This is an archived version of the Handbook. For the current version, please go to training.cochrane.org/handbook/current.

A.8  Data and analyses

Results of studies included in a review are organized in a hierarchy: studies are nested within (optional) subgroups, which are nested within outcomes, which are nested within comparisons.

 

RevMan automatically generates forest plots illustrating data, effect estimates and results of meta-analyses (where selected) from the data entered into the ‘Data and analyses’ structure. The author is able to control whether, and how, meta-analyses are performed.

 

Note: The ‘Data and analyses’ should be considered as supplementary information because they may not appear in some formats of the published review. Key forest plots (containing data for each study) may be selected to be always included with the full text of the review by selecting them as figures (see Section A.9). The full published Cochrane review in the CDSR will, however, contain all of the ‘Data and analyses’ section as a series of forest plots or tables.

 

Authors should avoid listing comparisons or outcomes for which there are no data (i.e. have forest plots with no studies). Instead, authors should note in the text of the review that no data are available for the comparisons. The main outcomes in a review should be included in a ‘Summary of findings’ table irrespective of whether data are available from the included studies.

 

Comparison

The comparisons should correspond to the questions or hypotheses under ‘Objectives’.

 

Outcome

Five types of outcome data are possible: dichotomous data, continuous data, ‘O – E’ and ‘V’ statistics, generic inverse variance (estimate and standard error) and other data (text only).

 

Subgroup

Subgroups may relate to subsets of studies (for example, studies done before and after the publication of the CONSORT statement) or to a sub-division of the outcome (for example, short-term, medium-term, long-term).

 

Study data

Data for each study must be entered in a particular format specific to the type of outcome data (e.g. a sample size, mean and standard deviation for each group for continuous data).

See also