Statistical and graphical evidence synthesis methods in health technology assessment

This thesis focusses on the challenges relating to clinical- and cost-effectiveness analysis in Health Technology Assessment (HTA). It includes methodological developments, both statistical and presentational, in evidence synthesis aiming to address those challenges. In HTA, analysts often face problems with limited availability of data required to inform economic model. This thesis proposes innovative evidence synthesis approaches to address this challenge, illustrated in two examples. Bivariate random-effects meta-analysis (BRMA) and network meta-analysis (NMA) were used to synthesise all available evidence to predict progression-free survival (PFS), in metastatic prostate cancer. This enabled the specification of a three-state Markov model previously limited to two states when PFS was not recorded. In the second example, a scenario in multiple sclerosis is considered where utility data for the trials included in a HTA were not available and external utility data from a single study was used instead. This thesis illustrates how BRMA can be applied to include all available evidence to inform utility estimates for use in a cost-effectiveness analysis. NMA, allowing for a simultaneous and coherent comparison of multiple interventions, is increasingly used in HTA. However, due to the inherent complexity of presenting NMA results, it is important to ease their interpretability. A review of existing methods of presenting NMA results in HTA reports revealed that there is no standardised presentational tool for their reporting. Novel presentational approaches were developed which are presented in this thesis. The original contributions of this thesis are the innovative approaches to incorporate historical data to predict and increase the precision of parameter estimates for cost-effectiveness analysis to better inform health policy decision-making; and three novel graphical tools to aid clear presentation and facilitate interpretation of NMA results. Ultimately, the hope is that the graphical tools developed will be recommended in updated guidance setting the standards for future HTAs