Build an evidence review that another scientist can audit

A literature review becomes unreliable long before the prose is written. The decisive moments are the choice of question, the search boundary, the exclusion rule and the treatment of contradiction.

Write the protocol while it can still constrain you

Define the population, intervention or exposure, comparator, outcomes, date range, languages and eligible study designs before retrieval. Record databases, queries and the planned synthesis. PRISMA 2020 provides a 27-item reporting framework and revised flow diagrams, while making clear that reporting quality is not the same as methodological quality (Page et al.). That distinction should remain visible.

ARiDA treats the protocol as part of the governed case. Search results are retained as records. Deduplication preserves provenance across copies. Screening decisions carry reasons, including every full-text exclusion. A later reviewer can distinguish “not found,” “found and excluded” and “included but judged weak.”

Build the table before the conclusion

The evidence table should capture population, intervention, comparator, outcome definition, follow-up, effect estimate, uncertainty, study design, limitations and source. Those fields force unlike studies to remain unlike. A randomized trial and a retrospective cohort may inform the same question without deserving the same authority.

Contradictory results stay side by side. ARiDA’s evidence layer can track support, opposition, coverage and source authority without asking a language model to average them into consensus. The writer synthesizes the accepted table and points to the records behind each material claim.

Treat the review as a decision input

A useful review ends by stating what the evidence can change. Does it alter target confidence, identify a safety veto, narrow a patient segment or justify an experiment? If the evidence is insufficient, name the missing study rather than filling the space with cautious prose.

The released article or brief is one view of the evidence set. The durable product is the protocol, record flow, inclusion logic and table that another scientist can reopen when the literature changes.