Systematic reviews of published evidence: miracles or minefields?

The review article has been a regular feature of medical journals for many years. As most decisions in health care cannot be fully informed by consulting the results of a single empirical study, review articles are needed to provide an overview of all the relevant findings on a particular topic, or within a given field. Traditionally such articles have been written in a narrative style by subject experts, and have described the epidemiology, diagnosis, treatment and the likely outcomes of specific diseases and conditions. As was summarised in the previous article in this series [1], researchers in the late 1980s and early 1990s revealed that"whilst such reviews were educational, they were commonly incomplete, opinionated and selective in the data that they referenced [2]. For these reasons they are now thought not to provide a sound basis for treatment or other patient management decisions. In recent years, in favour of a more rigorous approach, the systematic review has emerged in health care publications as an identifiable entity [3], aiming to produce a comprehensive and reliable overview of the results of all available relevant studies.