Potential costs of the IPD approach

It is often said that IPD reviews are much more time-consuming and expensive than other forms of systematic review [28]. However, this assumption is often based on comparison with meta-analyses of the published data that have been done with minimum effort, rather than with high quality systematic reviews, such as those carried out by the Cochrane Collaboration [6]. Furthermore, since resource comparisons were first discussed, technological progress has enabled some of the labour-intensive aspects of IPD meta-analysis to be done more easily, quickly and cheaply.

Most organizations now store their trial data on computer and most researchers have access to e-mail. This makes data collation and provision less time-consuming both for trialists providing data and for the team receiving it. Both the time taken to transfer data and the effort involved in assembling the meta-analysis database are reduced. Queries about data can easily be done by e-mail. Maintaining contact is also much easier using e-mailing lists and website bulletins. Software advances have also meant that data are now much more easily transferred between different types of database package, and that the format in which data are supplied is seldom a problem. Packages themselves have become more adaptable with many useful features. Little specialist programming expertise is now required to set up and run new IPD projects.

A key but time-consuming aspect of IPD meta-analysis is contacting trialists and persuading them to participate and provide data. However, the time taken to negotiate the provision of summary data is likely to be similar and time spent on this activity can be partially offset against time taken to contact authors for clarification of information presented in published reports. One aspect of IPD meta-analyses that adds to the cost and which is not encountered in other forms of review is funding collaborators' meetings where the collaborative group of trialists providing data and the researchers organizing the project meet to discuss the results. Although not normal practice, such meetings could also be useful where trialists supply summary data.

0 0

Post a comment