Keeping a Review Up to Date: Living Evidence Workflows

A traditional review is a snapshot. It answers a question using the evidence available on the day the searches were run, and then it slowly goes out of date. In fast-moving fields, a review can be obsolete within months of publication. Living evidence approaches address this by updating continually as new studies appear. Even if you are not running a formal living systematic review, the underlying workflow ideas can help any team that needs to keep a body of evidence current, whether for clinical guidance, policy, product strategy or internal knowledge bases.

What makes a review "living"

A living review is continually updated, incorporating relevant new evidence as it becomes available. In practice this means:

• searches are re-run frequently, often monthly;
• new records are screened promptly;
• new studies are added to the synthesis when they meet inclusion criteria;
• conclusions are revised when the new evidence changes them;
• the current version is published openly, with a change history.

Living reviews make most sense when three conditions hold: the question is a priority for decision-making, the evidence is uncertain enough that new studies could change conclusions, and new research is being published at a meaningful rate.

Designing for updates from the start

Retrofitting a living workflow onto a finished review is painful. It is far easier if the original review is designed with updating in mind.

Stable search strategies. Document every search exactly, including platform, syntax and limits, so it can be re-run identically. Set up saved searches and alerts in each database.

Structured data. Store extracted data in a structured format, such as a spreadsheet or database with consistent fields, rather than in prose. Adding a new study then means adding a row, not rewriting paragraphs.

Reproducible analysis. If you pool results, script the analysis so that adding a study and re-running it takes minutes.

Modular writing. Write findings in sections that correspond to specific questions or outcomes, so that only affected sections need revision.

Traceable claims. Keep each statement in the review linked to the studies and passages that support it. When evidence changes, you can see exactly which statements are affected.

A practical update cycle

A monthly cycle might look like this:

1. Week 1: search. Re-run saved searches limited to records added since the last search date. Check registries and preprint servers.
2. Week 1 to 2: screen. Deduplicate against the existing library and screen new records using the same criteria and decision rules as the original review.
3. Week 2 to 3: extract and appraise. Extract data and assess risk of bias for newly included studies.
4. Week 3: re-analyse. Update tables, meta-analyses and certainty ratings.
5. Week 4: decide and publish. Decide whether conclusions have changed. If they have, update the text and key messages. If not, publish a short note stating that the review was updated and conclusions remain unchanged.

Not every cycle needs a full republication. Many living reviews distinguish between minor updates, where new studies do not change conclusions, and major updates, where they do.

Deciding when conclusions change

A living review needs explicit rules for when to revise its conclusions. Without them, conclusions may shift with every small study, or never shift at all. Useful triggers include:

• a change in the direction of an effect estimate;
• a change in statistical or practical significance;
• a change in the certainty of evidence rating;
• new evidence about harms;
• a new study large enough to dominate the pooled estimate.

Write these triggers into the protocol and apply them consistently.

Versioning and transparency

Readers of a living review need to know which version they are reading and what has changed. Good practice includes:

• version numbers and dates on every release;
• a change log listing new studies, revised analyses and altered conclusions;
• access to previous versions;
• clear statements of the search date for each version.

This transparency also protects authors: if a decision was made using an earlier version, the record shows what the evidence looked like at that time.

Sustaining the effort

The biggest risk to a living review is not methodological but organisational: the team runs out of time or funding and updates stop. Plan for sustainability by:

• automating what can be automated, such as search alerts and deduplication;
• using screening tools with machine-learning prioritisation where appropriate;
• distributing tasks across a team rather than relying on one person;
• defining a retirement rule, for example ending living status when evidence becomes stable or the question loses priority.

Living evidence beyond formal reviews

The same principles apply to many everyday knowledge tasks. A product team tracking competitor features, a policy unit monitoring evaluations of a programme or a research group maintaining a reading list for a lab can all benefit from stable searches, structured data, modular writing and a regular update cycle.

The key enabler is traceability. If every statement in a document is linked to its supporting sources, updating becomes a targeted operation instead of a rewrite. Research workbenches such as Yila (https://yila.ai) are designed around this idea, keeping summaries and comparison tables connected to the passages they rely on, so that when a source changes or a new one arrives, it is clear which conclusions need to be revisited.

Communicating updates to users

People who rely on a living review need to know when something important changes. Offer a simple subscription, such as an email list or a feed, that announces new versions with a one-paragraph summary of what changed and whether conclusions moved. Keep the summary at the top of the review itself as well. For guidance producers who build recommendations on the review, agree in advance how quickly they will be notified of major updates, so that recommendations do not lag behind the evidence for months.

Common pitfalls

• Criteria drift. Over many update cycles, screening criteria can quietly change. Re-read the decision rules before each cycle.
• Alert fatigue. Too many alerts lead to skipped updates. Tune searches to keep volumes manageable.
• Untracked changes. Small edits between versions should still appear in the change log.
• Overreacting to single studies. Use the pre-defined triggers rather than revising conclusions on impulse.

Conclusion

Living evidence workflows keep a review useful long after its first publication. Design for updates from the start, run a regular and documented cycle, define clear rules for when conclusions change, version everything and keep claims traceable to their sources. The result is a body of evidence that stays current, and that decision-makers can rely on as the research landscape moves.