How to Build a Screening Workflow in Zotero for Systematic Reviews

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-27

Quick Answer

• Build a reproducible Zotero screening workflow using custom tags, saved searches, and full-text PDF annotations to track inclusion decisions across hundreds of references.
• Start by defining your inclusion and exclusion criteria before importing references, then apply tags consistently at each screening stage.
• Zotero alone cannot replace a formal systematic review protocol, so pair your workflow with reporting guidelines and documented methods from authoritative sources.

Why a Structured Screening Workflow Matters

Systematic reviews depend on transparent, reproducible methods for selecting studies. When researchers screen hundreds of references manually without a structured approach, they risk inconsistent decisions, lost records, and an inability to report how many records were excluded and why. The National Library of Medicine Research Methods Resources gateway provides access to authoritative biomedical texts that describe systematic review methodology, including the importance of documenting each screening step.

A well-designed Zotero workflow gives you a practical way to manage this process within a free, widely used reference manager. The goal is not to replace dedicated systematic review software but to create a defensible audit trail that supports your final report. This matters because funding agencies and journals increasingly expect transparent reporting of how studies were selected. The EQUATOR Network maintains reporting guidelines that help researchers document their methods clearly, and following such guidelines strengthens the credibility of your review.

Core Principles of Screening in Zotero

Separate Storage from Decision-Making

Zotero stores references and attaches PDFs, but it does not inherently track screening decisions. You must build that structure yourself. The core principle is to use Zotero's native features, tags, collections, saved searches, and notes, as a decision-tracking layer on top of your reference library.

Design Your Tag System Before You Start

Decide your screening stages in advance. A typical two-stage screening process includes title and abstract screening followed by full-text review. Your tags should reflect these stages plus the outcome of each decision. Common tag categories include:

• Stage indicators: stage1, stage2, fulltext
• Decision outcomes: include, exclude, maybe
• Exclusion reasons: exclude-no-data, exclude-wrong-population, exclude-wrong-design, exclude-duplicate
• Status markers: screened, unscreened, awaiting-decision

Keep your tag vocabulary short and consistent. Write your tag definitions in a protocol document before you begin screening, and store that document in a Zotero note attached to a dedicated protocol item.

Use Collections for Workflow Stages

Zotero collections function as folders. Create separate collections for each screening stage:

• 01-Imported for all records before screening
• 02-Title-Abstract-Screening for records currently in stage one
• 03-Full-Text-Review for records that passed stage one
• 04-Included for final included studies
• 05-Excluded for all excluded records

A record can belong to multiple collections, so you can keep the original import collection intact while moving records through workflow collections. This preserves your complete record set for reporting.

Building Your Zotero Screening Workflow

Step 1: Define Your Protocol and Criteria

Before importing a single reference, write your research question and define inclusion and exclusion criteria. This protocol should specify the population, intervention, comparison, outcomes, and study designs you will accept. The NIH Grants and Funding website emphasizes that rigorous methods are central to funded research, and a written protocol is the foundation of rigorous screening.

Store this protocol as a Zotero item with a note containing your criteria. This makes the protocol part of your library and ensures it travels with your screening records.

Step 2: Import References into a Dedicated Library

Create a new Zotero library or a clearly named collection within your existing library for each systematic review. Do not mix screening records with your general reference collection. This separation prevents accidental contamination of your screening decisions with unrelated references.

Import references from databases such as PubMed, Embase, Web of Science, and Scopus. Zotero can capture references directly from your browser using the Zotero Connector, or you can import RIS and BibTeX files. After import, run a duplicate check using Zotero's built-in duplicate detection feature. Remove or merge duplicates before screening begins.

Step 3: Apply Your Tag System

Work through your imported records and apply stage and status tags. Start by tagging every record as unscreened and stage1. This gives you a clear starting point and lets you use saved searches to track progress.

Apply tags consistently. If you are screening with a team, agree on tag definitions in advance and test them on a small sample of records before full screening begins. Disagreements about tag meaning will corrupt your screening data.

