Abstract:
|
Meta-analysis is a statistical technique that allows clinical trial data to be combined in a coherent way, respecting randomization. Network meta-analysis is an extension to situations involving more than two treatments, where direct treatment comparisons might not be possible. Such an approach can be highly valuable when attempting to gain an understanding of safety data analysis involving rare events, where it is necessary to synthesize data from all relevant trials. Within the context of safety, this talk will discuss the use of informative prior information associated with a background event rate. This approach deals with the controversial issue of 0 v 0 trials in a coherent and transparent way. Approaches based prior elicitation and modeling of external information will be discussed and illustrated with examples.
|