How to Plan a Rapid Review: Methods, Shortcuts and Transparent Reporting

By Dr. Esmaeel Saeedy Robat, Founder of MetaSyn Academy · Published meta-analyst (Nature Human Behaviour, 2026)

A rapid review is not a systematic review with a few boxes deleted. It is an evidence-synthesis process in which selected methods are deliberately simplified or omitted so that evidence can reach a decision in time. That makes the planning problem harder, not easier: the team has to decide which safeguards can be streamlined, which should remain intact, and how the resulting uncertainty will be explained to the people using the review.^1,2

The field has moved away from treating rapid review as one uniform recipe. Cochrane’s updated guidance contains 24 recommendations for rapid reviews of effectiveness, but it explicitly allows teams to apply all or some restrictions according to the topic, timeframe and resources. Separate guidance now exists for rapid qualitative evidence synthesis and for rapid scoping, mapping and evidence-and-gap-map work. The common principle is transparent adaptation; the exact shortcut package depends on the underlying review design.^2,11,12

Appropriateness comes before acceleration

A genuine deadline matters because rapid review exists to support decisions that cannot always wait for a conventional review timeline. But the date on the calendar is only one part of the appropriateness decision. Current methods guidance asks what decision will be made, whether the question is urgent or high priority, the cost of waiting, whether the findings can actually be used on time, how complex the question is, and whether the team has enough review expertise and resources to produce something interpretable.³

This is why a thesis submission date and a reimbursement deadline are methodologically different, even when both are six weeks away. In the reimbursement case, a defined external decision may proceed with or without synthesized evidence. In the thesis case, the deadline is largely a property of the research project. It may justify narrowing the research question or choosing a feasible study design, but it does not by itself establish that systematic-review safeguards should be relaxed. The same is true of convenience, small-team size or a desire for faster publication. Those constraints are real, but the rapid-review appropriateness literature does not treat them as sufficient reasons on their own.³

There is no universal shortcut package

Rapid reviews vary because the expensive parts of evidence synthesis vary. For one question, screening 20,000 records may dominate the workload. For another, the search is manageable but outcome extraction is complex. A third may have only six studies but require difficult risk-of-bias judgments and a network meta-analysis. Applying the same set of shortcuts to all three reviews would be convenient, but methodologically weak.

The current search guidance makes this point directly. Search acceleration can come from spending less time conducting searches or from reducing the size of the result set. Because screening often consumes more resources than running the database search itself, the authors recommend investing effort in search planning and involving an information specialist so that a smaller set of sources can still be chosen intelligently and the strategy can balance sensitivity with precision.⁴ A narrower search is therefore not automatically a faster review if it creates a poorly targeted result set or forces the team to compensate later.

Search restrictions trade coverage for workload, not certainty for a known amount

The literature search is often where teams are most tempted to impose a simple rule such as “MEDLINE only,” “English only,” or “last five years.” Current guidance is more careful. For rapid reviews of effectiveness focused on randomized trials, Cochrane provides design-specific database recommendations; the broader rapid-search paper advises a small number of relevant information sources selected with an information specialist. Those recommendations should not be converted into one universal database count for education, environmental policy, qualitative synthesis or every other rapid product.^2,4

Language restriction illustrates why context matters. A meta-epidemiological study of 59 Cochrane intervention reviews found that excluding non-English publications had little effect on the conclusions in that particular sample, even though 31 studies were lost from the evidence base.¹⁸ That is useful empirical evidence, but it is not permission to assume English-only searching is harmless in every topic. The result came from clinical intervention reviews, and excluded studies were often small. In a global policy question, infectious-disease topic, traditional medicine review or LMIC-focused synthesis, the representation consequences may be much larger.

Date restrictions need the same discipline. A natural starting point can exist when a technology was introduced, a law changed, or an existing high-quality review provides a verified search endpoint. “The last five years” is different: it is a workload shortcut with no intrinsic relation to the evidence question. If the review uses such a limit, the protocol should say what earlier evidence may be lost and why the decision-maker accepted that trade-off.

Screening is where speed creates a measurable exclusion risk

Screening is one of the few rapid-review steps with a clear empirical demonstration of what can be lost. Gartlehner and colleagues randomized 280 participants to screen abstracts from two review topics. Single-reviewer abstract screening had a sensitivity of 86.6% and missed 13% of relevant studies; dual screening missed about 3%.⁶ The finding is important because it quantifies a real error mechanism, but it should not be turned into a universal “13% penalty.” Reviewer expertise, topic complexity, eligibility criteria, calibration and software can all change performance.

