The Meta-Analyses, A Challenge for Statisticians
Meta-analyses based on updated individual patient data provide the most comprehensive and reliable means of assessing the results of existing randomised clinical trials. It is the only reasonable way of performing time to event analyses, the best way of performing sub-group analyses and allows the review to use common prognostic and outcome variables. The detailed checking of data possible with this approach also improves the accuracy of the data included in the meta-analyses, allowing the integrity of the randomisation and follow up procedures to be assessed centrally. However, considerable expertise, time, effort and resource are required to carry out meta-analyses using individual patient data. They should not be undertaken lightly and might best be carried out by a secretariat on behalf of an international collaborative group. We hope that the guidance contained in this report will prove useful to such people.
Meta-analyses, Individual trials, Study design.