Reducing biases in systematic reviews
Just as biases can distort individual tests of treatments and lead to false conclusions, so they can also distort reviews of evidence. For example, researchers can simply ‘cherry pick’ those studies which they know will support the treatment claims they wish to make.
To avoid these problems, plans for systematic reviews, as for individual research studies, should be set out in research protocols. Protocols need to make clear what measures researchers will take to reduce biases and the effects of the play of chance during the process of preparing the reviews.
These will include specifying which questions about treatments the review will address; the criteria that make studies eligible for inclusion in the review; the ways in which potentially eligible studies will be identified; and the steps that will be taken to minimize biases in selecting studies for inclusion in the review, and for analysing the data.