This is an archived version of the Handbook. For the current version, please go to or search for this chapter here.

Table 16.4.a presents some results that might be available from a report of a cross-over trial, and presents the notation we will use in the subsequent sections. We review straightforward methods for approximating appropriate analyses of cross-over trials to obtain mean differences or standardized mean differences for use in meta-analysis. Review authors should consider whether imputing missing data is preferable to excluding cross-over trials completely from a meta-analysis. The trade-off will depend on the confidence that can be placed on the imputed numbers, and in the robustness of the meta-analysis result to a range of plausible imputed results.