T he highest ideal in medical practice is fully informed, evidence-driven practice. In pursuit of that ephemeral goal, clinical research in interventional cardiology is at an all-time high. One of the most rapidly growing domains of clinical research in interventional cardiology and other medical fields has been metaresearch-systematic reviews and the meta-analyses they feed. In this journal alone, for example, 13 meta-analyses were published between 2013 and 2015, and 7 of the 112 original articles that appeared in 2015 fell into this category.1-7 Given the proliferation of such studies, in this journal and in the field more broadly, it seems an opportune moment to reflect briefly on their place in the evolving literature on structural interventions.Although articles reporting the results of meta-analysis have become increasingly pervasive in recent years, the statistical methodology and discipline are not new, having been pioneered >40 years ago. Wielded judiciously and effectively, meta-analysis can be one of the most important tools available for integrating and demystifying clinical research. It is ideally suited for synthesizing data from multiple sources in the service of seeking clarity around important questions for which there is no clear answer-for example, when trials yield conflicting results, and there are multiple underpowered or unfocused studies. Targeted finely and executed expertly, meta-analysis can and should serve as some of the highest-level evidence for clinical decision making and for subsequent trial design. However, reality is not always in keeping with the ideal.There is increasing evidence that the meta-analysis machine is bloated. In the cardiovascular literature, both the publication of meta-analyses and their relative contribution to the literature expanded dramatically during the first decade of this century, with the number of meta-analyses increasing >500% from 2000 to 2012 and the ratio of randomized trials to meta-analyses falling from >10:1 to <3:1. 8 Although the sheer number of meta-analyses does not necessarily speak to appropriateness or necessity, there also seems to be increasing redundancy. For example, Siontis et al 9 found that, for two thirds of medical meta-analyses published in 2010, there was at least 1, and often multiple, overlapping meta-analysis on the same topic. To be sure, meta-analyses on the same topic can differ in important respects and provide unique and complementary value, but discordance between overlapping studies can muddy, rather than clarify, the question, belying the purpose of a meta-analysis. In an effort to streamline the pipeline of meta-analyses, creating transparency and reducing redundancy, investigators at the Center for Reviews and Dissemination at the University of York developed PROSPERO, a Web-based system for registration of systematic reviews that was activated 5 years ago.
10Overlapping and redundancy are not the only concerning observations. Just as important, there is the issue of quality. As Berlin and Golub 11 recently expl...