What is the difference between a literature review and a systematic review?
Whether the method is specified in advance and reproducible. A literature review surveys what has been written; a systematic review follows a pre-registered protocol designed to find and appraise all relevant evidence on a specific question.
Literature review (narrative review). A survey of published work on a topic, selected by the author's judgement.
Its strengths: breadth, context, synthesis across a field, and the ability to trace the development of ideas. Well suited to introducing a subject or framing a debate.
Its weakness: the selection is not transparent. The author chose which papers to include, by unstated criteria, and a reader cannot tell what was omitted or why. Confirmation bias operates freely, and two authors reviewing the same field can reach opposite conclusions while both being honest.
Systematic review. Addresses a specific, answerable question using a method fixed in advance:
A protocol is written and registered — frequently on PROSPERO — before the search begins, specifying the question, inclusion and exclusion criteria, databases, and analysis plan. This prevents adjusting criteria after seeing results.
A comprehensive search across multiple databases, with the full strategy published so it can be repeated.
Screening by two independent reviewers, with disagreements resolved by a third.
Risk of bias assessment for each included study, using validated tools.
Synthesis, narrative or statistical.
Full reporting of what was found, excluded and why, conventionally with a flow diagram, following reporting guidelines such as PRISMA.
Meta-analysis is the statistical combination of results across studies, producing a pooled estimate with greater precision than any single study. It is not the same as a systematic review — a systematic review may or may not include one, and a meta-analysis without systematic searching is of limited value.
Why systematic reviews sit high in evidence hierarchies: they reduce the influence of individual study selection, and they can detect publication bias — the tendency for positive findings to be published and null ones not.
Their limitations: they are only as good as the studies included; heterogeneous studies may not be combinable; and they take a long time, so they can be out of date on publication.
Scoping and rapid reviews occupy intermediate positions.