Background: The association between prospective registration and the overall reporting and methodological quality of systematic reviews (SRs) has been investigated. The impact of prospective publication of protocols on publicly reviewed journals and prospective registration only on the quality of SRs and meta-analysis (MAs) may be different. However, whether prospective publication of protocols could improve overall reporting and methodological quality of SRs and MAs still remains unknown.
Objectives: The primary purpose of this study is to investigate whether the overall reporting and methodological quality of SRs and MAs with published protocols are superior to those of MAs only registered.
Methods: PubMed database will be searched to identified (paired) non-Cochrane SRs and MAs of randomized controlled trials (RCTs) published from 2015 to 2019 in English. The main search terms are including: “meta-analysis”, “meta-analyses”, “systematic review”, “systematic reviews” and “random*”. The rest of relevant reviews will be divided into two groups: non-Cochrane SRs and MAs with published protocols and those only registered when we finish titles and abstracts screening. For each group, the first 60 eligible studies from each group will be randomly selected. Key information (such as, first author, number of authors, year of publication, journal, types of disease, details of intervention and control, sample size, number of included RCTs, funding, etc.) of SRs and MAs will be extracted separately and cross-checked. Two independent reviewers will assess the reporting and methodological quality of included studies using PRISMA and R-AMSTAR. Any disagreements will be dissolved through discussion. The assessment method of reporting quality based on each PRISMA item will be: "1" point for each "yes", "0.5" for each "partial", and "0" point for any other responses ("no" and "cannot answer"). The assessment method of methodological quality based on each R-AMSTAR item will be: "1" point for zero or one criteria met, "4" points for all criteria met and other situations for “2" or “3” points. For each PRISMA item, odds ratio (OR) with 95% confidence interval (95% CI) will be calculated to compare the complete compliance of two groups. The mean difference (MD) and 95% CI will be calculated for each item of R-AMSTAR to compare the methodological quality of two groups. Univariable and multivariable linear regression analysis will also be used to explore basic characteristics (such as, number of authors, year of publication, sample size, number of included RCTs, funding, etc.) whether are associated with the overall reporting quality and met
Results: This study is ongoing, and the result of meta-epidemiological study will be submitted to a peer-reviewed journal for publication.
Conclusions: This study will provide comprehensive information on whether the overall reporting and methodological quality of SRs and MAs with published protocols are better than those of MAs only registered.
Patient or healthcare consumer involvement: Not applicable.