Step 4: Create Saved Searches for Workflow Management

Zotero saved searches are dynamic views that update automatically as you apply tags. Create saved searches for each workflow state:

• Unscreened Stage 1: tag contains unscreened AND tag contains stage1
• Stage 1 Include: tag contains include AND tag contains stage1
• Stage 1 Exclude: tag contains exclude AND tag contains stage1
• Stage 1 Maybe: tag contains maybe AND tag contains stage1
• Awaiting Full Text: tag contains include AND tag contains stage2
• Final Included: tag contains include AND tag contains fulltext

These saved searches appear in your Zotero sidebar and update in real time. They give you an instant count of records at each stage, which is essential for tracking progress and reporting your screening numbers.

Step 5: Screen Title and Abstract Records

Open each record in the Unscreened Stage 1 saved search. Read the title and abstract against your inclusion criteria. Apply one decision tag and one stage tag to each record:

• Apply include and stage2 if the record passes title and abstract screening
• Apply exclude and a reason tag if the record clearly fails criteria
• Apply maybe and stage1 if you need full text to decide

Work through the saved search until it shows zero records. This systematic approach ensures you screen every record exactly once.

Step 6: Retrieve and Screen Full Texts

For records tagged include and stage2, retrieve full-text PDFs. Attach each PDF to its Zotero item. Zotero can extract metadata from many PDFs, but verify that the extracted details match the original record.

Screen each full text against your criteria. Apply include and fulltext tags for records that pass, or apply exclude with a reason tag for records that fail at this stage. Use Zotero's PDF reader to highlight relevant passages and add notes. These annotations become part of your audit trail and help you justify inclusion decisions later.

Step 7: Track Your Screening Decisions

Maintain a screening log as a Zotero note or a separate spreadsheet. Record the date, the record identifier, the screener name, and the decision for each record. This log is essential for reporting and for resolving disagreements between screeners.

The Committee on Publication Ethics Core Practices emphasize transparency and accountability in research. A complete screening log demonstrates that your review followed ethical and rigorous practices.

At a Glance: Zotero Screening Workflow Components

Practical Implementation Steps

Set Up Your Zotero Library

1. Create a new collection named for your review topic
2. Write your protocol with inclusion and exclusion criteria as a Zotero note
3. Define your tag vocabulary and store it in the protocol note
4. Import all references from your database searches
5. Run duplicate detection and merge or remove duplicates
6. Apply initial tags (unscreened, stage1) to all records
7. Create your saved searches
8. Begin title and abstract screening
9. Retrieve full texts for records that pass stage one
10. Screen full texts and apply final decisions
11. Maintain your screening log throughout

Team Screening Coordination

If multiple people screen records, assign each screener a unique set of records or use Zotero groups for shared libraries. The ORCID for Researchers page explains how persistent researcher identifiers support accurate attribution of scholarly work, which is relevant when documenting who screened which records.

For team screening, consider using Zotero's group library feature. Each screener can work in the same library, and tags can include the screener's initials to track who made each decision. Disagreements should be resolved by discussion or by a third reviewer, and the resolution should be recorded in your screening log.

Records and Measurements

What to Record

Your screening records should capture enough detail to report your review accurately. At minimum, record:

• Total number of records imported from each database
• Number of duplicates removed
• Number of records screened at title and abstract stage
• Number excluded at title and abstract stage with reasons
• Number of full texts retrieved
• Number of full texts excluded with reasons
• Final number of included studies

The EQUATOR Network provides reporting guidelines such as PRISMA that specify what screening information to report. Following these guidelines ensures your methods section is complete and reproducible.

Measuring Screening Progress

Use your saved searches to monitor progress. The record counts in each saved search tell you how many records remain at each stage. A common pattern is to see the Unscreened Stage 1 count decrease steadily while the Stage 1 Include and Stage 1 Exclude counts increase. If the unscreened count does not change, you have a workflow problem that needs attention.

Documenting Exclusion Reasons

Apply a specific exclusion reason tag to every excluded record. Common reasons include:

• exclude-wrong-population
• exclude-wrong-intervention
• exclude-wrong-outcome
• exclude-wrong-design
• exclude-no-fulltext
• exclude-duplicate
• exclude-nonenglish

These tags let you generate exclusion counts by reason, which you will need for your PRISMA flow diagram and methods section.

