Create a PRISMA 2020 flow by tracking counts at each step—identify, deduplicate, screen, assess, include—and fill the template with those numbers.
A PRISMA flow diagram shows the path your records take from search to inclusion. Readers see how wide you cast the net, how many records you removed as repeats, and why some reports did not make the cut. The chart looks simple. The work behind it is not. This guide walks you through the steps with plain language and a clean workflow.
PRISMA was built for systematic reviews, yet many literature reviews still borrow the same flow to be transparent. The 2020 update split the diagram into two routes: records from databases and registers, and records found by other routes such as citation chasing or websites. Picking the right template and logging numbers in real time keeps the chart honest and easy to finish.
Doing A PRISMA Flow Diagram For Literature Reviews: Quick Overview
You will gather records, remove repeats, screen titles and abstracts, retrieve full texts, assess eligibility, and then list what you included. That sequence becomes the backbone of the figure. Keep a tracking sheet open from the first search so you never guess later.
Before you start, download the official PRISMA 2020 flow diagram template that matches your plan (databases and registers only, or databases plus other methods). The template sits on the PRISMA site and mirrors the boxes you will fill.
| Box | What To Record | Where Numbers Come From |
|---|---|---|
| Records Identified | All hits from each source before filters or de-duplication | Database export totals; register counts; website logs |
| Duplicates Removed | Exact and near-duplicate records removed once | Reference manager or de-duplication tool logs |
| Records Screened | Titles/abstracts you actually screened | Screening tool or spreadsheet tick marks |
| Records Excluded | Titles/abstracts that did not pass screening | Screening notes; reason tags optional here |
| Reports Sought | Full texts you tried to retrieve | Interlibrary requests; publisher downloads; author emails |
| Reports Not Retrieved | Full texts you could not obtain | Unfilled requests; broken links; no response |
| Reports Assessed | Full texts you read for eligibility | Screening tool or spreadsheet |
| Reports Excluded | Full texts excluded with reasons | Eligibility log with reason codes |
| Studies Included | Individual studies that met criteria | Final inclusion list |
| Reports Included | All reports tied to the included studies | Study-report mapping file |
How To Create The PRISMA Flow Diagram Step By Step
Pick The Right Template
Choose the PRISMA 2020 diagram that matches your routes. One version is for databases and registers only. Another adds an “other methods” lane for sources such as trial registries, websites, preprint servers, citation chasing, or expert contacts. Saving the right file at the start avoids edits later.
Plan Your Tracking Sheet
Set up a sheet with columns for source, date, raw hits, duplicates, screened, excluded, reports sought, not retrieved, assessed, excluded with reason, studies included, and reports included. Add free-text notes for odd cases. Log numbers the same day you run each search.
Identification: Gather All Records
Export every hit from each database and register. If a site limits exports, pull in batches and keep a tally. Record the raw counts before any filters. For other routes, write down what you did and how many items you collected. Save search strings and dates so someone else can repeat the steps.
De-duplication: Remove Repeats
Merge all files into one library and run a duplicate check. Use both exact-match and fuzzy rules so near-identical items do not slip through. Record the number removed and keep a log of the method and tool you used. A second pass after screening can catch strays.
Screening: Titles And Abstracts
Decide inclusion and exclusion rules in advance. Work in pairs if you can. Tag each record as include, exclude, or unsure. The count of records screened equals the library after de-duplication. The count of records excluded equals the number tagged as exclude at this stage.
Retrieval: Get Full Texts
Fetch the PDF for every record marked include or unsure. Use library tools, publisher sites, or author contact. Record how many reports you sought and how many you could not get. Keep request IDs so you can show you tried.
Eligibility: Assess Reports
Read each full text against your criteria. Record excludes with one clear reason per report, such as wrong study type, wrong population, wrong outcome, or duplicate report of an included study. Keep the reason list short and consistent so the flow diagram stays tidy.
Included: Studies And Reports
Map reports to studies. One study can have several reports across years or journals. Record both the number of studies and the number of reports included. This distinction sits in the bottom two boxes of the diagram.
Adapting The PRISMA Flow For A Literature Review That Isn’t Systematic
If you are writing a narrative or scoping style review, you can still show a flow. Be clear that you used a PRISMA-style diagram for transparency, not as proof of a full systematic method. Describe what you did, where you searched, the dates, and why you made each choice.
Use the “other methods” lane when you harvest items from reference lists, topical websites, or expert lists. If you skipped a step, say so on the figure and in the text. The point is to make the path visible so readers can judge the reach of your search.
Counting Rules That Keep Numbers Straight
One Record, One Decision
Make decisions at the record level during title and abstract screening. Do not mark the same record include twice when it shows in two sources. The de-duplication step should leave you with a single row per record for this stage.
Track By Source Then Total
Keep subtotals per database, register, and other route. At the end, add a totals row. Subtotals help you find missing counts when the figure and sheet do not match.
