Core PRISMA and PRISMA-P elements addressing statistical considerations for systematic reviews
| Statistical considerations for reporting, with guidance regarding core details | Defined PRISMA-P item? | Defined PRISMA core item? |
|---|---|---|
| Describe criteria under which study data will be quantitatively synthesised. Outline considerations that will be undertaken by the research team to make decisions regarding whether formal meta-analyses of end point data will be performed. This should include clear mention of factors related to study population, intervention, comparators and outcomes of included studies. | √ | √ |
| Describe planned summary measures for meta-analyses and evaluations of heterogeneity. Specify considerations for the combining of study data using meta-analysis including type of model (random vs fixed effects), approach (eg, generic inverse variance, Mantel Haensel), effect measures (eg, mean differences, ORs) as well as plans for the assessment of statistical heterogeneity (eg, I2), handling of multiarm studies, and considerations for missing data. Any data transformations to be undertaken should be mentioned. | √ | √ |
| Address performance of additional analyses (eg, subgroup/sensitivity analyses). Describe additional analyses beyond the primary analysis that explore the robustness of findings and/or assess the potential for different effects in certain types of patients or studies. This includes subgroup analyses, sensitivity analyses, metaregressions and other such data analyses performed secondary to the primary analyses. Distinctions between prespecified and post hoc analyses should be made. | √ | √ |
| Describe exploration of metabiases. Describe any evaluations performed to address the potential for risk of bias across studies (eg, publication bias via funnel plots or statistical tests, selective reporting via inspection of trial protocols). Statistical details of the evaluations performed (eg, graphical components or statistical tests performed) should be described, while reasons any planned analyses were not performed should also be clearly stated (eg, minimal studies available, etc). | √ | √ |
| Describe study selection process. For transparency of the systematic process followed to identify studies, document the numbers of studies screened for eligibility at the title/abstract level and the full-text level. Provide a summary of the final number of studies included qualitatively and quantitatively, as well as a description of reasons for exclusion from full-text screening. Provision of an illustrative flow diagram is recommended. | √ | |
| Describe study characteristics. Provide characteristics of relevance extracted from included studies. A narrative description capturing the extent of similarities and differences in key features of study methods, interventions and patient populations should also be provided, including tabulations of the numbers of studies possessing characteristics of key relevance to the research question. | √ | |
| Present results of individual studies. Present both simple summary data (eg, number of events and patients for dichotomous measures or means, SD and sample size for continuous end points) as well as summary effect estimates (including CIs) for each study and end point. If feasible, use of forest plots is recommended. This enhances reproducibility, inspection of variations across studies, and enhances the ability to identify data collection errors. Specifics to report for different end point types and strategies to make use of web appendices are discussed in the PRISMA elaborations document. | √ | |
| Present findings from meta-analyses. Report summary estimates from all meta-analyses performed in the review along with CIs estimates of between-study heterogeneity assessed. Ensure that the numbers of studies and patients contributing to each synthesis are clear. Use of forest plots is recommended. | √ |
PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis; PRISMA-P, PRISMA for Protocols.