Choose Your Evidence-Synthesis Review Type: Decision Tools

A review selector should help you make a better methodological decision, not give a false impression that one click can diagnose the only correct design. This tool begins with the purpose of the synthesis and the unit of evidence. It then considers whether existing reviews are adequate, whether an externally imposed decision deadline justifies rapid methods, whether a living updating model is warranted, and whether the question points to a specialist methodology that should not be forced into a generic category.

The reason for this structure is simple. Evidence-synthesis labels do not all describe the same property. Current methods literature distinguishes evidence-synthesis types, such as systematic reviews, scoping reviews and overviews of reviews, from modes such as rapid and living approaches that can modify how a synthesis is conducted or maintained.¹ The updated Right Review project reaches the same practical problem from a different direction: it uses guiding questions, definitions and outputs to orient reviewers among many possible synthesis approaches rather than assuming that a short list of labels is self-explanatory.^2,3

T28 deliberately has a narrower job. It does not attempt to reproduce Right Review’s detailed method library, and it does not cover every specialist synthesis method. It gives an academically defensible starting point for common design decisions, makes the reasoning visible, and stops when the question requires more specialized judgment. The output is therefore a best-fit candidate, not a methodological verdict.

What the tool is actually deciding

The first three questions deal with the underlying architecture. If the aim is to answer a focused question using primary studies, a systematic-review pathway is usually the starting point, although the exact subtype may depend on whether the question concerns intervention effects, diagnosis, prognosis, prevalence, etiology, qualitative evidence or another domain. If the aim is to map concepts, evidence characteristics, methods or gaps, a scoping review may fit better. If the evidence unit is existing systematic reviews, an overview of reviews, often called an umbrella review, becomes plausible, but only after the quality, currency and overlap of those reviews are considered.^4,5,6

Rapid and living decisions come later because they answer different questions. Rapid review methods are used to produce evidence under a genuine time constraint through deliberate and transparent methodological adaptations; current Cochrane guidance does not define one universal shortcut package.^7,8 Living systematic reviews retain the underlying systematic-review methods while adding continuing surveillance and timely incorporation of new evidence. Entry into living mode depends on the importance and uncertainty of the question, the likelihood that important new evidence will appear, and the feasibility of maintaining the review.^9,10

How to interpret the result

A clear fit means your stated purpose and evidence unit point in the same direction and no specialist signal creates an obvious conflict. It does not mean that the review’s protocol is already correct. A systematic review still needs a question-specific framework, eligibility criteria, search methods, appraisal methods and synthesis plan. A scoping review still needs a defensible PCC-based scope and planned charting. An overview still needs explicit criteria for what counts as an eligible systematic review, review-level appraisal and a strategy for overlapping primary studies.

A likely fit means the main direction is defensible, but at least one issue should be resolved before the label is frozen. Common examples are an overview candidate where the currency of existing reviews has not been assessed, or a scoping candidate where the user also expects a strong effect conclusion. A multiple plausible approaches result means the purpose or evidence unit remains mixed. In that situation, narrowing the question is usually more useful than adding more labels.

A specialist methodological review recommended result is a deliberate stop. Qualitative evidence synthesis, diagnostic-test-accuracy reviews, prognosis reviews, prevalence reviews, realist synthesis, mixed-methods reviews, economic evidence synthesis, systematic maps and IPD projects each have methodological decisions that a short general selector cannot safely resolve. RETREAT, for example, was developed to help teams choose among qualitative evidence-synthesis methods using seven criteria rather than treating all qualitative synthesis as one method.¹¹

Six boundaries the tool will not blur

Worked examples: the same label can be right or wrong for different reasons

Example 1: a broad digital-literacy field

A research team wants to know what definitions of digital literacy have been used in secondary schools, which populations have been studied, what instruments have been used and where evidence gaps remain. The topic is broad, but breadth is not the decisive fact. The decisive fact is that the team wants to map concepts, measures and gaps. A scoping review is a defensible starting point. If the team instead asks whether a particular digital-literacy intervention improves a specified outcome compared with usual teaching, the purpose changes. That focused question points toward a systematic review even if the surrounding literature remains heterogeneous.

