Methods used in Cochrane reviews

Study selection

Generally speaking the editorial group expect the inclusion of studies to be confined to controlled clinical trials and randomised controlled trials. However, the scope of the group is such that there may be problems where interventions are difficult to evaluate using experimental methodology and, in these cases, reviewers are asked to make the case for extending inclusion criteria to other forms of study design in the development of the protocol.

The group requires the inclusion criteria developed for a review to be applied independently by at least two reviewers for each potentially relevant study. Disagreements that cannot easily be resolved are referred to a third person, and/or a member of the editorial team. Studies that, in the judgement of the reviewers, are seriously compromised by weaknesses in design, conduct or analysis are excluded. Details of such studies are reported in the excluded studies table within reviews.

Assessment of methodological quality

See MECIR conduct and reporting standards.

Data collection

The policy of the CDPLP is that all reviewers should develop a data extraction sheet tailored to their own reviews (see here for templates). At least two reviewers should extract data independently for each review. In the event the report of a trial lacks relevant information (method of randomisation, data) authors should be contacted for further details. Reviewers should seek help from the editorial base in contacting authors if this proves difficult.

Analysis

Core guidance is available in the Cochrane Handbook for Systematic Reviews of Interventions.