|
Activity Number:
|
350
- Statistical Issues In Drug Development - 1
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2018 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #329958
|
|
|
Title:
|
Network Meta-Analysis: On the Use of the Standard Contrast-Based Approach in Disconnected Networks
|
|
Author(s):
|
Audrey Béliveau* and Paul Gustafson
|
|
Companies:
|
University of Waterloo and University of British Columbia
|
|
Keywords:
|
network meta-analysis;
mixed treatment comparisons;
disconnected networks;
Bayesian;
contrast-based
|
|
Abstract:
|
Network meta-analysis is a methodology used to compare the efficacy and/or safety of multiple medical interventions by synthesizing data across clinical studies. Most of the current literature focuses on connected networks and although disconnected networks commonly arise, their analysis is usually avoided as there is not currently a gold-standard method for analyzing disconnected networks. A standard method for analyzing connected networks is the contrast-based approach of Lu and Ades (2004). This approach is deemed inappropriate for disconnected networks (e.g. see the empirical results of Goring et al., 2016) but there is currently a lack of theoretical work to justify this claim as a general assertion. In this paper, we show using a Bayesian framework that in general the contrast-based approach of Lu and Ades (2004) does not work in disconnected networks. We show that the posterior variance on the relative effects of disconnected treatment is not updated significantly from the prior distribution. This talk will provide a brief overview of our theoretical developments and an illustration of those theoretical results using simulated data.
|
Authors who are presenting talks have a * after their name.
