How to Revise a Literature Review After Your Research Question Changes

A changed research question does not usually mean the review is wasted. It means the review now has two jobs: keep the old work auditable, and rebuild the search and synthesis around the new question. The practical answer is to treat the revision as a controlled re-scoping exercise: restate the question, translate it into explicit eligibility criteria, rebuild the search from those criteria, re-screen in stages, and only then rewrite the synthesis. Cochrane's guidance on searching for and selecting studies is a useful spine here, because it ties every search decision back to eligibility criteria and asks you to document the process, deduplicate records, screen in stages, and rerun searches before publication.

This article gives you a sequence with checkpoints, a worked example, a reusable checklist, and stop rules so you can finish the revision and know when it is ready.

Before rewriting anything, work out what kind of change you are facing. The revision path differs depending on whether the question shifted in population, intervention or exposure, outcome, comparison, or scope.

Ask yourself:

• What exactly changed? Write the old question and the new question side by side in one sentence each.
• Which eligibility criteria are affected? If the change touches who counts as a study participant or what counts as an eligible study design, your search strategy must change too.
• What can be reused? Records you already retrieved, screening decisions that still match the new criteria, and background material that is not question-specific.
• What must be retired? Any included study that no longer meets the new criteria, and any synthesis claim built on it.

A quick diagnostic: if the new question changes eligibility criteria, you are doing a re-scope and need a new search. If it only changes emphasis or framing while eligibility stays the same, you may be doing a re-frame and can reuse more of the existing record set. Decide which one you are in before you open the draft.

Your first artifact is a frozen snapshot of the previous version. This is what makes the revision auditable and lets a coauthor or reviewer see why the review changed.

Create a version folder containing:

1. The old research question and eligibility criteria, dated.
2. The old search strategies exactly as run, with database, date, and any filters.
3. The record counts at each stage: retrieved, deduplicated, screened, included.
4. The screening decisions and reasons for exclusion at full-text stage.
5. The old synthesis and any tables or figures built from it.

Cochrane's emphasis on documenting the search process and deduplicating records supports keeping these counts explicit rather than reconstructing them later. If you used a reference manager or a project workspace, keep the frozen export alongside the new one so the two are never confused.

Checkpoint 1: A coauthor can read your snapshot folder and explain, without asking you, what the old review asked and how it was searched.

The new question must be operational before any search begins. Cochrane's guidance is to plan searches from eligibility criteria, so the criteria come first and the search strings follow.

Write the new question in a structured form appropriate to your field, then convert each element into an inclusion or exclusion rule. For each criterion, note whether it is new, unchanged, or retired relative to the frozen version. That three-way marking is what makes the revision traceable.

A useful discipline: for every criterion you add, ask what search concept it implies. If you add a new outcome, you likely need new outcome terms. If you narrow the population, you likely need to remove or tighten population terms.

Checkpoint 2: Every eligibility criterion maps to at least one search concept, and every search concept maps back to a criterion. Nothing in the search is there "just in case."

Now rebuild the strategy. Cochrane recommends combining free-text terms with subject headings and using multiple sources, and both PubMed and Google Scholar document features that help you do this deliberately.

• PubMed supports field tags, phrase searching, filters, Advanced Search, and saved searches, and its Search Details show how terms were translated and combined. Use Search Details to verify that your concepts were actually combined the way you intended.
• Google Scholar documents searching by author and title, date sorting and limits, Cited by and Related articles, citation export, and alerts. Use Cited by and Related articles to chase forward and sideways from a seed paper that fits the new question.
• Multiple sources matter because no single index covers everything; Cochrane's guidance to use multiple sources is the reason to run the new strategy in more than one place.

If you want a scholarly search interface to run and iterate the new strategy, WisPaper search is the natural entry point; readers who want a public explanation of the same capability can start from WisPaper Scholar Search.

Record for each source: the exact strategy, the date run, filters applied, and the number of records returned. Save searches where the platform supports it so the rerun before publication is a repeat, not a reconstruction.

Checkpoint 3: A colleague could rerun your strategy from your written record and get a comparable record set.

Cochrane describes screening in stages and linking multiple reports of one study. Apply both to the new record set.

1. Deduplicate the combined records and record the count removed.
2. Screen titles and abstracts against the new criteria, recording reasons for exclusion.
3. Screen full texts, again with recorded reasons.
4. Link reports so that multiple papers from one study are counted as one study, not several.

