Structured Content Authoring for Regulatory Submissions: A Readiness Checklist

What Structured Content Authoring Means in a Regulatory Setting

Most regulatory documents are still written the way they were twenty years ago. A medical writer opens last year’s protocol or clinical study report, saves a copy, and edits it. Numbers are typed in from source tables. The same study design description is retyped, with small differences, into the protocol synopsis, the protocol body, the statistical analysis plan, the investigator brochure summary, the registry entry, and later the clinical study report. Each copy is reviewed and approved on its own. When something changes, someone has to find every copy.

Structured content authoring (often shortened to SCA) changes the unit of work. Instead of the document, the unit is a component: a defined block of content with a known purpose, a known owner, and metadata that says what it is. A document becomes an assembly of components, some written once and reused, some written for that document only. The document still exists at the end, because regulators still review documents, but the source of truth moves from the file to the component.

Documents Versus Components

A peer-reviewed paper published in AAPS Open in 2023 describes this as a component-based approach to information management. It defines a content block as a container for a stack of related data elements that commonly appear together, and it explains that reusing an approved block across applications limits the need to author, review, and approve the same content again and again⁹. The same paper notes that blocks can be distributed across multiple eCTD sections to meet different filing requirements⁹.

That last point matters for regulatory teams. The electronic Common Technical Document (eCTD) is organized by module and section, not by fact. A manufacturing site address, a shelf-life claim, or a primary endpoint definition can appear in several sections and several submissions. In a document-first model, each appearance is a separate copy. In a component model, each appearance is a reference to one approved source.

Structured Content Versus Structured Data

Two related terms get used loosely, and it helps to separate them early.

• Structured content is text, tables, and figures broken into labeled, reusable pieces. A paragraph describing the rationale for a dose is structured content if it lives as a governed component with metadata, even though it is still prose.
• Structured data is information held as discrete, typed values with defined meaning: a trial phase code, a number of participants, a storage temperature. Structured data can be validated by a machine and exchanged between systems without a human reading it.

A 2022 article in the Journal of Pharmaceutical Sciences uses the term Structured Content and Data Management (SCDM) to cover both. It describes the full life cycle: data captured in source systems, gathered into a consolidated repository, and turned into semantically structured blocks with metadata. The authors state that automating regulatory authoring, which they call Structured Content Authoring, is feasible because SCDM makes data both human and machine readable¹⁰.

Most real programs need both. The narrative parts of a protocol or a quality overall summary will stay as prose for the foreseeable future. The facts inside that prose (doses, visit schedules, specifications, identifiers) are where structured data pays off, because those are the values that get copied, mistyped, and changed.

What Structured Content Authoring Is Not

Three misunderstandings come up in almost every early conversation.

• It is not a template. A Word template with styles and headings gives a consistent look. It does not give reuse, because the content inside each file is still a separate copy.
• It is not generative AI. AI drafting tools can work on top of structured content, and they tend to work better when the inputs are clean and labeled. But a model that writes a first draft from source documents is a different capability. Structured content is about where approved content lives and how it is reused.
• It is not a single product. Vendors sell component content management systems, and some use XML standards such as DITA (Darwin Information Typing Architecture), an OASIS standard originally built for technical documentation¹⁷. The platform matters, but the content model, reuse rules, and ownership decisions are what make it work. Those are organizational choices, not features.

Why the Question Is Live in 2026

Life sciences companies have discussed structured authoring for well over a decade. What is different now is that several regulators and standards bodies have published content models or structured formats for specific document types. None of them requires a sponsor to buy a structured authoring platform. Together, they make the case for building the underlying discipline.

ICH M11 Is Final, but Not Mandatory

ICH M11, the Clinical Electronic Structured Harmonised Protocol (CeSHarP), was adopted by the ICH regulatory members on 19 November 2025¹. It has three parts: a guideline that explains the design principles, a protocol template with a standard table of contents, headings, and instructions, and a technical specification that defines the data elements and technical attributes (definition, conformance, cardinality) needed for electronic exchange¹.

FDA announced the final guidance in the Federal Register on 22 May 2026. The notice states that it finalizes the December 2022 draft guidance and the June 2025 draft technical specification⁴. It also carries FDA’s standard language: the guidance is not binding on FDA or the public, and a sponsor can use an alternative approach if it satisfies the applicable statutes and regulations⁴.

