Views towards publishing protocols for systematic reviews as peer-reviewed articles: an international survey of authors of non-Cochrane systematic reviews

Tags: Oral
Rombey T1, Puljak L2, Allers K3, Ruano J4, Pieper D1
1Institute for Research in Operative Medicine, Witten/Herdecke University, 2Center for Evidence-Based Medicine and Health Care, Catholic University of Croatia, 3Department of Health Services Research, Carl von Ossietzky University Oldenburg, 4Department of Dermatology, Reina Sofía University Hospital

Background: To ensure that systematic reviews (SRs) are truly systematic it is essential to define their methods a priori, e.g. by registering them in PROSPERO, the international prospective register for SRs. As PROSPERO records are not peer-reviewed or indexed in bibliographic databases, it might be useful to publish a protocol for the SR as a peer-reviewed article in addition to registering the SR in PROSPERO. For Cochrane reviews, publishing a protocol is mandatory, but for most non-Cochrane SRs it is optional.

Objectives: Our aim was to explore views of authors of non-Cochrane SRs registered in PROSPERO towards publishing SR protocols as peer-reviewed articles.

Methods: Contact persons of all PROSPERO records for non-Cochrane SRs registered in 2018 (N=12,531) were invited to participate in an anonymous 5-minute online survey that was administered through SurveyMonkey. The main question addressed SR authors’ views towards publishing SR protocols as peer-reviewed articles. Data were analyzed descriptively.

Results: In total, 4,223/12,531 (33.7%) invitees responded, of which 3,739/4,223 (88.5%) completed the survey. Almost half of the respondents had published or planned to publish a protocol for the SR described in their PROSPERO record as a peer-reviewed article (1,811/4,054; 44.7%). The remaining 2,243/4,054 (55.3%) respondents stated that there was no published protocol for their SR and they have not tried or plan to publish one. Of those, 66.4% (1,456/2,192) stated that there was an unpublished protocol that they followed during the conduct of their SR. The respondents had inconsistent views towards publishing protocols as peer-reviewed articles; most agreed that external feedback from peer-reviewers increases SR quality (2,899/3,739; 77.5%) but at the same time agreed that publishing a protocol in a peer-reviewed journal is not necessary if the SR is registered in PROSPERO (2,399/3,739; 64.2%). Respondents’ views towards acceptable manuscript processing times for SR protocols were consistent but far below actual manuscript processing times for SR protocols.

Conclusions: Although PROSPERO records are not peer-reviewed, many SR authors seem to consider registration in PROSPERO sufficient. Hence, awareness about the benefits of additionally publishing a SR protocol as a peer-reviewed article should be raised.

Our findings might be useful to various stakeholders of SRs; for example, they could support the decision-making of funding agencies on formally requiring published protocols, of SR authors on publishing a protocol as a peer-reviewed article, or of publishers on transforming the way manuscripts for protocols are being processed.

Patient or healthcare consumer involvement: As this was a research-on-research study, which had no health-related outcome, we did not involve patients or health care consumers. However, as this was a survey of SR authors, we actively involved the people that our research was about and is intended to ultimately benefit.