Ricerc@Sapienza

In the era of evidence-based medicine, systematic reviews and meta-analyses are considered at the top of evidence hierarchy. Despite the almost exponential increase in the number of published meta-analyses over the course of the last decades, only a small minority of them is of high quality, with major flaws involving every aspect of the meta-analytic process. The strength of a meta-analysis is closely linked to the quality of the included studies. Once preliminary phases are completed, it is vital that selected papers undergo a thorough quality assessment, using the most appropriate tools among those available. Analytical approaches must be tailored to the nature of the extracted data and the specific purpose of the meta-analysis. Appraisal of heterogeneity is a key step to inform subgroup or meta-regression analyses. The solidity of the results of the main analysis (especially in meta-analyses of observational studies or studies with high heterogeneity) should be tested by means of pertinent sensitivity analyses. Finally, the investigators should be aware of the possibility of publication bias and make efforts to assess it using qualitative and quantitative methods. The aim of the second part of this expert review is to provide guidance on how to appropriately perform trial level meta-analyses, with particular focus on the quality assessment of the included studies, the choice of the appropriate statistical approach, the methods to deal with heterogeneity (including subgroup, meta-regression and sensitivity analyses), and the appraisal of publication bias.