Choosing a meta-analysis project as a medical student or resident

A systematic review or meta-analysis is one of the few research projects a student or resident can complete without a laboratory, a patient cohort or approval to collect new data. That is why it is popular, and why many questions already have several reviews.

The most common reasons these projects fail are a topic that is too broad, a topic that someone else has already done, and a timeline that ignores how long screening takes. This guide is about avoiding those three.

What makes a good student project

• A narrow question with a clear PICO, so that the search returns hundreds, not tens of thousands, of records.
• A topic your supervisor or a collaborating clinician knows well, so that someone can judge which studies are comparable.
• Enough primary studies to pool, but not so many that screening takes a year. Roughly 5 to 30 included studies is a workable range for a first project.
• A reason to exist: a new trial, a new population, a new outcome, or an update of a review that is several years old.
• A team of at least two reviewers, because screening and data extraction should be done in duplicate.

A realistic timeline

1. Weeks 1 to 3: choose the topic, check that it has not been done, and write the question.
2. Weeks 3 to 6: write the protocol, ideally with a librarian's help on the search strategy, and register it.
3. Weeks 6 to 12: run the searches, remove duplicates, and screen titles and abstracts and then full texts.
4. Weeks 12 to 20: extract data in duplicate, assess risk of bias, and resolve disagreements.
5. Weeks 20 to 26: analyse, grade the certainty of the evidence, and write the manuscript following PRISMA 2020.
6. After that: revise for submission. A first project often takes longer than this, especially when the number of records is large.

Where to look for ideas

• Clinical questions that come up on your ward or in your clinic and that your seniors answer differently.
• The future research and limitations sections of recent reviews and clinical guidelines.
• Recent randomized trials in your specialty: a new trial can change an old review's conclusion.
• Older reviews: check the date of the last search. An update that adds newer studies is a legitimate project, as our guide on updating a review explains.
• A topic-discovery tool such as SynthGap, which proposes candidate topics and checks them against existing reviews. Treat its suggestions as candidates to verify, not final answers.

Mistakes to avoid

• Starting before checking PubMed, the Cochrane Library, PROSPERO and Epistemonikos for existing or registered reviews.
• Choosing a question so broad, such as the effects of exercise on health, that no protocol could cover it.
• Skipping the protocol, then changing the question after seeing results.
• Screening alone.
• Leaving authorship and responsibilities to be agreed at the end.

Questions

Keep reading

General methodological guidance, not medical advice. SynthGap's suggestions are AI-generated from the published literature, so verify them before registering a protocol.