Check each study against your written criteria using the full record, then log an include or exclude decision with the criterion that drove it and flag borderline cases for a second reviewer.
Work through the record field by field
- Compare the population, design, intervention, and outcomes in the full record against each criterion you wrote in advance. Use the complete record, since the abstract may omit design details or outcome definitions. ClinicalTrials.gov, for example, defines structured fields for study design, eligibility, outcomes, and reported results that you can read directly.
- Record the decision as include or exclude and name the specific criterion that drove it. A note such as "excluded: wrong population" is weaker than "excluded: participants under 18, criterion P2 requires adults." The named criterion lets a second reviewer retrace your reasoning.
- Flag borderline cases for a second reviewer instead of resolving them alone. Mark which criterion is in question and what information would settle it, so the second reviewer starts from the same evidence.
Keep the audit trail consistent
Apply the same criteria to every record and document the selection process. The Cochrane Handbook covers planning searches and selecting studies for systematic reviews, and PRISMA 2020 provides a checklist and flow diagram for reporting what studies were found and included. Your decision log feeds directly into that reporting.
When the record is incomplete
If a field needed for a criterion is missing, mark the study as unresolved and note which field you need. Retrieve the full text or the registration record before deciding. Treat the absence of a detail in one source as a gap to fill, not as proof of ineligibility.