Common Failure Patterns

Inconsistent Tag Application

The most common failure is applying tags inconsistently. One screener might tag a record as exclude while another tags the same type of record as maybe. This corrupts your saved search counts and makes your screening data unreliable. Prevent this by writing clear tag definitions, testing them on a sample, and holding a calibration session before full screening begins.

Mixing Screening Records with General References

If you screen within your main Zotero library without a dedicated collection, you risk accidentally tagging unrelated references or losing track of your screening set. Always create a dedicated collection or library for each review.

Skipping Duplicate Detection

Database searches produce overlapping results. If you skip duplicate detection, you may screen the same study twice and count it twice in your results. Run Zotero's duplicate detection immediately after import and again after full-text retrieval.

Relying on Memory Instead of Tags

Some researchers try to remember which records they have screened. This fails with more than a few dozen records. Use tags and saved searches systematically so your workflow state is always visible in Zotero.

Not Keeping a Screening Log

Without a dated log of decisions, you cannot reconstruct your screening process or resolve disagreements. Your screening log is your primary evidence that the review was conducted systematically.

Limitations of Zotero for Systematic Reviews

Zotero is a reference manager, not dedicated systematic review software. It lacks built-in features for blinded screening, inter-rater reliability calculations, and automated risk-of-bias assessment. For large or complex reviews, you may need to supplement Zotero with specialized tools or spreadsheets.

Zotero also does not automatically generate PRISMA flow diagrams. You will need to create these manually from your screening log data. The EQUATOR Network can help you identify the appropriate reporting guideline for your review type.

Another limitation is that Zotero's saved searches work within a single library. If you use multiple libraries, you must recreate your saved searches in each one. Plan your library structure before you begin to avoid this complication.

Welfare and Safety Context

Data Management and Reproducibility

Systematic reviews are research outputs that should be reproducible. The NIH Data Management and Sharing Policy describes expectations for managing and sharing research data. While this policy applies to NIH-funded research, the underlying principle of transparent data management applies to all systematic reviews. Your screening records, including tags and decisions, are part of your research data and should be preserved and shared where appropriate.

Publication Ethics

The Committee on Publication Ethics Core Practices outline expectations for ethical research conduct, including accurate reporting of methods and results. A well-documented screening workflow supports ethical reporting because it lets you describe exactly how studies were selected. Fabricating or omitting screening details violates these ethical standards.

Researcher Identity and Attribution

The ORCID for Researchers page explains how ORCID iDs provide persistent identifiers for researchers. In team screening, using ORCID iDs in your screening log and publications ensures accurate attribution of who contributed to the review process.

Professional Escalation Criteria

When to Seek Additional Support

If your review involves more than several thousand records, or if you need blinded screening and inter-rater reliability statistics, consider using dedicated systematic review software in addition to Zotero. The National Library of Medicine Research Methods Resources gateway can help you identify methodological guidance for complex reviews.

When to Consult a Librarian or Methodologist

If you are unsure about your search strategy, screening criteria, or reporting requirements, consult a health sciences librarian or a systematic review methodologist. These professionals can help you design a search that is comprehensive and reproducible. The NIH Grants and Funding website also provides context on the rigor expected in funded research, which can guide your methodological decisions.

When to Revise Your Protocol

If you discover that your inclusion criteria are too broad or too narrow during screening, you may need to revise your protocol. Document any protocol changes with dates and reasons. This transparency is consistent with the reporting expectations described by the EQUATOR Network.

Building a Two-Person Screening Agreement Protocol

Screening hundreds of references becomes unreliable when two reviewers interpret the same inclusion criteria differently. A calibration exercise before full screening begins prevents this problem. This section describes a practical agreement protocol that uses Zotero tags and saved searches to measure inter-rater consistency and resolve disagreements before they corrupt your screening data.

Why Calibration Matters Before Full Screening

Systematic review methodology expects that screening decisions are reproducible across reviewers. The National Library of Medicine Research Methods Resources gateway provides access to biomedical texts that describe the rationale for dual screening and disagreement resolution. When two reviewers screen the same set of records independently, the level of agreement between them is a measurable indicator of whether your criteria are clear enough to apply consistently.

A calibration exercise is a small-scale test run. You select a sample of records, have two reviewers screen them independently, then compare decisions. The purpose is to find ambiguous criteria and inconsistent tag use before you apply them to the full set. This is different from pilot testing your search strategy, which tests whether your search finds relevant records. Calibration tests whether your screening decisions are consistent between people.

