Living Systematic Review

A living systematic review is a systematic review that is continually updated, incorporating new evidence as it becomes available rather than remaining a static document fixed at the time of publication. It maintains the methodological rigor of a traditional systematic review while adopting an ongoing surveillance and update process to keep the conclusions current.

Traditional systematic reviews represent the evidence at a single point in time, and their conclusions can become outdated within months in rapidly evolving fields. During the COVID 19 pandemic, for example, new clinical trials were published weekly, and static reviews could become outdated within weeks. Living systematic reviews address this problem by establishing a process for regular monitoring, screening, and incorporation of new studies, so that clinicians, policymakers, and guideline developers can draw on evidence that is current to the latest update. The Cochrane Collaboration, the World Health Organization, and several national guideline bodies have adopted living review methodology for high priority topics.

A living systematic review begins with a standard systematic review: a registered protocol, a comprehensive search, screening, quality assessment, data extraction, and synthesis. What distinguishes it is the explicit commitment to ongoing updates at predefined intervals (such as monthly or quarterly) or in response to triggers (such as the publication of a new large trial).

The update process involves running the search strategy at regular intervals to identify new studies published since the last update. New records are screened, appraised, and incorporated into the synthesis following the same methods as the original review. Statistical analyses are updated to include the new data, and the conclusions are revised if warranted. Each update is clearly versioned and dated, with a transparent record of what changed and why.

Several methodological and practical considerations distinguish living reviews from traditional ones. Statistical considerations include the risk of inflating the Type I error rate through repeated analyses. Sequential methods, analogous to interim analysis in clinical trials, have been proposed to address this, but their use is debated: an expert panel of Cochrane's Scientific Committee recommended against using them for the main analyses or main conclusions of updated Cochrane meta analyses in most circumstances, preferring to interpret each update by the size of the estimated effect and its uncertainty. Practical considerations include the need for sustained funding and staffing, the development of efficient workflows for monitoring and screening, and the establishment of criteria for when the review should transition from "living" to "static" (for example, when the research question is settled, when new evidence is no longer expected, or when resources are no longer available).

Technology plays an important role in making living reviews feasible. Automated search alerts, machine learning tools for screening, and living evidence platforms that maintain up to date databases all reduce the workload associated with continuous updating. Authorship and governance structures must be defined to ensure accountability across successive updates, particularly when team members change over time.

Living systematic reviews are most valuable when the research question is a high priority for clinical or policy decision making, when the evidence base is rapidly evolving, when current evidence is uncertain and new studies could change the conclusions, and when there is a commitment of resources to sustain the update process. They are less appropriate when the evidence base is mature and unlikely to change, when the topic is of low priority, or when resources for ongoing monitoring and updating are unavailable.

A common error is labeling a review as "living" without establishing the infrastructure for regular updates, resulting in a standard review with an aspirational label. Another mistake is updating the review at irregular or infrequent intervals, allowing gaps during which important new evidence is missed. Some teams fail to address the statistical implications of repeated analyses, potentially increasing the risk of false positive conclusions. Governance failures, including unclear authorship criteria, lack of funding continuity, and absence of criteria for transitioning to a static review, can cause living reviews to stall or produce inconsistent updates. Finally, not clearly versioning and dating each update makes it difficult for readers to know which version of the review they are using.

A hypothetical living systematic review of antivirals for COVID 19 is established in early 2020. The team runs monthly searches across MEDLINE, Embase, and the WHO COVID 19 database. Over 18 months, the review is updated 14 times, incorporating 63 new RCTs as they are published. Each version is numbered and dated (v1.0 for the original review, then v2.0 through v15.0 for the 14 updates), and three updates lead to changes in the summary conclusions (for example, when two large trials of a newer antiviral are published). Because the meta analyses are repeated at every update, the protocol states in advance how this will be handled: in line with Cochrane guidance, the team bases its conclusions on the size of the estimated effect and its uncertainty rather than on whether each new analysis crosses a significance threshold.