Network meta-analysis of randomized controlled trials

A number of treatments may be available to patients with the same health condition. Policy-maker, clinicians and patients may want to know what the optimal treatment is. This requires comparing the benefits and harms of all available treatments. However, not all the direct comparisons are available from randomized trials. This makes the multiple comparisons challenging, so statistical advance is well motivated.

Network meta-analysis is a statistical method to compare multiple treatments simultaneously within a single framework. It can be implemented within either a frequentist or a Bayesian framework.

This Bsc project will review the network meta-analysis methodology and develop an application of the method, and involves programming in WinBUGS and/or R.


  • Lu, G., Ades, A.E. (2006) Assessing evidence consistency in mixed treatment comparisons. Journal of the American Statistical Association, 101, 447-459.
  • Ciprinani et. al. (2013) Conceptual and technical challenges in network meta-analysis. Annals of Internal Medicine, 159(2):130-197.

Supervisor: Dr Yinghui Wei