original filing coordinates |
| 3.2 | Sub-artifact layer added | Re-map artifacts to sub-artifacts; preserve original file metadata |
| 3.2.1 | Maintenance refinements | Low; crosswalk and verify retrieval |
| 3.3.1 | Further refinements and clarifications | Low to moderate; validate any artifact reassignments |
| v4 | Latest revision under CDISC stewardship | Re-map only; do not re-date or orphan filed documents |
The “what changed” column reflects industry convention, not regulatory text; the migration-impact column reflects the validation and traceability duties cited above, which apply to any version change.
Proving completeness, contemporaneity, and inspection-readiness
A taxonomy files documents. Proof requires a layer above it: an expected-document list, contemporaneous filing, and risk-based QC.
Contemporaneity is a named obligation, not a courtesy. The EMA TMF guideline’s contemporariness section requires that the TMF have all documentation added in a timely manner during the trial and that timelines for submission and filing be defined in procedural documents or TMF plans. ICH E6(R3) §C.2.5 mirrors this, requiring timely collection and filing and noting that some essential records should generally be in place before the trial starts. A document filed six months late is a finding even if it is eventually present.
Completeness is provable when QC runs against an expectation. The EMA TMF guideline requires the sponsor and/or investigator to implement risk-based quality checks to ensure the TMF is kept up to date and that all essential documents are appropriately filed, checking that generated essential documents are present, filed in the right locations, added in a timely manner, correctly indexed, and access-controlled, and adds that the sponsor should ensure the TMF is readily available and directly accessible to the competent authority for inspection. The Reference Model grid is what makes “all essential documents present” a checkable statement: each expected artifact is either filed, justifiably dropped, or an open gap.
Inspection-readiness is the sum. ICH E6(R3) §C.1.3 states that essential records are used during inspections by regulatory authorities to assess trial conduct and the reliability of results, and ICH E6(R3) §C.2.6 requires that records remain directly accessible upon request. Map your grid to the essential-record obligations, file contemporaneously, QC against an expected-document list, and you can demonstrate completeness rather than assert it.
Where teams get it wrong
- Treating the model as compliance. Adopting all 250 artifacts does not make a trial compliant and a sparse TMF is not automatically wrong. The test is whether your essential records meet ICH E6(R3) and EMA obligations for your trial, not whether you matched a taxonomy.
- Dropping artifacts silently. Reducing content is allowed under a risk-proportionate approach, but the EMA guideline requires the reduction rationale be documented in the TMF. Silent drops read as gaps.
- Filing late and calling it done. Presence is not enough. Both in-scope sources require timely filing; contemporaneity is a separate, named obligation.
- Refiling on version upgrade. Re-dating documents to a new artifact structure can break version history and traceability. Re-map with a validated crosswalk; preserve original metadata.
- Pushing investigator-controlled records into the sponsor eTMF. Subject identifiers, the ID code list, and signed consents should stay under sole investigator/institution control per the EMA guideline.
- Filing without reconstruction. A complete file list that cannot reconstruct the trial fails the EMA reconstruction-and-no-explanation standard. Map artifacts to the decisions and activities they evidence, not just to document types.
Getting the assets and operationalizing them
The Reference Model Excel and its user guide (available from CDISC) give you the artifact grid; they are starting inputs, not an operating model. To operationalize: load the grid, run the tailoring checklist to set each artifact to keep/drop/add with rationale, derive an expected-document list from the “keep” and “add” rows, assign owning parties respecting the sponsor/investigator split, define filing timelines in your TMF plan, and stand up risk-based QC against the expected list. That turns a static taxonomy into a provable, inspection-ready operating model grounded in the essential-records obligations the regulations actually impose.
For deeper treatment of adjacent obligations, see our sibling explainers on ICH E6(R3) sponsor oversight, on monitoring and RBQM, and on TMF inspection-readiness.
Sources
- ICH E6(R3) Good Clinical Practice — ICH, version r3 (2025). https://www.ich.org/page/efficacy-guidelines
- EMA Guideline on the content, management and archiving of the clinical trial master file (paper and/or electronic) — EMA/INS/GCP/856758/2018, version 2018.
- DIA/CDISC TMF Reference Model, industry best-practice standard (not a regulation; cited here for structure only, not for regulatory requirements). https://www.cdisc.org
Written by
Aileen
Aileen writes practical guidance for clinical trial teams at GCP Blog.
Continue reading
Clinical Trial Close-Out Checklist: A Dependency-Gated Sequence Built on GCP and 21 CFR Part 312
Most published close-out checklists hand you the CASPER-style alphabetical pile of documents and stop at the visit. That is where audits start, not where they end. This piece treats close-out as what it actually is in practice: a chain of reconciliations where each link gates the next, and where the...
ReadThe Monitoring Visit Report as Oversight Evidence: A GCP-Audit-Ready MVR, Not a Filled-In Form
Most teams treat the MVR as a form: date, attendees, what was reviewed, a box for issues, file it, move on. That framing is the root cause of nearly every MVR finding raised at audit. The report is not the administrative residue of a visit. It is the sponsor's documented evidence that oversight of t...
ReadData Safety Monitoring Plan vs. Medical Monitoring Plan vs. DSMB Charter: Which One Your Trial Needs, Who Signs It, and How They Fit Together
A coordinator handed a new protocol and told to "produce the safety monitoring plan" usually starts from a downloaded NIH or academic PDF that conflates three different documents. The ranking results compound the confusion: they hand over one generic form labeled "DSMP / MMP" and call it done. That ...
Read