In this interview we ask Rebecca Hodder from the National Health and Medical Research Centre Early Career Research Fellow, the University of Newcastle, Australia and Nicole Martin, Managing Editor, Cochrane Heart, University College London, London, UK to tell us more about this Living Systematic Review, which looks at increasing fruit and vegetable consumption in children under five years old.
Why was this review selected to be a Living Systematic Review (LSR) originally, was it in response to a target audience need?
Rebecca: The original review was initiated by the author team, who are embedded within the Hunter New England Local Health District, to inform the delivery of healthy eating health promotion services to children. The original review identified 5 studies and little evidence of effectiveness to inform what interventions could be implemented within the Local Health District to increase the fruit and vegetable consumption of young children.
For the first update of this review, the number of studies had substantially increased from 5 to 50 studies and it was during this update that we were approached to participate in a pilot study to maintain our review as an LSR.
Our review was considered appropriate for an LSR, given it met three key criteria:
First, the review question was considered a priority for decision making. There is a growing burden of disease internationally as a result of inadequate fruit and vegetable consumption, we know that childhood is a critical period during which lifelong healthy eating behaviours are established, and evidence from this review has the potential to inform international childhood obesity prevention interventions (as well as those within our Local Health District).
Second, the quality of the evidence for each of the comparisons from the original review was assessed as very low. This indicated there was a lack of certainty in the evidence and the results and the conclusions of the review were likely to change with the addition of new studies, if any were identified.
Third, on the basis of the identification of 45 new studies and a further 5 ongoing studies since the publication of the original review, there was new research evidence available that may change the conclusions and recommendations.
The current review update was published in November 2019. It includes 78 studies with a further 16 ongoing studies yet to be synthesised, and the quality of evidence for comparisons range from very low to moderate quality evidence. As a result it remains appropriate for an LSR and we will continued to maintain it as an LSR.
Nicole: This review meets the three criteria that make it appropriate for a review to be living (according to the Cochrane LSR Methods Guidance):
- The topic is a priority for decision making
- There is uncertainty in the existing evidence base
- Emerging evidence that may impact on the conclusions
This review has applied living systematic review methods since September 2017. Since then monthly searches are being run and screened and three subsequent updates were published – in January 2018, May 2018 and November 2019.
The increase in included and ongoing studies over time shows that this is a very active research area. The large disease burden based on insufficient consumption of fruit and vegetables means there is a need to incorporate new evidence when it emerges to adequately inform policy makers and the public.
Is the review in partnership with any other organisations?
Rebecca: Whilst the review is not explicitly conducted in partnership with other organisations, the author team hold positions or affiliations with the University of Newcastle, Hunter New England Local Health District and Hunter Medical Research Institute (Australia.) Author roles within Hunter New England Local Health District include delivering health promotion services, including those focused on healthy eating, to children via various settings including childcare services, hospitals and community health. Our LSR provides current evidence regarding which interventions are most likely to be effective for improving child heathy eating which can then be implemented within the Local Health District.
What does the review tell us?
Rebecca: Despite the large number of studies that have been conducted, now 78 in total, there is still limited evidence regarding effective ways to increase the fruit and vegetable consumption of young children. Of the types of approaches that have been investigated, multicomponent interventions (e.g. those that combine parent nutrition education with preschool nutrition policy changes), seem to have the most promise. The review found that multicomponent programs probably increase fruit and vegetable intake by children (by 0.36 cups per day) based on the moderate quality of the evidence from 14 studies. The review also found child feeding interventions may increase, and it was uncertain whether parent nutrition education interventions alone increased, child fruit and vegetable consumption.
How is working on/updating an LSR different to a ‘normal’ Cochrane review update, does this present any challenges/opportunities?
Rebecca: The main differences between maintaining an LSR and conducting a ‘normal’ Cochrane review update is the frequency of searching for and synthesising data from new studies.
For an LSR, database searches, screening of identified records, and data extraction from eligible studies are conducted on a monthly basis. Data from newly identified studies are then incorporated into the existing review; results and conclusions are updated; and the review update re-published. For our LSR we aim to re-publish the review every 3-4 months.
The main challenge of LSRs for an author team, is the ongoing availability of authors to undertake the required monthly and other tasks within a short time frame, and often concurrently. Early on we identified a team of authors that had expertise in child nutrition, the conduct of systematic reviews and ongoing capacity to contribute to our LSR. For the most part we have maintained this team of authors since the inception of our LSR. Additionally, we have used a number of tech-enablers to reduce the workload for authors during the screening stage, including a machine learning classifier of randomised controlled trials and Cochrane Crowd, which combined have more than halved the number of studies to be screened.
The RCT classifier (available in the CRE-Web) identifies which records from a database search are most likely to be RCTs (10-100% likely to be an RCT) and which are not (0-9% likely to be an RCT). For our LSR, those records likely to be RCTs were then screened by the author team against all review eligibility criteria, whereas those unlikely to be RCTs were sent to Cochrane Crowd to be screened. Any records identified as RCTs by Cochrane Crowd members were sent back to the author team to be screened.
Other key enablers for the conduct of LSRs includes the support and strong partnership with your Cochrane editorial team and access to LSR expertise. For our LSR, the Cochrane Heart Group have conducted the monthly database searches and facilitated an expedited timeline for editorial review and publication which has made our LSR possible. We have also had ongoing guidance and advice from many experienced members of the LSR Support team formed as part of Project Transform regarding the conduct of LSRs.
Our LSR provides an opportunity to generate the most up to date evidence for practitioners and policy-makers and inform international efforts to improve the dietary intake of young children. We are also currently investigating some other opportunities, including ways our LSR can be linked with child nutrition guidelines, and also partnering with health policy-makers to ensure the research evidence generated by our LSR is translated into policy and can achieve population wide benefits. There are also a number of other tech-enablers available or in development to assist with LSRs both within and beyond Cochrane, including Screen4Me.
Nicole: From an editorial prospective, the main challenge an LSR presents relates to time.
Our team’s Information Specialist works on compiling the latest search results on a monthly basis. While some of this is covered by search alerts, other databases have to be searched afresh each time. Deduplication of records against each month’s yield as well as all previously retrieved records also takes time. The numbers of search results to manage each month are relatively small, but the frequency of this task can be a challenge.
The time for the editorial review of LSR updates is much reduced compared to our usual processes. This requires careful planning and for everyone involved to be committed to very quick turn-around times - the editorial team, peer reviewers, editors, copy editors, sign-off editor and review authors.
For the latest update published on 7 November 2019 the first draft of the review update was submitted to us on 3 October 2019. Within just over a month, comments from everyone mentioned above were obtained and the authors addressed those in two separate rounds of revisions.
This highlights another prerequisite for an LSR – a high quality of the initial submission.
Some specific aspects are slightly different to the usual review process. We aim to involve peer reviewers repeatedly for LSR update cycles as familiarity with the review contents is thought to support a quicker response time. This was done for the publications in January and May 2018 and we aim to apply this again, with a new set of peer reviewers, to this current round of publications in November 2019 and the subsequent two during the first half of 2020.
A similar principle applies to copy editing in that the same copy editor looks at only those elements of the review that have changed compared to the previously published version.
How often will the LSR be updated?
Rebecca: During 2020 we will continue to maintain our review as an LSR. This involve ongoing monthly searches and identification of new studies, and re-publication of the review every 3-4 months. The ‘What’s new’ section of the review is updated every month to indicate how many new studies have been identified to date that will be synthesised in the next review update.