4 More systematic reviews are being registered in PROSPERO each year, but records are seldom up-to-date

Abstract

Objectives Prospective registration of systematic reviews (SRs) is useful as it increases transparency and decreases risk of bias. It also aims to prevent unintended duplication of SRs and facilitates collaboration. Thus, prospective registration is widely recommended in current guidance on SRs. Furthermore, like publishing a SR protocol, registration may increase the methodological quality of the SR. The international prospective register for SRs, PROSPERO, is the world’s first open-access online facility to register SRs. Registration is free of charge and records can easily be updated by the authors. Then a new version of the PROSPERO record is published. Our primary objective was to determine the proportion of SRs registered in PROSPERO and the influence of publication year, review focus and country on that proportion. Our secondary objectives were a) to compare the basic characteristics of registered and non-registered SRs and b) to assess the up-to-dateness of the PROSPERO records. Method We searched PubMed for SRs published between 2011 (launch of PROSPERO) and November 2018. We included a random sample of n = 300 SRs that would be eligible for registration in PROSPERO. We excluded scoping reviews, umbrella reviews/overviews and SRs, for which it is mandatory to publish a protocol, e.g. Cochrane reviews. We extracted data on the following items: Country, focus (therapeutic, epidemiologic, diagnostic, prognostic or other), number of authors, and date of submission, publication and last search. If the SR was registered in PROSPERO, we also extracted the number of versions, status, and date of first and last version. All data were analysed descriptively. Our primary analysis was based on the whole sample, while our secondary analyses were based on a) the registered and non-registered SRs and b) the identified PROSPERO records. Results Overall,15% (44/300) were registered in PROSPERO. This proportion has increased steadily over time and was 29% (15/52) in 2018. Therapeutic SRs were more frequently registered than SRs with another focus, with 18% (22/123) and 12% (22/177), respectively. We did not observe a clear trend regarding the country. Median number of authors and time from submission to publication did not differ between registered and non-registered SRs. Median time from search to submission was slightly shorter in registered than in non-registered SRs (14 vs 17 weeks), but median time from search to publication was longer (51 vs 43 weeks). Of the 44 analyzed PROSPERO records, most (82%) have been registered before the date of last search. For 23 records (52%) only one version existed, which means they have never been updated. Only six records (14%) were up-to-date, i.e. their status was ‘completed and published’ and a link to the publication was available. Conclusions More SRs are being registered in PROSPERO each year, but only few records are up-to-date. PROSPERO should be further promoted among prospective authors of SRs. Current users of PROSPERO need to be reminded to keep their records up-to-date. This could be done through PROSPERO or by journals when authors submit a manuscript for a SR.