Keep the old screening decisions visible. Where a previously included study is now excluded, record the criterion that changed. That single field is often the most valuable line in the whole revision, because it explains the difference between the two versions.

Checkpoint 4: Every excluded full text has a reason tied to a named criterion, and every included study is linked to all its reports.

Once the included set is settled, rebuild the synthesis from the new evidence base. Patching old paragraphs is where revisions go wrong: a sentence written for the old question can survive into the new draft and quietly contradict the new criteria.

Work in this order:

• Evidence table first. One row per included study, with the fields your new question requires.
• Then the narrative. Write each synthesis claim against specific rows in the table.
• Then the gaps. State what the new question cannot answer from the included evidence.
• Then the framing. Update the introduction and rationale last, so they describe the review you actually did.

If you need to ask questions across the papers you have collected, Library QA supports asking questions based on papers uploaded to My Library or added from search results, which fits the evidence-table stage. Keep the answers as working notes and verify each against the source paper before it enters the manuscript.

Checkpoint 5: Every synthesis claim traces to at least one row in the evidence table, and no claim depends on a study excluded under the new criteria.

A revision is only auditable if someone else can follow it. Build the trail as you go, not at the end.

• Version the question and criteria with dates.
• Version the searches with source, date, strategy, and counts.
• Version the screening with stage counts and exclusion reasons.
• Version the synthesis so each draft points to the evidence table it was built from.

For longer revision work that spans searching, reading, and drafting, Scholar Agent is designed for extended research tasks with literature review use cases. If the revision is part of a larger effort with multiple outputs, Research Projects gives you a place to organize it as a project. Save or upload the papers you are keeping in My Library so the included set and the working set stay distinct.

Checkpoint 6: A reviewer can move from a sentence in the draft to the study row, to the screening decision, to the search that found it.

Suppose a review team is revising a review on a clinical topic. The original question asked about a broad population; a coauthor argues the review should focus on a narrower subgroup and add a new outcome.

1. Diagnose. The change touches population and outcome, so eligibility criteria change. This is a re-scope, not a re-frame.
2. Freeze. The team archives the old question, criteria, search strategies, counts, and synthesis in a dated folder.
3. Restate. The new question is written in structured form. The narrowed population and the added outcome are marked new; the comparison is marked unchanged; one old outcome is marked retired.
4. Rebuild the search. Population terms are tightened, outcome terms are added, and free-text terms are combined with subject headings. The strategy is run in more than one source, with dates and counts recorded. Search Details in PubMed confirm the concepts combined as intended.
5. Re-screen. Records are deduplicated, screened in stages, and full-text exclusions are tied to the new criteria. Two previously included studies are excluded because they fall outside the narrowed population; the reason is recorded against the population criterion.
6. Relink. Two reports of the same study are linked so the study counts once.
7. Rebuild the synthesis. The evidence table is rebuilt for the new population and outcome. The narrative is rewritten from the table. The retired outcome is removed from the results and from the discussion.
8. Rerun before submission. The searches are rerun to capture recent records, and the rerun date and counts are added to the trail.

At the end, the team can show a reviewer exactly which criterion change removed which study, and which search change found the new ones.

The revision is ready when all of the following are true:

• The new question and criteria are written, dated, and marked against the old version.
• The search was planned from the criteria, run in multiple sources, and documented well enough to rerun.
• Records were deduplicated, screened in stages, and exclusions carry reasons.
• Multiple reports of one study are linked.
• The synthesis was rebuilt from an evidence table, not patched from the old draft.
• Searches were rerun before publication, with the update recorded.
• A coauthor or reviewer can trace any claim back to a study and a search.

If any of these is missing, the revision is not finished, regardless of how polished the prose is.

Use these rules to decide when to pause and who to involve.

• Stop and escalate to the whole author team if the new question changes the review's scope enough that the target journal or audience no longer fits.
• Stop and escalate if coauthors disagree about eligibility criteria; resolve the criteria before any further screening, because screening against unsettled criteria produces decisions you will have to redo.
• Stop and re-plan the search if Search Details or your platform's translation view shows your concepts were combined differently from what you intended.
• Stop and re-check the evidence table if a synthesis claim cannot be traced to a row.
• Escalate to a librarian or search specialist if the new question requires sources or syntax your team does not routinely use.
• Do not submit until the pre-publication rerun is done and recorded.