Full-text screening is the second selection stage of a systematic review, where two reviewers read the complete article of every record that survived title and abstract screening and make a final include-or-exclude decision against the eligibility criteria. Unlike the first pass, this stage is strict and conclusive: each excluded study must be logged with a single, specific reason for exclusion that appears in the flow diagram.
What changes when you move from abstracts to full texts
At the abstract stage you keep anything that might qualify. At full text you have the entire methods and results section in front of you, so the information gaps that forced a lenient first pass are gone. You can now confirm the population, verify the comparator, check the outcome definition, and confirm the study design. This is the point where your inclusion and exclusion criteria are applied at full force, and where ambiguity is finally resolved one way or the other.
Retrieving and organising the full texts
Find every report you can
Before screening can begin you need the full article for every record that passed the first pass. Most are available through institutional subscriptions or open access, and not retrieved is itself a category that must be counted in the PRISMA flow diagram. Where a report cannot be located, document the attempt rather than quietly dropping it, since a high number of unretrievable reports is a threat to the completeness of the review.
Keep the work inside a structured platform
Manage retrieval and decisions inside a dedicated systematic review screening tool so that each reviewer’s vote, the reason attached to it, and the eventual resolution are all captured automatically. Spreadsheet-based screening tends to lose the audit trail that peer reviewers increasingly ask to see.
Screening in duplicate with explicit reasons
As with the first pass, two reviewers work independently and blind to each other. The settled answer to how many reviewers screening needs is two, plus a third for adjudication. The new discipline at full text is that every exclusion carries one reason, applied in a fixed hierarchy: wrong population is recorded before wrong outcome, for example, so that two reviewers excluding the same paper cite the same reason. A pre-agreed order keeps the exclusion table internally consistent.
Building the exclusion hierarchy
Because every excluded full text carries exactly one reason, the order in which reasons are applied has to be agreed before screening starts. A workable hierarchy maps directly onto the PICO elements, usually in this sequence: wrong study design, then wrong population, then wrong intervention or exposure, then wrong comparator, then wrong outcome, with duplicate or wrong publication type sitting outside the clinical order. The principle is to record the most fundamental mismatch first. A randomised-trial review that retrieves a cohort study on the right patients and outcome should exclude it as wrong design, not wrong outcome, because the design failure is decisive regardless of anything downstream. Agreeing the order in advance is what makes two reviewers excluding the same paper cite the same category, which in turn keeps the exclusion table internally consistent when it is tallied for reporting.
Settling disagreements
Two careful reviewers will still disagree, and those disagreements matter more at full text because the decision is final. The methods for resolving screening conflicts are discussion to consensus or referral to a third reviewer, and the rate of disagreement is worth tracking. You can quantify agreement before adjudication using inter-rater reliability for screening, which signals whether the criteria are operating cleanly.
Reporting the full-text stage
The output of this stage feeds directly into the lower half of the flow diagram: the number of reports assessed for eligibility, the number excluded with reasons broken down by category, and the final number of studies included. Because these counts are audited by reviewers and readers, the arithmetic has to reconcile exactly with what entered the stage. A flow diagram whose numbers do not add up is one of the most common reasons a submission is sent back.
A step-by-step full-text procedure
Run the stage as a fixed sequence so nothing is improvised once the decisions become final:
- Retrieve the full report for every record that passed the first pass, logging any you cannot obtain as a counted not retrieved category rather than dropping them.
- Load the reports into your chosen screening platform so votes, reasons, and resolutions are captured automatically.
- Have two reviewers assess each report independently and blind, applying the eligibility rules at full force.
- Attach a single reason from the agreed hierarchy to every exclusion, and resolve every disagreement before the record leaves the stage.
- Reconcile the arithmetic: reports assessed, reports excluded by reason, and studies included must add up exactly to what entered the stage.
Common full-text screening mistakes
Reviewers repeatedly lose marks for the same lapses. Recording more than one reason per exclusion makes the reason counts impossible to tally cleanly. Dropping unretrievable reports silently understates the records assessed and hides a real threat to completeness. Re-litigating the first pass by excluding studies for reasons that should have been settled at title and abstract wastes effort and muddies the audit trail. And screening without two independent reviewers at this decisive stage reintroduces the selection bias the method exists to remove, which is why the question of the number of screeners a review needs applies just as firmly here as at the first pass. A clean stage avoids all four by fixing the rules, the hierarchy, and the reviewer roles before the first report is opened.
From included studies to synthesis
Once the included set is fixed, the studies move forward to data extraction and risk of bias assessment, and from there into synthesis. Some reviewers also use the full-text stage to begin citation chasing from the reference lists of included articles, feeding any new records back through screening. Full-text screening is therefore the gate that defines the evidence base for everything that follows. If the volume of full texts is more than your team can assess in duplicate alongside its other work, our screening and extraction service can run the stage to a documented, two-reviewer standard and hand back a reconciled included set.