Systematic Reviews and Meta-Analyses Explained
What systematic reviews and meta-analyses are, how they pool studies, what they can and cannot fix, and how to judge whether a review is trustworthy.
Content last updated . This page has not been medically reviewed.
Two terms, often mixed up
A systematic review collects the research on one question using a plan written in advance. A meta-analysis is a statistical step that combines the numbers from two or more studies. A review may include a meta-analysis, but not always: if the studies are too different, the authors may only describe them side by side.How a systematic review works
Cochrane's handbook says systematic reviews seek to collate evidence that fits pre-specified eligibility criteria, aim to minimize bias, and use explicit methods documented in advance with a protocol. [1] The aim is to avoid choosing only the studies that fit a view. A reporting guideline, PRISMA 2020, has a 27-item checklist and flow diagram so readers can see how many studies were found, screened and included. [2]What a meta-analysis does
Meta-analysis is the statistical combination of results from two or more separate studies. [3] Pooling can give a more precise estimate than any one small study, and it lets readers see whether results agree.Limits to know
| Problem | Why it matters |
|---|---|
| Different studies pooled | Heterogeneity (differences in people, doses, designs, outcomes) limits how far conclusions generalize. [3] |
| Reporting and publication bias | Cochrane warns that meta-analyses can "mislead seriously" if reporting biases, study design and within-study biases are not carefully considered. Smaller studies can differ from larger ones because of publication bias or bias within studies. [3] |
| Poor studies in, poor review out | Pooling cannot repair weak designs. Observational studies stay vulnerable to confounding (observational studies and confounding). |
| Funding | Who sponsored the included studies and the review can matter (funding and conflicts of interest). |
Reviews are only as current as their search date. For fast-moving fields, a review can miss recent trials.
How to judge a review in five checks
- Was there a written protocol and clear eligibility criteria? [1]
- Does the report follow a standard such as PRISMA, with a flow diagram of studies? [2]
- Did the authors rate study quality and discuss differences between studies?
- Did they consider missing or unpublished results?
- Do the conclusions match what the included studies can support, including statements such as "the evidence is too limited to say"?
When reviews report "no good evidence," that is a finding in itself, not a gap in the review.
Related guides
Studies that reviews pool are explained in randomization, blinding and placebo. To see why single studies fail to repeat, read why promising findings sometimes fail. The emerging science evidence landscape shows where reviews sit among other evidence.Frequently asked questions
Is a meta-analysis always stronger evidence than one trial?
No. It can be more precise, but Cochrane warns it can mislead seriously if biases in the studies or in what gets reported are ignored. [3]
What is PRISMA?
A reporting guideline for systematic reviews, updated in 2020, with a 27-item checklist and flow diagrams. [2]
What does heterogeneity mean?
Differences between the studies being pooled; it affects how far a combined result can be generalized. [3]
Does a review settle a question for good?
No. New trials can change the picture, and a review reflects only the studies available when it searched.