The PRISMA 2020 flow

A systematic-review flow diagram that counts your own library, and tells you which numbers it worked out and which are yours.

PRISMA is the reporting standard for systematic reviews, and its flow diagram is the figure that shows how you got from everything you found to everything you kept. Quester builds it from your own library rather than from a blank template, so the counts start out true and you correct them rather than compiling them by hand.

The one thing to understand before you use it is that the figure has two kinds of box. Most are derived: Quester works them out from your sources and your questions, they are drawn with a solid border, and pressing Refresh counts rewrites them. Two are yours: nothing in Quester can know how many duplicates you removed or how many reports you failed to retrieve, so those boxes are drawn with a dashed border, you type into them directly, and nothing ever overwrites what you put there. The exclusion reasons you write are yours in the same way. The toolbar carries a legend saying exactly this while you work.

Where the numbers come from

Three counts are read from your library, and everything else is arithmetic on top of them.

Records identified is every source in the project. Records excluded at screening is the ones you have marked Archived. In Quester, a source’s read status is its screening decision, and Archived is the status that means screened out. Studies included is the ones connected to a live research question, because connecting a source to a question is the act of actually using it in the review.

From those three, the chain runs as the standard requires: records screened is the identified total less anything you removed before screening; reports sought is what survived screening; reports assessed is what you managed to retrieve; and reports excluded at eligibility is whatever was assessed but did not end up connected to a question. That last number is the one reviewers most often have to explain, which is why the box comes seeded with three blank reason lines.

Two details are worth knowing because they are decisions rather than accidents. A source that is Archived does not count as included even if a question still points at it, since you screened it out. And a question that has itself been archived does not pull anything into the review.

The boxes you fill in yourself

Four counts have no answer anywhere in Quester, and they live in the two dashed boxes.

Records removed before screening holds three of them: duplicate records, records a tool marked ineligible, and records removed for other reasons. Reports not retrieved holds the fourth. Type the numbers straight into the box, in the lines already there, keeping the (n = …) shape; that is what the figure reads back when it recalculates. These are the only numbers in the diagram stored nowhere but the box itself, so treat the text as the record.

Registers (n = 0) in the first box is also yours. Quester does not distinguish a register from a database, so if some of your records came from one, split the total by hand.

Refreshing

Refresh counts re-reads your library and redraws every derived box. Use it whenever the library has moved on: sources added, something newly archived, or a question that gained or lost a source. The figure will catch up, including the date in the caption.

It is safe. Refresh recalculates the derived boxes and leaves the two dashed ones exactly as you typed them, so your duplicate count survives. In the exclusions box it replaces only the number in the header and keeps every reason you wrote below it. It also reads your typed numbers back in before recalculating, which is why the totals stay consistent instead of drifting apart from what you entered.

The sources behind each number

Four boxes count a real set of sources rather than a difference, and each of those shows how many it is holding underneath its text. Click that count and a panel lists the studies themselves, with the year, type and read status of each; click any card to open the source. It works on records identified, records excluded at screening, studies included, and reports excluded at eligibility.

This is the difference between a figure you assert and a figure you can defend. When someone asks which twelve studies you excluded at eligibility, the answer is two clicks away rather than a reconstruction.

Editing, layout, and export

Every box is editable text, so you can rewrite any wording the standard requires you to adapt. Boxes can be resized when a reason list outgrows one, and the whole figure can be rearranged like anything else on the canvas, though the layout is the published PRISMA arrangement, and journals expect that shape, so there is rarely a reason to move things.

Quester draws the “databases and registers” version of the diagram, and deliberately leaves out its second identification column for records found by other means, such as citation searching, contacting organisations, or web sources. Nothing in a library distinguishes those, and a column of zeros you have to delete is worse than one you add yourself if your review needs it.

Export the finished figure with Download image from the toolbar, and save named versions as you go from History. The counts move as the library moves, so a snapshot taken at submission is worth keeping.

Next: References.