Systematic Reviews and Meta-Analysis.

Lead Research Organisation: University College London

Abstract

A number of clinical trials comparing the same treatments might not show which treatment is best. In this situation, the CTU Meta-analysis Group searches for all the trials and describes them - a systematic review. Then, we put the results of these trials together in a meta-analysis. This gives a better picture of how good or bad a treatment is, than any single trial.|Most researchers use information written about trials for systematic reviews and meta-analyses. If trials are not written about or not written about in much detail, this method is not so reliable. Instead, we ask all the doctors who did the trials, for all the relevant information on each patient. This so called individual patient data method is much more reliable, but can take much longer. We are one of only a few groups worldwide that do systematic reviews and meta-analyses in this way.|In the past we have looked at treatments in ovarian, lung, oesophageal, bladder, brain, cervical and endometrial cancer, and sarcoma, as well as in pregnancy, human prion disease and venous leg ulcers. Now we have projects in lung and colorectal cancer, and in stroke.|We also research different ways of doing systematic reviews and meta-analyses, to try to improve how they are done.

Technical Summary

The Meta-analysis Group of the MRC Clinical Trials Unit (CTU) aims to rigorously evaluate the effectiveness of interventions for cancer and other diseases/conditions using systematic reviews and meta-analysis, to inform both clinical practice and research. We also aim to improve the quality and reliability of systematic reviews and meta-analyses through associated methodological research.|Moderate improvements in outcome are often the best that can be expected of new interventions, but these can be clinically worthwhile and important in terms of public health. Where randomised trials recruit too few patients to detect such modest differences reliably, they may have inconclusive results, or by chance, have results in favour of a particular intervention. Collating and appraising such trials in a systematic review, provides an objective summary of all the available evidence. Combining the results of these trials in a meta-analysis, can give sufficient statistical power to obtain a reliable and precise estimate of effect.|The Group predominantly conducts systematic reviews and meta-analyses based on individual patient data (IPD MA). These involve the collection, checking and re-analysis of the original IPD from all the relevant trials worldwide, necessitating extensive collaboration. Compared to standard systematic reviews based on trial-level data extracted from publications or obtained from trial investigators, the collection of IPD can improve the quality of data and breadth of analysis, and so it is considered the gold standard approach. As IPD MAs are resource-intensive and take time to complete, we also use standard systematic reviews to identify questions that require the IPD approach, when a more timely answer is needed, or when resource is limited. We are one of few groups worldwide with expertise in IPD MAs and have an international reputation in their conduct.|The Group has completed IPD MAs and other systematic reviews of therapies for ovarian, lung, oesophageal, bladder, brain and cervical cancer, and soft tissue sarcoma, as well as human prion disease, pre-eclampsia and venous leg ulcers. We have ongoing projects in lung and colorectal cancer, and in stroke. We are piloting consumer involvement in our IPD projects. Methodological projects have included investigations of publication and patient exclusion bias, and the development of methods for analysing published time-to-event data. Ongoing projects include comparing meta-analyses based on summary data with those based on IPD, evaluating the methods for assessing treatment by patient covariate interactions and methods for analysing IPD.|We act as the main CTU link to the Cochrane Collaboration, with members of the Group active in a number of Cochrane Review Groups and Methods Groups. We also co-convene the IPD Meta-analysis Methods Group.

Publications

10 25 50