Five Review Types, One Platform, Choosing the Methodology
Pull-quote: “The most expensive mistake in evidence synthesis is not a bad search string. It is running the wrong kind of review for the question.”
The question determines the method
Evidence teams tend to have two default speeds: the full systematic review, chosen for its authority, and the deadline scramble, chosen by the calendar. Both defaults skip the decision that matters most. Review methodology is a contract about what the reader can conclude, and the right contract depends on the question, the state of the literature, and the decision the review must serve. There are five contracts worth knowing, and they are not interchangeable.
| Review type | The question it answers | Signature output |
|---|---|---|
| Full systematic review | What does all the evidence say about one focused question | PRISMA 2020 package with full synthesis |
| Rapid review | What does the evidence say, in time for this decision | Streamlined synthesis with shortcuts documented |
| Scoping review | What evidence exists here, and what does it look like | Evidence charting and gap identification |
| Umbrella review | What do the existing systematic reviews collectively show | Overview across reviews, with their quality assessed |
| Mapping review | Where is the evidence dense, and where is it missing | Categorized evidence map for planning |
A decision path
how well is the question defined?
│
├─ field unfamiliar or heterogeneous ────► scoping review first,
│ then decide what comes next
├─ focused clinical question
│ │
│ ├─ credible systematic reviews
│ │ already published ──────────────► umbrella review
│ │
│ ├─ decision needed in weeks ───────► rapid review,
│ │ every shortcut in writing
│ └─ full rigor required ────────────► full systematic review
│
└─ planning a portfolio or program ──────► mapping reviewWhat each choice costs
A full systematic review buys exhaustiveness and pays in time: the timeline is measured in months, sometimes past a year. A rapid review buys the calendar back by narrowing sources, dates, or languages, and it is only defensible when each narrowing is pre-specified and reported. A scoping review buys orientation but cannot answer effectiveness questions, and dressing its output in effectiveness language is a methodological foul. An umbrella review inherits the quality and the age of the reviews beneath it. A mapping review buys a planning instrument, not a conclusion.
What changes, and what never does
Across the five types, the variables are search exhaustiveness, screening depth, and synthesis ambition. The constants are non-negotiable: a protocol before the first search, deduplication into one screening queue, dual review on inclusion decisions, and a flow diagram that reconciles from records identified to studies included.
This is the reason the five methodologies work best as five workflows on shared infrastructure rather than five separate tools. Parallel search across the major biomedical databases, a deduplicated screening queue, dual-reviewer enforcement, and an append-only audit trail stay constant across all of them. What the methodology changes is scope and depth, recorded in the protocol. A rapid review on this footing is rapid because its shortcuts are explicit, not because rigor was quietly dropped where nobody would look.
The failure modes
Three patterns account for most methodology regret. The default SLR: six months of exhaustive synthesis delivered three months after the decision it was meant to inform. The undocumented rapid review: fast, useful, and indefensible the first time anyone asks what was excluded. The scoping review with conclusions: a map of the literature quietly presented as if it were a verdict on the intervention.
The fix for all three is the same piece of paper. A protocol that names the review type, states the question in a structured frame, and lists every planned deviation from full systematic method turns the methodology choice into something a reader can audit. A rapid review whose shortcuts were written down before screening began is a different document from one explained after the fact.
Closing
Choose the methodology by the question and the decision date, write the choice into the protocol, and let shared infrastructure hold the constants that make any of the five defensible: one queue, dual review, a reconciling flow diagram, and an audit trail. The review type is a contract with the reader. Sign the right one.