In the EU, EMA’s Committee for Medicinal Products for Human Use (CHMP) gave final adoption to the Step 5 guideline on 11 December 2025, with a date for coming into effect of 11 June 2026⁵. EMA describes the template as offering standardized content with both required and optional components⁶. Nothing in the adoption documents we reviewed requires sponsors to submit protocols in the M11 electronic format.

Sakara Digital covered what a machine-readable protocol changes for downstream clinical systems in September, in ICH M11 Is Final, Not Mandatory. This article looks at the other side: what an authoring organization needs in place to produce protocols, and eventually other documents, as structured content.

Structured Formats Are Arriving in Other Document Families

M11 is not an isolated case. Several parallel efforts point the same way.

• Quality (CMC) dossiers. The draft revision of the ICH quality section of the CTD, M4Q(R2), was released for consultation in 2025, with EMA’s comment period closing on 24 October 2025¹². The draft says its structure is set out with enough granularity to support digitalization and knowledge management, and that this granularity also supports newer approaches such as structured data management processes¹². It is still a draft and could change before adoption.
• FDA’s PQ/CMC project. FDA is developing structured data standards for pharmaceutical quality and CMC information currently submitted in eCTD Module 3, plus an exchange standard for sending that data to the agency¹³. The exchange standard is being built as an HL7 FHIR implementation guide. FDA is explicit that testing, balloting, and publishing versions of that guide do not constitute FDA policy, and that the agency has formal processes for announcing when it will accept data in a given format¹⁴.
• EU electronic product information (ePI). The EU ePI Common Standard, based on FHIR, has been adopted by the European medicines regulatory network. EMA describes ePI as a semi-structured format. A one-year pilot ran from July 2023 to August 2024, EMA published a draft roadmap in March 2026 tied to upcoming requirements under the new pharmaceutical legislation, and on 4 September 2026 it published guidance on submitting ePI in centralized procedures, ahead of ePI go-live¹⁵.
• US labeling. Structured authoring is already routine in one corner of US submissions. Structured Product Labeling (SPL) is an HL7 document markup standard that FDA adopted for exchanging product and facility information, and FDA maintains guidance on providing content of labeling in electronic format¹⁶. Labeling teams have worked with structured XML for years, and their experience is worth drawing on.

Industry Groundwork Has Moved Into Standards

Two industry efforts shaped much of the current thinking on clinical document reuse. TransCelerate’s Clinical Content and Reuse initiative, formerly the Common Protocol Template, developed shared content for reuse across the protocol, statistical analysis plan, and clinical study report in its Clinical Template Suite. TransCelerate states that its team reconvened in 2026 to align the Common Protocol Template with the M11 template and technical specification, and that as of March 2026 the initiative is complete, with no new solutions or updates planned⁸.

CDISC, working with TransCelerate’s Digital Data Flow initiative, maintains the Unified Study Definitions Model (USDM). CDISC’s roadmap shows that a phase of this work represented the draft M11 template in USDM and added model elements for populating SDTM trial design domains and clinical trial registry information from study design data⁷. For authoring teams, the message is that the content model for the protocol is now owned by ICH and maintained through formal standards bodies, not by a single sponsor or consortium.

The Readiness Checklist at a Glance

We use five readiness areas when we assess an organization’s position on structured content authoring. They are listed in the order we recommend addressing them. Tooling is last on purpose: a platform will enforce whatever model and rules you give it, including poor ones.

The next sections work through each area in more detail, with specific checks you can use in an internal assessment.

Readiness Areas One and Two: Content Model and Templates

Area One: The Content Model

A content model is a written description of the components that make up a document family, what each one means, what type of content it holds, and how components relate to one another. It is the single most important readiness item and the one most often skipped.

The M11 technical specification is a useful example of what a mature content model looks like, even for teams that are not starting with protocols. For each item it records the term from the template, a data type, whether the item is a heading, data, or a value, a definition linked to a CDISC-maintained concept code, user guidance, a conformance rule, cardinality, relationships to other items, and rules for repeating and reuse². Conformance is expressed as required, optional, or conditional, and conditional items state their condition (for example, an amendment identifier applies only when there is an amendment)².

