Systematic Review

A review that answers a focused question by following an explicit protocol: a comprehensive, reproducible search; pre-defined inclusion/exclusion criteria; risk-of-bias appraisal of every included study; and a transparent synthesis — reported to the PRISMA standard. It may include a meta-analysis, but need not.

A systematic review assembles every eligible study on a focused question and appraises each for bias. If the studies are similar enough, the review may go on to pool them statistically (the gold diamond) — but the search, selection and risk-of-bias appraisal are the review; the diamond is the optional extra.

Check the protocol was pre-registered (e.g. PROSPERO); the search was comprehensive and included grey literature; selection was done in duplicate against pre-set criteria; and each study’s risk of bias was formally assessed. A slick forest plot built on biased primary studies is still biased.

A systematic review is far less prone to bias than a narrative review, where authors can cherry-pick supportive studies. Its reproducible method sits near the top of the evidence hierarchy — but only when the included evidence is sound.

• Focused question + explicit protocol (pre-registered)
• Comprehensive search · duplicate selection · risk-of-bias appraisal
• Reported to PRISMA; meta-analysis is optional

A Cochrane systematic review is the archetype — e.g. Perel et al., “Colloids versus crystalloids for fluid resuscitation in critically ill people” (Cochrane Database Syst Rev). It states a focused question, registers a protocol, searches multiple databases and trial registers, screens in duplicate, appraises each trial’s risk of bias, and then synthesises — concluding that resuscitation with colloids rather than crystalloids does not reduce, and may increase, the risk of death. A systematic review is only as trustworthy as the trials it includes — Cochrane’s structure controls reviewer bias, not the bias baked into the primary studies.

• Conflating “systematic review” with “meta-analysis” — the pooling step is optional and need not be present.
• Missing grey literature / publication bias — an incomplete search inflates the effect.
• Ignoring heterogeneity — pooling clinically dissimilar studies produces a misleading summary.