When you have created an outcome, you can define the details of the analysis by modifying the properties of the outcome.
Tip: You can easily change some of the properties for an outcome directly by using the toolbar buttons at the top right corner of the content pane to toggle between the available Effect Measure and Analysis Model options.
To modify an outcome
Right-click the outcome in the outline and choose Properties.
Use the General tab to modify the Name, Data Type and Group labels.
Note that changing the data type will mean that any data already entered will be lost.
Use the Analysis Method tab to set the Statistical Method, Analysis Model and Effect Measure options.
Use the Analysis Details tab to specify options for Totals, confidence intervals, and some advanced options.
Use the Graph tab to specify options for the graph appearance.
Click OK when finished.
Note: Outcomes of the type Other Data have a separate set of properties. These only allow you to specify the Name, and the number of rows and columns in the data table.
Use the outcome Name to describe the outcome.
You can choose between the following Data types:
Dichotomous: number of events and participants in the two groups.
Continuous: number of participants, mean and standard deviation for the two groups.
O-E and Variance: number of events and participants, and pre-calculated observed – expected (O-E) and variance.
Generic Inverse Variance: pre-calculated estimate of treatment effect and standard error of estimate.
Other Data: data of all other types, entered in a free text table.
The Group Labels (1 and 2) are used as headings in the data tables, and are published on those analyses graphs that are included as figures. The default labels are 'Experimental' and 'Control'. You can edit the labels for an individual outcome on the outcome properties, but you can also change the labels used in all outcomes for a comparison, see Comparison properties.
The Statistical Method options available depend on the data type, and in turn determine what analysis model and effect measure options are available.
The options for Totals determine whether the analysis will show totals for the entire outcome, for sub-categories only, or not show any totals.
Study confidence interval and Total confidence interval will be used as the default values when the review is published in The Cochrane Library. If you want to specify the default confidence intervals to be applied to any new outcome in RevMan use Settings: Statistical.
Advanced options:
Test for subgroup differences.
When 'Totals and subtotals' is selected the test is always expected to be valid, and cannot be turned off . When 'Subtotals only' is selected the subgroups might not be actual subgroups but different analyses (e.g. different comparisons), so in you may wean to switch off the test. When 'No totals' is selected the test is not valid because it is using the totals in the calculations.
Swap event and non-event.
Entered data are on log scale (generic inverse variance only).
Ratio measures such as odds ratios, risk ratios and hazard ratios should be entered as ln(effect size) and SE[ln(effect size)]. Non-ratio measures such as mean differences and risk differences should be entered in their original metric. The data entered will be used in the meta-analysis. However, when the results are presented, the ratio measures will be converted to their original metric (odds ratio etc.) and the information entered here will determine whether this happens.
Enter number of participants (generic inverse variance, for display only).
Select this check box if you want to enter the number of participants in the two compared groups into the data table. These data are provided for display purposes only and are not used in the meta-analysis.
Left Graph Label, Right Graph Label are used beneath the analysis graph. The defaults are 'Favours [experimental]', and 'Favours [control]'. These must be modified to accurately reflect the intervention and control. You can edit the labels for an individual outcome on the outcome properties, but you can also change the labels used in all outcomes for a comparison, see Comparison properties.
Tip: Remember to reverse the labels if the outcome is positive.
You can use the Scale option to set the graph scale to any numerical value. You can also modify the scale by using the slider beneath the graph, see Analysis graphs.
You can choose the order in which the studies appear by using the Sort by option. If you choose User Defined order, you can use the Move up and Move Down buttons on the outline toolbar to reorder the studies.