Why we want systematic reviews (SRs) alive

SRs have changed medicine for the better and allow us to make better informed decisions with our patients. While essential, the time and resources required to perform and publish SRs are major drawbacks. By the time a review is published, the search is often at least a year old. It takes from 2.5 to 6.5 years once a randomised controlled trial (RCT) is published for it to be included in a new review.1 This is likely to apply for other areas too, such as diagnosis, prognosis and aetiology.

Having taken a long time to create, SRs then have a limited life span: new studies are published, and the review’s conclusions expire. Twenty-five per cent of SRs are obsolete within 2 years and 50% within 5 years.4 So, despite the resource-intensive process of SRs, their life expectancy is limited, especially in rapidly emerging research fields.

Traditional ‘dead’ SRs invite others to revisit the same topic and to conduct new SRs when new data become available. This leads to multiple parallel SRs on the same topic: in Achilles tendinopathy treatments, there are >40 SRs despite only 82 treatment RCTs.2 Multiple SRs make it hard for clinicians to know where to look for the latest information.

Is there a life support system for SRs? Yes—LSRs. LSRs can form a unique, widely accepted and up-to-date source for clinicians, patients and policy makers to inform the management of a condition.

Challenges of LSRs

An LSR aims to provide clinicians with a single, high-quality comprehensive overview of all the evidence. Good LSRs should negate the need for multiple other reviews on the same topic. This could optimise the capacity of the research community and reduce research waste. For researchers, there are various incentives to do an LSR but also a number of challenges (see table 1).

When planning LSRs, it is important to define at what point you would change the conclusions. In areas where conclusions are based on a few studies to date, the impact of new studies will be larger than areas with numerous high-quality trials. If the data are synthesised using a meta-analysis, then there are also methodological considerations; for example, frequently repeated analyses can lead to false-positive results,1 and certain statistical approaches are needed to account for this.

Another challenge for LSRs is that they are labour and resource intensive. They can be likened to ‘painting the Forth Rail Bridge’. Painting this famous massive iron railway bridge in Scotland was such a long job that, by the time it was completed, the paint at the start had started to peel off again! Moreover, the painters just started at the beginning again. The Sisyphean task was finally completed in 2011 when new paint from oil rigs offered a more lasting solution.5 Technological advances may make LSRs easier—software for managing the review process, machine learning for search screening and data extraction, crowd sourcing and online only publications.6

Staying alive: when to publish the next version

How often LSRs should be updated is unknown. Cochrane proposes that the search should be repeated on a monthly basis.7 In areas where large amounts of new research emerge in rapidly changing fields, like COVID-19, this may be appropriate. A review on prediction models in the BMJ has already been updated twice in 6 months due to a wealth of new research being published.8 The first version comprised 27 original studies. The number of studies increased to 51 in the first update and further to 107 in the latest update from July 2020. For a slower-moving field, like Achilles tendinopathy, where two to seven trials were published yearly between 2013 and 2017,2 an annual update may be adequate.

LSRs are still in their infancy, and there is no consensus as to how new versions should be published and disseminated. Postpublication revisions, where older versions are added as online supplements, seem a good way where all versions are easy to find.8 Another way to provide easy access is by publishing the latest findings on a website where they are directly available to healthcare providers and patients without the typical delay in peer review.2

LSRs provide up-to-date overviews of the available scientific literature to inform clinicians and patients when making decisions together. The design can avoid the frustrations about finding the newest and most trustworthy evidence for clinicians. If you want to know more about LSRs, check out this excellent series of six brief 2–5 min webinars from Cochrane7; read this classic open-access introduction to the topic1; or take a deep dive and read Cochrane’s manual on LSRs.9