Introduction/Objective: The central collection of updated individual patient data (IPD) may represent a "gold standard" for the conduct of systematic reviews but for some it may lie in the "realms of luxury" because of the time and resources required. Empirical evidence is needed to help determine the relative value of this and other techniques for systematic reviews.

Methods: This presentation is the first step in a systematic review of studies comparing the results of meta-analyses based on IPD versus meta-analyses based on published data alone. It will be expanded to produce a full systematic review, which could be considered for inclusion in the methodology database on The Cochrane Library.

Results: There are at least five published comparisons of IPD versus published data meta-analysis. These have shown that the results of an IPD meta-analysis can be importantly different from those of a meta-analysis based solely on published data. However, the direction and size of the difference is not consistent between comparisons. Among the reasons for the differences are the inclusion of unpublished trials, improved follow-up information for published trials and greater power in subgroup analyses.

Discussion: IPD meta-analyses differ in many ways from published data meta-analyses, and in some ways from meta-analyses which are based on updated, aggregate data collected for both published and unpublished trials. Our preliminary findings confirm the importance of attempting to include all randomised participants from all relevant trials, with the maximum amount of follow-up information possible if as reliable an estimate as possible of the effects of different treatments is to be obtained. Our systematic review now needs to proceed to identify all of the comparisons of IPD meta-analyses versus published data meta-analyses regardless of the publication status of such studies.