9.4.10  A summary of meta-analysis methods available in RevMan

Table 9.4.a lists the options for statistical analysis that are available in RevMan. RevMan requires the author to select one preferred method for each outcome. If these are not specified then the software defaults to the fixed-effect Mantel-Haenszel odds ratio for dichotomous outcomes, the fixed-effect mean difference for continuous outcomes and the fixed-effect model for generic inverse-variance outcomes. It is important that authors make it clear which method they are using when results are presented in the text of a review, since it cannot be guaranteed that a meta-analysis displayed to the user will coincide with the selected preferred method.