How to Write a Clinical Trial Protocol
Turn a research question into a document that can be run, monitored, and defended.
What a how-to is not
A how-to is SPEQ’s practitioner method, not a procedure. It does not replace your own SOP, it is not a validated approach, and the judgement calls in it belong to your quality unit.
The protocol is the trial. Every downstream artefact — the consent form, the case report form, the analysis plan, the study report — either derives from it or contradicts it, and there is no third option. Most protocol problems are not scientific: they are places where the document asks a site to do something it never quite defined, and the deviation log fills up with the consequences.
- 1
Fix the question before the design
State the objective as a question the trial can answer, then choose the design that answers it. Teams that pick a design first end up writing objectives that describe what the design happens to produce, which is how a trial ends with a result nobody asked for.
- 2
Define eligibility so a site can apply it without calling you
Each criterion needs an unambiguous test — a value, a window, a documented diagnosis. Criteria phrased as clinical judgement ("adequate organ function", "clinically significant") generate screening deviations and inconsistent populations across sites, because ten investigators will draw the line in ten places.
- 3
Specify the endpoints and how they are measured
Name the primary endpoint, the instrument or assay that produces it, the timepoint, and who is blinded to what. An endpoint without a specified measurement method is a promise the analysis cannot keep, and it cannot be repaired after the first participant is dosed.
- 4
Build quality into the design rather than inspecting it in later
Identify the factors that genuinely matter to participant safety and result reliability, and design the procedures around those. Applying uniform intensity to everything spends monitoring effort on data that could not change a conclusion, which is the failure the risk-based approach exists to correct.
- 5
Write the schedule of assessments as an operational table
Every visit, every procedure, every window, with the acceptable tolerance stated. A schedule with no windows guarantees deviations, because participants do not arrive on the ideal day. State what is a deviation and what is within tolerance before the first one happens.
- 6
Define the safety reporting flow and the roles inside it
Say who assesses seriousness, who assesses causality and expectedness, who notifies whom, and against which reference document expectedness is judged. Sites cannot infer this, and a flow discovered during the first serious event is a flow that will be late.
- 7
Reconcile the protocol against everything derived from it
Before finalising, read the consent form, the case report form and the analysis plan against the protocol side by side. Each amendment reopens this check — a protocol amended without re-reconciling its dependents is the most common source of consent forms that describe procedures the trial no longer performs.
- !Eligibility criteria phrased as clinical judgement, so ten investigators apply ten different thresholds.
- !Visit windows omitted, which converts ordinary scheduling reality into a stream of protocol deviations.
- !Expectedness assessed with no named reference safety document, so causality assessments cannot be reproduced.
- !Amendments issued without re-reconciling the consent form and case report form against the new text.
How to Write a Clinical Trial Protocol: frequently asked questions
Common questions on write a clinical trial protocol.
How much operational detail belongs in the protocol rather than a manual?
Anything that changes what data mean belongs in the protocol; anything that only changes how a task is performed can sit in a manual of procedures. The test is whether a site doing it differently would alter the interpretation of a result — if so, it is protocol content, because the manual is not what regulators assess.
What does a quality-by-design approach change in practice?
It changes where effort goes. Rather than monitoring everything to the same depth, you identify the small set of factors critical to participant safety and to the reliability of the results, and concentrate design and oversight there. The point is not to do less; it is to stop spending scrutiny on data that could not change a conclusion.
When does a change need a formal amendment?
When it affects participant safety, the scientific interpretation of the trial, or the conduct of the study — those are substantial and require review before implementation. Administrative corrections do not, but the boundary is a documented judgement, not an assumption, and treating a substantial change as administrative is a finding.