Meta-analysis: mostly a confidence trick?

Meta-analyses are a necessary tool whenever there are multiple trials testing the same issue. However, their use has been expanded beyond the purpose for which this instrument had been originally designed, with some positive aspects. For example, the possibility of extracting events from trials designed for different purposes, and of combining them in a formal meta-analysis, provides the opportunity of a greater insight on rare adverse events. Meta-analyses have also been used to estimate cardiovascular effects from non-cardiovascular trials; this procedure, although possibly interesting, risks to be misleading. Reliable meta-analyses can be founded only on reliable trials; in addition, some specific methodological issues can affect the results of many meta-analyses. For all those reasons, the interpretation of most meta-analytical data should be cautious.