Systematic Review Protocol Planning and Registration

The search is not the beginning of the review

Running the first database search feels like the moment the review actually starts. It is not. By then the team should already have settled a much larger set of choices. The question decides what evidence is eligible. The eligibility criteria shape the search. The outcomes you plan to report govern what gets extracted, and the synthesis plan decides which data you need to collect at all.

Cochrane treats protocol development as a way to make these judgements before the findings can influence them. Its guidance asks teams to pre-specify the search, the risk-of-bias assessment, the synthesis, any subgroup analyses, and the certainty assessment where those methods apply.1 Planning has a practical payoff too: it shows early on what expertise and resources the review is going to need.2

A weak start spreads through the whole project. Vague eligibility criteria turn into screening conflicts. An outcome list without definitions produces extraction that no two people do the same way. And when the synthesis plan is written after the results arrive, readers have no way to tell which decisions came first and which were made with the findings already on the table.

A protocol connects scientific decisions with project operations

PRISMA-P is a 17-item reporting checklist for systematic review protocols. It covers administrative information, rationale and objectives, eligibility criteria, information sources, the search strategy, study records, outcomes, risk of bias, data synthesis, meta-biases, and certainty assessment.3 If you want the reasoning behind each item, along with worked examples of complete reporting, the explanation and elaboration paper supplies both.4

A checklist gets you complete reporting, nothing more. The team still has to pick methods that fit the question and the field. An intervention review, a qualitative synthesis, a diagnostic accuracy review, an environmental evidence synthesis, and a policy review will not share one design. That is what field-agnostic planning means in practice: the decision logic travels across disciplines while the methods stay matched to the evidence.

These four parts depend on each other. A sound scientific method stalls when nobody owns the tasks or the team has no access to specialist support. A polished schedule just moves the wrong method faster. Registration, for its part, will preserve an incomplete plan with perfect accuracy. The protocol’s job is to hold the scientific and operational layers together.

Registration creates a public record, not a quality certificate

Prospective registration lets readers find ongoing reviews, spot unplanned duplication, compare the published review with its earlier plan, and inspect major changes. PROSPERO was built around exactly those goals: public records that capture the core attributes of a review, plus an audit trail of major methodological changes.5 The wider case for registration rests on transparency, reproducibility, and protection against decisions driven by results.6

Pick a registry or repository that fits the review type, the subject area, and the stage of the work. Its fields will never hold the full method. The protocol stays the detailed source of truth, and where the platform allows it, the public record should point readers to that document.

Registration is not a promise that nothing will change. New information can expose an ambiguous criterion, a data source that turns out to be unavailable, or an analysis the evidence cannot support. Cochrane asks authors to document and report changes to protocol methods.2 PRISMA-P asks protocol authors to say up front how they will record important amendments.3

Undocumented drift separates the registered plan from the published review

A registration number says nothing about whether the team actually followed the registered methods. Readers need to see the plan and the finished work side by side. In a 2021 sample of 100 non-Cochrane systematic reviews with PROSPERO records, 90 differed from their record in at least one eligibility component, and the publications rarely explained why.7 That was one sample, so it should not be read as a prevalence estimate for reviews in general. What it does show is the exact problem an amendment trail exists to solve.

Some differences are sound responses to things the authors could not have known at the start. Others change which evidence enters the synthesis. Timing, rationale, approval, and disclosure are what separate the two. This is why a team should keep every protocol version instead of overwriting the original plan.

Reporting guidance supports the plan; it does not replace conduct guidance

Each guideline has one defined job. PRISMA-P covers the protocol. PRISMA 2020 covers the report of a completed systematic review.8 PRISMA-S adds detail for reporting literature searches.9 PRESS supports peer review of electronic search strategies.10

The conduct standard is a separate choice, and it has to match the review question. Cochrane, JBI, Campbell, CEE, or another established handbook may fit, depending on the evidence and the decision setting. Name the guidance you are using in the protocol, say where you adapted it, and do not claim compliance that the final methods cannot back up.