Summary tabular data provided by trialists

For summary data supplied by trialists, data tables should be scrutinized for any obvious errors or inconsistencies and where possible cross-checked with any publications. However, where data have been brought up-to-date or subdivided by patient types that were not reported in the publication, this may not be possible. These types of data are undoubtedly the most difficult to scrutinize as they have neither the benefit of peer review nor the detailed checking possibilities of IPD. Where such data supplement published data, common elements should be cross-checked with publications for consistency. Where summary data are from unpublished trials, the information supplied has largely to be taken at face value although some checks for internal consistency can be carried out. For example, the total numbers of individuals and total numbers of events should be the same for different subgroups supplied. It has been noted that there have been instances of meta-analyses of unpublished summary (and therefore unchallengeable) data, that show more favourable results than similar analyses of published data, and that have been produced in situations where there are vested financial interests [48].

0 0

Post a comment