Selecting a Calibration Sample

Choose a sample that represents the range of records you expect to encounter. A random sample of 20 to 30 records is a reasonable starting point for most reviews, but the sample should include records that are clearly eligible, clearly ineligible, and borderline. If your search returned a mix of study designs, populations, and publication types, your calibration sample should reflect that mix.

Do not select only easy records. The purpose of calibration is to find where reviewers disagree, and disagreement is most likely on records that are ambiguous. If your sample contains only obvious inclusions and exclusions, you will not learn where your criteria need refinement.

Running the Calibration Exercise

Each reviewer works independently in their own Zotero library or in a shared group library with a personal tag prefix. The ORCID for Researchers page explains how persistent identifiers support accurate attribution of scholarly work, which is relevant when you need to document which reviewer made which decision.

Create a saved search that shows only the calibration sample. You can do this by applying a unique tag such as calibration to every record in the sample. Each reviewer then screens the sample using the same tag system described earlier in this workflow, applying include, exclude, or maybe tags to each record.

Do not discuss decisions during the calibration screening. The purpose is to measure independent agreement, and discussion during screening will inflate your agreement estimate. Screen independently, then compare results.

Measuring Agreement with a Simple Decision Matrix

After both reviewers finish the calibration sample, export the decisions to a spreadsheet. Create a two-column table with reviewer A decisions in one column and reviewer B decisions in the other. Count the number of records where both reviewers applied the same decision tag. Divide that number by the total number of records in the sample to get your percentage agreement.

A more informative approach is to build a decision matrix that shows where disagreements occur. The matrix has reviewer A decisions as rows and reviewer B decisions as columns. The diagonal cells show agreements. The off-diagonal cells show disagreements. For example, if reviewer A tagged a record as include and reviewer B tagged the same record as exclude, that record appears in the include-exclude cell of the matrix.

The matrix tells you the pattern of disagreement. If most disagreements are between include and maybe, your criteria may be too vague for borderline records. If disagreements are between exclude and maybe, reviewers may be applying different thresholds for what counts as clearly ineligible.

Interpreting Disagreement Patterns

A high agreement percentage, above 90 percent, suggests your criteria are clear and your tag definitions are consistent. You can proceed to full screening with confidence.

An agreement percentage between 75 and 90 percent indicates that some criteria need clarification. Review the records where reviewers disagreed and discuss why each reviewer made their decision. The goal is to identify the specific criterion that caused the disagreement and to refine the wording of that criterion in your protocol note.

An agreement percentage below 75 percent means your criteria are not ready for full screening. You need to revise the criteria, update the tag definitions, and run a second calibration exercise on a new sample. Do not proceed to full screening until agreement improves.

Documenting the Calibration Outcome

Record the calibration results in your screening log. Include the date, the sample size, the number of agreements, the percentage agreement, and a summary of the disagreements and how they were resolved. This documentation is part of your audit trail and supports the transparency expectations described by the Committee on Publication Ethics Core Practices.

Store the calibration spreadsheet as an attachment in Zotero. You can attach it to the protocol item or to a dedicated calibration item. This keeps all screening documentation in one place.

Using Saved Searches to Track Agreement

You can use Zotero saved searches to identify records where reviewers disagree. If each reviewer applies a tag with their initials, such as reviewer1-include and reviewer2-exclude, you can create a saved search that shows records where reviewer one tagged include and reviewer two tagged exclude. This gives you a dynamic view of all disagreements during the calibration exercise.

The same approach works during full screening if you use a group library. Create saved searches for each disagreement combination. These saved searches give you an instant count of records that need discussion and resolution.

Resolving Disagreements

When reviewers disagree on a record, the standard approach is discussion followed by consensus. If discussion does not resolve the disagreement, a third reviewer makes the final decision. The EQUATOR Network provides reporting guidelines that describe how to report the number of disagreements and the resolution process in your methods section.

Record the resolution for every disagreement. The screening log should show which records were disputed, which reviewers disagreed, and how the disagreement was resolved. This level of detail is expected in systematic review reporting.

