Can you do a meta-analysis alone? The team and roles you need

You can do all of the work alone, but you cannot do all of the checking alone. A meta-analysis involves dozens of small decisions: is this study eligible, which number is the right one, how biased is this trial. Standards ask for two people to make the important ones independently, because one person's misreading or preference is hard to see from the inside.

This guide sets out the roles in a typical team, what the standards require and what to do if you are a student or resident with nobody else to share the work.

The roles

One person can hold more than one role, and a small team can cover all of them.

Content expert: knows the clinical question, helps to define the PICO and judges whether studies are comparable.
Lead reviewer: runs the project, keeps the protocol and the record of decisions.
Second reviewer: screens and extracts independently, so that errors and disagreements show up.
Information specialist: builds and checks the search strategy and runs it. Published reviews with a librarian as co-author have been found to report better search strategies.
Statistician: advises on the analysis plan, and is most useful at the protocol stage rather than at the end.
Arbiter: a third person who settles the disagreements the two reviewers cannot.

What the standards say

For Cochrane intervention reviews, the MECIR standards make it mandatory for at least two people to work independently when deciding whether studies meet the eligibility criteria and when extracting outcome data. Two independent screeners at the title and abstract stage are desirable but not mandatory, and duplicate extraction of study characteristics is highly desirable.

That rule does not bind every review, but it shows the reasoning: duplicating a decision reduces mistakes and the effect of one person's biases. Journals that follow PRISMA 2020 expect you to say how many reviewers carried out each step and whether they worked independently (items 8, 9 and 11).

Let SynthGap run the first pass

Enter an area. SynthGap proposes candidate topics, checks each against existing systematic reviews, and lists the primary studies behind it. Start on the Free plan: no card needed, and every run shows every topic it finds.

If you are working alone

Find a second person for at least the key decisions: a classmate, a supervisor or a colleague who can screen a sample and check your extraction of the numbers that enter the analysis.
Double-screen a sample, for example the first few hundred records or a random fraction, and measure how often you agree. It shows how reliable the rest of your screening is likely to be.
Say plainly in the methods which steps were done by one person. A single-reviewer review is a limitation to declare, not to hide.
If you use an AI tool as a second reviewer, validate it, keep a person in charge and report it. See AI screening for systematic reviews.
Get methods and statistics advice at the protocol stage, when a change is cheap.

Who counts as an author

Authorship follows the ICMJE criteria: a substantial contribution to the work, drafting or revising it, approving the final version and agreeing to be accountable for it. Everyone who meets all four should be an author, and others who contributed should be acknowledged. Agree this at the start and write down who does what.

Sources

The figures and rules on this page come from these sources.

Cochrane. Methodological Expectations of Cochrane Intervention Reviews (MECIR), standards C39 (selecting studies), C45 and C46 (extracting data).
Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71.
Rethlefsen ML, Farrell AM, et al. Librarian co-authors correlated with higher quality reported search strategies in general internal medicine systematic reviews. Journal of Clinical Epidemiology 2015;68:617-626.
International Committee of Medical Journal Editors. Defining the role of authors and contributors. Recommendations, section II.A.
Borah R, Brown AW, Capers PL, Kaiser KA. Analysis of the time and workers needed to conduct systematic reviews of medical interventions using data from the PROSPERO registry. BMJ Open 2017;7:e012545.

Questions

Do I need a statistician?

You can often manage a straightforward pairwise meta-analysis with the software and the Cochrane Handbook, but ask for statistical advice when you write the protocol, especially if you plan subgroup analyses, meta-regression or an unusual design.

Can my supervisor be the second reviewer?

Yes, if they will actually screen and extract independently and not just approve your decisions. Independence means they make their own judgement before they see yours.

How many people should be in the team?

In an analysis of 195 completed reviews registered on PROSPERO the average was about five authors. Smaller teams can do the work, but each person then carries more of it, and the checking needs planning.

Does the librarian need to be an author?

If they meet the ICMJE criteria, yes. If they only run a search to your specification, an acknowledgement may be the right credit. Decide at the start and put it in writing.

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.