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

What Is a Systematic Review?

A systematic review is a research method that identifies, evaluates, and synthesizes all available evidence on a specific question using predefined, documented, and reproducible methods. It follows a registered protocol, searches multiple databases systematically, screens and selects studies against set eligibility criteria, assesses the risk of bias in each included study, and brings the findings together — either narratively or statistically. For a full walkthrough of each of these steps, see our companion guide, What Is a Systematic Review?

What Is a Meta-Analysis?

A meta-analysis is a statistical technique that combines the quantitative results of multiple studies addressing the same question into a single pooled estimate of effect, along with a measure of how much the individual studies agree with each other. It increases statistical power beyond what any one included study could achieve alone, and can reveal a more precise estimate — or a real difference between subgroups — that individual studies were each too small to detect on their own.

The Key Difference

A systematic review is a research process: a method for exhaustively finding and evaluating evidence. A meta-analysis is a statistical output: a calculation performed on a subset of that evidence, when the numbers allow it. Every credible meta-analysis is built on a systematic review's search and selection process — but not every systematic review produces, or is meant to produce, a meta-analysis.

Systematic Review vs. Meta-Analysis at a Glance

When a Systematic Review Does Not Include a Meta-Analysis

A systematic review skips meta-analysis when the included studies are too different to combine meaningfully — different populations, interventions, outcome measures, or study designs — or when there are too few comparable studies to pool. In these cases, the review reports its findings narratively instead: describing patterns, agreements, and contradictions across studies in prose, sometimes supported by structured summary tables, without calculating a single pooled number. This is a normal, expected outcome, not a lesser one — guidance such as Synthesis Without Meta-analysis (SWiM) sets out how to report a narrative synthesis with the same rigor expected of a meta-analysis.

When Meta-Analysis Is Appropriate

Meta-analysis is appropriate when the included studies are similar enough — clinically (comparable populations, interventions, and outcomes), methodologically (comparable designs and risk-of-bias profiles), and statistically (low enough heterogeneity, often assessed with the I² statistic) — that pooling them produces a meaningful number rather than an average of genuinely different things. Reviewers also choose between a fixed-effect model, which assumes all studies estimate one true, identical effect, and a random-effects model, which allows the true effect to vary across studies — the more common and typically more conservative choice when between-study differences are expected.

Narrative Synthesis

A narrative synthesis brings together the findings of included studies in prose rather than through a pooled statistic — describing what studies found, where they agree, where they conflict, and what might explain the differences. It's not a lower standard of evidence than a meta-analysis; it's the more honest choice when the underlying studies genuinely aren't similar enough to average together. A well-reported narrative synthesis still follows a structured, pre-planned approach, rather than an ad hoc summary of whatever stood out to the reviewer.

Heterogeneity and Why Pooling May Be Inappropriate

Heterogeneity is the variation in results between the included studies that goes beyond what would be expected from chance alone. It's often estimated with the I² statistic, which describes what proportion of the observed variation reflects real differences between studies rather than random error. When heterogeneity is high and can't be explained — for example, through subgroup analysis — pooling the studies into a single number can be misleading: the resulting average may not represent any of the actual studies well, and can obscure a real difference between subgroups rather than revealing it. In that situation, many methodologists recommend a narrative synthesis, or a meta-analysis restricted to a more homogeneous subset of the studies, over forcing a pooled estimate.

Common Misconceptions

A few misunderstandings come up often enough to be worth naming directly. "Systematic review" and "meta-analysis" are not interchangeable terms — a systematic review is the review process, a meta-analysis is one possible statistical output of it. A systematic review is not automatically weaker evidence if it doesn't include a meta-analysis — for genuinely heterogeneous studies, skipping pooling is the methodologically correct choice, not a shortcut. And a meta-analysis is not a shortcut around doing a systematic review first — pooling studies that were never systematically identified and appraised is a well-documented source of misleading conclusions in the methodological literature.

Worked Example

The example below is a hypothetical illustration, not a real Tezyar project or a published study.

Imagine a systematic review asks: does structured nurse-led discharge education reduce 30-day hospital readmission in adults with type 2 diabetes? After a systematic search and screening process, the review includes 12 studies. Of these, 7 use a comparable randomized design, measure readmission the same way, and report enough data to pool — so the reviewers run a meta-analysis on those 7, producing a single pooled estimate of effect. The other 5 studies use varied designs and outcome definitions that don't fit the same statistical model, so their findings are reported narratively alongside the pooled result, rather than forced into the same calculation. The systematic review itself still covers all 12 studies; the meta-analysis is a component within it, applied only where it was statistically appropriate.