Low rate of protocol registration of systematic reviews published in high-impact-factor journals: a meta-epidemiological study




Poster session 4


Tuesday 25 October 2016 - 15:30 to 16:00


All authors in correct order:

Tsujimoto Y1, Tsujimoto H2, Kataoka Y3, Kimachi M1, Yamamoto Y1, Fukuhara S1
1 Department of Healthcare Epidemiology, Graduate School of Medicine and Public Health, Kyoto University, Japan
2 Department of Cell Growth and Differentiation, Graduate School of Medicine, Kyoto University, Japan
3 Hospital Care Research Unit, Department of Respiratory Medicine, Hyogo Prefectural Amagasaki General Medical Center, Japan
Presenting author and contact person

Presenting author:

Yasushi Tsujimoto

Contact person:

Abstract text
Background: Moher et al investigated the characteristics of systematic reviews (SRs) in 2007 and revealed that only 11.4% of non-Cochrane SRs used protocols. The idea of protocol registration to avoid publication bias of SRs was widely disseminated as the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement in 2009. The freely accessible international prospective register of systematic review (PROSPERO) database was launched for facilitation of protocol registration in 2011. Here, we investigated adherence to protocol registration of SRs after PRISMA statements.

Objectives: To assess the adherence to protocol registration in highly cited journals

Methods: We selected the top 10 journals classified by the Journal Citation Reports 2013 as the general internal medicine journals associated with the highest impact factors from August 2009 to June 2015.
We included SRs for any interventions. We excluded reviews that addressed diagnostic test accuracy, meta-epidemiology or were updates. We also excluded SRs that included non-randomized studies or observational studies and Cochrane SRs, since they have published protocols.
We assessed adherence to protocol registration as our primary outcome. Factors related to protocol registration were investigated using Fisher's exact test.

Results: We found 1584 articles and screened 438 full texts; 284 SRs were included for a detailed analysis. There were 224 (79%) non-registered or non-protocol-published SRs. Only 27 (9.5%) SRs were registered in PROSPERO, two (0.7%) SRs registered in other international registries and 31 (11%) SRs with published protocols. Protocol registration was associated with number of included studies of SRs (1-9,10-99, 99 <) (P for trend < 0.001), year (P for trend < 0.001) or financial support or grant (P = 0.002). Reference of PRISMA did not related to protocol registration (P = 0.77). Although 123 SRs referred to the PRISMA statement, only 27 (22%) of them registered the protocol.

Conclusion: Protocol registration of SRs is not still common six years after the PRISMA statement.