Network Meta-Analysis and Extrapolation of Survival Outcome Data

Health Technology Assessment (HTA) is used to inform policy decisions on whether to recommend treatments on the NHS, on the basis of whether health benefit gains represent good value for money. Impact of treatment on life-expectancy is usually a key driver of cost-effectiveness. Robust estimates of mean life expectancy differences between competing treatments are therefore required. This is challenging because: (i) randomised controlled trials (RCTs) typically follow up patients for only a few years; (ii) there may be multiple treatments and studies, requiring network meta-analysis; (iii) for cancer there are two related outcomes: progression-free and overall survival; (iv) treatment switching may occur after disease progression. Recently methods have been developed to extrapolate RCT evidence from a single trial over a lifetime, by calibrating it to registry data. This project will extend this approach to network meta-analysis of multiple trials for multiple treatments. This will involve flexible survival models (eg spline, fractional polynomial, or mixture models), consideration of joint modelling of progression-free survival and overall survival, and adjustment for treatment switching.

The student will learn complex cutting edge methods for network meta-analysis, analysis of survival data, combining evidence from different sources (RCT evidence and “real world evidence”), and treatment switching. They will develop new methods that combine all of these ideas, and apply them to recent examples in National Institute for Health and Care Excellence (NICE) technology appraisals. This project is suitable for a student with excellent mathematics/statistics skills. Extrapolation requires good understanding of disease natural history and mechanisms of action of treatments. The project will require the student to be able to communicate effectively with clinical experts to develop models that are clinically plausible.

The student will receive training in network meta-analysis, Bayesian methods, advanced survival analysis, and adjustment for treatment switching. The project is a collaboration between the Universities of Bristol and Exeter. They will have the opportunity to work on one or more current HTA projects at PenTAG (Exeter) to gain experience of methodology issues that arise in practice.