Association of protocol registration with methodological and reporting quality of systematic reviews and meta-analyses in medicine and health sciences fields in Ethiopia

Abstract

Background: With the exponential growth of published systematic reviews (SR) and meta-analyses (MA), there is a high potential for duplication of work and compromising quality. The national institute for health research (NIHR) recommend registration of SR and MA, which can reduce research waste and increase transparency and research quality. This study aimed to investigate the rate of protocol registration and its association with methodological and reporting quality.Methods: We searched SR and MA indexed in PubMed, EMABSE, Scopus and CINAHL databases until March 16, 2021. All SR and MA in medicine and health science fields associated with Ethiopia irrespective of the place of publication, authors' affiliation and design were included. Data on the background, methodological, and reporting characteristics were extracted. We applied Chi-Square or Fisher’s Exact test for categorical variables and Kruskal–Wallis test for continuous variables at alpha level 0.05 to examine association. All analyses were done using R 4.1.3 for macOS.Results: From the total of 349 SR and MA included, 100 (28.65%) registered their protocol. Significantly large proportion of SR and MA with registered protocol reported/conducted last search date (71.0%, p<0.05), duplicate screening (72.0%, p<0.005), duplicate data extraction (83.0%, p<0.005) and duplicate quality assessment (84.0%, p<0.005), and search within two years of publication (78.0%, p<0.05). Additionally, significantly large proportion of SR and MA with registered protocol used Joanna Briggs Institute's quality assessment tool (58.0%, p<0.05) and used Egger's test (89%, p<0.05) to assess publication bias. Moreover, SR and MA with registered protocol were less frequently cited (median = 3.50, p <0.001) and included a smaller number of authors (median = 4.00, p=0.007) than those without registered protocol.Conclusions: Nearly one-third of published SR and MA in medicine and health sciences fields in Ethiopia registered their protocol. SR and MA with registered protocol showed better methodological and reporting quality compared to those without registered protocol, but not for all quality criteria. To improve rate of prospective registration and ensure quality of SR and MA in medicine and health science fields, researchers should be trained in all aspects of SR and MA, and journals, funding agencies, governments and universities should encourage registration of SR and MA as an important indicator of transparency in research.

Ähnliche Arbeiten