Meta-Analysis

Use of the Appropriate Guidelines for Reporting a Systematic Review

Guidelines for meta-analyses have been established and accepted, and following such accepted guidelines is a requirement for publication in many journals. For meta-analyses of RCTs, these guidelines are compiled in a statement called the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; see Table 1). Although this statement focuses on randomized trials, it can also be used for other designs, such as service evaluation, and consists of a checklist and flow diagram. For observational studies, the appropriate recommendations are compiled in the Meta-analyses of Observational Studies in Epidemiology (MOOSE) statement. A disadvantage of these tools is that they only apply to specific aspects and types of reviews, as opposed to being global guidelines (see Table 1). As a result, a Measurement Tool to Assess Systematic Reviews has been developed to evaluate systematic reviews that include both randomized and/or nonrandomized studies (see Table 1). Finally, the National Institute of Health has produced a tool that can be applied to a broad range of systematic reviews and meta-analyses whether these are of observational or experimental studies (see Table 1). Given the range of alternatives, it is advisable to pick the option that best suits a particular meta-analysis.

Prospective Registration of a Protocol

Authors of meta-analyses typically register the review’s protocol with an appropriate repository or registry. In the case of Cochrane reviews, this is a mandatory step prior to completion of the meta-analysis. For other meta-analyses, an alternative is PROSPERO (see Table 1). PROSPERO is an online repository of protocols similar to clinical trial registries or protocols in the Cochrane Library, hosted by the Centre for Reviews and Dissemination at the University of York. These are shorter than Cochrane protocols and cover the study design of interest, the condition under review, the target population, any comparison group, relevant outcomes, and any proposed intervention. The proposed search terms, methods of data extraction, and methods of assessing study quality are also required. All accepted meta-analyses are given a registration number, which is then cited in the published meta-analysis. Open science framework and some clinical trial registries also accept protocols for meta-analysis (see Table 1). Another option is researchregistry.com, which unlike the previously mentioned repositories, charges a fee for each registration.

...