A statistical technique that combines the results of several studies addressing a set of related research hypotheses. The results of the included studies are analyzed as if they were the results of one large study.
The quality of a meta-analysis strongly depends on the quality (methodological soundness) of the included studies. A (good) meta-analysis of poorly designed studies will result in bad statistics, results, and conclusions.
Some (bad) meta-analyses can be misleading as they inaccurately pool data from primary prevention (disease prevention in healthy people) and secondary prevention (slowing down disease progression in patients) studies.
Additionally, a meta-analysis can be strongly biased by the scientists’ heavy reliance on published studies: as non-convincing (significant) studies often end up not getting published, the authors may tend to “select” data to get significant publishable results.