Meta-analyses or collaborative studies.

Abstract

In decision making, it is often necessary to select between different strategies, and the best decision relies heavily on precise quantitative estimates of risks and benefits of the different strategies. To make such estimates, large samples are needed, but large studies are very expensive and difficult to conduct, especially studies that involve several centers, often in different countries. Before such a study is initiated, other options should be fully explored; if data already exist, it may be possible to combine results into a single estimate. A meta-analysis may be easy to do, but it is difficult to do well; sometimes, a meta-analysis may be grossly misleading. Some of the options and pitfalls in doing a meta-analysis are revealed in the text. The options for a meta-analysis should be considered before a megastudy is launched, and meta-analyses are often a useful way of introducing past experience into inference making.