A systematic review vs meta-analysis comparison comes down to method versus maths. A systematic review answers a question by summarizing pre-specified evidence, while a meta-analysis statistically combines trial results and is always accompanied by a systematic review.
The distinction matters when a headline cites a “pooled” result. This guide shows how the review is built, how to read the pooled result and what to check before trusting it. For wider coverage of how studies reach the headlines, see medical news today.
The stages of a systematic review
A medical school library guide describes a systematic review as a piece of work that asks a research question and then answers it. What separates it from an ordinary literature summary is the order of work: the rules come first and the evidence second.
Stage one: fix the question and methods in advance
In a systematic review, the methods used to search for and analyze the data are transparent, reproducible and defined before searching begins. That sequence limits the temptation to pick studies that suit a preferred conclusion. A reader who wants to repeat the work should be able to follow the same steps and reach the same set of studies.
Stage two: search widely and report the search
The PRISMA statement recommends that authors report their complete search strategies for all databases, registers and websites. PRISMA is a minimum set of recommendations designed primarily to encourage transparent and complete reporting of systematic reviews. It does not include recommendations for designing and conducting literature searches, so a fully reported search is not automatically a good search. Readers still need to judge whether the databases and search terms were broad enough.
Stage three: screen studies independently
Eligibility screening decides which studies make it into the review, and the library guide notes that this is done by at least two people working independently. Independent screening protects against one person’s judgement quietly shaping the evidence base. A review that does not say who screened, or how disagreements were settled, gives readers less reason for confidence.
Stage four: synthesize the findings
Once studies are selected, their findings must be combined. Narrative synthesis combines the findings of multiple studies using words. Some systematic reviews instead present results using meta-analysis, which is covered next. Both routes sit inside a systematic review, so a review without a pooled number is not a weaker kind of review, only one that reports its results differently.
Systematic review vs meta-analysis: the working differences
A meta-analysis is a statistical method that combines the results of several trials into an average result. A systematic review is the wider process that finds and selects those trials. The table sets the two side by side.
| Feature | Systematic review | Meta-analysis | Planning use |
|---|---|---|---|
| What it is | A piece of work that asks a research question and answers it from pre-specified evidence | A statistical method that combines results of several trials | Ask whether a story is describing a whole review or only its pooled number. |
| How results are presented | Narrative synthesis (in words) or a meta-analysis | An average result with greater precision than individual studies | A review can be sound without a pooled figure. |
| Dependence | May or may not contain a meta-analysis | Always accompanied by a systematic review | A pooled result with no described search or screening deserves scepticism. |
| Main risk | Flawed methods, bias or uninformative design | Misleading if the pooled data are not sufficiently similar or are of poor quality | Check the inputs before the headline number. |
These distinctions come from the medical school library guide and a university library research guide.
The payoff from pooling is precision. Meta-analysis adds value because it can produce a more precise estimate of the effect of a treatment than any individual study. That benefit depends on the studies being comparable, which is why the review stage cannot be skipped.
How to read a forest plot
A forest plot is the standard picture of a meta-analysis. Forest plots display the estimated results from a group of studies plus a summary measure. Read it from the individual studies toward the summary.
- Study rows. Each row is one included study, so the number and variety of rows show how much evidence sits behind the result.
- Intervals. The plot shows the effect size and confidence interval for each study, as an ATLAS.ti methods guide explains. A wider interval reflects a less precise estimate, and studies whose intervals sit far from the others deserve a second look.
- Summary measure. The pooled result sits alongside the individual studies. It is only as trustworthy as the studies feeding it.
- Effect measure. Check the label. Typical effect sizes in systematic reviews are the odds ratio, the risk ratio, the weighted mean difference and the standardized mean difference, and each answers a slightly different question.
Before accepting the summary, compare it against the individual rows. If the studies broadly agree, the pooled figure is a fair shorthand. If they scatter, the average may hide more than it reveals, which leads directly to the question of similarity between studies.
Where reviews sit in the evidence hierarchy
Systematic reviews sit at the top of the evidence hierarchy, according to the library guide, because it is widely agreed that studies with rigorous methods give the most reliable answers. Researchers writing in a peer-reviewed methodology paper note that evidence syntheses are commonly regarded as the foundation of evidence-based medicine.
The top position is a statement about method, not a guarantee. The same paper warns that many systematic reviews are methodologically flawed, biased, redundant or uninformative. A well-run review outranks a single study, while a poorly run one can sit at the “top” in name only. When a story leans on a review’s rank, the useful question is how it was conducted.
Judging whether a review is reliable
Meta-analyses can be reliable, but reliability is earned by the review’s inputs and conduct rather than by the label. Four checks do most of the work.
Heterogeneity: are the studies similar enough?
Heterogeneity describes differences between the studies being pooled. Meta-analysis can be misleading if it is performed with data that are not sufficiently similar or are of poor quality. A review that pools trials of different designs or poorly matched outcomes may produce a tidy average that describes no real patient group. Reviews that explain why the studies were judged comparable are more convincing than those that simply report a number.
Publication bias and incomplete searching
Publication bias arises when the studies that are available to a reviewer are not a fair sample of all the studies that were done, which skews any pooled result built on them. A thorough review reports a complete search across databases, registers and websites, as PRISMA recommends, so readers can judge whether relevant work could have been missed. A thin or poorly described search is a warning sign.
Reporting standards: PRISMA and Cochrane
The PRISMA statement received a major update in 2021, known as PRISMA 2020, per the PRISMA explanatory paper. Its checklist items cover transparent reporting, including the full search strategies noted earlier. A review that states it followed PRISMA is easier to audit, although following a reporting guideline is not the same as good methods.
The Cochrane Library is the other name readers meet most often. A university library guide calls the Cochrane Database the leading resource for systematic reviews in health care and notes that Cochrane Reviews are also called systematic reviews. The ATLAS.ti guide points to the Cochrane Handbook as the place where best-practice steps for meta-analysis are outlined.
Method quality beyond the label
Quality varies even where standards exist. The authors also stress that endorsement of a guidance tool does not substitute for in-depth methodological training. Checklists help a reader ask better questions, but they do not replace judgement.
Testing a news story that cites a meta-analysis
Use these checks in order when a headline says a treatment works because “a meta-analysis found” it.
- Find the review. Confirm that a systematic review, with a described search and selection process, stands behind the pooled figure.
- Check the screening and search. Look for independent screening by at least two people and a fully reported search.
- Ask what was pooled. The headline outcome should match what the included studies measured, which is why it helps to understand primary endpoint vs secondary endpoint before accepting a pooled claim.
- Look at the forest plot. See whether the studies broadly agree and how wide the intervals are.
- Check the interventions. When a review compares products, make sure the products being pooled are truly equivalent; biosimilar vs generic drugs explains why that is not always the case.
A review that passes these checks supports a firmer reading of its result. One that fails several is better treated as a prompt for further reading than as a verdict.
Frequently asked questions
Does every systematic review include a meta-analysis?
No. A systematic review may report its findings in words through narrative synthesis, while a meta-analysis will always be accompanied by a systematic review.
Who coined the term meta-analysis?
The ATLAS.ti methods guide says the term was coined by Gene V. Glass in 1976.
What is PRISMA designed to do?
The guideline offers a baseline list of reporting items aimed mainly at making systematic reviews open and fully described. It does not cover how to design or conduct literature searches.
Why can a pooled average mislead?
Because it combines whatever it is given. If the underlying data are not sufficiently similar or are of poor quality, the average can misrepresent the evidence.