ICH and CDISC have signed an agreement for the maintenance and governance of ICH controlled terminology, and the M11 concepts are published as an NCI Thesaurus subset². ICH’s May 2026 explainer adds that each M11 data element has an object identifier (OID) linked to its code list, so electronic systems can identify it without ambiguity³. For a sponsor, this means the definitions for protocol content no longer have to be invented internally.

Content Model Checks

• Have you picked one document family to model first, rather than trying to model the whole dossier?
• Is there a written inventory of components for that family, with a plain-language definition for each?
• Does each component have a type (narrative text, table, structured value, figure, boilerplate) and a small, fixed set of metadata (product, study, indication, region, owner, status)?
• Where an external standard defines the same concept (M11 for protocols, CDISC terminology for study design, IDMP for product data), have you mapped to it instead of creating your own term?
• Have you agreed on the right level of granularity? Components that are too large give little reuse. Components that are too small create an assembly and review burden that authors will resist.

Area Two: Templates

In a structured environment, a template is not a formatted Word file. It is a recipe that says which components a document needs, in what order, which are required, which are optional, and which text is fixed.

The M11 guideline sets out design principles that apply well beyond protocols. It calls for common core content, content defined for electronic exchange, design for content reuse, and flexibility through both universal and optional text, with higher-level headings kept fixed while many lower-level sections can be added, removed, or changed¹. It also puts the information most needed to run the trial (synopsis, schema, schedule of activities) near the front, and states: “Unnecessary repetition is eliminated wherever possible.”¹

That last principle is worth taking seriously. Many sponsor templates repeat the same content in the synopsis, the body, and an appendix because that is how the previous template did it. A structured template should show each fact once and reference it elsewhere.

For clinical study reports, the CORE Reference user manual, developed by regulatory medical writing and statistics professionals and published in 2016, offers practical guidance on CSR content built on ICH E3, with the aim of producing reports that need minimal redaction before public disclosure¹⁸. It is a reasonable starting point when designing a structured CSR template, because disclosure is one of the main downstream uses of CSR content.

Template Checks

• Does each template reference components from the content model rather than containing free text?
• Is fixed text (text that must not change) clearly separated from text an author can edit?
• Are required, optional, and conditional sections marked, with the condition stated?
• Has anyone mapped your current protocol template section by section to the M11 template, and recorded the gaps?
• Is there a single owner for each template, and a change process for updating it?

Readiness Areas Three and Four: Reuse and Governance

Area Three: A Reuse Strategy You Can Write Down

Reuse is the main reason to do structured content authoring, and most of the design decisions belong to the business, not IT. The M11 technical specification offers a precise starting vocabulary. It separates repeating from reuse: repeating is defined as replication of a data element for new content, and “Reuse is defined as using verbatim content in more than one data element location in the protocol.”² Each item in the specification states whether it is repeated or reused and where².

Across a portfolio, the useful distinctions are wider than that. We suggest classifying every reusable component into one of three types.

• Locked reuse. The component appears verbatim everywhere it is used, and can only be changed at the source. Examples: a product’s approved description, standard safety reporting language, a company’s standard statement on data protection.
• Adapted reuse. The component starts from an approved source but can be edited for a specific use, and the link to the source is kept so reviewers can see what changed. Examples: an inclusion criterion adapted for a new population, a study design description shortened for a synopsis.
• Derived content. The component is generated from structured data rather than written. Examples: a schedule of activities table built from visit and procedure data, a list of manufacturing sites pulled from a product master record.

The practical payoff comes from locked reuse and derived content. The 2020 Journal of Pharmaceutical Sciences paper on CMC content describes the current state as a repetitive cycle of authoring, editing, and data verification that grows more complex as approved products change, with companies relying on manual change management across documents¹¹. Locked and derived components are what remove that cycle. Adapted reuse helps authors start faster, but each adapted copy still needs its own review.

How large is the benefit? Be careful with any single figure. The 2023 AAPS Open paper works through a hypothetical product scenario and estimates that authoring and submission timelines across the marketing application process could be reduced by as much as 50 percent compared with the traditional process⁹. That is a modeled estimate for an illustrative case, not a measured result, and the same paper notes that structured content was not yet a widespread industry standard or regulatory expectation⁹. Use it as a reason to measure your own baseline, not as a business case figure.

