A screening form turns your eligibility criteria into questions a reviewer can answer in seconds per record. Here is a screening form for a systematic review at both stages, title and abstract and full text, with versions for scoping and qualitative reviews. Copy them into a spreadsheet or your screening software, and pilot them first.
The forms assume your criteria are already written. If they are not, build them first with the inclusion and exclusion criteria builder.
Form 1: title and abstract screening (copy this)
Use one row or one copy per record. The questions follow PICO plus design and publication type.
TITLE AND ABSTRACT SCREENING FORM
Review: Reviewer: Date:
Record ID: First author, year: Source database:
Q1 Population matches the criteria? Yes / No / Unclear
Q2 Intervention or exposure matches? Yes / No / Unclear
Q3 Comparator matches (or none required)? Yes / No / Unclear
Q4 Outcome could be reported? Yes / No / Unclear
Q5 Study design is eligible? Yes / No / Unclear
Q6 Publication type is eligible? Yes / No / Unclear
(not an editorial, letter, commentary or protocol)
DECISION Include / Exclude / Maybe
RULE Any "No" -> Exclude
All "Yes" -> Include (goes to full text)
Any "Unclear" and no "No" -> Maybe (goes to full text)
Reason if excluded (optional at this stage):
Note:
Q4 asks whether the outcome could be reported. Abstracts often leave out secondary outcomes, so don't exclude on a missing outcome until you have the full text.
Form 2: full-text screening (copy this)
At full text, you count reports rather than records, and every exclusion needs one main reason so the reasons add up in your PRISMA flow diagram.
FULL-TEXT SCREENING FORM
Review: Reviewer: Date:
Report ID / citation:
Full text obtained? Yes / No (No = "report not retrieved", not an exclusion)
Same study as report(s): (link companion papers of one trial)
Check the criteria in this order; stop at the first "Not met":
1. Publication type Met / Not met / Unclear Page or quote:
2. Population Met / Not met / Unclear Page or quote:
3. Intervention/exposure Met / Not met / Unclear Page or quote:
4. Comparator Met / Not met / Unclear Page or quote:
5. Outcomes Met / Not met / Unclear Page or quote:
6. Study design Met / Not met / Unclear Page or quote:
7. Setting, language, date Met / Not met / Unclear Page or quote:
DECISION Include / Exclude / Awaiting information (author contacted)
Main exclusion reason (the first criterion not met):
Note:
Versions for scoping and qualitative reviews
Change the questions to match the framework. The decision rules stay the same.
| Review type | Framework | Questions at title and abstract | What changes at full text |
|---|---|---|---|
| Systematic review (effects) | PICO | Population, intervention, comparator, outcome, design, publication type | Strict check of design and outcome data |
| Scoping review | PCC | Population, concept, context, source type (any evidence, including reports and policy documents if your protocol allows) | No design or quality questions; check the source addresses the concept in the context |
| Qualitative evidence synthesis | SPIDER | Sample, phenomenon of interest, design, evaluation, research type | Does it report qualitative findings (themes, quotes) from qualitative data collection and analysis? For mixed methods, is the qualitative part reported separately? |
For a scoping review, a single question often does most of the work: "Does this source say anything about [concept] in [context]?" Keep the form short, because scoping searches are broad and record counts are high. See scoping review vs systematic review for when each applies.
Notes on using the forms
- Pilot first. Both reviewers screen the same 50 to 100 records, compare decisions and measure agreement. If Cohen's kappa is low, rewrite the unclear criteria before going on.
- Screen independently. Each reviewer completes the form without seeing the other's decision. Cochrane requires two independent reviewers for final inclusion decisions [1].
- Be inclusive early, strict late. Unclear goes forward at title and abstract; at full text, Unclear means you contact the authors or record "awaiting information".
- Use a fixed order of criteria at full text. The first criterion not met becomes the main reason. Write the order in your protocol.
- Log every decision. Keep the forms, or a screening decision log, so you can report how many reviewers screened each record and how disagreements were settled (PRISMA 2020 item 8) [2].