When to Repeat Calibration

Run a second calibration exercise if you revise your criteria after the first exercise. The second exercise should use a new sample of records that you have not screened before. This confirms that the revised criteria produce consistent decisions.

You should also run a calibration exercise if you add a new reviewer to the team mid-review. The new reviewer must demonstrate consistent application of the criteria before screening the full set.

Common Calibration Failure Patterns

The most common failure is skipping calibration entirely and moving directly to full screening. This produces inconsistent decisions that are difficult to correct after the fact. Re-screening the full set is time-consuming and may not be possible if you have already excluded records.

Another failure is discussing records during the calibration exercise. This inflates your agreement estimate and hides real disagreements. Screen independently, then compare.

A third failure is using a calibration sample that is too small or too homogeneous. A sample of five records that are all clearly eligible will not reveal the ambiguities that cause problems in the full set.

Calibration and Reporting Guidelines

The EQUATOR Network maintains reporting guidelines that specify how to describe the screening process in your methods section. These guidelines expect you to report the number of reviewers, the number of records screened by each reviewer, the number of disagreements, and how disagreements were resolved. Your calibration documentation provides the data for this reporting.

Calibration in the Context of Data Management

The NIH Data Management and Sharing Policy describes expectations for managing and sharing research data. Your calibration records are part of your research data. Preserve them and share them where appropriate to support reproducibility of your review.

Frequently Asked Questions

How do I create a saved search in Zotero for screening?

Click the gear icon in the Zotero sidebar and select New Saved Search. Name your search, then add conditions using the tag field. For example, set Tag contains unscreened and Tag contains stage1 to create a dynamic view of records awaiting title and abstract screening.

Can I use Zotero for a full systematic review?

Zotero can manage references and track screening decisions, but it lacks specialized features like blinded screening and inter-rater reliability calculations. For comprehensive reviews, use Zotero alongside dedicated systematic review software or a spreadsheet for statistical tracking.

How do I track exclusion reasons in Zotero?

Create a tag for each exclusion reason, such as exclude-wrong-population or exclude-no-fulltext. Apply the appropriate tag to each excluded record. Your saved searches can then count records by exclusion reason for your PRISMA flow diagram.

What is the best way to handle duplicates in Zotero?

Use Zotero's built-in duplicate detection feature immediately after importing references. Review the suggested duplicates and merge or remove them. Run duplicate detection again after full-text retrieval to catch any duplicates that were not obvious at the citation level.

How do I collaborate on screening with a team in Zotero?

Create a Zotero group library and add your team members. Use tags that include screener initials to track who made each decision. Hold a calibration session before screening begins to ensure consistent tag application across the team.

Do I need a protocol before I start screening?

Yes. Write your research question, inclusion criteria, exclusion criteria, and screening stages before importing references. Store this protocol as a Zotero note attached to a protocol item. This document guides your decisions and supports transparent reporting.

How do I generate a PRISMA flow diagram from Zotero?

Zotero does not generate PRISMA diagrams automatically. Use your saved search counts and screening log to calculate the numbers for each stage, then create the diagram manually or with a diagramming tool. The EQUATOR Network provides the PRISMA guideline and templates.

What should I do if my screening criteria change mid-review?

Document the change in your protocol note with the date and reason. Apply the revised criteria consistently to all remaining records. If the change affects already screened records, re-screen those records under the new criteria and update your screening log accordingly.

Using the Evidence

Related Articles

• Meta-Analysis vs Systematic Review: Choosing the Right Synthesis Method
• Protein Synthesis Step By Step
• Meta-Analysis vs Systematic Review: What's the Difference?
• Protein Synthesis: A Step-by-Step Guide to Transcription and Translation
• Molecular Cloning Workflow: Planning Validation Into Every Step

References and Further Reading

• Research Methods Resources. National Library of Medicine.
• EQUATOR Network. EQUATOR Network.
• Core Practices. Committee on Publication Ethics.
• NIH Grants and Funding. National Institutes of Health.
• ORCID for Researchers. ORCID.
• Data Management and Sharing Policy. National Institutes of Health.
• NCBI Literature Resources. National Center for Biotechnology Information.
• PubMed. National Library of Medicine.

This article is educational and does not replace institutional policy, professional advice, or applicable safety and regulatory requirements.