Reuse Checks

• Have you identified, by counting in real documents, the ten or twenty statements that are copied most often across your submissions?
• Is each reusable component classified as locked, adapted, or derived?
• When a locked source changes, can you list every document and submission that uses it?
• Is there a rule for what happens to documents already submitted when a source changes (for example, they stay as submitted, and only new versions pick up the change)?
• Have you agreed with regulatory affairs how reuse interacts with regional differences, so a component approved for one region is not reused in another region where it does not apply?

Area Four: Governance and Ownership

In a document-first world, ownership is simple: the author of the document owns it. In a component world, one component may appear in documents owned by clinical, regulatory, CMC, safety, and medical writing. Someone has to own the component itself, and everyone who uses it has to accept that they cannot change it on their own.

This is the area where programs most often stall. The technology works, the content model is sound, and then the first change to a shared component triggers a disagreement about who needs to approve it. Settle this before content moves into a new system.

Governance also has to cover the content model itself. When ICH or CDISC updates M11 terminology, or when an internal team wants a new component type, someone must decide whether and how the model changes. The M11 guideline notes that its template and technical specification are versioned documents that may be revised under a change control process as protocol requirements and technology change¹. Your internal model needs the same discipline, and a way to track which external version it is aligned to.

Governance Checks

• Is there a named owner for every component your pilot will reuse?
• Do functional leaders agree, in writing, that they cannot edit locked components used in their documents?
• Is there a documented process for changing the content model and templates, with a named decision-maker?
• Do you know which version of the M11 template and terminology your protocol model follows?
• Has quality assurance agreed how component-level approvals relate to document-level approvals?

Readiness Area Five: Tooling and Validation

Only once the first four areas are in reasonable shape does platform selection become productive. At that point the requirements are clear: the system has to hold the components you defined, enforce the reuse types you chose, support the approval paths you agreed, and produce documents that meet your publishing and submission standards.

What to Ask of a Platform

Validation

A structured authoring platform that holds content destined for regulatory submissions is a computerized system that needs a risk-based validation approach proportionate to its intended use. Where the system manages electronic records submitted to FDA or applies electronic signatures, 21 CFR Part 11 expectations for audit trails, access controls, and signatures apply. In the EU, EU GMP Annex 11 sets comparable expectations for computerized systems used in GMP activities, and GCP expectations apply to trial documentation. Most organizations handle this through their existing computerized system validation (CSV) procedures, typically following GAMP 5 risk-based principles.

Two points specific to structured authoring are easy to miss.

• Validate the assembly logic, not only the software. The rules that pull components into documents, resolve conditional sections, and populate derived tables are configuration. If a rule is wrong, every document built with it is wrong. Those rules need testing and change control like any other configuration.
• Validate the data feeds. Derived content is only as good as its source. If a schedule of activities is generated from a study design tool, the interface between the two systems is part of the validated scope.

If you plan to add AI drafting on top of structured content, treat it as a separate intended use with its own risk assessment. We covered that in Validating GenAI for Regulatory Document Authoring.

Tooling and Validation Checks

• Were platform requirements written from your content model, templates, reuse policy, and governance design?
• Has the GxP relevance of the platform been assessed and documented?
• Are assembly rules and data interfaces inside the validation scope?
• Can the platform export the structured content your standards call for, or is there a plan for how it will?
• Have real authors tested it on a real document before the purchase decision?

How ICH M11 Connects: Starting With the Protocol

For most sponsors, the protocol is the best place to start structured content authoring, and M11 is the reason. Four features make it a practical first document family.

The Content Model Already Exists

The hardest readiness item, the content model, is largely provided. The M11 template sets the structure, headings, and instructions, and the technical specification defines each item with conformance, cardinality, relationships, concept codes, and reuse rules¹². You will still need internal extensions for sponsor-specific content, but you are adapting a published model rather than inventing one.

Reuse Is Designed In

The M11 guideline lists design for content reuse as a core principle. It describes the protocol as a rich source of information that can be reused in clinical trial management and review, for publishing on clinical trial registries, and for standardized data capture¹. The technical specification’s design principles include a focus on relevant content use and reuse, and enabling development of a data model and an open, non-proprietary exchange message standard¹. The downstream consumers of protocol content (the statistical analysis plan, the case report form design, the registry entry, the clinical study report) are exactly where reuse delivers value.

