Systematic Review vs Meta-Analysis: What's the Difference?

Ask ten researchers whether their project is “a systematic review” or “a meta-analysis” and several will answer as if the two phrases mean the same thing. They don’t, and the difference isn’t pedantic — it determines what methods section you write, what your protocol commits you to, and whether a single pooled number is even an honest way to summarize what you found.

What a systematic review actually is

A systematic review is a structured, explicit, reproducible process for answering a research question using the totality of relevant evidence — not a hand-picked set of studies that happen to support a point. The Cochrane Handbook and the PRISMA 2020 statement both define it by its process: a pre-registered protocol, a comprehensive and documented search strategy, dual independent screening against stated eligibility criteria, structured data extraction, a risk-of-bias assessment for every included study, and a synthesis of findings — reported transparently enough that another team could follow the same steps and check the result.

Nothing in that definition requires pooling numbers into one estimate. The synthesis step can be narrative, tabular, or statistical — which one is appropriate depends entirely on the studies you actually find.

What a meta-analysis actually is

A meta-analysis is a statistical method: it takes an effect estimate and its uncertainty (a standard error) from each study and combines them into one pooled effect, typically weighted by each study’s precision (inverse-variance weighting), using a fixed-effect or random-effects model. The output is a single number with a confidence interval — usually displayed as the diamond at the bottom of a forest plot — plus a heterogeneity statistic (I², Cochran’s Q, τ²) describing how much the individual studies actually agreed with each other.

A meta-analysis needs comparable numbers from multiple studies. It does not, by itself, require a systematic search, documented screening, or risk-of-bias assessment — those are what a systematic review contributes to a meta-analysis, not something the statistical technique provides on its own.

How they actually relate

In practice: most published meta-analyses sit inside a systematic review, because that’s what justifies treating the pooled studies as a fair summary of the evidence rather than a cherry-picked subset. But the reverse isn’t true — plenty of rigorous systematic reviews contain no meta-analysis at all, because pooling would have been misleading.

When a systematic review should not include a meta-analysis

Pooling is a modelling choice, not an automatic next step once studies are included. The Cochrane Handbook is explicit that studies should only be combined when they are similar enough — in population, intervention, comparator, and outcome definition — that a single number is a meaningful answer to one question. Common reasons to skip pooling:

• Included studies measure genuinely different populations, interventions, or outcome definitions that a single pooled number would misrepresent.
• Heterogeneity is high and not explained by pre-specified subgroups, methodology, or risk of bias.
• Too few studies report compatible outcome data to compute a stable estimate.
• Most included studies carry a high risk of bias, so a pooled estimate would be falsely precise.

When that’s the case, the right move isn’t to force a pooled estimate — it’s to report a structured narrative synthesis instead. The SWiM (Synthesis without Meta-analysis) reporting guideline exists specifically for this: it sets out how to synthesize and report findings transparently without a forest plot at the end.

Deciding which one your project needs

Decide this during protocol-writing, before screening starts — not after you’ve seen which studies came back.

• You need a systematic review whenever the goal is a comprehensive, reproducible answer to a research question - regardless of whether the studies turn out to be poolable.
• You need the meta-analysis component specifically when you want one combined effect estimate, and only once you’ve confirmed the included studies are clinically and statistically similar enough to justify it.
• Plan for both, commit to neither in advance - a protocol should state the intent to pool if studies are sufficiently similar, with narrative synthesis as the pre-specified fallback, rather than assuming pooling will work out.

References

Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi.org/10.1136/bmj.n71. Cochrane Handbook for Systematic Reviews of Interventions (current version). training.cochrane.org/handbook. Campbell M, McKenzie JE, Sowden A, et al. Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline. BMJ. 2020;368:l6890. doi.org/10.1136/bmj.l6890.

Continue your review workflow

• Not sure a systematic review is even the right design? Check scoping review vs systematic review first.
• Start with a free protocol template and decide your synthesis plan - pooled or narrative - before screening begins.
• Run title and abstract screening against your eligibility criteria, then extract included studies with the data extraction template.
• Check heterogeneity before you pool anything, using the free I² calculator, or read the full heterogeneity and effect-size guide.
• Document the whole process with a PRISMA 2020 flow diagram.
• Start a free EvidenceFlow review to keep protocol, screening, extraction, and analysis on one project - pooling the data only when it's actually appropriate to.