Why a Platform Needs More Than One Protocol Layer
A platform trial may evaluate several interventions over time, with some features shared across arms and others necessarily different. If every detail is copied into one large protocol, repeated material can become difficult to maintain. If each arm is described in an independent protocol, shared design features can drift apart or become ambiguous. The document structure therefore has a statistical as well as an administrative purpose: it should make clear which rules apply across the platform, which apply to a particular arm, and where their interaction is defined.
A master protocol describes the platform-wide framework. An arm-specific protocol appendix adds or refines information for a particular intervention within that framework. Shared statistical documents, such as a statistical analysis plan or a platform statistical charter, specify common analytical conventions and identify where analyses differ by arm. These documents are complementary: no single document should be expected to repeat every detail, but a reader must be able to determine the governing rule for each relevant part of the design.
What Belongs in Each Document
The master protocol is the natural location for features that should have one platform-wide meaning. Depending on the design, these may include the overall objectives, the general participant pathway, common eligibility criteria, the platform structure, common outcome definitions, and procedures that apply to all active arms. It should state how arm-specific material relates to the common framework, including which document controls when a subject is addressed in more than one place.
An arm-specific appendix should contain details that would be misleading or unworkable if written as if they were identical across interventions. Examples include the investigational product and its administration, intervention-specific safety information, any justified additional eligibility criteria, and operational details unique to that arm. It should identify the arm and version it governs, distinguish additions from replacements, and avoid silently changing a platform-wide rule.
Shared statistical documents connect the overall design to its analysis. They can define common analysis populations, outcome variables, time origins, censoring conventions, missing-data handling, and reporting formats. Where a method differs for one arm, the document should describe the difference and its rationale rather than implying that a common method applies. A statistical document does not replace the protocol: it elaborates statistical implementation within the approved design and should not quietly introduce a new design feature.
| Document | Primary scope | Useful content | Common boundary to state |
|---|---|---|---|
| Master protocol | Platform-wide | Common framework, shared procedures, and cross-arm definitions | Which provisions may be refined by an arm appendix |
| Arm-specific appendix | One intervention or arm | Intervention details, arm-specific procedures, and justified additions | Whether a provision supplements or replaces a common rule |
| Shared statistical document | Common statistical implementation | Analysis definitions, common methods, and explicit arm-level variations | Which approved design document the analysis implements |
This allocation is not a universal regulatory template. The exact document set depends on the platform, sponsor, applicable requirements, and operational needs. The essential test is functional: can a site, analyst, monitor, or reviewer locate the applicable rule without guessing which document takes precedence?
A Consistency Rule for Overlapping Provisions
Document layering works only when the layers compose coherently. A simple way to express this is to treat the master protocol as defining a common set of requirements and the arm appendix as adding requirements for that arm. For eligibility, let \(C_M\) be the set of participants meeting the master protocol criteria and \(C_j\) the set meeting any additional criteria in arm \(j\)'s appendix. If the appendix adds restrictions without changing the common criteria, the arm's eligible set is:
The intersection follows because a participant must satisfy both the common requirements and the arm-specific additions. If the appendix instead intends to replace a common rule, the document must identify that as an approved exception and specify its scope; otherwise, a reader cannot know whether to take the intersection or treat one statement as superseding another. This is a small but important derivation: without a declared relationship, two individually clear provisions can yield an indeterminate combined rule.
The same logic applies beyond eligibility. For example, a platform-wide outcome definition and an arm-specific assessment schedule can coexist if the appendix specifies how the schedule implements the common definition. They conflict if the appendix changes the meaning, time origin, or ascertainment window without saying so. A shared statistical document can then translate the agreed definitions into analysis conventions, but it cannot resolve an undocumented design conflict by itself.
Worked Example: Combining Common and Arm-Specific Eligibility
Suppose a hypothetical platform master protocol requires participants to be adults aged 18 to 75 years and to have a confirmed diagnosis. The appendix for Arm A adds an age range of 40 to 70 years because the intervention is being evaluated only in that population. Assume the appendix states that this range is an additional restriction, not a replacement.
The master-eligible ages are \(18 \leq \text{age} \leq 75\), and the arm-specific ages are \(40 \leq \text{age} \leq 70\). Applying the intersection rule gives:
Thus, a 35-year-old with the confirmed diagnosis meets the age criterion in the master protocol but is not eligible for Arm A. A 65-year-old with the diagnosis meets both age requirements, subject to all other criteria. A 78-year-old fails the common criterion and cannot enter Arm A. The conclusion depends on the explicit statement that the appendix adds a restriction; if it merely listed a different range without explaining its relationship to the master protocol, the documents would be ambiguous.
Keeping the Platform Statistically Coherent
A coherent platform does not mean that every arm has identical treatment procedures or identical analyses. It means that the shared concepts remain stable where they are intended to be shared, and that departures are visible and justified. A statistical document should therefore distinguish three things: platform-wide definitions, arm-specific definitions, and design features that require a common interpretation across both.
For example, if a primary outcome is defined platform-wide, the document should state the outcome and its assessment window in a way that applies to every relevant arm. An appendix may specify how the outcome is collected for its intervention, but should not alter the endpoint silently. If an arm genuinely needs a different endpoint or time window, the difference must be explicit, and the shared statistical document must make clear that its analysis is not simply the common analysis with a different treatment label.
A useful justification for keeping these links explicit is that an analysis result is interpretable only when its population, outcome, and time frame can be traced to the approved design. If the master protocol defines the population one way, an appendix narrows it without identifying the change, and the analysis plan uses the original population definition, the documents no longer describe one analysis. The problem is not merely editorial: the analysis population may not match the participants or question specified for the arm.
