Designing a Clinical Trial Protocol

A clinical trial protocol is not the same document as a general research protocol, a grant proposal’s research plan, or an IRB application narrative, even though all four overlap heavily. CASRAI’s guide to what a research protocol is covers the concept across disciplines and funders generally; the guide to writing a protocol narrative for IRB submission covers what a human-subjects reviewer needs to make a Common Rule risk/benefit finding; and CASRAI’s practical guide to endpoints, sample size, randomization, and the SAP covers the statistical mechanics in depth. This guide sits between those: it is about the clinical trial protocol as a single governing document — the sections a sponsor and investigator are expected to include, why those sections exist, and how the finished protocol relates to the trial’s registration and its statistical analysis plan.

Because a clinical trial protocol is reviewed by an IRB/independent ethics committee, a sponsor’s quality system, and in most cases a regulator, its structure is not a matter of house style the way a manuscript’s might be. Two ICH guidelines set the framework almost every institutional and sponsor protocol template is built around: ICH E6(R2) (Good Clinical Practice), whose Section 6 specifies the content a clinical trial protocol and its amendments must contain, and ICH E8(R1), General Considerations for Clinical Studies, which sets out the “quality by design” framework for identifying what is actually critical to a trial’s ability to answer its question reliably, before drafting begins. See CASRAI’s full Good Clinical Practice (GCP) guide for how these guidelines fit into the broader GCP compliance picture.

The governing standards: ICH E6 Section 6 and ICH E8(R1)

ICH E6(R2) Section 6, “Clinical Trial Protocol and Protocol Amendment(s),” is the closest thing the field has to a required table of contents for a protocol. Its subsections run from general trial information and background, through trial objectives and purpose, trial design, selection and withdrawal of subjects, treatment of subjects, assessment of efficacy, assessment of safety, and statistics, to direct access to source data, quality control and quality assurance, ethics, data handling and record keeping, financing and insurance, and publication policy. A sponsor is not required to reproduce this exact numbering, but a reviewer — whether an IRB, a sponsor’s own quality function, or a regulator — will expect to find each of these questions answered somewhere in the document. ICH E6(R3), the current core GCP revision (finalized January 2025, phasing in across regulators through 2025), restructures GCP around principles and annexes rather than E6(R2)’s numbered sections, but does not remove the underlying expectation that a protocol document contain the same substantive content; during the transition, most sponsors are still working from E6(R2)’s Section 6 structure or an equivalent institutional template built on it.

ICH E8(R1), “General Considerations for Clinical Studies,” reached ICH Step 4 in 2021 and works one level up from Section 6’s content checklist. Rather than prescribing sections, it asks the sponsor and investigator to identify, before drafting, the factors that are actually critical to quality for that specific trial — the design choices, data, and processes whose failure would most threaten the trial’s ability to protect participants and produce a reliable answer — and to build the protocol and the trial’s quality management around those factors specifically, rather than applying uniform scrutiny to everything. This “quality by design” approach is why two protocols for very different trials (a large multi-site confirmatory trial and a small early-phase safety study) can look structurally similar under Section 6 while differing enormously in how much detail and oversight each section actually needs. E8(R1) also reinforces that fitness-for-purpose trial design — not just regulatory box-checking — is the standard a protocol is judged against.

Core sections of a clinical trial protocol

The sections below follow the substance of ICH E6(R2) Section 6, in the order most institutional and sponsor templates use. As with any protocol, an actual submission should follow the sponsor’s or institution’s current template rather than a generic outline — templates vary in exact section numbering and in what supporting appendices they require.

Background and rationale

What is already known about the condition and the investigational product or intervention, the gap the trial addresses, and why the trial is justified now — including a summary of relevant nonclinical and prior clinical data. This section is what an IRB/IEC and a regulator use to judge whether the trial’s anticipated benefit could plausibly justify exposing participants to its risks, so it needs to state the evidentiary basis explicitly rather than assert significance in the abstract.

Objectives, purpose, and endpoints

The trial’s primary and secondary objectives, translated into specific, pre-specified endpoints. This is the section a protocol shares most directly with its statistical analysis plan: endpoint selection (primary vs. secondary vs. exploratory, and the data type — binary, continuous, or time-to-event — behind each) drives the sample-size calculation and the analysis methods downstream. CASRAI’s guide to designing a clinical trial: endpoints, sample size, randomization, and the SAP covers endpoint typology, composite and surrogate endpoints, and sample-size mechanics in full; this section of the protocol document simply needs to state the chosen endpoints precisely enough that the reader could tell in advance what result would answer the question.

Trial design

The overall design — randomized or non-randomized, controlled or uncontrolled, parallel-group or crossover, blinded or open-label, single-site or multi-site — along with the trial’s phase and, where relevant, any adaptive elements. See CASRAI’s Clinical Study Design guide for how these design categories relate to one another and to the trial’s phase.

Eligibility criteria

Inclusion and exclusion criteria, written as concrete, checkable rules rather than general descriptions. Loosely stated eligibility criteria are one of the most common reasons a protocol comes back for revision, both because they make the risk/benefit analysis harder to evaluate and because they make it harder to confirm, mid-trial, that a specific participant was actually eligible. Where the population includes a group that triggers additional protections — children, pregnant women, or another population likely to be vulnerable to coercion or undue influence — the protocol needs to say so explicitly and describe the corresponding safeguards.

Intervention and comparator

The investigational product or intervention itself — dose, formulation, route, schedule — and, for a controlled trial, the comparator: placebo, an active comparator, or standard of care. Sourcing an active comparator is its own operational undertaking, distinct from sourcing the investigational product, and is covered in CASRAI’s dictionary entry on comparator sourcing. The protocol needs to specify not just what the comparator is but where it comes from, how it is blinded or matched to the investigational product where applicable, and how its supply chain is managed to the same GCP and, where the comparator is itself a regulated product, GMP standard as the investigational product.

