The SPIRIT checklist specifies what a clinical trial protocol should contain, and the current version is SPIRIT 2025, which carries 34 items. Its name stands for Standard Protocol Items: Recommendations for Interventional Trials. It replaces SPIRIT 2013 and was developed alongside the CONSORT 2025 update, which is why the two now align item for item in most places.
A protocol is the only document in a trial's life that is written before anybody knows the answer, and that is exactly where its value sits. Every methodological protection a trial claims later, that the primary outcome was not chosen after seeing the data, that the analysis was not adjusted until it worked, that the subgroup finding was planned, rests on a dated document that said so in advance. SPIRIT is the list of things that document has to pin down.
It helps to be concrete about the failure the guideline prevents. A trial with three plausible outcome measures and no prespecified primary has, in effect, three chances to find an effect, and a reader has no way to know how many were looked at. A trial whose analysis adjusts for covariates chosen after unblinding has fitted the model to its own results. Neither of these requires bad faith; both happen through ordinary drift when nothing was written down.
Once results exist, no amount of careful writing can restore the distinction between confirmatory and exploratory. That is why the three hardest SPIRIT items are the three that cannot be repaired retrospectively: the primary outcome definition, the sample size justification, and the statistical analysis plan.
The items follow the structure of a protocol document. Described in our own words, they cover the following. The authoritative wording and official templates sit with the CONSORT and SPIRIT group and should be taken from there.
Administrative information. The title identifying the design and population, the trial registration number and registry with the data set, the protocol version and date, sources of funding, the roles and responsibilities of authors, sponsor and any committees, and the sponsor's role in design, analysis and the decision to publish.
Introduction. The background and rationale including a summary of relevant studies and the justification for the comparator, the specific objectives or hypotheses, and the trial design including allocation ratio and the framework, whether superiority, non-inferiority, equivalence or exploratory.
Methods, participants and interventions. The study setting and list of sites, eligibility criteria including criteria for sites and the individuals delivering the interventions, the interventions described in replicable detail, the circumstances under which they may be modified or discontinued, strategies for improving adherence and how it will be assessed, and any concomitant care permitted or prohibited.
Methods, outcomes and timeline. Completely defined primary and secondary outcomes including the measurement variable, the analysis metric, the aggregation method and the timing, with an explanation of clinical relevance; the participant timeline of enrolment, interventions and assessments; the sample size with all the assumptions behind it; and the recruitment strategy for reaching target numbers.
Methods, assignment. How the allocation sequence is generated including any stratification or blocking, the allocation concealment mechanism, who implements assignment, and who is blinded plus the circumstances under which unblinding is permitted.
Methods, data collection and management. The plans for assessing and collecting outcome and other data, plans to promote retention and complete follow-up including how data from participants who discontinue will be handled, data management including entry, coding and range checks, and confidentiality arrangements. Any plans for biological specimens are covered here too.
Methods, statistics and monitoring. The statistical methods for primary and secondary outcomes, methods for any additional analyses including subgroup and adjusted analyses, the definition of the analysis population and how missing data will be handled, the composition and role of any data monitoring committee, plans for interim analyses and stopping rules, plans for collecting and reporting harms, and the frequency and procedures for auditing.
Ethics and dissemination. Plans for seeking ethics approval, communicating protocol amendments, obtaining informed consent including for ancillary studies, declarations of interest, who has access to the final data set, provisions for ancillary and post-trial care, plans for dissemination to participants and the public including authorship and any reproducible research arrangements, and model consent materials.
These are separate obligations and confusing them causes real problems at publication. A registry entry, on a platform such as ClinicalTrials.gov or ISRCTN, is a structured public record created before enrolment, and the registration item asks for its number and data set. A protocol is the full methodological document. A registry entry contains a fraction of the 34 items and cannot substitute for one.
The practical consequence is that a journal asking for your protocol will not accept a registry link, and that discrepancies between the two get noticed. If the primary outcome in the registry differs from the protocol, expect to explain which is authoritative and when it changed.
How SPIRIT and CONSORT work as a pair
The alignment between SPIRIT 2025 and CONSORT 2025 is deliberate. Most protocol items have a direct reporting counterpart, so a reader holding both documents can check the trial that happened against the trial that was planned. Sample size justified in the protocol, sample size reported in the paper. Primary outcome defined in advance, primary outcome reported first. Analysis population specified, analysis population used.
This is also why CONSORT asks for changes to the trial design and to outcomes after commencement, with reasons. Amendments are normal, and a protocol with a clean dated amendment log turns what would otherwise look like outcome switching into documented, defensible study management. Keep the log as you go; reconstructing it at submission is both harder and less convincing.
The same protocol-then-report structure appears across research designs: PRISMA-P and PRISMA 2020 for systematic reviews, and PREPARE and ARRIVE in preclinical animal work. Observational research is the notable exception, which is part of why STROBE leans on authors to distinguish prespecified analyses from later ones inside the report itself. The EQUATOR Network library indexes the current versions of all of them.
Publishing the protocol, and completing the checklist
Protocols are publishable in their own right, and several journals peer review them as standalone articles. Doing so is worth considering for three reasons: it timestamps your methods in a citable form, the peer review arrives while you can still act on it, and it removes any later argument about what was planned.
Complete the checklist against the final protocol version, recording the page and section for each item rather than ticking it, and name the version you followed. Where an item does not apply, say so with a reason. The items that most often cannot be pointed at a page are the analysis population definition, the missing data strategy, and the plan for collecting harms, and each of those is far cheaper to settle now than after the first participant is enrolled.