Worked Example: Locating a Time-Window Difference
In a hypothetical platform, the master protocol defines the primary outcome as symptom improvement assessed at 12 weeks after randomization. Arm B requires an additional clinical visit at week 8 for safety monitoring. The arm appendix says that this extra visit does not replace the week-12 outcome assessment.
The statistical document should retain the common primary outcome time point of 12 weeks for Arm B. The week-8 visit is an arm-specific operational addition, not a change to the primary outcome. If the analysis instead used the week-8 measurement for Arm B without an approved amendment and a clear statement of the revised estimand and analysis, the platform would be comparing results based on differently timed outcomes.
The document check is therefore concrete: the master protocol supplies the common outcome definition; the appendix adds a safety visit and explicitly preserves the common assessment; the statistical document implements the week-12 analysis. Each document has a distinct role, and their combined instructions agree.
Choosing a Document Architecture
A master protocol with arm-specific appendices is not the only possible structure. A monolithic protocol places all common and arm-specific information in one document. This can make precedence straightforward because there is one primary source, but repeated sections may become long and every arm change may require revising a large document. Separate standalone protocols can be simpler for teams working on one intervention, but common provisions may diverge and platform-wide relationships can be harder to see.
A layered structure aims to balance these concerns. It centralizes common rules, isolates genuine arm differences, and uses shared statistical documents to state how the design is analyzed. It is most useful when the platform has meaningful shared procedures and enough arm-level variation to make full duplication cumbersome. It is less helpful when appendices proliferate, duplicate the master text, or require readers to consult many files for a basic instruction.
There is no statistical guarantee that a particular architecture produces valid inference. Validity still depends on the design, assumptions, implementation, and analysis. Document architecture supports those conditions by making the intended design traceable and reducing the risk that investigators implement different versions of what was meant to be one platform.
Worked Example: Resolving a Conflicting Control Statement
Consider a hypothetical master protocol that says all participants are followed for 30 days after treatment discontinuation. An arm appendix says participants in Arm C are followed for 60 days after discontinuation because of a treatment-specific safety concern. The appendix does not state whether the 60-day period replaces or supplements the common follow-up.
The first step is not to guess which statement is more specific. The document set needs an explicit relationship. If the intent is longer follow-up for Arm C, the amendment should state that the 60-day requirement replaces the common 30-day duration for Arm C, while the common follow-up procedures apply otherwise. If the intent is an additional safety contact while the common follow-up remains unchanged, it should say that instead.
For the replacement interpretation, a concise rule would be: “For Arm C, follow-up after treatment discontinuation is 60 days; all other arms follow the master protocol's 30-day duration.” This makes the scope visible and preserves the common rule for other arms. The shared statistical document can then specify how the follow-up data are represented and analyzed, consistent with that approved design. Until the ambiguity is resolved, there is no defensible basis for treating either duration as the governing instruction.
Diagnostics, Failure Modes, and Limitations
A practical review should test the links among documents rather than proofreading each file in isolation. Signs of a poorly functioning architecture include duplicated language with different versions, arm appendices that silently redefine shared terms, statistical documents that specify procedures absent from the approved design, and references to appendices that do not identify the applicable version. A reader should also be able to distinguish a true exception from an example or explanatory note.
A useful review process is to inventory the key concepts—eligibility, outcomes, visit timing, safety follow-up, analysis populations, and statistical definitions—and record the authoritative location for each. For every arm-specific difference, verify that the appendix states whether it supplements, narrows, or replaces the master provision. Then check that the shared statistical document reflects the resulting design, rather than relying on informal assumptions.
These checks detect documentary inconsistency, not every design flaw. A perfectly cross-referenced set of documents can still specify a scientifically inappropriate outcome or an analysis that performs poorly. Conversely, a sound design can be implemented incorrectly if sites or analysts cannot determine which version governs. Document coherence is necessary for reliable implementation, but it is not a substitute for scientific review, operational testing, or appropriate statistical design.
Common Mistake
A common mistake is to assume that the arm-specific appendix automatically overrides the master protocol because it is more specific. That reasoning is unsafe: specificity alone does not establish precedence, and the appendix may be intended only to add a procedure. The correction is to state the relationship explicitly—addition, restriction, or replacement—and identify the scope of any exception. If the relationship is unclear, the conflict must be resolved through the applicable document-governance process before implementation, not inferred by each site or analyst.
Check Your Understanding
Answer each question, then compare with the brief solutions below.
- The master protocol requires a confirmed diagnosis. An arm appendix adds a biomarker-positive criterion and says it is additional. What rule determines eligibility for that arm?
- A shared statistical document gives one common outcome time point. An arm appendix adds a visit but does not say whether the outcome time changes. What should be checked before analysis?
- Name one advantage and one risk of separate standalone protocols compared with a master protocol and appendices.
- A specific appendix statement conflicts with a master-protocol statement, and neither document states precedence. Is it safe to assume the appendix controls? Why?
- What can a cross-document consistency review establish, and what can it not establish?
Solutions. (1) Participants must meet both the master criteria and the additional biomarker criterion; the eligible set is their intersection. (2) Clarify whether the added visit is operational only or changes the outcome definition, and ensure the approved design and statistical document agree before analysis. (3) Standalone protocols may be easier for an arm team to use, but common provisions can diverge; a layered protocol can centralize common rules, but may become hard to navigate if it duplicates material or has unclear precedence. (4) No. Specificity does not itself establish authority or whether the appendix adds or replaces the master rule; the relationship must be made explicit. (5) It can establish whether provisions and references agree across documents. It cannot prove that the scientific design or statistical method is appropriate.