Cochrane’s updated guidance therefore recommends partial dual independent screening and agreement assessment before moving to single screening. It gives a proportion such as 20% and agreement such as κ≈0.8 as examples.² The methodological idea is more durable than the numbers: calibrate the team on enough records to understand disagreement, predefine what acceptable agreement means for this project, and preserve a mechanism for detecting false exclusions. If reviewers disagree substantially during calibration, acceleration should not proceed simply because the deadline remains unchanged.

Extraction can be streamlined without making one person the only line of defence

Current rapid-review team guidance recommends a piloted extraction form, a minimal set of required data, one person extracting and another checking for correctness and completeness.⁵ This differs from simply assigning the entire task to one reviewer. Verification matters most for data that directly feed the result: effect estimates, denominators, variance measures, outcome definitions and information that affects eligibility or risk-of-bias judgments.

A protocol can also reduce extraction breadth. If the commissioner needs five decision-critical outcomes, extracting dozens of secondary outcomes because they happen to appear in the papers may waste the rapid timeline. The safe restriction is prospective: identify the variables required to answer the question, explain why other fields are not needed, and avoid deciding what is “important” only after seeing which results are statistically attractive.

Risk of bias and certainty do not disappear because the review is rapid

For rapid reviews of effectiveness, Cochrane retains risk-of-bias assessment. The accelerated staffing model is one assessor with another person verifying the judgments, and outcome-level assessment may be focused on outcomes most important for decision-making.^2,5 This is a useful example of a principled shortcut: it removes reviewer duplication while retaining the evaluative function that makes the effect estimate interpretable.

Certainty assessment is slightly different. The dedicated guidance recommends full GRADE for Cochrane rapid reviews if time and resources allow. When full implementation is not feasible, teams can focus on the main intervention and comparator, restrict ratings to critical benefits and harms, use one reviewer with second-person verification, or reuse trustworthy certainty ratings from an incorporated high-quality systematic review.⁷ The right conclusion is not “GRADE is mandatory in every rapid review,” nor is it “GRADE can be omitted because the review is rapid.” The protocol should state what level of certainty assessment the decision needs, what is feasible, and what users lose if it is absent.

The underlying design still matters. Rapid scoping reviews are not rapid effectiveness reviews; formal risk-of-bias or GRADE assessment is not automatically required merely because the word rapid appears in the title. Rapid qualitative evidence synthesis uses different appraisal and confidence concepts. This is why an adaptation template must branch by design rather than treating one health-intervention workflow as the definition of rapid evidence synthesis.^11,12

Meta-analysis is not the enemy of speed

Another common misconception is that a rapid review should avoid meta-analysis. The updated Cochrane guidance says to consider meta-analysis when appropriate and resources permit, and it applies the same methodological standards used in a systematic review.² The evidence-synthesis guidance makes the same distinction: every rapid review should synthesize the evidence, but the form of synthesis depends on the studies and the question.⁸

The statistical calculation is often not the slow part. Harmonizing outcome definitions, checking extracted numbers, resolving multiplicity, deciding whether studies are clinically comparable and planning heterogeneity analyses can consume much more time. If those tasks are required to answer the decision question, removing the meta-analysis does not necessarily produce a better or faster synthesis. Conversely, forcing a meta-analysis into heterogeneous data because a commissioner wants one number is equally indefensible. Rapid status changes the timetable, not the conditions under which pooling is meaningful.

Decision-makers should shape the question, not dictate unsafe methods

Rapid reviews are closely connected to knowledge users because their value depends on reaching a decision in time. Cochrane’s knowledge-user guidance describes opportunities to involve patients and public partners, healthcare providers and policymakers during planning, conduct and knowledge translation. It also acknowledges that meaningful involvement consumes time and resources.⁹ The practical implication is not that every rapid review needs every stakeholder group. It is that the people who will use the review should be identifiable, and the team should decide where their input improves decision relevance.

Methodological authority needs a separate line. A commissioner can say that evidence is required by 15 October, that mortality and hospital admission are the outcomes driving the policy, and that a 300-page report would be unusable. Those are legitimate decision requirements. The commissioner should not simply declare that risk-of-bias assessment must be removed or that one database is enough. The review team has to translate the deadline into a defensible package of adaptations.

