Abstract
Objective: To investigate the differences in main characteristics, reporting and methodological quality between prospectively registered and non-registered systematic reviews.
Methods: PubMed was searched to identify systematic reviews of randomized controlled trials published in 2015 in English. After title and abstract screening, potentially relevant reviews were divided into three groups: registered non-Cochrane reviews, Cochrane reviews, and non-registered reviews. For each group, random number tables were generated in Microsoft Excel, and the first 50 eligible studies from each group were randomly selected. Data of interest from systematic reviews were extracted. Regression analyses were conducted to explore the association between total R-AMSTAR or PRISMA scores and the selected characteristics of systematic reviews.
Results: The conducting and reporting of literature search in registered reviews were superior to non-registered reviews. Differences in nine of the 11 R-AMSTAR items were statistically significant between registered and non-registered reviews. The total R-AMSTAR score of registered reviews was higher than non-registered reviews (MD=4.82, 95%CI: 3.70, 5.94). Sensitivity analysis by excluding the registration related item presented similar result (MD=4.34, 95%CI: 3.28, 5.40). Total PRISMA scores of registered reviews were significantly higher than non-registered reviews (all reviews: MD=1.47, 95%CI: 0.64-2.30; non-Cochrane reviews: MD=1.49, 95%CI: 0.56-2.42). However, the difference in the total PRISMA score was no longer statistically significant after excluding the item related to registration (item 5). Regression analyses showed similar results.
Conclusions: Prospective registration may at least indirectly improve the overall methodological quality of systematic reviews, although its impact on the overall reporting quality was not significant.
Methods: PubMed was searched to identify systematic reviews of randomized controlled trials published in 2015 in English. After title and abstract screening, potentially relevant reviews were divided into three groups: registered non-Cochrane reviews, Cochrane reviews, and non-registered reviews. For each group, random number tables were generated in Microsoft Excel, and the first 50 eligible studies from each group were randomly selected. Data of interest from systematic reviews were extracted. Regression analyses were conducted to explore the association between total R-AMSTAR or PRISMA scores and the selected characteristics of systematic reviews.
Results: The conducting and reporting of literature search in registered reviews were superior to non-registered reviews. Differences in nine of the 11 R-AMSTAR items were statistically significant between registered and non-registered reviews. The total R-AMSTAR score of registered reviews was higher than non-registered reviews (MD=4.82, 95%CI: 3.70, 5.94). Sensitivity analysis by excluding the registration related item presented similar result (MD=4.34, 95%CI: 3.28, 5.40). Total PRISMA scores of registered reviews were significantly higher than non-registered reviews (all reviews: MD=1.47, 95%CI: 0.64-2.30; non-Cochrane reviews: MD=1.49, 95%CI: 0.56-2.42). However, the difference in the total PRISMA score was no longer statistically significant after excluding the item related to registration (item 5). Regression analyses showed similar results.
Conclusions: Prospective registration may at least indirectly improve the overall methodological quality of systematic reviews, although its impact on the overall reporting quality was not significant.
| Original language | English |
|---|---|
| Pages (from-to) | 45-55 |
| Journal | Journal of Clinical Epidemiology |
| Volume | 93 |
| Early online date | 31 Oct 2017 |
| DOIs | |
| Publication status | Published - Jan 2018 |
Keywords
- Prospective registration
- Quality
- Systematic reviews
- Meta-epidemiological study