[Durgical](https://durgical.com/en/) › [DurgiPedia](https://durgical.com/en/durgipedia/) › [Epidemiology](https://durgical.com/en/epidemiology/) › [Evidence-based medicine](https://durgical.com/en/epidemiology/areas/ebm/) › Systematic review

# Systematic review

- Study design

A systematic review gathers all the studies on a question with explicit, reproducible methods to reduce bias.

Updated Nov 5, 2026

## Key points

- A prior protocol, ideally registered: question, eligibility criteria and planned analyses set before seeing the data.
- An exhaustive search of several databases, with no restriction by language or publication status when possible.
- Selection and data extraction by two independent reviewers, with a record of every decision.
- Risk of bias of each study assessed with a tool chosen in advance.
- Narrative or quantitative synthesis (meta-analysis); reported with PRISMA and its flow diagram.

On this page

- The basics Key points
- What it is
- How it is built
- Worked example
- Systematic review versus narrative review
- Markers
- When to use it
- Common pitfalls
- Memory aid

- Study Test yourself

- Connections Connections

## What it is

A systematic review answers a defined question by gathering all the evidence that meets eligibility criteria set in advance, with explicit methods to search, select, assess risk of bias and synthesise the studies. It starts from a protocol written before the results are known. When the studies are similar enough, the synthesis can include a meta-analysis.

## How it is built

- PICO question and protocol.
- Search of databases, trial registries and grey literature.
- Two-stage selection: title and abstract, then full text.
- Data extraction and risk-of-bias assessment.
- Synthesis, certainty rating (GRADE) and conclusions.

## Worked example

- Data (made up): a review looks for trials of an exercise programme for chronic low back pain.
- Search of 3 databases: 1 850 records; 450 duplicates are removed: 1 850 − 450 = 1 400.
- Title and abstract: 1 340 are excluded; 1 400 − 1 340 = 60 remain for full-text reading.
- Full text: 48 are excluded (other population, no control group); 60 − 48 = 12 trials are included.
- Two independent reviewers agree on 55 of the 60 texts (≈ 92 %); the 5 disagreements are settled by a third reviewer.
- Risk of bias: 4 of the 12 trials (33 %) are at high risk for lack of allocation concealment.
- Reading: the review starts from 1 850 records and keeps 12 trials; each number goes into the PRISMA flow diagram.

## Systematic review versus narrative review

| Feature | Systematic review | Narrative review |
| --- | --- | --- |
| Question | Specific (PICO) | Broad, a topic |
| Search | Exhaustive and reproducible | Unstated, at the author's discretion |
| Selection | Explicit criteria, two reviewers | Implicit |
| Study quality | Risk of bias assessed | Rarely assessed |
| Synthesis | Systematic, sometimes meta-analysis | Qualitative summary |
| Risk of bias of the review | Low | High |

## Markers

- SR
- Protocol
- PRISMA
- Risk of bias

## When to use it

- It is the starting point of clinical practice guidelines and health technology assessments.
- Before reading 20 separate articles on a question, check whether a recent systematic review already exists.

## Common pitfalls

- Searching one database only, or in English only: studies are missed and bias creeps in.
- Changing the criteria after seeing the results without saying so.
- Adding up poor-quality studies without assessing their risk of bias: garbage in, garbage out.

## Memory aid

Plan, Fish, Filter, Judge, Synthesise: five steps a narrative review skips.

## Test yourself

Think of the answer, then tap the question to see it.

## What sets a systematic review apart from a narrative review?

A specific question, a prior protocol, an exhaustive and reproducible search, explicit selection criteria and assessment of risk of bias.

## Why is the protocol written before the results are known?

So that criteria and analyses are not fitted to what is found; it limits bias and lets others check the plan.

## Does a systematic review always include a meta-analysis?

No: only when the studies are similar enough; otherwise the synthesis is narrative.

## Which guideline is used to report a systematic review?

PRISMA, with its checklist and its flow diagram of study selection.

[Add to my cards (4)](https://durgical.app/estudiar/?add=/en/epidemiology/systematic-review/)

## Connections

## Related 6

- [Meta-analysis](https://durgical.com/en/epidemiology/meta-analysis/)
- [Forest plot](https://durgical.com/en/epidemiology/forest-plot/)
- [Publication bias and the funnel plot](https://durgical.com/en/epidemiology/publication-bias-and-the-funnel-plot/)
- [PICO question](https://durgical.com/en/epidemiology/pico-question/)
- [Reporting guidelines: CONSORT, STROBE, PRISMA, STARD](https://durgical.com/en/epidemiology/reporting-guidelines-consort-strobe-prisma-stard/)
- [The GRADE system](https://durgical.com/en/epidemiology/the-grade-system/)

[← All of epidemiology](https://durgical.com/en/epidemiology/)

Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.

Sources: [Chapter 1: Starting a review](https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-01) (Cochrane, 2026); [The PRISMA 2020 statement: An updated guideline for reporting systematic reviews](https://www.equator-network.org/reporting-guidelines/prisma/) (EQUATOR Network, 2021).

## **Spotted a mistake?** We are always improving for the community. Tell us what is wrong and our team will review it. Report a mistake

Representative model, for educational purposes only.

---
Source: https://durgical.com/en/epidemiology/systematic-review/
License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Cite as: DurgiPedia (durgical.com), https://durgical.com/en/epidemiology/systematic-review/