Statistical analysis plan (reference)

The protocol’s statistics section states the trial’s statistical approach at a level sufficient for ethical and regulatory review — the primary analysis population, the general analysis method, and the sample-size justification — but the full statistical analysis plan (SAP) is typically finalized as its own document, in more granular detail, before database lock and before unblinding. Treating the SAP as a separate, more detailed document rather than folding all of its content into the protocol is standard practice specifically because the SAP often needs refinement (multiplicity strategy, missing-data handling, exact model specification) after the protocol itself has already gone through ethics review, and re-approving the entire protocol for every SAP refinement would be disproportionate. What the protocol and the SAP must never do is diverge on anything that changes what was ethically reviewed — the endpoints, the population, and the primary analysis approach need to match between the two documents.

Safety monitoring

How adverse events and serious adverse events are defined, graded, and reported; the stopping rules, if any, that would halt or modify the trial for safety reasons; and who is responsible for ongoing safety oversight. For trials with meaningful risk, that oversight function is often an independent Data Safety Monitoring Board (DSMB), which reviews unblinded interim safety (and sometimes efficacy) data on a schedule specified in the protocol and its own charter, separate from the trial’s day-to-day investigator and sponsor team. The protocol needs to specify not just that safety will be monitored, but the concrete mechanism — reporting timelines to the sponsor and IRB/IEC, the DSMB’s composition and authority (advisory vs. binding stopping authority), and the specific criteria that would trigger a safety review outside the regular schedule.

Protocol amendments and deviations

A protocol is expected to be followed as written, but it is not expected to be static: a protocol amendment is a documented, prospectively IRB/IEC-approved change made before it takes effect, while a protocol deviation is an unplanned departure from the already-approved protocol, documented after the fact and assessed for whether it affected participant safety or data integrity. Anticipating likely sources of amendment during drafting — an eligibility criterion that may need loosening once recruitment data comes in, for example — does not remove the need for a formal amendment later, but a protocol that is realistic about where change is likely tends to need fewer major, disruptive amendments than one that is not.

Where a pilot study fits before the full protocol is finalized

Many of the specific values a protocol has to commit to — the effect size and variability behind a sample-size calculation, the feasibility of a recruitment target, whether an assessment instrument or procedure actually works as intended in the target population — are often not reliably known until they have been tested on a small scale first. A pilot study is run specifically to generate that information before the full trial protocol is locked, rather than to test the trial’s actual hypothesis. Running a pilot does not remove the need for a full protocol under ICH E6/E8 for the definitive trial that follows — it is a distinct, usually smaller-scale, separately reviewed study whose findings feed into the design decisions (sample size, procedures, eligibility thresholds) that the full protocol then commits to.

How the protocol relates to trial registration

A finalized protocol is also the source document for the trial’s public registration entry. Under FDAAA 801, an applicable clinical trial must be registered on ClinicalTrials.gov within 21 days of enrolling its first participant, and under ICMJE’s publication-eligibility policy, registration is expected at or before first participant consent — both deadlines that fall very close to, or before, when the protocol itself is finalized. Key registration data elements (the primary and secondary outcome measures, eligibility criteria, the intervention and comparator arms, and the target enrollment) are drawn directly from the protocol, which is why a protocol amendment that changes any of these elements generally requires a corresponding update to the registry entry. CASRAI’s full guide to clinical trial registration and reporting compliance covers the ICMJE, FDAAA, and NIH registration and results-reporting requirements, including how they differ from one another, in depth.

Frequently asked questions

What’s the difference between designing a clinical trial protocol and writing one for IRB submission?

They overlap heavily but serve different immediate readers. Protocol design, as covered here, is about assembling the full set of scientific, operational, and statistical content ICH E6(R2)/E8(R1) expect a clinical trial protocol to contain. Writing specifically for IRB submission — covered in CASRAI’s separate guide — is about presenting that same content so a human-subjects reviewer, who may not be a specialist in the field, can make the specific risk/benefit and consent findings the Common Rule requires. A clinical trial protocol needs to satisfy both audiences at once.

Do ICH E6(R2) and ICH E8(R1) apply to every clinical trial?

ICH GCP (E6) specifically governs interventional trials of regulated drugs and biologics (with a related but distinct standard, ISO 14155, covering device trials); ICH E8(R1)’s general considerations framework is written broadly enough that many sponsors and institutions apply its quality-by-design principles to trial design more generally, even for studies outside ICH’s formal drug/biologic scope. Purely observational, non-regulated research is typically governed by separate human-subjects frameworks such as the Common Rule rather than by ICH GCP directly.

Does the protocol need to contain the full statistical analysis plan?

No. The protocol’s statistics section states the analysis approach at a level sufficient for ethical and scientific review; the full SAP is typically a separate, more granular document finalized later, before database lock and before unblinding. What matters is that the two documents never diverge on anything that changes what was ethically reviewed — the endpoints, population, and primary analysis approach.

Is a pilot study required before writing the full trial protocol?

Not universally, but it is common practice wherever a key design input — expected effect size, recruitment feasibility, whether a procedure works as intended — is not already known with confidence from prior data. A pilot study is a distinct, separately reviewed study; running one doesn’t reduce what the definitive trial’s own protocol needs to contain.

What happens to the trial registration if the protocol is amended?

If the amendment changes a data element captured at registration — outcome measures, eligibility criteria, target enrollment, or the intervention/comparator arms, for example — the registry entry generally needs to be updated to match. Registries are not automatically synchronized with protocol amendments; updating the registration is a separate, affirmative step the responsible party has to take.

Designing a Clinical Trial Protocol: Structure, ICH E6/E8, and Registration