systematic review

Cochrane defines a systematic review as an attempt to collate all the empirical evidence that fits pre-specified eligibility criteria in order to answer a specific research question, using explicit and systematic methods selected with a view to minimising bias. Each part of that sentence is doing work. The question comes before the reading. The rules for what counts come before the search. The search itself has to be documented well enough that another team could run it and arrive at the same set of papers.

That is what separates it from an ordinary literature review, where an author gathers the studies they know of and describes them. The difference is not thoroughness but accountability: a systematic review makes every judgement visible, including the studies it excluded and why. Reviewers are expected to make every effort to find all the relevant research, to account for bias within the studies they include, and to reach their conclusions from the whole identified body of work rather than from the part that agrees.

A meta-analysis is not the same thing. It is the statistical step of combining results into a single pooled estimate, and it is one of several methods a systematic review might use to synthesise what it found. A systematic review can be published without one, where the studies are too different to pool honestly. The protocol matters for the same reason the pre-specified question does: registering the plan before the results are known records the judgements in advance, allows the methods to be reviewed on their merits, and prevents decisions made after seeing the data from shaping what the review concludes.