Study And Report Are Not The Same
A single study can generate many journal articles, conference papers, or preprints. The bottom line of the diagram asks for both counts. Build a two-column map so you can see which reports link to each study.
Use Whole Numbers Only
Counts in a flow diagram are integers. If you sample titles during a pilot, do not scale up by a fraction for the final chart. Log the pilot in your notes, then screen the full set.
Keep A Date Stamp On Every Row
Add a column for the run date and platform version. Platforms change, and search results shift. A date stamp helps another reader replicate your steps.
Data Management With Reference Managers And Screening Tools
Set Up A Master Library
Pick one primary tool for storage. Zotero and EndNote are common. Covidence and Rayyan can help with screening. Whatever you choose, keep a backup in a neutral format such as RIS or CSV.
Use Persistent IDs
When a record has a DOI, PubMed ID, clinical trial ID, or arXiv ID, store it in a dedicated field. These IDs speed up de-duplication and help with report-study mapping later.
Name Your Imports
Append the source and date to each import file name. Example: “medline_2025-09-10.ris”. Clear names let you retrace steps even months later.
Document De-duplication Logic
Write down the match rules you used. Include fields compared and the sequence of passes. If your tool offers an audit log, export it and save it with your sheet.
Keep Decisions In Sync
If you move records between tools, include the decision fields in your export. You do not want to re-tag thousands of rows because a field was missing.
Reporting Choices And Wording On The Figure
Stick to the box labels in the official template. Minor edits for tense and spacing are fine, yet the core terms should stay the same so readers recognize the layout. Avoid long labels inside the boxes; keep detail in the caption or methods section.
Reasons for exclusion work best when they are short and non-overlapping. If a report misses several criteria, pick the main reason. Many teams use study type, population, outcome, timeframe, or duplicate report as their core set.
Decimals and percentages do not belong in the boxes. Keep those for narrative text if you need them. The figure itself tells the story through counts alone.
| Problem | Fix | Flow Box Affected |
|---|---|---|
| Records screened count does not match de-duplicated total | Re-run de-duplication and reconcile by source | Records Screened |
| Excluded reasons overlap | Pick one primary reason per report and standardize wordings | Reports Excluded |
| Study vs report counts mixed | Make a study-to-report map before final tallies | Studies Included / Reports Included |
| Other methods not logged | Create a mini log for citation chasing, websites, and contacts | Identification (Other Methods) |
| Full texts missing | Show requests sent and mark “not retrieved” once efforts stop | Reports Not Retrieved |
Checks Before You Finalize The Figure
Pause and read the whole sheet from top to bottom. Every transition should make sense: raw hits lead to screened records, screened records lead to reports sought, reports sought lead to reports assessed, and so on. If a jump looks odd, hunt it down now, not after peer review starts.
Ask a teammate who did not build the sheet to trace one record from start to finish. A fresh pair of eyes catches gaps you miss after hours of data entry. Keep that feedback near the sheet so the story behind each number stays clear months later.
- Totals by source add up to the grand total.
- Counts in the figure match the latest sheet version.
- Reason codes are short and used one at a time.
- Box labels match the official template terms.
- The caption names the routes, date range, and template version.
Good Practices That Save Time And Stress
Name Reasons For Exclusion
Agree on a short reason list before you read full texts. Four to six codes are enough in most topics. Clear codes keep the final figure clean and prevent long footnotes.
Keep Search Notes
Store strategies, dates, platforms, and export settings in one file. Add screenshots when a site will not export settings. These small notes pay off when you double-check counts.
Update Runs And Amends
If you rerun searches, add the new counts to the same sheet with a new date. Keep the older rows so your audit trail stays intact. If criteria change mid-stream, add a row to your sheet that states the change and the date.
PRISMA Flow Diagram For Literature Review: Worked Mini-Example
Say your database search returns 2,450 records across three platforms. You also add 80 items from trial registries and 40 from citation chasing. After merging, your library shows 2,570 items. You remove 630 duplicates, leaving 1,940 records to screen.
During screening you mark 1,250 as exclude. You seek full texts for 690 records and fail to retrieve 18. You assess 672 reports. Of those, 540 do not meet criteria. You keep 132 reports that describe 89 individual studies. Those two counts sit in the bottom boxes of the figure.
Template Links And Handy References
Download the official PRISMA 2020 flow diagram templates and pick the route that fits your review. The same site links to checklists and the detailed explanation paper.
Read the PRISMA 2020 statement for the full set of items that sit behind the figure. The paper shows how the new diagram distinguishes records, reports, and studies.
If you want strong search notes, add the PRISMA-S extension. It lists what to write down about sources, strings, and updates so your flow counts match your methods.
Finish the figure only after the last inclusion decision today. Cross-check every box against your sheet, then paste the counts into the template. Save a copy of the blank file and your filled version clearly. Store the sheet and the filled figure in the project folder with backups. Add the figure to your review with a short caption that states the date of the final search window and the routes used.