Skip to Main Content

Systematic review

A step by step guide to doing a systematic review or other forms of Evidence Synthesis

Evidence synthesis review types

This page describes the most common types of review done at Imperial and presents a table comparing aims and features of Narrative, Systematic, Scoping, Rapid and Umbrella reviews.

Main review types

Differences between review types

 

Narrative

literature review

Systematic review

Scoping Review

Rapid Review

Umbrella Review

Time and resources

Low, can be relatively quick to completeHigh, can take 6-18 monthsModerate to High, 3-18 monthsLow to Moderate, typically a few weeks to 5 monthsModerate to High, 2-8 months 

Aim/output

To provide a broad, descriptive summary of existing literature on a topic, often to contextualise an issue or support an argument. Output is a narrative summary, not a formal synthesis.To answer a narrow, pre-defined research question by identifying, appraising and synthesising all available evidence. Output is a comprehensive, reproducible synthesis (with meta-analysis where data allow) to inform practice, policy or clinical decisions.To map the breadth and nature of evidence on a topic, clarify key concepts and identify gaps in the literature. Output is a descriptive overview/map of the evidence base, often used to determine feasibility of a future systematic review.To answer a focused question using systematic review methods, but with abbreviated or streamlined steps to produce evidence quickly for time-sensitive decisions (e.g. emerging health threats, urgent policy questions).To summarise findings across multiple existing systematic reviews/meta-analyses on related questions. Output is a high-level synthesis of the review-level evidence, useful for comparing interventions or getting a broad picture quickly.

Best suited for

Introductions, background sections, or informing readers broadly on a subject where methodological rigour is not the primary goal.Questions needing definitive, high-certainty evidence to change practice, policy or guidelines.Exploratory questions, emerging topics, or scoping the literature before committing to a full systematic review.Situations needing robust evidence under real time or resource constraints. Comparing multiple interventions/outcomes, or getting a rapid, broad overview when several systematic reviews already exist on the topic.

Research question

Broad or narrow topic e.g. What are the main causes of burnout among US healthcare workers?Narrow and focused e.g. What is the impact of anti-amyloid drugs on adults with mild cognitive impairment or mild dementia due to Alzheimer's disease?Broad or narrow, often exploratory e.g. What QoL questionnaires are available for paediatric patients following tonsillectomies with or without adenoidectomies for chronic infections or SDB?Similar to a systematic review, but scoped to be answerable within a compressed timeframe (weeks rather than months).Broad, comparative e.g. What do systematic reviews tell us about the comparative effectiveness of interventions for chronic lower back pain?

A priori protocol

Not required Explicit, formal protocol requiredExplicit, formal protocol required

Recommended but often abbreviated; may not always be registered given time constraints

 

Formal protocol recommended 

Reviewers 

Single or multiple reviewersMultiple reviewers required to reduce error and biasMultiple reviewers required to reduce error and bias

Multiple reviewers preferred, though some steps (e.g. screening) may be done by a single reviewer with partial verification to save time

 

Multiple reviewers required to reduce error and bias

Scope of literature

Selective, not exhaustive. May only use one database or search toolFocused and exhaustive within a defined criteria. Multiple databases (and potentially grey literature sources) are usedFocused, can be exhaustive or aim for diversity. Multiple databases (and potentially grey literature sources) are used

Focused, but limited. Fewer databases and sources searched than a full systematic review to save time

 

Focused and exhaustive within a defined criteria. Multiple databases (and potentially grey literature sources) are used

Search strategy

Not formal or systematic. No specified or reported strategyExplicit, systematic, and transparent search strategy for each database, reported in reviewExplicit, systematic, and transparent search strategy for each database, reported in review

Explicit, transparent but streamlined. May limit date ranges, languages, or number of databases; still reported in review

 

Explicit, systematic, and transparent search strategy for each database, reported in review

Study selection criteria

Selection criteria is not formally defined. Studies can be "cherry picked" to support the author's argumentStrict and defined a priori. All studies that fit the inclusion criteria are selected and are reviewed at title abstract and full-text level by more than one reviewerStrict or more flexible, defined a priori. All studies that fit the inclusion criteria are selected and are reviewed at title abstract and full-text level by more than one reviewer

Strict and defined a priori, but screening may be accelerated (e.g. single-reviewer screening)

 

Strict and defined a priori. Only systematic reviews/meta-analyses meeting inclusion criteria are selected

Quality appraisal

No formal quality assessment of evidenceFormal and systematic critical appraisal using defined criteria (e.g. Cochrane Risk of Bias assessment)Not mandatory. Reviewers may decide to assess quality depending on the purpose of the review

Often included but abbreviated; may use a simplified checklist rather than a full formal appraisal

 

Formal appraisal of included reviews' quality (e.g. AMSTAR-2)

Data extraction

No standardised data extraction forms. Data may not be extractedStandardised data extraction formsStandardised data extraction forms

Standardised data extraction forms, though the set of extracted variables may be reduced

 

Standardised data extraction forms

Synthesis method

Narrative synthesis, not suitable for meta-analysisQuantitative or qualitative. Meta-analysis if data is homogenousQuantitative or qualitative, not suitable for meta-analysis

Narrative or quantitative synthesis, depending on time available; formal meta-analysis less common

 

Narrative or quantitative synthesis comparing results across included reviews

Reporting

No reporting guidelines mandatedMust report review methods via PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)Must report review methods via PRISMA-ScR guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews)

Should report via PRISMA guidelines, with a note on which steps were abbreviated and why

 

Should report via PRISMA guidelines, adapted for review-of-reviews reporting

Potential for bias

High: selective, non-exhaustive searching and no formal quality appraisal. Low: formal protocol, exhaustive multi-database search, dual reviewers and mandatory quality appraisal (see Quality Appraisal row) are designed to minimise selection, reporting and reviewer bias.Moderate: exhaustive search and dual reviewers reduce selection bias, but because quality appraisal is not mandatory (see Quality Appraisal row), risk of bias within included studies may go unassessed.

Moderate: streamlined searching and screening trade rigour for speed, risking missed studies or reviewer errors.

 

Depends on the quality of included reviews - appraisal of review quality (see Quality Appraisal row) helps flag this, but findings can inherit bias present in the underlying systematic review.

 

Last Updated: Nov 17, 2026 11:58 AM