The adaptation matrix is an audit trail, not a risk calculator

The strongest rapid-review protocol is not the one with the largest list of shortcuts. It is the one in which each departure from the standard method can be traced to a reason. A practical matrix can record the review step, planned adaptation, rationale, anticipated consequence and mitigation. This corresponds closely to Cochrane’s recommendation to describe the selected rapid approach, restricted methods and potential limitations so readers can understand how those restrictions affect interpretation.²

What the matrix should not do is assign a fabricated “low,” “moderate” or “high” bias score to every shortcut. The evidence base is too uneven. Single-reviewer abstract screening has direct empirical accuracy data, but the magnitude of evidence loss from omitting a particular database or grey-literature source depends heavily on the topic. Some restrictions may miss studies without materially changing a conclusion; others can erase a whole regional evidence base. A descriptive consequence such as “may reduce comprehensiveness; magnitude uncertain” is more honest than an unsupported red score.

Transparent reporting is part of the method

Reporting is especially important in a rapid review because the reader cannot infer which parts of the conventional process were compressed. PRISMA 2020 remains the general reporting standard for systematic reviews. PRISMA-RR is still listed by EQUATOR as under development as of 9 August 2026, so it should not be cited as a finalized checklist.^14,16

In the meantime, Stevens and colleagues have published interim rapid-review reporting guidance. The paper identifies a minimum set of six rapid-review-specific items in addition to existing guidance, including methods-related items and publication-ethics considerations. Among the important methodological points are reporting a priori iterative methods, describing what distinguishes the rapid review from a conventional systematic review, and explaining knowledge-user involvement.¹³ The final journal issue is 2025 even though the article was published online in 2024, so citation dates should be handled carefully.

This is why protocol planning and reporting should be connected. If the protocol records every restriction, rationale, mitigation and planned iterative rule, the final report does not have to reconstruct those decisions from memory. Deviations can be added as they occur. The result is not a claim of “PRISMA-RR compliance.” It is a transparent account of what was planned, what changed and what the rapid approach means for interpretation.

Sometimes the right rapid-review decision is to renegotiate

Some deadlines cannot support the question originally requested. An extremely broad effectiveness question may require too many searches, too much screening or too many outcomes. A commissioner may request a one-database search, unverified screening, no appraisal and no certainty assessment because all four appear to save time. The correct response is not automatically to accept the package and place a limitation paragraph at the end.

Methodologically, narrowing scope is often safer than removing independent safeguards across every stage. A team can prioritize the population, intervention, comparison or outcomes that actually drive the decision. It can build on a current high-quality systematic review rather than starting again. It can negotiate a staged product, with a rapid initial synthesis followed by a fuller update. Or it can extend the deadline. The appropriateness framework asks reviewers to consider the risk of an incomplete rapid synthesis against the risk of no synthesis and the cost of waiting.³

If no defensible method fits the requested timeline, the team should say so. That conclusion is not a failure of rapid review. It is the point of methodological planning.

Different decisions justify different adaptations

Where the protocol template fits

The paired Rapid review protocol template turns the principles in this guide into an audit trail. It starts with the decision need and underlying synthesis design, then asks the team to record the method retained or adapted, why the adaptation is necessary, the consequence that may remain and the verification used to reduce avoidable error. It also separates prospective adaptations from later protocol deviations.

The tool deliberately does not generate a quality percentage or label the review “valid.” There is no validated equation that converts a set of rapid-review shortcuts into a probability of correctness. A protocol can be complete and still contain a method that needs expert reconsideration. The safe endpoint is therefore ready for methodological review, not certified.

Questions about rapid-review planning

A rapid review is strongest when the trade-offs are visible before the results

Rapid review planning is not a search for the fastest possible method. It is a negotiation between a real decision deadline and the minimum methodological work needed to make the evidence interpretable. Some standard methods can be retained. Some can be staffed differently. Some can be narrowed to critical outcomes. Some restrictions may be defensible only for a particular topic. Others create uncertainty that cannot be quantified in advance.

The protocol should make those choices visible before the team knows what the studies say. That is what separates deliberate acceleration from post hoc convenience. When the review is published, readers should be able to see the same chain: why rapid conduct was chosen, what changed from the full method, how errors were mitigated, what limitations remain, and how much confidence the decision-maker should place in the result.