Example 2: twelve systematic reviews on the same intervention

The existence of twelve reviews makes an overview possible, not automatically preferable. The team first needs to know whether those reviews are current, whether they address sufficiently compatible questions, whether their methods are trustworthy and how much their primary studies overlap. If most reviews recycle the same trials or omit important recent evidence, an updated or de novo systematic review may give a cleaner answer. T28 therefore treats review adequacy as a gate rather than treating the number of reviews as a vote for overview methodology.

Example 3: a government evidence request due in eight weeks

An external decision-maker needs evidence on a defined question before a policy meeting. If a systematic-review architecture is appropriate to the question, rapid mode may be justified. The resulting project is not simply a standard review completed more quickly. The protocol should state which methods are being adapted, why they are acceptable for the decision, and what limitations those decisions create. If the policy question is still too broad to support a focused synthesis, a rapid scoping or mapping approach may be more suitable; dedicated rapid guidance for these broad-review methods is now available.¹⁶

Example 4: an oncology question with new trials every month

Evidence velocity alone is not enough to select living mode. The question also needs to matter to decision-makers, current evidence should retain meaningful uncertainty, new studies should have a realistic chance of changing conclusions, and the review team must have a way to maintain searching, screening, appraisal, analysis and communication. If those conditions are present, a living systematic review may be appropriate. If the team cannot sustain the work, a planned periodic update is more defensible than calling the review living and allowing it to lapse.

Example 5: patient experiences of long-term treatment

If the purpose is to understand experiences, values, meanings or social processes, a conventional systematic-versus-scoping choice may be too crude. A scoping review could map the qualitative literature, but a question that asks what the experiences mean, how themes relate, or how explanatory concepts can be developed points toward qualitative evidence synthesis. T28 names the specialist family and stops. It does not choose meta-ethnography, thematic synthesis, framework synthesis or another QES method on the user’s behalf.

Static fallback when JavaScript is unavailable

The interactive result is a convenience layer. The core logic remains readable without JavaScript. Use the table below as a static fallback and record any unresolved branch in the protocol-development notes.

Before you freeze the review label

Use the result as the beginning of a protocol conversation. The review team should be able to explain, in ordinary language, why the chosen design can answer the intended question. That explanation should remain defensible even if the fashionable label disappears. If the explanation is merely “this is what similar papers call it,” the design is not yet settled.

• Write the purpose in one sentence without naming a review type.
• State the unit of evidence and why that unit is appropriate.
• Clarify whether the required output is an inferential answer, a map, an interpretation, or a review-level synthesis.
• If an overview is being considered, examine the quality, currency, compatibility and overlap of existing reviews before committing.
• If rapid mode is being considered, identify the external decision deadline and document the methods that may need adaptation.
• If living mode is being considered, document why continued surveillance is worthwhile and how maintenance could eventually stop.
• If the tool raises a specialist flag, use the specialist methodology before finalizing the protocol.
• Only after these decisions are stable should the final review label be treated as settled.

What this tool cannot decide

T28 does not assess whether your question is important, whether your eligibility criteria are coherent, whether your search strategy is sensitive enough, whether your appraisal tool matches the included study designs, or whether your planned synthesis is statistically and conceptually defensible. It cannot inspect the actual evidence base. An overview result does not mean that the available reviews are good enough to synthesize. A systematic-review result does not tell you whether the evidence should be pooled. A living flag does not guarantee that the infrastructure for living review maintenance exists.

The tool also does not provide numerical confidence. There is no validated gold-standard dataset that would justify saying a project is, for example, “87% scoping review.” Review selection involves methodological judgment about purpose, evidence and intended inference. Descriptive categories such as clear fit, likely fit, multiple plausible approaches and specialist methodological review recommended are more defensible than fabricated probability.

Frequently asked questions

Continue from selection to method