Systematic Review Screening Form and Eligibility Criteria

A protocol criterion is not yet a screening rule

Eligibility criteria define intended evidence boundaries, but reviewers must still evaluate whether individual records cross those boundaries. Broad protocol terms like “adult population,” “usual care,” “prospective design,” or “relevant outcomes” invite varying interpretations among qualified reviewers. Without explicit operational rules, final study inclusion depends on individual reviewer bias rather than protocol standards.

A screening form translates protocol criteria into actionable screening rules without altering scientific scope. Each criterion must define observable inclusion indicators, explicit exclusion triggers, mixed-sample rules, and handling procedures for missing data. The form logs criterion versions, reviewer IDs, screening stages, citation evidence, original votes, conflict statuses, and final resolutions to establish a reviewable audit trail.

While PRISMA 2020 requires reporting study selection processes and reviewer arrangements, it does not mandate specific screening software or forms.^1,2 Conduct standards derive from core synthesis manuals. Cochrane intervention reviews require two independent reviewers for final full-text eligibility decisions, while dual screening at title/abstract is recommended.^3,4 JBI synthesis guidelines specify distinct dual-screening arrangements across synthesis types.⁵

Write criteria that two reviewers can apply consistently

Operational criteria must address single eligibility concepts. Combining population, intervention, and design into a single screening question obscures disagreement causes. Structuring composite rules into discrete questions with standardized options helps reviewers identify clear exclusions quickly. Polanin and colleagues recommend objective, single-concept screening questions paired with pilot training for large evidence reviews.⁶

During full-text evaluation, log decision notes citing specific page numbers, tables, registry fields, or companion papers. Direct citations provide far greater transparency than generic labels like “fails criteria,” fulfilling PRISMA Item 16b requirements for reporting excluded studies.²

Pilot rules before launching full screening

Select a representative test batch, have reviewers apply draft rules independently, evaluate discrepancies, and refine ambiguous wording. Document test sample size, rule modifications, approval dates, and criterion versions. Apply pilot testing standards aligned with your review protocol.

Include straightforward inclusions, clear exclusions, sparse abstracts, borderline reports, mixed cohorts, ambiguous designs, and known companion papers in test batches. Independent reviewer evaluation exposes subjective phrasing early, preventing lead reviewer bias from overriding objective protocol rules.

Inter-rater agreement metrics describe pilot consistency, but high kappa values do not guarantee criterion validity. High agreement can occur in batches dominated by irrelevant records, whereas a single disagreement on a rare study design can alter review scope. Analyze disagreement patterns and freeze dated screening rules only after reviewers demonstrate consistent application across complex reports.

Title/abstract and full-text screening serve distinct purposes

Missing details at the title/abstract stage do not confirm study ineligibility. Preliminary screening should remain intentionally over-inclusive because complete reports often contain missing method details.³ During full-text appraisal, evaluate all companion reports before assigning final inclusion. If essential details remain missing, log the paper as “awaiting classification” and follow protocol author-contact steps.

Link companion reports before assigning final study inclusion

A single trial may produce protocols, conference abstracts, trial register entries, primary outcome papers, secondary analyses, and long-term follow-up publications. Reviewers should not drop secondary papers as duplicates during title screening, nor should they analyze them as independent trials. Assign unique IDs to reports and link them to a central study ID prior to final eligibility decisions.^3,4

Linked papers resolve reporting gaps: a protocol paper may define study design, a main trial paper may detail population characteristics, and a registry record may verify recruitment dates. Log supporting citations for each inclusion call and retain all companion reports for extraction. When companion reports present conflicting data, document the discrepancy and follow protocol hierarchy rules.

Distinguish workflow statuses from eligibility outcomes: “Report Not Retrieved” indicates full text was unavailable; “Awaiting Classification” indicates missing author responses; “Ongoing Study” tracks active trial recruitment. Forcing unresolved papers into exclusion categories creates artificial completeness while hiding evidence uncertainty.

Treat screening logs as decision audits, not simple vote tallies

Compare independent screening calls only after both reviewers evaluate assigned batches. Retain original votes when resolving conflicts. Consensus records should document disputed criteria, evidence evaluated, participating reviewers, adjudicator involvement, and final decisions. When disagreements reveal ambiguous criteria, update screening instructions via formal amendment and evaluate if previously screened records require re-review.

Empirical evaluations show single-reviewer screening misses more eligible studies than dual independent screening, though error rates vary by topic complexity and reviewer experience.⁷ Any deviation from dual independent screening requires protocol justification, quality control checks, and transparent methods reporting. Document verification samples or second-reviewer audits where full dual screening is not feasible.

Text-mining and machine-learning tools prioritize records or assist preliminary classification. However, tool performance varies across fields.⁸ Document software versions, tasks, training datasets, stopping rules, human verification steps, and screening impacts. Final inclusion calls remain the responsibility of human reviewers.

Manage screening rule changes as controlled protocol amendments

Screening often reveals unexpected study designs or reporting formats. Clarifying instruction wording without altering scientific scope is acceptable, but changing population, intervention, design, or publication criteria constitutes a formal protocol amendment. Document rationale, decision dates, updated version numbers, and affected citations. Applying updated rules only to future records introduces time-dependent selection bias.

Before exporting final screening datasets, verify that every record contains a criterion version, stage, reviewer ID, decision, and date; every full-text exclusion includes primary reasons and citations; all conflicts show resolution logs; and all included reports link to primary study IDs.

Scope & Technical Limitations

Browser storage provides convenient drafting support but does not replace secure team database storage. Store master datasets, criterion versions, reviewer assignments, conflict logs, and automation settings in governed team workspaces. Final eligibility decisions require accountable human review.

Complete study selection with reproducible decisions and auditable handoffs

A defensible screening system translates protocol criteria into objective rules, pilots rules on test samples, preserves independent dual-screening votes, resolves conflicts transparently, and logs citation evidence. This process produces an auditable record explaining how evidence boundaries were applied.

Freeze screening datasets only after verifying reviewer assignments, unresolved citations, full-text reasons, and study-report links. Use the full-text exclusion log to compile publication tables and the PRISMA flow tool to reconcile final counts. Downstream outputs should derive directly from this primary decision ledger.