A Meta-analysis is a study which combines the results of other studies (usually drawn from the published literature) and synthesize their summaries and conclusions. It is often an overview of many clinical trials.

Researchers analyzed 48 meta-analyses and reviews and discovered that about half contained a poor or incomplete selection of published papers. As much as 20% of the reviews contained "strong indications" of what the researchers termed "publication bias". In some of the reviews, the publication bias was so great that it actually changed the results to the opposite conclusion of what was reached.

This type of bias occurs in meta-analyses because studies with results that are significant, interesting, or from large well-funded studies are more likely to be submitted, published, or published more rapidly than other work. Therefore, a meta-analysis based on a literature search will thus include such studies differentially, and the resulting bias may invalidate the conclusions.

The study's authors suggest that researchers should be more diligent in their study selection, in order to prevent or reduce the chances of such potential problems occurring in the future.

British Medical Journal 2000;320:1574-1577.