If you have a question about this talk, please contact Alison Quenault.
Abstract: Meta-analysis is usually univariate, estimating the mean treatment effect for the outcome of primary interest. However, many clinical studies will also measure secondary outcomes. Multivariate meta-analysis allows us to take these secondary outcomes into account, and also allows us to include studies where the primary outcome is missing. How much is gained by taking these secondary outcomes into account? In most statistical problems we get more accurate estimates if we include other data which are correlated with the main object of interest. But in meta-analysis, the contribution of any particular study’s secondary outcomes depends not only on the between-outcome correlations but, crucially, on how typical the study’s research design is of the meta-analysis as a whole. A graphical interpretation will be suggested, and illustrated using a recent meta-analysis of ten anti-hypertension clinical trials.
This talk is part of the MRC Biostatistics Unit Seminars series.