GUIDES
G04 OF 07 · 7 MIN

Scoping review, systematic review, meta-analysis

Written and reviewed by Reza D, OTR/L, MOT, CPACC · Published 9 September 2026 · Free to read

Four things get called a review, and only one of them is designed to answer “does this work?”. Telling them apart takes about ten seconds and changes how much weight the paper can carry.

What is the difference between a scoping review, a systematic review and a meta-analysis?

A scoping review maps what research exists on a topic. It asks how much there is and what shape it takes, not whether the treatment works. A systematic review asks a specific question, searches exhaustively for every study that answers it, and appraises their quality. A meta-analysis is a systematic review that goes one step further and combines the results statistically into a single pooled estimate. A narrative review is an expert writing about a topic, with no stated search and no reproducible method.

Comparison

The question each is built to answer, and what it cannot tell you.
Review type	The question it answers	What it cannot tell you
Scoping	What research exists here, and what does it cover?	Whether anything works. Studies are mapped, not weighed, so a scoping review counts good and poor studies alike.
Systematic	What does all the evidence on this specific question say?	A single number. It may conclude the studies are too different or too weak to combine — which is a finding, not a failure.
Meta-analysis	Taken together, how big is the effect?	Whether the pooled number applies to your patient. An average across trials can hide populations where it does nothing.
Narrative	What does this expert think, and why?	Anything about completeness. Nothing obliged the author to find studies that disagreed with them.

The quickest test

Look for the search. A systematic review or meta-analysis will name the databases, the dates and the terms, and report how many records were screened and how many survived. A scoping review usually does this too. A narrative review does not, because it never searched systematically — and that is the whole distinction. If you cannot find the numbers, you are reading an opinion, however well informed.

A scoping review can find that there is nothing

This is the most useful thing a scoping review does, and the reason not to dismiss it as the weakest of the three. Establishing that a question has barely been studied is a real result, and it is one no other design delivers.

Worked example

Scoping review · occupational therapyScoping review of occupational therapy for distal radius fractures, including complications2026 · scoping reviewFive studies, total, on occupational therapy for wrist fractures — and those five focused almost entirely on range of motion and grip strength rather than on whether people got back to doing things. The finding is the gap. An OT reading this now knows that the outcome they care about has essentially not been measured.Read the summary

Worked example

Scoping review · the absence as the findingEmergency ambulance healthcare for people with intellectual disabilities: a scoping review2026 · scoping reviewNo study anywhere captured the actual experience of a person with an intellectual disability calling an ambulance. That sentence is worth more to a service planner than a pooled effect size would be.Read the summary

What a scoping review cannot do is tell you a treatment works. It counts studies; it does not weigh them. A scoping review reporting that twenty-eight studies found benefit is reporting a count, not a verdict, and the twenty-eight may include work a systematic review would have excluded on the first pass.

A systematic review that finds nothing has still worked

A systematic review is built to survive its own answer. If the honest conclusion is that the evidence is too thin, too inconsistent, or too poor to support the treatment, the method has done its job.

Worked example

Systematic review · a null result · free to readInclusion of participants with limited English proficiency in autism research2026 · systematic reviewAcross 1,007 autism studies, only 19% included participants with limited English proficiency, and just 18% of randomised trials did. The review answers its question precisely — and the answer is that most of the evidence base was built without these families in it, which changes how every other paper in that field should be read.Read the summary

How do you read a meta-analysis without reading the forest plot?

A forest plot is one row per study: a square for that study’s result, a horizontal line for its confidence interval, and a diamond at the bottom for the pooled estimate. Three things tell you most of what you need.

Where the diamond sits. If it crosses the line of no effect, the pooled result did not reach significance, whatever the abstract emphasises.
How wide everything is. Wide lines mean small or imprecise studies. A narrow diamond built from wide lines deserves less confidence than its width suggests.
Whether the squares agree. Studies scattered on both sides of the line mean the treatment did different things in different places, and the average may describe none of them. This is what heterogeneity means, and it is usually reported as I².

Worked example

Meta-analysis · the average hides the answerBrain–computer interface for upper limb functional recovery post-stroke2026 · meta-analysisPooled across sixteen trials, adding brain–computer interface training to standard rehabilitation improved upper limb scores. But the benefit was clearest in subacute stroke and much less apparent in chronic stroke — so the single pooled number describes a patient group you may not be treating. The subgroup is the finding; the average is the headline.Read the summary

The same shape appears in a hand therapy meta-analysis in the index: heat therapy reduced short-term pain and improved grip strength, and did not improve overall hand function. Pooling three outcomes gives three answers, and only one of them is the one a patient would notice.

A narrative review is not a systematic review

Thirteen papers here are narrative reviews. They can be excellent — an expert synthesising a field is genuinely useful — but nothing obliged the author to look for evidence that contradicted them, and there is no way for you to check what was left out. On this site they rate low certainty for that reason, not because the writing is poor.

The tell is the same one as above: no search strategy, no number of records screened, no exclusion criteria. If those are missing, read it as an argument rather than as a summary of the evidence.

What to do with each

Scoping review: use it to find out what exists and what has never been studied. Do not use it to justify a treatment choice.
Systematic review: the strongest single thing you are likely to find. Read its conclusion about the quality of the evidence, not just its conclusion about the treatment.
Meta-analysis: find the subgroup that matches your patient before you accept the pooled number.
Narrative review: read it for orientation and for the reference list, then go and read the studies.

Reviews rate moderate on this site and narrative reviews rate low — the guide on what the certainty rating means explains why no single paper reaches high. If the review you are reading turns out to be a pilot or a case series in disguise, the companion piece on studies that cannot prove anything covers those.

OTHER GUIDES

G01What the certainty rating on a research summary actually meansEvery paper on this site carries one of three words: moderate, low, or very low. Here is where each comes from, what it does not mean, and what it should change about how you read the summary underneath it.G02MCID vs MDC: telling whether a change on an outcome measure is realTwo numbers decide whether a score has moved: one asks whether the change is bigger than the measurement error, the other whether it is big enough for the patient to care. They are not the same, and one is useless without the other.G03Pilot, feasibility, protocol, case report: what a study that cannot prove anything still tells youA pilot, a feasibility study, a case report and a protocol each answer a real question. None of those questions is whether the treatment works — and reading them as if it were is the commonest way to misread the literature.G05What a qualitative study can tell you, and what it cannotIt does not measure whether a treatment works. It finds out what something is like, in enough detail that you recognise it. That is a different question, with answers a trial cannot produce.G06Reading a validation study: the numbers that decide whether an assessment is usableA validation study asks whether an assessment measures what it claims, in the people you are using it on. It answers with a handful of numbers, and two of them decide whether the tool is usable at all.G07"No significant difference" is not "no difference"It means the study did not find an effect. It does not mean there is no effect. Those are different claims, and only one of them is usually supported by the paper in front of you.

Written and reviewed by Reza D, OTR/L, MOT, CPACC. Guides are the only pages on Applied Evidence written by a person. Paper summaries are generated by software and marked as such. Last reviewed 9 September 2026.

← All guides and the A–Z of labels