What is a systematic review?

A systematic review answers a clearly defined question by finding, appraising, and bringing together all the studies relevant to it — using methods that are explicit, planned in advance, and reproducible. That last word is the point: another team following your stated methods should be able to arrive at the same set of studies and the same conclusion.

How it differs from an ordinary literature review

A traditional narrative review reflects what its author happened to read and found convincing. A systematic review removes that discretion. Before looking at any results, you fix the question, the criteria for including a study, the databases to search, and how each study will be assessed. Because the process is set down first and reported in full, the reader can see exactly how the conclusion was reached — and where it might be fragile.

The main stages

1. Ask a clear question. A well-scoped question usually names the population, the intervention or exposure, any comparison, and the outcome — the PICO frame is the common shorthand.

2. Write a protocol. Set out the methods in advance and, ideally, register them publicly (for example on PROSPERO) so the plan is on record before results can influence it.

3. Search systematically. Search several databases with a documented strategy, so the search can be rerun and checked rather than taken on trust.

4. Screen against your criteria. Two reviewers independently screen titles, abstracts, and then full texts, resolving disagreements by discussion — a guard against one person's blind spots.

5. Appraise the studies. Assess each included study for risk of bias with a recognised tool, so the quality of the evidence is weighed, not just its quantity.

6. Synthesise and report. Bring the findings together — narratively, or with a meta-analysis where the data allow — and report the whole process transparently, following the PRISMA guideline.

When a meta-analysis fits

A meta-analysis is the statistical step that combines results across studies into a single estimate. It is part of some systematic reviews, not all of them: it is only appropriate when the studies are similar enough in question, design, and outcome that pooling them is meaningful. When they are not, combining them produces a precise-looking number that means very little. A good review knows when not to pool.

Why it matters

Done well, a systematic review is one of the most reliable ways to turn a scattered, sometimes contradictory literature into something a clinician, a policymaker, or a funder can act on. Done carelessly, it lends false authority to a weak conclusion. The methods exist to keep the first from becoming the second — and they can be learned.