A worked example (invented)
Both records are invented for illustration. The review asks whether planned night-shift naps improve alertness in adult shift workers.
| Record | Q1 Population | Q2 Intervention | Q5 Design | Decision |
|---|---|---|---|---|
| "A 30-minute nap and vigilance in ICU nurses on night shifts: a crossover trial" | Yes | Yes | Yes | Include (to full text) |
| "Napping habits of medical students: a survey" | No (students, no shift work) | Unclear | No (survey) | Exclude: wrong population |
| "Strategies against fatigue in air traffic control" (no abstract) | Yes | Unclear | Unclear | Maybe (to full text) |
The second record fails two questions. Following the rule, the first "No" in the form's order, population, is its reason.
How to screen with these forms in AI for Research
The Systematic Review workspace builds the form from your review: the criteria come from the Protocol tab and the exclusion reasons from the review type.
- Set the review type, framework and Include when / Exclude when criteria on the Protocol tab.
- Add records on the Search tab: PubMed, Europe PMC and OpenAlex searches, or uploaded RIS, BibTeX, CSV or PubMed .nbib files. Duplicates are set aside by DOI, PubMed id and title.
- On the Screening tab, choose Title and abstract. Each record shows its title, authors, abstract and DOI or PubMed link. Press I to include, E to exclude (with a reason from the list) or M for maybe.
- Switch to Full text. Records you included or marked maybe at title and abstract are there. Read the paper through the DOI link and decide again.
- Optional: Suggest for the next 25 (1 credit) adds an AI suggestion with a one-sentence reason to each record. It is labeled as a suggestion; your vote is what counts.
- Screening somewhere else, or on paper? On the Report tab, download the screening form, the exclusion-reason list and the decision log as Word or PDF. They are built from your protocol's criteria and the review type's reasons, so a team that screens in a spreadsheet or another tool uses the same form. No credits.
| Review type | Exclusion reasons offered in the workspace |
|---|---|
| Systematic, rapid, living | Wrong population; wrong intervention or exposure; wrong comparator; wrong outcomes; wrong study design; wrong setting; wrong publication type; not in an included language; outside the date range; full text not available; duplicate report |
| Meta-analysis | The systematic list plus "no usable outcome data for pooling" |
| Scoping | Wrong population; wrong concept; wrong context; wrong source type; not in an included language; outside the date range; full text not available; duplicate report |
| Qualitative evidence synthesis | Not qualitative data; wrong sample; wrong phenomenon of interest; wrong study design; wrong publication type; not in an included language; full text not available; duplicate report |
To add a second reviewer, put the review in a project and share it as can edit, or tick Use the AI as a second reviewer. The workspace then needs two decisions per record and lists disagreements as conflicts. The full workflow is on the systematic review software page.
What these forms and the workspace do not do
-
The workspace does not show your own form questions. It records Include, Exclude or Maybe and one reason from the review type's fixed list. Keep the Yes/No/Unclear answers in your own sheet if your protocol needs them, or start from the Word export and add your questions to it.
-
Decisions made outside the workspace do not flow back in. The Word and PDF documents are for teams screening elsewhere; their answers are not imported.
-
Decisions are not hidden from each other. Once your co-reviewer, or the AI, has decided on a record, that decision shows under the abstract. Screen before you read it.
-
Full texts are not attached. You open the paper through its DOI or PubMed link; there is no PDF upload at the screening stage.
-
"Not retrieved" has no separate box. "Full text not available" is recorded as an exclusion reason, so move those reports to "reports not retrieved" when you draw the final PRISMA diagram.
Who it is for
Students and teams running their first review, especially PhD candidates (AI tools for PhD students) and clinical teams (AI for medical research). If you already screen in another tool, compare it in Rayyan alternative or Covidence alternative.
Learn more
- What is a systematic review: where screening sits in the eight steps.
- Cohen's kappa for screening: measuring agreement in your pilot.
- How to make a PRISMA flow diagram: turning screening counts into the figure.
- Systematic review AI: validating AI screening and reporting it.
Sources
- Lefebvre C, Glanville J, Briscoe S, et al. Chapter 4: Searching for and selecting studies. In: Higgins JPT, Thomas J, et al. (eds). Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. Cochrane, 2024. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-04
- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. https://doi.org/10.1136/bmj.n71
Last updated 2026-11-13