Downstream Links Are Being Built by Standards Bodies

CDISC’s USDM work has represented the M11 template in a study definitions model and added elements to populate SDTM trial design domains and registry information⁷. TransCelerate has aligned its Common Protocol Template with M11⁸. A sponsor who structures protocols to M11 is structuring them in a form that other tools and standards are being built to read.

The Timing Is Forgiving

Because neither FDA nor EMA has mandated the M11 format, there is no filing deadline forcing a rushed implementation⁴⁵. That gives sponsors room to pilot, learn, and adjust. It also means there is no external pressure to finish, which is why a clear internal case matters.

What M11 Does Not Solve

M11 scope is the interventional clinical trial protocol for medicinal products, across phases and therapeutic areas¹. The guideline states that it does not specify processes for developing and maintaining a protocol, and that it does not tell sponsors how to design a good trial¹. It does not cover the investigator brochure, the CSR, nonclinical documents, or CMC. For those, you will still need your own content models, drawing on CORE Reference for CSRs¹⁸, the draft M4Q(R2) structure for quality content¹², and FDA’s PQ/CMC data elements where relevant¹³.

A Realistic View of Effort and a Phased Start

The brief for any structured content program should be honest about effort. Most of the work is not software. It is modeling content, agreeing on ownership, rewriting templates, cleaning up existing content before it moves, and changing how authors and reviewers work. That work falls on scarce people: senior medical writers, regulatory leads, and CMC experts who are also writing live submissions.

Where the Effort Goes

• Content modeling. Even with M11 as a base, someone has to map current templates to it, decide which sponsor-specific extensions are needed, and document definitions. This needs experienced authors, not only analysts.
• Content clean-up. Existing content usually has several versions of the same statement. Before it becomes a locked component, someone has to decide which version is correct and retire the others. This is slow and requires scientific judgment.
• Governance design. Agreeing on ownership across functions takes time and senior attention. It cannot be delegated to a project team.
• Author and reviewer change. Reviewers who are used to reading a whole document now need to understand that locked content was already approved and should not be edited in place. Without that shift, review cycles do not get shorter.
• Validation and integration. Platform validation, interfaces to source systems, and publishing integration follow normal CSV effort.
• Running two ways at once. For a long period, some documents will be structured and some will not. The organization needs to support both, and the hybrid period is where most frustration arises.

A Phased Start

How to Tell Whether It Is Working

Pick measures before the pilot starts, and keep them simple. Useful ones include the time from protocol concept to approved protocol, the number of review cycles on reused sections, the number of inconsistencies found in quality control across protocol, statistical analysis plan, and registry entry, and the time needed to assess the impact of a change to a shared component. Report both the gains and the new burdens (for example, time spent on component governance), because leadership will ask about both.

Structured content authoring also connects with other data programs. Product data prepared for IDMP, study design data in CDISC standards, and submission planning for eCTD 4.0 all benefit from the same discipline of defined, owned, reusable information. If those programs are running in parallel, align their content models early so the organization does not end up with three definitions of the same product. Our earlier pieces on IDMP progress and eCTD 4.0 migration planning cover those dependencies.

Conclusion

Structured content authoring is no longer a speculative idea for regulatory documents. ICH M11 gives the clinical protocol a published content model with defined data elements, conformance rules, and reuse rules, and FDA and EMA have both adopted it, without making it mandatory. Similar structured approaches are in progress for quality content, product information, and labeling. The organizations that benefit will be those that treat structured content as a discipline first: a clear content model, templates built on it, a written reuse policy, named owners for shared content, and only then a platform chosen and validated to support all of that. The effort is real, most of it falls on experienced authors and functional leaders rather than IT, and it is best spent on one well-chosen document family with measurable outcomes.

Sakara Digital works with pharma and biotech organizations on regulatory content strategy, data governance, and the validation of the systems that support them. If you are weighing where structured content authoring fits in your submission plans, or how to use ICH M11 as a first step, we are happy to have that conversation and offer an independent view of where to start.

For Further Reading