Living Systematic Reviews: Keeping Evidence Current

A traditional systematic review is a snapshot. It answers a question using the studies available on the day the searches were run, and from that moment it begins to age. In fast-moving fields the gap between the search date and publication can already be long enough for important new trials to appear, and within a few years many reviews are out of date. Living systematic reviews were developed to close that gap. A living review is continually updated, incorporating relevant new evidence as it becomes available, so that readers always see a current synthesis. This guide explains what living reviews are, when they are worth the effort, how to plan and run one, how to communicate changes, and how a source-linked assistant such as Yila can help you screen and summarise new studies without losing track of where each finding came from.

What makes a review living

A living systematic review uses the same rigorous methods as a conventional review: a protocol, comprehensive searches, explicit eligibility criteria, duplicate screening, risk of bias assessment and transparent synthesis. What makes it living is the commitment to continual surveillance and updating. Searches are rerun frequently, often monthly, new studies are screened promptly, and the review is revised whenever new evidence could change its conclusions.

The living approach is not simply a review that is updated more often than usual. It involves a defined update frequency, explicit triggers for incorporating new evidence, a plan for when the review will stop living, and a publication format that allows readers to see what changed and when. Without these features, a review that is updated occasionally is better described as an updated review.

When a living review is worth it

Living reviews demand sustained effort, so they should be reserved for questions where they add real value. Three conditions are commonly cited. The question should be a priority for decision making, so that current evidence matters to patients, clinicians or policymakers. There should be important uncertainty in the existing evidence, so that new studies could plausibly change conclusions. And new research should be likely to emerge at a pace that makes frequent updating worthwhile.

The COVID-19 pandemic illustrated all three conditions. Treatments and vaccines were urgent priorities, early evidence was highly uncertain, and trials were reporting weekly. Living reviews and living guidelines became essential tools for keeping recommendations aligned with a rapidly changing evidence base. In more stable fields, where few new trials appear each year, a conventional review updated every few years may be entirely adequate.

Planning the protocol

The protocol for a living review should specify everything a conventional protocol includes, plus the arrangements for keeping the review alive. These include how often searches will be run, how quickly new studies will be screened, what will trigger an update to the published review, how changes will be communicated and under what circumstances the review will be retired.

Update triggers deserve careful thought. Some teams update the published review whenever any new eligible study is found. Others update only when new evidence could change the conclusions, for example when a large trial is published or when the certainty of evidence for a key outcome could shift. A clear rule prevents both endless minor revisions and long delays in incorporating important findings.

The protocol should also plan for retirement. A review may stop living when the question is no longer a priority, when the evidence has become sufficiently certain that new studies are unlikely to change conclusions, or when funding ends. Stating these criteria in advance makes the decision transparent.

Searching continually

Continual searching is the engine of a living review. Search strategies are developed and validated as for any systematic review and then rerun at the planned interval. Many teams set up automated alerts from bibliographic databases and trial registries so that new records arrive without manual effort. Preprint servers may be included where timely evidence matters, with clear rules about how preprints are handled when the peer reviewed version appears.

Search strategies should be reviewed periodically. Terminology evolves, new interventions appear and new databases become relevant. A strategy that was comprehensive at the start may miss important studies a year later if it is never revisited.

Screening and extraction at pace

The volume of records in a living review is usually modest in any single month, but the work never stops. Teams need a workflow that allows prompt screening and extraction without burning people out. Machine learning classifiers can help by ranking records by likely relevance or by excluding clearly irrelevant ones, such as non-randomised designs when only trials are eligible. Crowdsourcing platforms have also been used to screen records quickly.

A source-linked assistant can support the work by summarising new studies with links to the passages that support each extracted detail. When a new trial arrives, a reviewer can quickly see its design, population, interventions and outcomes, and then check the original report to confirm. Every extracted value should be verified against the source before it enters the analysis, because errors in a living review propagate through every subsequent version.

Updating the synthesis

When new studies are added, the meta-analyses, risk of bias assessments and certainty of evidence ratings must be revisited. Repeated meta-analysis raises a statistical concern: testing the same question again and again increases the chance of a spurious finding. Some living reviews use methods such as trial sequential analysis or Bayesian approaches to account for repeated updating, while others rely on GRADE assessments that consider imprecision and the totality of evidence. Whatever the approach, it should be specified in the protocol.

The certainty of evidence should be reassessed with each meaningful update. A conclusion that rested on low certainty evidence may become moderate or high as larger trials report, or an early promising effect may shrink as more data accumulate. Readers need to understand not only the current estimate but how confident they can be in it.

Communicating changes

Readers of a living review need to know what is new. Clear version histories, dated updates and summaries of what changed help them see whether conclusions have shifted. Many living reviews publish a short update note with each version, describing new studies added, changes to effect estimates and any changes to conclusions or certainty.

Publication formats matter. Some journals and platforms support versioned publication, where each update is linked to previous versions and has its own citation. Others publish living reviews as a series of articles. Online summaries and dashboards can make the current state of evidence accessible to decision makers between formal updates.

Linking to living guidelines

Living reviews are most powerful when they feed directly into living guidelines. When a guideline panel receives updated evidence promptly, recommendations can be revised quickly rather than waiting years for a full guideline update. This requires close coordination between the review team and the guideline developers, shared standards for evidence assessment and clear processes for deciding when new evidence warrants a change in recommendation.

Resourcing and sustainability

The greatest challenge for living reviews is sustainability. Conventional reviews are funded as projects with a defined end, whereas living reviews require ongoing staffing and infrastructure. Teams should plan for turnover, document their processes thoroughly and use shared tools so that new members can contribute quickly. Automating routine tasks, such as running searches and deduplicating records, frees people for the judgement-intensive work of appraisal and synthesis.

Funders are increasingly recognising the value of living evidence, but secure long-term funding remains rare. Being explicit about the resources needed, and about the retirement criteria, helps funders and teams make realistic commitments.

An illustrative example

Imagine a team maintaining a living review of drug treatments for a common respiratory infection. Each month, automated searches return around two hundred new records. A classifier removes obvious non-trials, two reviewers screen the remainder within a week, and any eligible trial is extracted and appraised within the following fortnight. If a new trial could change the certainty of evidence for mortality, the team updates the meta-analysis and publishes a new version with a short note explaining what changed. Twice a year the team reviews its search strategy and checks whether the question still meets the criteria for remaining living.

Common pitfalls

Living reviews fail in predictable ways. Teams underestimate the workload and fall behind, so the review stops being current while still claiming to be living. Update triggers are vague, leading to inconsistent decisions. Version histories are unclear, so readers cannot tell what changed. Repeated analyses are conducted without considering the statistical implications. And reviews continue living long after the question has been settled, consuming effort that could be used elsewhere.

Conclusion

Living systematic reviews keep evidence synthesis aligned with the pace of research in fields where current answers matter. They require the same rigour as conventional reviews, plus clear plans for continual searching, prompt screening, update triggers, transparent communication and eventual retirement. Tools that summarise new studies with links to their sources can lighten the workload, provided every extracted detail is checked against the original report. Used for the right questions and resourced realistically, living reviews give decision makers what they most need: the best available evidence, today.