Systematic review, meta-analysis, scoping review or umbrella review: which one?

The terms are often used as if they were interchangeable, but they are not. A systematic review is a method for finding, appraising and summarising all the evidence on a question. A meta-analysis is a statistical technique for combining numerical results, and it is done inside a systematic review when the studies are similar enough to combine.

Choosing the right type at the start decides the protocol, the tools, the reporting standard and where the review can be registered and published.

The four main types

Systematic review: a defined question, a reproducible search, explicit eligibility criteria, appraisal of each study and a structured synthesis, reported to PRISMA 2020. It may or may not include a meta-analysis.
Meta-analysis: a statistical combination of results from similar studies. A meta-analysis without a systematic search is not a systematic review, and its result depends on whichever studies happened to be picked.
Scoping review: maps the extent and type of evidence on a broad topic, without judging study quality or pooling results. It follows PRISMA-ScR and JBI guidance.
Umbrella review: a review of systematic reviews and meta-analyses, used to summarise what is known across many questions in a field. It needs a way to handle the overlap of primary studies between reviews, and an appraisal tool such as AMSTAR 2.

At a glance

Systematic review: answers a focused question about effects, risks, accuracy or prognosis. Pools results: sometimes. Appraises the quality of each study: yes. Reporting guideline: PRISMA 2020. Can be registered on PROSPERO: yes.
Meta-analysis: answers how large an effect is, across studies that can be combined. It is a step inside a systematic review. Reporting guideline: PRISMA 2020.
Scoping review: answers what evidence exists on a broad topic. Pools results: no. Appraises the quality of each study: usually not. Reporting guideline: PRISMA-ScR. Can be registered on PROSPERO: no.
Umbrella review: answers what systematic reviews say across a field. Pools results: rarely. Appraises the quality of each review: yes. Reporting guideline: PRIOR.

Let SynthGap run the first pass

Enter an area. SynthGap proposes candidate topics, checks each against existing systematic reviews, and lists the primary studies behind it. Start on the Free plan: no card needed, and every run shows every topic it finds.

Which should you choose?

Is your question narrow enough to be answered by comparing a defined intervention, exposure or test in a defined group? If so, choose a systematic review.
Do enough studies report a comparable numerical outcome? If so, add a meta-analysis. If not, use a narrative or structured synthesis and say why you did not pool.
Is your question broad, for example what is known about a topic or what kinds of evidence exist? Then a scoping review is the right type.
Are there already many systematic reviews on the area with different conclusions? Then an umbrella review can clarify, though it will not add new primary evidence.
Is an older review on the same question out of date? Then an update, or a living review, may be the best use of your time.

What this means for registration and publication

Systematic reviews with a health-related outcome can be registered on PROSPERO, ideally before screening begins and in any case before data extraction starts. Scoping reviews cannot be registered there, and teams often register them on the Open Science Framework instead. Journals expect the reporting checklist that matches the type: PRISMA 2020 for systematic reviews and meta-analyses, PRISMA-ScR for scoping reviews, and PRIOR for overviews of reviews. Choose the type before you write the protocol, because changing it later usually means starting again.

Where SynthGap fits

SynthGap helps at the start: it proposes candidate topics, checks them against existing systematic reviews, and reads the studies behind the topics you choose to write a critical report. It does not conduct the review for you, and its suggestions are AI-generated, so verify them before you register a protocol.

Questions

Is every systematic review a meta-analysis?

No. Many systematic reviews do not pool results because the studies are too different, and report a structured narrative synthesis instead. That is a legitimate result, not a failure.

Can I do a meta-analysis without a systematic review?

You can calculate a pooled estimate from a convenient set of studies, but it is not a reliable summary of the evidence, because the studies were not selected by a reproducible, comprehensive search.

How long does each type take?

It varies with the number of records. A systematic review with meta-analysis usually takes many months, a scoping review is similar, and an umbrella review can be quicker if the underlying reviews are few. Estimating the number of records you will have to screen is the best planning step.

Keep reading

General methodological guidance, not medical advice. SynthGap's suggestions are AI-generated from the published literature, so verify them before registering a protocol.