Table 1.
Key characteristics of the living systematic review pilot teams, processes and publication models
| Review topica | Anti-coagulation in people with cancer (3 related LSRs) | Fruit and vegetable consumption in children | Delayed antibiotics for respiratory infections | Zika virus and adverse neurological outcomes | Adherence to guidelines in traumatic brain injury | Epidemiology of traumatic brain injury |
|---|---|---|---|---|---|---|
| No. of authors maintaining LSRb | 4 | 2 | 2 | 4 | 2 | 4 |
| Search or other support | Information specialist to develop and run searches; ongoing LSR methods expert support | Librarian to develop searches only | Information specialist to develop and review searches only; ongoing LSR methods expert support | |||
| Direct funding for personnel | Yes (Part-time RA for authors) | Yes (Part-time RA for authors, stipend for editorial group) | No | Yes (Three funded positions [various roles] for authors) | Yes (Authors funded as part of broader work program, plus specific part-time LSR methods expert) | |
| Journal/Editorial Group; Publisher | Cochrane Gynaecological, Neuro-Oncology and Orphan Cancers; Cochrane | Cochrane Heart; Cochrane | Cochrane Acute Respiratory Infections; Cochrane | F1000Research; F1000 | Journal of Neurotrauma; Mary Ann Liebert, Inc. | |
| Search frequency | Monthlyc | Daily or monthlyd | Three-monthlye | |||
| Technological enablersf | Machine classifier and crowd-sourcing to identify RCTsg | Nil | Automation and machine learning algorithms to identify RCTs, with some data output automationh | Nil | ||
| Communicating review status to reader | Monthly statement to reader about review status (i.e. search date, new studies found, update plans) published in the ‘What’s New’ section of the review, via an article amendment | Daily updates (search date, new studies found) via study websitei | 3–6 monthly updates, describing results of new evidence found, available as supplementary material in online version of article | |||
| Editorial and peer review of status updates | No formal editorial review; no peer review | No formal editorial review; no peer review | Editorial review and copy-editing; no peer review | |||
| Process for integration of new evidence (citation/DOI status) | Full re-publication of review, with new citation and DOIj | New version of the review published, with linked citation and DOI (intended)k | Full re-publication of review, or short commentary article, with a new citation and DOI (intended)k | |||
| Editorial and peer review of new versions/publications | Standard editorial and peer review processes apply (may qualify for ‘selective’ peer review per Cochrane policy); same peer reviewers approached | Standard editorial and peer review processes apply; same peer reviewers approached (intended)k | Not confirmed (likely standard editorial and peer review processes apply) (intended)k | |||
| Trigger for integration of new evidence | When new evidence identified that changes review conclusions (intended)k | Every 4 months (irrespective of impact of new evidence) | When new evidence identified (irrespective of its impact) (intended)k | Every 6 months (irrespective of impact of new evidence) (intended)k | When new evidence identified that changes review conclusions, but no more frequently than yearly (intended)k | |
DOI digital object identifier, LSR living systematic review, RA research assistant, RCT randomised controlled trial
aEach team produced one LSR unless otherwise stated
bMeaning authors who contributed to the ongoing review tasks associated with maintaining the LSR (this may or may not have included the entire author team who contributed to the ‘baseline’ review)
cElectronic databases ± clinical trials registries all searched monthly, with remaining non-database sources, such as journal hand searching, websites and conference proceedings searched every 6 months
dDaily searches for PubMed, Embase and LILACS databases, with monthly searches for all other sources
eAll sources (including non-database sources) searched at this frequency
f‘Technological enablers’ refers to both computer technology and more efficient models of human contribution to increase the efficiency and sustainability of the systematic review enterprise (adapted from Thomas 2017 J Clin Epi 91:31-37)
gCochrane Crowd is a citizen science platform used to screen titles and abstracts to identify relevant citations (RCT’s) and exclude irrelevant citations. Machine classifiers exclude irrelevant citations automatically by using algorithms to predict how likely a new citation is to be describing a RCT
hSearches in some databases (PubMed, Embase and LILACS) are automated. De-duplication of citations is automated. Machine learning algorithm suggests a decision for inclusion based on title and abstract. All existing predefined tables and figures can be updated by running a script that re-renders these tables and figures
iStudy website is the Zika Open Access Project, available at: https://zika.ispm.unibe.ch/home
jThis process was implemented in one Cochrane Review only (Fruit and vegetable consumption in children). It was the intended process to be used in the remaining Cochrane Reviews but they did not reach the trigger for integration of new evidence, so their reviews were not re-published during the pilot period
k’Intended’ refers to the fact that this was the agreed process and/or trigger for integrating new evidence but that it was not undertaken during the pilot period