University of Cambridge > > MRC Biostatistics Unit Seminars > Models for baseline and treatment effects in meta-analysis

Models for baseline and treatment effects in meta-analysis

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact nikolaos.demiris.

The widely accepted approach to meta-analysis assumes that the trial-specific “baseline” effects are nuisance variables. This is well-suited to inference on the relative treatment effects. However, for cost-effectiveness analyses (CEAs) it is necessary to incorporate evidence on absolute effects. The RCT evidence is one possible source of data on absolute effects, and may sometimes be the preferred source. This presents a difficulty: if the trial baselines are nuisance parameters they cannot be used to inform a CEA analysis. But if we put a model on the baselines to feed into the CEA , then this changes the estimates of the relative treatment effects.

A “have-your-cake-and-eat-it-too” solution is tentatively proposed, in which the WinBUGS “cut” function is used to make a duplicate copy of treatment effect parameters which have been estimated in the standard way. Then, using the same data a second time, a model is put onto the trial baselines, while using the duplicate treatment effects as “priors”.

Examples involving mixed treatment comparisons will be presented. We compare the results obtained with different modelling assumptions, and consider alternative ways of developing predictions for a CEA from the posteriors. We also explore whether this approach might be used to “link” disconnected networks of randomised evidence.

This talk is part of the MRC Biostatistics Unit Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2019, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity